Task Order 68HERH20F0407 under
Contract EP-C-17-017
External Panel Peer Review of EPA's
Draft "IRIS Toxicological Review of
Perfluorodecanoic Acid [PFDA, CASRN 335-
76-2] and Related Salts"
FINAL PEER REVIEW REPORT
October 2023
Submitted to:
U.S. Environmental Protection Agency
Office of Research and Development
Center for Public Health and Environmental Assessment
Attn: Vicki Soto
Soto.Vicki (5) E PA.gov
Submitted by:
ERG
561 Virginia Road
Building 4 (Suite 300)
Concord, MA 01741
*ERG
www.erg.com
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
CONTENTS
1.0 INTRODUCTION
1
1.1 Background
1
1.2 Peer Review Process
1
2.0 SUMMARY OF KEY REVIEWER COMMENTS BY CHARGE QUESTION
2
3.0 REVIEWER RESPONSE TO CHARGE QUESTIONS
15
3.1 The Toxicological Review for PFDA describes and applies a systematic review protocol for
identifying and screening pertinent studies. The protocol is described in brief detail in
Section 1.2.1 (Literature Searching and Screening) and in full detail in Appendix A
3.2. For each health effect considered in the assessment and outlined below, please comment
on whether the available data have been clearly and appropriately synthesized to
describe the strengths and limitations, including whether the presentation and analysis of
study results are clear, appropriate, and effective to allow for scientifically supported
syntheses of the findings across sets of studies. Please comment on whether the study
confidence conclusions for the PFDA studies are scientifically justified, giving appropriate
consideration to important methodological features of the assessed outcomes. Please
specify any study confidence conclusions that are not justified and explain any alternative
study evaluation decisions. For each, please also comment on whether the weight-of-
evidence decisions for hazard identification have been clearly described and scientifically
justified. Note that the data from studies considered informative to the assessment are
synthesized in the relevant health effect-specific sections and available in the Health
Assessment Workspace Collaborative (HAWC) 31
3.3 For PFDA, no RfC was derived for inhalation exposures. An RfD is derived based on
studies by Budtz-Jorgensen and Grandjean (2018) and Grandjean et al. (2012) showing
decreased serum antibody concentrations for both tetanus and diphtheria in children
(male and female) at age seven years and PFDA measured at age five years and
developmental effects (i.e., reduced birth weight in humans) from the Wikstrom (2020)
study. Given the close proximity of the developmental and immune PODs and resulting
osRfDs and because these effects are observed during the developmental period, they
are selected as co-critical effects supporting the RfD. Are the selection of the studies for
the immune (Budtz-Jorgensen and Grandjean, 2018) and developmental (Wikstrom,
2020) effects for use in deriving the RfD values for PFDA scientifically justified? Are the
modeling approaches appropriate? 99
3.4 In addition, for PFDA, an RfD for less-than-lifetime ("subchronic") exposures is derived.
No subchronic RfC was derived. The same studies and outcomes were chosen for use in
deriving the lifetime and subchronic RfDs. Are the selection of these studies and these
effects for the derivation of the subchronic RfD for PFDA scientifically justified? 126
3.5 Appendix G identifies the potential for pharmacokinetic (PK) differences across species
and sexes as a key science issue and lays out a hierarchy for using relevant PK data in
extrapolating doses between laboratory animals and humans. Section 3.1 evaluates and
(Systematic Review Protocol for the PFAS IRIS Assessments). Please:
15
i
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
synthesizes the PK data in relevant species and sexes, and among human lifestages, up to
the derivation of key PK parameters used in the subsequent analysis. However, the
evaluation of existing PBPK models and a one-compartment PK model found that these
options were not sufficiently reliable for use. Given the information available on potential
interspecies differences in PFDA PK, EPA applied a data-derived extrapolation factor
(DDEF) to POD values from toxicity studies in laboratory animals to estimate
corresponding human equivalent doses (HEDs) in the derivation of the respective RfDs.
Similarly, the estimated human clearance (CL) was used to convert internal dose POD
(PODint) values from epidemiological analyses to corresponding HEDs 146
3.6 EPA has evaluated and applied where appropriate uncertainty factors to account for
intraspecies variability (UFH), interspecies differences (UFA), database limitations (UFD),
duration (UFS), and LOAEL-to-NOAEL extrapolation (UFL) for PFDA 155
3.7 The Toxicological Review concludes there is inadequate information to assess
carcinogenic potential for PFDA and that this descriptor applies to oral and inhalation
routes of human exposure. Please comment on whether the available human, animal and
mechanistic studies, and the analysis presented in the Toxicological Review are
scientifically justified and clearly described 164
3.8 Given the conclusion there was inadequate information to assess carcinogenic potential
for PFDA, the Toxicological Review does not derive quantitative estimates for cancer
effects for oral or inhalation exposures. Is this decision scientifically justified and clearly
described? 166
4.0 ADDITIONAL COMMENTS
168
APPENDIX A LIST OF REVIEWERS
A-l
APPENDIX B CHARGE TO REVIEWERS
B-l
APPENDIX C MEETING AGENDA
C-l
ii
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
1.0 INTRODUCTION
This report documents the results of an external independent peer review of the U.S. Environmental
Protection Agency's (EPA's) draft "IRIS Toxicological Review of Perfluorodecanoic Acid [PFDA. CASRN 335-76-
21 and Related Salts." ERG, a contractor to EPA, organized this review and developed this report. The peer
review included a virtual meeting that was open to the public as observers and included an opportunity for
oral public comment (in addition to opportunity for the public to submit written comments to EPA via the
PFDA docket).
Section 1.0 provides background about the review. Section 2.0 provides a high-level summary of reviewer
comments. Section 3.0 presents reviewer final individual post-meeting comments. In Section 3.0, reviewers'
final written comments are organized by charge question and presented exactly as submitted, without
editing or correction of typographical errors (if any). Appendices A, B, and C, respectively, provide the list of
reviewers, EPA's charge to reviewers, and the peer review meeting agenda.
1.1 Background
In 2023, ERG organized and managed an external peer review of EPA's draft "IRIS Toxicological Review of
Perfluorodecanoic Acid [PFDA, CASRN 335-76-2] and Related Salts," developed in support of the Agency's
online database, the Integrated Risk Information System (IRIS). IRIS assessments contain information about
chemicals that encompasses hazard identification and dose-response assessment, two of the four steps in
the human health risk assessment process. When used by risk managers in combination with information on
human exposure and other considerations, IRIS assessments support the Agency's regulatory activities and
decisions to protect public health. There is no existing IRIS assessment for PFDA. EPA's draft Toxicological
Review of PFDA is based on a comprehensive review of the available scientific literature on the noncancer
and cancer health effects in humans and experimental animals exposed to PFDA or related salts.
1.2 Peer Review Process
During the first stage of the peer review process, ERG assembled a set of experts interested in serving as
reviewers, who collectively spanned three key areas of expertise required by EPA for this and four other
related peer reviews: environmental epidemiology, experimental toxicology, and the use of quantitative
methods (e.g., dose-response modeling, PBPK model development) important for the derivation of toxicity
values in human health assessments of environmental chemicals. To identify candidates, ERG used standard
search processes and considered experts nominated by the public in response to a Federal Register Notice
(FRN) requesting nominations. After considering comments on these candidates submitted by members of
the public in response to a second FRN, ERG assembled a final pool of 20 experts from which to select
reviewers for this and the four other related peer reviews. For this PFDA review, ERG selected the following
nine experts after confirming they had no conflict of interest for this review:
• John L. Adgate, Ph.D., MSPH
• Courtney C. Carignan, Ph.D.
• Elaine M. Faustman, Ph.D., DABT (Panel Chair)
• Jeffrey W. Fisher, Ph.D.
• Panagiotis G. Georgopoulos, Ph.D.
• Joseph T. Haney, Jr., M.S.
• Alan M. Hoberman, Ph.D., DABT
1
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Angela M. Leung, M.D.
• R. Thomas Zoeller, Ph.D.
See Appendix A for a detailed list of reviewers.
ERG provided reviewers with the draft PFDA toxicological review document and with EPA's charge to
reviewers (Appendix B), which asked reviewers to address each of the eight questions and multiple sub
questions and to categorize their advice to EPA into three tiers:
• Tier 1: Necessary Revisions - Use this category for any revisions you believe are necessary to
adequately support and substantiate the analyses or scientific basis for the assessment conclusions.
• Tier 2: Suggested Revisions - Use this category for any revisions you encourage EPA to implement to
strengthen the analyses or scientific basis for the assessment conclusions, or to improve the clarity
of the presentation in the PFDA Toxicological Review.
• Tier 3: Future Considerations - Use this category for any advice you have for scientific exploration
that might inform future work. While these recommendations are generally outside the immediate
scope or needs of the PFDA Toxicological Review, they could inform future reviews or research
efforts.
For their consideration, ERG also provided reviewers with written public comments submitted to EPA's
docket for this review and with a table developed by EPA that listed published literature identified (1) during
EPA literature search updates after release of the draft PFDA toxicological review for public comment or (2)
in public comments received through the EPA docket.
Working individually, each reviewer prepared written pre-meeting comments in response to the charge
questions, and ERG compiled and distributed these preliminary comments to all reviewers a few days prior
to the peer review meeting to help them prepare for discussions at the meeting.
ERG organized and facilitated a virtual peer review meeting, which took place via Zoom.gov on July 10, 11
and 13, 2023. The meeting was open to members of the public to attend as observers and provided an
opportunity for members of the public to make an oral comment. During this meeting, reviewers discussed
and commented on EPA's draft PFDA Toxicological Review, with discussion structured by EPA's charge
questions. Appendix C provides the meeting agenda. After the meeting, reviewers prepared their individual
final post-meeting comments (see Section 3.0) and ERG prepared a high-level summary (Section 2.0) of
reviewer comments (Appendix B).
2,0 SUMMARY OF KEY 1EFIEWE1 COMMENTS BY CHARGE QUESTION
This section provides a high-level summary of reviewer comments organized by charge question. For
reference, EPA's charge questions are shown in italic font.1 Note that:
• The summary focuses on key comments. Therefore, it summarizes selected Tier 1, 2, and 3
comments, as well as key comments that reviewers did not categorize into a tier. For all tiered
comments, the tier in which the reviewer categorized the summarized comment is provided in
parentheses. If no tier is indicated, the reviewer did not assign a tier to that comment.
1 For simplicity, the italicized text does not include charge question text that provides background or summarizes EPA's
rationale. The italicized text does include charge question text that poses a question or otherwise requests comment.
For the full text of EPA's charge to reviewers, see Appendix B.
2
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Occasionally, the summary includes a key comment a reviewer made during the peer review
meeting but did not include in his or her final written comments. In such cases, the summary
clarifies that the reviewer made the comment during the meeting.
Please see Section 3 for the full text of all reviewer post-meeting comments, including comments not
included in this summary and the original and compete text of comments that are included.
Literature Search Methods and Documentation
la. Comment on whether the literature search strategy and screening criteria for PFDA are appropriate and
clearly described.
• While several reviewers provided Tier 1 comments with suggested revisions, all nine reviewers
agreed that the literature search strategy and screening criteria are clearly described and the
process is well documented (Adgate, Carignan, Faustman, Fisher, Georgopoulos, Haney, Hoberman,
Leung, Zoeller).
• Four reviewers commented that the description of the exclusion criteria should be expanded,
particularly around populations, exposures, comparators, and outcomes (PECO) exclusion and Figure
2-1 (Faustman [Tier 2], Carignan [Tier 1], Georgopoulos, Leung [Tier 1]).
• Three reviewers commented that the exposure section should be updated (Adgate [Tier 3], Carignan
[Tier 1], Georgopoulos [Tier 1]).
lb. Identify additional peer-reviewed studies of PFDA that EPA should consider incorporating prior to
finalizing the assessment.
• Leung, Zoeller, Fisher, Haney, and Hoberman did not identify any additional studies of PFDA for EPA
to consider prior to finalizing the assessment.
• Carignan and Georgopoulos each provided a few studies for EPA to consider, and Faustman
identified several review articles.
• Carignan noted a lack of literature on mammary gland development and identified several peer-
reviewed studies on the subject (Tier 1).
• Two reviewers (Fisher [Tier 2] and Georgopoulos [Tier 3]) noted that a list of ongoing studies on
PFDA should be recorded and that EPA should describe how these studies would be considered as
results become available.
1c. Review EPA's characterization and provide tiered recommendations regarding which studies, if any, would
have a material impact on the draft's conclusions and should be incorporated into the assessment before
finalizing, as well as your interpretation of the impact of those studies to be incorporated.
• Adgate, Carignan, Fisher, Georgopoulos, Haney, Hoberman, Leung, and Zoeller did not note any
material impact of the new studies on the draft's conclusions.
• Faustman commented that EPA should clarify how the updated literature review will be
incorporated into the document (Tier 1). Adgate similarly suggested that EPA should enumerate
additional studies and describe how they fit into the larger picture of the assessment.
3
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Georgopoulos commented that new studies may help identify data gaps for future research (this
comment was not tiered) and recommended developing a plan for the systematic and regular
updating of database that tracking information relevant to the Toxicological Reviews (Tier 3). Fisher
(Tier 2) also noted that tracking ongoing studies may be beneficial. Faustman recommended
clarifying how additional studies marked for inclusion would be included in the draft (Tier 2).
Noncancer Hazard Identification
2. For each health effect considered in the assessment and outlined below, please comment on whether
the available data have been clearly and appropriately synthesized to describe the strengths and
limitations, including whether the presentation and analysis of study results are clear, appropriate, and
effective to allow for scientifically supported syntheses of the findings across sets of studies. Please
comment on whether the study confidence conclusions for the PFDA studies are scientifically justified,
giving appropriate consideration to important methodological features of the assessed outcomes. Please
specify any study confidence conclusions that are not justified and explain any alternative study
evaluation decisions. For each, please also comment on whether the weiaht-of-evidence decisions for
hazard identification have been clearly described and scientifically justified.
2a. Liver effects.
• Seven reviewers (Leung, Carignan, Faustman, Fisher, Georgopoulos, Hoberman, Zoeller)
commented that available data are clearly and appropriately synthesized and study confidence
conclusions are scientifically justified and appropriate, including with regard to characterizing the
PPPRa.
• Adgate disagreed with the characterization of evidence of liver effects in humans as "slight" and
noted that evidence is between "slight" and "moderate." He thought that additional discussion of
the coherence between animal and human studies would be helpful (Tier 2).
• While Haney generally noted agreement with EPA's text on liver effects, he suggested that EPA
make revisions to clarify the selection of species deemed to be the most biologically
representative of humans such that the same or similar effects are expected in humans (Tier 1).
Haney also commented that EPA should explicitly show that the Hall et al. (2012) criteria for
adversity are met (Tier 2).
2b. Immune effects.
• Faustman and Leung agreed with the EPA's overall conclusions and had no Tier 1 or 2
recommendations.
• Five reviewers (Adgate, Carignan, Fisher, Georgopoulos, and Hoberman) agreed with EPA's
overall conclusions but also made some suggestions for improvement.
• Zoeller commented that EPA needs to explain its reasoning more fully in this section, including
adding further discussion of confounding by other PFAS and the use of diphtheria and tetanus
antibodies as an index of immune suppression (Tier 2).
• Adgate noted a need for further discussion of the sensitivity and uncertainty inherent in the cited
4
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
human epidemiology studies and for further comment on the generalizability of these studies to
U.S. populations (Tier 1).
• Carignan provided several comments on the number, quality, and presentation of studies in this
section of the review document:
o She recommended clarifying the number and quality of studies on immunostimulation
and autoimmunity to explain why these outcomes were excluded (Tier 1).
o She recommended including the dichotomous titer outcome from Timmerman et al. (2021)
in Table 3-12 (Tier 1).
o She commented that it was confusing that the final data set used for developing a
reference dose is not included in the document. This, along with a prominent clarification
about units, should be added to Table 3-12 (Tier 1).
• During the peer review meeting, multiple reviewers (Adgate, Fisher [Tier 2], Zoeller [Tier 2],
Haney [Tier 1], and Hoberman) noted that discussion of mixtures and confounding needed to be
further developed in the document. Georgopoulos noted that confounding effects from other
exposures could impact estimates used to derive RfDs (Tier 1), and Haney argued that issues of
mixtures and confounding in the epidemiology studies, combined with the low effect estimates,
precluded the use of these studies for quantitative risk assessment and derivation of toxicity
factors (Tier 1).
• Haney disagreed with EPA's conclusion on the strength of evidence for human
immunosuppression from the epidemiological studies, citing concerns about the strength of
unadjusted associations and the necessity of considering confounding from correlated PFAS
exposures, which further reduced the strength and significance of associations (Tier 1). In
addition, Haney did not agree that the measures of immunosuppression (titers) were clinically
significant, particularly for rare diseases such as diphtheria and tetanus (Tier 1).
• Carignan agreed with the use of immune titers as an endpoint for the epidemiology studies and
suggested that EPA prominently state their rationale (Tier 1).
2c. Developmental effects.
• Fisher, Georgopoulos, and Faustman agreed with what EPA had written in this section and had no
suggested Tier 1 or 2 revisions.
• Adgate agreed with the strength of evidence and noted that the structure of the evidence
integration section should be revised to increase the section's coherence (Tier 2).
• Carignan, Hoberman, and Leung agreed with the conclusions made by the EPA and thought that
study data were clearly and appropriately synthesized and conclusions were scientifically
justified. They also provided several suggested revisions.
o Carignan disagreed with the confidence rating for Bach et al. (2016), recommended the
inclusion of a new study (Padula et al., 2023), and recommended that EPA include an
5
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
indicator for studies that used gestational age-adjusted birth weight (Tier 1).
o Hoberman suggested adding discussion of pubertal development as a developmental
endpoint (Tier 1).
o Leung suggested identifying the possible exposure windows for each study (Tier 1).
• Zoeller commented that the argument that uncertainty around the relationship between sample
timing differences and reported fetal growth restriction deficits needs to be strengthened,
particularly in the areas of weight-of-evidence designation (Tier 1).
• Haney noted that since there is considerable uncertainty and only 'slight' evidence provided by
the epidemiology studies, he recommends using the Harris and Birnbaum (1989) mouse study to
develop RfDs rather than a single epidemiology study as is currently done by the EPA (Tier 1).
2d. Male reproductive effects.
• Adgate, Carignan, Georgopoulos, Haney, Hoberman, Leung, and Zoeller agreed with EPA's
conclusions and would have liked to have longer-term data.
• Carignan recommended adding a discussion of the non-monotonicity of testosterone effects and
how this may explain inconsistency among male reproductive endpoints in human studies (Tier
2).
• Haney noted that characterization of the body weight reduction for rats as "moderate" was
inconsistent with typical characterization of body weight reduction and should be further
explained (Tier 1).
• Fisher did not comment on this topic.
2e. Female reproductive effects.
• Adgate, Carignan, Faustman, Fisher, Georgopoulos, Hoberman, Zoeller, and Haney agreed with
EPA's conclusions and would have liked longer-term data and more mechanistic studies.
• Carignan noted that lactation duration should be included as an outcome as the pregnancy and
postpartum periods are a sensitive developmental window for mammary gland development
(Tier 1).
• Leung commented that the report's conclusion that there is a likely association between PFDA
exposure and female reproductive toxicity may be too strong given the sparse data of the single
rat study (Tier 2).
• Haney commented that characterization of the body weight reduction for rats as "moderate" was
inconsistent with typical characterization of body weight reduction and should be further
explained (Tier 1).
6
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
2f. Cardiometabolic effects.
• All nine reviewers agreed with EPA's conclusions.
• Adgate added that the justification would be more coherent if there were a logic tree to walk the
reader through the pathways by which PFDA affects each contributor to overall cardiometabolic
health (Tier 2).
• Similarly, Leung suggested that EPA add lipid subfractions, where available, in addition to total
cholesterol levels, as some components are protective while others are associated with adverse
outcomes (Tier 2).
2g. Neurodevelopmental effects.
• All reviewers agreed with EPA's conclusions.
• Adgate commented that adding ages or age categories and gender, when feasible and clinically
relevant, to Table 3-38 would increase the utility of this table in clearly synthesizing the studies
(Tier 2).
2h. Endocrine, urinary, and other noncancer effects.
• Adgate, Georgopoulos, Hoberman, Haney, and Carignan agreed with EPA's conclusions.
• Faustman and Leung noted agreement with EPA's conclusions on urinary and other non-cancer
endpoints.
• Faustman commented that designation of thyroid studies as "deficient" because of non-fasted
measurements or variation in time-of-day was inappropriate; Leung and Zoeller concurred (Tier
1).
• Faustman also commented that the interpretation of studies regarding thyroid pathway
responses was different from other recent PFAS IRIS documents and other current PFAS research
(Tier 1). In the peer review meeting, Fisher seconded this and noted that he thought that EPA had
interpreted the thyroid data incorrectly (Tier 1). Zoeller concurred. These three reviewers
disagreed with EPA's conclusion that the thyroid data was "incoherent" (Tier 1).
• Leung commented that studies of thyroid hormones in pregnant people and non-pregnant people
should be considered separately; exposure period should also be stated, particularly for studies
concerning maternal and/or neonatal exposure (Tier 1).
Noncancer Toxicity Value Data Selection and Modeling
3. Are the selection of the studies for the immune (Budtz-Jorqensen and Grand jean, 2018) and
developmental (Wikstrom, 2020) effects for use in deriving the RfD values for PFDA scientifically
justified? Are the modeling approaches appropriate?
• Seven reviewers (Adgate, Carignan, Faustman, Fisher, Georgopolous, Hoberman, and
Zoeller) agreed that Butdtz-Jorgensen and Grandjean (2018) and Wikstrom (2020) were
7
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
scientifically justified and that the modeling approaches were appropriate for deriving
the RfD values for PFDA for immune and developmental effects, respectively.
• Leung did not comment in response to this question.
• Haney commented that the selection of the studies was not scientifically justified and
appropriate for modeling RfDs for immune and developmental effects of PFDA.
• Carignan added that future research should consider mixtures and that EPA should
acknowledge dermal exposure routes; Adgate and Georgopoulos agreed (Tier 3).
3a. If so, please provide an explanation.
• Adgate commented that these studies were reliable sources given the underlying data,
but that EPA should present the data more clearly, including comment on clinical
relevance, uncertainties, and justification of the choices made in the derivation of the
RfDs (Tier 2). Faustman, Fisher, and Leung concurred on this point during the meeting.
• Carignan added that selection of co-critical effects is reasonable given the proximity of
the points of departure for the effects and that they are both observed during the
developmental period.
• Georgopoulos recommended that EPA add additional discussion of confounding in the
selected studies into the document (Tier 1). Fisher agreed; he recommended that EPA
acknowledge potential confounding by PFAS mixtures in the immune studies and
explore the possibility of assigning relative potency for each PFAS molecule or leaning
into computational toxicology methods (Tier 3).
3b. If not, please provide an alternative study(ies) or effect(s) that should be used to support the
derivation of the lifetime RfD and detail the rationale for use of such an alternative.
• Haney did not agree with the use of these studies to develop RfDs for immune and
developmental effects. He noted that for immune effects, Budtz-Jorgensen and Grandjean
(2018) reported inconsistent and/or insignificant associations between antibody
concentrations and PFDA, did not adjust for confounding from associated other PFAS, defined
the "clinically protective" level of serum antibodies as a higher concentration than typical for
the assay, and also noted no evidence of increased incidence of tetanus or diphtheria, despite
NHANES data indicating that the U.S. population geometric mean serum level of PFDA is
higher than the POD. Further, Haney commented that EPA did not adequately consider other
studies with null findings. Haney suggested sorting study data extracted by NOAEL/LOAEL,
BMD, or a similar criterion to identify the most sensitive effects based on "less problematic"
studies (Tier 1). For birthweight, Haney commented that EPA should use a metanalysis or data
from Harris and Birnbaum (1989) rather than Wikstrom (2020), given the weight of evidence
designation for the epidemiological literature as "slight." Further, confounding from correlated
PFAS is of concern when using Wickstrom (2020) to calculate RfDs (Tier 1).
8
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
3c. As part of the recommendations in "a" or "b" above, please comment on whether the effects selected
are appropriate for use in deriving the lifetime RfD, including considerations regarding adversity (or
appropriateness in representing an adverse change) and the scientific support for their selection. Please
also see charge guestions 2b and 2c.
• Carignan, Hoberman, and Faustman agreed that the selected effects are appropriate for use in
deriving the lifetime RfD; Faustman suggested adding an acknowledgment of the broader range of
pharmacokinetic differences associated with pregnancy that could affect the interpretation of the
epidemiology studies beyond what is currently discussed for the birth weight endpoint (Tier 2).
3d. Are the benchmark dose modeling (BMP) modeling approaches, selection and justification of
benchmark response levels, and selection of the BMP models used to identify each POD for toxicity
value derivation scientifically justified and clearly described?
• Carignan, Adgate, Faustman, Hoberman and Zoeller agreed that the BMD modeling
approaches, selection and justification of BMRs, and selection of BMD models used to identify
each POD for toxicity value derivation are scientifically justified and clearly described.
• Fisher commented that while some aspects are clearly described, the issue of mixtures and
confounding from other associated PFAS is not adequately addressed. EPA should add text on
the strengths and weaknesses of the assumptions and methodology using non-epidemiological
language. Georgopoulos expressed a similar concern about not adjusting for confounding from
correlated PFAS, including PFNA (Tier 1).
• Haney opined that the modeling approaches, model selection process, and BMRs used to
derive PODs are scientifically justified; however, the draft assessment is not using the
demonstrated severity/adversity of the effect itself (decreased antibodies) as support for the
lower BMR for immune effects and is using an internally inconsistent definition of an adverse
effect level to justify the BMR for immune effects. Haney disagreed with the justification EPA
provided for the use of the lower BMR for immune effects and recommended removing the
justification (Tier 1). Haney also suggested using Harris and Birnbaum (1989) as the basis of an
osRfD based on decreases in fetal bodyweight(Tier 1).
3e. For liver, male reproductive and female reproductive effects, quantitative information was limited to
studies in animals exposed to PFDA for 28 days and little to no information was available to evaluate the
effects of chronic exposure on these health hazards. Therefore, the derivation of lifetime organ-specific (os)
RfD values was not attempted for liver, male reproductive and female reproductive effects. However, these
endpoints were considered for the derivation of subchronic osRfDs. Does the provided scientific rationale
support this decision? Pleaseexplain.
• Six reviewers (Carignan, Faustman, Georgopoulos, Haney, Hoberman, and Zoeller) agreed with the
scientific rationale to consider deriving subchronic osRfDs but not lifetime osRfDs due to the
limited quantitative information.
• Carignan commented that future research should involve longer-term studies appropriate for
deriving lifetime osRfDs (Tier 3).
9
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Haney suggested increasing the uncertainty factor range to estimate chronic osRfDs (Tier 2).
• Fisher commented that the Kim et al. (2019) PBPK model should have been modified for PFDA to
derive subchronic osRfDs, as a PBPK model includes organs and tissue groups in a physiologically
relevant manner and can predict internal organ dosimetry (Tier 1).
3f. Given the lack of studies on inhalation exposure to PFDA, no reference concentration (RfC) is derived.
Please comment on this decision.
• All nine reviewers agreed with EPA's decision not to derive an RfC for inhalation exposure to
PFDA due to the lack of data. However, several reviewers suggested exploring methods of
gaining information about inhalation exposure to PFDA:
o Adgate suggested an analysis of the physical properties of PFDA and the potential
resulting air concentrations in indoor environments (Tier 3).
o Carignan suggested that information on inhalation as an exposure route should be
gleaned from other PFAS with similar properties (Tier 3).
o Haney suggested that EPA clearly state that they have not identified a reliable PBPK/PK
model for route-to-route extrapolation to support the decision not to derive a RfC (Tier
2).
o Fisher suggested that EPA consider using multi-path dosimetry software with a PBPK
model to simulate inhalation uptake of particles containing PFDA (Tier 3).
4. In addition, for PFDA, an RfD for less-than-lifetime ("subchronic") exposures is derived. No subchronic
RfC was derived. The same studies and outcomes were chosen for use in deriving the lifetime and
subchronic RfDs. Are the selection of these studies and these effects for the derivation of the subchronic
RfD for PFDA scientifically justified?
• Carignan, Faustman, Georgopoulos, Hoberman, and Zoeller agreed with the selection of the
studies and effects for the derivation of the subchronic RfD for PFDA.
• Adgate commented that the derivation of a subchronic RfD is not strictly necessary considering
that the exposure will be chronic since the chemical is highly persistent (Tier 3).
• Citing the same concerns as in his response to charge question 3, Haney disagreed with the
appropriateness of these studies to derive subchronic RfDs for immune and developmental
endpoints.
• Leung refrained from comment, explaining that toxicological modeling is not her area of
expertise.
• All reviewers noted similarities in their responses to charge question 3 with their responses to
this question.
10
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
4a. If so, please provide an explanation.
• In general, reviewers referred to their explanations provided for charge question 3.
4b. If not, please provide an alternative study(ies) or effect(s) that should be used to support the
derivation of the subchronic RfD and detail the rationale for use of such an alternative.
• Haney suggested the use of Harris and Birnbaum (1989) as the basis of a subchronic osRfD based
on decreases in fetal body weight (Tier 1). Haney also recommended conducting a more
complete evaluation of alternative studies and effects, with one option for such an evaluation
being to sort study data by duration, NOAEL/LOAEL, and BMD or a similar criterion that would
allow EPA to identify the next most sensitive effects based on more robust studies that are
adequate for subchronic RfD derivation (Tier 1).
4c. As part of the recommendations in 4a or 4b, please comment on whether the effects selected are
appropriate for use in deriving the subchronic RfD, including considerations regarding adversity (or
appropriateness in representing an adverse change) and the scientific support for their selection.
• Fisher suggested adding any existing vitro studies to Table 5-16 for organ-specific endpoints (Tier
2).
4d. Please comment on the other subchronic osRfDs (i.e., for liver, male reproductive, and female
reproductive effects).
• Faustman, Georgopoulos, and Hoberman noted agreement with EPA's selected endpoints and
with the choice of studies used to derive the subchronic osRfDs.
• Haney found the rationale provided by the EPA for these subchronic osRfDs to be reasonable; he
further suggested that the uncertainty factor could be increased from 1,000 to 3,000 to estimate
chronic osRfDs using these studies (Tier 2).
4e. Given the lack of studies on inhalation exposure to PFDA, no subchronic RfC is derived. Please
comment on this decision.
• Adgate, Carignan, Faustman, Fisher, Georgopoulos, Hoberman, and Zoeller agreed with the
EPA's decision not to calculate a subchronic RfC due to the lack of data.
• Haney suggested that EPA clearly state in the document that they have not identified a reliable
PFDA PBPK/PK model from which they could perform route-to-route extrapolation (Tier 2).
Noncancer Toxicity Value Pharmacokinetic Extrapolation and Uncertainty Factors
5a. Is applying the estimated DDEF values for PFDA scientifically justified for conversion of POPs from animal
toxicity studies to HEDs? If not, please provide an explanation and detail on a more appropriate approach.
• Adgate, Faustman, Haney, and Hoberman commented that the approach is scientifically justified,
given the shortcomings of the underlying data.
11
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Carignan and Georgopoulos agreed that the approach is scientifically justified, but the presentation
in the document is complicated and the justification is not readily apparent in the document text
(Tier 1).
• Fisher disagreed with the use of DDEFs and encouraged EPA to pursue a PBPK model beyond what
has been published by EA authors (such as Bernstein et al., 2021) and to consider harmonization or
read across from other PFAS (Tier 1).
• Georgopoulos and Haney commented on the uncertainty introduced because serum PFDA levels
would not reach steady-state by the end of the 28-day dosing period in the animal studies.
Georgopoulos suggested using average measured PFDA serum levels over the course of the study or
the maximum PFDA levels at the end of the dosing period to determine the POD for benchmark
dose modeling (Tier 1).
• Leung and Zoeller refrained from commenting on charge question 5, explaining that toxicological
modeling is not within their areas of expertise.
5b. Is application of the human CL to estimate HEDs from PODint values scientifically justified? If not, please
provide an explanation and detail on a more appropriate approach.
• Faustman and Hoberman commented that the application of the human CL to estimate HEDs from
PODint values is scientifically justified.
• Georgopoulos commented that the approach is justified, assuming that a reliable value of human
clearance can be estimated. In the document, EPA assumes that the ratio of fecal to urinary
clearance is the same in humans and rats; this has not been established and requires further
discussion in the document (Tier 1).
• Carignan recommended that EPA indicate where assumptions are based on animal data and discuss
the associated uncertainties in the document (Tier 1).
5c. Have the uncertainties in the DDEFs and human CL been adequately evaluated and described? In
answering this question, please provide an explicit recommendation on whether or not EPA should expand its
adjustment for menstrual fluid loss as outlined in (U.S. EPA, 2023, HERO ID 11181055) prior to finalizing the
assessment. As these newer data are from other PFAS, note that such an expansion would be based on the
assumption that the pharmacokinetic effect of pregnancy and lactation on PFDA is similar to that of the
other PFAS (i.e., a read-across based interpretation).
• Fisher and Haney commented that the uncertainties had been adequately evaluated and described.
• Adgate, Carignan, and Georgopoulos recommended that the document discuss uncertainty further
(Tier 1), including a section on evidence integration and the magnitude and directionality of
uncertainty.
• Faustman commented that EPA should not expand its adjustment for menstrual fluid loss as the
adjustment would rely on many assumptions about menstrual flow. At the meeting, Carignan
concurred with the notion that the adjustment relies on many assumptions, and in her written
12
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
comments, Carignan noted that if EPA does adjust for menstrual fluid loss, the agency will need to
consider that not all women menstruate, birth children, or breastfeed (Tier 1).
• Fisher, Haney (Tier 1), Hoberman (Tier 3), and Georgopoulos commented that EPA should expand its
adjustment for menstrual fluid loss.
• Faustman noted that the document lacks a discussion of PFDA transfer via breast milk (Tier
1).
6. EPA has evaluated and applied where appropriate uncertainty factors to account for
intraspecies variability (UFH), interspecies differences (UFA), database limitations (UFp), duration
(UFS), and LOAEL-to-NOAEL extrapolation (UFl_) for PFDA.
• Leung refrained from commenting, explaining that this topic was not her area of expertise.
6a. Is uncertainty in the derivation of the toxicity values scientifically justified and clearly
described? Please describe and provide comments, if needed.
• Adgate, Faustman, Haney, and Hoberman agreed that uncertainty in the derivation of the
toxicity values was scientifically justified and clearly described.
• Georgopoulos noted several inconsistencies regarding uncertainty factors for intraspecies
variability, interspecies differences, database limitations, duration, and LOAEL-to-NOAEL
extrapolation within the main document and appendices. Georgopoulos recommended
ensuring that the values of pharmacokinetic properties listed in the document match those
reported in their cited sources (Tier 1).
• Carignan and Zoeller refrained from commenting, explaining that this question was not
within their areas of expertise.
6ai. Please comment specifically on whether the methods used to derive toxicity values for PFDA
appropriately account for uncertainties in evaluating the pharmacokinetic differences between the
experimental animal data and humans.
• Fisher noted that the lack of fit to pharmacokinetic data for repeated exposures suggests that acute
dosing CL values do not predict repeat exposures; this will need to be further studied, perhaps using
computational toxicology methods (Tier 3).
• Georgopoulos noted that the Bernstein et al. (2021) methods of reparametrizing the Kim et al.
(2019) PBPK model should be included in the document (Tier 1).
• Haney commented that there may not be an acceptable way for EPA to further quantitatively
account for uncertainty surrounding interspecies differences in pharmacokinetics based on currently
available data and methods.
13
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
6b. For immune effects, a UFS of 1 and 3 were considered to account for extrapolation from less
than lifetime human data; ultimately a UFS of 1 was selected. Does the provided scientific rationale
support this decision ?
• Carignan, Faustman, Georgopoulos, Haney, and Hoberman agreed that the scientific
rationale supports the decision on UFs.
• Zoeller opined that the justification of a UF of 1 for sensitive life stage was not clear (Tier
2).
6c. For liver effects, a value of 3 is applied to extrapolate between effects in laboratory animals and
in humans during the derivation of the subchronic RfD. Please comment on whether the available
animal and mechanistic studies support this conclusion and whether the analysis presented in the
Toxicoloaical Review is clearly documented.
• Adgate, Carignan, Fisher, Georgopoulos, Haney, Hoberman, and Zoeller agreed that there
was adequate support for an uncertainty value of 3 to extrapolate between the effects in
laboratory animals and humans during the derivation of the subchronic RfD.
• Faustman did not support the use of a value of 3. She commented that an uncertainty
factor of 10 would be more appropriate to extrapolate the available data for chronic
considerations due to the lack of underlying data. Faustman did support a UFA greater
than 1 to account for the complexities of PPARa and non-PPARa receptor involvement
(Tier 1).
6d. For liver, male reproductive, and female reproductive effects, a default value of 10 is applied for the UFs
when extrapolating from 28-day animal data to a subchronic exposure. Does the provided scientific
rationale support this decision? Please explain.
• Adgate, Carignan, Faustman, Georgopoulos, Hoberman, and Zoeller agreed that the scientific
rationale provides clear support for this decision.
• Fisher suggested considering the use of an animal/human PBPK model to gain insights into
subchronic and chronic exposures and accumulation in the body over time (Tier 2).
• Haney noted that EPA should consider whether data from structurally related PFAS could help to
further inform the UFs values. Hoberman seconded this (Tier 2).
6e. Are the provided rationales for the remaining uncertainty factors (UFj, UFp, UFH) scientifically justified
and clearly described (to inform the UFH, the assessment evaluates and considers the available evidence on
potential susceptibility to PFDA within different populations or lifestaaes, including any potential impacts
from early life exposure to PFDA on children's health or health later in life, although few studies on
susceptibility were available)? If not, pleaseexplain.
• Adgate, Carignan, Faustman, Georgopoulos, Haney, Hoberman, and Zoeller agreed that the
rationales for the remaining uncertainty factors were scientifically justified and clearly described.
14
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Fisher noted that EPA could, using language accessible to audiences without a background in
epidemiology, explain the epidemiologic tools used to tease out the health effects of PFDA when
PFDA is part of a mixture of exposures (Tier 2).
Carcinogenicity Hazard Identification and Toxicity Value Derivation
7. The Toxicoloaical Review concludes there is inadequate information to assess carcinogenic
potential for PFDA and that this descriptor applies to oral and inhalation routes of human exposure.
Please comment on whether the available human, animal and mechanistic studies, and the analysis
presented in the Toxicoloaical Review are scientifically justified and clearly described.
• Adgate, Carignan, Faustman, Georgopoulos, Haney, Hoberman, Leung, and Zoeller agreed
that there currently is inadequate information to assess the carcinogenetic potential of
PFDA.
• However, Faustman also commented on observations of chromosomal abnormalities and
genotoxicity; these endpoints should not be ignored. Faustman encouraged EPA to clearly record
where DNA impacts are reviewed in the document and note where these topics were discussed in
multiple sections (e.g., if a study outcome was discussed in the male reproductive section as well
as the carcinogenicity hazard identification section). Leung agreed with Faustman on these points
(Tier 1).
• Fisher refrained from commenting.
8. Given the conclusion there was inadequate information to assess carcinogenic potential for PFDA, the
Toxicological Review does not derive guantitative estimates for cancer effects for oral or inhalation
exposures. Is this decision scientifically justified and clearly described?
• All nine reviewers agreed that the decision not to derive quantitative estimates for cancer effects
for oral or inhalation exposures was scientifically justified and clearly described.
3.0 REVIEWER RESPONSE TO CHARGE QUESTIONS
3.1 The Toxlcologlcal Review for PFDA describes and applies a systematic review protocol for
identifying and screening pertinent studies. The protocol is described in brief detail in
Section 1,2,1 (Literature Searching and Screening) and in full detail in Appendix A
(Systematic Review Protocol for the PFAS IRIS Assessments). Please:
a. Comment on whether the literature search strategy and screening criteria for PFDA are
appropriate and clearly described.
b. Identify additional peer-reviewed studies of PFDA that EPA should consider
incorporating prior to finalizing the assessment.
EPA synthesized the literature published through April 2022 in the external review draft
and has been monitoring newly identified studies (i.e., studies identified by EPA or the
public that meet the PECO criteria or otherwise inform key assessment conclusions, but
which were not addressed in the external review draft, for example due to publication
after April 2022). EPA will characterize these studies in a document that will be provided
to the peer review panel and the public and, following the review, included as an Appendix
15
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
to the assessment prior to finallzatlon. The characterization will focus on EPA's Judgment
of whether the studies would have a material impact on the conclusions (i.e., identified
hazards or toxicity values) in the external review draft. Following receipt of this additional
document after the review is underway, please:
c. Review EPA's characterization and provide tiered recommendations regarding which
studies, If any, would have a material impact on the draft's conclusions and should be
incorporated into the assessment before finalizing, as well as your interpretation of the
impact of those studies to be incorporated.
Literature Search Methods and Documentation
Reviewer
Comments
Adgate
a. Tier 3: In general, the literature search strategy and screening criteria used are clearly
described and the process is well documented and therefore transparent due to a
comprehensive protocol. Links to Appendix A in the "Supplemental Information"
document did not work, and it is confusing that there's are two Addenda to the main
document (i.e., "Tox Profile") which the Agency should rename as that is easily
confused with Appendix.
Some of the systematic review methods can be clarified, particularly the studies that
were put in the potentially useful for future consideration category. An improved
version of the document will clarify the process for deciding what's important and the
goals of future analyses and describing the main uncertainties that these.
Per Dr Georgopoulos's comment, what would be useful in the document is a better
summary of the exposure information, particularly as relevant to the epidemiology
studies and animal studies, i.e., some summary of NHANES serum levels to put the
IRIS analysis in context.
c. No additional comment as the rationale provided by EPA is reasonable but
enumerating the additional studies and how they fit into the larger picture, even if
not strongly supporting the assessment (example below), should be done as feasible.
Bailey JM, Wang L, McDonald JM, Gray JS, PetrieJG, Martin ET, Savitz DA, Karrer TA,
Fisher KA, Geiger MJ, Wasilevich EA. Immune response to COVID-19 vaccination in a
population with a history of elevated exposure to per- and polyfluoroalkyl substances
(PFAS) through drinking water. J Expo Sci Environ Epidemiol. 2023 Jun 19. doi:
10.1038/s41370-023-00564-8. Epub ahead of print. PMID: 37337047.
Carignan
a. The literature search strategy and screening criteria for PFDA appear appropriate
and clearly described with the exception of the Tier 1 comments noted below.
Section 1.2, Page 28, Line 25: "are provided in Appendix A." Recommend updating
to 'Appendix A and B,' checking that the link for Appendix A works and that the
dates in Appendix B are current. [Tier 1]
Section 1.1.4, Page 24 Line 36: Correction needed, first says 2002-2001 then 2000-
2001.
16
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Section 1.2, Page 28, Line 28: Literature appears to be misspelled. [Tier 2]
Section 1.2, Page 29, Line 18: Check that 'SEMs' is previously defined. [Tier 2]
Section 1.2, Page 29, Table 1-4:
In the Evidence column for Populations, Human: Recommend adding text
similar to for Animal, "(including preconception, in utero, lactation,
peripubertal, and adult stages)" or otherwise assure inclusion of sensitive
developmental windows including those relevant to mammary gland
development (e.g., duration of lactation as an outcome), fertility, infant
exposure, etc. [Tier 1]
Should consider alteration in mammary gland function (including lactation
duration) as the mammary gland is a sensitive endpoint for other PFAS (e.g.,
PFOA). Mammary gland development has at least two sensitive exposure
windows: early life and pregnancy/postnatal.
In the Evidence column for 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 progression of events contributing towards
toxic effects will be tracked as potential supplemental material [e.g., for
evaluating key science issues; Section 2.4 of the protocol])." Recommend
clarifying here or elsewhere how vaccine immune titer is a functional measure
of adaptive immune response and therefore a phenotypic outcome. [Tier 1]
Sections 1.2.2. (Evaluation of Individual Studies) and 1.2.3. (Additional
Epidemiology Considerations) are clear and highlight important points regarding
bias and confounding.
Section 1.2.4., Page 33, Line 34: 'all health effects for animal studies and some
health effects for epidemiologic studies.' Since details are in Appendix A, suggest
highlighting here the health effects included or excluded for epidemiologic studies.
[Tier 1]
Figure 2-1: More explanation needed for the 595 studies excluded as 'not PICO
relevant'. [Tier 1]
The exposure assessment sections (Sections 1.1.3: Environmental Fate and
Transport and 1.1.4: Potential for Human Exposure, including Populations and
Lifestages with Potentially Greater Exposure) should be expanded and updated to
provide important and relevant context. Specific recommendations are provided:
Section 1.1.3, Page 24 Line 11: "Yoo et al. (2011) estimated a grass-soil
accumulation factor (grass concentration divided by soil concentration) of 0.10 for
PFDA, based on samples collected from a site with bio-solids-amended soil." Should
update with results from Blaine et al. 2013, which calculated a BAF of 0.52 for
lettuce grown in industrially impacted soil. [Tier 1]
Andrea C. Blaine, Courtney D. Rich, Lakhwinder S. Hundal, Christopher Lau,
Marc A. Mills, Kimberly M. Harris, and Christopher P. Higgins. Uptake of
17
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Perfluoroalkyl Acids into Edible Crops via Land Applied Biosolids: Field and
Greenhouse Studies. Environ. Sci. Technol. 2013, 47, 14062-14069
Section 1.1.4, Page 24 Line 25: "Gebbink et al. (2015) modelled exposure to PFDA
among the adult general population. 'Intermediate' exposure (i.e., based on median
inputs for all exposure parameters) from direct and indirect (i.e., precursor) sources
was estimated to be 67 pg/kg-day. Of the pathways evaluated (i.e., ingestion of
dust, food, water; inhalation of air), direct intake of PFDA in the diet accounted for
the largest portion of exposure for the intermediate scenario." Should note that this
estimate is for the general population. [Tier 1]
Here are two additional references that also support this statement.
Haug LS, Huber S, Becher G, Thomsen C. 2011. Characterization of human
exposure pathways to perfluorinated compounds-comparing exposure
estimates with biomarkers of exposure. Environ Int 37:687-693.
Tittlemier SA, Pepper K, Seymour C, Moisey J, Bronson R, Cao XL, et al.
2007. Dietary exposure of Canadians to perfluorinated carboxylates and
perfluorooctane sulfonate via consumption of meat, fish, fast foods, and food
items prepared in their packaging. J Agric Food Chem 55:3203-3210.
Section 1.1.4, Page 25 Line 2: "For example, Lankova et al. (2013) detected PFDA in
10% of human milk samples collected from 50 Czech women at concentrations
ranging from <6 to 12 pg/mL indicating that breastmilk is a potential route of
exposure for infants." More recently, Zheng et al (2021) detected PFDA in 94% of
breast milk samples collected in 2019 from 50 U.S. mothers at median
concentration of 7.4 and maximum concentration of 697 pg/mL. A meta analysis by
LaKind et al. (2023) found that breast milk concentrations of some PFAS often
exceed children's drinking water screening values, regardless of geographic
location. A time trend analysis of breast milk samples collected from women in
Sweden identified an increasing trend for a subset of women in Stockholm (n= 50)
over the past 17 years. [Tier 1]
Guomao Zheng, Erika Schreder, Jennifer C. Dempsey, Nancy Uding, Valerie Chu,
Gabriel Andres, Sheela Sathyanarayana, and Amina Salamova. Per- and
Polyfluoroalkyl Substances (PFAS) in Breast Milk: Concerning Trends for Current-
Use PFAS. Environmental Science & Technology 2021 55 (11), 7510-7520. DOI:
10.1021/acs.est.0c06978
LaKind JS, Naiman J, Verner MA, Leveque L, Fenton S. Per- and polyfluoroalkyl
substances (PFAS) in breast milk and infant formula: A global issue. Environ Res.
2023 Feb 15;219:115042. doi: 10.1016/j.envres.2022.115042. Epub 2022 Dec 16.
PMID: 36529330; PMCID: PMC9872587.
Nyberg E , Awad R , Bignert A, Ek C, Sallsten G , Benskin JP . Inter-individual,
inter-city, and temporal trends of per- and polyfluoroalkyl substances in human
milk from Swedish mothers between 1972 and 2016. Environ Sci Process Impacts.
2018 Aug 16;20(8): 1136-1147. doi: 10.1039/c8em00174j. PMID: 29987291.
Section 1.1.4, Page 25 Line 4: "Exposure can also occur through hand-to-mouth
18
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
transfer of materials containing these compounds (ATSDR, 2018b) or in infants
through ingestion of formula reconstituted with contaminated drinking water." Should
note limited data on formula itself. Lakind et al (2023) reviewed the limited data on
PFAS in formula and noted, "Regarding infant formula PFAS levels, there is a general
dearth of information. At present, in any geographic location, it is not possible to
state with any certainty that infant formula will have lower PFAS levels compared to
breast milk. The same is true for drinking water used to reconstitute infant formula..."
[Tier 1]
LaKind JS, Naiman J, Verner MA, Leveque L, Fenton S. Per- and polyfluoroalkyl
substances (PFAS) in breast milk and infant formula: A global issue. Environ Res.
2023 Feb 15;219:115042. doi: 10.1016/j.envres.2022.115042. Epub 2022 Dec 16.
PMID: 36529330; PMCID: PMC9872587.
Section 1.1.4, Page 25 Line 9: "...such as firefighters or individuals who install and treat
carpets." Should also note workers in the PFAS manufacturing and other PFAS using
industries (e.g., textile, leather, plastics, paper). [Tier 1]
Franziska Heydebreck, Jianhui Tang, Zhiyong Xie, and Ralf Ebinghaus.Emissions of
Per- and Polyfluoroalkyl Substances in a Textile Manufacturing Plant in China and
Their Relevance for Workers' Exposure. Environmental Science &
Technology 2016 50 (19), 10386-10396. DOI: 10.1021/acs.est.6b03213
Hakon A. Langberg, Hans Peter H. Arp, Gijs D. Breedveld, G0ril A. Slinde, Ase
H0iseter, Hege M. Gr0nning, Morten Jartun, Thomas Rundberget, Bj0rn M.
Jenssen, Sarah E. Hale. Paper product production identified as the main source of
per- and polyfluoroalkyl substances (PFAS) in a Norwegian lake: Source and
historic emission tracking. Environmental Pollution. 2021. Volume 273.
https://doi.Org/10.1016/i.envpol.2020.116259.
Section 1.1.4, Page 25 Line 11: "...professional ski waxing may be more highly exposed
because PFAS in dust may become airborne and inhaled during this process."
Recommend removing 'in dust' as ski waxing often involves heating and therefore
there is also the potential for inhalation of volatile PFASs. [Tier 2]
Section 1.1.4, Page 25 Line 14: "Populations living near fluorochemical facilities where
environmental contamination has occurred may also be more highly exposed (ATSDR,
2018a). Yamada et al. (2014) estimated exposure to PFDA and other PFAS among high
seafood consumers and high freshwater fish consumers in France." Recommend
describing important fate and transport mechanisms: direct discharge, WWTP
effluent to DW and fish, WWTP biosolids to agriculture. Dry deposition to crops and to
ground/drinking/irrigation water. [Tier 1]
It would be extremely useful to include a figure that shows exposure pathways for
PFAS. [Tier 1]
Section 1.1.4, Page 25, Table 1-2: Should note that PFDA blood levels were estimated
to increase 18% per month among breastfeeding infants. [Tier 1]
Mogensen UB, Grandjean P, Nielsen F, Weihe P, Budtz-J0rgensen E. Breastfeeding
as an Exposure Pathway for Perfluorinated Alkylates. Environ Sci Technol. 2015
19
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Sep 1;49( 17): 10466-73. doi: 10.1021/acs.est.5b02237. Epub 2015 Aug 20. PMID:
26291735; PMCID: PMC6190571.
Section 1.1.4, Page 26, Line 9: "...PFAS-treated carpets or other textiles." Recommend
editing to say 'and textiles' since carpets aren't really textiles. [Tier 2]
Section 1.1.4, Page 26: Paragraph starting at Line 8: A table would be helpful showing
location, year, detection limits, frequencies, percentiles and maximum for each study.
Comparisons and overall takeaway are difficult as currently presented. [Tier 2]
Section 1.1.4, Page 26, Line 24: Should also note results of water monitoring for the
ATSDR multi-site study and other locations with considerable point-source PFAS
contamination. [Tier 1]
Section 1.1.4, Page 27, Line 25: "in Decatur, AL where wastewater treatment sludge
had been applied." Suggested rephrase to provide useful context, "...where sludge
had been applied from a wastewater treatment plant receiving effluent from PFAS
manufacturing." [Tier 2]
Section 1.1.4, Page 27: Recommend adding a section on foods that notes pathways
(irrigation water, biosolids and dry deposition), highlights examples, and notes
challenges as well as recent advancements including analytical chemistry methods.
[Tier 1]
Section 1.1.4, Page 27, Sections on AFFF and Military:
Should note impacts on surrounding community drinking water systems,
recreational water bodies, fish, wildlife and agriculture. [Tier 2]
Should note PFAS accumulation on the surface microlayer of waterbodies and in
surface water foam. [Tier 1]
Schwichtenberg T, Bogdan D, Carignan CC, Reardon P, Rewerts J, WanzekT,
Field JA. PFAS and Dissolved Organic Carbon Enrichment in Surface Water
Foams on a Northern U.S. Freshwater Lake. Environ Sci Technol. 2020 Nov
17;54(22): 14455-14464. doi: 10.1021/acs.est.0c05697. Epub 2020 Nov 8.
PMID: 33164508.
Section 1.1.4, Page 27, Table 1-3: Should update to include more recent data on PFDA
in environmental media at PFAS impacted military installations. [Tier 1]
Section 1.1.4, Page 28, Line 9: Should specify whether this is a mean or median
concentration of 0.014 ng/g. [Tier 2]
Section 1.1.4, Page 28, Line 19: Should also note findings from the recent review of
EPA fish data. [Tier 1]
Barbo N, StoiberT, Naidenko O, Andrews D. Locally caught freshwater fish
across the United States are likely a significant source of exposure to PFOS and
other perfluorinated compounds. 2023. Environmental Research. 220.
https://doi.Org/10.1016/i.envres.2022.115165
b. A recent high quality study investigating birth outcomes (e.g., birth weight) should
20
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
be included:
Padula et al. Birth outcomes in relation to prenatal exposure to per- and
polyfluoroalkyl substances and stress in the environmental influences on child
health outcomes (ECHO) program. March 2023. Environmental Health
Perspectives. 131(3). https://ehp.niehs.nih.gov/doi/10.1289/EHP10723
Studies of lactation duration (an indicator of adverse mammary gland
development) should be included.
Timmermann CAG, et al. Shorter duration of breastfeeding at elevated
exposures to perfluoroalkyl substances. 2017. Reprod Toxicol 68:164-170.
doi:10.1016/j.reprotox.2016.07.010.
Romano ME et al. Maternal serum perfluoroalkyl substances during
pregnancy and duration of breastfeeding. 2016. Environ Res. 149:239-246.
DOI: 10.1016/j.envres. 2016.04.034
For additional context: https://ehp.niehs.nih.gov/doi/full/10.1289/ehp.125-A17
c. This reviewer did not identify any studies that would have a material impact on the
draft's conclusions.
Faustman
a. This reviewer found the literature search strategy and screening criteria for PFDA as
appropriate and clearly described and in alignment with IRIS methodologies. The
methodology used was also aligned with the methods used for many if not all of the
PFAS related compounds and that provides a good comparison across these IRIS
reviews.
Figure 2-1 presents an important summary of this process including listing of the
source for the searches ( PubMed, Toxline, etc.) and the disposition of all the
identified items. This figure closely follows the text which explains how many were
unique (not duplicates) and how many were excluded at each stage. Of those texts
included for full screening rationale for 443 texts were given. Thirty five were
excluded as not relevant to PECO , were reviews, were abstracts only and or
without full text. This information was important in order to track the process as
was the breakdown of the 262 texts that met the PECO criteria as well as 374 which
were tagged as supplemental.
For this reviewer, the first exclusion of 595 texts as not relevant for PECO remains
a mystery and this reviewer would like to see a summary breakdown for the
rationale for exclusion similar to that provided later in the full text screening
materials. Again, this rationale could be at a very high level but it provides at least
two things to the reviewer, how effective the search and search terms are at
identifying relevant literature but also provides an insight as to what types of
literature, endpoints, and information is in general in this larger information
population and can sometimes provide hints at what new areas are forthcoming.
This request is a Tier 2 Suggested Revision.
b. This reviewer did not separately identify critical missing literature specific for PFDA
for this review. However, 1 will be attaching several recent papers that are reviews
21
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
on the impact of PFAS related compounds for male reproductive toxicity and for
cancer (including testicular cancer).
Clarification: Please see some of the references that are cited in these reviews regarding
sperm effects. The reviewer recognizes that these references include review type articles
but also references to other PFAS male reproductive effects that should at least be
discussed to provide context and to be responsive to public comments.
• Front Endocrinol (Lausanne). 2023 Feb 20;14:1114463. doi:
10.3389/fendo.2023.1114463. eCollection 2023. Toxic effects of per- and
polyfluoroalkyl substances on sperm: Epidemiological and experimental
evidence; Zhangbei Sun, Yiqian Wen, Binhui Wang, Shiyi Deng, Fan Zhang,
Zhendong Fu, Yangyang Yuan, Dalei Zhang
Abstract
As emerging organic contaminants, per- and polyfluoroalkyl substances (PFASs) have
aroused worldwide concern due to their environmental persistence, ubiquitous
presence, bioaccumulation, and potential toxicity. It has been demonstrated that
PFASs can accumulate in human body and cause multiple adverse health outcomes.
Notably, PFASs have been detected in the semen of human, posing a potential hazard
to male fecundity. This article reviews the evidence about the toxic effects of
exposure to PFASs on male reproduction, focusing on the sperm quality.
Epidemiological studies showed that PFASs, such as perfluorooctanoic acid (PFOA)
and perfluorooctane sulfonic acid (PFOS), were adversely associated with the semen
parameters in humans, including sperm count, morphology and motility. Experimental
results also confirmed that PFAS exposure led to testicular and epididymal damage,
therefore impairing spermatogenesis and sperm quality. The mechanisms of
reproductive toxicity of PFASs may be involved in blood-testosterone barrier
destruction, testicular apoptosis, testosterone synthesis disorder, and membrane lipid
composition alteration, oxidative stress and Ca2+ influx in sperm. In conclusion, this
review highlighted the potential threat of exposure to PFASs to human spermatozoa.
Keywords: male fecundity; per-and polyfluoroalkyl substances; reproductive toxicity;
sperm; testosterone.
• Int J Environ Res Public Health; 2021 Apr 5;18(7):3794. doi:
10.3390/ijerphl8073794. Perfluoroalkyl Chemicals and Male Reproductive
Health: Do PFOA and PFOS Increase Risk for Male Infertility? Pheruza Tarapore,
Bin Ouyang
Abstract
Poly- and perfluoroalkyl substances (PFAS) are manmade synthetic chemicals which
have been in existence for over 70 years. Though they are currently being phased out,
their persistence in the environment is widespread. There is increasing evidence
linking PFAS exposure to health effects, an issue of concern since PFAS such as
perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) bioaccumulate in
humans, with a half-life of years. Many epidemiological studies suggest that,
worldwide, semen quality has decreased over the past several decades. One of the
22
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
most worrying effects of PFOS and PFOA is their associations with lower testosterone
levels, similar to clinical observations in infertile men. This review thus focuses on
PFOS/PFOA-associated effects on male reproductive health. The sources of PFAS in
drinking water are listed. The current epidemiological studies linking increased
exposure to PFAS with lowered testosterone and semen quality, and evidence from
rodent studies supporting their function as endocrine disruptors on the reproductive
system, exhibiting non-monotonic dose responses, are noted. Finally, their
mechanisms of action and possible toxic effects on the Leydig, Sertoli, and germ cells
are discussed. Future research efforts must consider utilizing better human model
systems for exposure, using more accurate PFAS exposure susceptibility windows, and
improvements in statistical modeling of data to account for the endocrine disruptor
properties of PFAS.
Reviews that could be used to support consistence of findings across the PFAS
compounds and also PFDA but not directly studies. Can address the uncertainty in
observations only with PFDA.
• Front. Endocrinol., 09 March 2022; Sec. Reproduction, Volume 12 - 2021
https://doi.org/10.3389/fendo.2021.799043. Assessment of the Emerging
Threat Posed by Perfluoroalkyl and Polyfluoroalkyl Substances to Male
Reproduction in Humans. Leah Calvert, Mark P. Green, Geoffry N. De luliis,
Matthew D. Dun- Brett D. Turner. Bradley O. Clarke. Andrew L. Eamens, Shaun D.
Roman and Brett Nixon
Per-fluoroalkyl and polyfluoroalkyl substances (PFAS) are a diverse group of synthetic
fluorinated chemicals used widely in industry and consumer products. Due to their
extensive use and chemical stability, PFAS are ubiquitous environmental
contaminants and as such, form an emerging risk factor for male reproductive health.
The long half-lives of PFAS is of particular concern as the propensity to accumulate in
biological systems prolong the time taken for excretion, taking years in many cases.
Accordingly, there is mounting evidence supporting a negative association between
PFAS exposure and an array of human health conditions. However, inconsistencies
among epidemiological and experimental findings have hindered the ability to
definitively link negative reproductive outcomes to specific PFAS exposure. This
situation highlights the requirement for further investigation and the identification of
reliable biological models that can inform health risks, allowing sensitive assessment
of the spectrum of effects of PFAS exposure on humans. Here, we review the
literature on the biological effects of PFAS exposure, with a specific focus on male
reproduction, owing to its utility as a sentinel marker of general health. Indeed, male
infertility has increasingly been shown to serve as an early indicator of a range of co-
morbidities such as coronary, inflammatory, and metabolic diseases. It follows that
adverse associations have been established between PFAS exposure and the
incidence of testicular dysfunction, including pathologies such as testicular cancer and
a reduction in semen quality. We also give consideration to the mechanisms that
render the male reproductive tract vulnerable to PFAS mediated damage, and discuss
novel remediation strategies to mitigate the negative impact of PFAS contamination
and/or to ameliorate the PFAS load of exposed individuals.
23
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Papers addressing the carcinogenic potential of other PFAS related compounds
• IARC has classified PFOA as "possibly carcinogenic to humans" (Group 2B), based
on limited evidence in humans that it can cause testicular and kidney cancer,
and limited evidence that it can cause cancer in lab animals. Mar 21, 2023
• Perfluorooctanoic Acid (PFOA), Perfluorooctane Sulfonate ... American Cancer
Society https://www.cancer.org/
• PFAS Exposure and Risk of Cancer - NCI - National Cancer Institute (.gov).
https://dceg.cancer.gov/research/waht-we-study/pfas Perfluorooctanoic acid
(PFOA), the most well-studied PFAS, has been classified as a possible human
carcinogen based in part on limited epidemiologic ...See: Kidney Cancer,
Testicular Cancer, Prostate Cancer
References that emphasize differences in the appropriate measure to use for
evaluating Testicular effects from PFAS compounds.
• The Pan et al references do suggest that semen versus serum levels of PFAS
compounds are a better comparison to use for such pharmacokinetic and sperm
effect studies and may provide an explanation for variable comparisons
observed in the literature.
Environmental Pollution 266 (2020) 115330
Contents lists available at ScienceDirect
< ; S Environmental Pollution
ELSEVIER journal homepage: www.elsevier.com/locate/envpol
Exposure to per- and polyfluoroalkyl substances (PFASs) in serum -f)
versus semen and their association with male reproductive SSS
hormones'*
Qianqian Cui a, Yitao PanJinghua Wanga, Hongxiu Liu b, Bing Yao c, Jiayin Dai '¦'
a Key Laboratory of Animal Ecology and Conservation Biology. Institute of Zoology, Chinese Academy of Sciences, Beijing, 100101, China
b School of Public Health, Tongji Medical College, Huazhong University of Science and Technology. Wuhan, 430000. Hubei. China
c Reproductive Medical Center, NanjingJinling Hospital. Nanjing University, School of Medicine, Nanjing 210002, Jiangsu, China
REVIEWER comments on these reference-.;. Please note some of these citations are
reviews however because of the observation of testicular cancer from other PFAS
related comments the references that discuss sperm quality cited in the above
references suggest that more use of these references is needed as noted in my
comments on male reproductive endpoints and in the comments regarding cancer risks.
Inclusion would also help to address some of the public comments that were raised.
c. This reviewer appreciated the detailed information characterizing the updated
literature searches. Figure 1 was useful in looking at where the new studies mapped
on the HAWC visual. In general, this reviewer agreed with the agency regarding what
endpoints the additional studies might impact. The impacts of the additional data on
the supplemental evidence is of significance for the ADME considerations. For
example, Louisse et al 2022, where the determination of 0AT4 transporter
24
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
involvement was determined as potentially providing insight on the reduced human
urinary clearance of PFDA was important. Likewise, the paper by Yao et al 2023 which
estimated infant urinary CL of PFDA from matched blood and urine samples. The
literature review says that this helps to address the potential for lifestage CL
difference and this reviewer would say yes, it is a start however because the
commentary on EPA use of this reference does not provide a detailed review of this
paper and the number of matched samples available nor the range of levels of PFDA
in the samples analyzed it is difficult to fully interpret. Clarification on how this will be
incorporated in the review, as an addendum? Foot note on the relevant page of the
revised IRIS report, etc.? Tier 1 recommendation is to clarify overall how the recent
updated literature review will be incorporated into the current draft, Tier 2
recommendation deals with clarification of how those papers identified as YES, will
be incorporated into the specific relevant sections of the report?
Fisher
a. It appears that the search strategy was adequate and there are plans to update the
search. Listing any major research activities that are ongoing would be useful (Tier
2) because of the number of people contributing to the literature.
b. The updated EPA studies covered the few papers that 1 found relevant to PFDA.
Georgopoulos
a. The search strategy and screening criteria for PFDA literature related to
pharmacokinetics and health effects are clearly described in Section 1.2 (pp. 1-7 to
1-16) and Appendix B of the EPA Review document, as well as in the updated
Protocol available at
https://cfpub.epa.gov/ncea/iris drafts/recordisplav.cfm?deid=345065
The methods used for the PFDAToxicological Review were appropriate and
consistent with scientific standards and practices.
Note: It should be recognized that the process for identifying and screening
pertinent studies for the PFDAToxicological Review (as well as for the Toxicological
Reviews for other PFAS) is inherently challenging. The fact that multiple health
effects are associated (or are potentially associated) with exposures to individual
PFAS and PFAS mixtures has led to numerous studies worldwide, other completed
and many on-going, with multiple related publications appearing at an increasing
rate. This makes it necessary to develop a process of efficiently "aligning" the
information presented in peer-reviewed documents, such as the present
Toxicological Review, with the content of regularly updated online resources, such
as the Health & Environmental Research Online portal
(https://heronet.epa.gov/heronet/), the Health Assessment Workspace
Collaborative (HAWC) portal (https://hawc.epa.gov/assessment/100500073), or
the CompTox Chemicals Dashboard (https://comptox.epa.gov/dashboard) so as to
allow new information to appear on the online portals without creating
inconsistencies with statements included in the documents.
Even without accounting for inconsistencies with information appearing on the
online resources (e.g. Table 1-1 on page 1-1 and corresponding information
available on the CompTox Chemicals Dashboard), the current PFDA Toxicological
Review document should address inconsistencies in the information presented in
25
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Figure 2-1 "Literature search for perfluorodecanoic acid and related salts" (page 2-
2): in that figure it is stated that 1,057 records were identified (after duplicate
removal) for "Title & Abstract Screening" and 595 of those were excluded as "Not
relevant to PECO." That results in 462 (=1,057-595) records, while the "Full Text
Screening" box in the Figure lists 443 records, leaving 19 (=462-443) records
"missing". From the 443 "Full Text Screening" records 35 are excluded as "not
relevant to PECO," resulting in 408 (443-35) records: however, the "Studies
Meeting PECO" box lists only 262 records, leaving 146 (=408-262) records
"missing." Furthermore, the "Tagged as Supplemental" box states in its header
that the number of records identified is n=374, but if one adds the records listed in
individual categories within that box [mechanistic or MOA (n = 142), ADME 31),
exposure assessment or qualitative exposure only (n = 125), mixture-only (n = 6),
non-PECO route of exposure (n = 88), case report or case study (n = 0), other (n
=112)] gets a total of 504 records, creating a discrepancy of 130 (=504-374)
records.
Another point to be addressed here is that though the Technical Review states that
1,057 records were identified (after duplicate removal) for PFDA, the HERO portal
PFDA page
(https://heronet.epa.gov/heronet/index.cfm/proiect/page/proiect id/2614) lists
currently (June 2023) over 2,100 references.
On page 1-10 (lines 11-13) of PFDAToxicological Review it is stated that "Literature
inventories for studies meeting PECO criteria and studies tagged as "potentially
relevant supplemental material" during screening were created to facilitate
subsequent review of individual studies or sets of studies by topic-specific
experts." It would be very useful to identify and systematically organize and
screening Supplemental PFDA-related studies with respect to both the methods
used (in vitro, in silico, in vivo, etc.) and the bioactivity endpoints considered.
Although in silico, in vitro and non-mammalian model organism studies may not
have the same weight as human epidemiological and rodent laboratory and
studies, they can provide valuable mechanistic insights, both for hypothesis
development and for animal study evaluation, as well as corroborate hypotheses
derived from rodent and epidemiological studies when their results consistently
"point to the same direction". It would also be useful to identify and explicitly list
consistencies as well as inconsistencies appearing not only in the "PECO literature"
but also in the "supplemental literature". Examples are (a) the characterization of
PPAR binding affinities of PFDA (and of other PFAS) presented in the studies of
Ishibashi et al. (2019), Khazaee et al. (2021), Evans et al. (2023), Sun et al. (2023),
etc., and (b) the characterization of PFDA binding affinities with serum proteins
presented in Allendorf et al. (2019) and Allendorf (2021), Chen et al. (2020), etc.
Finally, it would be useful if exposure and risk assessment activities of regulatory
agencies, both in the US and internationally, collecting and/or organizing
information relevant to PFDA, were mentioned and cited in the Toxicological
Review: representative examples are
• the European Chemicals Agency document on PFDA (ECHA, 2016)
• The European Human Biomonitoring (HBM4EU) Platform
26
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
https://www.hbm4eu.eu/what-we-do/european-hbm-platform
• the Interstate Technology and Regulatory Council (ITRC) Per- and
Polyfluoroalkyl Substances Technical and Regulatory Guidance (ITRC, 2022)
• the CDC/ATSDR PFAS Multi-site Study (MSS)
(https://www.atsdr.cdc.gov/pfas/activities/studies/multi-site.html)
• the (13) PFAS studies listed at the HHEAR (Human Health Exposure
Analysis Resource) data Center (https://hheardatacenter.mssm.edu)
• etc.
b. A representative selection of peer-reviewed studies for potential consideration by
EPA is provided after my responses to the charge questions.
c. EPA evaluated a substantial number of additional studies, that were not included in
the draft of theToxicological Review initially provided to the panel, for incorporation
in the Toxicological Review and characterization of impact on specific conclusions
(i.e., identified hazards or toxicity values). According to the information provided by
the EPA, "186 studies were submitted by the State of New Jersey Department of
Environmental Protection and the Natural Resources Defense Council (NRDC)" and
of these 186 studies "119 studies had been previously identified and can be found
in the HERO database" while the "remaining 67 new studies were screened for PECO
criteria and evaluated for potential incorporation and impact on the assessment's
conclusions". EPA further clarified that "only studies that would notably impact the
primary EPA draft judgments (i.e., the health effects identified as human health
hazards and the final reference values) in the Step 4 draft will be added to the
Toxicological Review by EPA prior to finalization." Studies meeting assessment PECO
criteria were: 6 studies on immune effects, 10 studies on developmental effects, 6
studies on cancer, 6 studies on neurodevelopment, 5 studies on male reproductive
effects, 11 studies on female reproductive effects, 6 studies on urinary system
effects, 26 studies on cardiometabolic effects, 7 studies on endocrine effects, and
11 studies on other effects. Studies meeting select categories of supplemental
evidence were 7 ADME studies and 20 mechanistic studies, including non-PECO
routes of exposure. EPA indicated that it plans to incorporate only 4 of these
additional studies in the revision of the Technical Review: all 4 are supplemental
evidence studies: 3 are ADME studies (Louisse et al., 2022; Wang et al., 2018; Yao et
al., 2023) and 1 is a mechanistic study (Wang et al., 2022). EPA's rationale for
incorporating these 4 studies in the final Technical Review, is that, though they do
not affect draft judgments, they provide useful information and address important
data gaps. This rationale is reasonable and appropriate. EPA indicated that all other
additional studies will not be included in the final Review for reasons ranging from
presenting inconsistent, indeterminate, imprecise or unreliable results, to not
providing significant new data that would affect the draft assessments. Again, EPA's
rationale for this decision is in general reasonable and appropriate; however,
though this may not be a primary objective of the Toxicological Review, there is
value in identifying areas where studies and associated data are inconsistent or
inconclusive, potentially facilitating the development of new studies and application
27
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
of alternative methodologies that could address these inconsistencies.
Suggested Revisions and Future Considerations
• Tier 1 Necessary Revision: Update the subsection on Potential for Human
Exposure (subsection 1.1.4) with up-to-date references (as currently the most
recent reference cited in the subsection is from 2018) and include basic
summary information on distributions of population serum PFDA levels
(available from NHANES for the US) and on temporal trends for these
distributions.
• Tier 1 Necessary Revision: Check Table 1-1 for consistency with information
currently available on online (on portals such as EPA's CompTox Chemicals
Dashboard); evaluate the feasibility of expanding Table 1-1 by including values
of PFDA properties that are critical for is pharmacokinetics (e.g., binding
affinities to serum proteins).
• Tier 2 Suggested Revision: Include references to documents (and online
portals) relevant to PFDA risk characterization that have been developed by
US and international regulatory agencies; compile a summary (e.g., in the
form of a table or a brief Appendix) of established or proposed values for
metrics of reference doses/concentrations if such metrics are identified.
• Tier 3 Future Consideration: Develop a plan for the systematic and regular
updating of databases (such as HERO and HAWC) tracking information
relevant to the Toxicological Review(s); furthermore, specify criteria for new
information that would require re-evaluation and updating of the contents
and conclusions of the Toxicological Review(s).
Haney
a. The literature search strategy and screening criteria for PFDA appear appropriate
and clearly described. As indicated in Section 4.1 of Appendix A, 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. In February 2018, the literature search was updated and included
all PFAS nomenclature from the initial search as well as a broader non-date-limited
search of several new PFAS synonyms that were identified after the original search.
Full details of the July 2017 and February 2018 search strategies are presented in
Addendum B to Appendix A of the draft assessment. Additional relevant literature
not found through database searching was identified through other means (see p.
4-2 of Appendix A), and literature searches have been updated throughout the
assessment's development to identify newly published literature.
As indicated in Section 4.2 of Appendix A, PECO (populations, exposures,
comparators, and outcomes) criteria or predefined inclusion and exclusion criteria
(i.e., the latter were used for the initial search; see p. 4-5 of Appendix A for
exclusion criteria in addition to PECO) 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 and abstract level, these criteria were then used to
28
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
determine inclusion or exclusion of a reference based on the full text. In addition to
including studies that meet PECO criteria, other studies containing material
potentially relevant to the assessments objectives and specific aims were tracked
during the screening process as "potentially relevant supplemental material" for
potential incorporation into the assessment to address key science issues, etc. (see
pp. 4-5 and 4-6 of Appendix A). See Appendix A of the draft assessment for
additional information. Again, the literature search strategy and screening criteria
for PFDA appear overall to be appropriate and clearly described.
b. This reviewer personally knew of no additional peer-reviewed studies of PFDA that
the assessment should have incorporated. However, some public comments had
recommendations for the EPA's consideration (e.g., see the Literature Search and
Screening section of compiled public comments). If not already considered under
subsection "c" below, EPA should consider (Tier 1 necessary revision) whether or not
the results of the studies recommended by public commenters should be
incorporated into the assessment utilizing relevant practical criteria (e.g., based on
their potential impact (if any) on data gaps, assessment conclusions, and/or toxicity
values). However, it appears that EPA has already done so (see EPA characterization
of studies identified after public release of the draft IRIS Toxicological Review of
Perfluorodecanoic Acid (PFDA, CASRN 335-76-2) and Related Salts). Additionally, panel
member Dr. Carignan identified a newly published study, Padula et al. (2013).
c. In regard to studies published after release of the public comment draft assessment,
the EPA should apply reasonable and practical rationales regarding whether study
results should be included in the final assessment based on their potential impact (if
any) on data gaps, assessment conclusions, and/or toxicity values (Tier 1 necessary
revision). In fact, it appears that EPA has already done so (see EPA characterization of
studies identified after public release of the draft IRIS Toxicological Review of
Perfluorodecanoic Acid (PFDA, CASRN 335-76-2) and Related Salts). However, as
mentioned above, panel member Dr. Carignan identified a newly published study,
Padula et al. (2013).
Hoberman
The literature search strategy and screening criteria for PFDA was found to be
appropriate and clearly described. As noted in the review, all literature was tracked in
the U.S. EPA Health and Environmental Research Online (HERO). The database includes:
• PubMed (National Library of Medicine)
• Web of Science (Thomson Reuters)
• Toxline (National Library of Medicine)
• TSCATS (Toxic Substances Control Act Test Submissions)
In addition, relevant literature not found through database searching was identified by:
• Review of studies cited in any PECO-relevant studies and published journal
reviews; finalized or draft U.S. state, U.S. federal, and international
assessments (e.g., the draft 15 Agency for Toxic Substances and Disease
Registry [ATSDR] assessment released publicly in 16 2018).
29
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• In addition, studies included in ongoing IRIS PFAS assessments (PFHxA, PFHxS,
17 PFNA, PFDA) were also scanned for any studies that met PFBA PECO
criteria. Searches of published PFAS SEMs (Carlson et al., 2022; Pelch et al.,
2022) starting in 2021. Review of studies submitted to federal regulatory
agencies and brought to the attention of EPA. For example, studies submitted
to EPA by the manufacturers in support of requirements under the Toxic
Substances Control Act (TSCA). Identification of studies during screening for
other PFAS.
• For example, epidemiology studies relevant to PFDA were sometimes
identified by searches focused on one of the other four PFAS currently being
assessed by the Integrated Risk Information System (IRIS) Program. Other gray
literature (e.g., primary studies not indexed in typical databases, such as
technical reports from government agencies or scientific research groups;
unpublished laboratory studies conducted by industry; or working
reports/white papers from research groups or committees) brought to the
attention of EPA.
This reviewer could find no additional peer-reviewed studies of PFDA that EPA
should consider incorporating prior to finalizing the assessment.
• Tier 1: Necessary Revisions -None - An additional literature update has been
conducted and was provided as supplemental information.
• Tier 2: Suggested Revisions - With the limited number of studies on PFDA
some mention of the location of summary charts with all PFAS effects
showing a comparison across all PFAS would be useful.
• Tier 3: Future Considerations - Continual updating of database as new
relevant articles are published. The need for an updated literature search
based on the time between the searches used for the Toxicological Review of
PFDA, CASRN 335- 76-2.
Leung
a. The report is clear in describing the PECO literature search and screening process for
papers related to PFDA exposure. The initial search was conducted in 2017 and has
been updated yearly since; the last full update was conducted in June 2023. The
output was screened by two independent reviewers and the entire process was
captured in DistillerSR.
Although the PECO search was rigorous, includes both the appropriate library
reference data sources and other complementary mechanisms, with search results
tracked in the HERO database, 1 appreciate the public comments noting that some
human and animal studies as listed in the PFAS-Tox Database were not included,
which appears mostly due to the more broad PECO criteria in the latter. 1 agree with
the comments made in the reviewer panel public meeting to add the reasons for why
any studies may have been excluded to Figure 2-1 (Tier 1 Recommendation). It is also
noted that select studies of interest, but which did not fulfill PECO criteria, were
inventoried as potentially relevant supplemental material; the listed reasons for
assigning them as such are appropriate.
30
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
b. The search process appears appropriate. 1 do not have any additional studies to
recommend.
c. 1 agree with the EPA's disposition of these additional new studies and of studies
received from the open public comment period; none of the studies would
substantially change the current drafts' judgement for each of the health effect
categories. As noted, some of the studies regarded as supplemental evidence,
particularly for the ADME of PFDA, are worth noting for their possible mechanistic
data, but would not change the draft's assessment.
Zoeller
a. The literature search strategy and screening criteria for PFDA is described in section
1.2. The Agency strategy appears appropriate and clearly described. This search
strategy ended in June of 2022. The Agency has updated this search and identified a
number of new publications related to the PECO criteria for inclusion. However, the
Agency has not deemed that these papers will substantially alter the risk analysis.
b. The Agency's update and analysis appears well-justified and thorough.
c. No Recommendation.
3,2, For each health effect considered in the assessment and outlined below, please comment
on whether the available data have been clearly and appropriately synthesized to describe
the strengths and limitations, including whether the presentation and analysis of study
results are clear, appropriate, and effective to allow for scientifically supported syntheses
of the findings across sets of studies. Please comment on whether the study confidence
conclusions for the PFDA studies are scientifically justified, giving appropriate
consideration to important methodological features of the assessed outcomes2. Please
specify any study confidence conclusions that are not justified and explain any alternative
study evaluation decisions. For each, please also comment on whether the weight-of-
evidence decisions for hazard identification have been clearly described and scientifically
justified. Note that the data from studies considered informative to the assessment are
synthesized in the relevant health effect-specific sections and available in the Health
Assessment Workspace Collaborative (HAWC).
a. For liver effects, the Toxicological Review concludes that the available evidence
indicates PFDA exposure is likely to cause liver effects in humans given sufficient
exposure conditions, on the basis of a series of short-term studies in rats and mice
demonstrating consistent and coherent effects with a clear biological gradient. The
liver findings for PFDA were similar to those for other structuraIly-re 1 ated long-chain
PFAS and determined to be adverse and relevant to humans.
L Additional considerations influenced the liver effects hazard identification
decisions. Appendix A (Systematic Review Protocol for the PFAS IRIS Assessments)
outlines the human relevance of hepatic effects in animals that involve PPAHa
receptors as a key science issue. To the extent supported by the PFDA literature
(and to a lesser extent, literature for other PFAS), the Toxicological Review
2 The Toxicological Review provides an overview of individual study evaluations within each evidence synthesis section, and the
results of those outcome-specific evaluations are made available in the Health Assessment Workplace Collaborative linked here
HAWC. Note that a "HAWC FAQ for assessment readers" document, linked here (scroll to the bottom of the page, and the document
is available for download under "attachments"), is intended to help the reviewer navigate this on-line resource.
31
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
evaluates the evidence relevant to the potential involvement of PPARct and non-
PPARa pathways with respect to the reported liver effects. The Toxicological
Review ultimately concludes evidence from in vivo and in vitro studies support a
potential role for multiple pathways operant in the induction of hepatic effects from
PFDA exposure, although how those pathways interact within a mode of action
(MOA) cannot be specifically determined.
b. For immune effects, the Toxicological Review concludes that the available evidence
indicates PFDA exposure is likely to cause immunosuppression in humans given
sufficient exposure conditions, primarily on the basis of consistent evidence of reduced
antibody responses from two epidemiological studies in children and one study in
adults. Although some evidence for coherent immunomodulatory responses consistent
with immunosuppression was identified in short-term animal studies, the animal
evidence overall is uncertain. The Toxicological Review concludes the immune effects
are considered relevant to humans as the judgment is based on studies in humans.
i. For nearly all epidemiology studies of PFDA, there is potential that exposure to
other highly correlated PFAS could contribute to the observed effects. The evidence
synthesis for potential PFDA-induced immune effects included evaluation of the
potential for confounding across PFAS as well as other sources of confounding and,
based on the available data, determined that residual confounding could explain
part of the observed effect, but concern was minimal, and it was unlikely to fully
explain the associations seen in the literature.
c. For developmental effects, the Toxicological Review concludes that the available
evidence indicates PFDA exposure is likely to cause developmental effects in humans
given sufficient exposure conditions, based primarily on consistent findings of dose-
dependent decreases in fetal weight in mice gestationally exposed to PFDA supported
by some coherent evidence of decreased birth weight from studies of exposed humans
in which PFDA was measured during pregnancy, although uncertainties in the available
epidemiological evidence reduced the impact of these latter findings. The Toxicological
Review concludes the developmental effects in mice are considered relevant to humans
given similar findings of fetal growth restriction in mice and humans.
L As described in question 3,c and footnote to 3,c, the evidence synthesis for potential
PFDA-induced developmental effects considered potential confounding factors and
concluded that confounding across PFAS or from other potential sources of bias
(e.g., pregnancy hemodynamics in studies where PFDA was measured during or
after pregnancy) introduce significant uncertainty. These sources of uncertainty
ultimately reduce the strength of the available human evidence to slight for an
evidence base that might otherwise be interpreted as moderate.
d. For male reproductive effects, the Toxicological Review concludes that the available
evidence indicates PFDA exposure is likely to cause male reproductive effects in
humans given sufficient exposure conditions, based on coherent evidence in adult male
rats exposed to PFDA for 28 days. Although no direct information on the human
relevance of the animal evidence is available, the findings in animals are presumed to
be relevant based on the conserved role of androgen-dependent pathways in male
productive functions across species.
e. For female reproductive effects, the Toxicological Review concludes that the available
evidence indicates PFDA exposure is likely to cause female reproductive effects in
humans given sufficient exposure conditions, based primarily on coherent evidence
32
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
from a 28-day study in adult female rats. Although human studies are available
examining associations between PFDA and female reproductive toxicity (e.g.,
fecundity), the results were mostly null, possibly due to their low sensitivity for
observing effects. The Toxicological Review concludes the female reproductive effects
are considered relevant to humans given that mechanisms of female reproduction are
similar between rats and humans.
f. For cardiometabolic effects, the Toxicological Review concludes that the available
evidence suggests but is not sufficient to infer that PFDA exposure may have the
potential to cause cardiometabolic effects in humans given sufficient exposure
conditions, based on associations between PFDA and serum lipids, adiposity,
cardiovascular disease, and atherosclerosis in a few epidemiological studies. However,
the evidence is largely inconsistent across studies, which adds considerable
uncertainty. Evidence in experimental animals was indeterminate.
g. For neurodevelonmental effects, the Toxicological Review concludes that the available
evidence suggests but is not sufficient to infer that PFDA exposure may have the
potential to cause neurobehavioral effects in humans given sufficient exposure
conditions, based on associations between PFDA and outcomes related to attention and
behavior in epidemiological studies. However, the evidence is largely inconsistent
across studies, which adds considerable uncertainty. No evidence was found in
experimental animals to inform this outcome (indeterminate).
h. For mdjoerroe jilpa^ (i.e., hematological, respiratory,
digestive, dermal, musculoskeletal, and nervous systems), the Toxicological Review
concludes there is inadequate evidence to determine whether PFDA exposure has the
potential to cause these effects in humans on the basis of the sparsity of available
evidence.
Noncancer Hazard Identification
Reviewer
Comments
Adgate
a. Liver effects: Tier 2: Page 3.-49, line 4: meets the "criteria for adversity" is an odd
way of saying the evidence supports the finding of "adverse effects in the liver" in
animals.
Tier 2: Page 3-50 and surrounding text/evidence Integration. The overall
determination that "slight evidence" of liver effects in humans (exposure duration
not mentioned) is inconsistent given the positive effects observed 4 of 5 studies for
PFDA and ALT. The finding of an association is largely consistent with the extensive
literature on other longer chain legacy PFAS and liver effects. The subsequent
sentence saying that the results lack coherence and mentions other (unnamed)
biomarkers in humans is therefore unclear. Overall, this section would be clearer if
it clearly separated the determinations from the human and the animal data, then
discussed integration and consistency (or lack thereof) between the two data types.
A key point on this issue would be to clearly state the relevance of a 28-day animal
study for the overall determination and its relevance in humans given the time
needed to reach steady-state.
33
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Authors should also note that given the extensive evidence from related PFAS I
would say the determination lies somewhere in between slight and moderate. I
agree that the mechanistic evidence supports the observed effects, which increases
confidence in the conclusions and thus the writing on weight of evidence
determination (WOE) needs to be clearer on what a scientifically justified finding is
for this endpoint.
b. Immune Effects: Tier 2: Table 3-12. Address header issues as noted in general
comments above.
Clarification: Table 3-12. Look at the document. If any table breaks across pages
(and there are several) you can easily search "Table" and find them) the headers
need to be there.
Tier 1: Page 3-90 and surrounding text. The overall determination that evidence
indicates that the association between immunosuppression and PFDA exposure is
known with moderate confidence is a reasonable synthesis, though this needs to be
more clearly linked to several caveats. While I concur with the finding of increased
likelihood of immunosuppression in humans, the clarity can be improved as
sensitivity to detect effects and uncertainties inherent in these studies should be
more clearly acknowledged and integrated into the write up. Noting that this is
consistent with other PFAS is also an important point that cannot be understated.
What is also missing from this write up is a clear statement on why the findings
from the Faroe Islands cohorts, which have several unique characteristics and
dietary habits, are generalizable to the United States. While the evidence
integration and related tables address the potential for confounding and note that
effects are significant despite relatively low sensitivity to detect them (thus
increasing confidence that the findings are real), a clearer characterization of the
effect of multiple correlated PFAS on the immunosuppression attributed to PFDA is
needed. There are several highly correlated PFAS (and other persistent
contaminants) in the serum of most any environmental epidemiology study,
including those in the Faroe Islands, so confounding is a legitimate concern, but
there are limits to both the data and adjustment process that are well known. The
pros (and cons) of multiple confounder adjustment/potential for overfitting are all
issues that should be explicitly addressed in the write up as part of the summary
judgment. A more explicit discussion of the strengths, for example, of looking at
titers of multiple vaccine endpoints in more than one cohort are needed as well as a
discussion of how relatively small changes in titer levels at the measurement
timepoints are justifiable as key endpoints. The justification for using the response
level is not completely clear, and should be compared to other options, e.g., a 10%
BMR, so the authors can discuss and illustrate the significance of this choice.
I agree that the short-term animal studies and mechanistic evidence are coherent
and generally support the findings, and that this increases confidence in the
conclusions for this section.
c. Developmental effects: Tier 2: The characterization of an association between PFDA
and a wide range of developmental effects as slight in human and moderate in
animals is an reasonable synthesis of a large and varied dataset. The evidence
34
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
integration section lacks coherence as its sequencing of topics and determinations
could be clearer: it is easier to understand the findings by consulting Table 3-24.
Table 3-24 is a reasonable synthesis of the major strengths and limitations and the
determinations about WOE and are, for the most part, clear and scientifically
justified.
Tier 1: Page 156, line 24: The sentence starting with "Although" discussing potential
bias needs to be rewritten for clarity as it is unclear what it means.
Tier 2: Page 156, ~line 30-35: The text starting at line 30 seems like a more
appropriate place to start the evidence integration section as it identifies the main
endpoints and overall confidence.
d. Male reproductive effects: Tier 3: The finding of indeterminacy for the human data
on male reproductive effects is appropriate given the limitations of the few
available studies and short duration of exposure. The finding that PFDA likely causes
reproductive effects in male humans is a reasonable synthesis and extrapolation
from single short-term animal study available and thus the scientific justification
clear, though 28 days is likely too short a duration to meaningfully observe all
possible effects for this endpoint.
e. Female reproductive effects: Tier 3: The finding of indeterminacy for the human
data on female reproductive effects is appropriate given the scientific limitations of
the available epidemiology studies and the short-term exposure study in animals.
While there are many epidemiology studies, general lack of data in this area is
striking and there are no studies of key endpoints (e.g., fecundity) in animals. The
finding that PFDA likely causes reproductive effects in female humans is a
reasonable synthesis and extrapolation from a single short-term animal study
available and thus the scientific justification clear, though 28 days is likely too short
a duration to meaningfully observe all possible effects for these endpoints.
f. Cardiometabolic effects: Tier 2: The finding of slight evidence for an association
between PFDA and cardiometabolic effects in humans is a reasonable conclusion
based on a synthesis of the available studies, as is the finding of indeterminate
results based on the evidence from a study in animals.
The justification of would be more coherent if there was a logic tree that walked the
reader through the various pathways by which PFDA affects each examined
contributor to overall cardiometabolic health.
The finding that evidence suggests that PFDA can potentially affect cardiometabolic
health is reasonable and thus the scientific justification for these findings is as clear
as feasible given the limited and mixed evidence base.
g. Neurodevelopmental effects: Tier 2: The finding of slight evidence for an association
between PFDA and neurodevelopmental effects in humans is a reasonable
conclusion based on a synthesis of the available studies, as is the finding of
indeterminate results in animals. The finding that evidence suggests that PFDA can
potentially affect neurodevelopment is reasonable and thus the scientific
justification for these findings is as clear as feasible given the limited and mixed
evidence base. Table 3-38 can be strengthened if it has more information on
35
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
participant ages/age categories and breaking out results by gender as feasible when
clinically relevant (e.g., ADHD tends to be more often diagnosed in boys vs girls).
h. Endocrine, urinarv, and other noncancer effects: Tier 3: The finding of
indeterminacy for an association between PFDA and endocrine, urinary, and other
noncancer effects in humans is a reasonable conclusion based on the analysis of the
available human and animal studies. The finding that there is inadequate evidence
across human animal and other data is clear and scientifically justified.
Carignan
a. Liver effects: This reviewer agrees with the conclusions made bv USEPA that PFDA
exposure is likely to cause liver effects in humans and that evidence from in vivo
and in vitro studies support a potential role for multiple pathways operant in the
induction of hepatic effects from PFDA exposure. Study data are clearly and
appropriately synthesized to describe strengths and limitations. Study findings are
clearly, appropriately and effectively detailed to allow for scientifically supported
syntheses of the findings across sets of studies. Animal study comparisons for PFDA
across dose and by sex were presented and supported by the limited information
available from human studies. Confidence and conclusions for the studies are
scientifically justified, giving appropriate consideration to potential involvement of
PPARa and non-PPARa pathways with respect to the reported liver effects. Weight-
of-evidence decisions for liver effects are clearly described and scientifically
justified.
This reviewer agrees with the overall conclusion that hepatic effects of PFDA in
rodent studies should be considered adverse and relevant to humans.
Table 3-5: Should note units of the median exposure levels (e.g., ppb). [Tier 2]
Stratification by sex and menopause status would be useful for future studies. [Tier
3]
Exposure reduction, medical monitoring and other public health interventions are
needed for communities exposed to PFDA to better protect against and monitor for
liver related health outcomes. Future priorities should simultaneously focus on
legacy (e.g., PFDA) and current use PFASs for such public health interventions. [Tier
3]
b. Immune effects: This reviewer agrees with the conclusions made bv USEPA that
PFDA exposure is likely to cause immunosuppression in humans and that residual
confounding is unlikely to fully explain the associations seen in the literature. Study
data are clearly and appropriately synthesized to describe strengths and limitations.
Most of the study findings are clearly, appropriately and effectively detailed to
allow for scientifically supported syntheses of the findings across sets of studies.
Recommendations for improvement are detailed below. Confidence and
conclusions for the studies are scientifically justified, giving appropriate
consideration to the timing of the exposure and outcome measurements. Weight-
of-evidence decisions for immune effects are clearly described and scientifically
justified.
P 95, Line 18: Should clarify the number and quality of studies on
36
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
immunostimulation and autoimmunity to explain why these outcomes were
excluded. [Tier 1]
Page 96, Line 15: The first comma should be deleted. "The analysesLin
Grandjean..."
Page 99, Line 6: This seems reasonably rigorous. Perhaps note that mixtures of
PFAS can have cumulative effects. [Tier 2]
Page 99, Line 12: Effects should be plural "examining effects in children". [Tier 2]
Table 3-12: Should include the dichotomous titer outcome from Timmerman et al.
(2021). [Tier 1]
Table 3-12: Recommend also including a figure showing effect estimates and 95%
confidence intervals, ranked by confidence level. [Tier 2]
Table 3-12: It is confusing that the final data set used for developing a reference
dose is not included. I understand that it is because the units are different, but
should find a way to prominently clarify in this section and table, and clearly show
those results, as it is confusing for the reader. [Tier 1]
Immune titer section: Should prominently state that vaccine immune titer is a
functional measure of adaptive immune response and therefore an important
indicator of clinically relevent immunotoxicity. [Tier 1]
Page 102, Line 2: 'in' is grammatically incorrect. "Another prospective birth cohort m
examined... " [Tier 2]
Page 102, Line 18: Conversely, those with chronic health concerns may be more
susceptible and thus a reasonable model for investigating such effects. [Tier 2]
Future research on PFAS exposure and effects on adaptive immune response in
humans (e.g., vaccine immune titer) should utilize a birth cohort study design,
consider exposure across early life, include participants with a wide range of
exposure levels, investigate both legacy and current use PFASs, utilize the most
appropriate exposure measures (e.g., serum for legacy PFAS), measure titer at
consistent time points across subjects, and consider sex differences. [Tier 3]
Given the sensitivity of humans to immune effects of PFAS, indication of health
effects using any toxicological model (computational, in vitro, animal or human) for
current use PFASs should be acted on promptly for the protection of public health.
[Tier 3]
c. Developmental effects: This reviewer agrees with the conclusions made by USEPA
that PFDA exposure is likely to cause developmental effects in humans, and that
pregnancy hemodynamics are a potential source of bias that reduces confidence in
studies where exposure is measured after the first trimester. Study data are clearly
and appropriately synthesized to describe strengths and limitations. Most of the
study findings are clearly, appropriately and effectively detailed to allow for
scientifically supported syntheses of the findings across sets of studies.
Recommendations for improvement are detailed below. Confidence and
37
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
conclusions for the studies are scientifically justified, with the exception of the
continuous measure reported by Bach et al. (2016) as discussed below. Appropriate
consideration was given to potential bias of pregnancy hemodynamics. Weight-of-
evidence decisions for developmental effects are clearly described and scientifically
justified.
[Tier 1] Should include robust new study from combined U.S. birth cohorts,
which reported a significant negative association for PFDA with birthweight in
the combined analysis (Table 2, PFDA measured in any trimester), a sensitivity
analysis for PFDA in first trimester maternal serum (Table SI), and stratified by
infant sex (Table 6).
Padula et al. Birth outcomes in relation to prenatal exposure to per- and
polyfluoroalkyl substances and stress in the environmental influences on
child health outcomes (ECHO) program. March 2023. Environmental Health
Perspectives. 131(3). https://ehp.niehs.nih.gov/doi/10.1289/EHP10723
[Tier 1] Bach et al. (2016) did not log transform maternal PFDA serum levels in
the continuous analysis, therefore the confidence rating in that analysis should
be reduced or excluded.
Should note that gestational age adjusted birth weight is the preferred measure
for the birth weight outcome, and indicate which studies utilized that approach.
[Tier 1]
d. Male reproductive effects: This reviewer agrees with the conclusions made by
USEPA that PFDA exposure is likely to cause male reproductive effects in humans
based on evidence in adult male rats. Study data are clearly and appropriately
synthesized to describe strengths and limitations. Most of the study findings are
clearly, appropriately and effectively detailed to allow for scientifically supported
syntheses of the findings across sets of studies. Confidence and conclusions for the
studies are scientifically justified given the similarity of the mechanism of male
reproduction between rats and humans. Weight-of-evidence decisions for female
reproductive effects are clearly described and scientifically justified.
Figure 3-64: The clear and substantial non-monotonic dose response for
testosterone (NTP 2018) is interesting and notable, with an increase in
sperm concentration at the lower dose followed by a significant linear
decline at subsequent increasing doses. Human studies should be discussed
in this context for all male reproductive endpoints. [Tier 2]
Evidence integration, Page 225, Line 2 and Table 3-26. Non-monotonicity of
testosterone effects should be noted as a possible explanation for
inconsistency among male reproductive endpoints in the human studies,
which rarely considered such a dose response. [Tier 1]
Recommend integrating evaluation of non-monotonicity of dose response
into the literature review and evidence integration. [Tier 2]
Future studies should consider possible non-monotonicity of associations
with all male reproductive endpoints. Previous studies could re-analyze
38
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
data with this consideration and consider combining cohorts to improve
sample size and exposure range. [Tier 3]
e. Female reproductive effects: This reviewer agrees with the conclusions made by
USEPA that PFDA exposure is likely to cause female reproductive effects in
humans, and that existing human studies have low sensitivity. Study data are
clearly and appropriately synthesized to describe strengths and limitations. Most
of the study findings are clearly, appropriately and effectively detailed to allow for
scientifically supported syntheses of the findings across sets of studies.
Recommendations for improvement are detailed below. Confidence and
conclusions for the studies are scientifically justified given the similarity of the
mechanism of female reproduction between rats and humans. Weight-of-evidence
decisions for female reproductive effects are clearly described and scientifically
justified. Recommendations for improvement are detailed below.
Figure 3-53: It is unclear which are male and which are female, these should be
labeled. [Tier 1]
Should note that while the histopathology study found no effects on the mammary
glands, timing of the exposure and outcome was not in a sensitive developmental
window (in utero or pregnancy). [Tier 2]
The lactation duration outcome should be included in this section, as the
pregnancy and postpartum period is a sensitive developmental window for
mammary gland development (as is in utero/early life). It should also be noted
that mammary gland development is a sensitive outcome for other PFAS (e.g.,
PFOA). [Tier 1]
Timmermann CAG, et al. Shorter duration of breastfeeding at elevated
exposures to perfluoroalkyl substances. 2017. Reprod Toxicol 68:164-170.
doi: 10.1016/j.reprotox. 2016.07.010.
Romano ME et al. Maternal serum perfluoroalkyl substances during
pregnancy and duration of breastfeeding. 2016. Environ Res. 149:239-246.
DOI: 10.1016/j.envres.2016.04.034
For additional relevant references:
https://ehp.niehs.nih.gov/doi/full/10.1289/ehp.125-A17
The discussion and evidence stream (Table 3-29) should be revised to reflect the
following:
The statement of 'unclear biological relevance of increases' in testosterone in the
single (high confidence) animal study (NTP 2018) should be revised to note
coherence with the finding of continuous diestrous as high levels of androgens in
women causes infrequent, irregular or non-existent menstruation. [Tier 1]
The finding of high testosterone and a continuous phase of diesterous in the NTP
(2018) study is also consistent with the finding of decreased progesterone in
Leydig tumor cells (Zhao et al. 2017). High levels of testosterone in women are
associated with levels of progesterone, irregular menstruation and decreased
39
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
fertility. [Tier 1]
f. Cardiometabolic effects: This reviewer agrees with the conclusions made bv USEPA
that PFDA exposure is suggestive of cardiometabolic effects in humans, and of
clear associations with serum lipids. Study data are clearly and appropriately
synthesized to describe strengths and limitations. Most of the study findings are
clearly, appropriately and effectively detailed to allow for scientifically supported
syntheses of the findings across sets of studies. Confidence and conclusions for the
studies are scientifically justified. Weight-of-evidence decisions for
cardiometabolic effects are clearly described and scientifically justified.
g. Neurodevelopmental effects: This reviewer agrees with the conclusions made bv
USEPA that PFDA exposure is suggestive of neurodevelopmental effects in
humans, but not sufficient to infer causation given inconsistent findings in
epidemiologic studies and an absence of animal studies. Study data are clearly and
appropriately synthesized to describe strengths and limitations. Most of the study
findings are clearly, appropriately and effectively detailed to allow for scientifically
supported syntheses of the findings across sets of studies. Recommendations for
improvement are detailed below. Confidence and conclusions for the studies are
scientifically justified. Weight-of-evidence decisions for neurodevelopmental
effects are clearly described and scientifically justified.
High quality research is needed on neurodevelopmental effects in animal studies.
[Tier 3]
Future research should also investigate neurodegenerative outcomes. [Tier 3]
h. Endocrine, urinarv, and other noncancer effects: This reviewer agrees with the
conclusions made by USEPA that PFDA exposure is currently inadequate for
endocrine, urinary and other noncancer effects. It should be made clear that an
exception to this is conclusion is effects on reproductive hormones, which are also
part of the endocrine system and for which effects were previously discussed
(male and female reproduction). [Tier 1]
The thyroid hormone section would benefit from a table and/or figure grouped by
confidence level and noting limitations. [Tier 2]
Clarification: Human evidence, but this comment could apply to animal as well.
Established determinations of thyroid effects for other PFASs (e.g., PFOA and
PFOS) should be noted. [Tier 2]
Faustman
a. Liver effects: This reviewer is in agreement with the "available data
synthesis" of data input for this non-cancer endpoint. The human epi and
animal in vivo data in the report is reviewed and the strengths and
weaknesses of each study is presented and those reviews support the study
confidence score as well as the strengths and limitations described for each
study. Information on cross species consistency as well as dose response
and effect level discussion support Section 3 Non Cancer Hazard
Identification processes for hepatic endpoints (and subsequent non cancer
40
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
endpoint evaluations). The presentation style and thoroughness of the
hepatic endpoint evaluations contributes to the clarity and transparence in
all aspects of this organ system review. Dose response information for the
individual studies is presented methodically in Tables such as in Figures 3-5,
3-7 and 3-9 (examples). In addition, information on specifics of the severity
and incidence of various hepatocyte lesions in tested species is given in
Tables such as 3-6.
This reviewer did support basic assumptions of relevance statements such
as assumed relevance of animal studies unless mechanistically proven to be
non relevant for humans.
The overall synthesis of these data across platforms and species and hepatic
relevant endpoints are then presented in a weight of evidence table (Table
3-11) that provides a very clear evidence path for excellent reviews of
available data and provide both in text explanation as well as in Table 3-3 an
excellent study by study evaluation.
The section 3.2.1 Hepatic effects provide a robust discussion of what serum
enzyme biomarkers are assumed to be indicative of a potential adverse
outcome versus adaptive response providing dose response information,
cross species consistency and mechanistic discussions to support the
assumption of adversity for human responses.
Likewise for the mechanistic evaluations of PPAR a versus non PPAR a dependent
responses following PFDA, this reviewer supports the initial discussion of and
support of using both in vivo and in vitro data and Computational Toxicological data
to support the decision for not being able to discount the non PPAR a dependent
PFDA response pathways. As mentioned in this PFDA report this decision is also
supported by the decision for other longer chain PFAS compounds.
b. Immune effects: The IRIS team have very carefully detailed the complexity of
potential immune system impacts of PFDA in Section 3.2.2. Due to this complexity
of immunotoxicity responses, the report lays out the definitions and response types
that will be reviewed for this IRIS report. Consistent with WHO/IPCS definitions and
with the availability of data, this hazard identification reviewed available data for
both immunosuppression and sensitization and allergic response. This reviewer
agrees with their decision to not examine immunostimulation and or autoimmune
disease response based on lack of references for these responses. Overall, this
reviewer agrees with the assessments of the available data and the study
confidence conclusions. I also agree with the lists of potential strengths and
weaknesses of the studies available and noted in particular that most to many of
the human epidemiology studies for immunosuppression and infectious disease are
reported from country cohorts outside of the US with an emphasis of detailed
findings from the Faroe Islands. This reviewer also noted and agreed that "cross-
stream coherence" existed with evidence of immunosuppression in both animals
and humans (Table 3-19). It was noted that children and in utero fetuses maybe
susceptible populations. Minimal information is available on MOA but both animal
and human data points to possible NFkB pathway involvement as well as
modification of cytokine response. Many questions remain for this endpoint but this
41
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
reviewer is not requesting any Tier 1 or 2 recommendations at this time and agrees
with both the study confidence conclusions and weight of evidence tables. A Tier 3
recommendation is for further clarification of how the WHO guidelines differ from
some of the other texts on immunotoxicity as this reviewer believes that this will be
an important area of expanding research on response across the PRAS universe of
chemicals. Also following COVID it is understandable that we are looking more
comprehensibly about the immune system responses.
c. Developmental effects: The draft IRIS report identifies 46 epidemiological studies
of relevance for assessing developmental impacts. The report notes that many of
these have some information on fetal growth decreases and that in general many
of these observations are suggestive of fetal growth restriction. In addition,
several rodent studies support these observations. Again, USEPA has presented a
thoughtful and critical evaluation of these studies. I concur with the observations
that the Human Epidemiology studies are suggestive for associations of fetal and
postnatal growth restrictions however these are appropriately labeled as slight.
Observations of fetal growth reduction findings in rodent studies provides cross-
stream coherence which I agree with. This evidence for growth reduction is
observed as dose related changes in the rodent models. The report discusses the
evidence for this reduction at non-toxic versus maternally toxic levels of exposure
and Figures 3-56 and 3-57 and Table 3-23 show evidence for these reductions
occurring at less than maternally toxic doses. This reviewer agreed with these
findings and concurred with the evidence stream summary for rodent growth
retardation during development as moderate and together agreed with the
statements that the overall inferences and summary judgment as likely evidence
for growth retardation in humans. Evidence for other endpoints following PFAS
exposure such as spontaneous abortion, gestational duration, etc did not support
there use in the development of toxicology values. These details were presented
in Table 3-24 and these are detailed and rationale is presented in a clear manor.
The authors discuss reasons for variability in response between the human and
rodent models and speculate that there maybe hemodynamic changes across
gestation. The evidence and magnitude of these hemodynamic changes are
consistent with pregnancy related changes however this reviewer remained
skeptical for these suggestions and look forward to additional research that
USEPA had will provide that could resolve this issue. Minimal information was
obtained by the zebra fish study other than to acknowledge that growth
restriction will manifest its self under high exposures esp in sensitive populations
such as children.
Interpreting the complexity of the human epidemiological studies considering
PFDA association with developmental endpoints has been challenging and the
USEPA discussions on gestational duration and pre term birth illustrate their
methodical evaluation ( see Figures 3-50, 3-51and 3-52 as examples). In many
cases the available animal studies helped to support when dose response
relationships were present and also when these relationships were complicated
by significant maternal toxicity. This reviewer agrees with the USEPA conclusions
that complications of when PFAS compound exposures were taken across varied
lifecourse times and variations in the ability to associate impacts with specific
compounds such as PFDA remain unresolved for these endpoints in particular.
42
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
This reviewer will make a Tier 3 recommendation that in the future when the
USEPA does a cumulative risk assessment across the multiple PFAS related
compounds that they are reviewing, these developmental endpoints in particular
are re-considered for mixture effects.
d. Male reproductive effects: This reviewer agrees with the approach that
USEPA used in their review of male reproductive effects. Their evidence
profile for PFDA exposure and male reproductive effects is given as Table 3-
26 and that table identified the availability of both human, animal and
some mechanistic evidence. The human evidence overall was labeled with
an evidence stream judgment of Indeterminate for semen evaluation,
pubertal development and reproductive hormonal changes and this
reviewer agreed with this designation and with the reasons provided for
that judgement. Despite having observed decreased motility with increased
PFDA exposure in one study, 3 others did not show a similar relationship. In
addition, decreased testosterone was seen in one adult study and one
study of adolescents however other studies of both medium and low
confidence resulted in "unexplained inconsistencies" in the overall studies
available for assessment. The availability of a 28 day animal study, albeit
inadequate for evaluating the full spermatogenesis cycle, provided
moderate evidence for related effects in a dose dependent manner for
sperm impacts, testicular histopathology, reproductive hormones and
organ weights. When combined with mechanistic data on altered Leydig
cell androgen function and Tox Cast observations following PFDA exposure,
this assessment provided plausible mechanistic evidence for how these
effects were occurring.
(Please see notes regarding the sperm DNA fragmentation study discussed
under the cancer endpoint. There should be cross referencing with those
finding under this toxicity endpoint. Tier 1 recommendation is to ensure
that this cross- referencing is done and sited as mechanistic evidence
supportive of the current conclusion in this IRIS review.)
e. Female reproductive effects: This reviewer agrees with the USEPA
assessment of this endpoint. Despite the presence of various human
studies looking at endpoints(such as reproductive hormones, fecundity,
puberal development, menstrual cycle and endometriosis) relevant to
female reproductive function, these studies were identified by the
evidence stream summary as "indeterminate" and this reviewer agreed
with this designation since these studies had unexplained inconsistencies
and limited sensitivity to determine dose response relationships and
associations. The availability of a 28 day rodent study, albeit limited in the
lifecourse that it examined, provided moderate evidence for dose response
for estrous cycle, organ weight and altered hormone levels and provided
context for the assumed relevance of animal studies for human and
designation of evidence integration of likely. Mechanistic studies that
showed the ability of PFDA to inhibit progesterone production provided
supportive context for the in vivo animal studies. It is striking to this
reviewer the lack of adequate toxicological studies to cover important life
43
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
stages such as that covered in normal development and reproductive
assessments. Given the widespread exposure to the PFAS family of
compounds and consistent impacts on important reproductive endpoints,
that these compounds have not been prioritized for additional in vivo
characterization. This reviewer supports and adds as a Tier 3
recommendation the need for future studies on "the effect of PFDA on
female fertility and pregnancy outcomes in exposed animals from sub-
chronic, chronic, development, or multigenerational studies as well as in
vivo or cell culture mechanistic studies." (see page 3-204, lines 19-25).
f. Cardiometabolic effects: The USEPA presents a review of five of 6 medium
confidence human epidemiological studies that showed an increase in serum total
cholesterol with higher PFDA exposures. Other serum lipid changes were not this
consistent and lack of coherence was noted across many of the serum lipid
assessments. The evidence stream judgement was labeled as slight due to the
positive associations noted above. However most other observations were
inconsistent or lacked coherence. The animal studies were indeterminate due to
species differences and lack of coherence. This reviewer was supportive of the
USEPA's conclusions.
g. Neurodevelopmental effects: USEPA identified 13 human studies of
neurodevelopmental outcomes in humans and five out of six studies that looked at
behavioral issues and/ or attention problems were positively associated (note that
some studies were from the same cohort and USEPA detailed when separate
cohorts were evaluated). USEPA stated that the ADHD diagnosis studies were the
most clinically relevant but these had "inverse findings" for association with PFDA.
Figure 3-76 and Table 3-38 (medium confidence studies only) summarize the details
of these studies and look at specific outcomes such as diagnosed ADHA, behavioral
regulation, externalizing problems and various behavioral summary scores. Note
that only Niu et al, 2019 was identified by USEPA as having sensitivity while the
other studies were deficient due to "limited exposure contrast". In summary this
evidence stream judgment was slight and the evidence integration summary
judgement was suggestive. There was not a neurodevelopmental study in animals
available nor were any references cited for in vitro or Comp Tox data. Also, no
mention of consistency or inconsistency of neurodevelopmental outcomes from
other structurally related PFAS compounds cited. Tier 2 recommendation is made
by this reviewer to clarify (ie add one or two sentences) that state or modify the
availability of any supportive evidence such as cross coherence with animal or
mechanistic data and a reference to consistency with other PFAS related long chain
compounds.
h. Endocrine, urinary, and other noncancer effects: Consistent with studies of other
PFAS compounds, the USEPA focused on thyroid function and response. The
document reports 23 human studies that have examined thyroid hormones and
PFDA exposure and Figure 3-77 evaluates these studies. The text describes
several factors that resulted in a designation of "deficient" for outcome
ascertainment and several of these are consistent with other sections of this
report such as a study being under powered and unable to ascertain association
with exposure with needed sensitivity (7 of the studies listed in Figure 3-77 as
44
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
listed as having adequate sensitivity). Several other factors that were used to
identify "deficient" studies are concerning. These include factors such as only
studies with fasting thyroid measurements being deemed non deficient and also
considering studies that did not take thyroid measures at the same time of day
for all measurements deficient. It is good that these studies were not excluded
but designation as "deficient" definitely affected interpretation during the
epidemiology study evaluation. Other factors that are more concerning regard
the fact that interpretation of the studies of this PFAS compound, PFDA was not
consistent with the interpretation of other recent PFAS IRIS documents as well
as not clearly supported by more recent references in the literature. For
example, prior PFAS compounds have been reported as having consistent effects
on thyroid hormones and yet these changes were not always associated with
concurrent or "anticipated" changes in TSH. Lines 27-32 on page 3-250 refer to
these differences and this is listed as "inconsistent" and the resulting impact of
this designation has resulted in listing as "indeterminate" on Table 3-41 whereas
this reviewer would have designated this evidence stream as "slight" or higher.
Tier 1 recommendation from this reviewer is to re-examine this study
evaluation of the epidemiology studies for thyroid effects given the comments
listed above and the comments received from the public on this topic as well
for consistency in interpretation of the pathway responses from related recent
IRIS documents on other related PFAS compounds such as PRHxA, etc.
Similar concerns are raised by this reviewer for the interpretation of the animal
studies and for the interpretation of the mechanistic studies. For example, Table
3-41 lists a high confidence study available for evaluation and this study showed
consistency for decreased fT4 in both male and female rats and showed a dose-
response for deceased TSH in males and decreased fT4 in males and females at
levels (of 1.25 mg/kg-d. No change in tT4 was observed. These observations
were designated as "lack of expected coherence across thyroid measures with
any currently available understanding of adverse thyroid-related changes" and
this designation is inconsistent with previously observed and discussed PFAS
associated and observed changes. Tier 1 recommendation from this reviewer is
to re-examine the in vivo animal studies as this evidence stream would suggest
an evidence steam judgement as moderate.
Mechanistic data cited in Table 3-41 include information from studies that "PFDA
may impair" HPT axis response, PFDA may impair binding of thyroid hormones,
and PFDA may alter thyroid activity in rats. This supplemental data is limited due
to route of exposure ip and high doses but does illustrate a potential for
receptor binding. See comments in the preceding paragraph.
Note that the animal study discussion and summary in Table 3-41 notes that no
histopathology was observed but frequently such a suggestion continues to have
a comment about the length of time that observations were made ie a 28 day
study or longer as such pathologies would not necessarily be expected in the
shorter studies. Tier 1 recommendation is to clarify this in the table and the
text on lines 1-11 on page 3-255.
Lines 32 to 38 (page 3-263) and lines 1-4 (page 3-264) are important to note
45
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
regarding the potential impact of such thyroid effects during sensitive life-stages.
This reviewer applauds the inclusion of such information.
Tier 1 recommendation to USEPA is to reevaluate the designation of
"inadequate evidence" in Table 3-41 for the evidence integration summary
judgement given the recommendations and discussions above and by other
reviewers and public commentors. It is the opinion of this reviewer given the
suggested changes above that the overall designation would be "suggestive" of
higher.
Urinary and non cancer endpoints: This reviewer did not have significant
comments about the summary of potential Kidney toxicity or general toxicity
associated with PFDA studies. High doses used in the general toxicity studies
diminished the use of these studies and this reviewer agrees with the decision
not to use these in the evidence integration process. This reviewer noted that
the duration of the animal studies would not be long enough to definitely
diagnosis presence of histopathological changes in the kidney or most other
organs.
Fisher
a. Liver effects: Yes, liver effects are relevant for human exposures. An extensive effort
by EPA. MOA discussion is important.
a.i. Yes, the PPARa and non-PPARa issues are relevant for humans. Good review.
Clear.
b. Immune effects: Yes, these studies mav be relevant to humans.
b.i. 1 think this is a huge issue that is not adequately addressed, mixtures. A more
rigorous methodology is required to examine the potencies of PFAS for specific
endpoints. Statistics should be supported by biological plausibility (Tier 2).
Clarification: An example of a rigorous method would be direct experimentation
with a similar mixture, either in vitro or in vivo which would support the approach
used. 1 do not know if this data is available. 1 suspect in vitro methods would be
the best.
c. Developmental effects: Yes, the evidence does indicate that developmental effects
may be possible given sufficient exposures to PFDA.
c.i. Slight evidence in humans seems adequate.
d. Male reproductive effects: 1 am not qualified to answer this question.
e. Female reproductive effects: Yes, given sufficient exposure conditions (if thev exist),
may cause female reproductive effects in humans.
f. Cardiometabolic effects: 1 agree with this statement. It would be useful to include
some background on why this was evaluated based on MOA. (Tier 2).
g. Neurodevelopmental effects: 1 agree with this this statement.
46
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
h. Endocrine, urinarv, and other noncancer effects: 1 agree with this conclusion of
inadequate evidence. The endocrine evidence (thyroid) discussion seems dated
when talking about classical thyroid biomarkers (serum TSH and thyroid hormones).
There are many examples in the literature where serum free T4 in humans and total
and/or free T4 in animals decrease in response to a chemical or drug and TSH
appears unchanged or normal. 1 recommend that EPA form a focus group to address
the expectations for thyroid hormone perturbations and related adverse toxicity.
This is an important topic in light of high-throughput efforts looking at thyroid
molecular initiating events. 1 think the EPA interpreted the thyroid data incorrectly
(the data are ok) given the growing body of evidence about the thyroid system.
Clarification: Animal studies and even human studies with compounds that target
the thyroid system result in outcomes that are not classic (decrease in serum T4 and
increase in serum TSH). The results are often not considered or understood. It
would be very useful for the EPA offices to have the same view about thyroid
endpoints (Tier 2).
Georgopoulos
a. For liver effects, the Toxicological Review appropriately concludes that the
available evidence indicates PFDA is likely to cause liver effects in humans given
sufficient exposure conditions, based on short-term studies in rats and mice
demonstrating consistent and coherent effects with a clear biological gradient.
The liver findings for PFDA were similar to those for other structurally related
long-chain PFAS and determined to be adverse and relevant to humans. The
Toxicological Review also appropriately evaluated evidence relevant to the
potential involvement of PPARa and non-PPARa pathways with respect to the
reported liver effects to conclude that available data support a potential role for
multiple pathways operant in the induction of hepatic effects from PFDA
exposure, although the interaction of those pathways within a mode of action
(MOA) cannot be specifically determined.
b. For immune effects, the Toxicological Review appropriately concludes that the
available evidence indicates PFDA is likely to cause immunosuppression in
humans given sufficient exposure conditions. EPA reached this decision primarily
based on consistent evidence of reduced antibody responses from two
epidemiological studies in children and one study in adults. The document also
presents literature evidence from short-term animal studies for coherent
immunomodulatory responses consistent with immunosuppression, but the
animal evidence overall is uncertain. As the assessment is based on studies in
humans, the Toxicological Review appropriately concludes that the immune
effects are considered relevant to humans. The Toxicological Review document
also presented and discussed concerns that potentially exposures to other co-
occurring ("highly correlated") PFAS (and other immunomodulating agents) could
contribute to observed immune effects and concluded that confounding by other
factors, including PFAS co-exposures, was unlikely to fully explain the associations
of immune effects with PFDA that are reported in the literature. As a qualitative
conclusion this can be considered appropriate, but confounding effects can
impact estimates of PFDA levels used in deriving RfDs.
• Note: 1 am discussing the issue of confounding by other PFAS exposures in
47
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
my response to Charge Question 3, as confounding may impact the
calculated values of metrics involved in the derivation of the RfD for
immune and developmental effects.
c. For developmental effects, the Toxicological Review appropriately concludes that
the available evidence indicates PFDA is likely to cause developmental effects in
humans given sufficient exposure conditions, based primarily on consistent
findings of dose-dependent decreases in fetal weight in mice gestationally
exposed to PFDA, supported by some coherent evidence of decreased birth
weight from studies of exposed humans in which PFDA was measured during
pregnancy, although uncertainties in the available epidemiological evidence
reduced the strength of the available human evidence to slight for an evidence
base that might otherwise be interpreted as moderate. The Toxicological Review
reasonably concludes that the developmental effects in mice are considered
relevant to humans given similar findings of fetal growth restriction in mice and
humans.
d. For male reproductive effects, the Toxicological Review appropriately
concludes that theavailable evidence indicates that PFDA is likely to cause
male reproductive effects in humans given sufficient exposure conditions,
based on coherent evidence from a study that exposed adult male rats to
PFDA for 28 days. The findings from the animal study were reasonably
presumed to be relevant based on the conserved role of androgen-
dependent pathways in male productive functions across species.
e. For female reproductive effects, the Toxicological Review appropriately concludes
that the available evidence indicates that PFDA exposure is likely to cause female
reproductive effects in humans given sufficient exposure conditions, based on
coherent evidence from a study that exposed adult female rats to PFDA for 28
days. The document discussed available human data showing mostly null
associations between PFDA and adverse female reproductive outcomes but
concluded that this was probably due to low study sensitivity for observing effects.
The Toxicological Review reasonably concludes the female reproductive effects are
considered relevant to humans given that mechanisms of female reproduction are
similar between rats and humans.
f. For cardiometabolic effects, the Toxicological Review appropriately concludes that
the available evidence suggests but is not sufficient to infer that PFDA exposures
may have the potential to cause cardiometabolic effects in humans given sufficient
exposure conditions, based on associations between PFDA and serum lipids,
adiposity, cardiovascular disease, and atherosclerosis, as there is considerable
uncertainty and the evidence is largely inconsistent across available
epidemiological studies. EPA also appropriately concluded that evidence in
experimental animals was indeterminate.
g. For neurodevelopmental effects, the Toxicological Review appropriately
concludes that the available evidence suggests but is not sufficient to infer that
PFDA exposure may have the potential to cause neurobehavioral effects in
humans given sufficient exposure conditions, based on associations between
PFDA and outcomes related to attention and behavior in epidemiological studies.
48
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
However, the evidence is largely inconsistent across studies, which adds
considerable uncertainty. As no information was found for relevant experimental
animal studies, EPA appropriately concluded that neurodevelopmental effects
evidence in animals was indeterminate.
h. For endocrine, urinarv, and other noncancer effects (i.e., hematological,
respiratory, digestive, dermal, musculoskeletal, and nervous systems), the
Toxicological Review appropriately concludes that there is inadequate evidence
to determine whether PFDA exposure has the potential to cause these effects in
humans.
Haney
a. Liver effects: The studv confidence conclusions appear scientifically justified, giving
appropriate consideration to important study attributes. For each study in the
relevant figures of Section 3.2.1 (e.g., Figure 3-8), the consideration ratings appear
overall consistent with (i.e., provide the scientific justification needed for) the
overall study confidence level rating. Accordingly, the study confidence conclusions
appear scientifically justified, with EPA having given appropriate consideration to
important study attributes.
This question (under "2") also concerns whether the presentation and analysis of
study results is clear, appropriate, and effective, a step just upstream of synthesis
that if well done, then allows for appropriate syntheses of the endpoint-specific
findings. In this reviewer's opinion, the overall presentation and analysis of study
results appears clear, appropriate, and effective. Additionally, it appears that by and
large, the available data on hepatic effects are clearly and appropriately synthesized
to describe the strengths and limitations. P. 3-50, lines 18-19 state that, "Taken
together, the available evidence indicates that PFDA exposure is likely to cause
hepatotoxicity in humans given sufficient exposure conditions (see Table 3-11)."
Table 3-11 is the evidence profile table for hepatic effects, which among other
information contains both factors that increase certainty and factors that decrease
certainty along with evidence stream (i.e., human, animal, mechanistic/
supplemental) judgments/rationales and a summary judgment. Obviously, the text
of the document (Section 3.2.1) also contains information relevant to and
supporting the weight-of-evidence (WOE) for hepatic effects. Importantly,
recommendations of the Hall et al. (2012) paper were considered by the EPA in
assessing the adversity of observed hepatic effects (e.g., in discussing sufficiently
supporting histological evidence). However, while p. 3-49 (lines 1-4) state that,
"Overall, application of the recommendations from Hall et al. (2012) clearly
supports the conclusion that PFDA exposure has multiple and coherent effects on
liver histopathology, serum biomarkers and liver weights in exposed animals
(primarily rats) that meet the criteria for adversity", this is not clear from the
preceding text. As a Tier 2 suggested revision, EPA should explicitly show how the
Hall et al. (2012) criteria for adversity are met. For example, while fold increases in
biomarkers of liver dysfunction (e.g., bilirubin, bile acids/salts) are explicitly
discussed on p. 3-48 (lines 36-38), the fold increases needed to fulfill a second
criterion (e.g., in ALT or ALP) are not explicitly discussed on lines 34-35 (e.g., a 2- to
3-fold increase in ALT or biomarkers of hepatobiliary damage such as AST, ALP and
y-glutamyltranspeptidase [yGT]; Tables 3-7 and 3-8 are relevant). Additionally, as a
49
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Tier 2 suggested revision, the EPA should consider additional tables and/or figures
that would help readers visualize important EPA conclusions, such as "coherent
changes in serum biomarkers, histopathology, and liver weights" cited in Table 3-11.
This being said, taken together, the weight of the available scientific information
presented (e.g., evidence of PPARa-independent pathways) provides support that
assuming sufficiently high exposure over a sufficiently long duration (i.e., "given
relevant exposure circumstances"), PFDA exposure is likely to cause hepatic toxicity
in the general human population, which includes potentially susceptible
subpopulations (e.g., those with pre-existing liver disease). Though there is some
room for improvement (e.g., additional tables/figures), the bases for this WOE
decision are clearly described in the text (Evidence Integration) and Table 3-11 of
the document.
Last but not least, in regard to PFDA being likely to cause hepatic effects in humans
given sufficient exposure conditions, footnote 9 (p. 3-50) states that, "The
"sufficient exposure conditions" are more fully evaluated and defined for the
identified health effects through dose-response analysis in Section 5." For such
dose-response analyses conducted on animal data, this begs questions such as...
"What laboratory animal species, if any, has a dose-response for hepatic effects that
is predictive (after dosimetric adjustments) of the same or similar effects in
humans?" While the document states that the "likely to cause" conclusion is "based
primarily on coherent liver effects in rats (and, to a lesser extent, mice)" (p. 3-50,
lines 19-21), no attempt is made to establish rodents as good laboratory animal
models for PFDA-induced liver effects in humans, much less identify one species as
the better animal model for predicting the same or similar effects in humans.
Scientific justification should be provided for why the species (i.e., the rat)
ultimately providing the basis of the organ-specific toxicity factor (i.e., osRfD) is
expected to have a dose-response more predictive of that in humans than other
species (i.e., the mouse) for which data are available (Tier 1 necessary revision). It is
not scientifically robust to simply state "that without evidence to the contrary, the
human relevance of animal findings is assumed" (p. 1-14, lines 20-21), and this
statement could be equally applied to negative findings in a given laboratory animal
species. When not addressed scientifically, use of data from a given animal species
for dose-response assessment applied to humans can be a large and key area of
uncertainty (e.g., where significant interspecies differences in sensitivity exist in the
absence of data to inform identification of the most human-relevant laboratory
animal species) that pertains directly to the meaningfulness of the resultant toxicity
factor itself and for credibly informing risk/hazard management decisions.
The National Research Council (NRC) has advised that proper characterization of
uncertainty is essential in risk assessment since an assessment that omits or
underestimates uncertainty can leave decisionmakers with a false sense of
confidence in estimates of risk (NRC 1983,1994,1996,2002). Use of an animal model
as a surrogate for humans is an aspect of uncertainty that should be adequately
addressed and characterized in an assessment (USEPA 2005a). However, the
document does not address the uncertainty regarding the most appropriate animal
model and whether a similar dose-response may or may not be expected in humans
(after appropriate dosimetric adjustments). As a Tier 1 necessary revision, EPA
should: (1) attempt to scientifically justify whether the mouse or rat is likely most
50
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
biologically representative of humans such that the same or similar effects (hepatic,
etc.) are expected in humans at similar closes when converted to human equivalent
doses (HEDs); and (2) in the event (1) cannot be established with sufficient scientific
confidence, acknowledge within the assessment that the choice of the most
appropriate laboratory animal model for prediction of PFDA-induced adverse
effects in humans has not been scientifically established (i.e., is not "settled
science") but rather species selection is based on policy. This is a major area of
uncertainty, but it is not without precedent that it be explicitly acknowledged in the
associated uncertainty section (e.g., USEPA 2021).3 it is imperative for EPA to
acknowledge whether or not the dose-response data for a species selected for
osRfD derivation (most often the most sensitive species) is known or likely to be
(e.g., based on greater relevant biological similarities) the most predictive available
for similar effects in humans or if dose-response data from another species might
be equally relevant to humans and result in a significantly different (e.g., higher)
osRfD. Such evaluations and statements are Tier 1 necessary revisions that would
increase transparency and thereby improve influential EPA dose-response
assessments, significantly improve uncertainty sections, and provide a more
complete picture of the extent to which important/key aspects of an assessment
are "settled science." These comments on laboratory animal species also apply as
Tier 1 necessary revisions to other effects where animal data form the basis of an
osRfD (e.g., male and female reproductive effects for subchronic osRfDs).
a.i. In regard to rodent hepatic effects involving PPARa, the bottom line from
Appendix A seems to be that "human relevance is the default, and
mechanistic evidence will need to be compelling and strong to conclude
otherwise (i.e., to conclude that findings in animals are not relevant to
humans)" (p. 9-14, lines 25-27). As opposed to WOE for establishing the
unlikely relevance of an effect to humans, this default appears to establish a
scientific burden of proof of being able to confidently rule out the potential of
non-PPARa pathways as possible contributors to hepatic effects. Accordingly,
it makes the following comment almost moot. Section 2.4.2 of Appendix A
states [emphasis added] that, "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..." (p. 2-25, lines 1-4). This seems to contradict
the draft assessment statement, "PFDA can activate the human PPARa in
vitro" (p. 3-50, line 5). EPA should resolve this apparent discrepancy (Tier 1
necessary revision).
b. Immune effects: EPA has given consideration to important study attributes. For
each study in the relevant figures of Section 3.2.2 (e.g., many of the figures from
Figures 3-10 through 3-23), the consideration ratings appear overall consistent
with (i.e., provide the scientific justification needed for) the overall study
confidence level ratings in the draft assessment. However, I do have comments
below regarding whether EPA has given full consideration to the epidemiological
3 For example, in Section 2.2.8 (p. 2-20), the IRIS assessment for tert-butyl alcohol (USEPA 2021) acknowledges that,
"Most importantly, which animal species or sexes might be more comparable to humans is unknown."
51
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
study weaknesses/limitations and whether the associated results should be used
for quantitative risk assessment and toxicity factor (e.g., RfD) derivation.
Additionally, this question (under "2") concerns whether the presentation and
analysis of study results is clear, appropriate, and effective, a step just upstream
of synthesis that if well done, then allows for appropriate syntheses of the
endpoint-specific findings. The overall presentation of study results appears clear
and various strengths and limitations are described. Table 3-19 is the evidence
profile table for immune effects, which among other information contains some
factors that increase and decrease certainty along with evidence stream (i.e.,
human, animal, mechanistic/supplemental) judgments/rationales and a summary
judgment. Obviously, the text of the document (Section 3.2.2) also contains
information relevant to the WOE for immune effects. As a Tier 2 suggestion, the
EPA should consider additional tables and/or figures that would help readers
visualize important EPA conclusions, such as "coherent evidence of potential
immunosuppression in rats and mice at doses >0.089 mg/kg-d across two
high/medium confidence studies" cited in Table 3-19.
P. 3-92, lines 35-37 state that, "Altogether, considering the available evidence
from human, animal and mechanistic studies, the evidence indicates that PFDA
exposure is likely to cause adverse immune effects, specifically
immunosuppression, in humans, given sufficient exposure conditions (see Table
3-19)." EPA indicates that this hazard judgment is driven primarily by consistent
evidence of reduced antibody response from human epidemiological studies
(mostly from two birth cohort studies)..." (pp. 3-92 and 3-93, lines 37-1). However,
it is not clear that EPA has sufficiently considered adversity for the primary basis of
this hazard judgment, fully considered the weaknesses/limitations of this
evidence¦, or fully considered the lack of supporting evidence from studies on
potentially increased incidences of disease. For example, even without considering
potential confounding, 3/4 odds ratios (ORs) for PFDA and antibody
concentrations falling below the generally considered protective level of 0.1
lU/mL for tetanus and diphtheria in children ages 5 years (n=510) or 7 years
(n=386) contain 1, indicating that the WOE from this key study cohort is for no
statistically significant associations with less-than-protective serum antibody
concentrations in children (see eTable 4 of Grandjean et al. 2012).4 Consistent
with this, the more recent Grandjean et al. (2017a) study states [emphasis added]
that, "With many antibody concentrations being close to the assumed clinically
protective level of 0.1 lU/mL, logistic regression showed only weak tendencies for
antibody levels below the limit to be associated with serum PFAS
concentrations."5 Also consistent with a WOE for no statistically significant
associations (much less being able to say "effects" as there are problems with
causal attribution to PFDA), 6 of 8 confidence intervals (95% CIs) for both tetanus
4 The confidence interval (CI) for the one statistically significant OR of 1.36 (age 5, diphtheria) is (1.04,1.77), with the
lower end of the CI practically equal to 1 (eTable 4 of Grandjean et al. 2012).
5 Also, while Grandjean et al. (2017b) state that, "At age 5,152 (44%) children had antibody concentrations lower than
the protective level of 0.1 lU/mLfor diphtheria and 126 (36%) for tetanus", this appears inconsistent with Table 1 of
that study, which shows that the 25th percentiles for diphtheria and tetanus serum antibody concentrations were 0.1
lU/mL
52
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
and diphtheria serum antibodies included 0% change per 2-fold increase in
maternal and age 5 serum PFDA (see Table 3 of the study). Additionally, in
Grandjean et al. (2012), PFDA was correlated with other PFAS (e.g., PFOS, PFOA,
PFNA) that had some associations with antibody concentrations falling below the
protective level of 0.1 lU/mL (see Table 2 and eTable 4 of Grandjean et al. 2012).6
Moreover, the level of serum antibodies corresponding to a clinically protective
level appears to be assay specific. For the toxin binding inhibition (ToBI) assay
apparently used in the Faroe Islands studies, >0.01 lU/mL is considered to be the
clinically protective level, not the value of >0.1 lU/mL indicated by study authors.
Considering the comments above, this means that the reported decreases in serum
antibodies are even less likely to be biologically/clinically significant. This is not
surprising given the rarity of tetanus/diphtheria cases, particularly in those fully
vaccinated, and the WOE for PFDA not being associated with statistically significant
increases in the incidences of diseases based on the epidemiological literature (all
discussed below). To say the least, all this brings into serious question the validity of
the EPA's assumptions regarding the clinical relevance/adversity of these serum
antibody endpoints.
The clinically protective level cited by Grandjean et al. was >0.1 lU/mL. However:
• Grandjean et al. (2012) reported that "serum concentrations of antibodies
against the tetanus toxoid were measured in coded samples by the Statens
Serum Institut using enzyme-linked immunosorbent assay...", citing
Hendriksen et al. (1988);
• Hendriksen et al. (1988) describes the ToBI assay, which is a modified ELISA;
and
• WHO (2017) indicates that for a modified ELISA, clinical protection is achieved
at >0.01 lU/mL, not >0.1 lU/mL as indicated by Grandjean et al.7
That is, the WHO (2018, 2017) cites Hendriksen et al. (1988) as "a toxin binding
inhibition (ToBI) assay has been reported and demonstrated to show good
correlation with the neutralization assay (correlation coefficient = 0.95)", and for
modified ELISA assays such as this further indicates [emphasis added] that, "The
minimum amount of circulating antibody that, in most cases, ensures immunity to
tetanus is assay-specific. Using in vivo neutralization tests or modified enzyme-
linked immunosorbent assays (ELISA), concentrations exceeding 0.01 lU/ml are
usually considered protective, whereas antibody concentrations of at least 0.1-0.2
lU/ml are defined as protective when using standard ELISA techniques." WHO
(2018) also discusses and illustrates the timing of primary and booster vaccinations
and durations of protection in the context of the minimum putatively protective
6 A 2-fold increase in PFOS and PFOA concentrations at age 5 years was associated with odds ratios between 2.38 (95%
CI, 0.89 to 6.35) and 4.20 (95% CI, 1.54 to 11.44) for falling below a clinically protective level of 0.1 lU/mLfor tetanus
and diphtheria antibodies at age 7 years.
7 For example, Grandjean et al. (2012) refers to a "clinically protective level of 0.1 lU/mL" several times and further
states, "An antibody concentration greater than 0.1 lU/mL is considered an important indicator of protection in
accordance with the public health rationale for routine vaccinations."
53
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
level of 0.01 IU/mL.8 Thus, the protective level cited by Grandjean et al. for the
assay used is 10-fold higher than the protective level cited by WHO (2017), even
further calling into question the biological/clinical significance and adversity of the
reported results.
In regard to confounding and the ability to causally attribute associated effects to
PFDA:
• Grandjean et al. (2012) state [emphasis added], "Although all of the 5 PFCs
measured showed negative associations with antibody levels, the
overlapping confidence intervals and the lack of comparative toxicology
studies prevent inference in regard to causal attribution... PFOS (most likely
the linear isomer) and PFOA appear to be the main culprits."
• The more recent Grandjean et al. (2017a) study states [emphasis added],
"Owing to the intercorrelations between the serum PFAS concentrations,
further analysis of the possible role of individual PFASs was not pursued, and
the observed associations may reflect the effects of the PFAS mixtures."
• Similarly, Grandjean et al. (2017b) state [emphasis added], "The close
correlations prevented meaningful adjustment for concomitant PFAS
exposures."
Thus, it appears that effects may neither rise to the level of adversity nor be
attributable specifically to PFDA. Co-exposures to other PFAS (at a minimum) that
are not/cannot be adequately accounted for in the analyses are likely to be
significant confounders in these epidemiological studies, especially because PFAS
exposures are correlated, they are chemically-similar compounds, and there
appears to be little variation in exposure (i.e., low exposure contrasts) for the single
PFAS being assessed (e.g., Table 2 of Grandjean et al. 2012, Table 1 of both
Grandjean et al. 2017a and 2017b). For example, Grandjean et al. (2012) shows that
PFOA and PFOS had a correlation coefficient of 0.5 in the blood sera of 5-year olds
and interquartile range (IQR) differences in blood sera concentrations of less than
1.6-fold each (e.g., 75th percentile blood concentration of PFOA/25th percentile
blood concentration of PFOA), and PFDA and PFNA had a correlation coefficient of
0.78 in the blood sera of 5-year olds and IQR differences in blood sera
concentrations of less than 1.9-fold each (see Table 2 of the study).
Despite Grandjean et al. (2017b) stating that the close correlations prevented
8 WHO (2018; pp. 14-15) states [emphasis added] that 'To illustrate the kinetics of immunity among children >1 year,
adolescents and adults following primary and booster vaccination with TTCV, Figure 2 provides a schematic
diagram of the typical response. A single dose of TT in the absence of priming induces little, if any, protection. Two to
four weeks after the second dose, the mean level of tetanus antitoxin typically exceeds the minimum putatively
protective level of 0.01 IU/mL. One year after the second dose, the mean antibody levels are expected to decline and
may fall to the protective threshold level. After each subsequent dose of vaccine, immunity is boosted, then persists
above the protective threshold for a time, and then wanes over time. Putatively protective levels of immunity are
induced by a primary series of three TTCV doses and immunity typically persists for at least 5 years. After the third
dose, each additional booster dose given after at least a one-year interval increases tetanus antitoxin levels and further
prolongs the duration of immunity. Immunity may persist for approximately 10 years after the fourth dose of TTCV and
for at least 20 years after the fifth dose."
54
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
meaningful adjustment for concomitant PFAS exposures, Budtz-J0rgensen and
Grandjean (2018a) attempts to do just that for benchmark dose, and the results
appear to help demonstrate the effects of confounding co-exposures and/or the
inability to properly adjust for them. Table 2 of Grandjean et al. (2017b) reports
the change (in percent) of the pre-booster serum-antibody concentrations at age
5 years associated with a doubling of the serum concentration of PFDA. Results
for cohorts 5, 3, and joint results are in the negative direction for tetanus at age 5
but none are statistically significant, which already points to the unreliability of an
effect having been demonstrated by these results and the unreliability of any RfD
based on these results or association. Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports benchmark results for the five prenatal PFAS concentrations in
regard to antibody concentrations at age 5 years (pre-booster) both unadjusted
and adjusted for PFOS/PFOA co-exposures. For tetanus antibodies, while
unadjusted benchmark doses for three models for PFDA (linear, piecewise,
conservative) appear to show excellent agreement (BMDs of 0.11-0.25 ng/mL)
with insignificant reliance on choice of model, when adjusted for PFOS/PFOA the
three models' best estimates of the PFDA serum concentrations associated with a
5% change go to infinity (although BMDLs can still be estimated). For diphtheria,
Table 2 of Grandjean et al. (2017b) reports statistically significant changes for
cohort 3 and joint results for the pre-booster serum-antibody concentrations at
age 5 years associated with a doubling of the serum concentration of PFDA.
However, similar to results for tetanus, Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports that when benchmark results are adjusted for PFOS/PFOA, two of
the three models' best estimates of the PFDA serum concentrations associated
with a 5% change go to infinity (although BMDLs can still be estimated). While
Table 1 of Budtz-J0rgensen and Grandjean (2018a; benchmark results for the age-
5 serum concentrations of five PFASs in regard to tetanus and diphtheria antibody
concentrations at age 7 years) provides no benchmark doses for PFDA that go to
infinity, Table 2 results point to the unreliability of this endpoint and should be
considered along with the other issues raised above. For example, the odds ratios
for PFDA and inadequate antibody concentrations for diphtheria and tetanus at 7
years was not statistically significant (see eTable 4 of Grandjean et al. 2012).
Key BMD results in the draft assessment itself demonstrate the importance of co-
exposures to other PFAS. Tables C-l and C-3 of the draft assessment (below)
provide BMD results for the critical effects used for RfD derivation (e.g., see Table
ES-1).
Table C-l. Results specific to the slope from the linear analyses of PFDA
measured in serum at age 5 years and log2(tetanus antibody concentrations)
measured at age 7 years in a single-PFAS model and in a multi-PFAS model
from fBudtz-lorgensen and Grandiean. 2018h1.
Exposure
Model
shape
PFOS&
PFOA
adjusted
Slope (P)
per ng/mL
in serum
SE(M
ng/m Lin
serum
Slope (P) fit
Lower bound
slope (plB) per
ng/mL in serum
PFDA at Age 5
Linear
No
-1.55
0.602
p = 0.01
-2.55
PFDA at Age 5
Linear
Yes
-0.98
0.681
p = 0.15
-2.10
55
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
PFOS and PFOA had correlation coefficients of 0.39 and 0.35 with serum PFDA at
age 5, respectively (Table 2 of Grandjean et al. 2012). Despite these relatively low
correlation coefficients (Mukaka 2012), Table C-l shows that just controlling for co-
exposures from these two PFAS (PFOS, PFOA) resulted in significant impacts on
slope ((B) and slope fit. The slope estimate for PFDA was reduced 37% and PFDA is
no longer a significant predictor of tetanus antibody concentrations (p=0.15). The
most likely explanation (Occam's razor) is classic confounding as PFOS and PFOA are
documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean 2018a),
and the existence of some chance that correction for these co-exposures could
create some confounding is not a scientifically robust justification for dismissing the
important implications of the results of adjustments for PFOS/PFOA that the study
authors themselves (Budtz-J0rgensen and Grandjean 2018a) thought it important to
adjust for, and with good reason. These results demonstrate the statistical
unreliability of serum PFDA predicting tetanus antibody concentrations when just
two other PFAS are controlled for. Table C-3 concerns the critical effect for the RfD
based on diphtheria antibody concentrations.
Table C-3. Results specific to the slope from the linear analyses of PFDA in
serum measured at age 5 years and logz (diphtheria antibodies) measured at
age 7 years from Table 1 in a single-PFAS model and in a multl-PFAS model
from fBudtz-lorgensen and Grandiean. 2018hl.
Exposure
Model shape
PFOS & PFOA
adjusted
Slope (P) per
ng/mL in
serum
SE(P)
ng/mL in
serum
Slope (P) fit
Lower bound
slope (Plb)
per
ng/mL in
serum
PFDA at Age 5
Linear
No
-0.894
0.561
p = 0.11
-1.82
PFDA at Age 5
Linear
Yes
-0.297
0.635
p = 0.64
-1.35
The implications of these results are worse. First, even when evaluated alone
without accounting for co-exposures to relatively low correlated PFOS and PFOA,
serum PFDA is not a significant predictor of diphtheria antibody concentrations
(p=0.11). Table C-3 further shows that controlling for co-exposures from these
two PFAS (PFOS, PFOA) resulted in significant impacts on slope ((B) and slope fit.
The slope estimate for PFDA was reduced 67% and serum PFDA became a worse
nonsignificant predictor of diphtheria antibody concentrations (p=0.64). The most
likely explanation (Occam's razor) is classic confounding as PFOS and PFOA are
documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean 2018a),
and the existence of some chance that correction for these co-exposures could
create some confounding is not a scientifically robust justification for dismissing
the important implications of the results of adjustments for PFOS/PFOA that the
study authors themselves (Budtz-J0rgensen and Grandjean 2018a) thought it
important to adjust for, and with good reason. Thus, when co-exposures are taken
into account for two modestly correlated PFAS (PFOS, PFOA), serum PFDA is not a
significant (i.e., reliable) predictor of these critical effects serving as the basis of
the RfD (i.e., decreases in serum tetanus and diphtheria antibody concentrations).9
9 Knowing this, disparate results are not particularly surprising, such as the 18.7% increase in tetanus antibodies
predicted for children (age 13) with a 2-fold increase in serum PFDA based on the same study and type of analysis used
for the RfD critical effects (see Table 3-12, p. 3-59).
56
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Moreover, Budtz-J0rgensen and Grandjean (2018a) also controlled for PFOS or
PFOA when deriving BMDs/BMDLs for the other, and EPA did use those co-
exposure-adjusted results for RfD derivation (i.e., PFOA adjusted for PFOS, PFOS
adjusted for PFOA; see Tables B-l and B-2 in Sections B.l.l and B.1.2 of the PFOA
and PFOS draft assessments USEPA 2021a,b) without expressing any similar
concerns about creating confounding by adjusting for these co-exposures
(low/moderate correlation coefficient of 0.50 (Mukaka 2012); see Table 2 of
Grandjean et al. 2012) or citing Weisskopf et al. (2018) and Weisskopf and
Webster (2017). Rather, these concerns have been selectively cited for PFDA in an
attempt to provide some rationale for dismissal of the co-exposure-controlled
results (e.g., slope ((B) values are reduced; serum PFDA is a nonsignificant
predictor of tetanus and diphtheria antibody concentrations) and thus for
selection of BMDLs uncontrolled for PFOS and PFOA co-exposures (see Tables C-2
and C-4 of the draft), but this is not a scientifically robust rationale and is
furthermore inconsistent with EPA's draft assessments for PFOA and PFOS (USEPA
2021a,b). Lastly, it is further noted that Weisskopf et al. (2018) indicates: (1)
sometimes, depending on causal structure, the inclusion of multiple exposure
variables in a model can amplify the amount of bias in a regression estimate
compared to analyzing single exposures; and (2) this potential amplification of
biases increases with stronger correlations between mixture components. To
demonstrate that this can occur in some cases, the study authors used "highly
correlated exposures" (e.g., r2=0.9), whereas the correlation coefficients between
PFDA and PFOS/PFOA are low (0.35 and 0.39, respectively). The results of
Weisskopf et al. (2018) do not constitute reasonable doubt that for these PFDA
results, the confounding from not adjusting for co-exposures to documented
immunotoxicants (PFOS, PFOA) is significantly greater than the potential
amplification of biases that remains undemonstrated under the same or similar
circumstances.EPA should reevaluate the issues raised above in regard to
implications for their draft PFDA assessment (Tier 1 necessary revision).
Regarding potential immunosuppressive effects by PFDA, effects that rise to the
level of adversity would be expected to result in increased incidences of disease,
reflecting lower immunity and lower resistance to disease in the real world.
However, consistent with the WOE for no statistically significant associations with
antibody concentrations falling below the generally cited protective level of 0.1
lU/mL based on results from Grandjean et al., almost all ORs in Table 3-13 of the
draft assessment include 1, indicating that the WOE from studies on PFDA and
infectious disease in humans is for no statistically significant associations.
Consistent with this, host resistance was unaffected by PFDA based on the limited
animal study data available (p. 3-73, lines 8-9), and host resistance assays are
considered highly relevant to the evaluation of immunotoxicity in the context of
human health assessment (p. 3-72, lines 15-18; IPCS 2012). Host. Again, EPA
should reevaluate the adversity of these presumed antibody level effects,
including the association with PFDA itself, and do so within the context of
potential confounding, other limitations, and available human/animal data on
disease incidence (Tier 1 necessary revision), as this has important implications
for the hazard judgment and the strength of human evidence descriptor for
immunosuppression (listed as "moderate" in Table 3-19).
57
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Moreover, it is noted that the PODs for immunosuppressive effects from these
epidemiological studies range from 2.57E-04 mg PFDA/L blood serum to 7.02E-04
mg PFDA/L blood serum (BMDLi/2sd values from Table 5-8, pp. 5-16 to 5-17), and
when intrahuman variability is considered (through application of a UFH of 10), the
resulting values range from 2.57E-05 to 7.02E-05 mg PFDA/L blood or 0.0257 to
0.0702 ng PFDA/L blood serum.10 Data from NHANES show that geometric means
(GMs) representative of the U.S. population are well above these blood serum
levels (see Appendix A to these comments).11 Most notably, 2005-2018 population
GMs range from 0.154-0.355 ng/L, which are 2.2- to 13.8-fold higher than the
PODs adjusted for intrahuman variability (cited above). Despite these
exceedances, tetanus and diphtheria appear to be quite rare in the U.S.
population. The average annual number of tetanus cases in the U.S. from 2009-
2018 was 29, with the CDC attributing most cases to individuals who either have
not been vaccinated or who are not current on their boosters (e.g., only 3% of the
cases from 2001-2008 were in people who had received a complete tetanus toxoid
series with the last dose within 10 years; Tiwari et al. 2021). Tetanus also appears
rare in U.S. children specifically, occurring primarily in older adults. Per Liang et al.
(2018):
"During 2001-2016, three neonatal tetanus cases and 459 non-neonatal tetanus
cases were reported to the National Notifiable Diseases Surveillance System
(NNDSS). The median age for non-neonatal cases was 44.0 years (range: 2-95
years)... The risk for both tetanus disease and mortality was higher among
persons aged >65 years than among persons aged <65 years. Tetanus occurs
almost exclusively among persons who are unvaccinated or inadequately
vaccinated or in those whose vaccination histories are unknown or uncertain."
The incidence of U.S. diphtheria cases is even more rare. The CDC reported only 14
cases from 1996 through 2018 (Acosta et al. 2021). Thus, consistent with the WOE
for the lack of statistically significant associations from the epidemiology study
data discussed above and despite the NHANES blood serum data showing
exceedances of the draft assessment human PODs adjusted for intrahuman
variability (e.g., toxicodynamic) for most of the U.S. population for a prolonged
period of time (see the 50th percentile concentrations in Appendix A), U.S.
surveillance disease incidence data are not supportive of adversity. That is, U.S.
surveillance disease incidence data do not support that serum PFDA (or any other
serum PFAS) is suppressing tetanus and diphtheria vaccine responses and leaving
people vulnerable to infection from these diseases.12
Additionally, it appears that EPA has not fully considered all the relevant evidence
or the weaknesses/limitations of the epidemiological evidence, which in turn are
relevant for the hazard judgment. Table 3-19 indicates that human data provide
10 Based on BMDi/2sd values (Table C-9, p. C-16), these values are 0.0385 to 0.226 ng PFDA/L blood serum.
11 See NHANES Biomonitoring Data Tables at https://www.cdc.gov/exposurereport/data tables.html. Budtz-J0rgensen
and Grandjean (2018a) also acknowledge that, "Our BMDL results, both before and after adjustment are generally
below current exposure levels..."
12 The apparent lack of adversity/ consequence for the effects reported for tetanus and diphtheria certainly does not
provide support for an expectation of adversity/consequence for other effects not measured/observed (e.g., for
vaccines for other diseases and their incidences).
58
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
"moderate" evidence and that "the inconsistent and low confidence evidence on
infectious disease did not influence this judgment." However, this points to the
fact that EPA has not duly considered the implications of the null findings on
human and laboratory animal infectious disease and other relevant considerations
(e.g., some discussed above) for the scientific WOE, which is not a scientifically
supportable approach as it does not consider all relevant data, directly relevant
human data in particular. EPA should consider such null findings (and other
relevant considerations) in their WOE (Tier 1 necessary revision). Combined with
the "slight" human data for sensitization and allergic response, the "slight"
laboratory animal data for immunosuppression, and the "indeterminate" animal
data for sensitization and allergic response (Table 3-19), it does not appear that
PFDA exposure is "likely to cause" adverse immune effects in humans is sufficiently
supported. EPA should reevaluate this determination (Tier 1 necessary revision) as
"may cause" might very well be the better supported hazard judgement, and also
reconsider their use of the serum antibody endpoints for quantitative risk
assessment/derivation of toxicity factors (Tier 1 necessary revision). This would
appear more consistent with the data discussed above and recent conclusions by
the Australian government (FSANZ 2021) and the U.S. Agency for Toxic Substances
and Disease Registry (ATSDR 2021).
The Australian government (FSANZ 2021) has concluded that associations of PFAS
with immunological endpoints do not provide a suitable basis for quantitative risk
assessment:
"In summary, new epidemiological studies provide some evidence of statistical
associations between PFAS blood levels and impaired vaccine response,
increased susceptibility to infectious disease and hypersensitivity responses.
However the data are insufficient to establish causal relationships and it cannot
be ruled out with reasonable confidence that the observed statistical
associations may have been due to confounding, bias or chance. On the basis of
the uncertainties and limitations in the evidence base, immunomodulation is
not currently considered suitable as a critical endpoint for quantitative risk
assessment for PFAS."
Similarly, ATSDR (2021) found the epidemiology literature inadequate for use as
the basis of deriving minimal risk levels (MRLs)for PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive targets
for many of the perfluoroalkyls; however, there are two major limitations to
establishing dose-response relationships for these effects and using the
epidemiological studies to derive MRLs: accurate identification of
environmental exposure levels producing increased risk for adverse effects
(exposure estimates and routes of exposure) and likely co-exposure to mixtures
of perfluoroalkyls. Other limitations include the cross-sectional design of the
majority of epidemiological studies and the potential that reverse causality
contributes to the observed associations... In summary, the epidemiological
databases for several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with adverse
effects and possible interactions between compounds preclude use of these
59
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
data to derive MRLs."
Based on the information in the draft assessment and reviewed elsewhere¦, this
reviewer finds it difficult to disagree with the recent conclusions of the Australian
government (FSANZ 2021) and ATSDR (2021) that the epidemiology literature (e.g.,
on PFAS blood levels and impaired vaccine response) is inadequate for
quantitative risk assessment and use as the basis for deriving toxicity factors (e.g.,
RfDs). Unreliable/inadequate data (human or otherwise) cannot be justified
scientifically for use in dose-response assessment by how the associated
unreliable POD(s) compares to that based on reliable data definitively showing
cause-and-effect for unquestionably adverse effects, unconfounded by significant
co-exposures to similar chemicals. All this is not to say that PFAS are incapable of
causing immune effects, but rather most importantly that the epidemiological
data are insufficient for dose-response assessment due to significant issues such
as (but not limited to) confounding by co-exposures to other PFAS that have not
been (and perhaps cannot be) adequately adjusted for, and as a consequence any
effects observed are mixture effects. A counter argument appears to be that real-
world exposures are to mixtures, but nevertheless, it is not scientifically defensible
or realistic to attribute the effects of a mixture of chemically-similar chemicals to a
single component (i.e., co-exposures to other components of the mixture
contributing to the observed effects would have to be able to be adequately
adjusted for).13 Again, EPA should reconsider their use of the serum antibody
endpoints for quantitative risk assessment and derivation of toxicity factors (Tier 1
necessary revision).
b.i. See the comments above, some of which concern confounding while others
concern important issues such as adversity, the WOE against statistically
significant associations with serum antibody concentrations below the
generally cited protective level, the overall WOE against PFDA-associated
immunosuppressive effects (i.e., increased disease incidence) in humans, and
other important considerations. Here is simply noted that associations are not
causation, and residual confounding being judged "unlikely to fully explain the
associations" is not a scientific criterion for presuming that: (1) associated
effects are in fact attributable to the one chemical exposure in question (i.e.,
confounding not being known to account for 100% of an association or effect
is not scientifically-defensible grounds to then causally attribute the mixture
effect of exposure to a group of similarly acting compounds to just one
component); or for presuming that (2) the dose-response data, without
causation being demonstrated and with residual confounding explaining some
unknown part of the association(s), are suitable for quantitative dose-
response assessment and derivation of toxicity factors (e.g., RfDs). The
13 For example, just as there are thousands of PFAS, there are numerous hydrocarbon components of gasoline that
people are exposed to as a mixture, and even though they number fewer than the number of PFAS, it still would not be
scientifically defensible to derive a toxicity factor for just one component, toluene for example, attributing the totality
of the mixture effects observed solely to toluene following exposure to gasoline (e.g., even if two co-exposures such as
ethylbenzene and xylenes were adjusted for).
60
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Australian government (FSANZ 2021) and ATSDR (2021) would seem to concur
(see the quotes provided above).
As an additional example, following review of the relevant literature, FSANZ
(2021) stated [emphasis added]:
"While these studies provide limited evidence of statistical associations, a
causal relationship between increased PFAS blood levels and impaired
vaccine response cannot be established with reasonable confidence. The
evidence for an association between increasing PFAS blood levels and
impaired vaccine response is insufficient for quantitative risk assessment
on the basis of substantial uncertainties and limitations including:
• the small number of studies and participants, and mostly cross-sectional
design of studies such that conclusions around causality should be drawn
with caution.
• limited dose-response information with most studies investigating a
narrow range of blood levels associated with background levels of PFAS
exposure.
• inconsistency in antibody response to vaccines between different PFAS
congeners which cannot explained by study design.
• potential for confounding by other known environmental
immunotoxicants such as PCBsfor which inverse associations with blood
serum antibody concentrations against tetanus and diphtheria have
previously been reported in the child populations living in the Faroe islands
(Heilmann et al. 2010).
• uncertainty about the clinical relevance, if any, of the observed statistical
associations to susceptibility to infectious disease."
FSANZ (2021) adds that this conclusion is consistent with the recent decisions
of the German Human Biomonitoring Commission (Holzer et al. 2021;
Schumann et al. 2021), ATSDR (2018, 2021), and a number of earlier opinions
from national agencies and bodies such as Danish EPA (2016), Expert Health
Panel for PFAS (2018), and Kirk et al. (2018). See FSANZ (2021) for references.
Furthermore, the importance of the fact that the study authors themselves
acknowledge the inability to causally attribute associated effects to PFDA
cannot be overstated:
• Grandjean et al. (2012) state [emphasis added], "Although all of the 5
PFCs measured showed negative associations with antibody levels, the
overlapping confidence intervals and the lack of comparative toxicology
studies prevent inference in regard to causal attribution... PFOS (most
likely the linear isomer) and PFOA appear to be the main culprits."
• The more recent Grandjean et al. (2017a) study states [emphasis
added], "Owing to the intercorrelations between the serum PFAS
concentrations, further analysis of the possible role of individual PFASs
61
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
was not pursued, and the observed associations may reflect the effects
of the PFAS mixtures."
• Similarly, Grandjean et al. (2017b) state [emphasis added], "The close
correlations prevented meaningful adjustment for concomitant PFAS
exposures."
It appears that EPA may not have fully considered the implications of these
study author-cited limitations for use of the data for quantitative risk
assessment and derivation of toxicity factors (e.g., RfDs).
Considering my comments under this subsection and "2.b" above, I seem to
concur with the Australian government (FSANZ 2021) and ATSDR (2021)
that the epidemiology literature on PFAS (e.g., PFDA) blood levels and
impaired vaccine/antibody response is inadequate for dose-response
assessment and deriving toxicity factors (e.g., RfDs). Part of this shared
opinion is based on confounding considerations which are the subject of
this question subsection (2.b.i). Unreliable/inadequate data (human or
otherwise) cannot be justified scientifically for use in dose-response
assessment by how the associated unreliable POD(s) compares to that
based on reliable data definitively showing cause-and-effect for
unquestionably adverse effects, unconfounded by significant co-exposures
to similar chemicals. As indicated above, EPA should reconsider their use of
the serum antibody endpoints for quantitative risk assessment/derivation
of toxicity factors (Tier 1 necessary revision).
c. Developmental effects: For the most part, the available data on developmental
effects are clearly and appropriately synthesized to describe the strengths and
limitations. P. 3-158, lines 18-25 state that, "Taken together, the available evidence
indicates that PFDA exposure is likely to cause developmental toxicity in humans
given sufficient exposure conditions (see Table 3-24). This conclusion is based
primarily on findings of dose-dependent decreases in fetal weight in the only
available toxicology study, with mice gestationally exposed to PFDA doses >0.5
mg/kg-day and supported by evidence of decreased birth weight from studies of
exposed humans in which PFDA was measured during pregnancy, primarily with
median PFDA values ranging from 0.11 to 0.46 ng/mL. The conclusion is further
supported by coherent epidemiological evidence for biologically related effects
(e.g., decreased postnatal growth and birth length)." There are comments below
concerning the coherence or incoherence of key epidemiological data for birth
weight (e.g., Wikstrom et al. 2020, Bach et al. 2016, and other studies). Table 3-24 is
the evidence profile table for developmental effects. As quoted above, EPA's hazard
conclusion for developmental effects primarily relies on in vivo animal data (i.e., the
mouse study of Harris and Birnbaum 1989), for which Table 3-24 contains factors
that increase certainty for fetal growth but no factors that decrease certainty. Based
on over 45 different epidemiological studies included in the draft assessment, the
evidence of an association between PFDA exposure and developmental effects in
humans is considered only "slight" (p. 3-156, lines 10-11). Despite the "slight"
evidence in humans, Table ES-1 indicates that RfDs were nevertheless calculated
based on decreased birth weight in male and female children (Wikstrom et al.
62
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
2020). While data in humans, as the species of ultimate interest, are usually
preferred as the basis for derivation of toxicity factors, in this case it appears that
the mere "slight" totally of the evidence for developmental effects from the over 45
different epidemiological studies included in the draft assessment should not be
considered sufficient for quantitative dose-response assessment (e.g., birth weight),
but rather only for potentially supportive information for hazard identification. EPA
should reconsider their use of these epidemiological data for quantitative risk
assessment/derivation of toxicity factors (Tier 1 necessary revision). EPA Figure 3-
28 (p. 3-108), reproduced below, helps illustrate the weakness and incoherence of
the overall epidemiological database for demonstrating decreased birth weight,
particularly statistically significantly decreased birth weight, associated with PFDA
levels and how no one epidemiological study could possibly be representative of
these inconsistent and disparate results or the results of a meta-analysis of relevant
studies.
63
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
frirtlf
<%*•.*#»>* jJCi'+JtnKLrn* S*
:**•<•>* i'.s. r-nj
ttlf. 111
2**51 oSmhv SSTiir
r m PO>^|iW^<\htK»i
w»:nwnv< k,—¦ .«•»
K^uaM.»um
i» .»* Cni>-1'
r'« muta'XiKMiiwitT. ftu
[.JUi i'.lKll; au
0(1
!>v»a O
(MOW C<
Cktotw
u Wl-W";*!.)!.
Vilxiiiit. CvMnlwn KnuntN-/ M*
aat uniw c H*r ii.-v
BM HMfe (f»». « >4
MT«UM —h
~4a*»»K wn?B»*'\At*HllP Pi-t.
Otl(N« Cw
r, >rty«K. I'ltM, i)
Figure 3-28. Overall study population mean birth weight results for 22 PFDA
epidemiological studies*-®, [results can be viewed by clicking the HAWC link).
Abbreviation: BWT = Birth Weight
*- Studies are sorted first by overall study confidence level then by Exposure Window examined.
b- Mene et al. l'201Si pooled samples from umbilical cord blood and maternal plasma during the first and second
trimesters. The remaining studies were all based on either one umbilical or maternal sample.
L If a study presented regression coefficients for continuous exposure with multiple exposure units, only one unit
change is shown (e.g., (Bach et al. 2016 i. with the exception of [Li et a I.. 2017). which displays both IQ.R and In-
unit (ng/mL) values.
"• The results displayed here for mean birth weight among 587 overall population participants in the POPUP Cohort
are from a larger population of participants [Swedish Environmental Protection Agency. 2017) compared to a
sample size of 381 m their 2018 publication Gvilenhammar et al. (20181.
' Xu et al. (2019a|i results are truncated for the 210.7 gram increase; the complete 95% CI ranges from -314.3 to
735.8 grams.
Similarly, Figure 3-26 (p. 3-104) below illustrates how the results of Bach et al.
(2016), the other high confidence prospective cohort for trimester 1-2 with
adequate study sensitivity, had a nonmonotonic response, no statistically
significant quartile results for the overall population, and a regression coefficient
of +0.03 per each In-unit (ng/mL) (compared to -0.147 per each In-unit increase
for Wikstrom et al. 2020), all of which does not support the Wikstrom et al. (2020)
study results or their use in toxicity factor derivation. The Bach et al. (2016) and
Wikstrom et al. (2020) studies are the first two entries in Figure 3-26, respectively.
64
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Rsgr»4si«i coefflctom
SEl <2007-20101 i
Ounmi8 4
Ifrunl (ngimL)
•0 065 Quartile 3
-0179 Quartile 4
¦o 14? iiHmcnymL)
3 «w«.5 POM-BIW
moiha-irfaniBafS
Canadian (tealitiy i>i'«»1 LoiUludhaf
Development (CHILD) Study
(2010-2012). Canada. 414
• eichwiBs n wannarazBO BWT1
© t raisntie n stsrcorsizea SWT] c
M 95% eortlOoncc i ntcrvM
Figure 3-26.PFDA and birth weight z-scores (overall population)". Refer to
Birth Weight-Z for details on the individual study evaluation review.
Moreover, the BMD5RD values for decreased birth weight based on Wikstrom et al.
(2020) range from 0.54 to 0.71 ng/mL (note that BMD5RD values are not treated
like NOAELs for purposes of RfD derivation; Table C-10, p. C-22).14 By comparison,
PFDA blood serum concentrations in quartile 4 of the Bach et al. (2016) study
range from 0.43 to 2.87 ng/mL (see Table 1 of the study). Despite these quartile 4
blood serum concentrations being up to 5.3-fold higher than the BMD5rd values
for decreased birth weight based on Wikstrom et al. (2020), birth weight and birth
weight z-scores for quartile 4 in Bach et al. (2016) were actually increased
(nonsignificantly) (see Table 3 of the study). Again, this is from the other high
confidence, prospective cohort for trimester 1-2 with adequate study sensitivity.
These results from Bach et al. (2016), obtained from the other high confidence
study in the same PFDA blood serum concentration range (and higher) compared
to EPA's BMD5rd values, are inconsistent with and unsupportive of EPA's use of the
PODs for adverse birth weight effects based on Wikstrom et al. (2020) for RfD
derivation. Sex-specific results from Bach et al. (2016) do not provide strong
support for use of Wikstrom et al. (2020) results for quantitative dose-response
assessment either (e.g., Figure 3-29, p. 3-109 of the draft assessment). Other
important comments on Wikstrom et al. (2020) appear under question "2.c.i"
below (e.g., confounding).
Given the drastically different results across epidemiological studies and the mere
"slight" evidence of developmental effects across more than 45 such studies, EPA
should reconsider use of a single epidemiological study (Wikstrom et al. 2020) for
dose-response assessment of birth weight and RfD derivation and consider a
meta-analysis and/or using the more definitive dose-response data from the
mouse study for dose-response assessment of birth weight and RfD derivation
(Tier 1 necessary revision). It is remarkable that per EPA, the single mouse study
by Harris and Birnbaum (1989) gives rise to a greater level of evidence for
developmental effects ("moderate") than the results of over 45 epidemiological
studies. This fact alone, in this case, justifies use of the animal data for dose-
response assessment and RfD derivation. Based on EPA Figure 3-56 (p. 3-152),
14 For comparison, the BMDL5rd values range from 0.31 to 0.37 ng/mL (Table C-10, p. C-22).
65
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
reproduced below, the dose-response data for fetal body weight (as an example)
appear well suited for benchmark dose analysis.
Animal Description
Fatal Body Weight Harris. 1986. 3858729 High confidence Gestational Oral (GO 10-13) F1 Mouse. C57BU6n ("Z.) signiScant litter
PFDA Fetal Body Weight
Gestational Oral (GO 6-15)
% Statistically significant
% Percent control response
Figure 3-56. PFDA fetal body weight after gestational exposure (results can be viewed by clicking the HAWC link).
More specifically, Figure 3-56 shows statistically significant and progressive
decreases/trends in fetal body weight with dose across a range that includes the
>5% decrease historically used by regulatory agencies as the demarcation of
adversity in dose-response assessment (i.e., as a >5% reduction in fetal body
weight is usually considered adverse, a benchmark response of 5% is typically used
in benchmark dose analysis), from -1% to -50% and including doses that resulted
in decreases of -4% and -6% (Table 3-23, p. 3-153). Furthermore, the changes in
fetal body weight were at doses not associated with maternal toxicity (p. 3-151,
lines 5-6).
Harris and Birnbaum (1989) is a medium/high-confidence study (p. 3-150, lines 9-
12), and EPA should consider mouse data from the Harris and Birnbaum (1989)
study for use as the primary basis for osRfD development (Tier 1 necessary
revision) based on developmental growth effects consistent with: (1) this mouse
study providing the primary basis for EPA's developmental effects hazard
conclusion; (2) the lack of factors that decrease certainty for fetal growth as
evaluated in the mouse study (see Table 3-24); and (3) the extensive
epidemiological database merely being able to provide support for this mouse
study with what amounts to "slight" evidence for developmental effects across
over 45 epidemiological studies. Again, per the draft assessment, it is the mouse
data that primarily support that assuming sufficiently high exposure over a
sufficiently long duration (i.e., "given sufficient exposure conditions"), PFDA
exposure is likely to cause developmental toxicity in the general human
population, which includes potentially susceptible subpopulations (e.g.,
developing fetuses of pregnant women). Moreover, unreliable/inadequate data
(human or otherwise) cannot be justified scientifically for use in dose-response
assessment by how the associated unreliable POD(s) compares to that based on
reliable data definitively showing cause-and-effect for unquestionably adverse
effects, unconfounded by significant co-exposures to similar chemicals. So all this
is not to say that PFAS are incapable of causing developmental effects, but rather
most importantly that the epidemiological data are insufficient for dose-response
assessment due to significant issues such as (but not limited to) confounding by
co-exposures to other PFAS that have not been (and perhaps cannot be)
66
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
adequately adjusted for, and as a consequence any effects observed are mixture
effects. A counter argument appears to be that real-world exposures are to
mixtures, but nevertheless, it is not scientifically defensible or realistic to attribute
the effects of a mixture of chemically-similar chemicals to a single component (i.e.,
co-exposures to other components of the mixture contributing to the observed
effects would have to be able to be adequately adjusted for).15
c.i. I wholly agree that the strength of the available human evidence is "slight",
at least partly due to potential confounding. Given the slight human
evidence, above I suggest use of the mouse data (Harris and Birnbaum
1989) instead of epidemiological study data (Wikstrom et al. 2020) for dose-
response assessment and osRfD derivation based on developmental growth
effects (Tier 1 necessary revision). To do so would make EPA's assessment
somewhat more consistent with ATSDR (2021), who found the
epidemiology literature inadequate for use as the basis of deriving MRLs for
PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive
targets for many of the perfluoroalkyls; however, there are two major
limitations to establishing dose-response relationships for these effects
and using the epidemiological studies to derive MRLs: accurate
identification of environmental exposure levels producing increased risk
for adverse effects (exposure estimates and routes of exposure) and
likely co-exposure to mixtures of perfluoroalkyls. Other limitations
include the cross-sectional design of the majority of epidemiological
studies and the potential that reverse causality contributes to the
observed associations... In summary, the epidemiological databases for
several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with
adverse effects and possible interactions between compounds preclude
use of these data to derive MRLs."
To be entirely consistent with this conclusion from ATSDR (2021) and my own
comments and conclusions under question 2.b (above), the EPA assessment
would further need to omit the highly uncertain RfD derivations based on
PFDA associations with decreases in antibody levels/vaccine response (Tier 1
necessary revision) reported in some epidemiological study results that
appear overall to be inconsistent, non-adverse, and uncertain in nature (e.g.,
causal attribution problems, unsupported by epidemiological disease
incidence studies).
Getting back to use of Wikstrom et al. (2020) for dose-response assessment of
birth weight and RfD derivation, in addition to important comments above
under "2.c" (e.g.., inconsistency of Bach et al. 2016 with Wikstrom et al. 2020
15 For example, just as there are thousands of PFAS, there are numerous hydrocarbon components of gasoline that
people are exposed to as a mixture, and even though they number fewer than the number of PFAS, it still would not be
scientifically defensible to derive a toxicity factor for just one component, toluene for example, attributing the totality
of the mixture effects observed solely to toluene following exposure to gasoline (e.g., even if two co-exposures such as
ethylbenzene and xylenes were adjusted for).
67
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
and the weakness/incoherence of the overall epidemiological database for
demonstrating decreased birth weight, particularly statistically significant
decreases), Tables 2 and 3 from that study appear below (unaltered; see
license at https://creativecommons.Org/licenses/bv/4.0/) and demonstrate:
(1) significant co-exposures to other PFAS (e.g., PFOS, PFOA) associated with
birth weight deficits, an ATSDR concern and significant uncertainty; and (2)
few statistically significant ORs for decreased body weight in the PFDA
exposure quartiles; that is, only 2/9 of the ORs showed statistically significant
birth weight decrements, which were for quartile 4 where both PFOS and
PFOA were certainly co-exposures and also showed statistically significant
decreases for birth weight. In regard to this latter point concerning potential
confounding by co-exposure to other PFAS, which was cited by ATSDR as a
significant uncertainty, Wikstrom et al. state [emphasis added]:
"Another limitation is the compound-by-compound approach.
Theoretically, a health outcome is simultaneously influenced by multiple
environmental factors. Nevertheless, the exposure to several PFASs may
be correlated with each other due to common sources. Our findings were
consistent across different PFAS compounds, and we regard correction for
multiple comparisons overly conservative to be suitable for the
investigations on such interrelated compounds. If the single compound's
level could represent the levels of several other compounds, our findings
based on single compound analyses may still shed some light on the joint
effects of multiple PFAS compounds. However, carefully designed
statistical models, such as mixture-based approaches within the PFAS
compound class, should be explored in follow-up studies."
Thus, the study authors are acknowledging confounding by other PFAS as an
important limitation of their compound-by-compound approach as joint effects
of multiple PFAS compounds may be occurring such that mixture-based
approaches should be explored for data analysis. The compound-by-compound
approach used in this study gives rise to significant uncertainty that precludes
use of these data for dose-response assessment. It cannot be confidently said
scientifically that the presumed effects are due to PFDA exposure (e.g., PFOS
and PFOA were certainly co-exposures and also showed statistically significant
decreases for birth weight). Consistent with my opinion on the significant
uncertainty and confounding associated with these Wikstrom et al. results and
the weakness/limitations of the epidemiological evidence overall, Section F.3
of the draft assessment states that, "In the six studies using mutually adjusted
PFAS approaches to address coexposures, there was not consistent evidence
for birth weight deficits associated with increased exposure to PFDA." (p. F-27,
lines 13-14), and acknowledges that "there is considerable uncertainty due to
potential confounding by co-occurring PFAS in the existing literature." (p. F-27,
lines 30-32). Indeed, significant co-exposure to multiple PFAS is not the
exception but the rule, and is just the condition to result in significant bias
68
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
away from the null for adverse effects.16 The high confidence study of Luo et
al. (2021) is an example that serves as a cautionary tale on the obvious
importance of PFAS co-exposures, which reported large statistically significant
birth weight deficits (-97 g; -178, -16 per each In-unit PFDA increase) in a
single-pollutant PFDA model, but results were null and their direction reversed
in the multipollutant model with a nonsignificant increase in birth weight
associated with PFDA (Table F-2).17
As the data from epidemiological studies provide only "slight" evidence of
developmental effects, including data from Wikstrom et al. (2020) that appears
too uncertain and unsuitable for quantitative dose-response assessment, I
suggest use of the mouse data (Harris and Birnbaum 1989) for dose-response
assessment and osRfD derivation based on developmental growth effects (Tier
1 necessary revision). Unreliable/inadequate data (human or otherwise)
cannot be justified scientifically for use in dose-response assessment by how
the associated unreliable POD(s) compares to that based on reliable data
definitively showing cause-and-effect for unquestionably adverse effects,
unconfounded by significant co-exposures to similar chemicals.
Tabic 2, Prenatal exposure to eight PFAS, measured
as maternal serum concentrations (ng/mL) during early pregnancy.
Compound
Geometric mean 195% CI)
Mu jian Hi 'I
95th %
LOD
Above LOD {%>
PFOS
535 !521-5.501
538 (3.97-7.60:»
1034
0.06
100
PFQA
I &l< 11 5t» 1 bS 1
1 M ! II 1 _ "Ml
3.18
0.02
100
PFHxS
131 II10 1 35 i
1 2" i I r 1 44
2.94
0.03
100
PFNA
054 1053 0
II1". i U"""
1.02
0.01
100
PFDA
01b PZ*j 0 I"
! J- \ 1! 1 Q 1 4
050
0.02
100
PFUnDA
nn |011 0 221
0.44
0.02
995
PPHpA
0 -MHO T 0.019;
i L TD u V
0.077
0.01
73.9
PFDoDA
0 02" in 0>J 0.027)
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Table 3. Associations between prenatal PFAS exposure3 and birth weight6, together with odds ratios for birth weight small for gestational age in
1533 children.
All children
Girls
Boys
BW (g)
SGA
BW (g)
SGA
BW (g)
SGA
p (95% CI)
OR (95% CD
p (95% a)
OR (95% O)
& (95% CI)
OR (95% CO
PFOS
Per In-unit
-46 ( 88: -3)
1.19 (0.87; 1.64)
-85 (-145; 25)
1.40 (0.83; 235)
-13 (-73; 47)
1.08 (0.72 1.63)
Q1
Reference
Reference
Reference
Reference
Reference
Reference
02
-27 ( 89; 35)
0.69 (0.43
1.08)
- 32 (-115; 52)
0.89 (0.39; 203)
28 (-118; 63)
1.26 (0.67; 237)
Q3
-22 (¦ 84; 41)
0.79 (0.53
1.18)
-51 (-137; 34)
0.82 (0.36; 2.03)
5 (-86; 96)
0.86 (0.45; 1.67)
04
PFOA
Per In-unit
-80 ( 144; -16)
1.56 (1.09
2.22)
-142 (-231; -54)
2.05 (1.00; 4.21)
28 (-119; 63)
1.30 (0.70; 240)
-68 (-112; 24)
1.43 (1.03, 1.99)
-86 (-145; 26)
1.96 (1.18; 3.28)
49 (-113; 15)
1.16 (0.75; 1.78)
Q1
Reference
Reference
Reference
Reference
Reference
Reference
02
27 (- 35; 89)
0.77 (0.45
1.32)
30 (-55; 115)
1.00 (0.40; 251)
26 (-66; 116)
0.67 (0.34; 1.31)
03
-41 ( 106; 23)
0.96 (0.57
1.61)
36 (-124; 52)
1.64 (0.71; 3.83)
44 (-139; 50)
0.66 (0.33; 1.29)
04
PFHxS
Per In-unit
90 ( 159; 91)
1.44 (0.86
2.40)
136 ( 231; 40)
233 (1.00; 5.43)
47 (-147; 54)
1.04 (0.54; 201)
-0.1 ( 38; 38)
0.96 (0.72; 1.27)
-14 ( 68; 39)
1.14 (0.73; 1.80)
-13 (-67; 41)
0.84 (0.58; 1.22)
01
Reference
Reference
Reference
Reference
Reference
Reference
02
-4 (- 66; 58)
1.37 (0.86
2.20)
30 (-56; 116)
1.77 (0.78; 3.99)
39 (-129; 50)
1.24 (0.69; 223)
03
-15 (-78; 48)
0.89 (0.54; 1.47)
28 (-59: 115)
1.05 (0.44; 249)
-51 (-141; 39)
0.82 (0.44; 1.54)
04
PFNA
Per In-unit
6 ( 69; 57)
1.04 (0.63; 1.69)
-16 (-104; 71)
1.76 (0.79; 3.90)
1 (-90; 92)
0.73 (0.38; 1.41)
-46 (-89; -4)
1.38 (1.02, 1.87)
-52 (-117; -2)
1.34 (0.85; 211)
50 (-113; 14)
1.42 (0.94; 217)
01
Reference
Reference
Reference
Reference
Reference
Reference
02
7 (-55; 69)
0.83 (0.49
1.38)
-2 ( 86; 82)
0.66 (0.29; 1.52)
15 (76; 106)
0.97 (0.50; 1.89)
03
39 (102; 24)
1.14(0.70
1.85)
-49 (-137; 38)
1.39 (0.66; 290)
28 (-119; 64)
0.95 (0.50; 1.82)
04
PFDA
Per In-unit
-33 ( 96; 31)
1.23 (0.77
1.99)
66 (-153; 20)
1.22 (0.59; 253)
1 (-94; 95)
1.24 (0.66; 233)
58 ( 103;-13)
1.46 (1.06, 2.01)
-69 (-133; -6)
1.62 (0.98, 267)
47 (-11217)
1.36 (0.90; 2.07)
01
Reference
Reference
Reference
Reference
Reference
Reference
02
23 ( 85; 39)
1.03 (0.62; 1.69)
-42 (-126; 42)
0.86 (0.37; 200)
-8 (-99; 82)
1.18 (0.63; 223)
03
39 ( 101; 23)
1.07 (0.65; 1.76)
-74 (-160; 13)
1.20 (0.54; 267)
-8 ( 98; 81)
0.99 (0.52 1.89)
04
-69 (-132; -5)
1.50 (0.94; 2.38)
-116 ( 204; 27)
1.95 (0.94; 4.06)
-27 (-118; 64)
1.21 (0.66; 223)
PFUnDA
Per In-unit
-13 (- 49; 22)
1.21 (0.92; 1.58)
-24 (-75; 27)
1.08 (0.70; 1.67)
-6 (-55; 42)
1.29 (0.82; 1.83)
01
Reference
Reference
Reference
Reference
Reference
Reference
02
-67 (-153; 19)
1.24 (0.77; 2.01)
-67 (—153;19)
2.09 (0.94; 4.63)
41 (-48; 130)
0.90 (0.48; 1.68)
03
-42 ( 128; 44)
0.85 (0.51
1.43)
-42 (-128; 44)
1.01 (0.42; 2.44)
46 (-44; 136)
0.81 (0.43; 1.56)
04
-46 (-110; 17)
1.52 (0.95; 2.44)
-93 (-183; 3)
1.92 (0.86; 4.25)
-10 (-100; 80)
1.36 (0.76; 246)
PFHpA
Per In-unit
-1 (-24; 21)
1.06 (0.90; 1.25)
4 (-34; 26)
1.06 (0.83, 1.36)
0.03 (-33; 32)
1.07 (0.86; 1.37)
01
Reference
Reference
Reference
Reference
Reference
Reference
02
0 ( 62; 61)
0.92 (0.57; 1.48)
17 (-70; 103)
0.92 (0.44; 1.92)
-15 (-102 73)
0.94 (0.50; 1.75)
03
3 (~ 59; 65)
0.78 (0.48; 1.27)
21 (-106; 65)
0.75 (0.35; 1.65)
23 (-65; 112)
0.78 (0.42; 1.45)
04
31 (-31; 93)
1.25 (0.85; 1.84)
27 (-57; 112)
1.31 (0.66; 260)
33 (-58; 124)
1.15 (0.65; 204)
All analyses were adjusted for maternal weight, parity (three categories) and cotinine levels. Analyses including
for sex and analyses of BW were adjusted for GA. SGA was defined as BW 10th percentile for sex and GA
^Associations with PFAS are presented per In-unit and by quartiles of exposure, as related to
bBirth weight (g) and odds ratios (adjusted) for birth weight small for gestational age
both boys and girts were in addition adjusted
d. Male reproductive effects: The studv confidence conclusions appear scie
ntifically
justified, giving appropriate consideration to important study attributes. For each
study in the relevant figures of Section 3.2.4
e.g., Figures 3-58 to 3-61), the
consideration ratings appear overall consistent with (i.e.,
provide the scientific
justification needed for) the overall study confidence level rating. Accordingly, the
study confidence conclusions appear scientifically justified, with EPA having given
appropriate consideration to important study attributes.
This question (under
"2") also concerns whether the presentation and
analysis of study results is clear, appropriate, and effective, a step just
upstream of synthesis that if well done, then allows for appropriate
syntheses of the endpoint-specific findings. Although 1 have two comments
70
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
below, the presentation and analysis of study results appears clear,
appropriate, and effective overall. Additionally, it appears that generally,
the available data on male reproductive effects are clearly and
appropriately synthesized to describe the strengths and limitations. P. 3-
185, lines 17-18 state that, "Taken together, available evidence indicates
that PFDA is likely to cause male reproductive effects in humans under
sufficient exposure conditions (see Table 3-26). This conclusion is based
primarily on a constellation of coherent evidence from a high confidence
study in animals exposed to 0.625-2.5 mg/kg-day for 28 days, with some
support for biological plausibility provided by mechanistic evidence from
i.p. and cell culture models. Although no direct information on the human
relevance of the animal evidence is available, many aspects of the male
reproductive system are conserved across species, and the limited
sensitivity in human studies may explain the lack of associations observed."
Table 3-26 is the evidence profile table for male reproductive effects, which
among other information contains both factors that increase certainty and
factors that decrease certainty along with evidence stream (i.e., human,
animal, mechanistic/supplemental) judgments/rationales and a summary
judgment. Obviously, the text of the document (Section 3.2.4) also contains
information relevant to and supporting the WOE for male reproductive
effects.
My comment regarding presentation and analysis concerns p. 3-183, lines
19-22 that state, "The body weight reductions in male rats observed in the
28-day gavage study at 1.25-2.5 mg/kg-day are consistent with moderate
body weight changes (21-38%) that are not associated with confounding
effects from overt systemic toxicity in supplemental studies tailored to
examine that potential linkage." While described as "moderate" body
weight changes, reductions of 21-38% are well above those typically
considered adverse (e.g., >10% in adult animals) and would be associated
with doses above the maximum tolerated dose (MTD) and overt toxicity. As
such, the draft assessment should further elaborate upon, explain, and/or
clarify the statement cited above (Tier 1 necessary revision).
Lastly, in regard to "no direct information on the human relevance of the
animal evidence" being available, Tier 1 necessary revision comments
under question "2.a" on the most human-relevant laboratory animal
species also apply here for male reproductive effects.
e. Female reproductive effects: The study confidence conclusions appear scientifically
justified, giving appropriate consideration to important study attributes. For each
study in the relevant figures of Section 3.2.5 (e.g., Figures 3-66 and 3-67), the
consideration ratings appear overall consistent with (i.e., provide the scientific
justification needed for) the overall study confidence level rating. Accordingly, the
study confidence conclusions appear scientifically justified, with EPA having given
appropriate consideration to important study attributes.
This question (under "2") also concerns whether the presentation and
analysis of study results is clear, appropriate, and effective, a step just
71
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
upstream of synthesis that if well done, then allows for appropriate
syntheses of the endpoint-specific findings. Although I have three
comments below, the presentation and analysis of study results appears
clear, appropriate, and effective overall. Additionally, it appears that
generally, the available data on female reproductive effects are clearly and
appropriately synthesized to describe the strengths and limitations. P. 3-
204, lines 12-19 state that, "Taken together, the available evidence
indicates that PFDA is likely to cause female reproductive toxicity in
humans under sufficient exposure conditions (see Table 3-29). This
conclusion is based primarily on evidence from a high confidence study in
rats exposed to doses ranging from 1.25-2.5 mg/kg-day PFDA for 28 days.
The PFDA-induced disruption of estrous cyclicity observed in female rats
from the NTP study (NTP, 2018) and its implications for infertility can be
considered relevant to humans given that the mechanisms responsible for
regulating female reproductivity (e.g., estrous cyclicity in rats and
menstrual cycling in humans) are similar between rats and humans
(Goldman et al., 2007; Bretveld et al., 2006)." Table 3-29 is the evidence
profile table for female reproductive effects, which among other
information contains both factors that increase certainty and factors that
decrease certainty along with evidence stream (i.e., human, animal,
mechanistic/supplemental) judgments/rationales and a summary
judgment. Obviously, the text of the document (Section 3.2.5) also contains
information relevant to and supporting the WOE for female reproductive
effects.
My first comment regarding presentation and analysis concerns pp. 3-198
to 3-199, lines 17-5 that state, "Although decreased body weight in female
rats was observed at the same doses (body weight decreases were 12-36%
at >1.25 mg/kg-day; refer to Section 3.2.10 on General toxicity effects for
more details) as effects on estrous cyclicity, it is unclear if these effects are
related and the effect on female reproductive function is
disproportionately more severe and concerning than the moderate changes
in body weight." While described as "moderate" body weight changes,
reductions of 12-36% are well above those typically considered adverse
(e.g., >10% in adult animals) and would be associated with doses above the
MTD and overt toxicity. As such, the draft assessment should further
elaborate upon, explain, and/or clarify the statement cited above (Tier 1
necessary revision), especially how these effects may be related.
My second comment concerns p. 3-199, lines 5-12 that state [emphasis
added], "Although body weight has been shown to fluctuate during the
different estrous stages and weight loss has been shown to correlate with
disrupted estrous cyclicity in rats (Tropp and Markus, 2001), it is not
possible to determine if the decreases in body weight in female rats might
be responsible for the effects on estrous cyclicity observed in the NTP (2018)
study. Furthermore, even though no changes were observed on other
stages of the estrous cycle (i.e., proestrus and metestrus), the effects of
PFDA on estrus and diestrus are still considered biologically relevant given
the potential influence that the lack of cyclicity may have on fertility,
72
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
regardless of whether the observed decrease in body weight may have
partially contributed to these changes." EPA should clarify (Tier 1 necessary
revision) how it was determined that decreases in body weight only "may
have partially contributed" to estrous cycle effects given the statement
[emphasis added] that "it is not possible to determine if the decreases in
body weight in female rats might be responsible for the effects on estrous
cyclicity observed." Additionally, it is important for EPA to transparently
determine whether or not the effects on estrous cyclicity may be secondary
to the observed decreases in body weight because if so, these effects
should not be further considered for dose-response assessment (i.e., only
the body weight effects) and this should be explicitly stated in the
assessment (Tier 1 necessary revision). This is not to argue against the
adverse nature of these effects (e.g., as supported on pp. 3-203 to 3-204),
but rather to recognize the importance of whether these effects are a
direct effect of PFDA (i.e., a specific MOA) or merely a secondary effect due
to body weight decreases more generally.
Lastly, Tier 1 necessary revision comments under question "2.a" on the
most human-relevant laboratory animal species also apply here for female
reproductive effects.
f. Cardiometabolic effects: The study confidence conclusions appear scientifically
justified, giving appropriate consideration to important study attributes. For each
study in the relevant figures of Section 3.2.6 (e.g., Figures 3-69 to 3-74), the
consideration ratings appear overall consistent with (i.e., provide the scientific
justification needed for) the overall study confidence level rating. Accordingly, the
study confidence conclusions appear scientifically justified, with EPA having given
appropriate consideration to important study attributes.
This question (under "2") also concerns whether the presentation and analysis of
study results is clear, appropriate, and effective, a step just upstream of synthesis
that if well done, then allows for appropriate syntheses of the endpoint-specific
findings. Overall, the presentation and analysis of study results appears clear,
appropriate, and effective. Additionally, it appears that by and large, the available
data on cardiometabolic effects are clearly and appropriately synthesized to
describe the strengths and limitations. P. 3-235, lines 26-32 state that, "Overall,
evidence suggests that PFDA exposure has the potential to cause cardiometabolic
effects in humans under sufficient exposure conditions (see Table 3-38). This
conclusion is based on evidence of an association between PFDA exposure and
certain cardiometabolic outcomes (serum lipids, adiposity, cardiovascular disease,
and atherosclerosis) in a small number of epidemiological studies with median
exposure levels from 0.1-0.4 ng/mL; however, issues with inconsistency across
studies raise considerable uncertainty. Moreover, evidence in animals is sparse
and largely uninterpretable regarding its relevance to humans." Table 3-38 is the
evidence profile table for cardiometabolic effects, which among other information
frequently contains both factors that increase certainty and factors that decrease
certainty along with evidence stream (i.e., human, animal) judgments/rationales
and a summary judgment. Obviously, the text of the document (Section 3.2.6) also
contains information relevant to and supporting the WOE for cardiometabolic
73
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
effects.
g. Neurodevelopmental effects: The study confidence conclusions appear
scientifically justified, giving appropriate consideration to important study
attributes. For each study in the relevant figures of Section 3.2.7 (i.e., Figure 3-76),
the consideration ratings appear overall consistent with (i.e., provide the scientific
justification needed for) the overall study confidence level rating. Accordingly, the
study confidence conclusions appear scientifically justified, with EPA having given
appropriate consideration to important study attributes.
This question (under "2") also concerns whether the presentation and analysis of
study results is clear, appropriate, and effective, a step just upstream of synthesis
that if well done, then allows for appropriate syntheses of the endpoint-specific
findings. Overall, the presentation and analysis of study results appears clear,
appropriate, and effective. Additionally, it appears that by and large, the available
data on neurodevelopmental effects are clearly and appropriately synthesized to
describe the strengths and limitations. P. 3-245, lines 5-13 state [emphasis added]
that, "The evidence for potential neurodevelopmental effects in humans is
considered slight. Associations between PFDA exposure and outcomes related to
attention and behavior were reported in multiple epidemiological studies, though
there was inconsistency between these findings and the more clinically relevant
measure of ADHD diagnosis. Results for other neurodevelopmental effects were
largely inconsistent, though poor sensitivity due to limited exposure contrast may
explain the lack of association in some studies. No animal toxicity studies are
available. Altogether, based on the available human studies, the evidence suggests
that PFDA exposure might cause neurodevelopmental effects in humans under
sufficient exposure conditions (see Table 3-40)." Table 3-40 is the evidence profile
table for neurodevelopmental effects, which contains both factors that increase
certainty and factors that decrease certainty along with evidence stream (i.e.,
human) judgments/rationales and a summary judgment. Obviously, the text of the
document (Section 3.2.7) also contains information relevant to and supporting the
WOE for neurodevelopmental effects.
h. Endocrine, urinary, and other noncancer effects: The study confidence conclusions
appear scientifically justified, giving appropriate consideration to important study
attributes. For each study in the relevant figures of Sections 3.2.8 (Endocrine
Effects; Figures 3-77, 3-78, 3-80, 3-82), 3.2.9 (Urinary Effects; Figures 3-84 and 3-
85), and 3.2.10 (General Toxicity; Figure 3-88), the consideration ratings appear
overall consistent with (i.e., provide the scientific justification needed for) the
overall study confidence level rating. Accordingly, the study confidence
conclusions appear scientifically justified, with EPA having given appropriate
consideration to important study attributes. Section 3.2.11 (Other Health Effects)
had no such study evaluation heatmap figures.
• This question (under "2") also concerns whether the presentation and
analysis of study results is clear, appropriate, and effective, a step just
upstream of synthesis that if well done, then allows for appropriate
syntheses of the endpoint-specific findings. Overall, the presentation and
analysis of study results appears clear, appropriate, and effective.
74
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Additionally, it appears that overall, the available data on these effects are
clearly and appropriately synthesized to describe the strengths and
limitations. Sections 3.2.8 through 3.2.11 contain the following statements
[emphasis added]:
• "Taken together, there is inadequate evidence across human, animal, and
mechanistic data to determine whether PFDA exposure would cause
endocrine effects in humans. This conclusion is based on inconsistent
evidence from human studies and from a single high confidence rat study
investigating PFDA doses 2.5
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
that decrease certainty along with evidence stream (i.e., human, animal,
mechanistic/supplemental) judgments/rationales and a summary judgment.
Sections 3.2.10 (General Toxicity) and 3.2.11 (Other Health Effects) contain no such
tables. Obviously, the text of the document in Sections 3.2.8 through 3.2.11 also
contains information relevant to and supporting the WOE for these effects.
Hoberman
a. Liver effects: For liver effects, the available data have been clearlv and appropriately
synthesized to describe the strengths and limitations. On a weight-of-evidence
approach for hazard identification the conclusion that liver effects in rodents
appears relevant to humans and supports the possibility of liver toxicity in humans
at relevant exposures is clearly described and scientifically justified. The conclusion
is supported by the data from rats and mice tested in short-term and subchronic
studies, and rats in a developmental toxicity study.
The presentation and analysis of study results was found to be clear, appropriate
and effective, allowing for scientifically supported syntheses of the findings across
the various sets of studies.
This reviewer agrees that the available evidence indicates that PFDA exposure is
likely to cause hepatotoxicity in humans given sufficient exposure conditions. This
conclusion is based primarily on coherent liver effects in rats (and, to a lesser
extent, mice) exposed to doses >0.156 mg/kg-day for 28 days. The available
mechanistic information overall provides support for the biological plausibility of
the phenotypic effects observed in exposed animals as well as the activation of
relevant molecular and cellular pathways across human and animal models in
support of the human relevance of the animal findings (page 3-50).
(page 3-49) The evidence in humans based on liver biomarkers indicates slight
evidence of an association between PFDA exposure and hepatic effects. Positive
associations between exposure to PFDA and ALT were observed in four of five
studies of adults. A lack of coherence across clinical markers in humans reduces the
strength of the evidence.
The evidence for PFDA-induced liver effects is the strongest based on the short-
term animal studies that demonstrated coherent effects across serum biomarkers,
histopathology, and organ weight. Increases in serum biomarkers of
hepatocellular/hepatobiliary injury (ALT, AST, ALP, bile salts/acids and bilirubin)
(NTP, 2018) and liver weights were reported in male and female rats at >0.156
mg/kg-day after 28-day gavage exposure (Frawley et al., 2018; NTP, 2018). The
evidence for increased liver weights was consistent across several species (rats and
mice), strains (S-D, Wistar, Fischer F344, C57BL/6N, C57BL/6J and B6C3F1/N) and
exposure designs (gavage and dietary. At higher doses (>0.5 mg/kg-day), a
consistent pattern of hepatocellular lesions was observed in S-D rats that included
cytoplastic alterations and vacuolization, hypertrophy, and necrosis (Frawley et al.,
2018; NTP, 2018). The pattern of hepatocellular changes showed a progression in
severity within and across lesions with an increase in exposure dose, which adds
certainty to the interpretation of the evidence.
76
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
In combination with the histopathological findings, alterations in serum biomarkers
and liver weights support the development of adverse liver effects in rats after
continuous PFDA exposure. The evidence base is limited in that there is an absence
of studies via relevant exposure routes with durations longer than 28 days
examining potential hepatic effects of PFDA exposure.
Analysis of mechanistic and supplementary data from in vivo and in vitro rodent
models provide experimental (e.g., liver weight changes after i.p. exposure) and
biological support for the phenotypic effects reported in the short-term oral studies
summarized above. Exposure to PFDA was associated with the activation of several
molecular signaling pathways and altered cellular functions hypothesized to be
involved in the MOA for liver toxicity of related perfluorinated compounds.
Additionally, the evidence for PFDA-mediated liver effects implicates both PPARa-
dependent and independent mechanisms. The activation of PPARa in the MOA for
non-cancer liver effects in rodents has implications to human health assessment
based on perceived differences in PPARa response between rats/mice versus
humans. PFDA can activate the human PPARa in vitro but it exhibits less sensitivity
towards the human isoform in comparison to other mammalian species. PFDA also
interacts with other nuclear receptors and cell signaling pathways relevant to its
potential mechanism of hepatotoxicity in both human and animal models.
Some hepatic responses in animals occurred, at least in part, independent of PPARa
or were found to be activated in human in vitro assays or animal models that are
more relevant to humans with respect to PPARa sensitivity. These observations are
consistent with studies in PPARa null and humanized animals for other long-chain
PFAS such as PFOA, PFHxS and PFNA that suggest non-PPARa mechanisms of liver
toxicity. Given that the precise role of PPARa in the non-cancer liver effects of PFDA
remains largely unknown and the possible involvement of PPARa-dependent and
independent pathways, the effects observed in animals are considered potentially
relevant to humans (Soldatow et al., 2013).
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - Longer term studies in rats would confirm
trends observed in liver toxicity in the shorter-term studies used in for this
assessment.
b. Immune effects: For immune effects, the available data have been clearly and
appropriately synthesized to describe the strengths and limitations. Two
epidemiological studies in children and one study in adults demonstrated consistent
evidence of reduced antibody responses while the evidence of immunosuppression
in animal studies was not clear. The presentation and analysis of study results was
found to be clear, appropriate and effective, allowing for scientifically supported
syntheses of the findings across the various sets of studies.
(Page 3-92) The available evidence from human, animal and mechanistic studies,
indicates that PFDA exposure is likely to cause adverse immune effects, specifically
immunosuppression, in humans, given sufficient exposure conditions. The hazard
77
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
is driven primarily by consistent evidence of reduced antibody response from
human Toxicological Review of Perfluorodecanoic Acid and Related Salts
epidemiological studies (mostly from two birth cohort studies) at levels of 0.3
ng/mL (median exposure in studies observing an adverse effect).
The evidence in animals showed coherent immunomodulatory responses at
>0.089 mg/kg-day that Pare consistent with potential immunosuppression and
supportive of the human studies, although issues with overt organ/systemic
toxicity raise concerns about the biological significance of some of these effects.
A small number of studies conducted via i.p. injection and in vitro exposure in
human and rodent cell culture models added some support for the biological
plausibility of the observed phenotypic effects. While there is some evidence that
PFDA exposure might also have the potential to affect sensitization and allergic
responses, the human evidence underlying this possibility is uncertain and without
support from animal or mechanistic studies. Based on the antibody response data
in humans, children and young individuals exposed during critical developmental
windows may represent a potential susceptible population for the
immunosuppressive effects of PFDA.
Studies in humans and animals exposed to PFDA are available for the evaluation of
potential immunosuppression and sensitization or allergic responses. The
evidence of an association between PFDA exposure and immunosuppressive
effects in human studies is moderate. This is based on largely consistent decreases
in antibody response following vaccination (against two different infectious
agents) in two medium confidence studies describing results from two
independent birth cohorts in the Faroe Islands with outcome measurement in
childhood.
Reduced antibody response is an indication of immunosuppression and may result
in increased susceptibility to infectious disease (IPCS, 2012). The antibody results
present a consistent pattern of findings that higher prenatal, childhood, and adult
serum concentrations of PFDA were associated with suppression of at least one
measure of the anti-vaccine antibody response to common vaccines in two well-
conducted birth cohorts in the Faroe Islands and supported by a low confidence
study in adults. An inverse association was observed in 21 of 26 evaluations, with a
minimum of a 2% decrease in antibody concentration per doubling of PFDA
concentration at levels consistent with the general population in NHANES; six of
these evaluations were statistically significant and exhibited a large magnitude of
effect (i.e., >18% decrease in response). These associations were observed despite
poor study sensitivity, which increases confidence in the findings.
There is some remaining uncertainty resulting from variability in the response,
including positive associations in a few exposure-outcome combinations,
differences in the responses by age of exposure and outcome measures as well as
timing of vaccination (initial and boosters), from potential confounding across
PFAS, and from inconsistency in two other medium confidence studies with
outcome measurement in adults and cross-sectional exposure measurement in
children.
78
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Overall, the evidence supports an association with immunosuppressive-type
effects. These results are consistent with hazard identification conclusions from
the NTP (2016) monograph on immunotoxicity associated with exposure to PFOS
and PFOA, which concluded that PFOA and PFOS are presumed to be an immune
hazard to humans based largely on evidence of suppression of antibody responses
in both human and animal studies (NTP, 2016). Although no effects were observed
in T-dependent antibody responses with PFDA in one rat and one mouse study
(both high confidence), other immunomodulatory responses were observed in
animals that indicate potential for immunosuppression (see summary of animal
evidence below for more details). The database of animal studies examining PFDA-
induced immunosuppressive responses consists of two high or medium
confidence studies in B6C3F1/N mice (Frawley et al., 2018) and/or S-D rats
(Frawley et al., 2018; NTP, 2018) exposed via gavage for 28 days. PFDA did not
elicit a strong immunotoxic response in animals as evidenced by the absence of
treatment-related effects in a host resistance assay and most immune function
assays (NK cell activity and T-dependent antibody responses to SRBC, mixed
leukocyte response and DTH response to C. albicans). Nevertheless, coherent
responses that suggest potential immunosuppression by PFDA exposure were
observed, which is consistent with the human evidence.
The immunomodulatory responses included dose-related decreases in phagocytic
activity of rat liver macrophages (MPS activity) at > 0.25 mg/kg-day and in immune
cell population counts in mouse spleen at >0.089 mg/kg-day (Frawley et al., 2018),
but issues regarding overt organ toxicity (increased liver weight and hepatocyte
necrosis and spleen atrophy, respectively) introduce significant uncertainty
(Frawley et al., 2018).
Morphological changes occurred in the bone marrow (hypocellularity) and thymus
(atrophy and lymphocyte apoptosis) of rats at PFDA doses associated with
systemic toxicity (i.e., decreased body weights at >1.25 mg/kg-day) (NTP, 2018);
the changes are consistent with the wasting syndrome that PFDA elicits and could
represent secondary effects of the accompanying systemic toxicity.
The evidence for potential immunosuppression from short-term animal studies is
considered slight. Mechanistic evidence from a high-dose, i.p. injection study is
supportive of potential PFDA induced immunosuppression (i.e., decreased
antibody and DTH responses) in rats at >20 mg/kg (Nelson et al., 1992).
Furthermore, an in vitro study using stimulated human primary and cultured
leukocytes suggests that PFDA is capable of inhibiting NF-kB transcription and
suppressing cytokine production (Corsini et al., 2012), which may be relevant to its
mechanisms of immunotoxicity.
Limitations in the mechanistic information include issues interpreting the
exposure context (i.e., acute, high-dose exposure) of the i.p. injection study and
general lack of studies in animal and human models that can provide support for
the biological plausibility of putative immunosuppression observed in human and
animal studies.
There is slight evidence for sensitization and allergic responses from studies in
humans, but notable limitations and uncertainties in the evidence base remain. In
79
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
human studies, the available evidence for infectious disease and hypersensitivity
was less consistent than the evidence on immunosuppression and had more
uncertainties resulting from a limited number of studies, unexplained
heterogeneity in outcome or results, variable exposure assessment approaches
that considered exposure at different times in relation to outcomes, and in some
cases self-reported outcomes.
For asthma, two of the three available studies reported no association with PFDA
exposure. However, significant associations with asthma diagnosis and asthma-
related outcomes, including an exposure response gradient, were observed in one
well-conducted (medium confidence) study with adequate sensitivity (Dong et al.,
2013). This study also had the most specific outcome definition, based on asthma
incidence in the past year. These differences could account for the inconsistency
with other asthma studies, including the other medium confidence study which
examined "ever asthma". In addition, increases in biomarkers related to asthma
were reported in this study, providing biological plausibility to the apical
association. Still, the number of available studies is small, and poor sensitivity
makes the null results difficult to interpret.
In animals, the single, short-term, low confidence study that examined endpoints
relevant to sensitization and allergic responses reporting findings coherent with
immediate-type hypersensitivity (i.e., exacerbation of hypothermia and markers of
mast cell-mediated allergic inflammation in OVA-induced mice) (Lee and Kim,
2018); however, the high exposure dose used (21.4 mg/kg-day) raises significant
concerns about potential confounding effects by indirect systemic toxicity and
thus these coherent results were not interpreted to provide biological plausibility
for the findings in humans and the animal evidence was considered indeterminate
(Lee 34 and Kim, 2018).
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - Some indication if the suspected effects on the
immune system were in line with other PFOA would be helpful.
• Tier 3: Future Considerations -The absence of additional epidemiological
studies or any long-term/chronic exposure studies in animals examining
alterations in immune function or immune-related disease outcomes during
different developmental life stages represents a major source of uncertainty
in the immunotoxicity database of PFDA.
c. Developmental effects: For developmental effects, the available data have been
clearly and appropriately synthesized to describe the strengths and limitations. The
presentation and analysis of study results was found to be clear, appropriate and
effective, allowing for scientifically supported syntheses of the findings across the
various sets of studies.
This reviewer agrees that based on the available evidence PFDA exposure is likely
to cause developmental toxicity in humans given sufficient exposure conditions.
This conclusion is based on dose-dependent decreases in fetal weight in the single
toxicology study, that exposed mice during gestation to PFDA at doses >0.5
80
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
mg/kg-day and evidence of decreased birth weight from studies of exposed
humans with median PFDA values ranging from 0.11 to 0.46 ng/mL and
epidemiological evidence for biologically related effects (e.g., decreased postnatal
growth and birth length). (Page 3-156)
Based on over 45 different epidemiological studies included here the evidence of
an association between PFDA exposure and developmental effects in humans is
considered slight but was supported by the moderate evidence in animals. The
epidemiological evidence was strongest 13 and most consistent for fetal growth
restriction and in particular for birth-weight related measures, which were
assessed by the most accurate growth restriction measures available. Out of 28 in
total, 18 different studies showed some deficits for the overall population or for
both/either sex across various birth weight measures. For example, 11 out of 22
PFDA studies in the overall population reported some birth weight deficits; this
included 9 out of 14 medium and high confidence studies.
Data for growth measurements were mixed, but there appeared to be some
coherence across studies and other prenatal growth measures with different
postnatal growth parameters. For example, there was some consistency across 2
(one high and one low confidence) of the 3 postnatal weight studies with a
common examination window (~2 years of age). The evidence for other endpoints
was not as strong or consistent, including 10 of 17 birth length studies that
showed some associations.
Consistency of results varied across the developmental endpoints, the dearth of
birth weight and birth length results in the overall study populations based on
early or pre-pregnancy measures might be indicative of potential bias due to the
impact due to pregnancy hemodynamics on PFDA levels.
Despite consistent evidence of an association between PFDA and different BWT-
related measures, there is considerable uncertainty given that some sample
timing differences may explain some of the reported fetal growth restriction
deficits.
Across the outcomes, this set of developmental studies was of good quality and
generally had a low risk of bias, as out of the 45 studies across the six primary
endpoints [fetal growth restriction (including both birth weight and length
measures), gestational duration, postnatal growth, anogenital distance, birth
defects, and spontaneous abortions] were either medium or high overall
confidence.
Several studies demonstrated sufficient sensitivity to detect associations in the
overall population and across sub-groups. However, many studies lacked power
to detect statistical interactions or differences across populations especially those
based on stratified analyses. This often results from low exposure levels with
limited contrasts in many of the study populations, which may have diminished
the sensitivity of some studies to detect associations. As such, any null findings for
studies with endpoints which lacked sensitivity should not be interpreted as
supporting a lack of effect.
81
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Fetal growth restriction endpoints provided the strongest evidence for adverse
developmental effects among the available studies. In considering the dose-
dependence of the birth weight decreases, only one out of four medium or high
confidence studies with categorical PFDA exposure data showed an exposure-
response relationship. In addition, 9 of 14 medium or high confidence studies of
the overall population as well as 9 of 14 sex-specific results showing adverse
results based on continuous exposure also offer support for a biological gradient.
Exposure-response relationships were less evident for other endpoints that were
examined.
There was considerable variability in BWT deficits (P range: -29 to -101 g per In-
unit increases) in the overall population, with seven studies ranging from 31 to 59
g deficits per each I unit increase. The clinical significance of these changes may
not be immediately evident, but effects of this magnitude can increase the
number of infants at higher risk for other co-morbidities and mortality especially
during the first year of life. These population-level changes may have a large
public health impact when these mean birth weight deficits shift the birth weight
distribution to include more infants in the low-birth-weight category.
Decreased birth weight has been associated with long-term adverse health
outcomes (Osmond and Barker, 2000). Supporting the human evidence, the large
and dose-dependent effects on fetal body weight observed across two
independent experiments reported in the lone mouse study by Harris and 28
Birnbaum (1989) (medium confidence for this endpoint) are without evidence to
the contrary and thus provide moderate evidence coherent with the findings in
humans.
Following gestational PFDA exposure, decreases in fetal body weight with a
significant trend were consistently observed in both experiments at >0.5 mg/kg-
day, including doses (0.5-4 mg/kg-day) well below those that produced maternal
toxicity. The changes in fetal body weight were also large in magnitude with the
percent changes of up to 10% at the lower doses and ranging as high as 44-50%
at the highest doses tested in both experiments.
The rodent data for decreased fetal body weight are coherent with data from the
human studies in which the strongest and most consistent evidence was for fetal
growth restriction. Although an increased fetal incidence of several skeletal
variations (i.e., delayed braincase and phalanges ossification and absence of fifth
sternebrae) was observed, the delays in brain ossification, which started at >0.03
mg/kg-day, well below doses eliciting maternal toxicity, were most notable. This
change is potentially indicative of delayed development (which would be
coherent with the PFDA-induced changes on fetal body weight); however, the
significance of this variation (in terms of future adverse consequences), is
unknown, and malformations, which are known to be adverse, were not
observed.
PFDA was reported to be teratogenic in embryonic zebrafish (Truong et al., 2022;
Ulhaq et al., 2013). There were also statistically significant changes reported for
fetal viability in mice (i.e., increased % of resorptions per litter and reduced
number of live fetuses per litter) at the highest dose tested in the GD 6-15
82
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
experiment (Harris and Birnbaum, 1989); however, effects on fetal viability were
observed at the same doses as significant maternal toxicity, preventing the ability
to draw conclusions at these doses. A notable data gap exists, as animal studies
evaluating the effect of PFDA on postnatal development were not identified.
Some effects of PFDA on postnatal growth were observed in humans.
Additionally, effects on postnatal development (e.g., delayed eye opening;
reduced postnatal growth) have been observed in rodents exposed to other PFAS
such as PFOA, PFBS, PFBA.
• Tier 1: Necessary Revisions -Pubertal development is one of the four
endpoints of developmental toxicity and should be discussed in this section.
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - as noted in the review more investigation of
the potential mechanism that causes reductions on fetal growth in animal
and human studies would be useful.
d. Male reproductive effects: For male reproductive effects, the available data have
been clearly and appropriately synthesized to describe the strengths and
limitations. The presentation and analysis of study results was found to be clear,
appropriate and effective, allowing for scientifically supported syntheses of the
findings across the various sets of studies. (Page 3-184)
Evidence indicates that PFDA is likely to cause male reproductive effects in
humans under sufficient exposure conditions. This conclusion is based
primarily on a constellation of coherent evidence from a high confidence
study in animals exposed to 0.625-2.5 mg/kg-day for 28 days, with some
support for biological plausibility provided by mechanistic evidence from
i.p. and cell culture models. Although no direct information on the human
relevance of the animal evidence is available, many aspects of the male
reproductive system are conserved across species, and the limited
sensitivity in human studies may explain the lack of associations observed.
Uncertainties in the database of PFDA-induced male reproductive toxicity
includes the absence of subchronic, chronic, developmental, or
multigenerational studies testing these outcomes in animals (which,
overall, are anticipated to be more sensitive than the available short-term
study design), and a general lack of adequate epidemiological or
toxicological studies evaluating the potential for effects of early life PFDA
exposure on male reproductive system development.
The evidence of an association between PFDA exposure and male
reproductive effects in humans is limited to two medium (Tian et al., 2019;
Joensen et al., 2013) and one low confidence study (Zhou et al., 2016), with
findings suggesting potential decreases in testosterone, decreased sperm
motility, and anogenital distance with higher PFDA exposure. There are
concerns over inconsistency and imprecision, thus, the evidence is
considered indeterminate. The available evidence from a 28-day gavage
study in rats and supportive data from i.p injection and cell culture studies
83
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
in rodents provide moderate evidence of male reproductive toxicity in
animals with PFDA exposure. The 28-day rat study showed coherent effects
across several relevant endpoints, including sperm evaluations,
histopathology, hormone levels and organ weights (NTP, 2018), with most
effects observed at doses below those shown to cause overt toxicity.
Adverse histopathological changes were observed at doses associated with
body weight decrements of potential concern. The study methods were
considered high confidence for all endpoints other than sperm evaluations,
which were considered potentially insensitive due to an inadequate
exposure duration (i.e., biased towards the null; confidence is reduced
specifically in the interpreted reliability of null findings [i.e., sperm
motility]). A consistent pattern of decreased testicular and epididymal
sperm counts occurred at >0.625 mg/kg-day, but only the effects in the
epididymis were dose related.
Dose-related decreases in serum testosterone levels and testicular and
epididymal weights were also reported in rats at >0.625 mg/kg-day. The
reductions in sperm counts, serum testosterone levels and organ weights
are coherent with the mild degenerative changes found in testes and
epididymis at similar doses, particularly Leydig cell atrophy, which is
associated with androgen deficiency and decreased spermatogenesis
(Creasy et al., 2012).
Consistent effects on serum androgen levels, male reproductive organ
weights, and histopathology were observed in rodents exposed to high
doses of PFDA (>20 mg/kg) in, single, i.p. injection studies. The adverse
effects observed in the in vivo oral and i.p. exposure studies are biologically
consistent with a potential mechanism for PFDA-induced reproductive
effects in which alterations in Leydig cell functions result in decreased
steroidogenesis and androgen levels.
Limitations of the animal evidence base include the availability of only a
single, short-term oral exposure study in a single species, and uncertainties
regarding the potential impact of systemic toxicity, particularly with regard
to the observed histopathological effects. Significant reductions in body
weight were reported in the highest dose groups in the 28-day gavage
study (21% at 29 1.25 mg/kg-day and 38% at 2.5 mg/kg-day; (NTP, 2018).
Concern for nonspecific effects on the male reproductive system is
attenuated by the observed dose-related effects (i.e., sperm counts,
testosterone levels and organ weights) at a lower PFDA dose, not
associated with body weight changes (0.625 mg/kg-day).
An i.p. injection study that examined potential effects of PFDA-induced
"wasting syndrome" using pair-fed control rats observed androgenic
deficiency and male reproductive toxicity at 20 and 40 mg/kg that were
independent from severe body weight depression at the highest dose (72%
at 80 mg/kg) (Bookstaff et al., 1990).
84
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
in vitro models derived from the male reproductive system, and models
restricted to rodents, limits the ability of the available evidence to inform
potential pathways involved in PFDA- induced male reproductive toxicity
and to elucidate conserved mechanisms across species, including humans.
Nonetheless, the mechanistic information from acute i.p. and in vitro
animal studies is both consistent and coherent with the oral exposure
study evidence, and therefore, provides support for the biological
plausibility of the phenotypic responses.
In the absence of information to the contrary and given the conserved role
of androgen-dependent pathways in male reproductive functions across
species (including humans), the available evidence is relevant to humans. A
potentially susceptible population for PFDA-induced male reproductive
effects are young individuals exposed during critical developmental life
stages (e.g. the masculinization programming, which occurs prior to the
differentiation of androgen-sensitive tissues and determines penis size and
anogenital distance (Dent et al., 2015), although no such studies were
available in the current animal evidence base and few epidemiological
studies examining pubertal development and anogenital distance were
available. Androgens play a critical role in the normal development of the
male reproductive system and disruptions caused by exposures to
reproductive toxicants during gestation and early post-natal life stages can
lead to agenesis of the male reproductive system and/or infertility (Foster
and Gray, 2013; Sharpe, 2010; Scott et al., 2009).
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - Additional long term studies would be useful
to examine potential effects on reproductive organs.
e. Female reproductive effects: For female reproductive effects, the available data
have been clearly and appropriately synthesized to describe the strengths and
limitations. The presentation and analysis of study results was found to be clear,
appropriate and effective, allowing for scientifically supported syntheses of the
findings across the various sets of studies. (Page 3-202)
Evidence indicates that PFDA is likely to cause female reproductive toxicity in
humans under sufficient exposure conditions. This conclusion is based primarily on
evidence from a high confidence study in rats exposed to doses ranging from
1.25-2.5 mg/kg-day PFDA for 28 days.
The PFDA-induced disruption of estrous cyclicity observed in female rats from the
NTP study (NTP, 2018) and its implications for infertility can be considered
relevant to humans given that the mechanisms responsible for regulating female
reproductivity (e.g., estrous cyclicity in rats and menstrual cycling in humans) are
similar between rats and humans (Goldman et al., 2007; Bretveld et al., 2006).
There is indeterminate evidence of an association between PFDA exposure and
female reproductive effects in human studies, though the low confidence studies
85
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
that were available had concerns for study sensitivity which reduces the ability to
interpret the observed null findings.
A significant inverse association between PFDA and anogenital distance in girls
was observed in one study (see Developmental Effects), which is relevant to
female reproductive toxicity. The biological relevance of this effect on anogenital
distance is unclear given that an increase in this measure is considered adverse in
girls rather than a decrease per the U.S. EPA's Guidelines for Reproductive Toxicity
Risk Assessment.
The available reproductive hormone evidence for PFDA does not support an
association. Previous studies have shown an association between increased
testosterone and increased anogenital distance in women (Mira-Escolano et al.,
2014), however the human evidence is inadequate for examining PFDA-induced
effects on testosterone in women. Whereas increased testosterone was observed
in female rats in the NTP (2018) study, the study authors did not measure
anogenital distance given that there was no developmental exposure in the study.
The increased testosterone observed in female rats is considered relevant to
humans and given the known association between increased testosterone and
anogenital distance in women, an increase in anogenital distance rather than a
decrease would be expected in women exposed to PFDA. There is little biological
understanding of how hormonal perturbation or other biological processes might
result in a decrease in anogenital distance owing to PFDA exposure.
In addition to the outcomes, there is potential for two of the outcomes described
in the developmental section, preterm birth and spontaneous abortion, to be
related to female reproductive toxicity. The evidence for these outcomes was
inconsistent. Given that most of the evidence for female reproductive effects was
null or inconsistent, there is little clear indication of an association. However, the
exposure levels in most of the study populations were low, which resulted in low
sensitivity to detect an effect, and thus these findings should not be interpreted as
supporting a lack of effect.
The available data from a 28-day gavage study in rats provide moderate evidence
that PFDA exposure may cause female reproductive toxicity.. The evidence is
sparse. The data are from a single animal study that did not evaluate fertility,
pregnancy outcomes, multiple hormone levels (only testosterone was measured),
or markers of reproductive development. PFDA was observed to cause effects on
the following female reproductive parameters: organ weight (i.e., decreased
uterine weights at >1.25 mg/kg-day), hormone levels (i.e., increased testosterone
levels at >0.312 mg/kg-day), and estrous cycle (i.e., percentage of time spent in
estrus and diestrus at >1.25 mg/kg-day).
One factor increasing the strength of the evidence is the severity of the effect on
estrous cyclicity; specifically, that PFDA induced a continuous state of diestrus in
100% of rats treated at the highest dose tested (2.5 mg/kg-day), which could be
indicative of reductions or delays in fertility. However, some caution in the
interpretation of the higher dose effects is warranted given the significant
decreases in body weight, particularly at 2.5 mg/kg-day (36% decrease). Support
for the adversity and concerning nature of prolonged diestrus and its association
86
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
with infertility is provided by the following text in the U.S. EPA's Guidelines for
Reproductive Toxicity Risk Assessment:
• "Persistent diestrus indicates temporary or permanent cessation of follicular
development and ovulation, and thus at least temporary infertility/'
• "Pseudopregnancy is another altered endocrine state reflected by persistent
diestrus."
• "Significant evidence that the estrous cycle (or menstrual cycle in primates)
has been disrupted should be considered an adverse effect."
• "The greatest confidence for identification of a reproductive hazard should
be placed on significant adverse effects on sexual behavior, fertility or
development, or other endpoints that are directly related to reproductive
function such as menstrual (estrous) cycle normality, sperm evaluations,
reproductive histopathology, reproductive organ weights, and reproductive
endocrinology."
Prolonged diestrus is commonly reported in rodent models of impaired fertility (Li
et al., 2017; Caldwell et al., 2014; Miller and Takahashi, 2014; Mayer and Boehm,
2011) and continuous diestrus is observed during reproductive senescence in aged
female rats (Lefevre and Mcclintock, 1988).
There was also coherence between decreased uterus weight and increased
percentage of time spent in diestrus at >1.25 mg/kg-day. Previous studies have
shown that decreased uterus weight in rats is commonly observed during diestrus
(Westwood, 2008; Vasilenko et al., 1981; Walaas, 1952; Boettiger, 1946). In
addition to prolonged diestrus, PFDA decreased the percentage of time spent in
estrus (NTP, 2018), which could indirectly cause infertility given that rodents are
sexually receptive only during estrus (Goldman et al., 2007). The severe, PFDA-
induced decreased time spent in estrus is expected to result in decreased
opportunities for mating in the rats, and therefore reductions or delays in fertility.
Unfortunately, no multi-generational studies of PFDA were available to inform this
hypothesis. In the 28 day study, PFDA did not cause histopathological changes in
female reproductive tissues. Given the short-term duration of the lone animal
study, it cannot be reasonably ruled out that detectable histopathological effects
could have become apparent with a longer observation window. The short-term
duration of the lone animal study does not reduce confidence in the database for
PFDA-induced female reproductive effects given that biologically relevant effects
(e.g., prolonged diestrus) were still observed.
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - As the only animal data for female
reproductive effects is from a single short term toxicity study, there is a
need to investigate the effects of longer exposures in study designs that
include the effect of PFDA on female fertility and pregnancy outcomes and
87
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
possibly for more than one generations due to the continued presence of
PFDA in the body.
f. Cardiometabolic effects: For cardiometabolic effects, the available data have been
clearly and appropriately synthesized to describe the strengths and limitations. The
presentation and analysis of study results was found to be clear, appropriate and
effective, allowing for scientifically supported syntheses of the findings across the
various sets of studies. (Page 3-235).
Evidence suggests that PFDA exposure has the potential to cause cardiometabolic
effects in humans under sufficient exposure conditions. This conclusion is based
on evidence of an association between PFDA exposure and certain
cardiometabolic outcomes (serum lipids, adiposity, cardiovascular disease, and
atherosclerosis) in a small number of epidemiological studies with median
exposure levels from 0.1-0.4 ng/mL; however, issues with inconsistency across
studies raise considerable uncertainty. Moreover, evidence in animals is sparse
and largely uninterpretable regarding its relevance to humans
The evidence of an association between PFDA exposure and cardiometabolic
effects in humans is slight, with an indication of higher serum lipids, adiposity,
cardiovascular disease, and possible markers of atherosclerosis with higher PFDA
exposure. While most results were imprecise and not statistically significant,
exposure contrasts for PFDA in the study populations were relatively narrow,
which is interpreted to result in low sensitivity to detect an effect.
There is inconsistency across studies for similar outcomes, so there is considerable
uncertainty in the evidence. There is no evidence of an association with diabetes,
insulin resistance, and metabolic syndrome, but the null results are difficult to
interpret due to concerns for sensitivity.
The animal evidence is indeterminate given that the observed changes fail to
establish a coherent pattern of adverse cardiometabolic effects in animals
following short-term PFDA exposure.
The evidence in animals is limited to a high confidence study in rats exposed via
gavage for 28 days that examined cardiovascular histopathology, serum lipids and
heart weights (NTP, 2018). Dose-related decreases in triglyceride levels occurred
in males and females and cholesterol also decreased dose-dependently in females.
However, the biological significance of these responses is unclear. Absolute heart
weights decreased dose-dependently in rats at the highest doses (>1.25 mg/kg-
day) but confidence in the results is reduced by potential confounding with
decreased body weights and a lack of corroborative findings from
histopathological evaluations or other organ weight measures (relative heart
weight was unchanged).
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - A major limitation in the animal toxicity
database of this chemical is the lack of studies examining prolonged or
88
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
chronic oral exposures. In addition, for some cardiometabolic endpoints (i.e.,
serum lipids), it would be preferred if studies were available in models that
are more physiologically relevant to humans given species differences in
lipid metabolism between humans and rodents (Getz and Reardon, 2012;
Davidson, 2010).
g. Neurodevelopmental effects: For neurodevelopmental effects, available data have
been clearly and appropriately synthesized to describe the strengths and
limitations. The presentation and analysis of study results was found to be clear,
appropriate and effective, allowing for scientifically supported syntheses of the
findings across the various sets of studies. (Page 3- 245)
Based on human studies, the evidence suggests that PFDA exposure might cause
neurodevelopmental effects in humans under sufficient exposure conditions.
The evidence for potential neurodevelopmental effects in humans is considered
slight.
Associations between PFDA exposure and outcomes related to attention and
behavior were reported in multiple epidemiological studies, though there was
inconsistency between these findings and the more clinically relevant measure of
ADHD diagnosis.
Results for other neurodevelopmental effects were largely inconsistent, though
poor sensitivity due to limited lexposure contrast may explain the lack of
association in some studies. No animal toxicity studies are available.
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - None
h. Endocrine, urinary, and other noncancer effects: For endocrine, urinary and other
non-cancer effects, available data have been clearly and appropriately synthesized
to describe the strengths and limitations. The presentation and analysis of study
results was found to be clear, appropriate and effective, allowing for scientifically
supported syntheses of the findings across the various sets of studies. (Page 3- 262)
There is inadequate evidence across human, animal, and mechanistic data to
determine whether PFDA exposure would cause endocrine effects in humans. This
conclusion is based on inconsistent evidence from human studies and from a single
high confidence rat study investigating PFDA doses <2.5 mg/kg-day that reported
largely incoherent effects on thyroid hormone homeostasis and thyroid structure
(i.e., increased T3, decreased TSH and T4; increased thyroid weight; no
histopathology) that cannot be interpreted based on the currently available
evidence base.
There is indeterminate evidence of an association between PFDA exposure and
endocrine related effects in studies of exposed humans. The evidence is largely
null, but there are concerns for study sensitivity.
89
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
The observed associations are inconsistent across studies and not coherent
across thyroid hormones. There is indeterminate animal evidence of endocrine
toxicity- specifically, thyroid effects, with PFDA based on incoherent evidence
from a single high confidence short term study in rats (a second short term study
examined adrenal effects). PFDA was shown to cause changes in thyroid
hormone levels, some of which may be interpreted as suggestive of secondary
hypothyroidism, a phenotype characterized by decreased T4 and decreased or
normal levels of TSH (Lewinski and Stasiak, 2017); however, the PFDA data are
not entirely coherent with such a hypothesis.
In the NTP (2018) study, significant trends were reported for decreased TSH and
fT4 (but not tT4) in male rats at >0.312 mg/kg-day, while significant trends were
also reported for increased T3 (the latter findings are not coherent with
hypothyroidism). Likewise, in females, increased T3 and decreased fT4 was
observed at >1.25 mg/kg-day.
High dose PFDA exposure induced decreases in total T4 were consistently
observed in multiple, high dose i.p. studies in rats. The cause of secondary
hypothyroidism is thought to be due to impaired responsiveness of the
hypothalamus-pituitary-thyroid axis (Lewinski and Stasiak, 2017). Consistent
with this, PFDA was shown to impair the response of the hypothalamic-pituitary-
thyroid axis to TRH stimulation in rats from a high dose i.p. study (Gutshall et al.,
1989). These data provide mechanistic insight and biological plausibility for how
PFDA could be decreasing serum levels of T4. Furthermore, th >1.25 mg/kg-day
in male and female rats.
A previous study observed increased relative thyroid weight in a rat model of
methimazole-induced hypothyroidism (Soukup et al., 2001). Also, an enlarged
thyroid is a symptom of hypothyroidism in humans (IQEHC, 2014). In support for
PFDA induced changes on thyroid hormone homeostasis, structurally related
PFAS compounds (e.g., PFNA; PFOA) have been shown to effect thyroid
hormone levels in rodents. However, several aspects of the available animal
data decrease the strength or certainty of the evidence informing thyroid
effects, which was only available from a single oral exposure study. Whereas the
NTP (2018) study reported changes in fT4 and TSH in rats that may indicate
secondary hypothyroidism, there was an increase in T3 that cannot be
explained. Furthermore, there are no mechanistic studies that determined the
effect of PFDA on deiodinase activity that could offer insight on how PFDA
decreased T4 and TSH while increasing T3. Additionally, while T4 was decreased
in male and female rats from the NTP (2018) study, a consistent decrease in tT4
was not observed. However as noted above, fT4 not tT4 is the preferred
measure in adult animals. Whereas there was potential coherence between
decreased fT4 and increased thyroid weight in rats, it is unclear how thyroid
weight and T3 were increased in the absence of increased TSH or
histopathological changes.
Uncertainty is also associated with the mechanistic studies and supplemental
information. Specifically, inconsistent results were observed for effects on T3 in
rats exposed to PFDA via i.p. injection and results from the protein binding
90
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
studies (Gutshall et al., 1989) suggest that PFDA decreased protein binding of T4,
which could result in increased fT4 and decreased tT4, which is not consistent
with the results from the NTP (2018) study. The mechanistic database is also
limited in that there are no studies that investigated the effects of PFDA on
deiodinase activity.
The activities of thyroid-sensitive hepatic enzymes (e.g., L-glycerol-3-phosphate
dehydrogenase) were increased in rats exposed to PFDA via the i.p. route
suggesting that thyroid activity may not be decreased due to PFDA treatment. In
general, the interpretation and relevance of the mechanistic studies and
supplemental information to thyroid effects observed in the NTP (2018) study is
unclear given that these studies used doses that were much higher (i.e., 20-80
mg/kg-day, as compared to <2.5 mg/kg-day) and associated with overt systemic
toxicity. Additionally, the mechanistic studies and supplemental information are
of shorter duration and rats were exposed to PFDA via i.p. injection rather than
gavage as was done in the NTP (2018) study.
In addition to the uncertainty in the available evidence in adults, due to the
sparse evidence base available, concern remains for potential susceptible
populations to PFDA-induced endocrine effects in susceptible populations
including young individuals exposed during gestation, early childhood, and
puberty. Importantly, T3 and T4 levels play critical roles in bone growth and
brain development (O'Shaughnessy et al., 2019) at these various life stages.
However, at the present time few epidemiological studies and no animal
toxicological studies have addressed the potential for PFDA-induced effects in
these populations.
A primary delineating feature between adult animals was coherence with
increased relative thyroid weight and decreased T4 serum levels at and
developing offspring is that adults have a considerable reserve thyroid hormone
capacity whereas developing offspring do not. Thus, there is an elevated
concern regarding the potential for decreases in thyroid hormones during
developmental life stages due to the critical endocrine dependency of in utero
and neonatal development.
(Page 3 -279) Associations between PFDA exposure and impaired renal function
were reported in two low confidence epidemiological studies. However, there is
considerable uncertainty in the interpretation of these findings due to the
potential for reverse causation and some unexplained inconsistency in the
direction of association across studies. The evidence for potential urinary system
effects in experimental animals is limited to three high/medium confidence
studies in rats (Frawley et al., 2018; NTP, 2018) and one high confidence study in
mice with exposure for 28 days (Frawley et al., 2018). Although alterations in
BUN and creatine levels were observed at >1.25 mg/kg-day in rats, there is no
coherent pattern of effects (BUN levels increased and creatinine levels
decreased) or supportive information (i.e., histopathology) to determine the
toxicological relevance of the changes that occurred (NTP, 2018).
Histopathological examinations of rat kidney and urinary bladder were mostly
unremarkable across two studies (Frawley et al., 2018; NTP, 2018). Finally, the
91
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
interpretation of the absolute and relative kidney weight changes in rats at
doses >0.312 mg/kg-day is complicated by the lack of coherent histopathological
findings (Frawley et al., 2018; NTP, 2018), inconsistencies in the direction of
changes across experiments, and confounding effects from significant body
weight reductions at the highest doses tested (>1.25 mg/kg-day) (NTP, 2018). In
summary, the sparse and uncertain evidence from animal studies is considered
indeterminate. The absence of any long-term studies (subchronic/chronic) via
the oral route or other relevant routes of exposure increases uncertainty in the
evaluation of potential urinary system toxicity in animals following PFDA
exposure. Altogether, based on the available human and animal studies, there is
inadequate evidence to assess whether PFDA exposure can cause urinary system
effects in humans.
• Tier 1: Necessary Revisions -None
• Tier 2: Suggested Revisions - None
• Tier 3: Future Considerations - None
Leung
This Toxicological Review for PFDAs is overall clear and effective. This external reviewer
especially appreciates the individual heatmaps denoting the EPA's level of confidence,
as well as the summary tables listing the body of evidence separately for human and
animal studies, in a clear and organized fashion for each health effect. However, 1 would
recommend standardizing the layout of the heatmaps throughout the report; at
present, the listing of the individual studies is inconsistently listed either across columns
or rows, while the components for strength of evidence is on the other axis (it would be
much easier to interpret the various heatmaps by uniformly showing them in the same
format; Tier 2 Recommendation). The corresponding subsections on mechanistic
studies/supplemental evidence before the evidence integration text for each health
effect are excellent to support these complex data.
a. Liver effects: The Review summarizes the results of 8 human studies and 10 animal
studies that were included to examine this aggregate endpoint. Strengths and
limitations of the studies are clearly presented and well-organized. This particular
endpoint is challenging to study in humans, given the presence of likely multiple
measured and unmeasured confounders that are difficult to control for in
epidemiologic studies. The report has appropriately attempted to consider this
through reasonable confidence ratings of the human studies. The animal studies
overall are able to more strongly support the conclusions and appear mostly
consistent across the hepatic outcomes, with the exception of histopathologic
effects that were mostly low confidence studies. Furthermore, the review has also
undertaken relatively extensive supplemental analyses (Appendix) for a proposed
MOA/AOP approach, as well as analyses of relevant high-throughput screening
assays from the EPA's CompTox Chemicals Dashboard, to support the stated
conclusions.
Overall, the report's conclusions (that the available evidence suggesting PFDA
exposure has the potential to be associated with adverse hepatic effects in humans
under sufficient exposure conditions), appear to be scientifically justified.
92
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
b. Immune effects: The Review summarizes the results of several human studies and 2
animal studies that were included to examine this aggregate endpoint. Strengths
and limitations of the studies are clearly presented and well-organized. The Report's
conclusion that the available evidence suggesting that PFDA exposure is likely to be
associated with adverse immunologic outcomes is appropriate. Notably, the human
studies examining antibody response to vaccination, though they are of relatively
small sample size and vulnerable to confounding, are medium confidence studies
but show clinically significant and consistent decreases, even after adjustments for
other possible PFAS toxicant exposures. These are further supported by the two
available rat studies (high/medium confidence) demonstrating similar adverse
effects.
The body of available evidence for adverse sensitization and allergic response
effects is appropriately summarized to be less consistent.
Overall, the report's conclusions stating that the available evidence suggesting that
PFDA exposure has the potential to be associated with adverse immune effects
(specifically immunosuppression) in humans under sufficient exposure conditions,
appear to be scientifically justified.
c. Developmental effects: The Review summarizes the results of 46 human studies and
1 animal study that were included to examine this aggregate endpoint, which
overlaps with the female reproductive outcomes assessed. Strengths and limitations
of the studies are clearly presented and well-organized, including the additional
methodological considerations relevant to the complex physiology associated with
early fetal development during pregnancy. Given the multiple possible windows of
exposure for this outcome, I recommend stating when possible whether the
individual study evaluated and conclusions drawn were concerning maternal and/or
neonatal PFDA exposure, especially in the Table 3-24 summary of the weight of
evidence (Tier 1 Recommendation).
Overall, the report's conclusions stating that the available evidence suggesting that
PFDA exposure has the potential to be associated with adverse developmental
effects in humans under sufficient exposure conditions, appear to be scientifically
justified. Notable data gaps on this topic are appropriately acknowledged, including
the scarce data regarding postnatal development.
d. Male reproductive effects: The Review summarizes the results of 9 human studies
and 1 animal study that were included to examine this aggregate endpoint.
Strengths and limitations of the studies are clearly presented and well-organized.
The Review's conclusions stating that PFDA exposure is likely to cause male
reproductive effects (upon sufficient exposure conditions) in humans is based only
on 3 studies (two medium confidence, one low confidence); the conclusion is
primarily drawn instead from the single short-term (28 days) high confidence rat
study, which showed negative effects on all male reproductive parameters studied
except sperm evaluation, which could not be evaluated due to insufficient exposure
duration (NTP 2018). The report acknowledges the relative scarcity of available
human data to inform this endpoint, but the strength of the animal study across
93
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
multiple parts of the male reproductive axis appears to appropriately provide the
scientific support for the stated conclusions.
e. Female reproductive effects: The Review summarizes the results of 22 human
studies and 1 animal study that were included to examine this aggregate endpoint.
Strengths and limitations of the studies are clearly presented and well-organized. It
is appropriate that the literature search took into account a wide range of possible
PFDA exposure developmental timepoints, given the long window for exposure
effects in the female reproductive cycle.
As the outcomes characterizing this aggregate outcome are quite heterogeneous,
the Report proposes an there is indeterminate evidence of association when
assessing the human (mostly low confidence) studies and a moderate evidence of
association when assessing the single high confidence rat study (which was short-
term [28 days], and examined only select adverse female reproductive effects).
Based on the combination of these human and rat data, the Report overall
concludes that the available evidence shows that PFDA exposure is likely to be
associated with adverse female reproductive effects under sufficient exposure
conditions. Although there is similar reproductive physiology between rats and
humans, the Report's overall conclusion to suggest there is a likely association
between PFDA exposure and female reproductive toxicity may be too strong, given
the sparse data of the single rat study (a Tier 2 Recommendation is suggested to
considering softening this conclusion). Indeed, the proposed types of studies stated
at the end of the evidence integration text seem to be needed to more strongly
support this conclusion (Tier 3 Recommendation).
f. Cardiometabolic effects: The Review has synthesized the results of 22 human
epidemiologic studies and 1 animal study that were included to examine this
aggregate endpoint. The report is overall clear and effectively organized, and
summarizes the included studies' strengths and limitations. The Review delineated
reasons for noting why many studies were regarded only as medium confidence,
such as the inclusion of some studies that may have included non-fasting (less
accurate) serum lipid levels and potential lipid-lowering medication use.
The clinical relevance of Tables 3-30, 3-33, and 3-34 could be made easier to
interpret if the various measurement units for the various relevant outcomes (such
the lipid subfractions, fasting blood glucose, and waist circumference, respectively)
were provided for the studies that reported the I? effect estimate (Tier 2
Recommendation).
It would be additionally helpful in Section 3.2.6 to provide more details of the lipid
subfractions reported in the studies (e.g., the strength and directionality of
association between PFDA exposure and not just total cholesterol levels, but also
LDL, HDL, and triglycerides, since HDL elevation is protective while elevation of the
other subfractions are adverse outcomes); similarly, the evidence integration
section can be better clarified in this regard [Tier 2 Recommendation]).
Overall, the report's conclusions, stating that the available evidence suggesting that
PFDA exposure has the potential to be associated with adverse cardiometabolic
94
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
effects in humans under sufficient exposure conditions, appear to be scientifically
justified.
g. Neurodevelopmental effects: The synthesized results of the 13 human and 0 animal
studies that were included for this endpoint are clear and effectively organized. The
Review has appropriate taken into account the available studies' strengths and
limitations for this aggregate endpoint. Notably, only about half of the human
studies were regarded as having good or high confidence in the exposure
classification, as there was substantial variability in the timing of exposure (from
early pregnancy to as late as age 11 years in the offspring), thus there was
considerable heterogeneity across studies. Furthermore, the possible
neurodevelopmental outcomes ascertained in these studies were in themselves
quite heterogeneous. The report's conclusions, which state that the available
evidence is insufficient to deem whether or not there is an association between
PFDA exposure and adverse neurodevelopmental effects in humans, appear to be
scientifically justified.
h. Endocrine, urinary, and other noncancer effects: The Review summarizes the results
of available human studies (23 studies on thyroid effects, 14 on urinary effects) and
2 animal studies (1 on thyroid effects, 1 on adrenal effects, 4 on urinary effects, 8 on
general toxicity, 2 on other organ systems) that were included to examine this
combined endpoint of multiple organ systems.
Strengths and limitations of the studies are well-organized. However, the report
may consider the following:
1. In the methodological considerations of the endocrine studies, serum thyroid
hormone concentrations are not usually regarded to have diurnal variation,
and this may have thus been an unnecessary limitation (it is a Tier 1
Recommendation to not regard thyroid-related studies that failed to consider
diurnal variation as deficient); ** Table 3-77 already lists this; revise to state
that downgrading them not necessary
2. Any included studies of thyroid hormone status in pregnancy should not be
combined with the same assessments as non-pregnant adults (Tier 1
Recommendation), given the different reference ranges forTSH and of
changes in the binding of serum thyroid hormone levels in gestation;
3. Similar to the developmental effects outcome, given the multiple possible
windows of PFDA exposure on thyroid effects, it would be clearer to state
when possible whether the individual study evaluated and conclusions drawn
were concerning maternal and/or neonatal PFDA exposure (Tier 1
Recommendation).
Though it is fairly unlikely, reevaluation of the confidence of the included human
studies (in regards to the above considerations) may potentially alter the Report's
strength of evidence for adverse endocrine/thyroid effects.
95
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Overall, the Report's conclusion that there is inadequate available evidence to
support whether or not there are associations between PFDA exposure and urinary
and other noncancer effects is scientifically justified.
Zoeller
a. Liver effects: The Agency's analysis of PFDA impacts of hepatic toxicity is described
in section 3.2.1. This analysis is clearly stated, comprehensive and appears
scientifically defensible and clear. No recommendation.
a.i. This analysis was clear and defensible. No Recommendation.
b. Immune effects: The Agency's analysis was covered in section 3.2.2. The Agency
described the human and animal studies that provide information in this domain in
detail including a consideration of confounding due to the presence of other PFAS.
There was considerable discussion and clarification of this issue. The take-home
lesson was that the Agency needs to explain more fully what they have done and
what their thinking is. Tier 2 Recommendation. It may be useful for the Agency
explain more fully in the main document their thinking about the issue of
confounding by other PFAS (especially PFOS & PFOA). Although it is the goal of this
assessment to develop an RfD for PFDA as if it is the only chemical causing an effect,
mixture risk analysis procedures are being developed to explain the combined
effect of more than one chemical exposure. The Agency may choose to strengthen
their argument about PFDA using some of this logic. Tier 2 Recommendation. The
agency explained their thinking about the use of PFDA effects on diphtheria and
tetanus antibodies as an index of immune suppression, but this appeared to require
strengthening based on some of the discussion. Thus, the Agency may opt to
develop this thinking further in the main document.
c. Developmental effects: The effect of PFDA on development was covered in
section 3.2.3. The Agency identified 8 studies on postnatal growth, 12 studies on
gestational duration, 6 on fetal loss, 3 or anogenital distance, 2 on birth defects
and 31 on fetal growth restrictions. The Agency concludes that, "Based on over 45
different epidemiological studies included here the evidence of an association
between PFDA exposure and developmental effects in humans is considered slight
but was supported by moderate evidence in animals."
1 was not entirely convinced by the Agency's arguments. For example, "...9 out of
14 medium and high confidence studies" reported birth weight deficits. About this
endpoint, the Agency stated that, "Despite fairly consistent evidence of an
association between PFDA and different BWT-related measures, and more mixed
for other endpoints, there is considerable uncertainty given that some sample
timing differences may explain some of the reported fetal growth restriction
deficits examined here." The argument here appears to be that a number of high-
quality studies are disregarded - or "downgraded" - because there are no data to
rule out the possibility that the primary observation (i.e., fetal growth restriction)
is biased away from the null. Would the association between PFDA exposure and
developmental effects in humans be more important (stronger) if it was known
that sample timing differences were not confounding?
Tier 1 recommendation. The argument that uncertainty around the relationship
96
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
between sample timing differences and reported fetal growth restriction deficits
needs to be strengthened. Is it more likely than not to be a confounding variable
and what is the evidence for this opinion? If this isn't justified, then downgrading
an important observation seems unwarranted.
d. Male reproductive effects: The Agency covers the evidence indicating that
PFDA is likely to cause male reproductive effects in humans in section 3.2.4.
This conclusion was based primarily on a single high confidence study (NTP)
in animals with a number of coherent endpoints of male reproduction
being affected by PFDA exposure. Several epidemiological studies we less
informative, but the Agency clearly articulated why these results were less
informative. No recommendation.
e. Female reproductive effects: The effect of PFDA on female reproduction is covered
in section 3.2.5. Epidemiological studies were deemed inconclusive because of the
variability among studies and lack of coherent observations. A single high
confidence animals study identified a number of coherent endpoints reflective of
adverse effects on female reproduction. No recommendation.
f. Cardiometabolic effects: Cardiometabolic effects includes here risk factors for
cardiovascular disease including obesity, serum lipids, diabetes, and others. Thus,
this is a complex combination of studies in humans and animals. The Agency's
analysis was clear and logical. No recommendation.
g. Neurodevelopmental effects: The Agency's analysis of neurodevelopmental effects
was clear. No recommendation.
h. Endocrine, urinary, and other noncancer effects: In this case, "Endocrine" means
"Thyroid". The Agency clearly and thoroughly described the data in humans and
animals and justified their interpretation that the data were indeterminate.
There are several issues here.
The interpretation of the diurnal rhythm of thyroid hormones might be overly
restrictive in the interpretation of epidemiological studies. First, the time-of-day
differences in T3 and TSH reported by van Kerkhof (2015) do not seem large
enough to downgrade a study. Perhaps more important is the work by Andersen
et al. (JCEM 2002, 87(3):1068). They show that individual variation in T4 and TSH
is about 10% of the population variation. Thus, in a population study, individual
differences in the set-point around which T4 is regulated might be more
important that time-of-day. Still, this variability would bias the results toward
the null.
The Agency captured well the inconsistency in the human data. However, the
issue of "coherence" might be overstated. Specifically, the Agency notes that,
"Even in the studies that observed associations, there is not clear coherence
across outcomes, where one would expect a decrease in T4 and T3 to
correspond with an increase in TSH, or vice versa, though this could be explained
by secondary ("central") hypothyroidism...". First, serum T3 is not associated
with TSH, while T4 is negatively associated (negative feedback). So, it is not
necessarily surprising that T4 and T3 don't follow the same pattern with respect
97
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
to PFDA exposure. Moreover, animal data for other chemicals (e.g., PCBs or
PBDEs, or other PFAS) show that these chemicals can cause a reduction in serum
total and free T4 without a change in serum T3 or TSH. Although the mechanism
for this pattern is not understood, it is well documented. Thus, the Agency
should not expect that "coherence" based on patterns of thyroid hormones in
cases of disease or antithyroid drug therapy is observed.
The Agency's interpretation of the relationship between free and total T4 seems
overstated. Total T4 is dependent upon protein binding, but in the absence of
information about protein binding, these measures should not be ignored. Also,
the concept that free T4 "is available to be utilized by the body" is true on the
surface, but there is a dynamic equilibrium between the two fractions much as
there is in the case of "free" and protein bound oxygen.
While T3 is "the more active form" of the hormone, this is not to say that serum
T3 normally drives hormone action in tissues. A clear example is the case of
MCT8 deficiency in humans where blood levels of T3 can be elevated, but the
brain is deficient in thyroid hormone because the cellular transporter that
moves T4 across membranes is missing. These two forms - T4 and T3 - are
regulated differently and should not be considered in this concept of
"coherence".
Thus, the NTP observation in animals showing a dose response of PFDA and free
T4 decrement in male rats should not be downgraded because T3 is elevated, or
because TSH exhibits a decreasing trend. Moreover, thyroid gland
histopathology is driven by TSH, so the observation that thyroid histopathology
was not altered by PFDA is "coherent" with the observation that TSH levels were
not significantly altered. The increased weight of the thyroid gland would not be
expected, but this observation should not alter the conclusion that PFDA caused
a reduction in serum free T4 in males.
The observations of Gutshall et al., (1989) are interesting because it is quite
similar to the data for PCBs and PBDEs in that serum T4 is reduced, but thyroid
hormone-regulated genes (or enzymes) in liver are upregulated. In the case of
PCBs, there is evidence that some congeners can be hydroxylated and then
activate thyroid hormone receptor. Although we don't know the underlying
effects of PFDA, Ren et al. show that PFDA can bind to the thyroid hormone
receptor and while it didn't affect TR activity in GH-3 cells, the in vivo liver may
be a different environment. Finally, the observation by Gutshall et al. that a T4
administration to PFDA-treated rats did not produce a normal increase in serum
T4 indicates that the clearance rate of T4 is accelerated in PFDA-treated animals
(as the Agency noted -"biliary excretion"). This is also similar to other liver
enzyme inducers.
Ultimately, the Agency concluded that, "Taken together, there is inadequate
evidence across human, animaland mechanistic data to determine whether
PFDA exposure would cause endocrine effects in humans. This conclusion is
based on inconsistent evidence from human studies and from a single high
confidence rat study investigating PFDA doses <2.5 mg/kg-day that reported
largely incoherent effects on thyroid hormone homeostasis and thyroid structure
(i.e., increased T3, decreased TSH and T4; increased thyroid weight; no
98
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
histopathology) that cannot be interpreted based on the currently available
evidence base."
The interpretation of the human and animal data as "incoherent" is based on a
"model" of the thyroid hormone system responding to thyroidectomy or
goitrogenic chemicals/drugs (e.g., PTU). There is enough high-quality data about
the response of the thyroid hormone system responding to environmental
chemicals (see work of Mary Gilbert at EPA and Marta Axelstad of the Danish
Technical University) to recognize that the data in the NTP study is "coherent" to
the response to various chemical exposures. Moreover, the Agency does have a
policy of interpreting a decrease in serum T4 in a rat study as "adverse"
independent of other endpoints.
Tier 1 Recommendation. The Agency should incorporate the concept of
"coherence" in terms of the response to chemicals that do not act as a goitrogen per
se. Chemicals like PCBs or PBDEs as shown by EPA's Mary Gilbert is important to
incorporate.
Tier 3 Recommendation. The Agency should develop a strategy to incorporate new
information about the thyroid hormone system into an Agency-wide policy of the
way to perform a risk assessment for chemicals affecting the thyroid hormone
system.
3,3 For PF RfC was derived for inhalation exposures. An MD is derived based on studies
by Budtz-Jorgensen and Graiiiljean (2 nd Grandjean et al. (2012) showing decreased
serum antibody concentrations for both tetanus and diphtheria in children (male and
female) at age seven years and PFDA measured at age five years and developmental effects
(i.e., reduced birth weight in humans) from the Wikstrom (2020) study. Given the close
proximity of the developmental and immune PQDs and resulting osRfDs and because these
effects are observed during the developmental period, they are selected as co-critical
effects supporting the MD, Are the selection of the studies for the immune (Budtz-
Jorgensen and Grandjean, 2018) and developmental (Wikstrom, 2020) effects for use in
deriving the MD values for PFDA scientifically justified? Are the modeling approaches
appropriate?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effec it should be used to
support the derivation of the lifetime RfD and detail the rationale for use of such an
alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the
effects selected are appropriate for use in deriving the lifetime RfD, including
considerations regarding adversity (or appropriateness in representing an adverse
99
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
change) and the scientific support for their selection19. Please also see charge
questions 2b and 2c,
el. EPA used benchmark dose modeling (BMD) (U.S. 1 \ identify polnts-of-
departure (PODs) for PFDA. Are the HMD modeling approaches, selection and
justification of benchmark response levels, and selection of the BMD models used to
identify each POD for toxicity value derivation scientifically justified and clearly
described?
e. For liver, male reproductive and female reproductive effects, quantitative information
was limited to studies in animals exposed to PFDA for 28 days and little to no
information was available to evaluate the effects of chronic exposure on these health
hazards. Therefore, the derivation of lifetime organ-specific (os) RfD values was not
attempted for liver, male reproductive and female reproductive effects. However, these
end points were considered for the derivation of subchronic osRfDs. Does the provided
scientific rationale support this decision? Please explain.
f. Given the lack of studies on inhalation exposure to PFDA, no reference concentration
(RfC) is derived. Please comment on this decision.
Noncancer Toxicity Value Data Selection and Modeling
Reviewer
Comments
Adgate
Tier 2: The selection of the Budtz-Jorgensen and Grandjean (2018) for immune effects
and Wikstrom (2020) for developmental effects for use in deriving the RfD values for
PFDA is scientifically justified, with caveats on how the results should be interpreted in
the context of the IRIS modelling analysis (see my response to 2b and c).
Tier 2: Immune Effects: This section needs a clearer presentation of the extensive
analysis that EPA did on dose-response and the link to the values that are later a major
focus of Appendices. In particular, the presentation would be improved by providing a
more detail on clinical relevance and justification for the choices made and the
uncertainties associated with the determinations.
Tier 1: Note that in the Supplemental section Table F.3 (Page F-25) is inconsistent with
the text on Starling et al 2017—text says high confidence and table say low confidence.
It seems like given that this same paper is listed as high confidence in the next table the
right answer must be high confidence. More broadly this information on correlation
19
For the decreased antibody responses, Selgrade (Tox Sci 2007;100:328-332) suggests that these specific immunotoxic effects
may be broadly indicative of developmental immunosuppression impacting these children's ability to protect against a range of
immune hazards.
For developmental effects (i.e., fetal growth restriction), the human evidence was determined to be slight, primarily due to
potential confounding by hemodynamic changes among studies showing birth weight deficits. For the study (i.e., Wikstrom, 2020)
used to derive the developmental RfD, there is no presumed impact of pregnancy hemodynamics given the early sampling (96%
from trimester 1). However, unlike the Wikstrom (2020) study, some uncertainty remains across many of the available human
developmental studies given the predominance of associations that were detected were for studies with later pregnancy
sampling.
100
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
between PFASs is important for the discussion on confounding, so this summary
information would be better addressed in the main document.
d. Tier 2: The BMD approach is justified as a scientific process and fairly clearly
described.
e. Tier 2: The scientific rationale provided is reasonable given the underlying data.
f. Tier 3: Given the lack inhalation data from animal or human studies and the
likelihood that ingestion is the dominant human exposure pathway it is reasonable
that no RfC is estimated. PFAS inhalation exposure, however, is an area that has had
little research, and some planned future studies will likely explore this topic and
whether precursors that may be transformed in to PFDA under some circumstances
are an important consideration. At this time an analysis of PFDA physical constants
and potential resulting air concentrations in indoor environments would be
informative for decision-makings given PFDA's wide presence in consumer products.
Carignan
This reviewer agrees that in the absence of data an RfC cannot be derived.
Future research should investigate inhalation toxicity for PFDA, current use PFASs and
legacy PFAS mixtures to inform derivation of RfCs. [Tier 3]
This reviewer agrees that derivation of an RfD based on studies for the immune
(Budtz-Jorgensen and Grandjean 2018) and developmental (Wikstrom 2020) effects is
scientifically justified and that the modeling approaches are appropriate.
Future research should consider toxicity of PFAS mixtures, which may be cumulative
or synergistic, to inform derivation of RfDs. [Tier 3]
Table 4-1: Symbols (+ - +/-) need to be defined. [Tier 1]
a. Selection of co-critical effects is reasonable given the close proximity of the PODs
and that they are observed during the developmental period.
b. Not applicable.
c. Reduced antibody production in response to vaccinations is a sensitive, functional
test in humans that indicates global suppression of the adaptive immune response
and may indicate increased susceptibility to infectious diseases in general.
Infant body weight is also an important and sensitive outcome for development.
Lower birth weight has been linked to increased risk of neurodevelopmental effects
including impaired language development, reduced cognitive abilities and IQ, and
increased risk of cardiovascular disease and diabetes.
d. In the absence of a minimal biological response to set the BMR for the immune
endpoint, and given the monotonic dose response with no apparent threshold, use
of Vz SD is scientifically justified and clearly described in Appendix C.
Selection of 5% as the minimal biological change leading to an adverse effect is
scientifically justified for the developmental endpoint as it is clinically
101
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
predictive of health outcomes.
e. This reviewer agrees that in the absence of longer term studies, derivation of
lifetime osRfD values for liver, male reproductive and female reproductive effects
would be uncertain and limitation to derivation of subchronic osRfDs is a
reasonable approach and the scientific rationale supports this decision.
Future research should involve longer term studies appropriate for derivation of
lifetime osRfDs. [Tier 3]
f. This reviewer agrees that in the absence inhalation toxicity studies for PFDA, no
reference concentration (RfC) can be derived.
For PFASs with similar properties (e.g., long chain) read-across could be
considered where PFAS-specific data is unlikely to be generated. [Tier 2]
Additional comments:
Suggest providing prominent clarification in the immune section and this section
which studies shown in the immune review were used to create the BDML analysis
presented by Budtz-Jorgensen and Grandjean (2018). If available it would be helpful
if effect estimates and CIs for the combined cohort used for that study could be
provided. [Tier 2]
Suggest addition of a sensitive endpoint figure for the summary that helps highlight
data rich, moderate and poor areas. [Tier 2]
Suggest adding clarification to the RfD table where the values came from and how
they were derived (make things easier for your readers). [Tier 2]
Faustman
a. This reviewer agrees with USEPA's decision to use the two studies on serum
antibody response as the most sensitive endpoints along with a study of reduced
birth weight in humans for development of an RfD. This choice is consistent with
WHO guidelines that identify such immunological changes as relevant for
assessment of evidence for immunotoxicity and reduced body weight of children is
also well-recognized by WHO as well as USEPA as an important indicator of
alterations in the developmental trajectory. Using these three studies allowed for
the incorporation of two antibody responses one to diphtheria and tetanus and this
is supported by similarities in the dose response relationship and allowed for
assessments in both males and female offspring and at a developmental time period
where PFDA levels were assessed.
c. This reviewer agrees with the choice of both reductions in birth weight as well as
decreased antibody responses as well recognized endpoints of toxicological
importance.
Reduced birth weight is recognized as a clear sign of developmental toxicity in
both human and animal studies in long-standing guidance from the USEPA
developmental toxicity guidelines as well as WHO reports. It is recognized as a
clinically important parameter. In the birth weight studies in humans, USEPA
provides extended discussion on the hemodynamic changes during pregnancy
102
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
however stops short of actually accepting a pregnancy based TKTD model.
Biological discussion on such hemodynamic changes is again referred to in the
foot note below in this section of charge questions and continues to focus
primarily on volume of distribution differences however there are many
pregnancy-related changes in PK factors and for this reviewer very little attention
to these other factors is provided. An example is lack of attention to serum
albumin changes that could also significantly alter biologically active PFDA at
various stages of pregnancy. Also changes in gut motility and gut transit time also
changes yet these other changes are not equally evaluated. Tier 2
recommendation is to acknowledge the broader range of PK differences
associated with pregnancy that could also affect the interpretation of the human
studies beyond the focused "hemodynamic" currently discussed.
The EPA cites both the WHO guidelines for immunotoxicity as well as Selgrade,
2007 as examples of supportive evidence for identifying reduced antibody
response as significant endpoints of immunotoxicity esp. when occurring in
humans at early life stages. This reviewer supports the USEPA choice of these
endpoints for development of RfD values.
d. USEPA provides details about their choice of the Benchmark response appropriate
for the decrease in antibody response (continuous endpoint) in the main IRIS
document as well as in detail in Appendix C. In particular, their discussion includes
justification for quantal responses of a choice of a BMR of between 10% (standard
default value) to 1% based on modeled population level responses. They also
discuss the identification of biologically relevant response levels for continuous
responses based on standard deviation and variability. Appendix C cites biological
references for the tetanus and diphtheria antibodies as protective levels at 0.1
lU/blood concentration from vaccine producers (Grandjean et al, 2017) as well as a
discussion of uncertainty of immunological protection at specified levels of
circulating antibodies. USEPA states that "in the absence of a clear definition of an
adverse effect for a continuous endpoints like antibody concentrations" that they
calculated a BMR at both a SD of 1 as well as Vz SD. This discussion was very well
done with explanations of what endpoints and studies were modeled and what
models were used and why as well as how choices between factors were made. The
reviewers were also made aware of how USEPA obtained the datasets from the
relevant studies from Grandjean et al and provided a searchable reference of this
data posted in HAWC. This reviewer agrees with the transparency and choices that
USEPA has made in these modeling responses for the POD.
e. This reviewer agrees with the choice by USEPA to only use the data for these
endpoints for the derivation of the subchronic os RfD. It was clear that the
quantitative information for these endpoints was limited and human data
inconclusive for this application.
f. This reviewer would agree with this approach at this time due to lack of data.
Fisher
Selection of these studies are ok. The authors need to opine on the use of
these data for extrapolation of immune suppression for a population. Is the
suppression delayed or slowed or permanently suppressed? Is this an
103
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
adverse response that is recognized clinically with clinical studies to backup
the adverse findings? What are the problems with using a remote
population of people for use in the US? Is this the first time EPA has used
this endpoint for IRIS? Tier 2.
b. My comments are restricted to the use of data for PNDA within a known
serum mixture of PFAS. PNDA represents only a few percent of the total PFAS
concentrations measured in serum of pregnant women. Budtz-Jorgensen and
Greanjean (2018) acknowledge the mixture problem if one tries to tease out
individual molecules. One way to overcome this is to assign a relative potency
for each molecule, if possible. Please comment on relative potency for PFAS?
Is this possible? Does computational toxicology help? Tier 3.
c. 1 am not qualified to answer this.
d. Selected aspects of the BMD process are clearly described. As a mixture issue, there
should be a section with a header that pertains to mixtures. Text written on pages
1-11 and 1-12 are inadequate. This is a glaring weakness. The strengths and
weaknesses of the assumptions should be clearly articulated using non-
epidemiological language. From a toxicology perspective it appears unlikely that a
molecule (PNDA) which represents only a few percent of a total mixture
concentration in serum can be teased out for regulatory purposes. However, from
an epidemiology perspective and a BMDL perspective this was accomplished. This
needs to be explained in a lay language. Provide the strengths and weaknesses of
this methodology. Tier 1.
e. If subchronic RfDs are important, and even organ specific values, then using a GLP
compliant NTP 28-day study is warranted. Granted, additional toxicity studies would
provide more confidence. Listing all in vitro studies that may or may not support
these findings would be of value in the future, if not now. Tier 2. The cost of not
continuing to modify the Kim et al. published PBPK model for PFDA is problematic.
To derive subchronic osRfDs a PBPK model would be superior to any other
methodology because a PBPK model includes organs and tissue groups in a
physiologically relevant manner and can predict internal organ dosimetry (Tier 1).
f. Because of a lack of toxicity data for the inhalation route of exposure this route was
not evaluated. This makes sense. However, from an exposure perspective one can
simulate uptake of particles containing PFDA using physics based software, such as
multi-path particle dosimetry (MPPD) software and this can be interfaced with a
PBPK model.
Georgopoulos
a. The selection of the studies used in deriving the PFDA RfD values for immune
effects (Budtz-Jorgensen and Grandjean, 2018) and for developmental effects
(Wikstrom, 2020) is appropriate and scientifically justified. The rationale for
selecting these studies is based on a thorough assessment of available data
and information on immune and developmental effects from a substantial
range of studies (presented, respectively, in Sections 3.2.2 and 3.2.3 of the
Toxicological Review).
Note: Concerns regarding the modeling component are discussed in the
104
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
answers that follow.
b. N/A (see response "a", above)
c. The effects selected for deriving the lifetime RfD (antigen response for
immune effects and low birth weight for developmental effects), in
conjunction with the corresponding studies, are appropriate and scientifically
justified.
As stated in the Review (page 3-54, lines 29-32) "[t]he production of antigen-
specific antibodies in response to an immune challenge (e.g., vaccination in
humans or injection with an antigen [...] in rodents) is a well-accepted measure
of immune function included in risk assessment guidelines and animal testing
requirements for immunotoxicity," and (page 3-55, lines 3-5) "[rjeduced
antibody production is an indication of immunosuppression and may result in
increased susceptibility to infectious diseases generally (i.e., not limited to
those specifically studied)."
For the selected developmental effect (low birth weight), the Toxicological
Review assessed the human evidence to be slight, primarily due to potential
confounding by hemodynamic changes among studies showing birth weight
deficits. However, the study selected as the basis for the developmental
effects RfD (Wikstrom et al. (2020), evaluated serum PFAS in the first
trimester of pregnancy in 96% of study participants (and analysis of the impact
of excluding those subjects who were not sampled in the first trimester
showed no appreciable difference from the results including all subjects) thus
avoiding the effect of altered pregnancy hemodynamics.
d. EPA used benchmark dose modeling (BMD) to identify points-of-departure
(PODs) for PFDA immune and developmental effects. Though the Toxicological
Review follows established BMD procedures for this purpose (U.S. EPA, 2012),
there are issues requiring further consideration:
The most important concern in deriving the above-mentioned PODs is the
presence of confounding effects associated with co-exposures to other PFAS,
especially PFNA (and potentially to other immunomodulating agents). The
Toxicological Review explicitly recognizes this concern; a relevant discussion
(from page 3-57, line 34 to page 3-58, line 11) is quoted here:
"It is plausible that the observed associations with PFDA exposure could be
explained by confounding across the PFAS. Exposure levels to other PFAS in
the Faroe Islands populations were considerably higher (PFOS17 ng/mL,
PFOA 4 ng/mL, PFNA 1 ng/mL, PFDA 0.3 ng/mL at age 5 years in Grandjean
et al. (2012), and there was a high correlation between PFDA and PFNA (r =
0.78) and moderate correlations with PFOS and PFOA (r = 0.39 and 0.35,
respectively). The authors assessed the possibility of confounding in a follow-
up paper (Budtz-J0rgensen and Grandjean, 2018a) that reanalyzed data from
both Grandjean et al. (2012) and (Grandjean et al., 2017b) for benchmark
analysis. In this re-analysis, estimates were adjusted for PFOS and PFOA.
There were variable attenuation of the observed effect estimates across the
105
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
different analyses (though some of the adjusted estimates were not
estimable, likely due to collinearity), and PFNA was not adjusted for in these
models. However, associations with PFDA were stronger than for PFNA, and
adjustment by PFOS and PFOA did not eliminate the association, so
confounding by co-occurring PFAS is unlikely to fully explain the associations.
Overall, while it is not possible to rule out confounding across PFAS, the
available evidence suggests that it is unlikely to explain the observed effects.
Other sources of potential confounding, including possible co-exposures such
as PCBs, were controlled appropriately."
The above quoted conclusion, that "associations with PFDA were stronger than
for PFNA, and adjustment by PFOS and PFOA did not eliminate the association,
so confounding by co-occurring PFAS is unlikely to fully explain the
associations" may be qualitatively valid but needs further elaboration in the
context of deriving quantitative dose-response metrics to support
development of an RfD. The correlation between PFDA and PFNA was the
highest of the reported correlations in the studies discussed above (and in fact
such high correlations appear to be common in exposures studies); however,
while the impact of PFOA and PFOS on the associations of PFDA with
decreased vaccine response in the datasets used for RfD derivation was
evaluated, it was reported that it was not possible to do the same for the
corresponding impact of PFNA.
Furthermore, though information on correlations of PFDA with other PFAS is
not provided in Wikstrom et al. (2020), the same concern regarding
confounding effects of co-exposures to other PFAS, and in particular PFNA, also
applies to the derivation of the RfD for developmental effects; two relevant
statements from the Toxicological Review document and its Appendix F on
Additional Confounding Considerations (from page 3-99, lines 12-15 and page
F-27, lines 24-32, respectively) are quoted here:
"For fetal growth restriction and other developmental endpoints, there may
be more concern over potential PFAS co-exposure confounding due to PFNA
given higher correlations with PFDA and associations that are fairly
comparable in consistency and magnitude, as detailed in Appendix F."
"11 of 22 studies showed evidence of some 24 associations with PFDA and
mean birth weight in the overall population. Among these 11 studies [...] 7
showed deficits comparable in magnitude for PFNA and PFDA. Two studies
showed larger deficits for PFDA compared to PFNA, and three studies showed
larger deficits for PFNA compared to PFDA. Given these comparable results
seen in most of these studies for both PFNA and PFDA and the moderately
high correlations consistently reported between PFDA and PFNA, there is
considerable uncertainty due to potential confounding by co-occurring PFAS
in the existing literature."
EPA should address more clearly the uncertainties associated with the
potential impact of confounding (especially from co-exposures to PFNA) on the
derivation of the RfD for immune and developmental effects.
106
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
e. The fact that insufficient information was available to evaluate the effects of
chronic exposure on the liver, male reproductive organs, and female
reproductive organs, as quantitative information was limited to studies in
animals exposed to PFDA for only 28 days, makes the decision to not pursue
the development of lifetime organ-specific (os) RfD values for these effects the
appropriate one. The use of these endpoints for deriving subchronic osRfDs is
discussed in my answer to Charge Question 4.
f. The decision to not derive reference concentration (RfC) for PFDA is
appropriate and represents the only reasonable option given the lack of
relevant inhalation exposure studies.
Suggested Revisions
• Tier 1 Necessary Revision: Expand the discussion and consideration of confounding
factors (especially co-exposures to PFNA) in the context of their potential
quantitative impact on the derivation of the RfD for immune and developmental
effects. At a minimum clarify the rationale for evaluating the effects of PFNA co-
exposures as non-significant based on the analytical details of the regression models
performed by the authors of Budtz-Jorgensen and Grandjean 2018 (HERO 7276745)
Haney
a. As 1 do not agree with the selection of these studies and endpoints for RfD
derivation, please see my comments under subsection "b" below.
b. Immune Effect Endpoint
EPA indicates that the hazard judgment that "the evidence indicates that PFDA
exposure is likely to cause adverse immune effects, specifically
immunosuppression, in humans" (p. 3-92, lines 35-37) is driven primarily by
consistent evidence of reduced antibody response from human epidemiological
studies (mostly from two birth cohort studies)..." (pp. 3-92 and 3-93, lines 37-1).
However, it is not clear that EPA has sufficiently considered adversity for the
primary basis of this hazard judgment, fully considered the
weaknesses/limitations of this evidence, or fully considered the lack of supporting
evidence from studies on potentially increased incidences of disease. For example,
even without considering potential confounding, 3 out of 4 ORs for PFDA and
antibody concentrations falling below the generally cited protective level of 0.1
lU/mL for tetanus and diphtheria in children ages 5 years (n=510) or 7 years
(n=386) contain 1, indicating that the WOE from this key study cohort is for no
statistically significant associations with less-than-protective serum antibody
concentrations in children (see eTable 4 of Grandjean et al. 2012).20 Consistent
with this, the more recent Grandjean et al. (2017a) study states [emphasis added]
that, "With many antibody concentrations being close to the assumed clinically
protective level of 0.1 lU/mL, logistic regression showed only weak tendencies for
antibody levels below the limit to be associated with serum PFAS
20 The confidence interval (CI) for the one statistically significant OR of 1.36 (age 5, diphtheria) is (1.04,1.77), with the
lower end of the CI practically equal to 1 (eTable 4 of Grandjean et al. 2012).
107
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
concentrations."21 Also consistent with a WOE for no statistically significant
associations (much less being able to say "effects" as there are problems with
causal attribution to PFDA), 6 of 8 confidence intervals (95% CIs) for both tetanus
and diphtheria serum antibodies included 0% change per 2-fold increase in
maternal and age 5 serum PFDA (see Table 3 of the study). Additionally, in
Grandjean et al. (2012), PFDA was correlated with other PFAS (e.g., PFOS, PFOA,
PFNA) that had some associations with antibody concentrations falling below the
protective level of 0.1 lU/mL (see Table 2 and eTable 4 of Grandjean et al. 2012).22
Moreover, the level of serum antibodies corresponding to a clinically protective
level appears to be assay specific. For the ToBI assay apparently used in the Faroe
Islands studies, >0.01 lU/mL is considered to be the clinically protective level, not
the value of >0.1 lU/mL indicated by study authors. Considering the comments
above, this means that the reported decreases in serum antibodies are even less
likely to be biologically/clinically significant. This is not surprising given the rarity of
tetanus/diphtheria cases, particularly in those fully vaccinated, and the WOE for
PFDA not being associated with statistically significant increases in the incidences of
diseases based on the epidemiological literature (all discussed below). To say the
least, all this brings into serious question the validity of the EPA's assumptions
regarding the clinical relevance/adversity of these serum antibody endpoints.
The clinically protective level cited by Grandjean et al. was >0.1 lU/mL. However:
• Grandjean et al. (2012) reported that "serum concentrations of antibodies
against the tetanus toxoid were measured in coded samples by the Statens
Serum Institut using enzyme-linked immunosorbent assay...", citing
Hendriksen et al. (1988);
• Hendriksen et al. (1988) describes the ToBI assay, which is a modified ELISA;
and
• WHO (2017) indicates that for a modified ELISA, clinical protection is achieved
at >0.01 lU/mL, not >0.1 lU/mL as indicated by Grandjean et al.23
That is, the WHO (2018, 2017) cites Hendriksen et al. (1988) as "a toxin binding
inhibition (ToBI) assay has been reported and demonstrated to show good
correlation with the neutralization assay (correlation coefficient = 0.95)", and for
modified ELISA assays such as this further indicates [emphasis added] that, "The
minimum amount of circulating antibody that, in most cases, ensures immunity to
tetanus is assay-specific. Using in vivo neutralization tests or modified enzyme-
21 Also, while Grandjean et al. (2017b) state that, "At age 5,152 (44%) children had antibody concentrations lower than
the protective level of 0.1 lU/mLfor diphtheria and 126 (36%) for tetanus", this appears inconsistent with Table 1 of
that study, which shows that the 25th percentiles for diphtheria and tetanus serum antibody concentrations were 0.1
lU/mL.
22 A 2-fold increase in PFOS and PFOA concentrations at age 5 years was associated with odds ratios between 2.38 (95%
CI, 0.89 to 6.35) and 4.20 (95% CI, 1.54 to 11.44) for falling below a clinically protective level of 0.1 lU/mLfor tetanus
and diphtheria antibodies at age 7 years.
23 For example, Grandjean et al. (2012) refers to a "clinically protective level of 0.1 lU/mL" several times and further
states, "An antibody concentration greater than 0.1 lU/mL is considered an important indicator of protection in
accordance with the public health rationale for routine vaccinations."
108
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
linked immunosorbent assays (ELISA), concentrations exceeding 0.01 lU/ml are
usually considered protective, whereas antibody concentrations of at least 0.1-0.2
lU/ml are defined as protective when using standard ELISA techniques." WHO
(2018) also discusses and illustrates the timing of primary and booster vaccinations
and durations of protection in the context of the minimum putatively protective
level of 0.01 IU/mL.24 Thus, the protective level cited by Grandjean et al. for the
assay used is 10-fold higher than the protective level cited by WHO (2017), even
further calling into question the biological/clinical significance and adversity of the
reported results.
In regard to confounding and the ability to causally attribute associated effects to
PFDA:
• Grandjean et al. (2012) state [emphasis added], "Although all of the 5 PFCs
measured showed negative associations with antibody levels, the
overlapping confidence intervals and the lack of comparative toxicology
studies prevent inference in regard to causal attribution... PFOS (most likely
the linear isomer) and PFOA appear to be the main culprits."
• The more recent Grandjean et al. (2017a) study states [emphasis added],
"Owing to the intercorrelations between the serum PFAS concentrations,
further analysis of the possible role of individual PFASs was not pursued, and
the observed associations may reflect the effects of the PFAS mixtures."
• Similarly, Grandjean et al. (2017b) state [emphasis added], "The close
correlations prevented meaningful adjustment for concomitant PFAS
exposures."
Thus, it appears that effects may neither rise to the level of adversity nor be
attributable specifically to PFDA. Co-exposures to other PFAS (at a minimum) that
are not/cannot be adequately accounted for in the analyses are likely to be
significant confounders in these epidemiological studies, especially because PFAS
exposures are correlated, they are chemically-similar compounds, and there
appears to be little variation in exposure (i.e., low exposure contrasts) for the
single PFAS being assessed (e.g., Table 2 of Grandjean et al. 2012, Table 1 of both
Grandjean et al. 2017a and 2017b). For example, Grandjean et al. (2012) shows
that PFOA and PFOS had a correlation coefficient of 0.5 in the blood sera of 5-year
olds and IQR differences in blood sera concentrations of less than 1.6-fold each
(e.g., 75th percentile blood concentration of PFOA/25th percentile blood
24 WHO (2018; pp. 14-15) states [emphasis added] that "To illustrate the kinetics of immunity among children >1 year,
adolescents and adults following primary and booster vaccination with TTCV, Figure 2 provides a schematic
diagram of the typical response. A single dose of TT in the absence of priming induces little, if any, protection. Two to
four weeks after the second dose, the mean level of tetanus antitoxin typically exceeds the minimum putatively
protective level of 0.01 IU/mL. One year after the second dose, the mean antibody levels are expected to decline and
may fall to the protective threshold level. After each subsequent dose of vaccine, immunity is boosted, then persists
above the protective threshold for a time, and then wanes over time. Putatively protective levels of immunity are
induced by a primary series of three TTCV doses and immunity typically persists for at least 5 years. After the third
dose, each additional booster dose given after at least a one-year interval increases tetanus antitoxin levels and further
prolongs the duration of immunity. Immunity may persist for approximately 10 years after the fourth dose of TTCV and
for at least 20 years after the fifth dose."
109
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
concentration of PFOA), and PFDA and PFNA had a correlation coefficient of 0.78
in the blood sera of 5-year olds and IQR differences in blood sera concentrations
of less than 1.9-fold each (see Table 2 of the study).
Despite Grandjean et al. (2017b) stating that the close correlations prevented
meaningful adjustment for concomitant PFAS exposures, Budtz-J0rgensen and
Grandjean (2018a) attempts to do just that for benchmark dose, and the results
appear to help demonstrate the effects of confounding co-exposures and/or the
inability to properly adjust for them. Table 2 of Grandjean et al. (2017b) reports
the change (in percent) of the pre-booster serum-antibody concentrations at age
5 years associated with a doubling of the serum concentration of PFDA. Results
for cohorts 5, 3, and joint results are in the negative direction for tetanus at age 5
but none are statistically significant, which already points to the unreliability of an
effect having been demonstrated by these results and the unreliability of any RfD
based on these results or association. Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports benchmark results for the five prenatal PFAS concentrations in
regard to antibody concentrations at age 5 years (pre-booster) both unadjusted
and adjusted for PFOS/PFOA co-exposures. For tetanus antibodies, while
unadjusted benchmark doses for three models for PFDA (linear, piecewise,
conservative) appear to show excellent agreement (BMDs of 0.11-0.25 ng/mL)
with insignificant reliance on choice of model, when adjusted for PFOS/PFOA the
three models' best estimates of the PFDA serum concentrations associated with a
5% change go to infinity (although BMDLs can still be estimated). For diphtheria,
Table 2 of Grandjean et al. (2017b) reports statistically significant changes for
cohort 3 and joint results for the pre-booster serum-antibody concentrations at
age 5 years associated with a doubling of the serum concentration of PFDA.
However, similar to results for tetanus, Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports that when benchmark results are adjusted for PFOS/PFOA, two of
the three models' best estimates of the PFDA serum concentrations associated
with a 5% change go to infinity (although BMDLs can still be estimated). While
Table 1 of Budtz-J0rgensen and Grandjean (2018a; benchmark results for the age-
5 serum concentrations of five PFASs in regard to tetanus and diphtheria antibody
concentrations at age 7 years) provides no benchmark doses for PFDA that go to
infinity, Table 2 results point to the unreliability of this endpoint and should be
considered along with the other issues raised above. For example, the odds ratios
for PFDA and inadequate antibody concentrations for diphtheria and tetanus at 7
years was not statistically significant (see eTable 4 of Grandjean et al. 2012).
Key BMD results in the draft assessment itself demonstrate the importance of co-
exposures to other PFAS. Tables C-l and C-3 of the draft assessment (below)
provide BMD results for the critical effects used for RfD derivation (e.g., see Table
ES-1).
110
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Table C-l. Results specific to the slope from the linear analyses of PFDA
measured in serum at age 5 years and log2(tetanus antibody concentrations)
measured at age 7 years in a single-PFAS model and in a multi-PFAS model
from fBudtz-lflrgensen and Grandiean. 2018h1.
Exposure
Model
shape
PFOS &
PFOA
adjusted
Slope ((3)
per ng/mL
in serum
SE(P)
ng/m Lin
serum
Slope ((3) fit
Lower bound
slope (Plb) per
ng/mL in serum
PFDA at Age 5
Linear
No
-1.55
0.602
p = 0.01
-2.55
PFDA at Age 5
Linear
Yes
-0.98
0.681
p = 0.15
-2.10
PFOS and PFOA had correlation coefficients of 0.39 and 0.35 with serum PFDA at
age 5, respectively (Table 2 of Grandjean et al. 2012). Despite these relatively low
correlation coefficients (Mukaka 2012), Table C-l shows that just controlling for
co-exposures from these two PFAS (PFOS, PFOA) resulted in significant impacts on
slope ((B) and slope fit. The slope estimate for PFDA was reduced 37% and PFDA is
no longer a significant predictor of tetanus antibody concentrations (p=0.15). The
most likely explanation (Occam's razor) is classic confounding as PFOS and PFOA
are documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean
2018a), and the existence of some chance that correction for these co-exposures
could create some confounding is not a scientifically robust justification for
dismissing the important implications of the results of adjustments for
PFOS/PFOA that the study authors themselves (Budtz-J0rgensen and Grandjean
2018a) thought it important to adjust for, and with good reason. These results
demonstrate the statistical unreliability of serum PFDA predicting tetanus
antibody concentrations when just two other PFAS are controlled for. Table C-3
concerns the critical effect for the RfD based on diphtheria antibody
concentrations.
Table C-3. Results specific to the slope from the linear analyses of PFDA in
serum measured at age 5 years and log2 (diphtheria antibodies) measured at
age 7 years from Table 1 in a single-PFAS model and in a multi-PFAS model
from fRudt/ lorgensen and Grandiean. 2018b).
Exposure
Model shape
PFOS & PFOA
adjusted
Slope (P) per
ng/mL in
serum
SE(P)
ng/mL in
serum
Slope (P) fit
Lower bound
slope (Plb)
per
ng/mL in
serum
PFDA at Age 5
Linear
No
-0.894
0.561
p = 0.11
-1.82
PFDA at Age 5
Linear
Yes
-0.297
0.635
p = 0.64
-1.35
The implications of these results are worse. First, even when evaluated alone
without accounting for co-exposures to relatively low correlated PFOS and PFOA,
serum PFDA is not a significant predictor of diphtheria antibody concentrations
(p=0.11). Table C-3 further shows that controlling for co-exposures from these
two PFAS (PFOS, PFOA) resulted in significant impacts on slope ((B) and slope fit.
The slope estimate for PFDA was reduced 67% and serum PFDA became a worse
nonsignificant predictor of diphtheria antibody concentrations (p=0.64). The most
likely explanation (Occam's razor) is classic confounding as PFOS and PFOA are
documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean 2018a),
and the existence of some chance that correction for these co-exposures could
create some confounding is not a scientifically robust justification for dismissing
111
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
the important implications of the results of adjustments for PFOS/PFOA that the
study authors themselves (Budtz-J0rgensen and Grandjean 2018a) thought it
important to adjust for, and with good reason. Thus, when co-exposures are taken
into account for two modestly correlated PFAS (PFOS, PFOA), serum PFDA is not a
significant (i.e., reliable) predictor of these critical effects serving as the basis of
the RfD (i.e., decreases in serum tetanus and diphtheria antibody
concentrations).25
Moreover, Budtz-J0rgensen and Grandjean (2018a) also controlled for PFOS or
PFOA when deriving BMDs/BMDLs for the other, and EPA did use those co-
exposure-adjusted results for RfD derivation (i.e., PFOA adjusted for PFOS, PFOS
adjusted for PFOA; see Tables B-l and B-2 in Sections B.l.l and B.1.2 of the PFOA
and PFOS draft assessments USEPA 2021a,b) without expressing any similar
concerns about creating confounding by adjusting for these co-exposures
(low/moderate correlation coefficient of 0.50 (Mukaka 2012); see Table 2 of
Grandjean et al. 2012) or citing Weisskopf et al. (2018) and Weisskopf and
Webster (2017). Rather, these concerns have been selectively cited for PFDA in an
attempt to provide some rationale for dismissal of the co-exposure-controlled
results (e.g., slope ((B) values are reduced; serum PFDA is a nonsignificant
predictor of tetanus and diphtheria antibody concentrations) and thus for
selection of BMDLs uncontrolled for PFOS and PFOA co-exposures (see Tables C-2
and C-4 of the draft), but this is not a scientifically robust rationale and is
furthermore inconsistent with EPA's draft assessments for PFOA and PFOS (USEPA
2021a,b). Lastly, it is further noted that Weisskopf et al. (2018) indicates: (1)
sometimes, depending on causal structure, the inclusion of multiple exposure
variables in a model can amplify the amount of bias in a regression estimate
compared to analyzing single exposures; and (2) this potential amplification of
biases increases with stronger correlations between mixture components. To
demonstrate that this can occur in some cases, the study authors used "highly
correlated exposures" (e.g., r2=0.9), whereas the correlation coefficients between
PFDA and PFOS/PFOA are low (0.35 and 0.39, respectively). The results of
Weisskopf et al. (2018) do not constitute reasonable doubt that for these PFDA
results, the confounding from not adjusting for co-exposures to documented
immunotoxicants (PFOS, PFOA) is significantly greater than the potential
amplification of biases that remains undemonstrated under the same or similar
circumstances. EPA should reevaluate the issues raised above in regard to
implications for their draft PFDA assessment (Tier 1 necessary revision).
Regarding potential immunosuppressive effects by PFDA, effects that rise to the
level of adversity would be expected to result in increased incidences of disease,
reflecting lower immunity and lower resistance to disease in the real world.
However, consistent with the WOE for no statistically significant associations with
antibody concentrations falling below the generally cited protective level of 0.1
lU/mL based on results from Grandjean et al. (2012), almost all ORs in Table 3-13
of the draft assessment include 1, indicating that the WOE from studies on PFDA
25 Knowing this, disparate results are not particularly surprising, such as the 18.7% increase in tetanus antibodies
predicted for children (age 13) with a 2-fold increase in serum PFDA based on the same study and type of analysis used
for the RfD critical effects (see Table 3-12, p. 3-59).
112
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
and infectious disease in humans is for no statistically significant associations.
Consistent with this, host resistance was unaffected by PFDA based on the limited
animal study data available (p. 3-73, lines 8-9), and host resistance assays are
considered highly relevant to the evaluation of immunotoxicity in the context of
human health assessment (p. 3-72, lines 15-18; IPCS 2012). EPA should reevaluate
the adversity of these presumed antibody level effects, including the association
with PFDA itself, and do so within the context of potential confounding, other
limitations, and available human/animal data on disease incidence (Tier 1
necessary revision), as this has important implications for the hazard judgment,
the strength of human evidence descriptor for immunosuppression (listed as
"moderate" in Table 3-19), and whether an RfD should be based on these effects.
Moreover, it is noted that the PODs for immunosuppressive effects from these
epidemiological studies range from 2.57E-04 mg PFDA/L blood serum to 7.02E-04
mg PFDA/L blood serum (BMDLi/2sd values from Table 5-8, pp. 5-16 to 5-17), and
when intrahuman variability is considered (through application of a UFH of 10),
the resulting values range from 2.57E-05 to 7.02E-05 mg PFDA/L blood or 0.0257
to 0.0702 ng PFDA/L blood serum.26 Data from NHANES show that GMs
representative of the U.S. population are well above these blood serum levels
(see Appendix A to these comments).27 Most notably, 2005-2018 population GMs
range from 0.154-0.355 ng/L, which are 2.2- to 13.8-fold higher than the PODs
adjusted for intrahuman variability (cited above). Despite these exceedances,
tetanus and diphtheria appear to be quite rare in the U.S. population. The
average annual number of tetanus cases in the U.S. from 2009-2018 was 29, with
the CDC attributing most cases to individuals who either have not been
vaccinated or who are not current on their boosters (e.g., only 3% of the cases
from 2001-2008 were in people who had received a complete tetanus toxoid
series with the last dose within 10 years; Tiwari et al. 2021). Tetanus also appears
rare in U.S. children specifically, occurring primarily in older adults. Per Liang et al.
(2018):
"During 2001-2016, three neonatal tetanus cases and 459 non-neonatal tetanus
cases were reported to the National Notifiable Diseases Surveillance System
(NNDSS). The median age for non-neonatal cases was 44.0 years (range: 2-95
years)... The risk for both tetanus disease and mortality was higher among
persons aged >65 years than among persons aged <65 years. Tetanus occurs
almost exclusively among persons who are unvaccinated or inadequately
vaccinated or in those whose vaccination histories are unknown or uncertain."
The incidence of U.S. diphtheria cases is even more rare. The CDC reported only
14 cases from 1996 through 2018 (Acosta et al. 2021). Thus, consistent with the
WOE for the lack of statistically significant associations from the epidemiology
study data discussed above and despite the NHANES blood serum data showing
exceedances of the draft assessment human PODs adjusted for intrahuman
26 Based on BMDi/2sd values (Table C-9, p. C-16), these values are 0.0385 to 0.226 ng PFDA/L blood serum.
27 See NHANES Biomonitoring Data Tables at https://www.cdc.gov/exposurereport/data tables.html. Budtz-J0rgensen
and Grandjean (2018a) also acknowledge that, "Our BMDL results, both before and after adjustment are generally
below current exposure levels..."
113
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
variability (e.g., toxicodynamic) for most of the U.S. population for a prolonged
period of time (see the 50th percentile concentrations in Appendix A), U.S.
surveillance disease incidence data are not supportive of adversity. That is, U.S.
surveillance disease incidence data do not support that serum PFDA (or any other
serum PFAS) is suppressing tetanus and diphtheria vaccine responses and leaving
people vulnerable to infection from these diseases. The apparent lack of
adversity/ consequence for the effects reported for tetanus and diphtheria
certainly does not provide support for an expectation of adversity/consequence
for other effects not measured/observed (e.g., for vaccines for other diseases and
their incidences).
Additionally, it appears that EPA has not fully considered all the relevant evidence
or the weaknesses/limitations of the epidemiological evidence, which in turn are
relevant for the hazard judgment and whether an RfD should be derived based on
these effects. Table 3-19 indicates that human data provide "moderate" evidence
and that "the inconsistent and low confidence evidence on infectious disease did
not influence this judgment." However, this points to the fact that EPA has not
duly considered the implications of the null findings on human and laboratory
animal infectious disease and other relevant considerations (e.g., some discussed
above) for the scientific WOE, which is not a scientifically supportable approach as
it does not consider all relevant data, directly relevant human data in particular.
EPA should consider such null findings (and other relevant considerations) in their
WOE (Tier 1 necessary revision). Combined with the "slight" human data for
sensitization and allergic response, the "slight" laboratory animal data for
immunosuppression, and the "indeterminate" animal data for sensitization and
allergic response (Table 3-19), it does not appear that PFDA exposure is "likely to
cause" adverse immune effects in humans is sufficiently supported. EPA should
reevaluate this determination (Tier 1 necessary revision) as "may cause" might
very well be the better supported hazard judgement, and also reconsider their use
of the serum antibody endpoints for quantitative risk assessment/derivation of
toxicity factors (Tier 1 necessary revision). This would appear more consistent
with the data discussed above and recent conclusions by the Australian
government (FSANZ 2021) and the U.S. Agency for Toxic Substances and Disease
Registry (ATSDR 2021).
The Australian government (FSANZ 2021) has concluded that associations of PFAS
with immunological endpoints do not provide a suitable basis for quantitative risk
assessment:
"In summary, new epidemiological studies provide some evidence of statistical
associations between PFAS blood levels and impaired vaccine response,
increased susceptibility to infectious disease and hypersensitivity responses.
However the data are insufficient to establish causal relationships and it cannot
be ruled out with reasonable confidence that the observed statistical
associations may have been due to confounding, bias or chance. On the basis of
the uncertainties and limitations in the evidence base, immunomodulation is
not currently considered suitable as a critical endpoint for quantitative risk
assessment for PFAS."
114
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Similarly, ATSDR (2021) found the epidemiology literature inadequate for use as
the basis of deriving minimal risk levels (MRLs)for PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive targets
for many of the perfluoroalkyls; however, there are two major limitations to
establishing dose-response relationships for these effects and using the
epidemiological studies to derive MRLs: accurate identification of
environmental exposure levels producing increased risk for adverse effects
(exposure estimates and routes of exposure) and likely co-exposure to mixtures
of perfluoroalkyls. Other limitations include the cross-sectional design of the
majority of epidemiological studies and the potential that reverse causality
contributes to the observed associations... In summary, the epidemiological
databases for several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with adverse
effects and possible interactions between compounds preclude use of these
data to derive MRLs."
Based on the information in the draft assessment and reviewed elsewhere¦, this
reviewer finds it difficult to disagree with the recent conclusions of the Australian
government (FSANZ 2021) and ATSDR (2021) that the epidemiology literature
(e.g., on PFAS blood levels and impaired vaccine response) is inadequate for
quantitative risk assessment and use as the basis for deriving toxicity factors (e.g.,
RfDs). Unreliable/inadequate data (human or otherwise) cannot be justified
scientifically for use in dose-response assessment by how the associated
unreliable POD(s) compares to that based on reliable data definitively showing
cause-and-effect for unquestionably adverse effects, unconfounded by significant
co-exposures to similar chemicals.
The above comments provide relevant rationale for use of an alternative
study(ies) or effect(s) for lifetime RfD derivation. While the draft assessment
provides various but limited candidate PODs, due consideration of the comments
above appears to require a more complete evaluation of alternative
studies/effects (Tier 1 necessary revision). One method to begin such an
evaluation may entail sorting the study data extracted by NOAEL/LOAEL, BMD or
a similar criterion that would allow EPA to readily identify the next most sensitive
effects based on less problematic studies that are adequate for RfD derivation.
However, any newly derived toxicity factor should be subject to external expert
peer review (Tier 1 necessary revision).
Birth Weight Endpoint
EPA's hazard conclusion for developmental effects primarily relies on in vivo
animal data (i.e., the mouse study of Harris and Birnbaum 1989). Based on over 45
different epidemiological studies included in the draft assessment, the evidence of
an association between PFDA exposure and developmental effects in humans is
considered only "slight" (p. 3-156, lines 10-11). Despite the "slight" evidence in
humans, Table ES-1 indicates that RfDs were nevertheless calculated based on
decreased birth weight in male and female children (Wikstrom et al. 2020). While
data in humans, as the species of ultimate interest, are usually preferred as the
basis for derivation of toxicity factors, in this case it appears that the mere "slight"
115
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
totally of the evidence for developmental effects from the over 45 different
epidemiological studies included in the draft assessment should not be considered
sufficient for quantitative dose-response assessment (e.g., birth weight), but
rather only for potentially supportive information for hazard identification. EPA
should reconsider their use of these epidemiological data for quantitative risk
assessment/derivation of toxicity factors (Tier 1 necessary revision). EPA Figure 3-
28 (p. 3-108), reproduced below, helps illustrate the weakness and incoherence of
the overall epidemiological database for demonstrating decreased birth weight,
particularly statistically significantly decreased birth weight, associated with PFDA
levels and how no one epidemiological study could possibly be representative of
these inconsistent and disparate results or the results of a meta-analysis of
relevant studies.
Figure 3-28. Overall study population mean birth weight results for 22 PFDA
epidemiological studies3 e. (results can be viewed by clicking the HAWC link).
Abbreviation: BWT = Birth Weight
1 Studies are sorted first by overall study confidence level then by Exposure Window examined.
b" Mens et a I. [2Q1S1 pooled samples from umbilical cord blood and maternal piasma during the first and second
trimesters. The remaining studies were all based on either one umbilical or maternal sample.
c If a study presented regression coefficients for continuous exposure with multiple exposure units, only one unit
change is shown (e.g., (Bach et ai.. 2Q16). with the exception of (Li et al., 2017). which displays both IQ.R and In-
un'rt (ng/mL) values.
d The results displayed here for mean birth weight among 587 overall population participants in the POPUP Cohort
are from a larger population of participants fSwedish Environmental Protection Agency, 2017) compared to a
sample size of 3S1 in their 201B publication Gyilenhammar et al. f2Q18).
*• Xu et al. (2Q19a[i results are truncated for the 210.7 gram increase; the complete 95% CI ranges from -314.3 to
735.8 grams.
116
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Similarly, Figure 3-26 (p. 3-104) below illustrates how the results of Bach et al.
(2016), the other high confidence prospective cohort for trimester 1-2 with
adequate study sensitivity, had a nonmonotonic response, no statistically
significant quartile results, and a regression coefficient of +0.03 per each In-unit
(ng/mL) (compared to -0.147 per each In-unit increase for Wikstrom et al. 2020),
all of which does not support the Wikstrom et al. (2020) study results or their use
in toxicity factor derivation. The Bach et al. (2016) and Wikstrom et al. (2020)
studies are the first two entries in Figure 3-26, respectively.
Figure 3-26.PFDA and birth weight z-scores (overall population)3. Refer to
Birth Weight-Z for details on the individual study evaluation review.
Moreover, the BMD5rd values for decreased birth weight based on Wikstrom et al.
(2020) range from 0.54 to 0.71 ng/mL (note that BMD5rd values are not treated
like NOAELs for purposes of RfD derivation; Table C-10, p. C-22).28 By comparison,
PFDA blood serum concentrations in quartile 4 of the Bach et al. (2016) study
range from 0.43 to 2.87 ng/mL (see Table 1 of the study). Despite these quartile 4
blood serum concentrations being up to 5.3-fold higher than the BMD5rd values
for decreased birth weight based on Wikstrom et al. (2020), birth weight and birth
weight z-scores for quartile 4 in Bach et al. (2016) were actually increased
(nonsignificantly) (see Table 3 of the study). Again, this is from the other high
confidence, prospective cohort for trimester 1-2 with adequate study sensitivity.
These results from Bach et al. (2016), obtained from the other high confidence
study in the same PFDA blood serum concentration range (and higher) compared
to EPA's BMD5rd values, are inconsistent with and unsupportive of EPA's use of the
PODs for adverse birth weight effects based on Wikstrom et al. (2020) for RfD
derivation. Sex-specific results from Bach et al. (2016) do not provide strong
support for use of Wikstrom et al. (2020) results for quantitative dose-response
assessment either (e.g., Figure 3-29, p. 3-109 of the draft assessment).
Given the drastically different results across epidemiological studies and the mere
"slight" evidence of developmental effects across more than 45 such studies, EPA
should reconsider use of a single epidemiological study (Wikstrom et al. 2020) for
dose-response assessment of birth weight and RfD derivation and consider a meta-
analysis and/or using the more definitive dose-response data from the mouse study
for dose-response assessment of birth weight and RfD derivation (Tier 1 necessary
revision). It is remarkable that per EPA, the single mouse study by Harris and
28 For comparison, the BMDL5rd values range from 0.31 to 0.37 ng/mL (Table C-10, p. C-22).
117
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Birnbaum (1989) gives rise to a greater level of evidence for developmental effects
("moderate") than the results of over 45 epidemiological studies. This fact alone, in
this case, justifies use of the animal data for dose-response assessment and osRfD
derivation. Based on EPA Figure 3-56 (p. 3-152), reproduced below, the dose-
response data for fetal body weight (as an example) appear well suited for
benchmark dose analysis.
£rvd point Name Study Ni
CnnflrlAivo Raeult Units tmmbn-rlavt
PFDA Fetal Body Weight
Fetal Body Weight Harris. 1989. 3858729 High confidence Gestational Oral (GO 10-13) F1 Mouse. C57BU6n ( ;> significant littef
Gestational Oral (GO 6-15)
| Statistically significant
| Percent control respons
Figure 3-56. PFDA fetal body weight after gestational exposure (results can be viewed by clicking the HAWC link).
More specifically, Figure 3-56 shows statistically significant and progressive
decreases/trends in fetal body weight with dose across a range that includes the
>5% decrease historically used by regulatory agencies as the demarcation of
adversity in dose-response assessment (i.e., as a >5% reduction in fetal body
weight is usually considered adverse, a benchmark response of 5% is typically used
in benchmark dose analysis), from -1% to -50% and including doses that resulted
in decreases of -4% and -6% (Table 3-23, p. 3-153). Furthermore, the changes in
fetal body weight were at doses not associated with maternal toxicity (p. 3-151,
lines 5-6).
Harris and Birnbaum (1989) is a medium/high-confidence study (p. 3-150, lines 9-
12), and EPA should consider mouse data from the Harris and Birnbaum (1989)
study for use as the primary basis for osRfD development (Tier 1 necessary
revision) based on developmental growth effects consistent with: (1) this mouse
study providing the primary basis for EPA's developmental effects hazard
conclusion; (2) the lack of factors that decrease certainty for fetal growth as
evaluated in the mouse study (see Table 3-24); and (3) the extensive
epidemiological database merely being able to provide support for this mouse
study with what amounts to "slight" evidence for developmental effects across
over 45 epidemiological studies. Afterall, per the draft assessment, it is the mouse
data that primarily support that assuming sufficiently high exposure over a
sufficiently long duration (i.e., "given sufficient exposure conditions"), PFDA
exposure is likely to cause developmental toxicity in the general human
population, which includes potentially susceptible subpopulations (e.g.,
developing fetuses of pregnant women).
My suggestion to use the mouse data (Harris and Birnbaum 1989) instead of
epidemiological study data (Wikstrom et al. 2020) for dose-response assessment
and osRfD derivation based on developmental growth effects (Tier 1 necessary
revision) would make EPA's assessment somewhat more consistent with ATSDR
118
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
(2021),29 who found the epidemiology literature inadequate for use as the basis of
deriving MRLsfor PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive targets
for many of the perfluoroalkyls; however, there are two major limitations to
establishing dose-response relationships for these effects and using the
epidemiological studies to derive MRLs: accurate identification of
environmental exposure levels producing increased risk for adverse effects
(exposure estimates and routes of exposure) and likely co-exposure to mixtures
of perfluoroalkyls. Other limitations include the cross-sectional design of the
majority of epidemiological studies and the potential that reverse causality
contributes to the observed associations... In summary, the epidemiological
databases for several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with adverse
effects and possible interactions between compounds preclude use of these
data to derive MRLs."
Getting back to use of Wikstrom et al. (2020) for dose-response assessment of birth
weight and RfD derivation, Tables 2 and 3 from that study appear below (unaltered;
see license at https://creativecommons.Org/licenses/by/4.0/) and demonstrate: (1)
significant co-exposures to other PFAS (e.g., PFOS, PFOA) associated with birth
weight deficits, an ATSDR concern and significant uncertainty; and (2) few
statistically significant ORs for decreased body weight in the PFDA exposure
quartiles; that is, only 2 out of 9 ORs showed statistically significant birth weight
decrements, which were for quartile 4 where both PFOS and PFOA were certainly
co-exposures and also showed statistically significant decreases for birth weight. In
regard to this latter point concerning potential confounding by co-exposure to other
PFAS, which was cited by ATSDR as a significant uncertainty, Wikstrom et al. state
[emphasis added]:
"Another limitation is the compound-by-compound approach. Theoretically, a
health outcome is simultaneously influenced by multiple environmental factors.
Nevertheless, the exposure to several PFASs may be correlated with each other
due to common sources. Our findings were consistent across different PFAS
compounds, and we regard correction for multiple comparisons overly
conservative to be suitable for the investigations on such interrelated compounds.
If the single compound's level could represent the levels of several other
compounds, our findings based on single compound analyses may still shed some
light on the joint effects of multiple PFAS compounds. However, carefully designed
statistical models, such as mixture-based approaches within the PFAS compound
class, should be explored in follow-up studies."
Thus, the study authors are acknowledging confounding by other PFAS as an
important limitation of their compound-by-compound approach as joint effects of
29 To be entirely consistent with this conclusion from ATSDR (2021) and my own comments above, the EPA assessment
would further need to omit the highly uncertain RfD derivations based on PFDA associations with decreases in antibody
levels/vaccine response (Tier 1 necessary revision) reported in some epidemiological study results that appear overall
to be inconsistent, non-adverse, and uncertain in nature (e.g., causal attribution problems, unsupported by
epidemiological disease incidence studies).
119
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
multiple PFAS compounds may be occurring such that mixture-based approaches
should be explored for data analysis. The compound-by-compound approach used in
this study gives rise to significant uncertainty that precludes use of these data for
dose-response assessment and RfD derivation. It cannot be confidently said
scientifically that the presumed effects are due to PFDA exposure (e.g., PFOS and
PFOA were certainly co-exposures and also showed statistically significant
decreases for birth weight). Consistent with my opinion on the significant
uncertainty and confounding associated with these Wikstrom et al. results and the
weakness/limitations of the epidemiological evidence overall, Section F.3 of the
draft assessment states that, "In the six studies using mutually adjusted PFAS
approaches to address coexposures, there was not consistent evidence for birth
weight deficits associated with increased exposure to PFDA." (p. F-27, lines 13-14),
and acknowledges that "there is considerable uncertainty due to potential
confounding by co-occurring PFAS in the existing literature." (p. F-27, lines 30-32).
Indeed, significant co-exposure to multiple PFAS is not the exception but the rule,
and is just the condition to result in significant bias away from the null for adverse
effects.30 The high confidence study of Luo et al. (2021) is an example that serves as
a cautionary tale on the obvious importance of PFAS co-exposures, which reported
large statistically significant birth weight deficits (-97 g; -178, -16 per each In-unit
PFDA increase) in a single-pollutant PFDA model, but results were null and their
direction reversed in the multipollutant model with a nonsignificant increase in
birth weight associated with PFDA (Table F-2).31
As the data from epidemiological studies provide only "slight" evidence of
developmental effects, including data from Wikstrom et al. (2020) that appears
too uncertain and unsuitable for quantitative dose-response assessment, I
suggest use of the mouse data (Harris and Birnbaum 1989) for dose-response
assessment and osRfD derivation based on developmental growth effects (Tier 1
necessary revision). Unreliable/inadequate data (human or otherwise) cannot be
justified scientifically for use in dose-response assessment by how the associated
unreliable POD(s) compares to that based on reliable data definitively showing
cause-and-effect for unquestionably adverse effects, unconfounded by significant
co-exposures to similar chemicals.
30 To be a confounder, the co-occurring PFAS would need to be associated with both the PFAS of interest and the
outcome, but not an intermediate in the causal pathway; such PFAS would be considered positive confounders if
their effect estimate with the endpoint of interest is in the same direction as the primary PFAS of interest. If positive
confounders are not accounted for, the anticipation is that any resultant bias would be away from the null. (p. F-24,
lines 16-21). The stronger the association between co-exposures, the larger the effect size for the co-exposure of
interest (i.e., the greater the confounding) (p. F-25, lines 14-15).
31 The other high confidence study in Table F-2 (Starling et al. 2017) showed that adjustment for PFAS co-exposures in
the multipollutant model resulted in a large statistically significant increase in birth weight for PFDA (+97.5 g; 31.5,
163.6), which for some reason is not discussed in Section F.3 (PFDA and PFAS Coexposure Study Results) but serves as
yet another example of the obvious, commonsensical importance of accounting for PFAS co-exposures.
120
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Table 2. Prenatal exposure to eight PFA5, measured as maternal serum concentrations (ng/mL) during early pregnancy.
Compound
Geometric mean [95% CI]
Median (IQR)
95th %
LOD
Above LOD (%)
PF05
535 [521-5.50]
5.38 (3.97-7.60)
1034
0.06
100
PFOA
1.60 [156-1.65]
1.61 (1.11-230)
3.18
0.02
100
PFHxS
131 [128-1.35]
1.23 (0.86-199)
2.94
0.03
100
PFNA
054 [053-0.56]
0.53 (0.39-0.73)
1.02
0.01
100
PFDA
026 [026-0.27]
0.26 (0.19-034)
050
0.02
100
PFUnDA
021 [021-0.22]
0.23 (0.15-033)
0.44
0.02
995
PFHpA
0.018 [0X317-0.019]
0.02 (
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
readily identify the next most sensitive effects based on less problematic studies
that are adequate for RfD derivation. However, any newly derived toxicity factor
should be subject to external expert peer review (Tier 1 necessary revision).
c. As I do not agree with the selection of these studies/endpoints for RfD derivation,
please see my comments under subsection "b" above, many of which concern the
considerations of adversity and scientific support (as well as this question's
footnote). That being said, speaking more generally in regard to deriving a
lifetime RfD, lifetime RfDs must also be protective against adverse effects
occurring over a shorter duration of exposure, such as in childhood or during
development. Thus, if effects that occur over a shorter duration of exposure are
truly adverse, relevant to humans, and are the first to occur as the dose rate
increases (i.e., the critical effect), they can appropriately serve as the basis for a
lifetime RfD.
Lastly, since I suggest use of the mouse study of Harris and Birnbaum (1989)
above as the basis of an osRfD based on decreases in fetal body weight (with
various gestational exposures), I note that Figure 3-56 of the draft assessment
shows statistically significant and progressive decreases/trends in fetal body
weight with dose across a range that includes the >5% decrease historically used
by regulatory agencies as the demarcation of adversity in dose-response
assessment (i.e., as a >5% reduction in fetal body weight is usually considered
adverse, a benchmark response of 5% is typically used in benchmark dose
analysis), from -1% to -50% and including doses that resulted in decreases of -4%
and -6% (Table 3-23, p. 3-153). Furthermore, the changes in fetal body weight
were at doses not associated with maternal toxicity (p. 3-151, lines 5-6). Harris
and Birnbaum (1989) is a medium/high-confidence study (p. 3-150, lines 9-12).
d. Table 5-7 (pp. 5-13 to 5-18) contains the benchmark response (BMR) levels for
BMD modeling. For immuno-toxicity, the draft assessment indicates [emphasis
added] that in the absence of a clear definition of an adverse effect for a
continuous endpoint like antibody concentrations, a default BMR of 1 SD
change from the control mean may be selected, and a BMR lower than 1 SD
can be used if it can be justified on a biological and/or statistical basis (p. 5-5).
For decreased antibody tetanus vaccine response in children, the draft seems
to primarily use three considerations to justify a BMR of Vz SD (pp. 5-5 to 5-6):
(1) characterization as a developmental effect; (2) severity of the associated
disease for which the vaccine is designed to protect as a biological
consideration (but not severity/adversity of the effect itself); and (3) a BMR of
Vz SD results in 7.9% of the tetanus antibody distribution values being below a
0.1 lU/mL cutoff, which is 5.1% extra risk, showing that the generic BMR of Vz
SD can provide a reasonably good estimate of 5% extra risk (perhaps their
statistical consideration). Accordingly, the draft assessment: (a) is not using the
demonstrated severity/adversity of the effect itself (decreased antibodies) as
support for the lower BMR (i.e., Vz SD); and (b) is using an admittingly unclear
definition of an adverse effect (antibodies <0.1 IU/mL)32 in an attempt to
justify/support the lower BMR of Vz SD. (a) is problematic and (b) is internally
32 Per EPA's draft assessment (p. 5-5, lines 260-28).
122
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
inconsistent as the draft does not scientifically support the definition of <0.1
lU/mL as the adverse cutoff. These justifications/support for the BMR should
be removed (Tier 1 necessary revision).
Additionally, while the draft assessment suggests that specific immuno-toxic
effects observed in children may be broadly indicative of developmental
immunosuppression impacting these children's ability to protect against a
range of immune hazards (citing Selgrade 2007), this appears speculative on
EPA's part as the draft does not provide robust scientific support for such an
assumption. In fact, the draft does not even provide evidence that the
antibody decreases reported result in any inability for children to protect
themselves against the very diseases for which the antibody decreases are
reported (diphtheria and tetanus). This is true for both the study populations
as well as the U.S. population. For the U.S. general population, the CDC
reported that about 10% of children for ages 6- to 11-years old had tetanus or
diphtheria antibody titer measurements below 0.1 IU/mL33, and EPA argues
that even 0.1 IU/mL may not be protective (p. 5-6, lines 17-25). The implication
is that a significant percentage of U.S. children are at risk of contracting
tetanus or diphtheria. Yet as discussed in detail under question "3.b" in regard
to the adversity (or lack thereof) of the reported antibody effects, U.S.
surveillance disease incidence data do not support that serum PFDA (or any
other serum PFAS) is suppressing tetanus and diphtheria vaccine responses
and leaving people vulnerable to infection from these diseases. As a corollary
to EPA's reasoning, the lack of demonstrable real-world effects on the
incidences of these diseases when NHANES data show most blood serum levels
well above the PODs adjusted for intrahuman variability (see comments under
question "2.b") suggests that these specific effects observed in children
certainly may not be broadly indicative of immunosuppression impacting these
children's ability to protect against a range of other diseases (i.e., "immune
hazards"). In fact, this is consistent with the WOE from scientific studies on
PFDA and disease incidence, which is for no significant effects (again, see
comments under question "2.b" above). Also see comments above under
question "3.b" above citing rationale for the use of mouse data for reduced
fetal weight (Tier 1 necessary revision).
With all this being said, otherwise, the modeling approaches, model selection
process, and BMRs used to derive PODs for toxicity value derivation are scientifically
justified. Use of BMD modeling to the extent possible, guided by standard statistical
model fit criteria (+ visual inspection) for model selection, is essentially standard
scientific procedure inside (and outside) EPA (e.g., USEPA 2012a).34 Aside from the
33 CDC (Centers for Disease Control and Prevention). National Health and Nutrition Examination Survey, Immunizations.
Available at: https://www.cdc.gov/nchs/data/nhanes/databriefs/immuniza.pdf
34 An adequate fit is judged on the basis of a x2 goodness-of-fit p-value (p > 0.1), scaled residuals at the data point
(except the control) closest to the predefined BMR (absolute value <2.0), and visual inspection of the model fit. Among
all models providing adequate fit, the benchmark dose lower confidence limit (BMDL) from the model with the lowest
Akaike's information criterion (AIC) was selected as a potential POD when BMDL estimates differed by less than
threefold. When BMDL estimates differed by greater than threefold, the model with the lowest BMDL was selected to
account for model uncertainty (see Section C.2.1, p. C-24).
123
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
BMRfor immune effects (see comments above), reasonable scientific justifications
for the BMRs utilized are provided in Table 5-7 (pp. 5-13 to 5-18). Since I suggest
use of the mouse study of Harris and Birnbaum (1989) as the basis of an osRfD
based on decreases in fetal body weight (with various gestational exposures), I note
here that a 5% relative deviation BMR in markers of growth/development in
gestational studies (e.g., fetal weight) has generally been considered a minimally
biologically significant response level and has historical precedence for use as a
BMR (EPA 2012b, 2004, 2003).
e. Yes, the section Derivation of Candidate Lifetime Toxicity Values for the RfD (p. 5-
19) provides reasonable rationale for only considering liver effects data and
male/female reproductive effects data for subchronic osRfD derivation. More
specifically, p. 5-19 (lines 10-20) states [emphasis added]:
"For liver and male and female reproductive effects, quantitative information is
limited to studies in which animals were exposed for <28 days. For each of these
identified hazards, very little information is available to assess the extent to
which the specific changes caused by PFDA exposure for 28 days might be
expected to worsen with PFDA exposure for a lifetime. Separately, human
equivalent PODsfor these endpoints were much less sensitive (several orders of
magnitude) than the PODs for developmental and immune effects from the
epidemiology studies (see Table 5-9). As such, for liver, male reproductive, and
female reproductive effects, derivation of candidate lifetime values was not
attempted given the high degree of uncertainty associated with using PODs from
a 28-day rodent study to protect against effects observed in a chronic setting.
However, these endpoints were considered for the derivation of the subchronic
RfD (see Section 5.2.2)."
Taken together, these considerations provide a reasonable rationale to support
the decision not to attempt the derivation of lifetime osRfD values (only for
derivation of subchronic osRfD values). While the total composite uncertainty
factor for these effects is already considerable (1,000; Table ES-1), if lifetime
osRfDs based on these effects (i.e., liver, male and female reproductive effects)
are desired by EPA and/or considered useful for more complete risk assessments,
EPA could consider increasing the total composite uncertainty factor to the upper
end of their total uncertainty factor range (3,000) to estimate chronic osRfDs (Tier
2 suggestion).
f. Section 5.2.4 (p. 5-41, lines 35-36) states, "No studies examining inhalation
effects of short-term, subchronic, chronic or gestational exposure for PFDA in
humans or animals have been identified, precluding the derivation of an RfC."
Furthermore, it appears from Sections 3.1.6 and 3.1.7 of the draft assessment
that a reliable PBPK/PK model has not been identified for PFDA that could be
used for route-to-route extrapolation. If EPA has not identified a reliable PFDA
PBPK/PK model that could be used for route-to-route extrapolation, then EPA's
decision not to derive an RfC would be fully justified. However, this is not
explicitly stated in Section 5.2.4, which should clearly state whether or not EPA
has identified a reliable PFDA PBPK/PK model for route-to-route extrapolation
(Tier 1 necessary revision).
124
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Hoberman
The selection of the three studies covering the response to decreased
serum antibody concentrations for both tetanus and diphtheria in children
and reduced birth weight in humans to determine the RfD values for PFDA
is believed to be scientifically justified.
The modeling approach appears appropriate.
a. The modeling appears to be appropriate based on the nature of the endpoints and
fact that the changes in the endpoints occurred during development, a critical
period.
b. Not applicable.
c. The effects selected were appropriate for use in deriving the lifetime RfD
considering that the effects occurred during development of the immune system
and the effects were presumed continue throughout the lifetime of the affected
subjects. There are no long term or lifetime studies of these effects to confirm this
modeling.
d. The BMD modeling approaches, selection and justification of benchmark response
levels, and selection of the BMD models used to identify each POD for toxicity value
derivation scientifically were justified and clearly described.
e. Not deriving a lifetime organ-specific RfD for the liver, male and female
reproductive effects appears appropriate since there were no lifetime studies and
no fully supportive human studies for these endpoints.
The 28-day study in mice was supportive of subchronic os RFDs since four weeks of
exposure is an appropriate subchronic exposure.
f. With no inhalation studies of adequate quality to review not deriving an RfC
appears to be appropriate.
Leung
Toxicological modeling is not my area of expertise, and 1 defer this question to the other
external reviewers. From an overall perspective, 1 agree with the point made in the
panel meeting, that clearly conveying the additional data obtained from the authors of
the original papers would improve the clarity of the sections used to derive the RfD.
Zoeller
a. Derivation of the RfD values based on immune suppression and birth weight are
reasonable and the Agency has justified the use of the presented in Budtz-
Jorgensen and Grandjean and Wikstom. As the Agency points out, the endpoint of
decreased antidiphtheria and tetanus titers is not an adverse endpoint per se and
there is no minimal biological response that the Agency can identify to set the BMR.
Therefore, the choice of a Vz SD as discussed and justified in Appendix C.l.l. was
clear and reasonable. In addition, while this is an intermediate endpoint, it is clear
that diphtheria and tetanus represent potential severe including fatal infections that
can occur during development. Thus, there is significant justification for the use of
this intermediate endpoint to establish the RfD. The scientific support for this was
125
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
well-developed in C.l.l., including the recognition that these responses may be
broadly indicative of immune suppression.
Likewise, the use of Wikstrom (birth weight) was clear and scientifically justified. A
5% change was chosen as the minimal biological change leading to an adverse effect
because this endpoint occurs during a developmental stage and because this
endpoint is clinically predictive of health outcomes.
No recommendation.
d. This issue was clearly and thoroughly presented by the Agency. No
Recommendation.
e. It is reasonable that the Agency used these endpoints to derive subchronic RfDs
only. The endpoints captured for liver and male and female reproductive endpoints
are scientifically justified. No Recommendation.
f. The Agency was justified in not developing an RfC where no studies are available.
No Recommendation.
3,4 In addition, for PFDA, an MD for less-ttian-llfetlme ("subchror posures is derived. No
subchronic RfC was derived. The same studies and outcomes were chosen for use in
deriving the lifetime and subchronic RfDs, Are the selection of these studies and these
effects for the derivation of the subchronic MD for PFDA scientifically justified?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effect(s) that should be used to
support the derivation of the subchronic RfD and detail the rationale for use of such an
alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the
effects selected are appropriate for use in deriving the subchronic RfD, including
considerations regarding adversity (or appropriateness in representing an adverse
change) and the scientific support for their selection.
d. Please comment on the other subchronic osRfDs (i.e., for liver, male reproductive, and
female reproductive effects).
e. Given the lack of studies on inhalation exposure to PFDA, no subchronic RfC is derived.
Please comment on this decision.
Noncancer Toxicity Value Data Selection and Modeling
Reviewer
Comments
Adgate
Tier 3: Given that the same studies are used for both lifetime and subchronic RfDs, the
main issue is the significance of a short-term adverse effect that may or may not be
reversible relative to a lifetime adverse effect RfD derivation. Given the persistence of
PFDA leading to ~ lifetime exposure (even if declining) a subchronic RfD seems less
important and harder to understand than a lifespan RfD.
126
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
d. See above comment.
e. Tier 3: Given the lack inhalation data from animal or human studies and the
likelihood that ingestion is the dominant human exposure pathway it is reasonable
that no subchronic RfC is estimated. PFAS inhalation exposure, however, is an area
that has had little research, and some planned future studies will likely explore this
topic and the potential for precursors that are often present in the environment to
transform to PFDA. At this time an analysis of PFDA physical constants and potential
resulting air concentrations in indoor environments would be informative for
decision-makings given PFDA's presence in many consumer products.
Carignan
a. Use of the same studies to derive a subchronic RfD is reasonable and scientifically
justified given the persistence of PFDA.
d. The rationale and approach provided is reasonable.
e. In the absence of data it is reasonable not to derive a subchronic RfC for PFDA.
For PFASs with similar properties (e.g., long chain) read-across could be
considered where PFAS-specific data is unlikely to be generated. [Tier 2]
Faustman
a. This reviewer supports the inability to derive a meaningful subchronic RfC. This
reviewer does support the development of subchronic RfDs and in particular is
supportive of the choice of human studies for both the immunotoxicity and
developmental endpoints and likewise supportive of the choice to use the animal
study to derive the RfDs for hepatotoxicity and male and female reproductive
toxicity. These choices were well described and resulted from the evidence
summary tables at the end of each of the toxicity endpoint sections as well as the
relevant endpoint discussions in Section 5.
b. No alternative suggestions.
c. Section 5.2.1 details the choice of the studies used by USEPAto proceed for toxicity
value calculations. Table 5-2 details the immune antibiotic response data sets that
were used for the deriving subchronic endpoints for immunotoxicity. For example,
the combined birth cohort data was used for the POD derivation for the human
immunotoxicity endpoint as the log-log transformed data could not be used for
calculating the BMDs (see notes above about availability of the original data in
order to do these calculations). Examples for developmental toxicity outcomes are
provided in Table 5-3 and two studies Valvi et al 2017 and Wikstrom et al 2020 both
high confidence epi studies are available and were modeled. Animal studies were
cited as being supportive for the choice of these endpoints to model and Table 5-4
notes one animal study (Harris and Birnbaum, 1989) as modeled for decreased fetal
body weight. Notes provided in these tables summarize the rational for choosing
these studies for toxicity value calculations.
d. For hepatic effects, Table 5-1 details the endpoints for which PODs are derived for
the high confidence studies. Four endpoints were chosen and included increased
serum ALT and ALP values (NTP, 2018) as well as increased liver weight (NTP, 2018
and Fawley et al 2018) from two high confidence rodent studies. Rationale for
127
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
choosing these endpoints included the presence of dose response as well as
consistency across studies and biological significance of endpoints affected. Male
reproductive endpoints are presented in Table 5-5 and two endpoints with dose
response data for decreased absolute epididymis sperm counts and Leydig cell
atrophy were chosen from the NTP, 2018. Note that only the Leydig cell atrophy
endpoint study was identified as a high confidence study but the epididymis sperm
count endpoint from the same study was identified as low confidence due to
potential insensitivity however both endpoints were identified as biologically
important and both endpoints were modeled. Table 5-6 lists four endpoints from
the NTP, 2018 study that are available for female reproductive endpoints and all but
one endpoint are modeled due to the studies being high confidence, of biological
relevance and of interest due to the presence of dose-response data. Increased
testosterone as an endpoint was not modeled as the toxicological significance of
this endpoint in this study was unclear. This reviewer supports all of these choices.
e. This reviewer supports this lack of action due to lack of relevant studies to model.
Fisher
a. This has been done before, using the same study for subchronic and lifetime
exposures. It is not a great position to be in, but this is the state of the toxicology for
PNDA. The chronic RfD is more uncertain than the subchronic.
c. Table 5-16 lists organ specific endpoints for subchronic RfD consideration. The table
and text provide good descriptions about the evidence base supporting the toxicity,
along with confidence in the POD for a human and a summary confidence
statement based on these factors. If there are in vitro studies supporting the
toxicity endpoints (Confidence in evidence base supporting this hazard), it is
worthwhile to include. Tier 2.
d. There is a good discussion about these endpoints relative to an organ specific RfD.
The liver is the most studied endpoint in animals. The use of human data
(immune/developmental) is preferred by me over animal data.
e. This is the proper course of action. See comment above about exposure and the use
of MPPD simulation software.
Georgopoulos
a. The rationale for selecting the same studies and outcomes for deriving the
lifetime and the subchronic RfDs for PFDA is appropriate and scientifically
justified. As stated in the Toxicological Review (page 5-29, lines 27-32):
"Datasets considered for the subchronic RfD were based on endpoints advanced
for RfD derivation in Table 5-8. Given that the developmental and immune
effects were observed in humans exposed to PFDA during susceptible lifestages
(postnatal growth/development and immune system effects in children at ages
5-7), these endpoints were also considered for the derivation of candidate
subchronic values, applying identical uncertainty factors to those used for the
lifetime candidate values."
b. N/A.
c. The appropriateness of the selected effects in representing adverse changes is
128
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
discussed adequately in the corresponding sections of the Toxicological
Review.
d. The studies and effects selected for deriving other subchronic osRfDs (i.e., for
liver, male reproductive, and female reproductive effects) are also
appropriate; their relevance in representing adverse changes is discussed
adequately in the corresponding sections of the Toxicological Review. It should
be recognized that the available evidence base is the determining factor
limiting the ability to derive osRfDs. For example, regarding liver effects, the
Toxicological Review (page 3-49, lines 29-32) states:
"The evidence base is limited in that there is an absence of studies via
relevant exposure routes with durations longer than 28 days (i.e., no
subchronic and chronic exposure studies) examining potential hepatic effects
of PFDA exposure."
Given such constraints, the selected studies and effects represent the best
available options.
e. The decision to not derive a subchronic RfC is the only reasonable option since
no relevant inhalation exposure studies to PFDA have been identified.
Haney
This question states that, "The same studies and outcomes were chosen for use in
deriving the lifetime and subchronic RfDs." However, question "3.e" states, "the
derivation of lifetime organ-specific (os) RfD values was not attempted for liver, male
reproductive and female reproductive effects." Therefore, while correct for the final
RfD value, this question (4) would be incorrect in regard to candidate RfDs as liver and
male/female reproductive effects were not chosen for use in deriving osRfDs for
consideration in selecting the lifetime RfD. The cited sentence of this question "4" was
not abundantly clear and open for interpretation, hence this comment.
a. As with the lifetime RfD, 1 do not agree with the selection of these same studies and
endpoints for subchronic RfD derivation (i.e., decreased serum antibody
concentrations in Grandjean et al. 2012, decreased birth weight in Wikstrom et al.
2020). Please see my comments under subsection "b" below, which 1 feel compelled
to reiterate here.
b. Immune Effect Endpoint
EPA indicates that the hazard judgment that "the evidence indicates that PFDA
exposure is likely to cause adverse immune effects, specifically
immunosuppression, in humans" (p. 3-92, lines 35-37) is driven primarily by
consistent evidence of reduced antibody response from human epidemiological
studies (mostly from two birth cohort studies)..." (pp. 3-92 and 3-93, lines 37-1).
However, it is not clear that EPA has sufficiently considered adversity for the
primary basis of this hazard judgment, or fully considered the weaknesses/
limitations of this evidence. For example, even without considering potential
confounding, 3 out of 4 ORs for PFDA and antibody concentrations falling below
the protective level of 0.1 lU/mL for tetanus and diphtheria in children ages 5
years (n=510) or 7 years (n=386) contain 1, indicating that the WOE from this key
129
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
study cohort is for no statistically significant associations with less-than-protective
serum antibody concentrations in children (see eTable 4 of Grandjean et al.
2012).35 Consistent with this, the more recent Grandjean et al. (2017a) study
states [emphasis added] that, "With many antibody concentrations being close to
the assumed clinically protective level of 0.1 lU/mL, logistic regression showed
only weak tendencies for antibody levels below the limit to be associated with
serum PFAS concentrations."36 Also consistent with a WOE for no statistically
significant associations (much less being able to say "effects" as there are
problems with causal attribution to PFDA), 6 of 8 confidence intervals (95% CIs)
for both tetanus and diphtheria serum antibodies included 0% change per 2-fold
increase in maternal and age 5 serum PFDA (see Table 3 of the study).
Additionally, in Grandjean et al. (2012), PFDA was correlated with other PFAS
(e.g., PFOS, PFOA, PFNA) that had some associations with antibody
concentrations falling below the protective level of 0.1 lU/mL (see Table 2 and
eTable 4 of Grandjean et al. 2012).37
Moreover, the level of serum antibodies corresponding to a clinically protective
level appears to be assay specific. For the ToBI assay apparently used in the Faroe
Islands studies, >0.01 lU/mL is considered to be the clinically protective level, not
the value of >0.1 lU/mL indicated by study authors. Considering the comments
above, this means that the reported decreases in serum antibodies are even less
likely to be biologically/clinically significant. This is not surprising given the rarity of
tetanus/diphtheria cases, particularly in those fully vaccinated, and the WOE for
PFDA not being associated with statistically significant increases in the incidences of
diseases based on the epidemiological literature (all discussed below). To say the
least, all this brings into serious question the validity of the EPA's assumptions
regarding the clinical relevance/adversity of these serum antibody endpoints.
The clinically protective level cited by Grandjean et al. was >0.1 lU/mL. However:
• Grandjean et al. (2012) reported that "serum concentrations of antibodies
against the tetanus toxoid were measured in coded samples by the Statens
Serum Institut using enzyme-linked immunosorbent assay...", citing
Hendriksen et al. (1988);
• Hendriksen et al. (1988) describes the ToBI assay, which is a modified ELISA;
and
35 The confidence interval (CI) for the one statistically significant OR of 1.36 (age 5, diphtheria) is (1.04,1.77), with the
lower end of the CI practically equal to 1 (eTable 4 of Grandjean et al. 2012).
36 Also, while Grandjean et al. (2017b) state that, "At age 5,152 (44%) children had antibody concentrations lower than
the protective level of 0.1 lU/mLfor diphtheria and 126 (36%) for tetanus", this appears inconsistent with Table 1 of
that study, which shows that the 25th percentiles for diphtheria and tetanus serum antibody concentrations were 0.1
lU/mL
37 A 2-fold increase in PFOS and PFOA concentrations at age 5 years was associated with odds ratios between 2.38 (95%
CI, 0.89 to 6.35) and 4.20 (95% CI, 1.54 to 11.44) for falling below a clinically protective level of 0.1 lU/mLfor tetanus
and diphtheria antibodies at age 7 years.
130
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• WHO (2017) indicates that for a modified ELISA, clinical protection is achieved
at >0.01 lU/mL, not >0.1 lU/mL as indicated by Grandjean et al.38
That is, the WHO (2018, 2017) cites Hendriksen et al. (1988) as "a toxin binding
inhibition (ToBI) assay has been reported and demonstrated to show good
correlation with the neutralization assay (correlation coefficient = 0.95)", and for
modified ELISA assays such as this further indicates [emphasis added] that, "The
minimum amount of circulating antibody that, in most cases, ensures immunity to
tetanus is assay-specific. Using in vivo neutralization tests or modified enzyme-
linked immunosorbent assays (ELISA), concentrations exceeding 0.01 lU/ml are
usually considered protective, whereas antibody concentrations of at least 0.1-0.2
lU/ml are defined as protective when using standard ELISA techniques." WHO
(2018) also discusses and illustrates the timing of primary and booster vaccinations
and durations of protection in the context of the minimum putatively protective
level of 0.01 IU/mL.39 Thus, the protective level cited by Grandjean et al. for the
assay used is 10-fold higher than the protective level cited by WHO (2017), even
further calling into question the biological/clinical significance and adversity of the
reported results.
In regard to confounding and the ability to causally attribute associated effects to
PFDA:
• Grandjean et al. (2012) state [emphasis added], "Although all of the 5 PFCs
measured showed negative associations with antibody levels, the
overlapping confidence intervals and the lack of comparative toxicology
studies prevent inference in regard to causal attribution... PFOS (most likely
the linear isomer) and PFOA appear to be the main culprits."
• The more recent Grandjean et al. (2017a) study states [emphasis added],
"Owing to the intercorrelations between the serum PFAS concentrations,
further analysis of the possible role of individual PFASs was not pursued, and
the observed associations may reflect the effects of the PFAS mixtures."
• Similarly, Grandjean et al. (2017b) state [emphasis added], "The close
correlations prevented meaningful adjustment for concomitant PFAS
38 For example, Grandjean et al. (2012) refers to a "clinically protective level of 0.1 IU/mL" several times and further
states, "An antibody concentration greater than 0.1 IU/mL is considered an important indicator of protection in
accordance with the public health rationale for routine vaccinations."
39 WHO (2018; pp. 14-15) states [emphasis added] that "To illustrate the kinetics of immunity among children >1 year,
adolescents and adults following primary and booster vaccination with TTCV, Figure 2 provides a schematic
diagram of the typical response. A single dose of TT in the absence of priming induces little, if any, protection. Two to
four weeks after the second dose, the mean level of tetanus antitoxin typically exceeds the minimum putatively
protective level of 0.01 IU/mL. One year after the second dose, the mean antibody levels are expected to decline and
may fall to the protective threshold level. After each subsequent dose of vaccine, immunity is boosted, then persists
above the protective threshold for a time, and then wanes over time. Putatively protective levels of immunity are
induced by a primary series of three TTCV doses and immunity typically persists for at least 5 years. After the third
dose, each additional booster dose given after at least a one-year interval increases tetanus antitoxin levels and further
prolongs the duration of immunity. Immunity may persist for approximately 10 years after the fourth dose of TTCV and
for at least 20 years after the fifth dose."
131
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
exposures."
Thus, it appears that effects may neither rise to the level of adversity nor be
attributable specifically to PFDA. Co-exposures to other PFAS (at a minimum) that
are not/cannot be adequately accounted for in the analyses are likely to be
significant confounders in these epidemiological studies, especially because PFAS
exposures are correlated, they are chemically-similar compounds, and there
appears to be little variation in exposure (i.e., low exposure contrasts) for the
single PFAS being assessed (e.g., Table 2 of Grandjean et al. 2012, Table 1 of both
Grandjean et al. 2017a and 2017b). For example, Grandjean et al. (2012) shows
that PFOA and PFOS had a correlation coefficient of 0.5 in the blood sera of 5-year
olds and IQR differences in blood sera concentrations of less than 1.6-fold each
(e.g., 75th percentile blood concentration of PFOA/25th percentile blood
concentration of PFOA), and PFDA and PFNA had a correlation coefficient of 0.78
in the blood sera of 5-year olds and IQR differences in blood sera concentrations
of less than 1.9-fold each (see Table 2 of the study).
Despite Grandjean et al. (2017b) stating that the close correlations prevented
meaningful adjustment for concomitant PFAS exposures, Budtz-J0rgensen and
Grandjean (2018a) attempts to do just that for benchmark dose, and the results
appear to help demonstrate the effects of confounding co-exposures and/or the
inability to properly adjust for them. Table 2 of Grandjean et al. (2017b) reports
the change (in percent) of the pre-booster serum-antibody concentrations at age
5 years associated with a doubling of the serum concentration of PFDA. Results
for cohorts 5, 3, and joint results are in the negative direction for tetanus at age 5
but none are statistically significant, which already points to the unreliability of an
effect having been demonstrated by these results and the unreliability of any RfD
based on these results or association. Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports benchmark results for the five prenatal PFAS concentrations in
regard to antibody concentrations at age 5 years (pre-booster) both unadjusted
and adjusted for PFOS/PFOA co-exposures. For tetanus antibodies, while
unadjusted benchmark doses for three models for PFDA (linear, piecewise,
conservative) appear to show excellent agreement (BMDs of 0.11-0.25 ng/mL)
with insignificant reliance on choice of model, when adjusted for PFOS/PFOA the
three models' best estimates of the PFDA serum concentrations associated with a
5% change go to infinity (although BMDLs can still be estimated). For diphtheria,
Table 2 of Grandjean et al. (2017b) reports statistically significant changes for
cohort 3 and joint results for the pre-booster serum-antibody concentrations at
age 5 years associated with a doubling of the serum concentration of PFDA.
However, similar to results for tetanus, Table 2 of Budtz-J0rgensen and Grandjean
(2018a) reports that when benchmark results are adjusted for PFOS/PFOA, two of
the three models' best estimates of the PFDA serum concentrations associated
with a 5% change go to infinity (although BMDLs can still be estimated). While
Table 1 of Budtz-J0rgensen and Grandjean (2018a; benchmark results for the age-
5 serum concentrations of five PFASs in regard to tetanus and diphtheria antibody
concentrations at age 7 years) provides no benchmark doses for PFDA that go to
infinity, Table 2 results point to the unreliability of this endpoint and should be
considered along with the other issues raised above. For example, the odds ratios
for PFDA and inadequate antibody concentrations for diphtheria and tetanus at 7
132
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
years was not statistically significant (see eTable 4 of Grandjean et al. 2012).
Key BMD results in the draft assessment itself demonstrate the importance of co-
exposures to other PFAS. Tables C-l and C-3 of the draft assessment (below)
provide BMD results for the critical effects used for RfD derivation (e.g., see Table
ES-1).
Table C-l. Results specific to the slope from the linear analyses of PFDA
measured in serum at age 5 years and log2(tetanus antibody concentrations)
measured at age 7 years in a single-PFAS model and in a multi-PFAS model
from fBudtz-lorgensen and Grandiean. 2018hV
Exposure
Model
shape
PFOS &
PFOA
adjusted
Slope (p)
per ng/mL
in serum
SE(p)
ng/m Lin
serum
Slope (p) fit
Lower bound
slope (pLB) per
ng/mL in serum
PFDA at Age 5
Linear
No
-1.55
0.602
p = 0.01
-2.55
PFDA at Age 5
Linear
Yes
-0.98
0.681
p = 0.15
-2.10
PFOS and PFOA had correlation coefficients of 0.39 and 0.35 with serum PFDA at
age 5, respectively (Table 2 of Grandjean et al. 2012). Despite these relatively low
correlation coefficients (Mukaka 2012), Table C-l shows that just controlling for
co-exposures from these two PFAS (PFOS, PFOA) resulted in significant impacts on
slope ((B) and slope fit. The slope estimate for PFDA was reduced 37% and PFDA is
no longer a significant predictor of tetanus antibody concentrations (p=0.15). The
most likely explanation (Occam's razor) is classic confounding as PFOS and PFOA
are documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean
2018a), and the existence of some chance that correction for these co-exposures
could create some confounding is not a scientifically robust justification for
dismissing the important implications of the results of adjustments for
PFOS/PFOA that the study authors themselves (Budtz-J0rgensen and Grandjean
2018a) thought it important to adjust for, and with good reason. These results
demonstrate the statistical unreliability of serum PFDA predicting tetanus
antibody concentrations when just two other PFAS are controlled for. Table C-3
concerns the critical effect for the RfD based on diphtheria antibody
concentrations.
Table C-3. Results specific to the slope from the linear analyses of PFDA in
serum measured at age 5 years and log2(diphtheria antibodies) measured at
age 7 years from Table 1 in a single-PFAS model and in a multi-PFAS model
from fBudtz-lorgensen and Grandiean. 2018bV
Exposure
Model shape
PFOS & PFOA
adjusted
Slope ((3) per
ng/mL in
serum
SE(P)
ng/mL in
serum
Slope (P) fit
Lower bound
slope (pLB)
per
ng/mL in
serum
PFDA at Age 5
Linear
No
-0.894
0.561
p = 0.11
-1.82
PFDA at Age 5
Linear
Yes
-0.297
0.635
p = 0.64
-1.35
The implications of these results are worse. First, even when evaluated alone
without accounting for co-exposures to relatively low correlated PFOS and PFOA,
serum PFDA is not a significant predictor of diphtheria antibody concentrations
(p=0.11). Table C-3 further shows that controlling for co-exposures from these
133
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
two PFAS (PFOS, PFOA) resulted in significant impacts on slope ((B) and slope fit.
The slope estimate for PFDA was reduced 67% and serum PFDA became a worse
nonsignificant predictor of diphtheria antibody concentrations (p=0.64). The most
likely explanation (Occam's razor) is classic confounding as PFOS and PFOA are
documented immunotoxicants (e.g., per Budtz-J0rgensen and Grandjean 2018a),
and the existence of some chance that correction for these co-exposures could
create some confounding is not a scientifically robust justification for dismissing
the important implications of the results of adjustments for PFOS/PFOA that the
study authors themselves (Budtz-J0rgensen and Grandjean 2018a) thought it
important to adjust for, and with good reason. Thus, when co-exposures are taken
into account for two modestly correlated PFAS (PFOS, PFOA), serum PFDA is not a
significant (i.e., reliable) predictor of these critical effects serving as the basis of
the RfD (i.e., decreases in serum tetanus and diphtheria antibody
concentrations).40
Moreover, Budtz-J0rgensen and Grandjean (2018a) also controlled for PFOS or
PFOA when deriving BMDs/BMDLs for the other, and EPA did use those co-
exposure-adjusted results for RfD derivation (i.e., PFOA adjusted for PFOS, PFOS
adjusted for PFOA; see Tables B-l and B-2 in Sections B.l.l and B.1.2 of the PFOA
and PFOS draft assessments USEPA 2021a,b) without expressing any similar
concerns about creating confounding by adjusting for these co-exposures
(low/moderate correlation coefficient of 0.50 (Mukaka 2012); see Table 2 of
Grandjean et al. 2012) or citing Weisskopf et al. (2018) and Weisskopf and
Webster (2017). Rather, these concerns have been selectively cited for PFDA in an
attempt to provide some rationale for dismissal of the co-exposure-controlled
results (e.g., slope ((B) values are reduced; serum PFDA is a nonsignificant
predictor of tetanus and diphtheria antibody concentrations) and thus for
selection of BMDLs uncontrolled for PFOS and PFOA co-exposures (see Tables C-2
and C-4 of the draft), but this is not a scientifically robust rationale and is
furthermore inconsistent with EPA's draft assessments for PFOA and PFOS (USEPA
2021a,b). Lastly, it is further noted that Weisskopf et al. (2018) indicates: (1)
sometimes, depending on causal structure, the inclusion of multiple exposure
variables in a model can amplify the amount of bias in a regression estimate
compared to analyzing single exposures; and (2) this potential amplification of
biases increases with stronger correlations between mixture components. To
demonstrate that this can occur in some cases, the study authors used "highly
correlated exposures" (e.g., r2=0.9), whereas the correlation coefficients between
PFDA and PFOS/PFOA are low (0.35 and 0.39, respectively). The results of
Weisskopf et al. (2018) do not constitute reasonable doubt that for these PFDA
results, the confounding from not adjusting for co-exposures to documented
immunotoxicants (PFOS, PFOA) is significantly greater than the potential
amplification of biases that remains undemonstrated under the same or similar
circumstances.EPA should reevaluate the issues raised above in regard to
implications for their draft PFDA assessment (Tier 1 necessary revision).
40 Knowing this, disparate results are not particularly surprising, such as the 18.7% increase in tetanus antibodies
predicted for children (age 13) with a 2-fold increase in serum PFDA based on the same study and type of analysis used
for the RfD critical effects (see Table 3-12, p. 3-59).
134
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Regarding potential immunosuppressive effects by PFDA, effects that rise to the
level of adversity would be expected to result in increased incidences of disease,
reflecting lower immunity and lower resistance to disease in the real world.
However, consistent with the WOE for no statistically significant associations with
antibody concentrations falling below the protective level of 0.1 lU/mL based on
results from Grandjean et al. (2012), almost all ORs in Table 3-13 of the draft
assessment include 1, indicating that the WOE from studies on PFDA and
infectious disease in humans is for no statistically significant associations.
Consistent with this, host resistance was unaffected by PFDA based on the limited
animal study data available (p. 3-73, lines 8-9), and host resistance assays are
considered highly relevant to the evaluation of immunotoxicity in the context of
human health assessment (p. 3-72, lines 15-18; IPCS 2012). EPA should reevaluate
the adversity of these presumed antibody level effects, including the association
with PFDA itself, and do so within the context of potential confounding, other
limitations, and available human/animal data on disease incidence (Tier 1
necessary revision), as this has important implications for the hazard judgment,
the strength of human evidence descriptor for immunosuppression (listed as
"moderate" in Table 3-19), and whether an RfD should be based on these effects.
Moreover, it is noted that the PODs for immunosuppressive effects from these
epidemiological studies range from 2.57E-04 mg PFDA/L blood serum to 7.02E-04
mg PFDA/L blood serum (BMDLi/2sd values from Table 5-8, pp. 5-16 to 5-17) that
when intrahuman variability is considered (through application of a UFH of 10)
results in values of 2.57E-05 to 7.02E-05 mg PFDA/L blood or 0.0257 to 0.0702 ng
PFDA/L blood serum.41 Data from NHANES show that GMs representative of the
U.S. population are well above these blood serum levels (See Appendix A NHANES
tables).42 Most notably, 2005-2018 population GMs range from 0.154-0.355 ng/L,
which are 2.2- to 13.8-fold higher than the PODs adjusted for intrahuman
variability (cited above). Despite these exceedances, tetanus and diphtheria
appear to be quite rare in the U.S. population. The average annual number of
tetanus cases in the U.S. from 2009-2018 was 29, with the CDC attributing most
cases to individuals who either have not been vaccinated or who are not current
on their boosters (e.g., only 3% of the cases from 2001-2008 were in people who
had received a complete tetanus toxoid series with the last dose within 10 years;
Tiwari et al. 2021). Tetanus also appears rare in U.S. children specifically,
occurring primarily in older adults. Per Liang et al. (2018):
"During 2001-2016, three neonatal tetanus cases and 459 non-neonatal tetanus
cases were reported to the National Notifiable Diseases Surveillance System
(NNDSS). The median age for non-neonatal cases was 44.0 years (range: 2-95
years)... The risk for both tetanus disease and mortality was higher among
persons aged >65 years than among persons aged <65 years. Tetanus occurs
almost exclusively among persons who are unvaccinated or inadequately
41 Based on BMDi/2sd values (Table C-9, p. C-16), these values are 0.0385 to 0.226 ng PFDA/L blood serum.
42 See NHANES Biomonitoring Data Tables at https://www.cdc.gov/exposurereport/data tables.html. Budtz-J0rgensen
and Grandjean (2018a) also acknowledge that, "Our BMDL results, both before and after adjustment are generally
below current exposure levels..."
135
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
vaccinated or in those whose vaccination histories are unknown or uncertain."
The incidence of U.S. diphtheria cases is even more rare. The CDC reported only
14 cases from 1996 through 2018 (Acosta et al. 2021). Thus, consistent with the
lack of statistically significant effects from the epidemiology study data discussed
above and despite the NHANES blood serum data showing exceedances of the
draft assessment human PODs adjusted for intrahuman variability (e.g.,
toxicodynamic) for most of the U.S. population for a prolonged period of time
(see the 50th percentile concentrations in Appendix A), U.S. surveillance disease
incidence data are not supportive of adversity. That is, U.S. surveillance disease
incidence data do not support that serum PFDA (or any other serum PFAS) is
suppressing tetanus and diphtheria vaccine responses and leaving people
vulnerable to infection from these diseases. The apparent lack of adversity/
consequence for the effects reported for tetanus and diphtheria certainly does
not provide support for an expectation of adversity/consequence for other effects
not measured/observed (e.g., for vaccines for other diseases and their
incidences).
Additionally, it appears that EPA has not fully considered all the relevant evidence
or the weaknesses/limitations of the epidemiological evidence, which in turn are
relevant for the hazard judgment and whether an RfD should be derived based on
these effects. Table 3-19 indicates that human data provide "moderate" evidence
and that "the inconsistent and low confidence evidence on infectious disease did
not influence this judgment." However, this points to the fact that EPA has not
duly considered the implications of the null findings on human and laboratory
animal infectious disease and other relevant considerations (e.g., some discussed
above) for the scientific WOE, which is not a scientifically supportable approach
as it does not consider all relevant data, directly relevant human data in
particular. EPA should consider such null findings (and other relevant
considerations) in their WOE (Tier 1 necessary revision). Combined with the
"slight" human data for sensitization and allergic response, the "slight" laboratory
animal data for immunosuppression, and the "indeterminate" animal data for
sensitization and allergic response (Table 3-19), it does not appear that PFDA
exposure is "likely to cause" adverse immune effects in humans is sufficiently
supported. EPA should reevaluate this determination (Tier 1 necessary revision)
as "may cause" might very well be the better supported hazard judgement, and
also reconsider their use of the serum antibody endpoints for quantitative risk
assessment/derivation of toxicity factors (Tier 1 necessary revision). This would
appear more consistent with the data discussed above and recent conclusions by
the Australian government (FSANZ 2021) and the U.S. Agency for Toxic
Substances and Disease Registry (ATSDR 2021).
The Australian government (FSANZ 2021) has concluded that associations of PFAS
with immunological endpoints do not provide a suitable basis for quantitative risk
assessment:
"In summary, new epidemiological studies provide some evidence of statistical
associations between PFAS blood levels and impaired vaccine response,
increased susceptibility to infectious disease and hypersensitivity responses.
136
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
However the data are insufficient to establish causal relationships and it cannot
be ruled out with reasonable confidence that the observed statistical
associations may have been due to confounding, bias or chance. On the basis of
the uncertainties and limitations in the evidence base, immunomodulation is
not currently considered suitable as a critical endpoint for quantitative risk
assessment for PFAS."
Similarly, ATSDR (2021) found the epidemiology literature inadequate for use as
the basis of deriving minimal risk levels (MRLs) for PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive targets
for many of the perfluoroalkyls; however, there are two major limitations to
establishing dose-response relationships for these effects and using the
epidemiological studies to derive MRLs: accurate identification of
environmental exposure levels producing increased risk for adverse effects
(exposure estimates and routes of exposure) and likely co-exposure to mixtures
of perfluoroalkyls. Other limitations include the cross-sectional design of the
majority of epidemiological studies and the potential that reverse causality
contributes to the observed associations... In summary, the epidemiological
databases for several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with adverse
effects and possible interactions between compounds preclude use of these
data to derive MRLs."
Based on the information in the draft assessment and reviewed elsewhere, this
reviewer finds it difficult to disagree with the recent conclusions of the Australian
government (FSANZ 2021) and ATSDR (2021) that the epidemiology literature
(e.g., on PFAS blood levels and impaired vaccine response) is inadequate for
quantitative risk assessment and use as the basis for deriving toxicity factors (e.g.,
RfDs). Unreliable/inadequate data (human or otherwise) cannot be justified
scientifically for use in dose-response assessment by how the associated
unreliable POD(s) compares to that based on reliable data definitively showing
cause-and-effect for unquestionably adverse effects, unconfounded by significant
co-exposures to similar chemicals.
The above comments provide relevant rationale for use of an alternative
study(ies) or effect(s) for subchronic RfD derivation. While the draft assessment
provides various but limited candidate PODs, due consideration of the comments
above appears to require a more complete evaluation of alternative
studies/effects (Tier 1 necessary revision). One method to begin such an
evaluation may entail sorting the study data extracted by duration, then
NOAEL/LOAEL, BMD or a similar criterion that would allow EPA to readily identify
the next most sensitive effects based on less problematic studies that are
adequate for subchronic RfD derivation. However, any newly derived toxicity
factor should be subject to external expert peer review (Tier 1 necessary
revision).
Birth Weight Endpoint
EPA's hazard conclusion for developmental effects primarily relies on in vivo animal
137
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
data (i.e., the gestational exposure mouse study of Harris and Birnbaum 1989). Based
on over 45 different epidemiological studies included in the draft assessment, the
evidence of an association between PFDA exposure and developmental effects in
humans is considered only "slight" (p. 3-156, lines 10-11). Despite the "slight"
evidence in humans, Table ES-1 indicates that RfDs were nevertheless calculated
based on decreased birth weight in male and female children (Wikstrom et al. 2020).
While data in humans, as the species of ultimate interest, are usually preferred as the
basis for derivation of toxicity factors, in this case it appears that the mere "slight"
totally of the evidence for developmental effects from the over 45 different
epidemiological studies included in the draft assessment should not be considered
sufficient for quantitative dose-response assessment, but rather only for supportive
information for hazard identification. EPA Figure 3-28 (p. 3-108), reproduced below,
helps illustrate the weakness and incoherence of the overall epidemiological
database for demonstrating decreased birth weight, particularly statistically
significantly decreased birth weight, associated with PFDA levels and how no one
epidemiological study could possibly be representative of these inconsistent and
disparate results or the results of a meta-analysis of relevant studies.
138
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
UT"V cr* Jrtit!*.
CnjKekw Cvrfy•!j«
IfHwixmiifi
alv.'W M
IMmrit WVl'»*n Wp\ »t
4|»W *•!<¦ i
^ «' v - . u « . fm ... J.^
1135 P|0»P"<|
* -sri} liio.' KwUi/nli Cm
&MMtairwh»l*tLor< Ln
bniunl CHUT: UrJr
0(«X4I Cc»] Twtft*
huiitHM.:.
h*r*
Figure 3-28. Overall study population mean birth weight results for 22 PFDA
epidemiological studies3 c. (results can be viewed by clicking the HAWC link).
Abbreviation: BWT = Birth Weight
*¦ Studies are sorted first by overall study confidence level then by Exposure Window examined,
k Mens et al. (20181 pooled samples from umbilical cord blood and maternal pSasma during the first and second
trimesters. The remaining studies were all based on either one umbilical or maternal sample.
c If a study presented regression coefficients for continuous exposure with multiple exposure units, only one unit
change is shown (e.g., (Bach et al.. 2016). with the exception of [Li et a I.. 2017), which displays both IQ.R and In-
unit (ng/mL] values.
d- The results displayed here for mean birth weight among 587 overall population participants in the POPUP Cohort
are from a larger population of participants fSwedish Environmental Protection Agency. 2017) compared to a
sample size of 3SI in their 2018 publication Gyilenhammar et al. (20181.
*• Xu et al. (2019a) results are truncated for the 210.7 gram increase; the complete 95% CI ranges from -314.3 to
735.8 grams.
Similarly, Figure 3-26 (p. 3-104) below illustrates how the results of Bach et al. (2016),
the other high confidence, prospective cohort for trimester 1-2 with adequate study
sensitivity, had a nonmonotonic response, no statistically significant quartile results
for the overall population, and a regression coefficient of +0.03 per each In-unit
(ng/mL) (compared to -0.147 per each In-unit increase for Wikstrom et al. 2020), all
of which does not support use of the Wikstrom et al. (2020) study results. The Bach
et al. (2016) and Wikstrom et al. (2020) studies are the first two entries in Figure 3-
26, respectively.
139
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Study Population Overall Study Study sensitivity Design Exposure Regression Exposure
Confidence Window Coefficient Comparison Rejection cswtrkietii
Bachetai.2010. A«mwBnhC«ioii20M-M13i. |H«N Adwuals Cof-xl TnmMWrl-2 -0.1 Quarts 2
¦0.13 Ouwffle3
0.02 Quaroie4
O.OS livjillngrrCt
Wiiewiretai SElM/\«007-2010iSwiwen 1533 |Hig>y Anenume Cowt TnmeaeM-2 -0077 QuaniieZ
2020,6311677 moUiet-trfant pa is (Prospective)
-0-179 fXiarrHe4
•0147 in-um (nyn.)
Wanjetal .2016 Tam»n Mawn« ana InHMt Cofwt |H«n| AONjjMf Conor! Tr«ster3 0.04 In-unlWo'mO
M4SV>? Study (2000-2001 J, 223 |Pro»p«cllv»l nemm
*m>M«.2018. l-jiiwiMlirKl«:l9U4-'9aS),tlM<» |H«m IM'oina Colo I Iramtoia -OM l»u»l (NfM. >
Wlrk'iwr »l dl. Cai
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Emlpoint Name Study Name Outcome Study Design Animal Description Trend Test Response Dose
Confidence Result Units (mg'ki
Fetal Body Weight Hatris. 1989, 3858729 High confidence Gestational Cral (GD 10-13) Fl Mouse, C57BU&1 ( I ) significant litter 0
0.2S
1G
Gestational Oral (GO 6-15) Ft Mouse, CS7BUCn significant litter 0
-56 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10
l't.-ii:i.-it C.MilH: '-[-O]
Figure 3-56. PFDA fetal body weight after gestational exposure (results can be viewed by clicking the HAWC link).
More specifically, Figure 3-56 shows statistically significant and progressive
decreases/trends in fetal body weight with dose across a range that includes the >5%
decrease historically used by regulatory agencies as the demarcation of adversity in
dose-response assessment (i.e., as a >5% reduction in fetal body weight is usually
considered adverse, a benchmark response of 5% is typically used in BMD analysis),
from -1% to -50% and including doses that resulted in decreases of -4% and -6%
(Table 3-23, p. 3-153). Furthermore, the changes in fetal body weight were at doses
not associated with maternal toxicity (p. 3-151, lines 5-6).
Harris and Birnbaum (1989) is a medium/high-confidence study (p. 3-150, lines 9-12),
and EPA should consider mouse data from the Harris and Birnbaum (1989)
gestational exposure study for use as the primary basis for subchronic osRfD
development (Tier 1 necessary revision) based on developmental growth effects
consistent with: (1) this mouse study providing the primary basis for EPA's
developmental effects hazard conclusion; (2) the lack of factors that decrease
certainty for fetal growth as evaluated in the mouse study (see Table 3-24); and (3)
the substantial epidemiological database merely being able to provide support for
this mouse study with what amounts to "slight" evidence for developmental effects
across over 45 epidemiological studies. Afterall, per the draft assessment, it is the
mouse data that primarily support that assuming sufficiently high exposure over a
sufficiently long duration (i.e., "given sufficient exposure conditions"), PFDA exposure
is likely to cause developmental toxicity in the general human population, which
includes potentially susceptible subpopulations (e.g., developing fetuses of pregnant
women).
My suggestion to use the mouse data (Harris and Birnbaum 1989) instead of
epidemiological study data (Wikstrom et al. 2020) for dose-response assessment and
osRfD derivation based on developmental growth effects (Tier 1 necessary revision)
would make EPA's assessment somewhat more consistent with ATSDR (2021),43 who
found the epidemiology literature inadequate for use as the basis of deriving MRLs
43 To be entirely consistent with this conclusion from ATSDR (2021) and my own comments above, the EPA assessment
would further need to omit the highly uncertain RfD derivations based on PFDA associations with decreases in antibody
levels/vaccine response (Tier 1 necessary revision) reported in some epidemiological study results that appear overall
to be inconsistent, non-adverse, and uncertain in nature (e.g., causal attribution problems, unsupported by
epidemiological disease incidence studies).
141
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
for PFAS, noting:
"There are sufficient epidemiological data to identify possible sensitive targets
for many of the perfluoroalkyls; however, there are two major limitations to
establishing dose-response relationships for these effects and using the
epidemiological studies to derive MRLs: accurate identification of
environmental exposure levels producing increased risk for adverse effects
(exposure estimates and routes of exposure) and likely co-exposure to mixtures
of perfluoroalkyls. Other limitations include the cross-sectional design of the
majority of epidemiological studies and the potential that reverse causality
contributes to the observed associations... In summary, the epidemiological
databases for several perfluoroalkyls provide valuable information on hazard
identification; however, uncertainties regarding doses associated with adverse
effects and possible interactions between compounds preclude use of these
data to derive MRLs."
Getting back to use of Wikstrom et al. (2020) for dose-response assessment of birth
weight and RfD derivation, Tables 2 and 3 from that study appear below (Table 2
unaltered, Table 3 statistically insignificant results for PFDA associations with
decreased birth weight and small for gestational age are highlighted in red; see license
at https://creativecommons.Org/licenses/bv/4.0/) and demonstrate: (1) significant co-
exposures to other PFAS (e.g., PFOS, PFOA) associated with birth weight deficits, an
ATSDR concern and significant uncertainty; and (2) few statistically significant ORs for
decreased body weight in the PFDA exposure quartiles; that is, only 2 out of 9 ORs
showed statistically significant birth weight decrements, which were for quartile 4
where both PFOS and PFOA were certainly co-exposures and also showed statistically
significant decreases for birth weight. In regard to this latter point concerning potential
confounding by co-exposure to other PFAS, which was cited by ATSDR as a significant
uncertainty, Wikstrom et al. state [emphasis added]:
"Another limitation is the compound-by-compound approach. Theoretically, a
health outcome is simultaneously influenced by multiple environmental factors.
Nevertheless, the exposure to several PFASs may be correlated with each other
due to common sources. Our findings were consistent across different PFAS
compounds, and we regard correction for multiple comparisons overly
conservative to be suitable for the investigations on such interrelated
compounds. If the single compound's level could represent the levels of several
other compounds, our findings based on single compound analyses may still shed
some light on the joint effects of multiple PFAS compounds. However, carefully
designed statistical models, such as mixture-based approaches within the PFAS
compound class, should be explored in follow-up studies."
Thus, the study authors are acknowledging confounding by other PFAS as an important
limitation of their compound-by-compound approach as joint effects of multiple PFAS
compounds may be occurring such that mixture-based approaches should be explored
for data analysis. The compound-by-compound approach used in this study gives rise to
significant uncertainty that precludes use of these data for dose-response assessment
and RfD derivation. It cannot be confidently said scientifically that the presumed
effects are due to PFDA exposure (e.g., PFOS and PFOA were certainly co-exposures
142
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
and also showed statistically significant decreases for birth weight). Consistent with my
opinion on the significant uncertainty and confounding associated with these
Wikstrom et al. results and the weakness/limitations of the epidemiological evidence
overall, Section F.3 of the draft assessment states that, "In the six studies using
mutually adjusted PFAS approaches to address coexposures, there was not consistent
evidence for birth weight deficits associated with increased exposure to PFDA." (p. F-
27, lines 13-14), and acknowledges that "there is considerable uncertainty due to
potential confounding by co-occurring PFAS in the existing literature." (p. F-27, lines
30-32). Indeed, significant co-exposure to multiple PFAS is not the exception but the
rule, and is just the condition to result in significant bias away from the null for adverse
effects.44 The high confidence study of Luo et al. (2021) is an example that serves as a
cautionary tale on the obvious importance of PFAS co-exposures, which reported large
statistically significant birth weight deficits (-97 g; -178, -16 per each In-unit PFDA
increase) in a single-pollutant PFDA model, but results were null and their direction
reversed in the multipollutant model with a nonsignificant increase in birth weight
associated with PFDA (Table F-2).45
As the data from epidemiological studies provide only "slight" evidence of
developmental effects, including data from Wikstrom et al. (2020) that appears too
uncertain and unsuitable for quantitative dose-response assessment, I suggest use of
the mouse data (Harris and Birnbaum 1989) for dose-response assessment and
subchronic osRfD derivation based on developmental growth effects due to gestational
exposure (Tier 1 necessary revision). See previous comments about the apparent
amenability of these mouse data to BMD analysis. Unreliable/inadequate data (human
or otherwise) cannot be justified scientifically for use in dose-response assessment by
how the associated unreliable POD(s) compares to that based on reliable data
definitively showing cause-and-effect for unquestionably adverse effects,
unconfounded by significant co-exposures to similar chemicals.
Tabfe 2. Pn iidtd!
l-osure to eight PFAS,. n
1 ur d r maternal serum concent!dti i
,nq n^L dufin*] mi/ pregnancy.
Compound
Geometric mean (91,,
•> Ui Median 0QR)
LOD
Above LOD (%>
PFOS
535 [52]-5.501
s.L rQ
1034
0.06
100
PFOA
1.60 {1.56-1.65!
1 _ i
3.18
0.02
100
PFHxS
131 il.2S-1.35I
L iUb»- 1
2.94
0.03
100
PFNA
054 |0.53-0.56]
OiT lU^M J~ >
1.02
0.01
100
PFDA
0 28 (0 26--0.27I
0 in H C4t
oso
0.02
100
PFUnDA
021 (021-0.22]
qs m jh
0.44
0.02
995
PFHpA
0.018 (0.017 0.019!
01V i |_UD
0.077
0.01
73.9
PFDoDA
0.027 (0.026-0.027]
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Table i. Associations between prenatal PFAS exposure* and birth weight61,, together with odds ratios for birth weight small for gestational age in
1533 Children.
All children
Girts
Boys
BW(g)
SGA
BW(g)
SGA
BWIg)
SGA
£ 199% aj
OR (95% CIJ
0 195% aj
OR (95% CI)
$ 195% a)
OR 195% a)
PF OS
Per In-unit
-¦46 1-88; —3)
1.19 I&87; 1641
-85 1-145; -251
1.40 14183; 235)
-13 (-73; 47)
1.458 (0.72; 1.631
Q1
Reference
Reference
Reference
Reference
Reference
Reference
Q2
-27 1 —89; 35)
0.69 1043; 1.08)
-32 (-115; 52)
0.89 14139; 203)
-28 (-118; 63J
1.26 (0.67; 237)
Q3
-22 <-84; 41)
0.79 |Gl53; 1.181
-51 (-137; 34)
0.82 |036; 203)
5 (-86; 96j
4186 (0-45; 1.67)
04
-BO 1-144; -16)
156 11.09; 2221
-142 1—231; —54.1
2.05 11.00; 4211
-28 (-119; 63)
110 (0.70? 2401
PFOA
Per Irvunit
-68 {-112; -24)
1.43 ( 1j03 > 1.991
-86 (-145; -26)
1.96 11.18; 128)
-49 (-113; 15)
1.16 (0.75; 1.78)
01
Refeience
Reference
Reference
Reference
Refereixe
Reference
Q2
27 <-35; 891
0.77 14145; 1121
30 (—55; 115)
1.4)0 10.40; 251.1
26 1-66; 116)
4167 (014; 111)
03
-41 (-106; 23)
0.96 |057; L61)
-36 |-124; 52)
1.64 14171; 183)
-44 (-139; 501
066 (013; 1291
04
-90 1-159; -91)
1.44 1086; 240)
-136 (—231;—40)
213 11-00; 143)
-47 (-147; 54)
1.04 (054; 201)
PFHxS
Per Irvunit
-0,1 (—38; 381
0.96 10.72; 1.27)
-14 1-68; 39J
1.14 (4173; 1.80)
-13 (-67; 41)
4184 (058; 122)
01
Reference
Reference
Reference
Reference
Refereixe
Reference
02
-4 |-66; 581
117 10B6; 2201
30 (-56; 1161
1.77 14178; 199J
-39 (-129; 501
1.24 |0j69; 2231
03
-15 (—78; 4®
0.89 10l54; 1.47)
28 1-59; 115)
1.05 14144; 249)
-51 (-141; 39)
4182 (0.44; 1541
04
-6 (-69; 57)
1.04 1063; 1.69)
-16 (-104; 71)
1.76 |Q79; 190)
1 (-90; 92)
4173 (018; 1.41)
PFNA
Per In-unft
-46 1-89; -4)
1 JIB (1j02, 1JJ7)
-52 (-117; -2)
134 14185; 211)
-50 (-113; 14)
1.42 (0.94; 217)
01
Refeience
Reference
Reference
Reference
Refereixe
Reference
02
7 (—55; 63
0.83 10.49; 1.38)
-2 1-86; 82)
0.66 10.29; 112)
15 (76; 106)
4197 (050; 1.891
03
-39 (-102; 24J
1.14 1070; LBS)
-49 (-137; 38)
119 10.66; 290)
-28 (-119; 64)
095 (050; L82)
04
-33 1-96; 31)
1J3 10.77; 1.99)
-66 1-153; 20)
1.22 1059; 253)
1 (-94; 95)
124 (0j66; 233)
PFDA
Per In-untt
-58 (-103; -13)
1.46 11.06, 201J
-69 (-133; -61
1-62 [098,2671
-47 £-112517)
116 I0.9O; 2071
01
Reference
Reference
Reference
Reference
Refereixe
Reference
02
-23 I -EE; 33
1.03 1062; 1.691
-42 (-126; 42)
0.86 (037; 200)
-8 (-99; 83
1.18 10.63; 223)
03
-39 (-101; 23)
1.07 1065; 1.761
-74 (-160; 13)
12D (054; 267)
-8 (-98; 81)
4199 (052 1-891
04
-69 1-112; -51
150 10.94; 2381
-116 (-204; —27J
1.95 [4194; 406)
-27 (-118; 64)
1.21 |Di»6c 2231
PFUnDA
Per lrvuntt
-13 1-49; 221
1.21 10.92; 158)
-24 (-75; 27)
1.438 |4174>; 1.67)
-6 (-55; 42
129 (0.82; 1.83)
01
Refeience
Reference
Reference
Reference
Reference
Reference
02
-67 (-153; 19J
1.24 1077; 1011
-67 (—153; 19)
2.09 1Q94; 463)
41 1-48; 130)
0.90 (0.48; 1j68J
03
-42 (-128; 411
0.85 14151; 1.43)
-42 (-128; 44)
1.01 |042; 2441
46 1-44; 1361
4181 (0.43; 1561
04
-46 1-110; 171
152 (0.95; 2441
-93 (-183; -31
1.92 1086; 425)
-10 ( -100; 801
116 (0.76; 2461
PFHpA
Per Irvunit
-1 (-24; 21)
1-06 10.90; 125)
-4 (-34; 2Q
1.4)6 1083, 116)
—Q.Q3 (-33; 32)
1.07 (0.86; 117)
01
Refeience
Reference
Reference
Reference
Refereixe
Reference
02
-0 1-62; 61)
0.92 1057; 1.481
17 (-70; 143)
0.92 14144; 1.92)
-15 (-102; 73)
4194 (050; 1.751
03
3 |-59; 65)
0.78 1048; 1J27)
-21 (-106; 65)
0.75 1015; 1.65J
23 1-65; 112)
4178 (0.42; 1.45)
04
31 (—31; 93)
1-25 14185; 1.841
27 (-57; 112)
111 1066; 260)
33 (-58; 124)
1.15 (0j65; 2041
AII analyses were
adjusted for maternal weight, parity [three categories) arvd cotinirve levels. Analyses including both beys and girls were in
addition adj is ted
far sex and analyses of BW were adjusted for G\. SGA was d«ftr-ed as BW UMh pecentile for sex and GA
"A ucxiatkus with PFAS are presorted pe
In-urit and by quartiles of expos are, as related to
^Birth weight Jg'l
and odds ratios {adjusted! for birth weight small for gestational age
The above comments provide relevant rationale for use of an alternative study(ies)
or effect(s) for subchronic RfD derivation. While the draft assessment provides
various but limited candidate PQDs, due consideration of the comments above
appears to require a more complete evaluation of alternative studies/effects (Tier 1
necessary revision). One method to begin such an evaluation may entail sorting the
study data extracted by duration, then NOAEL/LOAEL, BMD or a similar criterion that
would allow EPA to readily identify the next most sensitive effects based on less
problematic studies that are adequate for subchronic RfD derivation. However, any
newly derived toxicity factor should be subject to external expert peer review (Tier 1
necessary revision).
c. As I do not agree with the selection of these studies/endpoints for lifetime or
subchronic RfD derivation, please see my comments under subsection "b" above,
many of which concern the considerations of adversity and scientific support.
144
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
That being said, speaking more generally in regard to deriving a subchronic RfD,
subchronic RfDs must be protective against adverse effects occurring over
subchronic and shorter durations of exposure, such as during childhood or during
development (e.g., critical windows of development during gestation). Thus, if
effects that occur over a subchronic or shorter duration of exposure are truly
adverse, relevant to humans, and are the first to occur as the dose rate increases
(i.e., the critical effect), they can appropriately serve as the basis for a subchronic
RfD.
Lastly, since 1 suggest use of the mouse study of Harris and Birnbaum (1989) as
the basis of a subchronic osRfD based on decreases in fetal body weight (with
various gestational exposures), 1 note that Figure 3-56 of the draft assessment
shows statistically significant and progressive decreases/trends in fetal body
weight with dose across a range that includes the >5% decrease historically used
by regulatory agencies as the demarcation of adversity in dose-response
assessment (i.e., as a >5% reduction in fetal body weight is usually considered
adverse, a BMR of 5% is typically used in BMD analysis), from -1% to -50% and
including doses that resulted in decreases of -4% and -6% (Table 3-23, p. 3-153).
Furthermore, the changes in fetal body weight were at doses not associated with
maternal toxicity (p. 3-151, lines 5-6). Harris and Birnbaum (1989) is a
medium/high-confidence study (p. 3-150, lines 9-12).
d. The section Derivation of Candidate Lifetime Toxicity Values for the RfD (p. 5-19)
provides reasonable rationale for only considering liver effects data and
male/female reproductive effects data for subchronic osRfD derivation. While the
total composite uncertainty factor for these effects is already considerable
(1,000; Table ES-1), EPA could consider increasing it to the upper end of their
total uncertainty factor range (3,000) to estimate chronic osRfDs.
e. Section 5.2.4 (p. 5-41, lines 35-36) states, "No studies examining inhalation
effects of short-term, subchronic, chronic or gestational exposure for PFDA in
humans or animals have been identified, precluding the derivation of an RfC."
Furthermore, it appears from Sections 3.1.6 and 3.1.7 of the draft assessment
that a reliable PBPK/PK model has not been identified for PFDA that could be
used for route-to-route extrapolation. If EPA has not identified a reliable PFDA
PBPK/PK model that could be used for route-to-route extrapolation, then EPA's
decision not to derive a subchronic or chronic RfC would be fully justified.
However, this is not explicitly stated in Section 5.2.4, which should clearly state
whether or not EPA has identified a reliable PFDA PBPK/PK model for route-to-
route extrapolation (Tier 1 necessary revision).
Hoberman
The selected studies to derive the subchronic RfD for PFDA were scientifically justified.
a. The studies used to derive the subchronic RfD were typically of 4 weeks in duration
and appropriate for use to derive a subchronic RfD.
b. Not applicable.
145
-------
Final PFDA Post-Meeting Comments Task Order 68HERH20F0407, Contract EP-C-17-017
c. The effects selected were appropriate for use in deriving the subchronic RfD,
including considerations regarding adversity (or appropriateness in representing an
adverse change) and the scientific support for their selection.
d. This reviewer agrees that "the subchronic osRfDs for liver, male reproductive and
female reproductive effects derived from short-term animal data were several
orders of magnitude higher than the subchronic osRfDs for immune and
developmental effects in humans; therefore, they were not considered to be
sufficiently protective for consideration in the selection of the overall subchronic
RfD. Also, in the case of male and female reproductive effects, confidence in the
respective osRfDs was lower compared to the immune osRfD (medium-low versus
medium) due to deficiencies in the evidence base for these health effects" (Page 5-
41)
e. This reviewer agrees that with "no studies examining inhalation effects of short-
term, subchronic, chronic or gestational exposure for PFDA in humans or animals
have been identified, precluding the derivation of an RfC." (Page 5-41)
Leung
Toxicological modeling is not my area of expertise, and 1 defer this question to the
other external reviewers.
Zoeller
a. The Agency clearly explained the use of the same studies for derivation of lifetime
and subchronic RfDs. The issues regarding adversity etc are developed under
Question 3 above. No Recommendation.
d. This is covered in Question 3 above.
e. This is a rational decision. No Recommendation.
3,5 Appendix G Identifies the potential for pharmacokinetic (PIC) differences across species
and sexes as a key science issue and lays out a hierarchy for using relevant PK data in
extrapolating doses between laboratory animals and humans. Section 3,1 evaluates and
synthesizes the PIC data in relevant species and sexes, and among human lifestages, up to
the derivation of key PK parameters used in the subsequent analysis. However, the
evaluation of existing PBPK models and a one-compartment PK model found that these
options were not sufficiently reliable for use. Given the information available on potential
interspecies differences in PFDA PK, EPA applied a data-derived extrapolation factor
(DDEF) to POD values from toxicity studies in laboratory animals to estimate
corresponding human equivalent doses (HEDs) in the derivation of the respective MDs,
Similarly, the estimated human clearance (CL) was used to convert internal dose POD
(PODint) values from epidemiological analyses to corresponding HEDs.
After publicly releasing the draft IRIS PFDA assessment, the EPA evaluated recently
published data for several other long-chain PFAS, described here (U.S. EPA, 2023, HERO ID
146
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
11181055), that are potentially relevant to evaluating PFDA dosimetry in women of
chiklbearing age (see question 5c below).
a. Is applying the estimated DDEF values for PFDA scientifically justified for conversion of
PQDs from animal toxicity studies to HEDs? If not, please provide an explanation and
detail on a more appropriate approach.
b. Is application of the human CL to estimate HEDs from PODInt values scientifically
justified? If not, please provide an explanation and detail on a more appropriate
approach.
c. Have the uncertainties in the DDEFs and human CL been adequately evaluated and
described? In answering this question, please provide an explicit recommendation on
whether or not EPA should expand its adjustment for menstrual fluid loss as outlined
in (U.S. EPA, 2023, HERO ID 11181055) prior to finalizing the assessment. As these
newer data are from other PFAS, note that such an expansion would be based on the
assumption that the pharmacokinetic effect of pregnancy and lactation on PFDA is
similar to that of the other PFAS (i.e., a read-across based interpretation).
Noncancer Toxicity Value Pharmacokinetic Extrapolation and Uncertainty Factors
Reviewer
Comments
Adgate
Tier 1: The presentation of the complex relevant data landscape on PFDA and
shortcomings of the single available PK model can be improved substantially. Given the
shortcomings of the underlying data from both animals and humans the DDEF approach
is justified in this case, with the caveat that there needs to be an Evidence Integration
section in the Appendix AND main PFDA document that lay out the logic of the choices
made more clearly. Most importantly, this section needs to explicitly lay out the
uncertainties as judged by the scientists who have done this extensive evaluation of the
existing data.
EPA's remarks, especially the summary provided by Dr Dzierlenga, should be used as a
basis for this improved Evidence Integration section. This El section should also contain
a table that explicitly summarizes the key uncertainties and judgments on the relative
importance of each uncertainty for model performance and for informing future efforts
to improve the existing model or create new models that can provide insight into PFDA
pharmacokinetics.
It is notable that the DDEF approach was chosen due to the uncertainties of the model
parameters and quality of the underlying data. The presentation could be improved if
the DDEF approach were applied to another, more data rich, PFAS (e.g., PFOA or PFOS)
to see how the DDEF performs under those circumstances.
Lastly, this section should clearly acknowledge that half-life for PFAS in general and
likely PFDA in particular is often represented by a mean that often understates the wide
observed variability in half-lives observed for many similarly persistent PFAS in humans.
At a minimum half-life should be reported as a mean or median, standard deviation
and/or geometric standard deviation, and, if possible, range of observed values.
147
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Carignan
a. As explained during the meeting the approach seemed reasonable, however it was
complicated and not readily apparent from reading the document. Modeling should
be sure to consider enterohepatic recirculation and kidney resorption. However,
this type of modeling is not my main area of expertise. [Tier 2]
b. It was not clear from the description that assumptions are based on animal data,
which introduces substantial uncertainty. [Tier 1]
c. This seems like a reasonable assumption.
It is important to consider that not all women menstruate, bear children or breast
feed. Those who do not would of course have lower clearance than those who do.
[Tier 1]
Of course these considerations are essential in order to protect the fetus, infants
and children.
Additional comments:
Discussion of the magnitude and directionality of uncertainty. [Tier 2]
Recommend adding a table of uncertainties and prioritization for evidence integration.
[Tier 2]
Faustman
a. This reviewer supported the actions by the USEPA to use DDEF values. Ideally, as
the report states a specific PBPK model would be ideal however this review was
convinced by the extensive discussion that any existing models are of limited
application.
b. This reviewer was supportive of the actions by the USEPA to use human CL values to
estimate HEDs. In particular, the discussion on distribution and known roles of Oat
transporters was of interest in the discussion of PFDA as there is increasing
information regarding developmental differences in the developmental time of
expression. This reviewer was also glad to see discussion on the degree of protein
binding of PFDA and how it affects both distribution and elimination on page 3-3,
however minimal discussion of pregnancy related changes in protein binding was
not especially emphasized (see earlier comments about hemodynamics during
pregnancy). This reviewer was looking for discussion of transfer of PFDA via breast
milk but minimal to no comments were made on this aspect of PK. This issue was
raised in the public comments but does need to be discussed in the document. A
Tier 2 recommendation is made to acknowledge the finding of differences in a
variety of pharmacodynamic parameters including protein binding during pregnancy
with linked comment on potential for breast milk lactational pharmacokinetic
considerations. Addition of references and several sentences would be suggested.
This reviewer was less convinced about the importance of menstrual blood loss and
increased elimination in women reportedly due to this process. It was listed as an
exercise to look at potential impact however there are so many factors that affect
population dynamics and potential loss by this pathway that further thought would
be need to determine human impact. Due to various underlying conditions, there is
148
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
tremendous variability across the female proportion of women and menstrual flow.
For example, many women do not experience regular periods. Would the USEPA
use a population distribution to determine this variation in more details. This
reviewer did not determine that this was a consideration that was "ready for prime-
time application "and was glad to see that this was not applied as an adjustment.
c. This reviewer did not identify other areas or factors to be examined.
This reviewer appreciated the efforts by USEPA to evaluate available PBPK models as
well as characterize both a two and one compartment model with Bayesian modeling
efforts. Due to the ultimate need to improve cross compound PFAS comparisons such a
model would provide important tools to accomplish these comparisons. This reviewer
however agreed with the ultimate decisions to apply a more simple approach in the
document due to lack of fully supported pregnancy based PBPK model availability.
Fisher
a. The reasons stated why data-derived extrapolation factors (DDEFs) were used are
because of the publication of a poor quality PBPK model for PFDA (Kim et al.
2019) and the inability to predicted plasma levels in rats gavage dosed with PFDA
daily for 28 days (NTP 2018) using a compartment model calibrated using single
dose PFDA PK data ( IV and oral). 1 think the EPA should continue to resolve these
issues and use a PBPK model (Tier 1) beyond what has been published by EPA
authors ( Bernstein et al. 2021). It would be useful to have a PBPK model that can
describe several PFAS compounds, both single doses and repeated dosing
scenarios. It would provide substantial utility in the future. It appears that the EPA
modeling group started down this path (Bernstein et al. 2021) using the Kim
model along with other PBPK models for PFOA and PFNA. 1 would call this
harmonization of existing models.
1 was struck by this finding because this is what we found when working with
perchlorate ( single dose versus multi-day drinking water exposures), which
inhibits thyroid uptake of iodide. Our conclusion was that the transporter which
transports iodide into the thyroid gland (sodium iodide symporter (NIS) protein)
was upregulated by the thyroid stimulating hormone, TSH. In this case with PFDA
there is potential involvement of transporters in the kidney and the liver, which
may be altered by PFDA itself. The liver was a target for toxicity, which is
discussed in detail in this report and may be associated with the altered PK
profiles of PFDA (biliary transporters, for example). The accumulated PFDA
concentrations (in plasma) were well below model predictions (Fig. G-8) even at
the lowest doses, suggesting alterations in binding profiles in plasma or tissues
and/or transporter proteins that are key for PFDA pharmacokinetics. Without
seeking resolve for the PFDA pharmacokinetics for repeated daily exposure in
animals 1 am skeptical of using a DDEF methodology that relies on the ratio of
calculated whole body total clearance rates (CLhuman/CLrat). More work is
needed to better understand the pharmacokinetics of PFDA for repeated animal
exposures in the context of a PBPK model. Tier 1. 1 would look at the protein
transporters involved and determine if there is experimental evidence for the up
or downregulation (gene level or protein), including PFOA and PFNA. 1 would
evaluate where the transporters are located and their functions. If protein binding
is saturated in plasma, the liver or other organs this could lead to increased
149
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
clearance. Is there experimental evidence with PFNA or PFOA? Is a 3X increase in
CL approaching the GFR? Is there evidence for changes in clearance with PFNA or
PFOA in rats exposed daily for weeks? This might be called read across, but I call
this harmonization.
b. HED= PODint (serum concentration of PFDA) x CLhuman. The downside is
that no dose-response information was obtained. This is a serum
concentration associated with a single response. The only way to
strengthen this would be to have all the individual data on serum
concentrations and immune response. Then a POD could be estimated,
perhaps (Tier 2).
Another assumption is that the PODint garnered from human blood
sampling is robust. The human CL values encompassing the values
reported by Fujii et al. (2015) and Zhang et al. (2013b) are valuable. I
would prefer healthy individuals with spot urine samples (86 individuals,
Zhang et al. (2013b)) over a smaller number of older hospitalized
individuals (bile/serum, n=5) with a 24 hr urine /blood samples from 10
young volunteers (Fujii et al (2015). Both studies should be used to
capture the potential variability in CL. I believe this leads to a probabilistic
type assessment of HEDs using human serum concentrations of PFDA._Tier
2.
c. Yes, the authors discuss the shortcomings of the studies used and how CL
was calculated.
The big problem is that protein transporters (liver and kidney) and serum and
tissue (?) protein binding are responsible for the slow whole body CL rate. A
PBPK model is greatly favored over the current approach because the key
biological factors can be explicitly described and used in predictions. Animals
are much different than humans. The use of allometric relationships is of
little or no value for across species scaling. Tier 2.
HERO 10410674 does have data on PFDA in the supplemental table (S2). In pregnancy
many changes occur that can affect the measured plasma levels of a drug or chemical.
In this case, albumin concentrations decrease (PFDA is highly bound to albumin),
plasma volume increases by up to 50%, GFR increases and renal blood flow. These
changes alone may account for the observed decreases in PNDA (in this paper) and
other PFAS. There are many good reviews of pregnancy, physiology, and
pharmacokinetics (Ward and Varner 2019, Clin Perinatol 46 383-398, Pariente et al.,
2016). Pariente et al. report pharmacokinetic parameter changes for drugs in
pregnancy. In lactation, lactational transfer and physiology both play a role in the
measured levels of PFAS and PNDA. I would not call it menstrual fluid loss during
pregnancy, but apparent pregnancy related changes in distribution and clearance. Table
S2 shows a slight trend of decreasing serum PFDA. Again what are the observations, if
any, for PFNA and PFOA?
150
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Clarification: Since there is data on PFDA as mentioned above, you do not need
read-across, use the data directly. Read-across is helpful to provide evidence of
going down the right path in terms of PK. Tier 2.
Because the CL or half-life is so long in humans for PFDA, menstrual fluid represents a
significant pathway for clearance or loss from the body because PFDA is highly bound to
protein in blood. You can calculate what fraction of total CL menstrual fluid represents.
Childbirth itself represents a significant loss of blood from the body.
If mechanistic models were used to interpret the PK for animal studies for extrapolation
to humans, more confidence would be achieved in the HEDs based on animal studies.
Georgopoulos
a. Applying estimated data-derived extrapolation factor (DDEF) values for conversion
of PODs from animal toxicity studies to human equivalent doses (HEDs) is in general
appropriate in cases when reliable pharmacokinetic models (either physiologically-
based or traditional) are not available for animal to human extrapolation. The
Toxicological Review reports (in Section 3.1 and in particular in subsection 3.1.6)
that EPA evaluated existing PBPK models and a one-compartment PK model and
concluded that these options were not sufficiently reliable for use. This conclusion is
appropriate, despite the fact that certain issues require clarification and additional
information that is not currently included in the Review (1 summarize related
concerns in my response to Charge Question 6). Since this is the case, EPA
proceeded to apply a data-derived extrapolation factor (DDEF) to point of departure
(POD) values from toxicity studies in laboratory animals for estimating
corresponding human equivalent doses (HEDs) in the derivation of the respective
RfDs. The DDEF essentially (assuming similar absorption fractions) represents the
ratio of human clearance (CLH) over animal (rat) clearance (CLA). Although, as
mentioned above, the use of the ratio of human to animal clearance factors for
interspecies toxicokinetic extrapolation (in the absence of reliable PK/PBPK models)
appropriate, in this case there are concerns that should be discussed and addressed
in the revision of the Toxicological Review.
The human clearance for PFDA was calculated by combining information from
human data reported in Fujii et al. (2015) and from animal (rat) data reported in
Kim et al. (2019); it involves critical assumptions that are discussed next, in my
answer to Charge Question 4b.
The calculation of the animal clearance for PFDA raises more serious concerns as it
was performed employing a steady-state assumption but using data from short-
term animal studies (involving only 28 days of exposure, i.e., duration much shorter
than the half-life of PFDA in either male or female rats). Serum PFDA levels will not
reach steady-state at the end of the 28-day dosing period studies, since a
reasonable approximation of steady-state is not attained until after 4 to 5 half-lives
of dosing. Using this approach when serum PFDA levels in rats have not reached
steady-state will result in overprediction of the HED. The Toxicological Review
recognizes the issue but proceeds with the calculation by assigning an uncertainty
factor to account for the fact that that the required conditions for steady-state are
not fulfilled; specifically the Review (page 3-24, lines 7-12) states:
151
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
"... application of a DDEF assumes that a steady-state concentration is reached,
equal to dose/CL. When the end-of-study PFDA levels observed by NTP (2018) are
compared to the corresponding dose/CL using the mean CL estimated for male
rats the resulting dose/CL values are 2.7 to 1.4 times higher than the data. Hence
the uncertainty in use of the male rat CL in the current analysis is considered to be
less than a factor of 3."
An alternative to assuming steady-state conditions could involve using average
measured PFDA serum levels over the course of the study, or maximum serum
PFDA level at the end of the dosing period to determine the POD for benchmark
dose modeling (BMD).
b. The approach of using the human clearance (CLH) for extrapolation of the serum
PFDA level point of departure (PODint) from human epidemiological studies via HED
= PODint x CLH is appropriate, assuming of course, that a reliable value of CLH can
be estimated. According to the Toxicological Review, for this purpose a "health-
protective estimate for human clearance" was used, equal to CLH = 2.6 x 10-5 L/kg-
d, i.e., "the value estimated for pregnant and breast-feeding women and their
female children." (page 3-24, lines 1-4).
It should be taken into account however that these estimates of human clearance
are based on critical assumptions, such as that the ratio of fecal/urinary clearance is
the same in humans and rats; specifically, as per the Toxicological Review (page 3-
16, lines 8-15) it was assumed:
"that the ratio of fecal/urinary clearance is similar in humans as in rats. Kim et al.
(2019) observed a mean fecal excretion 1.63 times higher than urinary excretion in
male rats, but only 0.742 times urinary excretion in female rats. Both of these
ratios are considerably lower than the ratio of 3.3 estimated by Fujii et al. (2015).
Given the uncertainty [...] for the estimated fecal clearance of Fujii et al. (2015),
these sex-specific ratios will be applied to the estimated human urinary clearance
from Fujii et al. (2015) (0.015 mL/kg/day) to obtain total estimated urinary plus
fecal clearance rates of 0.039 mL/kg/day in men and 0.026 mL/kg/day in women."
The Toxicological Review also states (page 3-18, lines 3-7) that:
"The estimated urinary clearance from Fujii et al. (2015) is considered particularly
reliable because 24-hour urine samples were used to determine the rate. Adding
the menstrual clearance of 0.030 mL/kg-day based on the results of Verner and
Longnecker (2015) yields total CL = 0.056 mL/kg-day for women between 12.4
(menarche) and 50 years of age, except during pregnancy and until menstruation
resumes postpartum."
The question whether the ratio of fecal/urinary clearance of PFDA is the same in
humans and rats deserves further inquiry and certainly contributes to the
uncertainty associated with the estimates of CLH.
c. Uncertainties and concerns associated with calculating the human clearance (CLH)
and the data-derived extrapolation factors (DDEFs) for animal-to human
extrapolation were discussed above, in my answers to Charge Questions 5a and 5b.
Regarding the specific question on whether or not EPA should expand its
152
-------
Final PFDA Post-Meeting Comments Task Order 68HERH20F0407, Contract EP-C-17-017
adjustment for menstrual fluid loss as outlined in (U.S. EPA, 2023, HERO ID
11181055) prior to finalizing the assessment, my recommendation is positive. It
should be pointed out that EPA recognizes (in
IRISPFDA_ChargeQuestion5c_Supplementallnformation.PDF, page 3) that the
overall impact of adopting this alternative approach is minor:
"If the alternative dosimetric approach is recommended by the peer reviewers,
this would result in no quantitative change to the draft RfD or subchronic RfD; the
highlighted minor change, an approximate 2-fold increase in the draft
Developmental osRfD, would be implemented..."
Suggested Revisions and Future Considerations
• Tier 1 Necessary Revision: Edit and expand the discussion on the estimation of
human clearance to state explicitly and clarify that that this estimation
incorporates uncertainties associated with assumptions that are based on
animal data regarding the ratio of fecal/urinary clearance of PFDA.
• Tier 2 Suggested Revision: Evaluate and, if feasible, apply options for
alternative animal to human extrapolation approaches that do not require
steady-state assumptions inconsistent with short-term study data.
Haney
a. Generally speaking, yes. As discussed in Section 3.1.7 of the draft assessment
(starting on p. 3-22), evaluation of a published PBPK model (Kim et al., 2019) and
a one-compartment PK model showed significant errors in the PBPK model and
that the simpler PK modeling approach did not reliably predict PFDA serum
concentrations. An alternative to PBPK/PK modeling for dosimetric extrapolation
is data-derived extrapolation factors (DDEFs). The estimated population average
values of CLtot for male and female rats, female mice, and male and female
humans are considered to be sufficiently sound for use in such extrapolations,
while use of default BW3/4 scaling would lead to significant errors in HED
calculations. Therefore, application of DDEFs calculated from the clearance values
listed in Table 3-3 (p. 3-18) of the draft was considered by EPA to be the next
preferred option for extrapolation from laboratory animal doses to HEDs. The
corresponding DDEFs appear in Table 3-4 (p. 3-23). This hierarchy of preferences
appears consistent with, and therefore is generally justified by, environmental
regulatory agency practice (e.g., EPA, TCEQ).
Additionally, it is noted that in the absence of good scientific information on
which laboratory animal species/sex/life-stage best predicts similar effects in
humans, regulatory agencies often use the most sensitive species/sex/life-stage
for each adverse effect. Consistent with this type of assumption, EPA used CL
values generally matched for animals and humans based on sex and life stage.46 In
the absence of additional information, this may be the most reasonable approach.
46 "Liver effects observed in adult female rats are assumed to be relevant to older women, hence the same CLH (0.026
mL/kg-day) will be used to extrapolate those. Liver and reproductive effects observed in adult male rats will be
extrapolated using the CLH for men, 0.039 mL/kg-day. Finally, reproductive effects observed in adult female rats will be
extrapolated using the CLH for women of reproductive age, 0.056 mL/kg-day." (p. 3-23, lines 16-20)
153
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
b. As 1 am not a PBPK/PK modeler, on this question 1 think it best to
ultimately defer to those with expertise in PBPK/PK and its
implementation in the context of dose-response assessment. 1 simply note
that the draft assessment indicates that application of a DDEF assumes
that a steady-state concentration is reached (p. 3-24, lines 7-8), that a
chemical "achieves a given steady-state concentration after approximately
4 to 5 half-lives" (Hallare and Gerriets 2022), and that PFDA half-lives are
relatively long depending on species (e.g., 44 days to 23 years per Table 3-
3; p. 3-18). Accordingly, it seems that if steady-state was not reached in
the underlying study made basis for an RfD, this extrapolation method
might be a source of significant uncertainty.
c. The draft assessment section Uncertainty in HED calculations for PFDA
discusses the evaluation of various uncertainties (pp. 3-24 to 3-25), which by
and large seems to adequately highlight those aspects of uncertainty. At the
same time, 1 note that the uncertainty appears appreciable. For example, this
section states [emphasis added], "While the PK model parameter estimates
seek to make the best use of the available chemical- and species-specific
data, there are also many uncertainties noted above, in particular for
humans." (p. 3-25, lines 3-5), and ultimately acknowledges that it is plausible
that human clearance is actually as high as that predicted by BW3/4 scaling (p.
3-25, line 10), which would result in HEDs 20- to 40-fold higher (p. 3-19, lines
33-35). Thus, there appears to be significant uncertainty.
EPA should expand its adjustment for menstrual fluid loss (as outlined in U.S.
EPA, 2023, HERO ID 11181055) prior to finalizing the assessment if EPA
judges the alternative/modified adjustment to be associated with less
uncertainty than that associated with not conducting such an adjustment
(Tier 1 necessary revision). The assumption that the PK effect of pregnancy
and lactation on PFDA is similar to that of the other PFAS likely to have
similar PK properties seems to be a reasonable one, and this new
alternative/modified adjustment would appear to result in a greater degree
of scientific refinement.
Hoberman
a. Applying the estimated DDEF values for PFDA was scientifically justified for
selection of the PODs when extrapolated from animal toxicity studies to HEDs.
b. The application of the human CL to estimate HEDs from PODint values is
scientifically justified.
c. The uncertainties for the DDEFs and human CL have been adequately evaluated.
However with the newer data from other PFAS, especially the adjustment for
menstrual fluid should considered on the assumption that the pharmacokinetic
effect of PFDA on pregnancy and lactation can be extrapolated using read-across
based interpretation.
Clarification: Tier 3 - All future evaluations should include the adjustment for menstrual
fluid for each PFAS.
154
-------
Final PFDA Post-Meeting Comments Task Order 68HERH20F0407, Contract EP-C-17-017
Leung
Toxicological pharmacokinetics is not my area of expertise, and 1 defer this question to
the other external reviewers.
Zoeller
a. This is outside my area of expertise,
c. This is outside my area of expertise.
3,6 EPA lias evaluated and applied where appropriate uncertainty factors to account for
Intraspecies variability (IJFH), interspecies differences (UFA), database limitations (IJFD),
duration (UFS), and LOAEL-to-NOAEL extrapolation (UFL) for PFDA,
a. Is uncertainty in the derivation of the toxicity values scientifically justified and clearly
described? Please describe and provide comments, if needed.
I, Please comment specifically on whether the methods used to derive toxicity
values for PFDA appropriately account for uncertainties in evaluating the
pharmacokinetic differences between the experimental animal data and humans?
b. For Immune effects, a UFS of 1 and 3 were considered to account for extrapolation from
less than lifetime human data; ultimately a UFS of 1 was selected. A UFs of 10 was not
considered as the developmental period is recognized as a susceptible lifestage for
these types of effects and therefore exposure during this time window can be
considered more relevant than exposure in adulthood (U.S. El >1). Also important
is the fact that, given PFDA's long half-life and the expectation that the children and
their mothers have been exposed to elevated levels of PFDA for many years, the
observed effects on immune response are considered to be the result of a cumulative,
prolonged exposure. Uncertainties with regards to additional susceptible life stages
(e.g., old age) are addressed as part of the UFD, Does the provided scientific rationale
support this decision? Please explain.
c. For liver effects, a value of 3 is applied to extrapolate between effects in laboratory
animals and in humans during the derivation of the subchronic RfD. Although PPARa
dependence might support a value of UFA = 1 If that were the sole pathway leading to
these effects, evidence for the involvement of non-PPARa pathways is available in the
PFDA database. Thus, uncertainty remains regarding the potential differences in
sensitivity across species due to the involvement of both PPARa-dependent and
PPARa- independent mechanisms. As such, the Toxicological Review concludes the
available data are not adequate to determine if humans are likely to be equally or less
sensitive than laboratory animals with respect to the observed liver effects and that a
value of UFA = 3 is warranted to account for the residual uncertainty in toxicodynamic
differences across species. Please comment on whether the available animal and
mechanistic studies support this conclusion and whether the analysis presented in the
Toxicological Review is clearly documented.
d. For liver, male reproductive, and female reproductive effects, a default value of 10 Is
applied for the UFS when extrapolating from 28-day animal data to a subchronic
exposure. Considering the potential for some health effects (prolonged diestrus, sperm
measures and increased liver weight) to worsen with increasing duration and the large
uncertainty associated with the lack of chem i ca I -sped fi c data to evaluate the effects of
subchronic exposure on liver, male reproductive and female reproductive outcomes,
the Toxicological Review concludes that application of a UFs of 10 is supported for the
155
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
purposes of deriving the subclironlc MD from the 28-day toxicity data. Does the
provided scientific rationale support this decision? Please explain.
e. Are the provided rationales for the remaining uncertainty facto L, UFD, IJFH)
scientifically justified and clearly described (to inform the IJFH, the assessment
evaluates and considers the available evidence on potential susceptibility to PFDA
within different populations or lifestages, including any potential impacts from early
life exposure to PFDA on children's health or health later in life, although few studies
on susceptibility were available)? If not, please explain.
Noncancer Toxicity Value Pharmacokinetic Extrapolation and Uncertainty Factors
Reviewer
Comments
Adgate
a. The derivation of tox values is clearly described and scientifically justified.
b. This is a reasonable scientific rationale that supports this decision. My response is
also consistent with my answer above on the question about subchronic RfDs and
their utility. 1 also concur with Dr Zoeller's comments on this topic.
c. The available animal data and the write up support this conclusion with enough
clarity to provide adequate scientific support for this decision.
d. The scientific rational here is clear and supported by this analysis.
e. The scientific rationale for these UFs is clear and scientifically justified.
Carignan
a. No comment, this is not my main area of expertise.
b. The scientific rationale supports application of a UFS of 1.
c. Available studies support this conclusion and the analysis is clearly documented.
d. The scientific rationale supports application of a UFS of 10.
e. The provided rationales are scientifically justified and clearly described.
Additional comments:
Suggest defining the abbreviations in Table 5-9 (this is buried and hard to find), explain
why the math doesn't add up (adding uncertainty factors as shown does not produce
the sum shown), and make it clear that the one to the right is the sum. [Tier 2]
Faustman
a. This reviewer thought the derivation of toxicity values was scientifically justified and
were clearly presented by USEPA. Please see the sections above that address these
choices. Tables 5-14 and 5-15 are very clear and provide documentation of these
choices.
b. This reviewer was supportive of the choice of uncertainty factors for immune
effects. The use of human data to set the immune response and to define the
response during early development supports the choice of 1 for UFS. Also, this
156
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
reviewer did feel that the immunotoxicity endpoint represented a biologically
important and sensitive endpoint for derivation of toxicity values.
c. This reviewer was not supportive of the use of the value of 3 to account for lack of a
chronic study in animals. A Tier 1 recommendation is for EPA to review this choice.
Please note that there was not even a 90 day study for extrapolation. This reviewer
feels that the absence of such studies severely compromised our ability to evaluate
more chronic impacts and would suggest a factor of 10 applied to extrapolate the
available data for chronic considerations. This reviewer was supportive of the use of
a value of UPA of greater than 1 to account for the complexities of PPAR alpha and
non PPAR alpha receptor involvement. This was also supported by observations
across other PFAS compounds. The factor of three is thus supported by complexity
of the toxicodynamics across species and the report provides justification.
d. This reviewer supports USEPA's choice of a default factor of 10 to account for the
lack of a chronic study. The text supports this choice.
e. This reviewer supports the rationales for the remaining uncertainty factors. In
regards to the potential lifestage susceptibility, the report clearly discusses how an
endpoint of altered antibody response in early childhood was considered a sensitive
and relevant endpoint to id developmental impacts.
Fisher
a. The lack of fit to pharmacokinetic data for repeated exposures suggests that the
acute dosing CL values do not predict repeated exposures. This is an unfortunate
situation that needs resolved to deal with PFDA pharmacokinetics in animals. 1 think
virtual simulations of populations would be helpful in the future (Tier 3), Monte
Carlo, for example.
b. 1 am not able to answer this question.
c. 1 am not an expert on this issue, but my readings indicate that PPARa is most likely
not an irrelevant endpoint for human health considerations. A UF of 3 is a good
compromise.
d. With a long half-life PFDA may accumulate in the body over time, that is, intake
exceeds excretion from the body. Thus, a lifetime exposure may result in an internal
exposure that is greater than in early life, if the rate of intake exceeds the excretion
rate. 1 recommend that a calculation be completed in this regard using an
animal/human PBPK model for PFDA to gain insights into subchronic and chronic
exposures. If no accumulation appears likely (quasi-steady state) (then justify your
UF by giving examples of chemicals that have 28-day, 90-day and 2-year toxicity
information. Is a UF of 10 justified? Tier 2.
e. 1 can see the rationale, more or less, for PFDA. 1 am stuck with the mixture issues
and the uncertainty about PFDA potency when found in a mixture. 1 think the
epidemiologic tools used to tease out its contribution need to be more transparent,
using a language that all can understand. Tier 2.
Georgopoulos
a. It should be recognized that the Toxicological Review tries to evaluate and
synthesize information from an extensive body of literature that is often
157
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
inconsistent and associated with significant uncertainties; this fact is also true
for the available information on the pharmacokinetics of PFDA for humans as
well as for laboratory animals. Unfortunately, certain statements across the
Toxicological Review and its Appendices reflect the inconsistencies in the
literature. A characteristic example is the following: In the first page of the
section on Pharmacokinetics (page 3-1, lines 18-23) one reads
"While female rats administered PFDA tended to have a higher dose-
normalized area under the plasma concentration time curve (AUC) than males,
(Dzierlenga et al., 2019) suggested that there was no sex difference in PFDA
half-life. However, calculating the average clearance across studies, doses, and
routes, the EPA obtains a value of 6.1 mL/kg-d in male rats and 4.3 mL/kg-day
in female rats, i.e., 30% lower in female rats. The elimination half-life of PFDA
is generally much longer in humans (4.5-12 years) than in rats (20-59 days) or
mice (63-222 days)."
There are issues regarding the consistency of pharmacokinetic information
within the above quote. In fact, the last sentence of the above statement is
repeated on page 3-56 (lines 10-11) but no relevant references are provided in
either case. The source is probably the article (included in the references of the
Toxicological Review) by Zhang et al. (2013b), who estimated the PFDA half-lives
being 4.5 years for women <50 years old and 12 years for males and women >50
years old.
In IRISPFDA_ChargeQuestion5c_Supplementallnformation.PDF (page 2) one can
read:
"...the half-life of PFDA in humans is estimated to be 10-20 years; see External
Review Draft"
This statement contradicts the last sentence of the earlier quote but agrees
(approximately) with the values of half-lives for humans reported in Table 3-3 of
the Toxicological Review (page 3-18) that are 21 years for men, 23 years for
women < 12.4 or > 50 years, and 10.6 years for women 12.4-50 years old. The
half-lives reported for laboratory animals in Table 3-3 are 72 days for male rates,
54 males for female rats, 44 days for male mice and 63 days for female mice.
One should compare these values for rats and mice with those reported on page
3-14 (lines 1-14) of the Toxicological Review:
"While Vanden Heuvel et al. (1991) also evaluated the elimination of PFDA in
rats, they did not report clearance values nor AUC values that could be used to
calculate clearance. The half-lives estimated from the decline in total body
burden (based on 14C activity) were 23 and 43 days in males and females,
respectively, while the half-lives based on blood concentrations were 22 and
29 days, respectively (Vanden Heuvel et al., 1991). These female half-lives are
comparable to the beta-phase half-lives reported for female rats by Dzierlenga
et al. (2019) (18-44 days), though somewhat lower than reported for female
rats by Kim et al. (2019) (50-75 days). The half-life estimates of Vanden Heuvel
et al. (1991) for male rats are between the alpha-phase (1.7-2.1 days) and
beta-phase values (80-110 days) reported by Kim et al. (2019), but much less
158
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
than those reported by Dzierlenga et al. (2019) (215-300 days beta- or single-
phase half-life). This range of half-life values reflects the fact that half-life
estimates are sensitive to noise in the experimental data and study design,
with Vanden Heuvel et al. (1991) having only measured elimination for 28
days, while Dzierlenga et al. (2019) measured plasma concentrations to 105
days and Kim et al. (2019) to 150 days."
The above quotes confirm that uncertainties in interspecies and intraspecies
differences are very significant; so, assumptions extrapolating pattern appearing
in an animal species (e.g., the relationship of urinary to fecal clearance) to
another should be questioned and evaluated. Pharmacokinetics of PFAS in
general are primarily driven by processes such as serum protein binding and
renal reabsorption, and these processes can differ dramatically between animal
models and humans. Dzierlenga et al. (2019) and Fujii et al. (2015) fitted rat and
human time-course data, respectively, with two-compartment pharmacokinetic
models, as PFDA displays the biphasic elimination pattern typical of many PFAS,
with a rapid decline in an initial (alpha) phase and a slower decline in a second
(beta) phase, calculating the two corresponding half-lives. However, this
"piecewise linear" approximation should not lead to a conclusion that there is a
"universal" set of two values for PFDA biological half-lives for a given
species/gender. PFAS kinetics in general involve saturable processes, with high
interspecies and interindividual variabilities, that may induce nonlinearities in
situations of high exposures (or significant co-exposures with other PFAS that
have common molecular targets).
Note al: The Toxicological Report (caption of Figure 3-2 on page 3-21) states
that:
"EPA's replication of the PBPK model exactly reproduces the PBPK model
results of Kim et al. (2019) for oral dosimetry hence is considered an accurate
reproduction of the model. The discrepancy between the PBPK model
prediction for a 1 mg/kg dose and the data demonstrates that the published
model structure and parameters are very inconsistent with the empirical data,
hence that there is a significant flaw in the model."
More specific information is provided on page 3-20, lines 25-33:
"The over-prediction (approximately three to four times higher than these key
pharmacokinetic data for male rats) of the IV data by the Kim et al. (2019)
model indicates that distribution into the body is significantly under-predicted
by the model, which was offset in the simulations of oral dosimetry data by
using an unrealistically low oral bioavailability. Initial efforts to re-fit the
model to the data did not produce acceptable fits to both the IV and oral dose
PK data and involved changing model assumptions in a way that would require
separate experimental validation before use. It was therefore determined that
the published model structure and underlying assumptions did not allow a
sufficiently sound calibration of the model to the PK data, given the currently
available data."
It would be useful to clarify how EPA replicated the model of Kim et al. (2019)
159
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
and specifically if the EPA PBPK templates (coded in R) provided in Bernstein et
al. (2021) were used for that purpose, as in fact these templates correct certain
errors present in the formulation of the Kim et al. (2019) model and provide a
valuable platform for model experimentation and refinement. Appendix G
provides extensive descriptions of the analyses performed for the one- and two-
compartment pharmacokinetic models; it would be informative to describe
what procedures were followed in adjusting the EPA PBPK template in the effort
to develop predictions matching the data reported in Kim et al. (2019). The
statement in the above quote that "[i]nitial efforts to re-fit the model to the
data did not produce acceptable fits to both the IV and oral dose PK data"
should not be interpreted that this is the main reason for not adopting the Kim
et al. (2019) PBPK model. In fact, using the PFDA model template provided in
Bernstein et al. (2021) one can achieve reasonable fits to both the IV and oral
dose PK data by adjusting simultaneously the values of a subset of model
parameters (specifically the fraction unabsorbed, lung-plasma partition
coefficient, transport maximum, and transport affinity constant), but this does
not imply that the "re-parameterized" model fits the data for the right reasons.
It should also be clarified whether EPA considered the possibility that
enterohepatic reabsorption of PFDA may have affected the bioavailability
measurements/estimates reported in Kim et al. (2019). Finally, it should be
recognized that even if a PBPK model with a new calibrated set of parameters
were considered acceptable for PFDA pharmacokinetics in rats, there is
currently no scientifically defensible way to extrapolate that model to humans
since the allometric scaling of parameters used by Kim et al. (2019) is not
applicable to the parameters associated with renal reabsorption processes.
Note a2: Regarding human pharmacokinetics, Fabrega et al. (2015) presented
PBPK modeling results for PFDA; it should be clarified if EPA considered this
source during the development of the Toxicological Review.
Note a3: Finally, it should be pointed out that EPA's httk (high-throughput
toxicokinetics) R package (September 22, 2022 version) includes a human PBPK
formulation for PFDA; however it appears that the PFDA half-life in that model is
approximately only 18 days. Information on whether EPA plans to align future
versions of httk with the findings and approaches adopted in the Toxicological
Review would be useful.
b. The selection of UFS=1 to account for extrapolation from less than lifetime human
data in assessing immune effects should be considered appropriate. The rationale
provided by EPA, i.e., exposures during the developmental period, which is a
susceptible lifestage forimmune effects, can be considered more relevant than
exposures in adulthood and that the observed effects on immune response are
considered to be the result of a cumulative, prolonged exposure. It is also
appropriate to address uncertainties associated with additional susceptible
lifestages (e.g., old age) as part of the UFD.
c. The selection of UFA = 3 for extrapolating liver effects from laboratory animals to
humans in the derivation of the subchronic RfD is appropriate and accounts for
residual uncertainty in interspecies toxicodynamic differences. The rationale
160
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
provided, i.e., the evidence for the involvement of both PPARa-dependent and
PPARa- independent pathways in liver effects, and the associated uncertainties
regarding potential differences in sensitivities across species, support the
conclusion that available data are not adequate to determine if humans are likely
to be equally or less sensitive than laboratory animals with respect to the
observed liver effects. The available animal and mechanistic studies support this
conclusion and the analysis presented in the Toxicological Review is clearly
documented.
d. The value UFS = 10 applied when extrapolating from 28-day animal data to
subchronic exposures for liver, male reproductive, and female reproductive
effects, is appropriate. The reasons provided, i.e., (a) the potential for effects such
as prolonged diestrus, sperm measures and increased liver weight to worsen with
increasing duration and (b) the large uncertainties associated with the lack of data
for evaluating effects of subchronic exposures on liver, male reproductive and
female reproductive outcomes, support this decision.
e. The rationales provided for the remaining uncertainty factors (UFL, UFD, UFH) are
scientifically justified and clearly described.
Suggested Revisions and Future Considerations
• Tier 1 Necessary Revision: Briefly summarize and explain the approach
followed for replicating and adjusting the PBPK model of Kim et al. (2019), i.e.,
using the R templates provided in Bernstein et al. (2021); incorporate this
information in Appendix G following the analysis of the -1 and 2-compartment
PK models. (Please see Answer to Charge Question 6a for more details).
• Tier 1 Necessary Revision: Explain (in Appendix G) why other published
models (e.g., Fabrega et al., 2015), were evaluated as not adequate for
supporting pharmacokinetic calculations.
• Tier 1 Necessary Revision: Check and ensure that values of pharmacokinetic
properties listed in the document match those reported in their cited sources:
an example is such as, e.g. the value of215-300 days for beta- or single-phase
half-life of PFDA in male rats attributed to Dzierlenga et al. (2019).
Haney
a. As described in section Derivation of Candidate Lifetime Toxicity Values for the
RfD of the draft assessment and EPA's A Review of the Reference Dose and
Reference Concentration Processes (U.S. EPA, 2002) and Methods for Derivation
of Inhalation Reference Concentrations and Application of Inhalation Dosimetry
(U.S. EPA, 1994), five possible areas of uncertainty and variability were
considered in deriving the candidate values for PFDA. An explanation of these
five possible areas of uncertainty and variability and the values assigned to each
as a designated UF to be applied to the candidate lifetime PODhed values are
listed In Table 5-9 (pp. 5-19 to 5-20) of the draft. Table 5-9 contains reasoned
justifications for the values used for the various UFs in deriving lifetime
candidate Rf Ds. The rationales are also clearly discussed in the text of that
section.
161
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Similarly, an explanation of the five possible areas of uncertainty and variability
and the values assigned to each as a designated UF to be applied to the
candidate subchronic PODhed values are listed in Table 5-13 (pp. 5-30 to 5-31) of
Section 5.2.3 (Subchronic Toxicity Values for Oral Exposure (Subchronic Oral
Reference Dose [RfD]) Derivation). Table 5-13 contains reasoned justifications
for the values used for the various UFs in deriving subchronic candidate RfDs.
The rationales are also clearly discussed in the text of Section 5.2.3.
a.i. Beyond simply adopting a relatively more conservative approach, as was done
in the draft assessment (e.g., DDEFS versus BW3/4 scaling), I am not sure there
is an acceptable way for EPA to further "account" for the total uncertainty
surrounding interspecies differences in PK (e.g., through quantitative
methods). However, an external PBPK/PK modeler, another peer review panel
member, or a public commenter may be aware of an alternative
method/approach.
b. Yes, EPA provides the rationale needed to support a UFS of 1 on pp. 5-20 (lines 7-18)
to 5-21 (lines 1-14) of the draft assessment. The reasoning is clear, multifactorial,
and strong. For example, the observed immune effects in children are considered
the result of prolonged exposure to PFDA and the enhanced
susceptibility/sensitivity of the developmental immune system, attenuating
concerns of potentially increased sensitivity with longer-term exposures.47
c. At present, it appears that available animal and mechanistic studies support use of a
UFa of 3 as "uncertainty remains regarding the potential differences in sensitivity
across species due to the involvement of both PPARa-dependent and PPARa-
independent mechanisms." EPA provides the rationale supporting a UFA of 3 for
liver effects on pp. 5-31 (lines 18-25) to 5-32 (lines 1-19) of the draft assessment. In
addition to PPARa-dependent MOAs, other MOAs may be active in PFDA-induced
liver effects48, and the uncertainty surrounding a potential multifaceted MOAfor
PFDA-induced liver effects supports the value of 3 selected for the UFA for the
purposes of deriving candidate subchronic RfDs for hepatic effects.
d. EPA provides a rationale supporting a UFS of 10 for these effects on pp. 5-32 (lines
20-38) to 5-33 (lines 1-12) of the draft assessment. EPA states (p. 5-33, lines 7-12),
"Considering the potential for some health effects (prolonged diestrus, sperm
measures and increased liver weight) to worsen with increasing duration and the
large uncertainty associated with the lack of any chemical-specific data on
whether the effects observed in the short-term study worsen after subchronic
47 However, as I do not agree with the selection of these studies and endpoints for RfD derivation, please see my
previous comments.
48 "Although PPARa appears to be an important mechanism of PFDA-induced liver toxicity in animals and reduced
sensitivity in PPAR activation in humans compared to rodents has been suggested, available evidence for PFDA in
PPARa null mice, human in vitro assays and in vivo animal models more relevant to humans with respect to PPARa
sensitivity (i.e., guinea pigs and Syrian hamsters) suggest that liver effects occur, at least in part, independent of PPARa
(see Summary of mechanistic studies for PFDA in Section 3.2.1). A plausible PPARa-dependent and independent MOA
for liver effects is also supported by studies in null and humanized animal models of structurally related long-chain
PFAS [C7-C9] (see Evidence for Other PFAS in Section 3.2.1), which are mostly lacking for PFDA (a few studies in null
mice but no humanized models)." (p. 5-32, lines 6-15).
162
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
exposure, a UFs of 10 is selected for the purposes of deriving candidate
subchronic toxicity values from the 28-day toxicity data." However, while this
quote refers to "chemical-specific data", it is unclear if EPA attempted to consider
whether data from structurally related PFAS could help inform the UFS value for
one or more of these effects (e.g., as they did to some extent for the UFA value for
liver effects). EPA should consider whether data from structurally related PFAS
could help inform the UFS value for one or more of these effects, or if they have
already done so, explicitly state this in Section 5.2.3 of the draft assessment (Tier 1
necessary revision).
e. Yes, as alluded to under subsection "a", Table 5-9 contains reasoned justifications
for the values used for the various UFs, including the UFL, UFD, and UFH, in deriving
lifetime candidate RfDs. The rationales are also clearly discussed in the text of that
draft assessment section (Derivation of Candidate Lifetime Toxicity Values for the
RfD). Similarly, Table 5-13 (pp. 5-30 to 5-31) of Section 5.2.3 (Subchronic Toxicity
Values for Oral Exposure (Subchronic Oral Reference Dose [RfD]) Derivation)
contains reasoned justifications for the values used for the various UFs, including
the UFl, UFDj and UFHj in deriving subchronic candidate RfDs. The rationales are
clearly discussed in the text of Section 5.2.3 of the draft.
Hoberman
a. It is scientifically justified to consider uncertainty in the derivation of the toxicity
values. The review clearly described these uncertainties. The method used to derive
toxicity values for PFDA seem to appropriately account for uncertainties in
evaluating the pharmacokinetic differences between the experimental animal data
and humans.
a.i. The methods used to derive toxicity values for PFDA appropriately account
for uncertainties in evaluating the pharmacokinetic differences between
the experimental animal data and humans.
Tier 1 necessary revision: The lack of consideration of women who are not
menstruating due to amenorrhea, exercise or being on contraception
should be noted in the analysis.
b. The scientific rationale supports the decision to use a UFS of 1 for immune effects.
c. The available animal and mechanistic studies support the conclusion and is clearly
documented. As noted on page 5-38,"Uncertainties remain due to the absence of
longer-term toxicity studies (28 d) and limited information from available
epidemiological studies and in vivo models to characterize the role of PPARa and
other signaling pathways in the mechanisms of hepatotoxicity of PFDA in both
humans and animals."
d. The UF of 10 is supported for the purpose of deriving the subchronic RfD as these
values are based on a single 28 day study in one species with no additional species
or any longer term study in mice. A 90 day study is normally considered to provide
better data for a subchronic RfD.
e. The provided rationales for the remaining uncertainty factors appear to
scientifically justified and are clearly described. With all the different populations
163
-------
Final PFDA Post-Meeting Comments Task Order 68HERH20F0407, Contract EP-C-17-017
and life stages that can occur with exposure to PFDA (neonates, children,
adolescents, adults) it would take a significant more studies to cover all these life
stages.
Tier 1 necessary revision: Data from structurally related PFAS could help inform the
UFS value for one or more of these effects.
Leung
Toxicological pharmacokinetics is not my area of expertise, and 1 defer this question to
the other external reviewers.
Zoeller
a. This is outside my area of expertise.
b. The justification for a UF of 1 for a sensitive life stage was not clear. Tier 2
Recommendation. The Agency should better support the justification for this UF.
c. The justification for a UFA = 3 was clear. No Recommendation.
d. The justification for this decision was clear. No Recommendation.
e. Yes, these were clear. No Recommendation.
3,7 The Toxicologic;!! Review concludes there is Inadequate Information to assess
carcinogenic potential for PFDA and that this descriptor applies to oral and inhalation
routes of human exposure. Please comment on whether the available human, animal and
mechanistic studies, and the analysis presented in the Toxicological Review are
scientifically Justified and clearly described.
Carcinogenicity Hazard Identification and Toxicity Value Derivation
Reviewer
Comments
Adgate
1 concur that that there is inadequate information to assess the carcinogenic potential of
PFDA at this time.
Carignan
This reviewer agrees that there is inadequate information to assess carcinogenic
potential for PFDA at this time.
To avoid readers confusing 'inadequate information' with 'not carcinogenic', recommend
that determinations for PFASs with similar properties (e.g., PFOA and PFOS) be noted.
[Tier 1]
Faustman
In general, this reviewer agrees with the conclusion that it is not possible at this
time to assess carcinogenic potential however this reviewer also felt "limited" in
options to explain how the observations of chromosomal abnormalities and
clastogenic activity are to be handled that are reviewed in this section. For
example, in Table 3-46 studies that showed positive genotoxicity in vitro and in in
vivo for endpoints such as DNA strand breaks, chromosomal aberrations, oxidative
damage and cell cycle changes were discounted as it was difficult to link with long
164
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
term impacts and a chronic endpoint of cancer but these are of concern for their
genotoxicity. Please see statement lines 9-11 page 3-289. Yes this is true for linkage
with cancer however such impacts can have significant health implications and
should be put into the correct hazard identification endpoint section. Were these
reviewed elsewhere in a genotoxicity section? 1 may have missed this. Alternatively,
this could be added to the male reproductive toxicology endpoints as supplemental
data.
The importance of this information is indirectly emphasized again in comments
from the public that were circulated to us in June 2023. On page 2 and 3 of Table 2
there are comments about missing references. Although the comments from NRCD
3-4 are identified as relevant to charge question lb and the literature search
approaches and findings, the specific example of a missing reference Pan et al (see
clarification below) that reports DNA fragmentation in semen in the section on
male reproductive toxicity were sperm impacts are identified is of relevance there
but also when combined with the genotoxicity and chromosomal aberrations above
(now currently discussed in cancer) would support additional, separate review of
genotoxicity. Please note that such "stove piping" can occur with the more
mechanistic data if careful cross inclusion of data is not undertaken. A Tier 1
recommendation is given here to first determine where in the hazard
identification process such DNA impacts will be reviewed and second to
undertake the evaluation of such data for the additional endpoint. This reviewer
would suggest to add to the supplemental information considered for reproductive
impacts.
Clarification: Pan Y, Cui Q, Wang J, Sheng N, Jing J, Yao B, Dai J. Profiles of
Emerging and Legacy Per-/Polyfluoroalkyl Substances in Matched Serum and
Semen Samples: New Implications for Human Semen Quality. Environ Health
Perspect. 2019 Dec;127(12):127005. doi: 10.1289/EHP4431. Epub 2019 Dec 16.
PMID: 31841032; PMCID: PMC6957285.
Fisher
1 am not qualified to answer this.
Georgopoulos
The Toxicological Review appropriately concludes that there is inadequate
information to assess carcinogenic potential for PFDA and that this descriptor
applies to both oral and inhalation routes of human exposure. This decision is
scientifically justified and clearly described in the Toxicological Review.
Haney
Yes, the available human, animal and mechanistic studies along with the analysis
presented in the Toxicological Review support the conclusion that "the evidence is
inadequate to assess carcinogenic potential of PFDA in humans" (Section 5.3, p. 5-
42). Briefly, as indicated in Section 3.3 of the document: (1) the available
epidemiologic evidence on PFDA and the risks of cancer is limited and generally
uninformative (p. 3-288, lines 33-34); (2) there are no long-term animal bioassay
studies available for PFDA (p. 3-289, line 2); and (3) in summary, PFDA does not
appear to elicit a strong genotoxic response as demonstrated by the lack of activity
in most assays described above, including mutagenicity tests in prokaryotic
organisms and mammalian cells; SCE and cell transformation assays in vitro; and
165
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
UDS, oxidative DNA damage and micronucleus assays in rats (p. 3-291, lines 4-7).
This limited evidence amounts to inadequate information to confidently assess the
carcinogenic potential of PFDA for any route of exposure. Accordingly, consistent
with EPA guidance (EPA 2005) to apply a standard descriptor as part of the hazard
narrative and to express a conclusion regarding the weight of evidence for the
carcinogenic hazard potential, a descriptor of inadequate information to assess
carcinogenic potential was applied by the EPA for oral and inhalation exposure to
PFDA and is clearly scientifically justified.
Hoberman
The available human, animal and mechanistic studies are appropriately analyzed and
clearly described, but these studies are inadequate to assess carcinogenic potential for
PFDA.
Leung
Overall, the Review's conclusions that there is inadequate information for a potential
clinical carcinogenic effect of PFDA exposure is scientifically justified, mostly due to the
confounding potential in the human studies and the overall lack of longterm studies for
this outcome, as has been clearly summarized. However, 1 agree with Dr. Faustman's
excellent points raised regarding a possible genotoxic effect of PFDA observed in some
studies, and agree with a Tier 1 recommendation to evaluate where in the hazard
identification process this endpoint will be evaluated.
Zoeller
Yes, this decision was well justified. No Recommendation.
3,8 Given the conclusion there was Inadequate Information to assess carcinogenic potential
for PFDA, the ToxicologlcaJ Review does not derive quantitative estimates for cancer
effects for oral or inhalation exposures. Is this decision scientifically justified and clearly
described?
Carcinogenicity Hazard Identification and Toxicity Value Derivation
Reviewer
Comments
Adgate
This scientific decision is clearly described and scientifically justified.
Carignan
This reviewer agrees estimates could not be derived.
To avoid readers confusing 'inadequate information' with 'not carcinogenic',
recommend that values for PFASs with similar properties (e.g., PFOA and PFOS) be
noted. [Tier 1]
Faustman
This reviewer agrees with this approach at this time. Nevertheless, this reviewer is
always concerned that the public confuses the category of "inadequate information to
assess carcinogenic potential" as a statement of "no carcinogenicity" and encourages
the EPA to be careful in how this information is translated to the public.
It is also the opinion of this reviewer that as more reviews of longer chain PFAS
compounds are published that discuss carcinogenic findings such as the IRAC listing of
166
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
PFOA as a Group 2B carcinogen and the NCI PFAS Exposure and Risk of Cancer
documents are published there will be more interest in determining how and when SAR
approaches and read across extrapolations across the many PRAS compounds will be
implemented. Thus this review proposes a Tier 3 recommendation that future
documents clearly document the chemical domain for when such cross compound
groups are conducted. It maybe helpful to enhance the Chemistry and structure
information for these individual PFAS reports in anticipation of such efforts. It could pull
from and highlight information already being generated within the Computational
Toxicology program. USEPA states the intention to implement. This reviewer applauded
the addition to Appendix D.2 and D.3 however some of these details almost get lost
when just located in the Appendix. Also, Appendix E provides such Comp Tox data
however it was interesting that seemed to be only put into context for informing the
liver effects as it would seem that these assays could also start to inform
neurodevelopmental and also endocrine such as thyroid endpoints. A Tier 2
recommendation is for the authors to add an explanatory paragraph that states how
the Computational Toxicology information, now primarily in the appendix, is or will be
incorporated across endpoints and effects. Such a paragraph is needed to ensure the
intended user of the current IRIS review is clear and can articulate how and when the
additional information from the appendix was used and also perhaps when it was not
used across endpoints.
Fisher
Yes, the lack of animal data or human data justifies this response currently.
Georgopoulos
Given the conclusion that there was inadequate information to assess carcinogenic
potential for PFDA, the only reasonable option is to not derive quantitative estimates
for cancer effects for oral or inhalation exposures. This decision is scientifically
justified and clearly described in the Toxicological Review.
Haney
Yes, the decision to not derive quantitative estimates for cancer effects for oral or
inhalation exposures is scientifically justified. Section 5.3 (p. 5-42) states:
"Considering the limitations in the evidence base across human, animal, and
mechanistic studies of PFDA (see Section 3.3) and in accordance with the
Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005), EPA concluded that
the evidence is inadequate to assess carcinogenic potential of PFDA in humans.
The lack of adequate carcinogenicity data for PFDA precludes the derivation of
quantitative estimates of either oral (oral slope factor, OSF) or inhalation
(inhalation unit risk; IUR) exposure."
Consistent with "lack of adequate carcinogenicity data for PFDA", the derivation of
quantitative risk estimates for either oral (oral slope factor, OSF) or inhalation
(inhalation unit risk; IUR) exposure is precluded. Furthermore, no robust scientific
foundation has been laid, critically reviewed and broadly accepted by the scientific
community for the use of any surrogate PFAS with carcinogenicity data (e.g., PFOA) for
this purpose. Thus, the decision not to derive quantitative estimates for cancer effects
for oral or inhalation exposures is both reasonable and scientifically justified.
167
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
In conclusion, it is obvious that the EPA has put a great deal of time and work into the
draft PFDA assessment. Similarly, the external scientific peer review panel has no doubt
spent significant time and effort in reviewing and providing thoughtful comments on
the draft assessment. At the same time, a great deal of scientific expertise and
resources (e.g., subject area experts, total review time available) exist outside of any
peer review panel or environmental regulatory agency. Additionally, the public is
obviously an important stakeholder in chemical dose-response assessments such as this
that will ultimately be utilized in the protection of public health. Consequently, in
addition to careful review of comments from the peer review panel, the EPA should
duly consider the scientific credibility of all public comments submitted so that the draft
assessment is the beneficiary of EPA staff having considered the most diverse set of
scientifically credible perspectives possible. Thank you for the opportunity to have peer
reviewed this important draft assessment.
Hoberman
The decision not to derive quantitative estimates for cancer effects for oral or
inhalation exposures is scientifically justified and described in the text.
Leung
The inability to derive quantitative estimates for carcinogenic effects associated with
PFDA exposure is scientifically justified and clearly described.
Zoeller
This decision was well justified. No Recommendation.
4.0 ADDITIONAL COMMENTS
Reviewer
Comments
Adgate
Tier 3: The overall PFDA document is the best that could be produced under the
circumstances. This review process and its findings, which relied on limited animal and
human data, demonstrates the key shortcoming of the IRIS chemical-by-chemical
assessment approach: the assessment of PFDA is complicated by the multiple PFAS
humans are routinely exposed to. Given it is hard to disentangle the effects of multiple
similarly persistent PFAS (including PFDA), the Agency should strongly consider treating
these PFAS as a mixture.
Tier 1: All Evidence integration sections for each health endpoint: subheadings for
human and animal determinations will improve the document, the clarity of the
integration process, and judgments presented in each section, especially in sections with
a large number of epidemiology studies. The Agency and section authors should pay
special attention to these sections, as this summary of the determinations made is the
most important part of each section devoted to a specific endpoint/set of endpoints.
Tier 2: The Table of Acronyms is missing several definitions (e.g., DDEF), among others.
This is especially important for relatively new or obscure acronyms that are not part of
the standard risk assessment lexicon. The authors should do a systematic search for any
that are defined but not included in the summary table.
Tier 2: Many Tables and Figures do not stand alone as individual interpretable units that
can be understood without consulting the text. This general editorial rule is widely
168
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
accepted by scientific journals and should be EPA's rule for determining if a table or
figure should be included and is interpretable by readers.
Clarification: This applies to nearly every table and figure. Apply the following rules
(which are the same one's any good scientific Journal uses):
• Does the title describe the contents?
• Are all acronyms defined? (master table of PFAS acronyms exempted)
• Are all headers attached/existing on every page?
• Figures: is everything (axes, colors, etc) clearly labelled.
• And, as the comment says: can a reader figure out everything in the table
WITHOUT consulting the text?
Tier 2: In several places key tables that could fit on one page break across pages (e.g.,
Tables 3.24, 5.1, 5.5, 5.9, etc.; this list is not exhaustive). For tables that have to break
across pages make sure column/sub column headers appear on every page, e.g., Table
3.20, 5-14 and others.
Tier 1: Numbering of the Appendices with key information is incorrect in many places,
e.g., page 1-10 references "Appendix A.6" - which can't be found in the Table of
Contents of the main document or PFDA Supplement.
Faustman
This reviewer was tremendously impressed with the careful, transparent and in-depth
IRIS review that the USEPA conducted for PFDA. On all levels from the transparency in
the systematic review processes, clarity of the evidence reviews and synthesis, enhanced
cross- talk across lines of evidence and detailed and thoughtful considerations on how
toxicokinetics and dynamics are included in the review were all noteworthy. 1 have
reviewed a series of the PFAS compounds and as these have been developed
sequentially, the incorporation of cross compound information has become clearer and 1
applaud the IRIS team for these advances. These current reviews should serve as
excellent examples of how these various integrations can occur. Please see my detailed
comments however 1 wanted to start with this statement as in so many of the discussions
in our open review sessions, the USEPA experts were available during the open
discussion, they were prepared and thoughtful and provided many clarifications and
were able to document additional points for our clarification. These additional details
support my conclusions in reading this draft that at all aspects of the project the team is
truly exceptional in their review and product. Thank you!
One other comment on the linkage with the references and the newly modified HAWC
system appears to be facilitating the translation of information from the original peer
reviewed literature. Also USEPA's discussion and follow-up with authors now
documented in retrievable documents in the reference system has also greatly
documented the process.
This reviewer also found the details in Appendix F.2 and F. 3 on confounding
directionality of co-exposures to be of interest.
169
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Editorial comment:
Please note that throughout the draft report the labeling for the evidence tables at the
end of each of the non-cancer endpoints sections were consistently labeled and that was
helpful to the reviewers however it appears that the evidence tables or figures were not.
Note that sometimes this later table was labeled as a Figure and in the following
examples different names were used including "heat maps" as well as evaluation results.
Please look across the following example tables/figures to make these more consistent:
Table 3-11 and 3-19, Table 3-41 and Figure 3-6.
Editorial recommendation Please note that Figure 3-86, page 3-274 on Kidney
histopathology... has a header that says "PFDA male reproductive organ
histopathology". Please confirm this header?
Georgopoulos
General Comment
Tier 3 Future Consideration: Future efforts and revisions of the assessment for PFDA
(and other PFAS) should consider cumulative risks in the context of real-world
population exposures to PFAS mixtures
Individuals and communities who will experience high PFDA exposures should reasonably
be expected to have above average (and above median) exposures to other PFAS,
including the major legacy PFAS. Furthermore, it is recognized that different PFAS share
multiple common Adverse Outcome Pathways. It is therefore very important for EPA to
continue developing a consistent, integrative, framework for cumulative risk assessments
of PFAS mixtures. Also, though the IRIS Program does not aim to address the exposure
component of risk assessment, it is still important to provide context for the metrics that
are developed in the Toxicological Reviews. This requires a concise but essential
characterization of the distributions and of the geographical and temporal trends of real-
world exposures to PFAS mixtures that include PFDA. As evaluation and assessment of
environmental monitoring is not within the scope of the IRIS program, this
characterization can consist of a summary of information from available biomarker data
(e.g., from NHANES, from on-going CDC/ATSDR biomonitoring studies and NIH-funded
studies across the US, and from international efforts in Canada, the European Union,
China, etc.).
Finally, it should be recognized that the authors of the draft PFDA Toxicological Review
have accomplished a most demanding task: they have developed a thorough and
balanced document and they deserve our thanks.
Selection of Editorial Notes and Recommendations on Specific Issues
• page xi, ABBREVIATIONS AND ACRONYMS: the list appears "generic" and it
looks like the majority of acronyms and abbreviations used in the present
Toxicological Review are not included; as one example CL is listed here as
"Confidence Limit" whereas CL is used as abbreviation for clearance across the
Review document. The list should be expanded if it is intended to be relevant
to the present Toxicological Review
170
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• page 1-1: please ensure that the information appearing in Table 1-1 is
consistent with the information provided on the CompTox Chemicals
Dashboard
• page 1-1: The 2018 reference to ATSDR should be updated
• page 1-8, line 17: PFBA should be PFDA
• page 1-10, line 29: the link to HAWC should be corrected as it takes the reader
to https://hawc.epa.gov/assessment/100500073/ which is ORD IRIS
Assessment PFBA (2022), instead of the PFDA assessment
• page 2-2, Figure 2-1: inconsistencies in listed numbers of references should be
addressed
• page 3-1, line 15: shorted chain of PFAAs should be short chain PFAS
• page 3-11, line 5: PFHxS should be PFDA
• page 3-12, line 4: "appreciable" should be either explained/documented or
deleted
• page 3-12, line 28: PFSA should be PFDA
• page 3-13, line 4: "compared to other short chain of PFAA compounds" should
be corrected
• page 3-14, line 19: why is PFDA compared to "short chained" PFAAs?
• page 3-15, lines 15-16: "these half-life estimates of half-life" should be
corrected
• page 3-15, line 37: PFFDA should be PFDA
• page 3-17, line 36: PFHxS should be PFDA
• page 3-25, line 10: the statement "it is plausible that human clearance is
actually that high" does not appear to be supported by any available data
• page 3-26, line 5: "and are associated" should be "are associated"
• page 3-44, line 6: pregnant should be pregnane
• page 3-67, line, line 15: "four studies study" should be "four studies"
• page 3-98, line 8: "examining examined" should be "examined"
• page 3-98: it would be better to be consistent in using numerals when referring
to the numbers of studies
• page 3-16, lines 31-33: the sentences are incomplete and require editing
• page 3-154, line 25: "Absence of the fifth sternebrae and delayed phalanges
ossific- and mortality at" should be edited/corrected
171
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• page 3-220, Figure 3-72: it should be explained why the study of Kang et al.
(2018) is listed as medium in this figure but is described as low confidence on
page 3-221 (line 29) and page 3-225 (Table 3-33)
• page 3-286, line 32: "to 28" should be "to 28 days"
• page 4-3, footnotes 14 and 15: "are not advance for" should be "are not
advanced for"
• page 5-21, line 11: it is not clear if the cited PFHxS study is unique or if there
are similar results from studies for other PFAS
• pages R-14, R-33, R-35 and elsewhere: on various occasions references to
articles published in the same year that have different authors with the same
last name, e.g. J. Yang and BY Yang, are cited incorrectly as Yang et al. (2022a)
and Yang et al. (2022b)
• page F-23: the first page of Appendix F is F-23; page numbering should be
corrected for this and the pages that follow
• page F-26, line 22: PDFA should be PFDA
• page G-12, line 2: "general" should be "generally"
Additional Studies for Consideration
HUMAN PHARMACOKINETICS
• Bernstein, A. S., Kapraun, D. F., & Schlosser, P. M. (2021). A model template
approach for rapid evaluation and application of physiologically based
pharmacokinetic models for use in human health risk assessments: A case
study on per-and polyfluoroalkyl substances. Toxicological Sciences, 182(2),
215-228
• Fabrega, F., Kumar, V., Benfenati, E., Schuhmacher, M., Domingo, J. L., & Nadal,
M. (2015). Physiologically based pharmacokinetic modeling of perfluoroalkyl
substances in the human body. Toxicological & Environmental Chemistry,
97(6), 814-827.
• Perez, F., Nadal, M., Navarro-Ortega, A., Fabrega, F., Domingo, J. L., Barcelo, D.,
& Farre, M. (2013). Accumulation of perfluoroalkyl substances in human
tissues. Environment international, 59, 354-362.
ANIMAL PHARMACOKINETICS
• Cao, H., Zhou, Z., Hu, Z., Wei, C., Li, J., Wang, L., ... & Liang, Y. (2022). Effect of
enterohepatic circulation on the accumulation of per-and polyfluoroalkyl
substances: evidence from experimental and computational studies.
Environmental Science & Technology, 56(5), 3214-3224.
172
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
IN VITRO STUDIES
• Allendorf, F., Berger, U., Goss, K. U., & Ulrich, N. (2019). Partition coefficients of
four perfluoroalkyl acid alternatives between bovine serum albumin (BSA) and
water in comparison to ten classical perfluoroalkyl acids. Environmental
Science: Processes & Impacts, 21(11), 1852-1863
• Allendorf, F. (2021). Equilibrium sorption of perfluoroalkyl acids (PFAAs) and
four of their alternatives in mammals (Doctoral dissertation, Dissertation, Halle
(Saale), Martin-Luther-Universitat Halle-Wittenberg, 2021).
• Amstutz, V. H., Cengo, A., Sijm, D. T. H. M., & Vrolijk, M. F. (2022). The impact
of legacy and novel perfluoroalkyl substances on human cytochrome P450: An
in vitro study on the inhibitory potential and underlying mechanisms.
Toxicology, 468, 153116.
• Chen, H., Wang, Q., Cai, Y., Yuan, R., Wang, F., & Zhou, B. (2020). Investigation
of the interaction mechanism of perfluoroalkyl carboxylic acids with human
serum albumin by spectroscopic methods. International Journal of
Environmental Research and Public Health, 17(4), 1319.
• Khazaee, M., Christie, E., Cheng, W., Michalsen, M., Field, J., & Ng, C. (2021).
Perfluoroalkyl acid binding with peroxisome proliferator-activated receptors a,
y, and 6, and fatty acid binding proteins by equilibrium dialysis with a
comparison of methods. Toxics, 9(3), 45.
• Modaresi, S. M. S., Wei, W., Emily, M., DaSilva, N. A., & Slitt, A. L. (2022). Per-
and polyfluoroalkyl substances (PFAS) augment adipogenesis and shift the
proteome in murine 3T3-L1 adipocytes. Toxicology, 465, 153044.
• Pierozan, P., Cattani, D., & Karlsson, O. (2022). Tumorigenic activity of
alternative per-and polyfluoroalkyl substances (PFAS): Mechanistic in vitro
studies. Science of The Total Environment, 808, 151945.
• Xie, M. Y., Lin, Z. Y., Liu, L. Y., Wu, C. C., Liu, Y. W., Huang, G. L., & Zeng, E. Y.
(2022). Use of glioma to assess the distribution patterns of perfluoroalkyl and
polyfluoroalkyl substances in human brain. Environmental Research, 204,
112011.
IN SILICO STUDIES
• Yu, S., Ren, J., Lv, Z., Li, R., Zhong, Y., Yao, W., & Yuan, J. (2022). Prediction of
the endocrine-disrupting ability of 49 per-and polyfluoroalkyl substances: In
silico and epidemiological evidence. Chemosphere, 290, 133366.
HUMAN EXPOSURE STUDIES THAT MEASURED BIOMARKERS
• Bao, J., Shao, L. X., Liu, Y., Cui, S. W., Wang, X., Lu, G. L., ... & Jin, Y. H. Target
Analysis and Suspect Screening of Per-and Poly-Fluoroalkyl Substances in
Paired Samples of Maternal Serum, Umbilical Cord Serum, and Placenta Near
173
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Fluorochemical Plants in Fuxin, China. SSRN
http://dx.doi.org/10.2139/ssrn.4091456
• Blomberg A.J., Noren E., Haug L.S., Lindh C, Sabaredzovic A., Pineda D.,
Jakobsson K., Nielsen C. (2023). Estimated Transfer of Perfluoroalkyl
Substances (PFAS) from Maternal Serum to Breast Milk in Women Highly
Exposed from Contaminated Drinking Water: A Study in the Ronneby Mother-
Child Cohort. Environ Health Perspect.l31(l):17005.
• Chen, X., Feng, X., Sun, X., Li, Y., Yang, Y., Shan, G., & Zhu, L. (2022). Quantifying
Indirect Contribution from Precursors to Human Body Burden of Legacy PFASs
Based on Paired Blood and One-Week Duplicate Diet. Environmental Science &
Technology.
• Claessens, J., Pirard, C., & Charlier, C. (2022). Determination of contamination
levels for multiple endocrine disruptors in hair from a non-occupationally
exposed population living in Liege (Belgium). Science of The Total Environment,
815, 152734.
• Fiedler, H., Sadia, M., Krauss, T., Baabish, A., & Yeung, L. W. (2022).
Perfluoroalkane acids in human milk under the global monitoring plan of the
Stockholm Convention on Persistent Organic Pollutants (2008-2019). Frontiers
of Environmental Science & Engineering, 16(10), 1-14.
• Li, N., Ying, G. G., Hong, H., & Deng, W. J. (2021). Perfluoroalkyl substances in
the urine and hair of preschool children, airborne particles in kindergartens,
and drinking water in Hong Kong. Environmental Pollution, 270, 116219.
• Macheka, L. R., Abafe, O. A., Mugivhisa, L. L., & Olowoyo, J. O. (2022).
Occurrence and infant exposure assessment of per and polyfluoroalkyl
substances in breast milk from South Africa. Chemosphere, 288, 132601.
• Peng, L., Xu, W., Zeng, Q., Sun, F., Guo, Y., Zhong, S., ... & Chen, D. (2022).
Exposure to perfluoroalkyl substances in waste recycling workers: Distributions
in paired human serum and urine. Environment international, 158, 106963.
• Xu, Y., Fletcher, T., Pineda, D., Lindh, C. H., Nilsson, C., Glynn, A., ... & Li, Y.
(2020). Serum half-lives for short-and long-chain perfluoroalkyl acids after
ceasing exposure from drinking water contaminated by firefighting foam.
Environmental Health Perspectives, 128(7), 077004.
• Xu, Y., Nielsen, C., Li, Y., Hammarstrand, S., Andersson, E. M., Li, H., ... &
Jakobsson, K. (2021). Serum perfluoroalkyl substances in residents following
long-term drinking water contamination from firefighting foam in Ronneby,
Sweden. Environment international, 147, 106333.
• Zheng, G., Schreder, E., Dempsey, J. C., Uding, N., Chu, V., Andres, G., ... &
Salamova, A. (2021). Per-and polyfluoroalkyl substances (PFAS) in breast milk:
Concerning trends for current-use PFAS. Environmental Science & Technology,
55(11), 7510-7520.
174
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
HUMAN EXPOSURE/RISK STUDIES THAT ONLY MONITOR ENVIRONMENTAL AND
MICROENVIRONMENTAL MEDIA
• Lin, H., Taniyasu, S., Yamashita, N., Khan, M. K., Masood, S. S., Saied, S., &
Khwaja, H. A. (2022). Per-and polyfluoroalkyl substances in the atmospheric
total suspended particles in Karachi, Pakistan: Profiles, potential sources, and
daily intake estimates. Chemosphere, 288, 132432.
• Sonego, E., Simonetti, G., Di Filippo, P., Riccardi, C., Buiarelli, F., Fresta, A., ... &
Pomata, D. (2022). Characterization of organophosphate esters (OPEs) and
polyfluoralkyl substances (PFASs) in settled dust in specific workplaces.
Environmental Science and Pollution Research, 1-15.
• Tang, J., Zhu, Y., Li, Y., Xiang, B., Tan, T., Lv, L., & Luo, Q. (2022). Occurrence
characteristics and health risk assessment of per-and polyfluoroalkyl
substances from water in residential areas around fluorine chemical industrial
areas, China. Environmental Science and Pollution Research, 1-11.
• Wang, F., Wang, W., Zhao, D., Liu, J., Lu, P., Rose, N. L., & Zhang, G. (2022).
Source apportionment and wet deposition of atmospheric poly-and per-
fluoroalkyl substances in a metropolitan city centre of southwest China.
Atmospheric Environment, 273, 118983.
• Wang, Y., Gao, X., Liu, J., Lyu, B., Li, J., Zhao, Y., & Wu, Y. (2022). Exposure to
Emerging and Legacy Polyfluoroalkyl Substances in the Sixth Total Diet Study-
China, 2016-2019. China CDC Weekly, 4(9), 168.
HUMAN POPULATION HEALTH STUDIES
• Cao, L., Guo, Y., Chen, Y., Hong, J., Wu, J., & Hangbiao, J. (2022). Per-
/polyfluoroalkyl substance concentrations in human serum and their
associations with liver cancer. Chemosphere, 296, 134083.
• Duan, Y., Sun, H., Yao, Y., Li, Y., Meng, Y., Lu, Y., ... & Chen, L. (2021). Serum
concentrations of per-/polyfluoroalkyl substances and risk of type 2 diabetes: A
case-control study. Science of The Total Environment, 787, 147476.
• Jin, H., Mao, L., Xie, J., Zhao, M., Bai, X., Wen, J.,... & Wu, P. (2020). Poly-and
perfluoroalkyl substance concentrations in human breast milk and their
associations with postnatal infant growth. Science of the Total Environment,
713, 136417.
• Liu, J. J., Cui, X. X., Tan, Y. W., Dong, P. X., Ou, Y. Q„ Li, Q. Q-, ... & Zhao, X. M.
(2022). Per-and perfluoroalkyl substances alternatives, mixtures and liver
function in adults: A community-based population study in China. Environment
International, 163, 107179.
• Nian, M., Zhou, W., Feng, Y., Wang, Y., Chen, Q., & Zhang, J. (2022). Emerging
and legacy PFAS and cytokine homeostasis in women of childbearing age.
Scientific Reports, 12(1), 6517.
175
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Velarde, M. C., Chan, A. F. O., Sajo, M. E. J. V., Zakharevich, I., Melamed, J., Uy,
G. L. B., ... & Gerona, R. R. (2022). Elevated levels of perfluoroalkyl substances
in breast cancer patients within the Greater Manila Area. Chemosphere, 286,
131545
• Zhang, Y. T., Zeeshan, M., Su, F., Qian, Z. M., Geiger, S. D., McMillin, S. E., ... &
Dong, G. H. (2022). Associations between both legacy and alternative per-and
polyfluoroalkyl substances and glucose-homeostasis: The Isomers of C8 health
project in China. Environment international, 158, 106913.
REVIEWS
• Appel, M., Forsthuber, M., Ramos, R., Widhalm, R., Granitzer, S., Uhl, M., ... &
Gundacker, C. (2022). The transplacental transfer efficiency of per-and
polyfluoroalkyl substances (PFAS): a first meta-analysis. Journal of Toxicology
and Environmental Health, Part B, 25(1), 23-42.
Review rodent and human toxicokinetics
• Zhu, Q„ Li, H., Wen, Z., Wang, Y., Li, X., Huang, T., ... & Ge, R. S. (2020).
Perfluoroalkyl substances cause Leydig cell dysfunction as endocrine
disruptors. Chemosphere, 253, 126764.
Reviews rodent experiments and human epidemiological studies
GENERAL METHODS FOR PFAS RISK ASSESSMENTS (NOT SPECIFIC TO PFDA)
• Chou, W. C., & Lin, Z. (2020). Probabilistic human health risk assessment of
perfluorooctane sulfonate (PFOS) by integrating in vitro, in vivo toxicity, and
human epidemiological studies using a Bayesian-based dose-response
assessment coupled with physiologically based pharmacokinetic (PBPK)
modeling approach. Environment International, 137, 105581.
• Neagu et al. (2021). Adverse outcome pathway in immunotoxicity of
perfluoroalkyls. Current Opinion in Toxicology, 25, 23-29.
LARGE SCALE BIOMONITORING PROGRAMS, FRAMEWORKS, REVIEWS and PROTOCOLS
FOR PFAS THAT ARE RELEVANT TO PFDA
• Calafat, A. M., Kato, K., Hubbard, K., Jia, T., Botelho, J. C., & Wong, L. Y. (2019).
Legacy and alternative per-and polyfluoroalkyl substances in the US general
population: paired serum-urine data from the 2013-2014 National Health and
Nutrition Examination Survey. Environment international, 131, 105048.
• Cousins, I. T., DeWitt, J. C., Gluge, J., Goldenman, G., Herzke, D., Lohmann, R.,
... & Wang, Z. (2020). Strategies for grouping per-and polyfluoroalkyl
substances (PFAS) to protect human and environmental health. Environmental
Science: Processes & Impacts, 22(7), 1444-1460.
• DeLuca, N. M., Angrish, M., Wilkins, A., Thayer, K., & Hubal, E. A. C. (2021).
Human exposure pathways to poly-and perfluoroalkyl substances (PFAS) from
176
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
indoor media: A systematic review protocol. Environment international, 146,
106308.
• ECHA (2016). European Chemicals Agency. Member State Committee Support
Document for identification of nonadecafluorodecanoic Acid (PFDA) and its
sodium and ammonium salts as a Substance of Very High Concern Because of
its toxic for reproduction (Article 57 C) and persistent, bioaccumulative, and
toxic (PBT) (Article 57 D) properties. Adopted on 2 December 2016.
https://echa.europa.eu/documents/10162/77191b64-4f79-0e53-0b89-
57098caf46c4
• EFSA Panel on Contaminants in the Food Chain (EFSA CONTAM Panel), Schrenk,
D., Bignami, M., Bodin, L., Chipman, J. K., del Mazo, J., ... & Schwerdtle, T.
(2020). Risk to human health related to the presence of perfluoroalkyl
substances in food. EFSA Journal, 18(9), e06223.
• EPA SAB (2022). United States Environmental Protection Agency Science
Advisory Board. Review of EPA's Analyses to Support EPA's National Primary
Drinking Water Rulemaking for PFAS. EPA-SAB-22-008. August 22, 2022.
• ITRC (2022) Interstate Technology and Regulatory Council Per- and
Polyfluoroalkyl Substances Technical and Regulatory Guidance. ITRC, 50 F
Street, NW, Suite 350, Washington, DC 20001.https://pfas-l.itrcweb.org/
• Lopez, M. E., Goen, T., Mol, H., N Li bier, S., Haji-Abbas-Zarrabi, K., Koch, H. M., ...
& Castano, A. (2021). The European human biomonitoring platform-Design and
implementation of a laboratory quality assurance/quality control (QA/QC)
programme for selected priority chemicals. International Journal of Hygiene
and Environmental Health, 234, 113740.
Haney
Reviewer's References (not cited in EPA's draft assessment):
• Acosta AM, Moro PL, Hariri S, Tiwari TSP. 2021. Chapter 7: Diphtheria. In:
Centers for Disease Control and Prevention's Epidemiology and Prevention of
Vaccine-Preventable Diseases. Hall E, Wodi AP, Hamborsky J, et al., eds. 14th
ed. Washington, D.C. Public Health Foundation, 2021. Available at:
https://www.cdc.gov/vaccines/pubs/pinkbook/dip.html
• ATSDR (Agency for Toxic Substances and Disease Registry). 2021. Toxicological
Profile for Perfluoroalkyls. U.S. Department of Health and Human Services.
Released May 2021. Available at:
https://www.atsdr.cdc.gov/toxprofiles/tp200.pdf
• FSANZ (Food Standards Australia and New Zealand). 2021. PFAS and
Immunomodulation Review and Update. Available at:
www.foodstandards.gov.au/publications/Documents/PFAS%20and%20lmmun
omodulatorv%20Review%20and%20Update%202021.pdf
177
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
• Hallare J, Gerriets V. Half Life. 2022. In: StatPearls [Internet], Treasure Island
(FL): StatPearls Publishing. Available at:
https://www.ncbi.nlm.nih.gov/books/NBK554498/
• Hendriksen CF, vd Gun JW, Nagel J, Kreeftenberg JG. 1988. The toxin binding
inhibition test as a reliable in vitro alternative to the toxin neutralization test
in mice for the estimation of tetanus antitoxin in human sera. J Biol Stand.
16(4): 287-297.
• Mukaka MM. 2012. Statistics corner: A guide to appropriate use of
correlation coefficient in medical research. Malawi Med J. 24(3): 69-71.
• Padula AM, Ning X, Bakre S, et al. 2023. Birth Outcomes in Relation to
Prenatal Exposure to Per- and Polyfluoroalkyl Substances and Stress in the
Environmental Influences on Child Health Outcomes (ECHO) Program.
Environ Health Perspect. Mar 131(3): 37006.
• Tiwari TSP, Moro PL, Acosta AM. 2021. Chapter 21: Tetanus. In: Centers for
Disease Control and Prevention's Epidemiology and Prevention of Vaccine-
Preventable Diseases. Hall E, Wodi AP, Hamborsky J, et al., eds. 14th ed.
Washington, D.C. Public Health Foundation, 2021. Available at:
https://www.cdc.gov/vaccines/pubs/pinkbook/tetanus.html
• United States Environmental Protection Agency (USEPA). 2021a. External
Peer Review Draft. Proposed Approaches to the Derivation of a Draft
Maximum Contaminant Level Goal for Perfluorooctanoic Acid (PFOA) (CASRN
335-67-1) in Drinking Water. United States Environmental Protection Agency,
Office of Water. EPA Document No: 822D21001.
• United States Environmental Protection Agency (USEPA). 2021b. External
Peer Review Draft. Proposed Approaches to the Derivation of a Draft
Maximum Contaminant Level Goal for Perfluorooctane Sulfonic Acid (PFOS)
(CASRN 1763-23-1) in Drinking Water. United States Environmental
Protection Agency, Office of Water. EPA Document No: 822D21001.
• World Health Organization (WHO). 2017. Tetanus Vaccines: WHO Position
paper - February 2017. Available at:
www.who.int/publications/i/item/tetanus-vaccines-who-position-paper-
februarv-2017
• World Health Organization (WHO). 2018. The Immunological Basis for
Immunization Series Module 3: Tetanus (Update 2018). Available at:
www.who.int/publications/i/item/9789241513616
Appendix A: NHANES Blood serum Data for PFDA
178
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Serum Perfluorodecanoic acid (PFDA) (1999-2000, 2003-2010)
CAS Number 335-76-2
Geometric mean and selected percentiles of serum concentrations (in (jg/L) for the U.S. population from the National Health
and Nutrition Examination Survey.
Demographic
Survey
Geometric Mean
50th Percentile
75th Percentile
90th Percentile
95th Percentile
Sample
Categories
|Years)
(95% CI)
(99% CI)
[95% a)
(95% CI)
(95% CI)
Size
Total population
99-00
¦
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Serum Perfluorodecanoic acid (PFDA) (2011 - 2018)
CAS Number 335-76-2
Geometric mean and selected percentiles of serum concentrations (in pg/L) for the U.S. population from the National Health
and Nutrition Examination Survey.
Demographsc
Survey
Geometric Wean
50th Percentile
75 th Percentile
90th Percentile
95th Percentile
Sample
Categories
(Years)
(95% CI)
(95% CI)
(95% CI)
(95% CI)
(95% CI)
Size
Total population
11-12
.190 (.181-220)
.190 (.170-210)
300 (270-.340)
.480 (-420-.590)
.690 (.600-.770)
1904
Total popiiabon
13-14
.185 (.165-208)
200 (200-.200)
.300 (300-.30Q)
.500 (.400-600}
.700 (€00-.900)
2168
Total popiiabon
15-16
.154 (.140-. 169)
.100 (.100-200)
.300 (200-.300)
-400 (-400-.60Q)
.700 (.5OO-.0OO)
1963
Total popiiabon
17-18
.163 (.178-.209)
200 (200-.200)
,300 (300-.300)
400 (-300-.500)
.600 (.500-900)
1920
Age 12-10 years
11-12
.146 (.126-. 168)
.150 (.120-. 170)
.200 (.180-230)
290 (240-.340)
.360 (290-.56Q)
344
Age 12-16 years
13-14
.136 (.122-152)
.100 (.100-.200)
200 (200-200)
.300 (2D0-.400)
.400 (.300-500)
402
Age 12-19 years
15-16
*
.100 (w limit of detection was too high to prov'de a vaiid result.
Biomonfaxiig Summary: gto.reWhion-onjtor-a.PFA.'S Rir/non»orinn.S^imarv htnrl
Factsheet htlps:/.'www.cclC-gov.'b-:cmon>toring/FFA5 FactSf-eet.html
180
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Reference:
See NHANES Biomonitoring Data Tables at:
https://www.cdc.gov/exposurereport/data tables.html
181
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Appendix A
List of Reviewers
A-l
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
External Peer Review of the EPA Draft "IRIS Toxicological
Review of Perfluorodecanoic Acid (PFDA) and Related Salts"
Monday, July 10, 2023: 10:00 AM - 4:00 PM EDT
Tuesday, July 11, 2023: 10:00 AM - 4:00 PM EDT
Thursday, July 13, 2023: 11:30 AM - 2:15 PM EDT
Virtual Meeting via Zoom.gov
Reviewers
John L. Adgate, Ph.D., MSPH
Professor
Colorado School of Public Health
Department of Environmental and Occupational
Health
University of Colorado Anschutz Campus
Aurora, CO
Panagiotis G. Georgopoulos, Ph.D.
Professor, Rutgers Biomedical and Health
Sciences
Department of Environmental and Occupational
Health and Justice
School of Public Health
Rutgers University
Piscataway, NJ
Courtney C. Carignan, Ph.D.
Assistant Professor
Department of Food Science and Human
Nutrition
Department of Pharmacology and Toxicology
Michigan State University
East Lansing, Ml
Joseph T. Haney, Jr., M.S.
Distinguished Toxicologist
Toxicology Division, Office of the Executive
Director
Texas Commission on Environmental Quality
Austin, TX
Elaine M. Faustman, Ph.D., DABT (Panel
Chair)
Professor, Environmental and Occupational
Health Sciences, and Director, Institute for Risk
Analysis and Risk Communication
School of Public Health
University of Washington
Seattle, WA
Jeffrey W. Fisher, Ph.D.
Senior Research Fellow
Computational Biology Division
ScitoVation
Durham, NC
Alan M. Hoberman, Ph.D., DABT
President
Argus International, Inc.
Horsham, PA
Angela M. Leung, MD, MSc
Associate Professor of Medicine
David Geffen School of Medicine
University of California Los Angeles
Los Angeles, CA
R. Thomas Zoeller, Ph.D.
Professor Emeritus
Department of Biology
University of Massachusetts Amherst
Amherst, MA
A-l
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Appendix B
Charge to Reviewers
B-l
-------
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Technical Charge to External Peer Reviewers
Contract No. EP-C-17-017
Task Order 68HERH20F0407
June 2023
External Peer Review of EPA's Draft IRIS Toxicological Review of
Perfluorodecanoic Acid [PFDA, CASRIM 335-76-2] and Related Salts
INTRODUCTION
The U.S. Environmental Protection Agency (EPA) is seeking a scientific peer review of the draft IRIS
Toxicological Review of Perfluorodecanoic Acid (PFDA) and Related Salts developed in support of the
Agency's online database, the Integrated Risk Information System (IRIS). IRIS is prepared and maintained
by EPA's Center for Public Health and Environmental Assessment within the Office of Research and
Development. IRIS assessments contain information about chemicals that encompasses hazard
identification and dose-response assessment, two of the four steps in the human health risk assessment
process. When used by risk managers in combination with information on human exposure and other
considerations, IRIS assessments support the Agency's regulatory activities and decisions to protect
public health.
There is no existing IRIS assessment for perfluorodecanoic acid (PFDA). The draft Toxicological Review of
PFDA is based on a comprehensive review of the available scientific literature on the noncancer and
cancer health effects in humans and experimental animals exposed to PFDA or salts of PFDA. The
systematic review protocol for PFDA and appendices for dose-response modeling, mechanistic
evaluations and pharmacokinetic information and other supporting materials are provided as
Supplemental Information (see Appendices A to I) to the draft Toxicological Review.
CHARGE QUESTIONS
In response to the numbered charge questions below organized by topic area (italicized headers), the
advice provided as part of this peer review would be most useful when prioritized to indicate its relative
importance as follows:
• Tier 1: Necessary Revisions - Use this category for any revisions you believe are necessary to
adequately support and substantiate the analyses or scientific basis for the assessment
conclusions.
• Tier 2: Suggested Revisions - Use this category for any revisions you encourage EPA to
implement to strengthen the analyses or scientific basis for the assessment conclusions, or to
improve the clarity of the presentation in the PFDA Toxicological Review.
• Tier 3: Future Considerations - Use this category for any advice you have for scientific
exploration that might inform future work. While these recommendations are generally outside
the immediate scope or needs of the PFDA Toxicological Review, they could inform future
reviews or research efforts.
B-3
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Literature Search Methods and Documentation
1. The Toxicological Review for PFDA describes and applies a systematic review protocol for
identifying and screening pertinent studies. The protocol is described in brief detail in Section
1.2.1 (Literature Searching and Screening) and in full detail in Appendix A (Systematic Review
Protocol for the PFAS IRIS Assessments). Please:
a. Comment on whether the literature search strategy and screening criteria for PFDA are
appropriate and clearly described.
b. Identify additional peer-reviewed studies of PFDA that EPA should consider
incorporating prior to finalizing the assessment.
EPA synthesized the literature published through April 2022 in the external reviewdraft and has
been monitoring newly identified studies (i.e., studies identified by EPA or the public that meet
the PECO criteria or otherwise inform key assessment conclusions, but which were not
addressed in the external review draft, for example due to publication after April 2022). EPA will
characterize these studies in a document that will be provided to the peer review panel and the
public and, following the review, included as an Appendix to the assessment prior to finalization.
The characterization will focus on EPA's judgment of whether the studies would have a material
impact on the conclusions (i.e., identified hazards or toxicity values) in the external review draft.
Following receipt of this additional document after the review is underway, please:
c. Review EPA's characterization and provide tiered recommendations regarding which
studies, if any, would have a material impact on the draft's conclusions and should be
incorporated into the assessment before finalizing, as well as your interpretation of the
impact of those studies to be incorporated.
Noncancer Hazard Identification
2. For each health effect considered in the assessment and outlined below, please comment on
whether the available data have been clearly and appropriately synthesized to describe the
strengths and limitations, including whether the presentation and analysis of study results are
clear, appropriate, and effective to allow for scientifically supported syntheses of the findings
across sets of studies. Please comment on whether the study confidence conclusions for the
PFDA studies are scientifically justified, giving appropriate consideration to important
methodological features of the assessed outcomes1. Please specify any study confidence
conclusions that are not justified and explain any alternative study evaluation decisions. For each,
please also comment on whether the weight-of-evidence decisions for hazard identification have
been clearly described and scientifically justified. Note that the data from studies considered
informative to the assessment are synthesized in the relevant health effect-specific sections and
available in the Health Assessment Workspace Collaborative (HAWC).
1 The Toxicological Review provides an overview of individual study evaluations within each evidence synthesis section, and the
results of those outcome-specific evaluations are made available in the Health Assessment Workplace Collaborative linked here
HAWC. Note that a "HAWC FAQ for assessment readers" document, linked here (scroll to the bottom of the page, and the
document is available for download under "attachments"), is intended to help the reviewer navigate this on-line resource.
B-4
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
a. For liver effects, the Toxicological Review concludes that the available evidence indicates
PFDA exposure is likely to cause liver effects in humans given sufficient exposure
conditions, on the basis of a series of short-term studies in rats and mice demonstrating
consistent and coherent effects with a clear biological gradient. The liver findings for PFDA
were similar to those for other structurally-related long-chain PFAS and determined to be
adverse and relevant to humans.
i. Additional considerations influenced the liver effects hazard identification
decisions. Appendix A (Systematic Review Protocol for the PFAS IRIS
Assessments) outlines the human relevance of hepatic effects in animals that
involve PPARa receptors as a key science issue. To the extent supported by
the PFDA literature (and to a lesser extent, literature for other PFAS), the
Toxicological Review evaluates the evidence relevant to the potential
involvement of PPARa and non-PPARa pathways with respect to the reported
liver effects. The Toxicological Review ultimately concludes evidence from in
vivo and in vitro studies support a potential role for multiple pathways
operant in the induction of hepatic effects from PFDA exposure, although
how those pathways interact within a mode of action (MOA) cannot be
specifically determined.
b. For immune effects, the Toxicological Review concludes that the available evidence
indicates PFDA exposure is likely to cause immunosuppression in humans given sufficient
exposure conditions, primarily on the basis of consistent evidence of reduced antibody
responses from two epidemiological studies in children and one study in adults. Although
some evidence for coherent immunomodulatory responses consistent with
immunosuppression was identified in short-term animal studies, the animal evidence
overall is uncertain. The Toxicological Review concludes the immune effects are
considered relevant to humans as the judgment is based on studies in humans.
i. For nearly all epidemiology studies of PFDA, there is potential that exposure to
other highly correlated PFAS could contribute to the observed effects. The evidence
synthesis for potential PFDA-induced immune effects included evaluation of the
potential for confounding across PFAS as well as other sources of confounding and,
based on the available data, determined that residual confounding could explain
part of the observed effect, but concern was minimal, and it was unlikely to fully
explain the associations seen in the literature.
c. For developmental effects, the Toxicological Review concludes that the available evidence
indicates PFDA exposure is likely to cause developmental effects in humans given
sufficient exposure conditions, based primarily on consistent findings of dose-dependent
decreases in fetal weight in mice gestationally exposed to PFDA supported by some
coherent evidence of decreased birth weight from studies of exposed humans in which
PFDA was measured during pregnancy, although uncertainties in the available
epidemiological evidence reduced the impact of these latter findings. The Toxicological
Review concludes the developmental effects in mice are considered relevant to humans
B-5
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
given similar findings of fetal growth restriction in mice and humans.
i. As described in question 3.c and footnote to 3.c, the evidence synthesis for
potential PFDA-induced developmental effects considered potential confounding
factors and concluded that confounding across PFAS or from other potential
sources of bias (e.g., pregnancy hemodynamics in studies where PFDA was
measured during or after pregnancy) introduce significant uncertainty. These
sources of uncertainty ultimately reduce the strength of the available human
evidence to slight for an evidence base that might otherwise be interpreted as
moderate.
d. For male reproductive effects, the Toxicological Review concludes that theavailable
evidence indicates PFDA exposure is likely to cause male reproductive effects in
humans given sufficient exposure conditions, based on coherent evidence in adult
male rats exposed to PFDA for 28 days. Although no direct information on the
human relevance of the animal evidence is available, the findings in animals are
presumed to be relevant based on the conserved role of androgen-dependent
pathways in male productive functions across species.
e. For female reproductive effects, the Toxicological Review concludes that the available
evidence indicates PFDA exposure is likely to cause female reproductive effects in humans
given sufficient exposure conditions, based primarily on coherent evidence from a 28-day
study in adult female rats. Although human studies are available examining associations
between PFDA and female reproductive toxicity (e.g., fecundity), the results were mostly
null, possibly due to their low sensitivity for observing effects. The Toxicological Review
concludes the female reproductive effects are considered relevant to humans given that
mechanisms of female reproduction are similar between rats and humans.
f. For cardiometabolic effects, the Toxicological Review concludes that the available evidence
suggests but is not sufficient to infer that PFDA exposure may have the potential to cause
cardiometabolic effects in humans given sufficient exposure conditions, based on
associations between PFDA and serum lipids, adiposity, cardiovascular disease, and
atherosclerosis in a few epidemiological studies. However, the evidence is largely
inconsistent across studies, which adds considerable uncertainty. Evidence in experimental
animals was indeterminate.
g. For neurodevelopmental effects, the Toxicological Review concludes that the available
evidence suggests but is not sufficient to infer that PFDA exposure may have the potential
to cause neurobehavioral effects in humans given sufficient exposure conditions, based
on associations between PFDA and outcomes related to attention and behavior in
epidemiological studies. However, the evidence is largely inconsistent across studies,
which adds considerable uncertainty. No evidence was found in experimental animals to
inform this outcome (indeterminate).
h. For endocrine, urinary, and other noncancer effects (i.e., hematological, respiratory,
digestive, dermal, musculoskeletal, and nervous systems), the Toxicological Review
B-6
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
concludes there is inadequate evidence to determine whether PFDA exposure has the
potential to cause these effects in humans on the basis of the sparsity of available
evidence.
Noncancer Toxicity Value Data Selection and Modeling
3. For PFDA, no RfC was derived for inhalation exposures. An RfD is derived based on
studies by Budtz-Jorgensen and Grandjean (2018) and Grandjean et al. (2012) showing
decreased serum antibody concentrations for both tetanus and diphtheria in children
(male and female) at age seven years and PFDA measured at age five years and
developmental effects (i.e., reduced birth weight in humans) from the Wikstrom (2020)
study. Given the close proximity of the developmental and immune PODs and resulting
osRfDs and because these effects are observed during the developmental period, they
are selected as co-critical effects supporting the RfD. Are the selection of the studies for
the immune (Budtz-Jorgensen and Grandjean, 2018)and developmental (Wikstrom,
2020) effects for use in deriving the RfD values for PFDA scientifically justified? Are the
modeling approaches appropriate?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effect(s) that should be used to
support the derivation of the lifetime RfD and detail the rationale for use of such an
alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the
effects selected are appropriate for use in deriving the lifetime RfD, including
considerations regarding adversity (or appropriateness in representing an adverse
change) and the scientific support for their selection2. Please also see charge questions 2b
and 2c.
d. EPA used benchmark dose modeling (BMD) (U.S. EPA, 2012) to identify points-of-
departure (PODs) for PFDA. Are the BMD modeling approaches, selection and
justification of benchmark response levels, and selection of the BMD models used to
identify each POD for toxicity value derivation scientifically justified and clearly
described?
e. For liver, male reproductive and female reproductive effects, quantitative information was
limited to studies in animals exposed to PFDA for 28 days and little to no information was
2
For the decreased antibody responses, Selgrade (Tox Sci 2007;100:328-332) suggests that these specific immunotoxic
effects may be broadly indicative of developmental immunosuppression impacting these children's ability to protect against
a range of immune hazards.
For developmental effects (i.e., fetal growth restriction), the human evidence was determined to be slight, primarily due to
potential confounding by hemodynamic changes among studies showing birth weight deficits. For the study (i.e., Wikstrom,
2020) used to derive the developmental RfD, there is no presumed impact of pregnancy hemodynamics given the early
sampling (96% from trimester 1). However, unlike the Wikstrom (2020) study, some uncertainty remains across many of the
available human developmental studies given the predominance of associations that were detected were for studies with
later pregnancy sampling.
B-7
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
available to evaluate the effects of chronic exposure on these health hazards. Therefore,
the derivation of lifetime organ-specific (os) RfD values was not attempted for liver, male
reproductive and female reproductive effects. However, these endpoints were considered
for the derivation of subchronic osRfDs. Does the provided scientific rationale support this
decision? Pleaseexplain.
f. Given the lack of studies on inhalation exposure to PFDA, no reference concentration
(RfC) is derived. Please comment on this decision.
4. In addition, for PFDA, an RfD for less-than-lifetime ("subchronic") exposures is derived. No
subchronic RfC was derived. The same studies and outcomes were chosen for use in deriving the
lifetime and subchronic Rf Ds. Are the selection of these studies and these effects for the
derivation of the subchronic RfD for PFDA scientifically justified?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effect(s) that should be used to support
the derivation of the subchronic RfD and detail the rationale for use of such an
alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the
effects selected are appropriate for use in deriving the subchronic RfD, including
considerations regarding adversity (or appropriateness in representing an adverse
change) and the scientific support for theirselection.
d. Please comment on the other subchronic osRfDs (i.e., for liver, male reproductive, and
female reproductive effects).
e. Given the lack of studies on inhalation exposure to PFDA, no subchronic RfC isderived.
Please comment on this decision.
Noncancer Toxicity Value Pharmacokinetic Extrapolation and Uncertainty Factors
5. Appendix G identifies the potential for pharmacokinetic (PK) differences across species and sexes as
a key science issue and lays out a hierarchy for using relevant PK data in extrapolating doses
between laboratory animals and humans. Section 3.1 evaluates and synthesizes the PK data in
relevant species and sexes, and among human lifestages, up to the derivation of key PK parameters
used in the subsequent analysis. However, the evaluation of existing PBPK models and a one-
compartment PK model found that these options were not sufficiently reliable for use. Given the
information available on potential interspecies differences in PFDA PK, EPA applied a data-derived
extrapolation factor (DDEF) to POD values from toxicity studies in laboratory animals to estimate
corresponding human equivalent doses (HEDs) in the derivation of the respective RfDs. Similarly, the
estimated human clearance (CL) was used to convert internal dose POD (PODint) values from
epidemiological analyses to corresponding HEDs.
After publicly releasing the draft IRIS PFDA assessment, the EPA evaluated recently published data
for several other long-chain PFAS, described here (U.S. EPA, 2023, HERO ID 11181055), that are
potentially relevant to evaluating PFDA dosimetry in women of childbearing age (see question 5c
B-8
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
below).
a. Is applying the estimated DDEF values for PFDA scientifically justified for conversion of PODs
from animal toxicity studies to HEDs? If not, please provide an explanation and detail on a more
appropriate approach.
b. Is application of the human CLto estimate HEDs from PODint values scientifically justified? If
not, please provide an explanation and detail on a more appropriate approach
c. Have the uncertainties in the DDEFs and human CL been adequately evaluated and described?
In answering this question, please provide an explicit recommendation on whether or not EPA
should expand its adjustment for menstrual fluid loss as outlined in (U.S. EPA, 2023, HERO ID
11181055) prior to finalizing the assessment. As these newer data are from other PFAS, note
that such an expansion would be based on the assumption that the pharmacokinetic effect of
pregnancy and lactation on PFDA is similar to that of the other PFAS (i.e., a read-across based
interpretation). See file named IRISPFDA_ChargeQuestion5c_Supplementallnformation.PDF
7. EPA has evaluated and applied where appropriate uncertainty factors to account for
intraspecies variability (UFh), interspecies differences (UFa), database limitations (UFd),
duration (UFs), and LOAEL-to-NOAEL extrapolation (UF|_) for PFDA.
a. Is uncertainty in the derivation of the toxicity values scientifically justified and clearly
described? Please describe and provide comments, if needed.
i. Please comment specifically on whether the methods used to derive toxicity
values for PFDA appropriately account for uncertainties in evaluating the
pharmacokinetic differences between theexperimental animal data and
humans?
b. For immune effects, a UFS of 1 and 3 were considered to account for extrapolation from
less than lifetime human data; ultimately a UFS of 1 was selected. A UFs of 10 was not
considered as the developmental period is recognized as a susceptible lifestage forthese
types of effects and therefore exposure during this time window can be considered more
relevant than exposure in adulthood (U.S. EPA, 1991). Also important is the fact that, given
PFDA's long half-life and the expectation that the children and their mothers have been
exposed to elevated levels of PFDA for many years, the observed effects on immune
response are considered to be the result of a cumulative, prolonged exposure.
Uncertainties with regards to additional susceptible life stages (e.g., old age) are addressed
as part of the UFD. Does the provided scientific rationale support this decision? Please
explain.
c. For liver effects, a value of 3 is applied to extrapolate between effects in laboratory
animals and in humans during the derivation of the subchronic RfD. Although PPARa
dependence might support a value of UFA = 1 if that were the sole pathway leading to
these effects, evidence for the involvement of non-PPARa pathways is available in the
PFDA database. Thus, uncertainty remains regarding the potential differences in sensitivity
across species due to the involvement of both PPARa-dependent and PPARa- independent
B-9
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
mechanisms. As such, the Toxicological Review concludes the available data are not
adequate to determine if humans are likely to be equally or less sensitive than laboratory
animals with respect to the observed liver effects and that a value of UFA = 3 is warranted
to account for the residual uncertainty in toxicodynamic differences across species. Please
comment on whether the available animal and mechanistic studies support this conclusion
and whether the analysis presented in the Toxicological Review is clearly documented.
d. For liver, male reproductive, and female reproductive effects, a default value of 10 is
applied for the UFS when extrapolating from 28-day animal data to a subchronic exposure.
Considering the potential for some health effects (prolonged diestrus, sperm measures and
increased liver weight) to worsen with increasing duration and the large uncertainty
associated with the lack of chemical-specific data to evaluate the effects of subchronic
exposure on liver, male reproductive and female reproductive outcomes, the Toxicological
Review concludes that application of a UFs of 10 is supported for the purposes of deriving
the subchronic RfD from the 28-day toxicity data. Does the provided scientific rationale
support this decision? Please explain.
e. Are the provided rationales for the remaining uncertainty factors (UF|_, UFD, UFH)
scientifically justified and clearly described (to inform the UFH, the assessment evaluates
and considers the available evidence on potential susceptibility to PFDA within different
populations or lifestages, including any potential impacts from early life exposure to PFDA
on children's health or health later in life, although few studies on susceptibility were
available)? If not, pleaseexplain.
Carcinogenicity Hazard Identification and Toxicity Value Derivation
8. The Toxicological Review concludes there is inadequate information to assess carcinogenic
potential for PFDA and that this descriptor applies to oral and inhalation routes of human
exposure. Please comment on whether the available human, animal and mechanistic studies,
and the analysis presented in the Toxicological Review are scientifically justified and clearly
described.
9. Given the conclusion there was inadequate information to assess carcinogenic potential for PFDA,
the Toxicological Review does not derive quantitative estimates for cancer effects for oral or
inhalation exposures. Is this decision scientifically justified and clearly described?
B-10
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
Appendix C
Meeting Agenda
C-l
-------
-------
Final PFDA Post-Meeting Comments
Task Order 68HERH20F0407, Contract EP-C-17-017
External Peer Review of EPA's "Draft IRIS Toxicological
Review of Perfluorodecanoic Acid and Related Salts"
(PFDA)
Monday, July 10, 2023: 10:00 AM - 4:00 PM EDT
Tuesday, July 11, 2023: 10:00 AM - 4:00 PM EDT
Thursday, July 13, 2023: 11:30 AM - 2:15 PM EDT
Virtual Meeting via Zoom.gov
Note: Daily meeting start times are fixed; discussion and break times may be adjusted by reviewers.
Final Agenda
DAY 1: Monday, July 10
10:00 AM Meeting Purpose, Peer Review Process & Reviewer lntros Jan Connery, ERG (facilitator)
10:20 AM U.S. EPA Office of Research and Development (ORD) Background Presentation
11:05 AM Public Comments Jan Connery, ERG
11:15 AM Reviewer Discussion Agenda and Process Jan Connery, ERG
11:25 AM BREAK
11:45 AM Chair Opening Remarks to Panel Peer Review Chair
11:55 AM Reviewer Discussions:
Literature Search Methods and Documentation
11:55 AM Charge Question 1: Literature Search Methods and Documentation Peer Review Pane!
Noncancer Hazard Identification
12:40 PM Charge Question 2fa): Liver Effects Peer Review Panel
1:10 PM BREAK
1:25 PM Charge Question 2fb1: Immune Effects Peer Review Panel
2:20 PM BREAK
2:35 PM Charge Question 2fc): Developmental Effects Peer Review Panel
3:30 PM Charge Question 2(dV. Male Reproductive Effects Peer Review Panel
4:00 PM ADJOURN Day 1
DAY 2: Tuesday, July 11
10:00 AM Day 1 Recap, Day 2 Agenda and Process Jan Connery, ERG
10:05 AM Reviewer Discussions:
Noncancer Hazard Identification (cont.)
C-3
-------
Final PFDA Post-Meeting Comments Task Order 68HERH20F0407, Contract EP-C-17-017
DAY 2: Tuesday, July 11 (cont.)
10:05 AM Charge Question 2fel: Female Reproductive Effects Peer Review Panel
10:35 AM Charge Question 2(f): Cardiometabolic Effects Peer Review Parte!
Reviewer Discussions (cont.):
Noncancer Hazard Identification (cont.)
10:55 AM Charge Question 2fg): Neurodevelopmental Effects Peer Review Panel
11:15 AM BREAK
11:35 AM Charge Question 2fh1: Endocrine. Urinary, and Other Effects Peer Review Panel
Noncancer Toxicity Value Data Selection and Modeling
11:50 AM Charge Question 3: Derivation of Lifetime RfD Peer Review Panel
12:55 PM BREAK
1:20 PM Charge Question 4: Derivation of Subchronic RfD Peer Review Panel
Noncancer Toxicity Value Pharmacokinetic Extrapolation and Uncertainty Factors
1:55 PM Charge Question 5: Pharmacokinetic Extrapolation Peer Review Panel
2:40 PM BREAK
3:05 PM Charge Question 6: Uncertainty Factors Peer Review Panel
3:55 PM Day 3 Preview Jan Connery, ERG
4:00 PM ADJOURN Day 2
DAY 3: Thursday, July 13
11:30 AM Day 2 Recap, Day 3 Agenda and Process Jan Connery, ERG
11:35 AM Reviewer Discussions:
Carcinogenicity Hazard Identification and Toxicity Value Derivation
11:35 AM Charge Question 7: Carcinogenicity Hazard Identification Peer Review Panel
11:55 AM Charge Question 8: Toxicity Value Derivation Peer Review Panel
General Discussion and Closing
12:10 PM Reviewer Integrative Comments and Discussion Peer Review Panel
12:40 PM BREAK
12:55 PM Individual Reviewer Recommendations Peer Review Panel
2:10 PM Closing Remarks EPA, ERG
2:15 PM ADJOURN DAY 3
C-4
------- |