EPA Response to External Peer Review of U.S. EPA’s Draft Acute Freshwater Aquatic Life Benchmarks for Eight Data-Limited PFAS: PFBA, PFHxA, PFNA, PFDA, PFBS, PFHxS, 8:2 FTUCA, and 7:3 FTCA


oEPA

United States	Office of Water	EPA-822-R-24-010

Environmental Protection Agency	4304T	September 2024

EPA Response to External Peer Review of U.S. EPA's

"Draft Acute Freshwater Aquatic Life Benchmarks
for Eight Data-Limited PFAS:

PFBA, PFHxA, PFNA, PFDA, PFBS, PFHxS,
8:2 FTUCA, and 7:3 FTCA"

U.S. Environmental Protection Agency
Office of Water, Office of Science and Technology
Health and Ecological Criteria Division
Ecological Risk Assessment Branch

September 2024

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

TABLE OF CONTENTS

I.	INTRODUCTION	1

II.	CHARGE TO PEER REVIEWERS	3

III.	PEER REVIEWER COMMENTS TABLE	4

I.	General Impressions	4

II.	Response to Charge Questions	16

III.	Specific Observations	54

Appendix A	66

li

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. INTRODUCTION

An independent contractor, Versar Global Solutions, for the Environmental Protection Agency
(EPA), coordinated an external letter peer review of the Derivation of Acute Protective
Freshwater Benchmarks for Selected PFAS through a New Approach Method report (January,
2024). The peer review was conducted for the EPA's Office of Water, Office of Science and
Technology.

Background

The U.S. Environmental Protection Agency (EPA) Office of Water is charged with protecting
human health and the environment from chemicals in water, under the purview of the Clean
Water Act (CWA). In accordance with this mission, the EPA developed acute aquatic life
freshwater benchmark values for eight PFAS: Perfluorobutanoic acid (PFBA),
Perfluorobutanesulfonic acid (PFBS), Perfluorononanoic acid (PFNA), Perfluorodecanoic acid
(PFDA), Perfluorohexanoic acid (PFHxA), Perfluorohexanesulfonic acid (PFHxS),
Hexadecafluoro-2-decenoic acid (8:2 FTUCA), and Pentadecafluorodecanoic acid (7:3 FTC A).
Aquatic life benchmarks, developed under 304(a)(2) of the CWA, are informational values that
the EPA generates when there are limited high quality toxicity data available and data gaps exist
for several families of aquatic organisms. The EPA developed acute benchmarks for these eight
PFAS using available freshwater species empirical test data in conjunction with the application
of a New Approach Method (NAM), specifically the EPA's Office of Research and
Development's (ORD) peer-reviewed web-based Interspecies Correlation Estimate tool (Web-
ICE; Version 4.0; https://www.epa.gov/webice/) (Raimondo et al. 2010). The EPA additionally
investigated the approach described in Giddings et al. (2019) to determine whether a data
binning approach based on similar chemical structure could be applied to the above PFAS to
derive protective values for carboxylic acid PFAS and sulfonic acid PFAS. The EPA also
conducted a comparison of its acute benchmark values to acute data and interim values used by
the Department of Defense in developing ecological screening values (Grippo et al, 2021).

Versar conducted an independent search for scientific experts with expertise in one or more of
the following disciplines: a) application of NAMs to the derivation of protective aquatic life
benchmark values; b) toxicity of PFAS to aquatic life; c) aquatic ecotoxicology; and d) the
acceptability of methods, statistical analyses and data interpretation applied to the determination
of data and methods acceptability.

As a result of this search, the contractor identified and contacted 25 experts and received eight
positive responses expressing interest and availability to participate. The remaining 17 experts
were not available during the peer review timeframe or did not respond to the invitation. For
each interested and available peer reviewer, the contractor evaluated their qualifications and
conducted conflict of interest (COI) screening to ensure that the experts had no COI.

1

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Versar selected the following five scientific experts to serve as peer reviewers:

David Buchwalter, Ph.D.

North Carolina State University

Anupama Kumar, Ph.D.

CSIRO Environment, Australia

Hakon Austad Langberg, Ph.D.

Norwegian Geotechnical Institute, Norway

Ryan Prosser, Ph.D.

University of Guelph, Canada

William Stubblefield, Ph.D.

Oregon State University

2

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

II. CHARGE TO PEER REVIEWERS

1)	Please comment on the overall clarity of the documents and construction as it relates to
assessing the effects and derivation of acute benchmarks for the eight selected PFAS.

2)	Please comment on each of the technical approaches used to derive the draft
benchmark values for the eight selected PFAS presented in EPA's Derivation of Acute
Protective Freshwater Benchmarks for Selected PFAS through a New Approach
Method (ICE-based approach using extrapolation [Sections 3-5], ICE-based approach
using scaled data [Appendix F], and PFAS benchmark calculations using a data binning
approach [Appendix G]).

a.	Are the technical approaches used to derive the benchmark values logical?

b.	Does the science support the conclusions?

c.	Are the approaches and resulting values consistent with the protection of aquatic
life?

d.	For the ICE-based models in particular, please compare and contrast the strengths
and weaknesses of the extrapolation approach [Sections 3-5] with those of the
approach using scaled data [Appendix F],

3)	Please comment on the empirical (direct test) and ICE-generated toxicity data used to
derive the benchmark values presented in the draft document.

a.	Were the data adequately used and sufficiently comprehensive to represent risks
to sensitive aquatic life?

b.	Were the data selected and/or excluded from the benchmark values derivation
appropriately utilized?

c.	Are there relevant data that you are aware of that should be included? If so, please
provide for derivation of benchmark values.

4) Are the derived benchmark values appropriately protective of sensitive aquatic life?

3

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. PEER REVIEWER COMMENTS TABLE

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

This report describes the process and results for acquiring draft acute recommended freshwater aquatic
life benchmarks for eight PFAS: PFBA, PFHxA, PFNA, PFDA, PFBS, PFHxS, 7:3 FTCA, and 8:2
FTUCA. Acute benchmark values were derived using the procedure described in EPA's "Guidelines for
Deriving Numerical National Water Quality Criteria for the Protection of Aquatic Organisms and Their
Uses" where sensitivity distributions based on toxicity databases are used to derive acute freshwater
criteria. Due to the lack of acceptable empirical data for constructing the sensitivity distributions,
Interspecies Correlation Estimation (ICE) models were applied to complete the toxicity databases. The
applied methods are transparent and the methods have undergone previous peer review. The approaches
and data used are reported in a clear manner, making it possible to examine the whole process of
acquiring the benchmarks.

Thank you for your comment.

Some clarification would, however, strengthen the report and make it less likely that readers will
misunderstand what is the appropriate use of these benchmarks:

1) It should be more clearly expressed in the summary that these benchmarks are for acute
exposure, i.e., to protect aquatic life in freshwater from acute toxic effects of PFAS. For
example, these benchmarks are not sufficient to prevent pollution of the environment, adverse
effects of chronic exposure to aquatic animals, human exposure, etc. This is important
information as the problems with PFAS pollution are often due to the combination of their
persistency, mobility, potential for bioaccumulation, and long-term toxicological effects.

Thank you for your comment. Text was added to
further clarify that the values expressed are for
acute exposure and effects, as well as to clarify
that the derived values do not address potential
chronic and/or bioaccumulative effects of these
PFAS chemicals.

2) The assumptions for the use of the ICE models should be summarized in the main report.

The ICE model construct and assumptions
associated with the ICE model application are
already discussed in Sections 2 and 3, and in the
cited literature documenting the application of the
ICE models. The application of ICE models to the
PFAS values does not inherently change or expand
beyond these already-documented model
assumptions.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

3) The resulting benchmark values should be discussed in more detail. Especially trends for acute
toxicity depending on PFAS group and chain length.

Thank you for your comment. Text was added to
Section 5.10 to further discuss and compare the
resulting benchmark values, including a discussion
of the benchmark values in relation to PFAS group
and chain length.

4) There is a lack of clear conclusions regarding the different methods explored for deriving the
benchmarks.

Thank you for your comment. Text was added to
Section 5.10 to further discuss and compare the
resulting benchmark values.

Establishing environmental standards for the protection human health and the environment is critical.
Unfortunately, our laws and regulatory approaches regarding the release of chemicals into the
environment have allowed for the generation of thousands of chemicals used in commerce without the
requirement that they be tested for safety. This leads us to our current situation where the pace of new
chemistries being introduced to the world is far outpacing our ability to evaluate their toxicity to all
forms of life including aquatic life. This document reflects this current state-of-affairs. PFAS as a
chemical class are ubiquitous with new chemistries being introduced a rapid pace. Few environmental
standards for their concentrations in surface waters have been established and toxicity data are limited.
Methods to extrapolate toxicity data are unfortunately necessary in light of this data-limited situation.
Here, the authors use a "New Approach Method" of Interspecies Correlation Estimation developed by
Raimondo and others as a way of generating predictions of toxicity to aquatic organisms such that acute
benchmarks for selected PFAS could be established based on the 1985 Guidelines for the Protection of
Aquatic Life. As these compounds are generally not acutely toxic, and the benchmarks proposed in the
document appear to be orders of magnitude higher than expected environmental concentrations, it is
unclear what the rationale is for proposing these benchmarks in the absence of chronic benchmarks. By
releasing these benchmarks as the only protective values available for the compounds in question, it is
possible that the discovery of environmental concentrations that are well below these benchmarks (but
could be chronically toxic) might not receive an appropriate response by states and tribes. I generally
like the ICE approach to fill data gaps such that environmental standards can be set. It just does not feel
like these acute standards are particularly relevant in light of the expected environmental concentrations.

Thank you for your comment. The acute toxicity
concentrations of these compounds is generally
expected to be higher than chronic, as is typical of
an acute value. However, characterization of what
is a protective acute concentration for chemicals of
concern still remains important. Text was added to
the document to further clarify the values provided
are for acute effects and do not account for
potential chronic and/or bioaccumulative effects.
Importantly, the EPA is working towards
approaches for the development of chronic values
for data-limited chemicals, including PFAS, which
we agree is important, but the NAMs tools to
develop chronic benchmarks were not available to
apply at this time. Additionally, there is a general
lack of chronic data, especially for these PFAS,
upon which to base chronic analyses. Additional
research on the chronic toxicity of a range of
PFAS would support development of additional
values.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

In general, the document requires the reader to consult some of the original literature on ICE models and
there is not enough explanation of these approaches contained within the document itself. I would
encourage the authors to add more technical information about how ICE models work, what their
limitations are, and perhaps build in some uncertainty factors given the language in the Forward section
of the document.

Thank you for your suggestion. Additional details
have been added to the summary description of
ICE models (Section 2), and the description of the
application of ICE models to the derivation of
PFAS benchmarks was expanded (Section 3). The
underlying ICE models have been
comprehensively documented in previous
publications and are available on a public-facing
website, and direct references to these resources
have been provided in the document. Additional
references have been added to the document to
facilitate access to these materials; however,
inclusion of these materials would greatly expand
the size of this document, and in our opinion,
would make it more difficult for the reader to
identify and interpret the specific application and
approach used for benchmark derivation.

The document is generally well-written and easy to follow, especially in the discussion of the data used
and derivation of the benchmark. Greater detail and perhaps some examples would help in the discussion
of the difference between "extrapolated" and "scaled" ICE-based data. It is not clear exactly what the
difference is between these, or at least the implications of the two approaches other than to increase the
number of species represented. Reviewing both the Raimondo et al (2010) and Willming et al (2016)
papers did not address this topic. Raimondo et al (202?) is "in review" and perhaps will address the
issues; however, the manuscript was not included in the review materials. The following discussion
applies to the document and approach and was not specifically addressed in the charge questions.

Thank you for your comment. Text was added to
the summary section to further differentiate
between the extrapolated and scaled ICE-based
models and to provide the reason for increased test
acceptability with the scaled approach. The paper
identified as "in review" has now been published
and the full reference (Raimondo et al. 2024) has
been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds. Integr

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

To sum up the issue, the problem is that there is insufficient high-quality empirical data available to
derive AWQC for the "selected" PFAS compounds. This issue is not new, data limitations in deriving
Ambient Water Quality Criteria (AWQC) have been an issue since shortly after the implementation of
the Clean Water Act (Kimerle et. al., 19851) and became a greater concern with the reduction of AWQC
data development at the EPA-ORD research laboratories. EPA previously proposed an approach to
address this issue in the Type II standards methodology developed as part of the Great Lakes Initiative
(GLI) published in 1995 (USEPA 40 CFR 9, 122, 123, 131, and 132, Final Water Quality Guidance for
the Great Lakes System; Final Rule, March 23 1995). EPA presented a method to develop Secondary
Maximum Concentrations (SMC) and the Secondary Continuous Concentrations (SCC) based on data
sets that were insufficient to satisfy the eight minimum data requirements (MDR) to derive a national
AWQC. Briefly, a secondary acute value (SAV) is calculated by dividing the lowest GMAV in the
database by a Secondary Acute Factor (SAF) that is designated in Table A-l in the document (ranging
from 4.3 to 21.9) based on the number of satisfied MDRs available for the compound. This approach is
somewhat crude and certainly lacks a great deal of technical basis; nonetheless, it probably should be
discussed in the current document. In addition, application of the GLI technique to the PFAS compounds
in this document has been previously conducted and presented in Grippo et al (2021)2. Resulting values
between the Grippo et al (2021) report and application of the method to EPA's data result in slightly
differing values, likely due to the acceptance and availability of different empirical data. In Table 1, a
comparison of the benchmarks reported in this document (using the ICE approach) is made with the
values calculated using the GLI approach (based on the empirical data reported in this document).

Thank you for the comment. Reference to the GLI
methodology and comparison to values presented
in Grippo et al. (2021) were added to the
document summary. The EPA compares the two
sets of values, the EPA's and Grippo et al.
(2021)(Section 5.10) and discusses the use of ICE
vs. assessment factors in developing benchmarks
or screening values (Section 5.1.2). The EPA
added additional discussion comparing these
approaches in the final document, including text
indicating that ICE uses PFAS data for the
derivation of data to fill the data gaps, while the
Grippo et al. (2021) approach uses assessment
factors that did not incorporate data for PFAS.

A further note is that the EPA's final values were
revised to include consideration of new, low
toxicity values from 2 species, the insect mayfly,
Neocloeon traingulifer (Soucek et al. 2023) and
the cladoceran Moina micrura (Razak et al. 2023),
while the Grippo et al. (2021) publication did not
include the consideration of these data for the
derivation of protective values for these PFAS.
The EPA final acute freshwater benchmark values

1 Kimerle, R. A., Werner, A. F., and Adams, W. J., "Aquatic Hazard Evaluation Principles Applied to the Development of Water Quality' Criteria, "Aquatic Toxicology and Hazard Assessment: Seventh Symposium, ASTA1STP854, R. D.
Cardwell, R. Purdy, and R. C. Bahner, Eds., American Society for Testing and Materials, Philadelphia, 1985, pp. 538-547

2 AL Grippo, J. Hayse, I. Hlohowslyj, and K. Picel. 2021. Derivation ofPF*4S Ecological Screening Values. Environmental Science Division. Argonne National Laboratory. September 2021.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Table 1. Comparison of benchmark values using the GLI and ICE-based derivation methods.

Chemical

Lowest empirical
value

GLI factor

Tier II GLI
calculated value
(mg/L)

EPA Benchmark
(Extrapolation)
FAV/2 (mg/L)

Perfluorobutanoic acid (PFBA)

110

S

13.75

83

Perfluorobutanesulfonic acid (PFBS)

1938

13

149.1

183

Perfluorononanoic acid (PFNA)

27.84

13

2.14

10.3

Perfluorodecanoic acid (PFDA)

32

8

4

7.9

Perfluorohexanoic acid (PFHxA)

140

8

17.5

75

Perfluorohexanesulfomc acid
(PFHxS)

Hexadecafluoro-2-decenoic acid (8:2

22.5
3.2

13
13

1.7
0.24

9.1
0.58

FTUCA)

Pentadecafluorodecanoic acid (7:3

0.959

13

0.074

0.18

FTCA)

are compared to the Grippo et al. (2021) values in
Table 5-30 (Reproduced as Table 1, below).

Table 1. Comparison of B
Quality-Based Ecological
Grippo et al. (2021) Using
Al

nchmark Outcomes with Water
Screening Values Calculated by
the Great Lakes Initiative (GLI)
iproach.

Chemical

EPA
Recommended
Acute
Benchmark
(mg/L)

Grippo et
al. (2021)
Calculated
Tier II GLI
Values
(mg/L)

Factor Difference
Between EPA
Acute
Benchmark and
Tier II GLI
Values

Carboxylic Acids

PFBA

5.3

13.75

2.6

PFHxA

4.8

17.5

3.6

PFNA

0.65

2.14

3.3

PFDA

0.50

4

8

Sulfonic Acids

PFBS

5.0

149.1

30

PFHxS

0.21

1.7

8

Grippo, M., J. Hayse, I. Hlohowskyj and K. Picel.
2021. Derivation of PFAS ecological screening
values. Final. September 2021. Environmental
Science Division, Argonne National Laboratory.

Razak, M.R., A.Z. Aris, A.H. Zainuddin, F.M.
Yusoff, Z.N.B. Yusof, S.D. Kim, and K.W. Kim.
2023. Acute toxicity and risk assessment of	

8

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

perfluorooctanoic acid (PFOA) and
perfluorooctanesulfonate (PFOS) in tropical
cladocerans Moina micrura. Chemosphere 313: 9
P-

Soucek, D.J., R.A. Dorman, E.L. Pulster, B.G.
Perrotta, D.M. Walters and J.A. Steevens. 2023.
Perfluorooctanesulfonate adversely affects a
mayfly (Neocloeon triangidifer) at
environmentally realistic concentrations. Environ.
Sci. Technol. Lett. DOI: DOI:
10.1021/acs.estlett.3c00056 <3March2023>

Other comments not specifically addressed in the charge questions are provided below:

• In at least three cases (i.e., PFBS, PFNA, and PFHxS) the derived criteria, did not comply with
the MDR minimum of n=8, it appears that EPA disregarded the MDR minimum for these
materials and calculated the benchmark with fewer MDRs, i.e., 7. This is not keeping with the
1985 guidance that states: "Similarly. if all required data are not available, a numerical criterion
should not be derived except in special cases. " This should be acknowledged in the text and
some statement regarding the minimum number of MDRs to calculate a benchmark addressed.

The EPA is aware that missing the final minimum
data requirement (MDR) could mean additional
uncertainty with the derived value. However, these
values are benchmarks, not criteria, and are
expected to be less certain due to the limited
empirical data on PFAS. Text was added to further
note the uncertainties and discuss the specific
MDRs for which data are missing. Further, the
goal of the EPA here is to provide the best
available scientific information for states, Tribes
and others to consider in their water quality
protection programs, rather than not providing any
information to support environmental protection.

• Table 2 summarizes the MDRs available for each of the 8 PFAS compounds. In all cases 62.5 to
75% of the MDR data used in deriving the SSD-based benchmark was estimated using the ICE
model method. Thus, most of the data used for MDRs and to develop SSDs, are estimated values.
It is interesting that in the best case, only 1 empirical data point was available among the 4 most

Thank you for the comment. As noted, the entire
purpose of developing benchmarks is to provide
information to states, Tribes and other interested
stakeholders on the toxicity of these chemicals
given that there are very limited direct empirical

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

sensitive species for 50% of the materials. The other 4 materials had no empirical data
represented among the most sensitive species. Although there may be good correlations between
species making estimation for one species based on data from another possible, questions remain
regarding the extent of the role that estimated values should play in the derivation of water
quality criteria, standards, or benchmarks.

Table 2. Numbers ofMDRs used in derivation of benchmarks









ff quantitively

Chemical

# of empirical

ff of estimated

Total MDRs met

accepted empirical

MDRs

MDRs

data in 4 lowest
species

Perfluorobutanoic acid (PFBA)

3

5

8

1

Perfluorobutanesulfonic acid

T

5



0

(PFBS)

Z

7

Perfluorononanoic acid (PFNA)

2

5

7

1

Perfluorodecanoic acid (PFDA)

3

5

8

0

Perfluorohexanoic acid (PFHxA)

3

5

8

1

PerQuorohexanesulfonic acid

1

a

5

1

(PFHxS)

z

j

Hexadecafluoro-2-decenoic acid



£.

O

0

(8:2 FTUCA)

z

0

e

Pentadecafluorodecanoic acid



r

o

0

(7:3 FTCA)

z

0

0

test data, including data for the most sensitive test
species.

After the draft was peer reviewed, the EPA
developed application factors to account for two
genera (the mayfly, Neocloeon and the cladoceran,
Moina) that were indicated by direct empirical test
data to have markedly greater sensitivity to PFOA
and PFOS. These application factors were used to
lower the PFAS values to account for these highly
sensitive genera, for which PFAS toxicity data are
not available. The pattern noted for the other direct
empirical test data does not invalidate the
outcome, but instead simply reflects the relative
sensitivity of these highly sensitive species for
which these direct test data were available. The
EPA acknowledges there is greater uncertainty
associated with the calculated benchmark values,
and that the presence of direct empirical test data
amongst the most sensitive test species would
provide a greater degree of certainty. However, the
absence of sensitive species test does not
undermine the validity of the resulting
benchmarks. These values were calculated based
on methods presented in the peer-reviewed
scientific literature and provide useful information
that states and Tribes can choose to use for the
protection of their surface waters.	

• If EPA is going to revise the AWQC minimum MDR data requirement (i.e., 8) for the purpose of
"Benchmark" derivation, then EPA should develop guidance regarding the minimum quantity

Thank you for your comment.

The EPA is not revising the AWQC MDRs for the

purposes of benchmark derivation, but is using a

10

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

and quality of empirical data required before a benchmark can be derived; this should potentially
include:

o a minimum number of empirical data required to be contained among the 4 lowest species
(should benchmarks should be derived based solely on "estimated" values?). Is there a
minimum percent of empirical to estimated data that should be met to establish a
benchmark?

o a requirement for empirical data with a representation among a base set of organisms that
have historically been shown to be sensitive to a range of toxicants, e.g., Ceriodciphnici
dubia, fathead minnow, would be useful. Kimerle et al (1985)3 suggested that a minimum
base data set composed of an algae, daphnid, and fish could consistently predict the most
sensitive species based on available data at the time. Use of the ICE database could help
to identify species consistently shown to be among those predicted to be most sensitive
among chemical groups with a common mode-of-action. The table below, composed
from the ICE-modeled data for PFAS compounds, suggests that a base data set composed
of a freshwater mussel, cladoceran, and fish, would cover the majority of predicted most-
sensitive species for PFAS compounds.

o a minimum number of empirically derived MDR data points should be established, and a
maximum number of ICE-estimated values that are allowed to be considered in an SSD
should be established.

publicly available, data-rich, peer-reviewed
modeling approach to fill in data gaps, in
consonance with the EPA's NAMs workplan,
which was created to support reduction of animal
testing while continuing to protect human health
and the environment, as noted by another peer
reviewer. Further, it is noteworthy that web-ICE is
based on empirical data for many species and
substances.

Although the presence of direct empirical test data
for the chemical of concern amongst the most
sensitive test species could potentially increase the
level of certainty in the resulting values, the
absence of a direct test data for a chemical within
the four most sensitive test values does not
invalidate the development of a benchmark values.
Instead, it may simply reflect an absence of direct
test data for the most sensitive test species for that
chemical. Accordingly, the absence of direct
empirical test data for a chemical amongst the
most sensitive species does not invalidate the
derivation of a benchmark value.

The EPA has also been conducting a separate
evaluation to characterize patterns of relative taxa
sensitivity across chemicals. This ongoing
evaluation of taxa relative sensitivity will further

3 Kimerle, R. A., Werner, A. F., and Adams, W. J., "Aquatic Hazard Evaluation Principles Applied to the Development of Water Quality Criteria, "Aquatic Toxicology and Hazard Assessment: Seventh Symposium, ASTMSTP854, R. D.
Cardwell, R. Purdy, and R. C. Bahner, Eds., American Society for Testing and Materials, Philadelphia, 1985, pp. 538-547.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Table 3. Comparison of the species sensitivity ranking from the ICE modelfor each of the PFAS compounds

Chemical

Species Sensitivity Rank from ICE model

1

Perfluorobutanoic acid
CPFBA)

Perfluorobutanesulfonic
acid (PFBS)
Perfluorononanoic acid
(PFNA)

Perfluorodecanoic acid
(PFDA)

Perfluorohexanoic acid
(PFHxA)

Perfhiorohexanesulfonic
acid (PFHxS)

Hexadecafluoro-2-
decenoic acid (8:2
FTUCA)

Pentadecafluorodecanoic
acid (7:3 FTCA)

Brachionus calyciflorus
(rotifer)
Amblema plicata

(mussel)
Amblema plicata
(mussel)
Caecidotea brevicauda

(isopod)
Brachionus calyciflorus
(rotifer")

Danio rerio (zebrafish)

Amblema plicata
(mussel)

Amblema plicata
(mussel)

Oncorhynchus mykiss

(rainbow trout)
Gammarus fasciatus

(amphipod)
Chydorus sphaericus

(cladoceran)
Mcropterus salmoides
(bass)

Amblema plicata
(mussel)

Jordanella floridae
(flagfish)

Palaemoneies
kadiakensis (grass

shrimp)
Macrobrachium
nipponense (river
	shrimp)	

Gammarus fasciatus
(amphipod)
Hyalella azteca
(amphipod)
Megalonias nervosa
(mussel)
Percaflaxescens
(yellow perch)
Gammarus fasciatus
(amphipod)

Daphnia magna
(cladoceran)

Chydorus sphaericus
(cladoceran)

Chydorus sphaericus
(cladoceran)

Amblema plicata
(mussel)
Ceriodaphnia dubia

(cladoceran)
Oncorhynchus mykiss

(rainbow trout)
Salvelinus fontinalis

(brook trout)
Chydorus sphaericus
(cladoceran)
Limnodrilus
hoffmeisteri
(oligochaete)

Megalonias nervosa
(mussel)

Megalonias nervosa
(mussel)

inform data considerations for the derivation of
future benchmark and criteria values.

In considering the peer reviewers comment that
"Kimerle et al (1985)4 suggested that a minimum
base data set composed of an algae, daphnid, and
fish". The EPA notes that an insect (the mayfly
Neocloeon tricmgulifer) is by far the most acutely
sensitive species for PFOS and among the most
acutely sensitive for PFOA (Soucek at al 2023),
while one cladoceran species (.Moina micrurd) was
the most acutely sensitive species to PFOA and
amongst the most sensitive species to PFOS
(Razak et al 2023) and was much more sensitive
than any other cladoceran species. Although data
on these very sensitive species were not available
for the 8 chemicals for which these acute PFAS
benchmarks were developed, the sensitivities of
these taxa were accounted for through the use of
an application factor that was derived following
the peer review. The calculated PFAS values were
divided by the application factor, to lower the
benchmark values and account for these highly
sensitive species. Additionally, although in most
cases not falling amongst the four most sensitive
species, empirical toxicity text data were available
for daphnids for seven of the eight chemicals
evaluated (PFBA, PFHxA, PFNA, PFDA, PFBS,

4 Kimerle, R. A., Werner, A. F., and Adams, W. J., "Aquatic Hazard Evaluation Principles Applied to the Development of Water Quality Criteria, "Aquatic Toxicology and Hazard Assessment: Seventh Symposium, ASTMSTP854, R. D.
Cardwell, R. Purdy, and R. C. Bahner, Eds., American Society for Testing and Materials, Philadelphia, 1985, pp. 538-547.

12

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE





8:2 FTUCA, 7:3 FTCA) and for fish for six of the
eight chemicals evaluated (PFBA, PFDA, PFBS,
PFHxS, 8:2 FTUCA, 7:3 FTCA), in addition to
web-ICE models more broadly representing these
taxa. For these analyses, the EPA used the most
current data set and models available.

The EPA considers it useful to present the best
available science, using publicly available, peer-
reviewed models, to develop benchmarks as
information for states, Tribes, stakeholders and the
public to consider, instead of providing no
information whatsoever.

Soucek, D.J., R.A. Dorman, E.L. Pulster, B.G.
Perrotta, D.M. Walters and J.A. Steevens. 2023.
Perfluorooctanesulfonate adversely affects a
mayfly (Neocloeon tricmgulifer) at
environmentally realistic concentrations. Environ.
Sci. Technol. Lett. DOI: DOI:
10.1021/acs.estlett.3c00056 <3March2023>.

Razak, M.R., A.Z. Aris, A.H. Zainuddin, F.M.
Yusoff, Z.N.B. Yusof, S.D. Kim, and K.W. Kim.
2023. Acute toxicity and risk assessment of
perfluorooctanoic acid (PFOA) and
perfluorooctanesulfonate (PFOS) in tropical
cladocerans Moina micrura. Chemosphere 313: 9

P-

13

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

• If EPA chooses to develop criteria/benchmarks for materials that have limited empirical data,
then a two-tiered approach should be adopted much like that previously adopted by EPA for the
GLI program. It is unclear from the document what the long-term intent and regulatory status
will be for "benchmarks." Those materials that do not have sufficient empirical data to permit
derivation of a "Tier 1" criteria, could be addressed by a "Tier II" benchmark, as suggested. The
Tier II benchmark could be derived using the proposed ICE-based methodology or a method like
that used for the GLI Secondary Acute Values. That said, will benchmarks serve the same
purpose as the current AWQC? They do not have the same scientific basis as "Tier I" AWQC,
but if adopted as "standards" by states and tribes, they will have the same regulatory/legal status.
This is briefly addressed in the document's forward; however, greater clarity regarding the
"scientific confidence" and "regulatory validity" could be provided. Questions regarding use of
the approach should be considered, for example, if a state developed a proposed standard for a
chemical based on limited empirical data and relying predominately on ICE-estimated data (e.g.,
7:3 FTCA), would EPA approve it?

The EPA is not developing a new two-tiered
approach for criteria. Benchmarks were derived
for these chemicals instead of criteria because
there were insufficient data to develop criteria,
recognizing that the resulting values have a greater
degree of uncertainty than criteria, as stated in the
document. Text has been added to the document to
further clarify the difference between aquatic life
criteria and benchmarks.

The EPA is providing these benchmark values as a
source of information that states can choose to
consider for use in the protection of their surface
waters. The EPA also provides 304(a) ambient
water quality criteria as recommendations for
states and Tribes to consider in their water quality
protection programs. Additionally, states and
Tribes can develop their own scientifically
defensible values and/or develop site-specific
values and submit them for approval as state water
quality standards. The EPA considers all values
states and Tribes develop and propose for use in
their water quality standards.

The draft Derivation of Acute Protective Freshwater Benchmarks for Selected PFAS Compounds
through a New Approach Method (NAM) was well organized and well written. The accuracy of each
element of the derivation process was satisfactory. The structure and writing of the draft document
clearly communicated the rationale and the process of deriving the benchmarks. I think that the proposed
benchmarks are reasonable and protective of aquatic life based on the acceptable empirical acute toxicity
data available for the eight PFAS. The major source of uncertainty for the derived benchmarks is the
lack of acceptable empirical acute toxicity data on freshwater primary producers and freshwater
invertebrates for the eight PFAS. I think the "New Approach Method" is reasonable when there is a lack

Thank you for your comment. The use of a NAM
in calculating protective values is not intended to
replace existing empirical data but is instead
intended to supplement the dataset when limited
direct test data are available and incorporate the
consideration of new approaches going beyond
those developed in EPA's Aquatic Life Criteria
Guidelines. The EPA's NAMs workplan supports

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

I. General Impressions

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

of acceptable empirical data, and a benchmark needs to be derived. However, the NAM should not
replace the derivation of benchmarks with empirical data.

reduction of animal testing while continuing to
protect human health and the environment.

This document provides draft Acute Protective Freshwater Benchmarks for the following eight PFAS
Compounds using New Approach Method (NAM):

1. Perfluorobutanoic acid (PFB A)

2. Perfluorobutanesulfonic acid (PFBS)

3. Perfluorononanoic acid (PFNA)

4. Perfluorodecanoic acid (PFDA)

5. Perfluorohexanoic acid (PFHxA)

6. Perfluorohexanesulfonic acid (PFHxS)

7. Hexadecafluoro-2-decenoic acid (8:2 FTUCA), and

8. Pentadecafluorodecanoic acid (7:3 FTC A)

The detailed methodology used for the derivation of benchmarks has been thoroughly explained. The
process used and results of a systematic review of available empirical toxicity data for aquatic organisms
identified via EPA's literature search for the eight PFAS has been adequately addressed.

Thank you for your comment.

The aquatic life benchmarks for the eight PFAS compounds have been developed using the empirical
and Web-ICE data for these chemicals and were calculated by applying statistical methods. This method
aligns with the EPA's objective to decrease reliance on animal testing by employing NAMS in toxicity
assessment. In addition, the EPA applied 'binning' approach to calculate protective benchmark values
for six PFAS, utilizing combined carboxylic acid (PFBA, PFNA, PFDA, PFHxA) and sulfonic acid
(PFBS, PFHxS) groupings to facilitate value derivation.

Thank you for your comment.

The use of estimated data suggests a proactive approach in addressing gaps in empirical data. It also
demonstrates agencies' commitment to methodological rigor and adaptability in the face of data
challenges when deriving protective values for PFAS compounds.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

II. Response to Charge Questions

Charge Question 1: Please comment on the overall clarity of the documents and construction as it relates to assessing the effects and derivation of acute benchmarks for the
eight selected PFAS.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The overall methodology and data used is clearly presented. The conclusions for the explored
approaches, i.e., the use of extrapolation versus the "scaled" approach and the use of the binning
approach should be made clearer. These conclusions should be included in the summary.

Thank you for your comment. Text was added to
the summary (Section 5.10) to further discuss
attributes and conclusions made about each
approach.

The report would be strengthened by including an introduction to PFAS and its uses (and emissions).
The brief introduction in the summary is very good, however a version with some more details in the
main report would be good to include. Further, I would encourage the authors to include a brief
discussion on the trends for the benchmarks for the different PFAS (including PFOA and PFOS)
compared to the scientific literature (i.e., trends for toxicity depending on group and chain length).

Thank you for your comment. A discussion of
PFAS (including PFOA and PFOS) toxicity based
on chain length was added to the document.
Additionally, references to the final PFOA and
PFOS criteria documents were added. These
documents provide detailed discussions of PFAS
sources, fate, and transport in the environment.

It would be useful to better explain what a Benchmark is related to a Water Quality Criterion. The
forward states that benchmarks are "less certain than Water Quality Criteria", but the reader should also
be informed about enforcement differences between benchmarks and WQC.

Thank you for your suggestion. Text has been
added to the document to further define
benchmarks and their intended application.
Recognizing there is greater uncertainty associated
with these values, as indicated by the
"benchmark" designation, the EPA provides these
values as a source of information. Benchmarks,
provide information that states and Tribes can
choose to use for the protection of their surface
waters.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 1: Please comment on the overall clarity of the documents and construction as it relates to assessing the effects and derivation of acute benchmarks for the
eight selected PFAS.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

In general, the document is clear and well-organized. The sections of the document follow a "template"
making the subsections parallel to each other making it easy to compare. As previously suggested, some
additional detail regarding the extrapolation and scaled estimation technique would be helpful. This may
be contained in the Raimondo et al (in review) document but it was not provided for this review. Also, a
brief discussion of the "binning" technique (Giddings et al 2019) was made in Section 5.10 and in
Appendix G and a comparison of the benchmarks derived using the binning vs ICE-based techniques is
provided, but discussion or assessment of the utility/advantages/disadvantages of the technique is not
provided.

The Raimondo et al. paper identified as "in
review" has now been published and the full
reference (Raimondo et al. 2024) has been added
to the reference list. Text was also added to the
summary (Section 5.10) to further discuss
attributes and conclusions made about each of the
approaches.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds. Integr
Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

I thought the overall clarity of the writing and construction of the document were good. I found the
structure logical and easy to follow in the progression of the derivation process. I would not recommend
any changes to the overall writing or structure of the document. There were a few spelling and
grammatical errors but all very minor.

Thank you for your comment.

Great job on the overall structure and organization of the document! The logical flow and seamless
transitions between sections significantly enhance the readability and understanding of the content.
Information has been laid out in appendices with detailed information on the approaches and examples
for deriving benchmarks. The list of Tables and Figures provides information on all the empirical data
and acceptable ICE models used in deriving aquatic life acute benchmarks of all eight compounds. The
lowest quantitatively acceptable empirical toxicity studies used to derive aquatic life benchmarks for
eight PFAS compounds were detailed in the appendix. Ranked GMAVs and FAVs have been provided
for all eight PFAS compounds. Data incorporated in SSDs have been listed and all figures are self-
explanatory.

Thank you for your comment.

It is apparent that considerable thought and effort were invested in crafting a document with a well-
considered and smooth progression.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 2: Please comment on each of the technical approaches used to derive the draft benchmark values for the eight selected PFAS presented in EPA's
Derivation of Acute Protective Freshwater Benchmarks for Selected PFAS through a New Approach Method (ICE-based approach using extrapolation [Sections 3-5], ICE-
based approach using scaled data /Appendix F], and PFAS benchmark calculations using a data binning approach [Appendix G[).

2. a. Are the technical approaches used to derive the benchmark values logical?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

All these 3 methods are logical. However, I am missing a discussion comparing these
methods. It is mentioned that the "scaled" approach is undergoing evaluation by
Raimondo et al. (in review), however no information from this work is reported. The
lack of a discussion and clear conclusions for the comparisons of these methods makes
it difficult for the reader.

The Raimondo paper identified as "in review" has now been published
and the full reference (Raimondo et al. 2024) has been added to the
reference list. Text was also added to the summary to further discuss
attributes and conclusions made about each of the approaches.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty
analysis and updated user guidance for Interspecies Correlation
Estimation (ICE) models and low toxicity compounds. Integr Environ
Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/ieam,4884

Text was also added to the summary (Section 5.10) to further discuss
attributes and conclusions made about each of the approaches.

I think the reader does not have enough information to evaluate ICE-based approaches
without consulting the original literature. The introduction to Web-ICE (p. 17) is
remarkably brief in explaining the technical approach and this section could be
expanded significantly. It is helpful that the reader is given references to read that point
to the successful applications of the approach, but this document should be self-
contained with respect to describing the technical approach in detail.

Thank you for your comment. Because the underlying approaches
represent publicly-available published methods, the EPA has chosen to
focus the document on describing the specific application of these
methods, rather than restating a description of the tools being used for
the evaluation. References were provided to facilitate access to these
documents.

The same criticism can be applied to the scaled data and data binning approaches.
These technical approaches should be explained in more detail particularly in terms of
the mechanisms by which these different approaches could yield different toxicity
estimates.

Text was added to the summary (Section 5.10) to further discuss
attributes and conclusions made about each of the approaches. However,
as noted above, because the underlying approaches represent publicly-
available published methods, the EPA has chosen to focus the document
on describing the specific application of these methods, rather than
restating a description of the tools used for the evaluation.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2. a. Are the technical approaches used to derive the benchmark values logical?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

In general, I favor action on the creation of environmental standards, even when data
are limiting. The approach of making toxicity predictions is logical and the process
follows the 1985 Guidelines. However, these guidelines are in need of modernization,
and it is unclear to me how aggregated, individual species toxicity tests (that ignore
dietary exposure pathways and species interactions) provide compelling evidence for
protecting aquatic communities in nature.

The EPA is continuing work on revising EPA's Aquatic Life Criteria
Guidelines. Consideration of NAMs-based approaches represents a
component of this revision process. This document focuses on the
consideration of short-term exposures based on the derivation of acute
toxicity values. Dietary exposure pathways and species interactions are
both more relevant to chronic exposures.

The approaches used by the three methods are logical and creative methods to address
the issue of data limitations. Although the calculated data are provided in the document
and in Appendix F, it is difficult for this reviewer to fully understand the technical
differences in the ICE-based approach using extrapolation or scaled data. Perhaps hands
on evaluation of the models or review of the Raimondo et al (in review) manuscript
would help. At the least, an example showing calculations both ways would help the
reader.

Figure 3-1 and supporting text were added to the benchmarks document
(Section 3) to depict side-by-side the extrapolation and scaled
approaches based on a selected example. This example figure is shown
in Appendix A of this document. The Raimondo paper identified as "in
review" has now been published and the full reference (Raimondo et al.
2024) has been added to the reference list. Text was also added to the
summary (Section 5.10) to further discuss attributes, conclusions made,
and differences between each of the approaches.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty
analysis and updated user guidance for Interspecies Correlation
Estimation (ICE) models and low toxicity compounds. Integr Environ
Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

PFAS benchmark calculations using the data binning approach presented in Giddings et
al (2019) are logical and provide a method to expand the quantity of empirical data
provided that the assumption of a common mode of action (MOA) is valid. Raimondo
et al (2010) notes that MOA-specific models are more robust and improve the fit of the
ICE model approach.

Thank you for your comment.

Approaching the problem of missing data using the ICE model and binning techniques
are more elegant than the previous GLI technique and are more generally scientifically
supportable.

Thank you for your comment.

I think the technical approaches taken to derive the benchmark values were logical. In
the absence of acceptable empirical data, the use of the ICE models to generate a data-

Thank you for your comment. Correct, the intent would be to use the
acceptable direct test data for each chemical, as available. In addition to

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2. a. Are the technical approaches used to derive the benchmark values logical?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

rich SSD is a logical approach. Obviously, the benchmarks could be re-evaluated if and
when acceptable empirical data is available to regulators.

the use of ICE models, the EPA has since developed application factors
to account for two genera (the mayfly, Neocloeon and the cladoceran,
Moina) that were indicated by direct empirical test data on PFOA and
PFOS to have markedly greater sensitivity. These application factors
were used to lower the PFAS values to account for these highly sensitive
genera, for which PFAS toxicity data for these eight benchmark
chemicals are not available.

The methodologies employed to determine the benchmark values are rational and sound
from a technical perspective. The available empirical data for the eight PFAS under
consideration fulfill only 2-3 Minimum Data Requirements (MDRs). Consequently, the
EPA opted to employ the peer-reviewed Interspecies Correlation Estimation (ICE)
models developed by Raimondo et al. in 2010. The primary objective of this application
was to provide acute toxicity data to fulfill MDRs in instances where direct toxicity
data were not at hand. The ICE models underwent rigorous evaluation based on
acceptance parameters, including mean square error (MSE), R2, and slope, as delineated
in Box 1. Only models meeting these predefined acceptance criteria were utilized in the
derivation of species-specific toxicity data. This data, when integrated with empirical
toxicity data, served to strengthen the process of establishing benchmark values.

Thank you for your comment. Following the peer review, model
selection has been further refined. Only ICE models based on freshwater
species were utilized. This is based on information indicating that
saltwater species may differ in sensitivity to PFAS from freshwater
species because of their ionic composition. Additionally, when more
than one ICE model was available for a predicted species, only the
model derived using the species with the closest taxonomic relationship
was used, based on the greater accuracy of prediction of models with
closer taxonomic relationships. The EPA considers these two
improvements to further support the scientific rogor of the benchmarks.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2. a. Are the technical approaches used to derive the benchmark values logical?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The EPA's investigation into the "binning" approach for establishing protective values
for grouped carboxylic acid PFASs and grouped sulfonic acid PFAS is grounded in the
precedent established by Giddings et al. in 2019. A similar methodology proved
successful for pyrethroids in that study. This strategic approach involved consolidating
chemicals with shared modes of action, offering advantages in scenarios where
limitations in available data present challenges to value determination. The calculated
values were based on the amalgamation of carboxylic acid compounds (PFB A, PFNA,
PFDA, PFHxA) and sulfonic acid compounds (PFBS, PFHxS), thereby substantiating
the derivation of these values. Calculated benchmark values for carboxylic acids and
sulfonic acids consistently demonstrated higher values when utilizing the SSD
generator in comparison to the Guidelines-based approach. However, these benchmarks
displayed variability in magnitude when contrasted with the ICE-based benchmark
values. This might be influenced by the constrained empirical datasets for certain PFAS
and the restricted number of data points available for the species employed in
normalization.

Thank you for your comment. We concur, most notably, the "binning"
approach incorporated the much larger empirical datasets provided by
the inclusion of PFOA and PFOS than were available for the other
PFAS compounds, which were evaluated using the ICE-based
approaches. This is likely to have influenced the outcome of the
evaluation, as noted by the reviewer.

The EPA also updated the acute benchmarks to reflect improvements in
use of the ICE models (see response immediately above) and also the
use of information indicting expected high sensitivity of two
invertebrate species to these eight data-limited PFAS, based on
empirical data for these sensitive species (Neocloeon tricmgirtifer and
Moina mi crura) with PFOA and PFOS. These refinements to the acute
benchmarks following peer review resulted in the benchmarks being
much closer in magnitude to the calculated binned values using EPA's
Aquatic Life Criteria Guidelines-based calculations. Table 2, provided
below and which is also included in the revised document, shows a
comparison of the benchmark outcomes compared with values that EPA
calculated using the approach by Giddings et al. (2019), using both
EPA's Aquatic Life Criteria Guidelines and SSD Generator-based
calculation methods.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2. a. Are the technical approaches used to derive the benchmark values logical?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Table 2. Comparison of Benchmark Outcomes with Acute Benchmarks Calculated Based on Approach
by Giddings et al. (2019) (Aquatic Life Criteria Guidelines and SSD Generator-Based Values).

Chemical

EPA
Recommended
Acute Benchmark
(mg/L)

Acute Benchm
Using Binnin

arks Calculated
g Approaches

Factor Difference Between
Acute Benchmark and Binning-
based Benchmark Values

Aquatic Life

Criteria
Guidelines-
Based Values
(mg/L)

SSD
Generator-
Based Values
(mg/L)

Aquatic Life

Criteria
Guidelines-
Based Values

SSD
Generator-
Based Values

Carboxylic Acids

PFBA

5.3

19

110

3.6

21

PFHxA

4.8

4.3

24

0.9

5.1

PFNA

0.65

0.35

2.0

0.5

3.0

PFDA

0.50

0.49

2.8

1.0

5.6

Sulfonic Acids

PFBS

5.0

15

75

3.0

15

PFHxS

0.21

0.11

0.56

0.5

2.7

Giddings, J.M., J. Wirtz, D. Campana and M. Dobbs. 2019. Derivation
of combined species sensitivity distributions for acute toxicity of
pyrethroids to aquatic animals. Ecotoxicol. 28: 242-250.

By stating that the derived benchmarks are considered less certain than ambient water
quality criteria, the authors acknowledge a level of uncertainty. This acknowledgment is
crucial in providing a realistic assessment of the reliability of the benchmarks.	

Thank you for your comment.

22

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.b. Does the science support the conclusions?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The applied methods have undergone peer review in previous publications. Previous testing of ICE
model performance indicate validity for the assumptions of this approach: 1) that the relationship of
inherent sensitivity between two species is conserved across chemicals, mechanisms of action, and
ranges of toxicity; and 2) that the nature of a contaminant that was tested on the surrogate reflects the
nature of the contaminant in the predicted species.

Thank you for your comment.

A deeper scientific understanding would require in-depth knowledge of the toxicological mechanisms.
Modes of action (MOA) specific models have previously been reported to be more robust. However,
such mechanistic understanding, including data for MOA specific toxic effects, is not available.
Therefore, the applied methodology can be considered as the best available option.

Thank you for your comment.

The science tells us that these compounds are not acutely toxic and the benchmark values could give
states and tribes a false sense of safety if they encounter high concentrations that are below the
benchmark. The language in the document is explicit about the application as a 1 hour maximum every 3
years, but in the absence of more environmentally relevant standards, I'm not sure what these
benchmarks do for environmental protection.

Thank you for your comment. The acute toxicity
concentrations of these compounds is expected to
be higher than chronic, as is typically the case for
acute vs chronic toxicity. However,
characterization of what is a protective acute
concentration for chemicals still remains
important, especially in the context of PFAS
releases. Text was added to the document to
further clarify the values provided are for acute
effects and do not account for potential chronic
effects and/or bioaccumulation. The EPA is
working towards approaches for the development
of chronic values for such data-limited chemicals,
but the NAMs tools to develop chronic
benchmarks were not available to apply at this
time. Additionally, there is a general lack of
chronic data, especially for these PFAS, upon
which to base chronic analyses. Additional
research on the chronic toxicity of a range of
PFAS would support development of additional
values.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.b. Does the science support the conclusions?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

I am not sure what is being asked since there are no specific conclusions called out in the document. The
approach of estimating species sensitivities to toxins is logical and scientifically supportable, provided
that sufficient underpinning data are available to support the models. Derivation of 1985 AWQC
compliant values (FAV/FCVs) are dependent on 5 datapoints; the four lowest Genus Mean Values
(GMAV/GMCV) and the total number of species represented in the database. The calculation is more
sensitive to the relationship of the 4 lowest values than to the total number of species represented. As an
example, Figure 1 provides an example of increasing the size of the database for a compound (related to
the number of available ICE models); increasing from 8 GMAVs to 42 results in a slightly less than a
doubling in the calculated FAV (8 to 14.5). So, the choice of extrapolated vs scaled ICE models may
result in a slight increase in the calculated benchmark due to acceptance of more models. Far more
important is the validity and relationship of the 4 lowest GMAVs. It is critical that these values be valid
and as accurate as possible. Relying on estimated values can introduce a large degree of uncertainty in
the resulting benchmark value.

15
j 14
13
12
11
10
9
8
7
6

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Number of species in database

Figure 1. Final Acute and Chronic values are sensitive to the number of species in the database

In this case, the conclusions are the benchmark
values. As noted by the reviewer, the lowest four
values play a significant role in determining the
final benchmark value. As previously discussed,
since the peer review the EPA has developed and
included the use of application factors to account
for two genera (the mayfly, Neocloeon and the
cladoceran, Moina) that were indicated by recent
direct empirical test data to have much greater
sensitivity to PFOA and PFOS. These application
factors were used to lower the benchmarks for the
8 data-limited PFAS to account for these
particularly sensitive genera, for which toxicity
data are not available.

Overall, yes, the conclusions are supported by the available science.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.b. Does the science support the conclusions?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

An important element of the science supporting the conclusions is the validation of the ICE models to
predict the acute toxicity of PFAS. In the benchmark document on page 17, an unpublished work by
Raimondo et al. is cited to support the validation of the ICE models to predict acute toxicity of PFAS.
The documents states, "/CE models have been developedfrom a broad range of chemicals (e.g., metals
and other inorganics, pesticides, solvents, and reactive chemicals) and across a wide range of toxicity
values and have been validated as accurate predictors of PFAS acute toxicity when model criteria
parameters are followed (Raimondo et al, in review)." As the unpublished manuscript by Raimondo et
al. is not available as part of this review, I am left to assume that the statement made in the benchmark
document about the validation of the ICE models to predict the acute toxicity of PFAS to be accurate.

The paper identified as "in review" has now been
published and the full reference (Raimondo et al.
2024) has been added to the reference list.

Another question is whether the statement " ...have been validated as accurate predictors of PFAS acute
toxicity... " is solely based on data with PFOS and PFOA?

This statement is referring to the validation of ICE
models across a broad range of chemicals.

This is a valuable contribution to the current scientific knowledge on the toxicity of PFAS compounds to
aquatic life, even while acknowledging the inherent uncertainties associated with using estimated data in
the derivation process.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.c. Are the approaches and resulting values consistent with the protection of aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

In my opinion, the approaches and resulting values are protective for acute toxic effects on aquatic
life as defined in the report (i.e., to be protective of 95% of freshwater genera potentially exposed to
these specific PFAS under short-term conditions of one-hour duration, if the one-hour average
magnitude is not exceeded more than once in three years).

Thank you for your comment.

However, it is important to note that the most problematic properties of PFAS are not the acute
toxicological effects but rather the combination of their persistency, mobility, potential for
bioaccumulation (for some PFAS), and long-term toxicological effects. Emissions of PFAS resulting
in recipient concentrations comparable to the threshold values in this report may (depending on the
volume of water in the recipient) be substantial sources of PFAS pollution to the environment. As
many PFAS (PFCA and PFSA) are both persistent and relatively mobile, such emissions may result
in problematic pollution of drinking water and wildlife with adverse long-term consequences for both
human health and the environment.

Thank you for your comment. The EPA agrees with the
reviewer. However, characterization of what is a
protective acute concentration for chemicals of concern
still remains important. Text was added to the document
to further clarify the values provided are for acute effects
and do not account for potential chronic effects and/or
bioaccumulation. The EPA is working towards
approaches for the development of chronic values for
data-limited chemicals, but the NAMs tools to develop
chronic benchmarks were not available to apply at this
time. There is a general lack of chronic data, especially
for these PFAS, upon which to base chronic analyses.
Additional research on the chronic toxicity of a range of
PFAS would support development of additional values.

See comments above. Adhering to an outdated (1985) understanding of toxicology and species
sensitivity differences remains an unfortunate state-of-affairs at EPA.

The EPA disagrees with the commenter's vague and
negative characterization of the current work. The
NAMs-based approach presented in this document,
although incorporating elements of the EPA's Aquatic
Life Criteria Guidelines (e.g., using sensitivity
distributions and fulfilling MDRs), represents the
application of an innovative approach to addressing data-
limited chemicals. The approach used in this evaluation
does not strictly "adhere" to the EPA's Aquatic Life
Criteria Guidelines, as doing so would have meant that
none of the protective aquatic life values presented
within the document could have been derived.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.c. Are the approaches and resulting values consistent with the protection of aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The use of empirical data on chemical toxicity,
consideration of intra and interspecies (and genus)
sensitivities, and use of new scientific tools to leverage
data are all aspects of this document's analysis and are
also common and current approaches to evaluating the
effects of chemicals on both ecological species and
humans that is applied across the globe. The
understanding of the toxicological activity of PFAS is an
ongoing international research endeavor, both for human
and ecological health.

As previously discussed, the EPA has also been
conducting a separate evaluation to characterize patterns
of relative taxa sensitivity across chemicals. This
ongoing evaluation of taxa relative sensitivity will
further inform data considerations for the derivation of
benchmark and criteria values.

The table below summarizes the benchmarks calculated using the various methods described in the
reviewed document and the values calculated using the GLI method. Values vary substantially, in
some cases as much as an order of magnitude. The Tier II GLI value frequently provided the lowest
calculated value; however, the values were not inconsistent with the other methods. Given the
minimal amount of empirical data available for these materials, it is difficult to identify if the values
are ''consistent with the protection of aquatic life. " Perhaps conducting an analysis with a data rich
compound (e.g., copper or a pesticide), using only a limited portion of the available data followed by
a comparison to the full AWQC database would give some insight into the comparability of the
benchmark and the standard AWQC approach.

Thank you for your comment. The values presented in
the reviewer's comment have been updated in the
revised document and a summary of these revised values
is presented in the table below. The values do vary
between those derived with the benchmark extrapolation
approach and the other approaches (up to a factor
difference of approximately 13 for all chemicals except
for PFBS using the Tier II approach, which differed by a
factor of approximately 30); however, most of the
differences were less than a factor of 10. Text has been
added to the summary section (Section 5.10) to compare
the values derived using the ICE-based, binning, and
Tier II GLI approaches and to discuss in greater detail

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.c. Are the approaches and resulting values consistent with the protection of aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Table 5. Summary of calculated benchmark values

Chemical

EPA
Benchmark
(Extrapolation)
(m.g/L)

EPA
Benchmark
(Scaled)
(m.g/L)

Binning
approach
(Guidelines-
based) (mg/L)

Binning
approach
(SSD-based)
(m.g/L)

Perfluorobutanoic acid
(PFBA)

Perfluorobutanesulfonic acid
(PFBS)

Perfluorononanoic acid
(PFNA)

Perfluorodecanoic acid
(PFDA)

Perfluorohexanoic acid
(PFHxA)

Perfluorohexanesulfonic acid
(PFHxS)

Hexadecafluoro-2-dec enoic
acid (8:2 FTUCA)
Pentadecafluorodecanoic acid
(7:3 FTCA)	

83

183

10.3

7.9

75

9.1

0.58

0.18

174

237
12
10
95
9.4
0.65
0.23

194

24
3.4
4.9
43
0.18

461

102
8.3
12

103
0.76

Tier II GLI
calculated
value (mg.L)

13.75
149.1
2.14

4
17.5
1.7
0.24
0.074

the differences between these values. As noted, values
did vary between the approaches used. However, all the
methodologies tested rely on relatively robust underlying
empirical datasets for their development. The ICE
models rely on underlying empirical test data used to
derive each model, the binning approach bases value
development on direct empirical datapoints for PFAS
that were grouped to determine the chemical-specific
benchmarks, while the Tier II approach is directly based
on available empirical data. The approach used to derive
the ICE-based benchmark values additionally used
empirical data for mayfly {Neocloeon triangulifer) and a
cladoceran (.Moina micrura) to account for these two
highly sensitive species that were not otherwise
represented in the underlying direct empirical or ICE
datasets. All approaches, however, do rely on the
extrapolation of these empirical data to derive the
benchmark values, which is likely to be leading to some
of the differences observed for these values. Consistent
with the suggestion by the reviewer, the EPA has been
exploring the evaluation of these approaches using
chemicals with more data-rich direct toxicity datasets.

28

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.c. Are the approaches and resulting values consistent with the protection of aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Chemical

EPA
Benchmark
(Extrapolation;
mg/L)

EPA
Benchmark
(Scaled;
mg/L)

Binning
Approach
(Guidelines-
based; mg/L)

Binning
Approach
(SSD- based;
mg/L)

Tier II

GLI
Value
(mg/L)

PFBA

5.3

110

13.75

PFHxA

4.8

6.0

4.3

17.5

PFNA

0.65

0.73

0.35

2.0

2.14

PFDA

0.50

0.65

0.49

2.8

PFBS

5.0

7.6

149.1

PFHxS

0.21

0.11

0.56

1.7

8:2

FTUCA

0.037

0.041

7:3 FTCA

0.012

0.015

Yes, I think the approaches and resulting values are consistent with the protection of aquatic life
based on the acceptable empirical data that was available to the assessors.

Thank you for your comment.

Yes, the approaches and resulting values align with the protection of aquatic life. The aquatic life
benchmarks for the eight PFAS compounds were established using empirical and Web-ICE data,
employing statistical methods for calculation. This approach aligns with the EPA's goal to reduce
reliance on animal testing by incorporating NAMS in toxicity assessment.

Thank you for your comment.

Detailed response as in 2a.

Thank you for your comment.

Limitation

ICE models have not been developed for chronic toxicity data and therefore only acute criteria were
developed.

Thank you for your comment. That is correct that ICE
models have been developed to address acute toxicity
only. The EPA is pursuing similar NAMs-based
approaches for the derivation of chronic values.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.d. For the ICE-based models in particular, please compare and contrast the strengths and weaknesses of the extrapolation approach [Sections 3-5] with those of the
approach using scaled data [Appendix FJ.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The use of extrapolation beyond the model range may result in large confidence intervals and hence,
many ICE models will not meet the acceptability parameters. However, in my opinion, this method is
the most intuitive. The "scaled" approach produced more models that met the acceptability parameters.
Nevertheless, based on the results, both these methods provided similar benchmark values.

Thank you for your comment. As noted, the
extrapolation method resulted in fewer models
meeting with acceptability parameters (outlined
in Box 1) than using the scaled approach
(between one and 64 more models accepted for
PFHxS and PFBS, respectively. However, there
was only a marginal increase in the Minimum
Data Requirements met using the scaled
approach (one additional MDR met for PFBS
and for PFHxS). Further, as noted by the
reviewer, both approaches resulted in similar
benchmark values. As discussed within the
document, EPA selected the extrapolation
approach as the primary approach for derivation
of the benchmark values.

It would strengthen the report to include some of the conclusions from Raimondo et al. (in review)
regarding the use of the scaling method.

The paper identified as "in review" has now been
published and the full reference (Raimondo et al.
2024) has been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.

Integr Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

We are not given enough technical information to make this comparison. The scaled approach
information we are given seems more like instructions for how to run the model when values fall out of
environmental realism than a detailed description of how it differs technically from the normal model.

Figures and supporting text were added to the
benchmarks document (Section 3) to compare
the extrapolation and scaled approaches side-by-
side, using a specific example. This example
figure is shown in Appendix A of this document.
Text was also added to the summary section
(Section 5.10) of the benchmarks document to
values derived using the ICE-based
(extrapolation and scaled), binning, and GLI
approaches. Additional details were provided to
further clarify the differences between the
extrapolated and scaled models. Finally, the
recently published paper by Raimondo et al.
(2024) also provides comparison of the scaled vs
extrapolation approach.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.
Integr Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

As stated above, the technical approach between these techniques is not clear. The implication of the
use of the scaled approach rather that the extrapolation method results in the acceptance of more
GMAVs, thus resulting in an increased benchmark.

Figure 3-1 and supporting text were added to the
benchmarks document (Section 3) to compare
the extrapolation and scaled approaches side-by-
side, using a specific example. This example
figure is shown in Appendix A of this document.
Text was also added to the summary section
(Section 5.10) of the benchmarks document to
values derived using the ICE-based
(extrapolation and scaled), binning, and GLI
approaches. Additional details were provided to
further clarify the differences between the
extrapolated and scaled models. Finally, the
recently published paper by Raimondo et al.
(2024) also provides comparison of the scaled vs
extrapolation approach. As noted by the
reviewer, the increase in number of models
accepted using the scaled approach does result in
slightly higher benchmark value.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.

Integr Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Based on the benchmarks derived using the two different approaches, the extrapolation approach
generated lower benchmarks across the eight PFAS compared to using scaled data. For the purpose of
the protection of aquatic life, the extrapolation approach would be more protective than the scaled
approach. I don't know if this would be the case for other groups of chemicals, but it appears that for
PFAS, the extrapolation approach is a more protective approach. In the absence of acceptable empirical
data, the more protective approach should be selected. This is critical to avoiding a type II error (i.e.,
false negative), which is an important consideration in risk assessment.

Thank you for your comment. Additional details
were provided to further clarify the differences
between the extrapolated and scaled models. The
recently published paper by Raimondo et al.
(2024) also provides comparison of the scaled vs
extrapolation approach.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.

Integr Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0,1002/iearn,4884

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

2.d. For the ICE-based models in particular, please compare and contrast the strengths and weaknesses of the extrapolation approach [Sections 3-5] with those of the
approach using scaled data [Appendix Ff.	

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Benchmarks derived using extrapolation approach:

Chemical1

PFBA

(mg/L)

PFBS

(ing/L)

PFNA

(mg/L)

PFDA

(mg/L)

PFHsA

(mg/L)

PFIUS

(mg/L)

8:2
FTUCA

(mg/L)

7:3
FTC A

(mg/L)

Magnitude

83

183

10.3

7.9

75

9.1

0.58

0.18

Duration

One hour average

Frequency

Not to be exceeded more than once in three years oil average

Benchmarks derived using scaled approach:

Chemical1

PFBA

(mg/L)

PFBS

(mg/L)

PFNA

(mg/L)

PFDA

(mg/L)

J'FUxA

(mg/L)

TOM

(mg/L)

8:2
FTUCA

(mg/L)

7:3
FTCA

(mg/L)

Magnitude

174

237

12

10

95

9.4

0.65

0.23

Duration

One hour average

Frequency

Not to be exceeded more than once in three years on average

Thank you for your comment. Additional details
were provided in Section 3 (pgs. 8-10) of the
document to further clarify the differences
between the extrapolated and scaled models. The
recently published paper by Raimondo et al.
(2024) also provides comparison of the scaled vs
extrapolation approach.

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.

Integr Environ Assess Manag. Accepted Author
Manuscript, https://doi.org/10.1002/ieam.4884

Based on the reported validation that has been conducted by Raimondo et al., both approaches seem
reasonable. It would be nice to be able to see the validation that has been conducted by Raimondo et al.,
but it appears that this manuscript is currently in review.

The paper identified as "in review" has now been
published and the full reference (Raimondo et al.
2024) has been added to the reference list.

34

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Raimondo, S., Lilavois, C. and Nelson, S.A.
(2024), Uncertainty analysis and updated user
guidance for Interspecies Correlation Estimation
(ICE) models and low toxicity compounds.
Integr Environ Assess Manag. Accepted Author
Manuscript. httDs://doi.or«/l0.1002/iearn,4884

The acknowledgment of potential challenges, such as large confidence intervals and potential limitations
in accepting ICE models beyond the model range, demonstrates transparency and a thorough
understanding of the modeling process.

Thank you for your comment.

The decision to select the "scaled" approach as an alternative approach for deriving benchmark values,
showcases a proactive and meticulous approach. The alternative scaled approach modifies toxicity
values, as needed, to align them with the ICE model range, avoiding the extrapolation of regressions
beyond the established model range. There is close agreement between the benchmark values calculated
using either approach (as listed in Table 5-26).

Thank you for your comment.

This consideration of alternative methods highlights a commitment to rigorous evaluation and
continuous improvement, reflecting a commendable scientific rigor in the approach to deriving
benchmark values.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 3: Please comment on the empirical (direct test) and ICE-generated toxicity data used to derive the benchmark values presented in the draft document.

3.a. Were the data adequately used and sufficiently comprehensive to represent risks to sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Yes. The data covers the eight MDRs, and hence a wide range of taxa with different characteristics in
aquatic ecosystems. This, combined with the statistical approach which is focusing on the lowest
GMAVs ensures that acute toxic effects on sensitive aquatic life are taken into account. For PFBS and
PFHxS, only seven and six, respectively, of the eight MDRs were fulfilled and hence, these benchmarks
are associated with greater uncertainty. This information should be included in the summary (could be
footnotes in Table Exl-1).

Thank you for your comment. Footnotes and text
were added to the noted summary to clarify this
point.

Probably not. There are too few empirical data to be secure in understanding which species in the real
world might be sensitive.

Thank you for your comment. It should be noted
that the ICE models are derived from empirical
test data, though as discussed, these data are
extrapolated to other species. The binning
approach is also based on empirical data, with the
assumption that the constituent empirical data
have similar or the same MO As.

As previously discussed, following the peer review
the EPA has developed application factors to
account for two species (the mayfly Neocloeon
tricmgulifer and the cladoceran species Moina
micrura) that were indicated by direct empirical
test data to have markedly greater sensitivity to
PFOA and PFOS. These application factors were
used to lower the PFAS values to account for these
particularly sensitive species, for which PFAS
toxicity data are not available. This adjustment
captures the sensitivity of the most sensitive
known species to acute effects of PFAS, yielding
protective values.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.a. Were the data adequately used and sufficiently comprehensive to represent risks to sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Further, the outcomes of the benchmark approach
using ICE data is corroborated by two other
methods for calculating protective values in data -
limited situations, the Giddings et al (2019)
binning approach and the Grippo et al (2021)
approach (using EPA's Aquatic Life Criteria
Guidelines-based approach for the SSD) to
calculate protective values. The protective values
calculated by all 3 methods fall within a factor of
30 of each other (except for PFBS), with most
falling within a factor of less than 10 of each other,
which in the stochastic world of environmental
science is remarkably close, providing further
support and validation of the EPA's approach
using web-ICE. See Table 2 above, and in Section
5.10.3 of the document.

As stated, the problem is a lack of empirical data and a reliance on data estimation techniques. The table
below provides the data for the 4 most sensitive species used to derive the benchmark values for the
PFAS materials. Actual empirical data are highlighted (4 of 32 data points, 12.5%); as you can see,
most of the data for the range of species are estimated values. Without additional confirmatory
experimental data, it is difficult to say anything about how comprehensive the data represent sensitive
aquatic organisms.

Thank you for your comment. The methodologies
tested both rely on relatively robust underlying
empirical datasets for their development. The ICE
models rely on underlying empirical test data used
to derive each model, while the binning approach
bases value development on direct empirical
datapoints for PFAS that were grouped to
determine the chemical-specific benchmarks. Both
approaches, however, do rely on the extrapolation
of these empirical data (to other compounds) to
derive the benchmark values. The EPA has been
exploring the evaluation of these approaches using
chemicals with other, more data-rich direct
toxicity datasets.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.a. Were the data adequately used and sufficiently comprehensive to represent risks to sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Table 4. Species sensitivity ranking for PFAS compounds. Empirical data are in bold, all others are ICE-estimaled values.

Chemical

Species Sensitivity Rank from ICE model1

1

Perfluorobutanoic acid
(PFBA)

Perfluorobutanesulfonic
acid (PEBS)
Perfluorononanoic acid
(PFNA)

Perfluorodecanoic acid
CPFDA)

Perfluorohexanoic acid
(PFHxA)

Perfluorohexanesulfomc
acid (PFHxS)

Hexadecafluoro-2 -
decenoic acid (8:2
FTUCA)

Pentadecafluorodecanoic
acid (7:3 FTCA)

Brachionus
calyci/lorus (rotifer)

Amblema plicata

(mussel)
Amblema plicata
(mussel)
Caecidotea brevicauda
(isopod)
Brachionus
calyci/lorus (rotifer)

Danio rerio (zebrafish)

Amblema plicata
(mussel)

Amblema plicata
(mussel)

Oncorhynchus mykiss

(rainbow trout)
Gammarus fasciatus

(amphipod)
Chydorus sphaericus

(cladoceran)
Micropterus salmoides
(bass)

Amblema plicata
(mussel)

Jordcmella floridae
(flagfish)

Palaemonetes
kadiakensis (grass

shrimp)
Macrobrachium
nipponense (river
	shrimp)	

Gammarus fasciatus
(amphipod)
Hyalella azteca
(amphipod)
Megalonias nervosa
(mussel)
Percaflavescens
(yellow perch)
Gammarus fasciatus
(amphipod)

Daphnia magna
(cladoceran)

Chydorus sphaericus
(cladoceran)

Chydorus sphaericus
(cladoceran)

Amblema plicata
(mussel)
Ceriodaphnia dubia

(cladoceran)
Oncorhynchus mykiss

(rainbow trout)
Salvelimis fontinalis

(brook trout)
Chydorus sphaericus
(cladoceran)
Limnodrilus
hojfmeisteri
(oligochaete)

Megalonias nervosa
(mussel)

Megalonias nervosa
(mussel)

1 It should also be noted that the 4 data points listed in bold are all based on nominal concentrations, so the accuracy of the ECiO
values may be questioned.

I think the available empirical was adequately used.

Thank you for your comment.

However, there was an overall lack of empirical acute toxicity data, particularly for freshwater primary
producers and freshwater invertebrates. For example, there was no acute toxicity data on the eight PFAS
for freshwater primary producers. Freshwater invertebrates were also under-represented in the empirical
data set. For example, there was only three empirical data points for PFHxS and all three were for
freshwater vertebrates. When empirical toxicity data on the eight PFAS was available for a freshwater
invertebrate species, it was usually Daphnia magna. Consequently, I don't think that the data is
sufficiently comprehensive to represent risk to sensitive aquatic life.

Thank you for your comment. The entire effort to
develop benchmarks for the eight PFAS using
web-ICE was pursued precisely because data are
limited, yet the need to protect the environment
from potential impacts of these chemicals remains.
It should be noted that the ICE models are derived
from an extremely extensive empirical test dataset
(>10,700 toxicity records); these data are
extrapolated to other species based on the ICE
models. The binning approach is also based on
empirical data, with the assumption that the	

38

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.a. Were the data adequately used and sufficiently comprehensive to represent risks to sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

constituent empirical data all have similar or the
same MO As.

As previously discussed, following the peer
review, the EPA has developed application factors
to account for two species (the mayfly Neocloeon
tricmgulifer and the cladoceran species Moina
micrura) that were indicated by direct empirical
test data to have markedly greater sensitivity to
PFOA and PFOS. These application factors were
used to lower the PFAS values to account for these
particularly sensitive species, for which PFAS
toxicity data are not available. This adjustment
captures the sensitivity of the most sensitive
known species to acute effects of PFAS, yielding
protective values.

The EPA agrees that additional data generation by
researchers on the toxicity of PFAS to aquatic life,
and other species would be useful to inform
development of protective values.

The EPA employed both empirical test data and ICE values, derived for missing Minimum Data
Requirements (MDRs), to determine acute freshwater benchmark recommendations for aquatic life. The
utilization of ICE-predicted values by various independent, international groups to establish protective
values for aquatic life confirms that values derived from ICE-generated Species Sensitivity Distributions
(SSDs) offer a consistent level of protection comparable to using directly measured laboratory data.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

The data that was deemed of sufficient quality to be used (i.e., quantitatively acceptable freshwater acute
toxicity studies in appendix A) were appropriately utilized as these data were used to produce the
sensitivity distributions. However, the qualitative freshwater acute toxicity studies in appendix C should
be included in a discussion about the derived benchmarks.

Thank you for your comment. Text was added to
provide a general discussion of the data in
Appendix C, and the relationship of the patterns
observed with these data relative to the
quantitatively-accepted data.

I think the authors did the best they could with the available data on hand. It is a shame that more
resources are not being deployed to generate more empirical data.

Thank you for your comment.

Much of the data accepted would not meet current standards for data acceptability or criteria derivation.
The authors have tried to maintain a degree of fidelity to the Stephan et al (1985) AWQC methodology;
however, several concerns exist with the data considered:

• The benchmark document states: "Toxicity studies accessed from the ECOTOX database were
further evaluated by Office of Water. Studies were evaluated for data quality as described by
EPA OW's data quality standard operating procedure (SOP), and consistent with OW's data
quality review approach U.S. EPA (1985), and EPA's Office of Chemical Safety and Pollution
Prevention (OPP)'s Ecological Effects Test Guidelines (U.S. EPA 2016c)." These documents
were not included in the reference list and were not provided.

Thank you for your comment. The EPA disagrees
with the peer reviewer's comment that the
accepted data "would not meet current standards
for data acceptability." The toxicity data were
thoroughly reviewed, first through the ECOTOX
data evaluation process and subsequently by OW's
extensive data quality review analyses, which
have been consistently applied in all criteria. The
data quality review for the PFAS benchmark
studies is discussed on pages 11-14 of the PFAS
acute benchmark report, and tables identifying
which tests were used quantitatively, qualitatively
or not used are detailed in Appendices A through
D of the report. Appendices C and D describe
deficiencies in the studies for the qualitative and
unused studies, respectively, regarding their lack
of quantitative use.

Please also see specific responses to the next five
comments about data acceptability. The noted
references have been added to the document.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

• In general, most of the accepted empirical studies are reported based on nominal exposure

concentrations rather than analytically measured concentrations; this is not consistent with state-
of-the-science standards of acceptability for empirical toxicology data. In fact, one of the studies
that measured exposure concentrations (Ding et al 2012) ultimately reported test endpoint data
(EC50) based on nominal concentrations rather than measured values. The reported analytical
data indicates that test concentrations differed from nominals by 10-20%, so the value reported
based on nominals is likely to be 10-20% off.

Toxicity tests used in EPA aquatic life criteria
documents are typically based on measured
chemical concentrations if such data are available.
However, for PFOA and PFOS, a substantial
proportion of the available data were reported by
the study authors only as nominal test
concentrations. The EPA completed a thorough
analysis of the data to determine whether nominal
(unmeasured) and measured concentrations are in
agreement with each other, based on an analysis of
available studies reporting both nominal and
measured concentrations in the EPA's 2024 Final
Aquatic Life Criteria documents. (U.S. EPA
2024a,b). PFOA and PFOS toxicity having pairs of
nominal and corresponding measured
concentrations were compared through: (1) linear
correlation analysis and; (2) an assessment of
measured concentrations as a percent of its paired
nominal concentration. The EPA's analyses
demonstrated that the linear correlation between
measured and corresponding nominal
concentrations show a high degree of precision
between paired concentrations across all test
conditions, in most cases within the test
acceptability threshold identified by the EPA's
OCSPP's Ecological Effects Test Guidelines.

These analyses were published in a peer-reviewed
journal in 2023 by Jarvis et al. As broadly
accepted by the international community,

OCSPP's Guidelines recommend measured
concentrations be within +/- 20% of nominal

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

concentrations to be used. Although a parallel
analysis could not be conducted based upon the
more limited datasets for these PFAS benchmarks,
the PFAS concentrations in test waters are
expected to remain relatively constant over the
course of acute exposures given its ability to resist
breakdown and transformation (Ahrens et al.
2011). Accordingly, PFAS toxicity tests were not
excluded from quantitative use in criteria
derivation on the basis of unmeasured test
concentrations alone based on results of the
analysis with PFOA and PFOS.

Finally, Ding et al.(2012) appears to have
calculated point estimates using nominal data over
measured data because only a portion of the test
replicates/treatments were actually measured. For
example, the PFAS Benchmarks document states,
"Although partial chemical analysis was
conducted, only nominal concentrations were
reported. To assess the test concentration
variability, all samples at a selected nominal
concentration were analyzed, with a relative
standard deviation of 6.5%. The measured
concentrations were between 82 and 91% of the
nominal concentration

Ahrens, L. 2011. Polyfluoroalkyl compounds in
the aquatic environment: a review of their
occurrence and fate. J. Environ. Monit. 13(1): 20-
31.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

Ding, G.H., T. Fromel, E.J. Van den Brandhof, R.
Baerselman and W.J.G.M. Peijnenburg. 2012.
Acute toxicity of poly- and perfluorinated
compounds to two cladocerans, Daphnia magna
and Chydorus sphaericus. Environ. Toxicol.
Chem.31(3): 605-610.

Jarvis, A.L., J.R. Justice, B. Schnitker and K.
Gallagher. 2023. Meta-analysis comparing
nominal and measured concentrations of
perfluorooctanoic acid and perfluorooctane
sulfonate in aquatic toxicity studies across various
experimental conditions. Environ. Toxicol. Chem.
42(11): 2289-2301.

USEPA (U.S. Environmental Protection Agency).
2024a. Aquatic life ambient water quality criterion
for PFOA-freshwater. EPA-842-R-24-002. Office
of Water. Washington, D.C.

USEPA (U.S. Environmental Protection Agency).
2024b. Aquatic life ambient water quality criterion
for PFOS-freshwater. EPA-842-R-24-003. Office
of Water. Washington, DC.

• Some of the toxicity data used in the derivation of the aquatic benchmarks comes from studies
that used non-native species (i.e., zebrafish, Danio rerio), which adds uncertainty associated with
the representativeness of such species to native North American aquatic fauna. Stephan et al
(1985) states: II. G. "Questionable data, data on formulated mixtures and emulsifiable
concentrates, and data obtained with non-resident species in North America or previously

While it remains uncertain if there are established
resident zebrafish {Danio rerio) populations in the
conterminous United States (USFWS 2018),
zebrafish are common ecotoxicity test organisms
that serve as taxonomic surrogates for untested

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

exposed organisms may be used to provide auxiliary information but should not be used in the
derivation of criteria." Appendix I Resident North American Species of Aquatic Animals Used
in Toxicity and Bioconcentration Tests defines zebrafish (Danio rerio) as "Non-resident" species
and therefore should not be included for criteria derivation. Use of Non-resident species is
briefly discussed in the report and a reference to US EPA 2018b is cited; however, this reference
is not included in the reference list. The zebrafish is in the family Cyprinidae, which all North
American native minnows (including the fathead minnow), shiners, and dace belong. Although
not native to North America, EPA seems to have decided that in the absence of suitable data on
native cyprinids, the zebrafish is an acceptable representative. However, given that zebrafish are
frequently among the more sensitive species and at least some studies with PFAS compounds
have suggested that fathead minnows may be more sensitive5, it would be good to have some
comparative additional data with NA species.

fish species and are also considered in effects
assessments conducted under the Toxic Substances
Control Act (TSCA) and the Federal Insecticide,
Fungicide, and Rodenticide Act (FIFRA).
Moreover, zebrafish data were used in deriving the
final aquatic life criteria for PFOA and PFOS
(USEPA 2024a, b), and for the derivation of other
aquatic life criteria (e.g., aluminum). USEPA
(2024a, b) specifically showed that inclusion of
non-North American species had a negligible to
slight impact on the final PFOA and PFOS aquatic
life criteria. The noted reference (i.e., USEPA
2018) was added to the document.

U.S. EPA (United States Environmental Protection
Agency). 2018. Final aquatic life ambient water
quality criteria for aluminum - 2018 (EPA-822-R-
18-001). Office of Water. Washington, D.C.

USEPA (U.S. Environmental Protection Agency).
2024a. Aquatic life ambient water quality criterion
for PFOA-freshwater. EPA-842-R-24-002. Office
of Water. Washington, D.C.

USEPA (U.S. Environmental Protection Agency).
2024b. Aquatic life ambient water quality criterion

5 Suski et al. 2023. Ecotoxicity and Accumulation of Perfluorononanoic Acid in the Fathead Minnow (Pimephalespromelas) and an Approach to Developing Protective Thresholds in the Aquatic Environment Through Species Sensitivity
Distribution. Environ Toxicol Chem. https://doi.org/10.1002/etc.5692

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

for PFOS-freshwater. EPA-842-R-24-003. Office
of Water. Washington, DC.

U.S. FWS. (U.S. Fish and Wildlife Service). 2018.
Zebra Danio (Danio rerio) Ecological Risk
Screening Summary.

• Some of the test methods used are not consistent with the 1985 guidance. EPA 1985 states that
"Acute EC50s that are based on effects that are not severe, such as reduction in shell deposition
and reduction in growth, are not used in calculating the Final Acute Value." The zebrafish tests
included in the benchmark document (Annunziato et al 2020) followed the OECD 236 method
and reported results based on a growth rather than survival endpoint. These data would not be
acceptable for derivation of an FAV based on the 1985 guideline.

The -96 hr. growth-based results were not used.
Despite the authors measuring growth, this test
provided information to identify an acute value as
a "< LC50" from a relevant acute exposure
duration. For example, the PFAS Benchmarks
document states, "Test concentrations of PFHxS
were too low to derive an LC50 value. Larval
growth (length) was assessed at test termination.
The author-reported a growth (weight) NOEC of
22.5 mg /.. and a 93-hr LCso >45 mg /.. This latter
value was acceptable for quantitative use in
deriving the recommended acute f'eshwater
PFHxS benchmark."

• In addition to the above concerns, it was noted that at least two studies (Ding et. al. 2012,
Annunziato et al 2020) that reported tests with PFBA, PFBS, PFNA, PFDA, and PFHxS
conducted their studies using dimethylsulfoxide (DMSO) as a carrier solvent. In EPA's current
test guidelines, it is recommended that if a carrier solvent must be used, "Preferred solvents are
dimethylformamide, triethylene glycol, methanol, acetone, or ethanol. Solvent use should be
avoided if possible. " DMSO is known to transport nonionized molecules thorough many
biological membranes (Jacob and Herschler 19856). Although the authors of the lab tests
conducted a "solvent control" this does not control for possible synergistic interactions of DMSO
acting as a membrane carrier, thus potentially increasing observed toxicity. Because these tests
represent a large portion of the quantitatively acceptable freshwater toxicity tests (20%, 7 of 36),

OECD (2019) indicated that when a solvent must
be used, the final concentration of the solvent used
should be minimized and not exceed 100 mg/L or
0.1 mL/L and should be the same in all test
vessels, excluding the dilution water control.
OCSPP 850.1000 and OCSPP 850.1075 similarly
report 0.1 mL/L as a level that should not be
exceeded when a solvent is used. These Test
Guidelines do not recommend excluding tests with
DMSO. Studies conducted by Ding et. al. (2012)

"Jacob, S. IV., & Herschler, R. (1986). Pharmacology of DMSO. Ciyobiology, 23(1), 14-27.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

EPA should consider the potential for inclusion of these data resulting in lower than desired
criteria, due to an overestimation of toxicity due to DMSO synergy.

and Annunziato et. al. (2020) are consistent with
these guidelines. These data were accordingly
considered acceptable for quantitative use.

OECD. 2019. Guidance Document on Aqueous-
phase Aquatic Toxicity Testing of Difficult Test
Chemicals. Report

ENV/JM/MONO(2000)6/REV 1. Organisation for
Economic Co-operation and Development. Series
on Testing and Assessment. No 23 (Second
Edition).

EPA should reassess the quality and acceptability of the available data for regulatory purposes.

The EPA disagrees with the peer reviewer's
comment and wishes to clarify the purpose of the
benchmarks. Importantly, these benchmarks are
not regulatory values; rather, they provide
information to states, Tribes and other
stakeholders.

The toxicity data were thoroughly reviewed, first
through the ECOTOX data evaluation process, and
subsequently by OW's extensive data quality
review analyses, which have been consistently
applied in all criteria. The data quality review for
the PFAS benchmark studies is discussed on pages
11-14 of the PFAS acute benchmark report, and
tables identifying which tests were used
quantitatively, qualitatively or not used are
detailed in Appendices A through D of the report.
Appendices C and D describe deficiencies in the
studies for the qualitative and unused studies,

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.b. Were the data selected and/or excludedfrom the benchmark values derivation appropriately utilized?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

respectively, regarding their lack of quantitative
use.

Notably, other peer reviewers agreed the EPA's
data review and selection process was clear,
logical and well-established.

Yes, the process of including and excluding empirical data for derivation of the benchmark values was
clearly explained, logical, and well established.

Thank you for your comment.

Quantitatively acceptable empirical acute toxicity data available for each of the eight PFAS was
tabulated for each individual study. All toxicity values, including LC values, EC values, NOECs,
LOECs, and species- and genus-mean values, were presented with four significant figures. This practice
avoided round-off errors in subsequent calculations. Studies that were determined to be qualitatively
acceptable as supporting information, but not acceptable for quantitative use were listed with
deficiencies in each study. Furthermore, studies that were deemed unsuitable for either quantitative or
qualitative were also cited. I endorse the choice to incorporate toxicity data for studies solely based on
unmeasured test concentrations. This decision is rooted in findings for PFOA and PFOS (U.S. EPA
2022a, b), leading the EPA to determine that nominal test concentrations effectively represent real PFAS
exposures in standard acute laboratory-based toxicity tests. In addition, Hoke et al.,2012
(https://doi.ors/10.1016/i.chemosphere.2011.12.066) also reported mean measured test concentrations
were similar (within 80-120% of nominal) to the targeted nominal test concentrations for fluorinated
acids with the exception of the 5:3 acid.

Thank you for your comment.

The authors applied the criteria recommended by Willming et al., 2016 to enhance models' reliability
and robustness (Box 1). Models adhering to these acceptance parameters were employed to generate
species toxicity data, which were then combined with empirical toxicity data to strengthen the derivation
of benchmark values. This approach demonstrated logical and consistent application of standard criteria
across all eight PFAS compounds.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.c. Are there relevant data that you are aware of that should be included? If so, please provide for derivation of benchmark values.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

I am not aware of data not already included in the study.

Thank you for your comment.

There is some mayfly data from the Soucek lab that does not seem to be acknowledged here. The
general lack of insect data is a systemic problem - particularly when a single midge is used to represent
the toxicity of an entire class or organisms that is likely close to 10,000 species in N. America.

Thank you for your comment. The EPA concurs
with the importance of additional insect data and
uses acceptable empirical toxicity test data
whenever they are available in the scientific
literature. As discussed in earlier responses,
following the peer review the EPA developed
application factors to account for two species (the
mayfly Neocloeon triangulifer and the cladoceran
species Moina micrura) that were indicated by
recent direct empirical test data to have markedly
greater sensitivity to PFOA and PFOS. These
application factors were used to lower the PFAS
values to account for these particularly sensitive
species, for which PFAS toxicity data are not
available, thus addressing the peer reviewer's
concern there is a lack of insect data, including the
sensitive mayfly N. triangulifer.

Several recent publications have critically reviewed the available data for PFAS compounds, e.g.,
Pandelides et al. 20237. The references below are just an example of amphibian references included in
one of the review articles, these include acute and chronic endpoints:

Abercrombie, S. A., de Perre, C., Choi, Y. J., Tornabene, B. J., Sepulveda, M. S., Lee, L.
S., & Hoverman, J. T. (2019). Larval amphibians rapidly bioaccumulate poly- and
perfluoroalkyl sub stances. Ecotoxicology and Environmental Safety, 178, 137—
145. https://doi.Org/10.1016/j.ecoenv.2019.04.022;

Thank you for your comment. The cited studies
and those within the cited review paper focus on
the evaluation of bioaccumulation, subacute,
subchronic, chronic responses, non-apical
endpoints, and/or mixtures and the reported data
are not directly applicable to the acute endpoints
evaluated in this document.

Pandelides Z, J Conder, Y Choi, EAllmon, T Hoskins, L Lee, J Hovemian, MSepiilveda. 2023. A Critical Review of Amphibian Per- and Polvfluoroalkyl Substance Ecotoxicity Research Studies: Identification of Screening Levels in
Water and Other Useful Resources for Site-Specific Ecological Risk Assessments. Environ Toxicol Chem. https://doi.org/10.1002/etc.5695

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.c. Are there relevant data that you are aware of that should be included? If so, please provide for derivation of benchmark values.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE



Ankley, G. T., Kuehl, D. W., Kahl, M. D., Jensen, K. M., Butterworth, B. C., & Nichols,
J. W. (2004). Partial life-cycle toxicity and bioconcentration modeling of
perfluorooctane sulfonate in the northern leopard frog (Ranapipiens).
Environmental Toxicology and Chemistry, 23, 2745. https://doi.org/10.1897/03-
667

Brown, S. R., Flynn, R. W., & Hoverman, J. T. (2021). Perfluoroalkyl substances
increase susceptibility of northern leopard frog tadpoles to trematode infection.
Environmental Toxicology and Chemistry, 40, 689-694.
https://doi.org/10.1002/etc.4678
Flynn, R. W., Chislock, M. F., Gannon, M. E., Bauer, S. J., Tornabene, B. J., Hoverman,
J. T., & Sepulveda, M. S. (2019). Acute and chronic effects of perfluoroalkyl
substance mixtures on larval American bullfrogs {Rana catesbeiana). Chemosphere,
236, 124350. https://doi.Org/10.1016/j.chemosphere.2019.124350
Flynn, R. W., Hoover, G., Iacchetta, M., Guffey, S., de Perre, C., Huerta, B., Li, W.,
Hoverman, J. T., Lee, L., & Sepulveda, M. S. (2022). Comparative toxicity of
aquatic per- and polyfluoroalkyl substance exposure in three species of amphibians.
Environmental Toxicology and Chemistry, 41, 1407-1415.
https://doi.org/10.1002/etc.5319
Flynn, R. W., Iacchetta, M., Perre, C., Lee, L., Sepulveda, M. S., & Hoverman, J. T.
(2021). Chronic per-/polyfluoroalkyl substance exposure under environmentally
relevant conditions delays development in northern leopard frog {Ranapipiens)
1 arvae. Environmental Toxicology and Chemistry, 40, 711-716.
https://doi.org/10.1002/etc.4690
Foguth, R. M., Hoskins, T. D., Clark, G. C., Nelson, M., Flynn, R. W., de Perre, C.,

Hoverman, J. T., Lee, L. S., Sepulveda, M. S., & Cannon, J. R. (2020). Single and
mixture per- and polyfluoroalkyl substances accumulate in developing northern
leopard frog brains and produce complex neurotransmission alterations.
Neurotoxicology and Teratology, 81, 106907.
https://doi.Org/10.1016/j.ntt.2020.106907
Fort, D. J., Mathis, M. B., Guiney, P. D., & Weeks, J. A. (2019). Evaluation of the
developmental toxicity of perfluorooctane sulfonate in the Anuran, Siliirana



49

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

3.c. Are there relevant data that you are aware of that should be included? If so, please provide for derivation of benchmark values.

REVIEWER

REVIEWER COMMENT

EPA RESPONSE



tropicalis. Journal of Applied Toxicology, 39, 365-374.
https://doi.org/10.1002/jat.3727
Hoover, G. M., Chislock, M. F., Tornabene, B. J., Guffey, S. C., Choi, Y. J., De Perre, C.,
Hoverman, J. T., Lee, L. S., & Sepulveda, M. S. (2017). Uptake and depuration of
four per/polyfluoroalkyl substances (PFAS) in northern leopard frog Ranapipiens
tadpoles. Environmental Science and Technology Letters, 4, 399-403.
https://doi.org/10.1021/acs.estlett.7b00339
Hoskins, T. D., Allmon, E. B., Flynn, R. W., Lee, L. S., Choi, Y., Hoverman, J. T., &
Sepulveda, M. S.(2022). An environmentally relevant mixture of perfluorooctane
sulfonic acid and perfluorohexane sulfonic acid does not conform to additivity in
northern leopard frogs exposed through metamorphosis. Environmental Toxicology
and Chemistry, 41, 3007-3016. https://doi.org/10.1002/etc.5486
Lech, M. E., Choi, Y. J., Lee, L. S., Sepulveda, M. S., & Hoverman, J. T. (2022). Effects
of per- and polyfluoroalkyl substance mixtures on the susceptibility of larval
American bullfrogs to parasites. Environmental Science cf- Technology, 56, 15953—
15959. https://doi.org/10.1021/acs.est.2c04574



4

I am not aware of relevant data that should be included in this process. An extensive and complete
review of available data has been conducted in preparation for this process of deriving benchmarks.

Thank you for your comment.

5

Below, a recent chronic study by Kadlec et al., 20203 has been listed as an additional reference, some
aspects may be relevant.

httDs://doi.org/10.1002/etc.5784 Sarah M. Kadlec, Will J. Backe, Russell J. Erickson, J. Russell
Hockett, Sarah E. Howe, Ian D. Mundy, Edward Piasecki, Henry Sluka, Lauren K. Votava, David R.
Mount (2023) Sublethal Toxicity of 17 Per- and Polyfluoroalkyl Substances with Diverse Structures to

Ceriodaphnia dabia, Hyalella azteca, and Chironomus dilutus

Thank you for your comment. The cited paper
contains subchronic studies which exceed the test
duration considered applicable for inclusion within
the acute dataset.

50

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 4: Are the derived benchmark values appropriately protective of sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

In my opinion, the derived benchmark values are protective for acute toxic effects on aquatic life as
defined in the report (i.e., to be protective of 95% of freshwater genera potentially exposed to these
specific PFAS under short-term conditions of one-hour duration, if the one-hour average magnitude is
not exceeded more than once in three years).

Thank you for your comment.

However, it is important to note that the most problematic properties of PFAS are not the acute
toxicological effects but rather the combination of their persistency, mobility, potential for
bioaccumulation (for some PFAS), and long-term toxicological effects. Emissions of PFAS resulting in
recipient concentrations comparable to the threshold values in this report may (depending on the volume
of water in the recipient) be substantial sources of PFAS pollution to the environment. As many PFAS
(PFCA and PFSA) are both persistent and relatively mobile, such emissions may result in problematic
pollution of drinking water and wildlife with adverse long-term consequences for both human health and
the environment.

Thank you for your comment. The EPA agrees
with the reviewer that the chronic effects of PFAS
are a concern. However, characterization of what
is a protective acute concentration for chemicals of
concern still remains important, particularly for
discharges. Text was added to the document to
further clarify the values provided are for acute
effects and do not account for potential chronic
effects and/or bioaccumulation. The EPA is
working towards approaches for the development
of chronic values for data-limited chemicals, but
the NAMs tools to develop chronic benchmarks
were not available to apply at this time.
Additionally, there is a general lack of chronic
data, especially for these PFAS, upon which to
base chronic analyses. Additional research on the
chronic toxicity of a range of PFAS would support
development of additional values.

As applied to a one-hour maximum concentration not to be exceeded every 3 years, the values are likely
protective to most species.

Thank you for your comment.

The benchmark methods attempt to maintain compliance with the EPA's 1985 method for derivation of
AWQC and to the extent that the 1985 method was "appropriately protective of sensitive aquatic life "
the benchmark approach should be as well. However, one major difference between the 1985 guidance
and the new benchmark approach is that the requirements for high-quality empirical data for a minimal
range of aquatic species have been reduced or eliminated. The benchmark approach seems to rely on
existing data or extrapolation of limited data from similar compounds. ICE methods are extremely useful

The intent of deriving benchmark values is not to
undermine the importance of using empirical test
data for the derivation of protective aquatic life
values (benchmarks or criteria). The EPA is
incorporating the use of direct test data to the
greatest extent possible when they are available.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 4: Are the derived benchmark values appropriately protective of sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

and important in estimating values for species where we cannot generate empirical data, e.g., T&E
species (Willming et al 2016). However, the benchmark approach proposed seems to minimize the
utility and need for empirical data. Modelling techniques that are based on robust empirical data are
extremely useful for supplementing extant data for species-of-concern that cannot be easily or cost-
effectively tested, or tested due to regulatory restrictions, but they should not suppliant the need for
chemical-specific empirical data. Is there a minimum amount of empirical data that are needed to derive
a benchmark? In theory, a single acute toxicity test may be sufficient, using ICE-models, to derive
regulatory benchmarks. To gain confidence in the proposed approach, EPA should conduct testing to
further confirm the accuracy of the ICE estimates, especially for the most sensitive species in the
benchmark data sets.

Instead, the objective is to employ methodologies
that would allow for the development of protective
aquatic life values for chemicals that might
otherwise be precluded from the derivation of
protective values based on traditional derivation
approaches and "minimum data requirements",
which would exclude the EPA's ability to derive
protective values. Evaluation and the appropriate
incorporation of evolving science and NAMs is
consistent with the good science clause in the
EPA's Aquatic Life Criteria Guidelines.

There are no specific direct test requirements that
have been established for deriving benchmarks.
However, as previously discussed, all methods
used incorporate underlying empirical test data,
albeit with extrapolation to derive benchmark
values. Consistent with the suggestion by the
reviewer, the EPA has been exploring the
evaluation of these approaches using chemicals
with more data-rich direct toxicity datasets.

I do have concerns about the lack of acceptable empirical acute toxicity data for freshwater primary
producers and freshwater invertebrates. I think the process of deriving benchmarks for the eight PFAS
described by the USEPA is appropriate for the empirical data that is available. I think they have done
their best with the data that is available to them.

Thank you for your comment.

The establishment of aquatic life benchmarks for the eight PFAS compounds involved the utilization of
empirical and Web-ICE data, incorporating statistical methods for calculation. This strategy is in
accordance with the EPA's objective of minimizing dependence on animal testing by integrating NAMS
into toxicity assessments.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Charge Question 4: Are the derived benchmark values appropriately protective of sensitive aquatic life?

REVIEWER

REVIEWER COMMENT

EPA RESPONSE

5

In stating that the benchmarks derived are regarded as less certain than ambient water quality criteria,
the authors are acknowledging a degree of uncertainty. This recognition is essential for offering a logical
evaluation of the benchmarks' reliability.

Thank you for your comment.

53

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

Front page

Title

As "PFAS" is an abbreviation for "per- and polyfluorinated
substances", the term "PFAS Compounds" would mean "per-
and polyfluorinated substances Compounds". Here and
throughout the document, I suggest to just use "PFAS" instead
of "PFAS Compounds", "PFAS substances", or "PFAS
chemicals"

The edit was made as suggested.

viii

Acronym
list

Several of the acronyms used in the report are missing.
Examples of missing acronyms are listed in the following,
however, there are likely more missing acronyms. This should
be reviewed and corrected before publication. Examples of
missing acronyms: SMAV, all the eight PFAS in focus (PFBA,
PFBS, PFHxA, PFHxS, etc.), EPA, DOD, MSE, SMAV

The missing acronyms were added as suggested.

Please define what is meant by the term Water Quality Criteria
in this context

The text is referring to 304(a)(1) Water Quality Criteria. This text first
identifies these criteria in order to contrast them with 304(a)(2) benchmark
values.

Please define the difference between "draft ambient water
quality benchmarks" and "Water Quality Criteria"

Benchmarks refer to the values derived in this document, while "Water
Quality Criteria" refers to 304(a)(1) Water Quality Criteria.

It is somewhat confusing for the reader to understand what was
done in the present study and what has been done previously. I
suggest making this clearer by using terms such as "in the
present study" or similar.

The edit was made as suggested.

Here, and throughout the document, it would be more logical
to present the eight PFAS sorted according to group and
number of perfluorinated carbon atoms in the structures. I.e.,
PFBA, PFHxA, PFNA, PFDA, PFBS, PFHxS, 7:3 FTC A, 8:2
FTUCA.

The edit was made as suggested.

It should be made clearer in the summary that the benchmarks
reported here are for acute exposure. I.e.: "...protective of 95%
of freshwater genera potentially exposed to the specific PFAS

Text was added clarifying parameters for frequency and duration.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

under short-term conditions of one-hour duration, if the one-
hour average magnitude is not exceeded more than once in
three years"

The sentence that starts with "EPA's "Guidelines for Deriving
NumericaF needs to be revised as it should likely be at least
two sentences.

The edit was made as suggested.

xii

Table Ex-
1-1

Is the superscript "1" referring to a footnote?

The superscript was corrected as suggested.

xii

Table Ex-
1-1

It would be more logical to present the eight PFAS sorted
according to group and number of perfluorinated carbon atoms
in the structures. I.e., PFBA, PFHxA, PFNA, PFDA, PFBS,
PFHxS, 7:3 FTCA, 8:2 FTUCA.

The edit was made as suggested.

xiii

Here, and throughout the document, I suggest using the terms
perfluoroalkyl carboxylic acids (PFCA) and perfluoroalkyl
sulfonic acids (PFSA) as this is the common terminology.

The edit was made as suggested.

xiii

A conclusion for the binning approach should be included in
the summary.

A conclusion about the binning approach was added to the summary.

14 (or 1?)

Page
numbers

The first page in the section "Background" should probably be
1 (it has page number 14 in the version I received)

The edit was made as suggested.

The acute water quality benchmark concentrations for PFOS
and PFOA should be stated here

The PFOS and PFOA final values were added to the summary discussion
for the PFAS values.

It would be good to include information on how the selection
of these eight PFAS was done. Why these exact substances?
Were other PFAS considered, but not included due to limited
information available?

The text (now on page 1) states "The above eight chemicals were selected
to represent a range of PFAS that are present in aquatic ecosystems and of
concern to stakeholders (e.g., states, Tribes, DOD). Another important

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

consideration for selection was the availability of both acute empirical and
ICE model toxicity data for these chemicals."

The reference U.S. EPA 1985 is not included in the reference
list.

Missing references were added to the document.

It would make it clearer for the reader to state "toxicity data"
instead of just "data"

The edit was made as suggested.

It would be good to include a reference to Table 4-2 to show
which MDRs are fulfilled for which PFAS.

The edit was made as suggested.

Would it be better to use the term "aquatic life benchmarks"
instead of "aquatic life values" ?

The edit was made as suggested.

Raimondo et al. 2023 is not in the References list. Is it the
same as "Raimondo et al., in review"?

Correct. This paper has now been published and the full reference
(Raimondo et al. 2024) has been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty analysis
and updated user guidance for Interspecies Correlation Estimation (ICE)
models and low toxicity compounds. Integr Environ Assess Manag.
Accepted Author Manuscript. httDs://doi.or«/l 0,1002/iearn,4884

Wilming et al 2016 does not mention PFOS and PFOA

The text was edited to remove reference to PFOA and PFOS.

Wilming et al 2016 is not included in the References list

Missing references were added to the document.

Wilming et al 2016 defines parameters for listed species. The
use of these as general parameters is likely unproblematic, but
it should be stated that these parameters are used in a slightly
different context here compared to in Wilming et al

The designation as a listed species does not materially affect the analysis
conducted by Willming et al. (2016), and reference to this difference may
introduce uncertainty regarding why it is being noted.

Bejarano and Wheeler, 2020 is not in the References list

Missing references were added to the document.

Figure 2-1

The figure is not referred to in the text. A reference to the
figure should be included where appropriate.

The edit was made as suggested.

Figure 2-1

The last sentence in the figure text is not written in bold and
appears to be incomplete. Should it be "...develop a log-linear
model"?

The edit was made as noted.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

Table 3-1

It would be good to include references in this table,
alternatively to refer to appendix A in the table text.

The edit was made as suggested.

Table 3-1

The heading "Toxicity" is a bit ambiguous. It would be more
intuitive to write e.g., EC50/LC50

The table heading was edited to indicate "Acute Toxicity".

The sentence "The Office of Water completed a Data
Evaluation Record (DER) for each species by chemical
combination from the studies identified by ECOTOX for the
eight PFAS compounds undergoing evaluation." is not clear to
me. What does "chemical combination" refer to in this
context?

The text was edited to clarify that the EPA was referring to chemical
mixtures (as opposed to tests with single compounds).

The sentence "Further, only single chemical toxicity tests with
PFOA were considered for possible inclusion in benchmark
derivation" is confusing. Is this statement correct?

The reference to PFOA was deleted. The text is otherwise correct as
stated.

The title for Appendix A used here ("Appendix A: Acceptable
Freshwater Acute PFOA Toxicity Studies") is not correct.

The title was corrected.

Were all toxicity values (e.g., LC, EC, NOEC, LOEC) treated
as the same? Please explain

The accepted empirical studies were all based on LC50 or EC50 values.
These studies are summarized in Appendix B. Data based on NOECs and
LOECs were not determined to be acceptable for quantitative use (see
Appendix C) and were not used for calculating the benchmark value.

The last sentence is missing a word ("is"?) (i.e., "of the
corresponding benchmarks is stated for each study at the
end"?)

The edit was made as suggested.

The EPA 1985 approach should be summarized here. At least
the calculation procedure (as detailed in EPA 1985) should be
stated. It should be clear to the reader why the four most
sensitive values are focused on.

Text was added to refer the reader to EPA's Aquatic Life Criteria
Guidelines for a more detailed description, in order to focus the reader on
the primary analyses being conducted within this document.

In the first sentence, I suggest reminding the reader that eight
MDR groups are required fulfilled.

The edit was made as suggested.

Please explain why the FAV was divided by two (what is the
reasoning behind this approach?)

This is a standard procedure for the derivation of acute criteria, as
described in EPA's Aquatic Life Criteria Guidelines, the purpose of which

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

is to estimate from an LC50/EC50 a concentration that would not impact
aquatic life. Additional text was added to the document to clarify this
point.

The sentence "A Web ICE model was not available to fulfill
the MDR for a third family in the phylum chordata." Is
somewhat confusing as it is according to the table already
fulfilled using empirical data.

The noted sentence was determined to not be necessary and was removed
from the document.

Here and for the other PFAS, explain why it is stated that
"GMAVs for the four most sensitive genera were within a
factor of 2.0 of each other" (i.e., it is according to the criteria
in the guidelines)

This statement provides a general indication of the similarity or
"agreement" between the four values used for calculating the final value,
and such a statement summarizing the similarity of these values is
typically provided as a general statement in criteria documents.

A brief discussion on the results would be appropriate. For
example:

- The long chained PFAS are more acute toxic than shorter
(PFBA, PFBS). This is in agreement with scientific literature.
Previously published values for PFOS and PFOA indicate that
these are the most toxic of the PFSA and PFCA, respectively.

- According to the results published here, FTUCA and FTCA
are the most toxic. Is this as expected?

A discussion of chain length and toxicity value magnitude was added to
the summary section as suggested.

The sentence "The resulting acute benchmarks, although
consistently hi "her, were also small, with each of the
benchmarks falling within a factor of <2.1 of one another,
indicating close agreement between values calculated using
either approach." Should be revised. What were small, the
difference between values calculated using the different
approaches?

The text (now on page 87) was edited to state "The differences between
the resulting acute benchmarks, although consistently higher using the
scaled approach, were also small,..."

A conclusion for the use of the binning method should be
included.

Further discussion about use of the binning approach was added to the
document.

References

The reference list needs to be updated as several references
used in the text are not included here.

Missing references were added to the document.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

Appendix
A

The references are not included in any Reference list. This
should be corrected.

Missing references were added to the document.

Appendix
B

Please explain why the four most sensitive values are
summarized

Text was added to clarify how the four values are used in the value
calculation.

C-2

Green alga

The text explaining the Deficiency for this study "Initially
identified as Quantitative" is confusing.

The text was edited to clarify the final decision for use of the value.

C-3

Green alga

Why is the test duration written in red?

The text was edited to correct to black lettering.

E-l

Only the Web-ICE version 3.3 is available online. I assume
that the small differences I found between online model
parameters and parameters in the Table is due to the updated
values in the v4 model?

That is correct. There are very small differences between the version 3.3
and 4.0 model outcomes based on changes (primarily increases) in the
underlying data.

F-l

The claim "In these situations, a user can either enter the
measured toxicity value (LC50/EC50) into the ICE model as
[j,g/L and allow the regression to extrapolate beyond the range
of the model or enter a "scaled" toxicity value (i.e. enter the
measured LC50 value as mg/L)." should be supported by a
literature reference.

The supporting paper has now been published and the full reference
(Raimondo et al. 2024) has been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty analysis
and updated user guidance for Interspecies Correlation Estimation (ICE)
models and low toxicity compounds. Integr Environ Assess Manag.
Accepted Author Manuscript. httDs://doi.or«/l 0,1002/iearn,4884

F-73

Using the scaled approach, the eight MDRs were fulfilled for
seven of the evaluated compounds (not six)?

The edit was made as suggested.

G-7

Footnote b

It would be more intuitive for the reader to use "Species x
SMAV", similar to what is done for PFAS

The edit was made as suggested.

G-9

The table lacks references to the footnotes

References to all footnotes were added to this table.

G-12

Figures G-l and G-2 are too small. Consider showing panel B
below panel A

The noted figures were moved to separate pages and the figure sizes were
increased.

Box

Some classes are quite species rich and a single representative
is likely not sufficient (see insects)

There was variation in the size of the database used and notable limitations
to the data available for some species groups (notably insects); however,
all acceptable data were considered for inclusion in the benchmark
calculations.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

When is Raimondo et al (in review) going to be released? It
seems like this should be available information for the reader

The supporting paper has now been published and the full reference
(Raimondo et al. 2024) has been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty analysis
and updated user guidance for Interspecies Correlation Estimation (ICE)
models and low toxicity compounds. Integr Environ Assess Manag.
Accepted Author Manuscript. httDs://doi.or«/l 0,1002/iearn,4884

throughout

Why are no extra uncertainty factors used other than FAV/2
given the uncertainties associated with the process?

Additional uncertainty factors are not used as part of EPA's Aquatic Life
Criteria Guidelines criteria derivation process, which was the basic
process used to derive the benchmark values, beyond the development of
the toxicity database. As previously discussed, following the peer review
the EPA has developed application factors to account for two genera (the
mayfly, Neocloeon and the cladoceran, Moina) that were indicated by
direct empirical test data to have markedly greater sensitivity to PFOA and
PFOS. These application factors were used to lower the PFAS values to
account for these particularly sensitive genera, for which PFAS toxicity
data are not available.

throughout

Within class extrapolation is a whole lot of biodiversity to
lump together

Data were grouped according to procedures presented in EPA's Aquatic
Life Criteria Guidelines, which was the basic process used to derive the
benchmark values, beyond development of the toxicity database.

Are benchmarks expected to carry the same weight as AWQC
if adopted as state or tribal standards?

As indicated in the document, the freshwater acute benchmarks are
informational values that states and Tribes could choose to consider in
their state water quality protection programs; however, the acute
benchmarks magnitudes are less certain than Water Quality Criteria since
they are based on both empirical and estimated toxicity data. Text has
been added to clarify the difference between benchmarks and water
quality criteria.

Should there be inclusion of EPA's GLI approach?

This document focuses on consideration of recently-developed NAMs-
based methods. Inclusion of the GLI approach is beyond the scope of this
document.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

It is difficult to evaluate this statement since the Raimondo et
al report is unpublished and not supplied.

The supporting paper has now been published and the full reference
(Raimondo et al. 2024) has been added to the reference list.

Raimondo, S., Lilavois, C. and Nelson, S.A. (2024), Uncertainty analysis
and updated user guidance for Interspecies Correlation Estimation (ICE)
models and low toxicity compounds. Integr Environ Assess Manag.
Accepted Author Manuscript. httDs://doi.or«/l 0,1002/iearn,4884

The document references USEPA 2016c for information on
how data were evaluated. The reference is not in the reference
list and was not provided. It is critical to assess the
acceptability of the empirical data accepted in Appendix A.
USEPA 1985 was not included in the reference list.

Missing references were added to the document.

The document references USEPA 2018b however, the
reference is not in the reference list and was not provided.

Missing references were added to the document.

Second

Period missing at the end of ".. .minimum data requirements
(MDRs) to calculate aquatic life criteria"

The edit was made as suggested.

xii

First

Space needed between "... (Guidelines)(U.S. EPA 1985)"

The edit was made as suggested.

xii

Table Ex-
1-1

Not clear how the "Duration" and "Frequency" were
determined for the recommended benchmarks. Is this standard
for USEPA acute freshwater aquatic life benchmarks?

The duration and frequency assumptions are consistent with those
presented in EPA's Aquatic Life Criteria Guidelines for acute values and
with those used for most criteria values.

First

A closing bracket missing in "• Perfluorohezanesulfonic acid
(PFHxS (CAS# 355464, 108427538, 3871996, 82382125)"

The edit was made as suggested.

First

The first word in f), g), and h) is not capitalized as in a) to e).

a) "insect (e.g., mayfly, dragonfly, damselfly, stonefly,
caddisfly, mosquito, midge, etc.)

b) family in a phylum other than Arthropoda or Chordata
(e.g., Rotifer a, Annelida, Mollusca, etc.)

The edit was made as suggested.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

c) family in any order of insect or any phylum not already
represented

Final
sentence
on the page

The final sentence of this section is "Benchmark values for the
eight PFAS using this alternative approach are summarized in
Section 5.10. "

I would specifically reference Table 5-26 in section 5.10 at the
end of the sentence above. It would make it easier for the
reader to find the benchmarks derived using the scaled
approach. Or reference section F.9 and/or Table F-29.

A reference to Section F.9 was added to the document.

First

The first sentence in the paragraph states, "Empirical studies
available for the eight PFAS were identified using the
ECOTOXicology KnowledgebaseI assume that the "eight
PFAS" refers to the compounds for which the benchmarks are
being set. However, later in the paragraph, there is a sentence
that states, "Further, only single chemical toxicity tests with
PFOA were considered for possible inclusion in benchmark
derivationPFOA is not one of the eight PFAS for which a
benchmark is being set, so it is not clear why acute toxicity
data for PFOA is being used. It was stated earlier in section 3
that validation was conducted using measured and predicted
values for PFOS and PFOA, but it is not clear at this point in
section 4 how acute toxicity data for PFOA will be used in
deriving benchmarks for the eight PFAS that are the focus of
this document. You may want to make that clear to the reader.

The text was edited and references to PFOA and PFOS were removed
from this section of the text.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

First

The paragraph makes references to whether this process should
consider studies that only report nominal concentrations of the
PFAS in the toxicity study. The rationale given for choosing to
consider studies that only report nominal concentrations is a
case study that was conducted with measured and nominal
concentrations for PFOS and PFOA. While the rationale is
sound for considering studies that only include nominal
concentrations of PFOS or PFOA, care should be taken to
extrapolate to the entire class of chemicals, i.e., PFAS. The
eight PFAS that are focus of this process have different
physical and chemical properties than PFOS and PFOA,
consequently, the probability of the nominal concentrations
being with 20% of the measured concentrations for the eight
PFAS may be different than PFOS and PFOA, which was 82
and 83%, respectively.

Toxicity tests used in EPA aquatic life criteria documents are typically
based on measured chemical concentrations only. For PFOA and PFOS, a
substantial proportion of the available data was reported as nominal test
concentrations. The EPA examined the issue of whether nominal
(unmeasured) and measured concentrations are in close agreement with
each other, based on an analysis of available studies reporting both
nominal and measured concentrations (U.S. EPA 2024a,b). PFOA and
PFOS toxicity having pairs of nominal and corresponding measured
concentrations were compared through: (1) linear correlation analysis and;
(2) an assessment of measured concentrations as a percent of its paired
nominal concentration. Linear correlation between measured and
corresponding nominal concentrations suggested a high degree of
precision between paired observations across all test conditions, in most
cases within the test acceptability threshold identified by the EPA's
OCSPP's Ecological Effects Test Guidelines. Similar results were also
observed by Jarvis et al. (2023). Although a parallel analysis could not be
conducted based upon the more limited datasets for the evaluated PFAS,
the PFAS concentrations in test waters are expected to remain relatively
constant over the course of acute exposures given its ability to resist
breakdown and transformation (Ahrens et al. 2011). Accordingly, PFAS
toxicity tests were not excluded from quantitative use in criteria derivation
on the basis of unmeasured test concentrations alone based on results of
the analysis with PFOA and PFOS.

Ahrens, L. 2011. Polyfluoroalkyl compounds in the aquatic environment:
a review of their occurrence and fate. J. Environ. Monit. 13(1): 20-31.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations

Reviewer

Page

Paragraph

Comment or Question

EPA Response

studies across various experimental conditions. Environ. Toxicol. Chem.
42(11): 2289-2301.

USEPA (U.S. Environmental Protection Agency). 2024a. Aquatic life
ambient water quality criterion for PFOA-freshwater. EPA-842-R-24-002.
Office of Water. Washington, D.C.

USEPA (U.S. Environmental Protection Agency). 2024b. Aquatic life
ambient water quality criterion for PFOS-freshwater. EPA-842-R-24-003.
Office of Water. Washington, DC.

First

Comma needed between "toxicity database" and "benchmark
values " in the first sentence on page 27.

The edit was made as suggested.

Second

It would be useful to report at some point the number of
acceptable empirical LC50 or EC50 used in each SSD that
were based on nominal concentrations. The data is available in
the document but the reader would have to take a great deal of
time to compile these numbers.

This information is summarized in Table 4-2 and all studies used
quantitatively are presented in Appendix A.

Second

Why not use the lowest acute value for a species instead of the
mean? Using the lowest acute value for a species would be a
more conservative approach in terms of protection of sensitive
species. The same question could be asked about the genus
mean acute values.

Mean species and genus values are used as part of EPA's Aquatic Life
Criteria Guidelines criteria derivation process, which was the basic
process used to derive the benchmark values, beyond the development of
the toxicity database.

Entire
document

Review the document to ensure that Greek letters are used
consistently, e.g., |ig/L vs. ug/L

Values were reviewed and changed to |ig/L, as appropriate.

Appendices
A to C

I thought these were very valuable appendices. They clearly
laid out the studies that were considered for inclusion in the
derivation of the benchmarks and why studies were eventually
not included.

Thank you for your comment.

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

III. Specific Observations



Reviewer

Page

Paragraph

Comment or Question

EPA Response

5

70 and 80

Figure 5-7
and Figure
5-8

Is this a bimodal response- as this is model based SSD, it is
challenging to confirm? Invertebrates and fish may have
different mode of action for 8:2 FTUCA and 7:3 FTC A

It is possible that results are showing a bimodal distribution for fish and
invertebrates, but it may be overstepping to draw any conclusions based
on the available data.

65

-------
External Letter Peer Review for Freshwater Benchmarks for Selected PFAS Using a New Approach Method

Appendix A

Example of Extrapolation vs Scaled Approach Value Calculation: I), magna with PFBA

EXTRAPOLATION APPROACH

SCALED APPROACH

HQ
3

2
_•

/

m

1

I

E

E
-

O

a

& -

O y

cr. 3.48

3

'— cn —













• ••

i*







iff?*'

•

* •
* * ••



/>'• • ~
V •



/ '

2	3.72 4

66

-------