A EPA
EPA/635/R-22/277Fb
www.epa.gov/iris
IRIS Toxicological Review of Perfluorobutanoic Acid (PFBA, CASRN 375-
22-4) and Related Salts
Supplemental Information—Appendices A through F
December 2022
Integrated Risk Information System
Center for Public Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Supplemental Information ofPFBA and Related Salts
DISCLAIMER
This document has been reviewed by the U.S. Environmental Protection Agency, Office of
Research and Development and approved for publication. Any mention of trade names, products, or
services does not imply an endorsement by the U.S. government or the U.S. Environmental
Protection Agency. EPA does not endorse any commercial products, services, or enterprises.
ii
-------
Supplemental Information ofPFBA and Related Salts
CONTENTS
APPENDIX A. SYSTEMATIC REVIEW PROTOCOL FOR THE PFAS IRIS ASSESSMENTS A-l
APPENDIX B. ADDITIONAL DETAILS OF SYSTEMATIC REVIEW METHODS AND RESULTS B-l
APPENDIX C. ADDITIONAL TOXICOKINETIC INFORMATION IN SUPPORT OF DOSE-RESPONSE
ANALYSIS C-l
C.l. USE OF HALF-LIVES OF EXCRETION FOR DOSIMETRIC ADJUSTMENTS C-l
C.2. MIXED MODELING TO ESTIMATE HALF-LIFE IN HUMANS C-7
APPENDIX D. BENCHMARK DOSE MODELING RESULTS D-l
D.l. BENCHMARK DOSE MODELING APPROACHES D-l
D.l.l. Modeling Procedure for Dichotomous Noncancer Data D-l
D.l.2. Modeling Procedure for Continuous Noncancer Data D-l
D.l.3. Modeling Procedure for Continuous Noncancer Developmental Toxicity Data D-3
D.1.4. Modeling Procedure for Dichotomous Noncancer Developmental Toxicity Data D-3
D.l.5. Data Used for Modeling D-3
D.2. RELATIVE LIVER WEIGHT—MALE RATS EXPOSED 90 DAYS (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-14
D.3. RELATIVE LIVER WEIGHT—P0 MICE (Das et al., 2008) D-23
D.4. LIVER HYPERTROPHY—MALE RAT (Butenhoff et al., 2012; van Otterdijk, 2007b) D-29
D.5. TOTAL T4—MALE RAT (Butenhoff et al., 2012; van Otterdijk, 2007b) D-32
D.6. INCREASED FETAL MORTALITY - MALE AND FEMALE Fi MICE (Das et al., 2008) D-36
D.7. DELAYED EYE OPENING—Fi MALE AND FEMALE MICE (Das et al., 2008) D-43
D.8. VAGINAL OPENING—Fi FEMALE MICE (Das et al., 2008) D-49
D.9. PREPUTIAL SEPARATION—Fi MALE MICE (Das et al., 2008) D-54
D.10. RELATIVE LIVER WEIGHT—MALE HUMANIZED PPARa MICE (Foreman et al., 2009) D-58
D.ll. RELATIVE LIVER WEIGHT—MALE RATS EXPOSED 28 DAYS (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-61
APPENDIX E. SUMMARY OF PUBLIC AND EXTERNAL PEER REVIEW COMMENTS AND EPA'S
DISPOSITION E-l
E.l. CHARGE QUESTION 1 - SYSTEMATIC REVIEW E-2
E.l.l. Overarching External Peer Reviewer Comments on Systematic Review E-2
E.l.2. Tier 1 Recommendations E-3
iii
-------
Supplemental Information ofPFBA and Related Salts
E.1.3. Tier 2 Suggestions E-4
E.1.4. Public Comments E-7
E.2. CHARGE QUESTION 2 - STUDY EVALUATION E-8
E.2.1. Overarching External Peer Reviewer Comments on Study Evaluation E-9
E.2.2. Tier 1 Recommendations E-9
E.2.3. Tier 2 Suggestions E-9
E.3. CHARGE QUESTIONS 3 AND 4 - HEPATIC EFFECTS E-ll
E.3.1. Overarching External Peer Reviewer Comments on Hepatic Effects E-ll
E.3.2. Tier 1 Recommendations E-ll
E.3.3. Tier 2 Suggestions E-ll
E.3.4. Public Comments E-15
E.4. CHARGE QUESTION 3 - THYROID EFFECTS E-16
E.4.1. Overarching External Peer Reviewer Comments on Thyroid Effects E-16
E.4.2. Tier 1 Recommendations E-17
E.4.3. Tier 2 Suggestions E-18
E.4.4. Public Comments E-20
E.5. CHARGE QUESTION 3 - DEVELOPMENTAL EFFECTS E-21
E.5.1. Overarching External Peer Reviewer Comments on Developmental Effects E-21
E.5.2. Tier 1 Recommendations E-22
E.5.3. Tier 2 Suggestions E-23
E.5.4. Public Comments E-24
E.6. CHARGE QUESTION 3 - REPRODUCTIVE AND OTHER EFFECTS E-25
E.6.1. Overarching External Peer Reviewer Comments on Reproductive and Other
Effects E-25
E.6.2. Tier 1 Recommendations E-25
E.6.3. Tier 2 Suggestions E-25
E.7. CHARGE QUESTION 5 - CANCER HAZARD E-26
E.7.1. Overarching External Peer Reviewer Comments on Cancer Hazard E-26
E.7.2. Tier 1 Recommendations E-26
E.7.3. Tier 2 Suggestions E-27
E.8. CHARGE QUESTION 6 - NONCANCER TOXICITY VALUE DATA SELECTION E-27
E.8.1. Overarching External Peer Reviewer Comments on Noncancer Toxicity Value
Data Selection E-27
E.8.2. Tier 1 Recommendations E-27
iv
-------
Supplemental Information ofPFBA and Related Salts
E.8.3. Tier 2 Suggestions E-28
E.8.4. Public Comments E-28
E.9. CHARGE QUESTION 7 - SUBCHRONIC REFERENCE DOSE E-29
E.9.1. Overarching External Peer Reviewer Comments on the Subchronic Reference
Dose E-30
E.9.2. Tier 1 Recommendations E-30
E.9.3. Tier 2 Suggestions E-32
E.10. CHARGE QUESTION 8 - NONCANCER TOXICITY VALUE DOSE-RESPONSE MODELING E-32
E.10.1. Overarching External Peer Reviewer Comments on Noncancer Toxicity Value
Dose-Response Modeling E-33
E.10.2. Tier 1 Recommendations E-33
E.10.3. Tier 2 Suggestions E-33
E.ll. CHARGE QUESTION 9 - TOXICOKINETICS E-34
E.ll.l. Overarching External Peer Reviewer Comments on Toxicokinetics E-34
E.ll.2. Tier 1 Recommendations E-34
E.ll.3. Tier 2 Suggestions E-35
E.11.4. Public Comments E-37
E.12. CHARGE QUESTION 10 - UNCERTAINTY FACTOR APPLICATION E-40
E.12.1. Tier 1 Recommendations E-40
E.12.2. Tier 2 Suggestions E-40
E.12.3. Public Comments E-42
E.13. CHARGE QUESTION 11 - CANCER TOXICITY VALUES E-44
E.13.1. Overarching External Peer Reviewer Comments on Cancer Toxicity Values E-44
E.13.2. Tier 1 Recommendations E-44
E.13.3. Tier 2 Suggestions E-44
APPENDIX F. QUALITY ASSURANCE FOR THE IRIS TOXICOLOGICAL REVIEW OF
PERFLUOROBUTANOIC ACID AND RELATED COMPOUND AMMONIUM
PERFLUOROBUTANOATE F-l
REFERENCES R-l
v
-------
Supplemental Information ofPFBA and Related Salts
TABLES
Table B-l. Perfluorobutanoic acid (PFBA) database search strategy B-l
Table B-2. Title/abstract-level screening criteria for the initial literature searches B-4
Table B-3. Example DistillerSR form questions to be used for title/abstract-level and full
text-level screening for literature search updates from 2019 B-6
Table D-l. Sources of data used in benchmark dose modeling of PFBA endpoints D-3
Table D-2. Data received from study authors for Das et al. (2008) on full litter resorptions (FLR) D-3
Table D-3. Data received from study authors for Das et al. (2008) on fetal death (litters without
full litter resorptions) combined with full litter resorptions D-4
Table D-4. Data received from study authors for Das et al. (2008)on delayed eye opening D-4
Table D-5. Data received from study authors for Das et al. (2008) on delayed vaginal opening D-7
Table D-6. Data received from study authors for Das et al. (2008) on delayed preputial
separation D-12
Table D-7. Dose-response data for relative liver weight in male rats following 90 day exposure
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-14
Table D-8. Benchmark dose results for relative liver weight in male rats exposed 90 days
—constant variance, BMR = 10% relative deviation (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-15
Table D-9. Benchmark dose results for relative liver weight in male rats exposed 90
days—nonconstant variance, BMR = 10% relative deviation (Butenhoff et al.,
2012; van Otterdijk, 2007b) D-17
Table D-10. Benchmark dose results for relative liver weight in male rats exposed 90
days—log-normal distribution, constant variance, BMR = 10% relative deviation
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-18
Table D-ll. Benchmark dose results for relative liver weight in male rats exposed 90
days—log-normal distribution, constant variance, BMR = 1 standard deviation
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-22
Table D-12. Dose-response data for relative liver weight in pregnant mice (Das et al., 2008) D-23
Table D-13. Benchmark dose results for relative liver weight in pregnant mice—constant
variance, BMR = 10% relative deviation (Das et al., 2008) D-24
Table D-14. Benchmark dose results for relative liver weight in pregnant mice—constant
variance, BMR = 1 standard deviation (Das et al., 2008) D-28
Table D-15. Dose-response data liver hypertrophy in male rats (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-29
Table D-16. Benchmark dose results for liver hypertrophy in rats—BMR = 10% extra risk
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-29
Table D-17. Dose-response data for liver hypertrophy (slight severity lesions) in male rats
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-32
Table D-18. Benchmark dose results for liver hypertrophy (slight severity lesions) in male
rats—BMR = 10% extra risk (Butenhoff et al., 2012; van Otterdijk, 2007b) D-32
Table D-19. Dose-response data for total T4 levels in male rats (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-32
Table D-20. Benchmark dose results for total T4 levels in male rats—constant variance,
BMR = 1 standard deviation (Butenhoff et al., 2012; van Otterdijk, 2007b) D-33
Table D-21. Benchmark dose results for total T4 levels in male rats—nonconstant variance,
BMR = 1 standard deviation (Butenhoff et al., 2012; van Otterdijk, 2007b) D-34
vi
-------
Supplemental Information ofPFBA and Related Salts
Table D-22. Benchmark dose results for total T4 levels in male rats—log-normal distribution,
constant variance, BMR = 1 standard deviation (Butenhoff et al., 2012; van
Otterdijk, 2007b) D-35
Table D-23. Dose-response data for increased fetal mortality (Das et al., 2008) D-36
Table D-24. Benchmark dose results for increased fetal mortality (male and female
mice)—BMR = 1% extra risk (Das et al., 2008) D-38
Table D-25. Dose-response data for delayed eye opening in male and female mice (Das et al.,
2008) D-43
Table D-26. Benchmark dose results for delayed eye opening in male and female
mice—constant variance, BMR = 5% relative deviation (Das et al., 2008) D-44
Table D-27. Benchmark dose results for delayed eye opening in male and female
mice—constant variance, BMR = 1 standard deviation (Das et al., 2008) D-48
Table D-28. Dose response data for delayed vaginal opening in female mice (Das et al., 2008) D-49
Table D-29. Benchmark dose results for delayed vaginal opening in female mice—constant
variance, 5% relative deviation (Das et al., 2008) D-49
Table D-30. Benchmark dose results for delayed vaginal opening in female mice—constant
variance, 1 standard deviation (Das et al., 2008) D-53
Table D-31. Dose-response data for delayed preputial separation in male mice (Das et al., 2008) D-54
Table D-32. Benchmark dose results for delayed preputial separation in male mice—constant
variance, BMR = 5% relative deviation (Das et al., 2008) D-54
Table D-33. Benchmark dose results for delayed preputial separation in male mice—constant
variance, BMR = 1 standard deviation (Das et al., 2008) D-58
Table D-34. Dose-response data for relative liver weight in male humanized PPARa mice
(Foreman et al., 2009) D-58
Table D-35. Benchmark dose results for relative liver weight in male humanized PPARa mice
—nonconstant variance, BMR = 10% relative deviation (Foreman et al., 2009) D-59
Table D-36. Dose-response data for relative liver weight in male rats following 28 day exposure
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-61
Table D-37. Benchmark dose results for relative liver weight in male rats exposed 28 days
—nonconstant variance, BMR = 10% relative deviation (Butenhoff et al., 2012;
van Otterdijk, 2007b) D-61
FIGURES
Figure C-l. Mouse AUC after oral doses of PFBA C-l
Figure C-2. Mouse Cmax after oral doses of PFBA C-2
Figure C-3. Rat AUC after oral doses of PFBA C-3
Figure C-4. Rat Cmax after oral doses of PFBA C-4
Figure C-5. Estimated human half-lives versus initial serum concentrations C-5
Figure D-l. Dose-response curve for the Exponential M3 model fit to relative liver weight in
male rats exposed 90 days (Butenhoff et al., 2012; van Otterdijk, 2007b) D-19
Figure D-2. Dose-response curve for the Exponential M4 model fit to relative liver weight in
pregnant mice (Das et al., 2008) D-25
Figure D-3. Dose-response curve for the Weibull model fit to liver hypertrophy in male rats
(Butenhoff et al., 2012; van Otterdijk, 2007b) D-30
-------
Supplemental Information ofPFBA and Related Salts
Figure D-4. Dose-response curve for the Nested-Logistic model fit to increased fetal mortality in
male and female mice (Das et al., 2008) D-39
Figure D-5. Dose response curve for the Hill model fit to delayed eye opening in male and female
mice (Das et al., 2008) D-45
Figure D-6. Dose response curve for the Hill model fit to delayed vaginal opening in female mice
(Das et al., 2008) D-50
Figure D-7. Dose response curve for the Exponential 3 model fit to delayed preputial separation
in male mice (Das et al., 2008) D-55
viii
-------
Supplemental Information ofPFBA and Related Salts
ABBREVIATIONS AND ACRONYMS
ACO
acyl-CoA oxidase
HAWC
Health Assessment Workspace
ADME
absorption, distribution, metabolism,
Collaborative
and excretion
HED
human equivalent dose
AFFF
aqueous film-forming foam
HERO
Health and Environmental Research
AIC
Akaike's information criterion
Online
ALP
alkaline phosphatase
HISA
highly influential scientific information
ALT
alanine aminotransferase
HPT
hypothalamic-pituitary-thyroid
AST
aspartate aminotransferase
IRIS
Integrated Risk Information System
atm
atmosphere
i.v.
intravenous
ATSDR
Agency for Toxic Substances and
IQ
intelligence quotient
Disease Registry
IQR
interquartile range
AUC
area-under-the-concentration curve
ISI
influential scientific information
BMD
benchmark dose
IUR
inhalation unit risk
BMDL
benchmark dose lower confidence limit
LLOQ
lower limit of quantitation
BMDS
Benchmark Dose Software
LN
log-normal
BMR
benchmark response
LOAEL
lowest-observed-adverse-effect level
BW
body weight
MBq
megabecquerel
Cavg
average concentration
MOA
mode of action
Cmax
maximum concentration
NCEA
National Center for Environmental
CA
Cochran-Armitage
Assessment
CAR
constitutive androstane receptor
NCV
nonconstant variance
CASRN
Chemical Abstracts Service registry
NIOSH
National Institute for Occupational
number
Safety and Health
CDR
Chemical Data Reporting
NIS
sodium-iodide symporter
CI
confidence interval
NOAEL
no-observed-adverse-effect level
CL
clearance
NPL
National Priority List
CLa
clearance in animals
NTP
National Toxicology Program
CLh
clearance in humans
OAT
organic anion transporter
CPAD
Chemical and Pollutant Assessment
OECD
Organisation for Economic Co-
Division
operation and Development
CPHEA
Center for Public Health and
OMB
Office of Management and Budget
Environmental Assessment
ORD
Office of Research and Development
CV
constant variance
OSF
oral slope factor
CYP450
cytochrome P450 superfamily
PC
partition coefficient
DAF
dosimetric adjustment factor
PBPK
physiologically based pharmacokinetic
DNA
deoxyribonucleic acid
PBTK
physiologically based toxicokinetic
DNT
developmental neurotoxicity
PECO
Populations, Exposures, Comparators,
DOD
Department of Defense
Outcomes
EPA
Environmental Protection Agency
PFAA
perfluoroalkyl acid
EOP
Executive Office of the President
PFAS
per- and polyfluoroalkyl substances
ER
extra risk
PFBA
perfluorobutanoic acid
FLR
full-litter resorption
PFBS
perfluorobutane sulfonate
FTOH
fluorotelomer alcohol
PFCA
perfluoroalkyl carboxylic acid
GD
gestation day
PFDA
perfluorodecanoic acid
GFR
glomerular filtration rate
PFHxA
perfluorohexanoic acid
GGT
y-glutamyl transferase
PFHxS
perfluorohexane sulfonate
GRADE
Grading of Recommendations
PFNA
perfluorononanoic acid
Assessment, Development, and
PFOA
perfluorooctanoic acid
Evaluation
PFOS
perfluorooctane sulfonate
GSH
glutathione
PK
pharmacokinetic
PND
postnatal day
ix
-------
Supplemental Information ofPFBA and Related Salts
POD
point of departure
TRI
Toxic Release Inventory
PODhed
human equivalent dose POD
TSCA
Toxic Substances Control Act
PPAR
peroxisome proliferator-activated
TSCATS
Toxic Substances Control Act Test
receptor
Submissions
PQAPP
Programmatic Quality Assurance
TSH
thyroid-stimulating hormone
Project Plan
TSHR
thyroid-stimulating hormone receptor
PT
prothrombin time
UCMR
Unregulated Contaminant Monitoring
PXR
pregnane X receptor
Rule
QA
quality assurance
UDP-GT
uridine 5'-diphospho-
QAPP
Quality Assurance Project Plan
glucuronosyltransferase
QMP
Quality Management Plan
UF
uncertainty factor
RBC
red blood cell
UFa
animal-to-human uncertainty factor
RD
relative deviation
UFc
composite uncertainty factor
RfC
inhalation reference concentration
UFd
database deficiencies uncertainty factor
RfD
oral reference dose
UFh
human variation uncertainty factor
RS
Rao-Scott
UFl
LOAEL-to-NOAEL uncertainty factor
SD
standard deviation
UFs
subchronic-to-chronic uncertainty
S-D
Sprague-Dawley
factor
SE
standard error
Vd
volume of distribution
TD
toxicodynamic
VOC
volatile organic compound
TH
thyroid hormone
WOS
Web of Science
TK
toxicokinetic
TPO
thyroid peroxidase
x
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX A. SYSTEMATIC REVIEW PROTOCOL FOR
THE PFAS IRIS ASSESSMENTS
1 A single systematic review protocol was used to guide the development of five, separate
2 IRIS PFAS assessments (i.e., PFBA, PFHxA, PFHxS, PFNA, and PFDA). This "systematic review
3 protocol for the PFAS IRIS assessments" was released for public comment and subsequently
4 amended. The amended protocol and prior revisions can be found at the following location:
5 http://cfpub.epa.gov/ncea/iris drafts/recordisplay.cfm?deid=345065.
6 When the assessment references a particular section or page number in Appendix A, please
7 refer to that section in the systematic review protocol linked above.
A-l
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX B. ADDITIONAL DETAILS OF
SYSTEMATIC REVIEW METHODS AND RESULTS
Table B-l. Perfluorobutanoic acid (PFBA) database search strategy
Search
Search strategy
Dates of search3
PubMed
Search
terms
375-22-4[rn] OR "Heptafluoro-l-butanoic acid"[tw] OR "Heptafluorobutanoic
acid"[tw] OR "Heptafluorobutyric acid"[tw] OR "Kyselina
heptafluormaselna"[tw] OR "Perfluorobutanoic acid"[tw] OR
"Perfluorobutyric acid"[tw] OR "Perfluoropropanecarboxylic acid"[tw] OR
"2,2,3,3,4,4,4-heptafluoro-Butanoic acid"[tw] OR "Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-"[tw] OR "Butanoic acid, heptafluoro-"[tw] OR
"Perfluoro-n-butanoic acid"[tw] OR "Perfluorobutanoate"[tw] OR
"2,2,3,3,4,4,4-Heptafluorobutanoic acid"[tw] OR "Butyric acid,
heptafluoro-"[tw] OR "Fluorad FC 23"[tw] OR "H 0024"[tw] OR "NSC 820"[tw]
OR «PFBA[tw] OR "FC 23"[tw] OR HFBA[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] OR polyfluoro*[tw] OR perfluoro-*[tw] OR
perfluoroa*[tw] OR perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] OR perfluoroh*[tw] OR perfluoron*[tw] OR
perfluoroo*[tw] OR perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw]))
No date
limit—7/19/2017
Literature
update
search
terms
(((375-22-4[rn] OR "Heptafluoro-l-butanoic acid"[tw] OR
"Heptafluorobutanoic acid"[tw] OR "Heptafluorobutyric acid"[tw] OR
"Kyselina heptafluormaselna"[tw] OR "Perfluorobutanoic acid"[tw] OR
"Perfluorobutyric acid"[tw] OR "Perfluoropropanecarboxylic acid"[tw] OR
"2,2,3,3,4,4,4-heptafluoro-Butanoic acid"[tw] OR "Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-"[tw] OR "Butanoic acid, heptafluoro-"[tw] OR
"Perfluoro-n-butanoic acid"[tw] OR "Perfluorobutanoate"[tw] OR
"2,2,3,3,4,4,4-Heptafluorobutanoic acid"[tw] OR "Butyric acid,
heptafluoro-"[tw] OR "Fluorad FC 23"[tw] OR "H 0024"[tw] OR "NSC 820"[tw]
OR «PFBA[tw] OR "FC 23"[tw] OR HFBA[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] OR polyfluoro*[tw] OR perfluoro-*[tw] OR
perfluoroa*[tw] OR perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] OR perfluoroh*[tw] OR perfluoron*[tw] OR
perfluoroo*[tw] OR perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw])) AND ("2017/08/01"[PDAT] : "2018/02/14"[PDAT])
8/1/2017-2/14/2018
B-l
-------
Supplemental Information ofPFBA and Related Salts
Search
Search strategy
Dates of search3
Web of Science
Search
terms
TS="Heptafluoro-l-butanoic acid" OR TS="Heptafluorobutanoic acid" OR
TS="Heptafluorobutyric acid" ORTS="Kyselina heptafluormaselna" OR
TS="Perfluorobutanoic acid" OR TS="Perfluorobutyric acid" OR
TS="Perfluoropropanecarboxylic acid" OR
TS="2,2,3,3,4,4,4-heptafluoro-Butanoic acid" OR TS="Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-" ORTS="Butanoicacid, heptafluoro-" OR
TS="Perfluoro-n-butanoic acid" ORTS="Perfluorobutanoate" OR
TS="2,2,3,3,4,4,4-Heptafluorobutanoic acid" OR TS="Butyric acid,
heptafluoro-" ORTS="Fluorad FC 23" ORTS="H 0024" ORTS="NSC 820" OR
(TS=(PFBA OR "FC 23" OR HFBA) AND TS=(fluorocarbon* OR fluorotelomer*
OR polyfluoro* OR perfluoro-* OR perfluoroa* OR perfluorob* OR
perfluoroc* OR perfluorod* OR perfluoroe* OR perfluoroh* OR perfluoron*
OR perfluoroo* OR perfluorop* OR perfluoros* OR perfluorou* OR
perfluorinated OR fluorinated OR PFAS OR PFOS OR PFOA))
No date
limit-7/20/2017
Literature
update
search
terms
((TS="Heptafluoro-l-butanoic acid" OR TS="Heptafluorobutanoic acid" OR
TS="Heptafluorobutyric acid" ORTS="Kyselina heptafluormaselna" OR
TS="Perfluorobutanoic acid" OR TS="Perfluorobutyric acid" OR
TS="Perfluoropropanecarboxylic acid" OR
TS="2,2,3,3,4,4,4-heptafluoro-Butanoic acid" OR TS="Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-" ORTS="Butanoicacid, heptafluoro-" OR
TS="Perfluoro-n-butanoic acid" ORTS="Perfluorobutanoate" OR
TS="2,2,3,3,4,4,4-Heptafluorobutanoic acid" OR TS="Butyric acid,
heptafluoro-" ORTS="Fluorad FC 23" ORTS="H 0024" ORTS="NSC 820") OR
TS=(PFBA OR "FC 23" OR HFBA) AND TS=(fluorocarbon* OR fluorotelomer* OR
polyfluoro* OR perfluoro-* OR perfluoroa* OR perfluorob* OR perfluoroc*
OR perfluorod* OR perfluoroe* OR perfluoroh* OR perfluoron* OR
perfluoroo* OR perfluorop* OR perfluoros* OR perfluorou* OR
perfluorinated OR fluorinated OR PFAS OR PFOS OR PFOA)) AND
PY=2017-2018
2017-2018
Toxline
Search
terms
( 375-22-4 [rn] OR "heptafluoro-l-butanoic acid" OR "heptafluorobutanoic
acid" OR "heptafluorobutyric acid" OR "kyselina heptafluormaselna" OR
"perfluorobutanoic acid" OR "perfluorobutyric acid" OR
"perfluoropropanecarboxylic acid" OR "2,2,3,3,4,4,4-heptafluoro-butanoic
acid" OR "butanoic acid 2 2 3 3 4 4 4-heptafluoro-" OR "butanoic acid
heptafluoro-" OR "perfluoro-n-butanoic acid" OR "perfluorobutanoate" OR
"2,2,3,3,4,4,4-heptafluorobutanoic acid" OR "butyric acid heptafluoro-" OR
"fluorad fc 23" OR "h 0024" OR "nsc 820" OR (( pfba OR "fc 23" OR hfba ) AND
(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro* OR
perfluorinated OR fluorinated OR pfas OR pfos OR pfoa ))) AND (ANEUPL
[org] OR BIOSIS [org] OR CIS [org] OR DART [org] OR EMIC [org] OR EPIDEM
[org] OR HEEP [org] OR HMTC [org] OR IPA [org] OR RISKUNE [org] OR
MTGABS [org] OR NIOSH [org] OR NTIS [org] OR PESTAB [org] OR PPBIB [org] )
AND NOT PubMed [org] AND NOT pubdart [org]
No date
limit-7/20/2017
B-2
-------
Supplemental Information ofPFBA and Related Salts
Search
Search strategy
Dates of search3
Literature
update
search
terms
@AND+@OR+("heptafluoro-l-butanoic
acid"+"heptafluorobutanoic+acid"+"heptafluorobutyric+acid"+"kyselina+hept
afluormaselna"+"perfluorobutanoic+acid"+"perfluorobutyric+acid"+"perfluor
opropanecarboxylic +acid"+"2 2 3 3 4 4
4-heptafluoro-butanoic+acid"+"butanoic+acid+2 2 3 3 4 4
4-heptafluoro-"+"butanoic+acid+heptafluoro-"+"perfluoro-n-butanoic
acid"+"perfluorobutanoate"+"2 2 3 3 4 4
4-heptafluorobutanoic+acid"+"butyric+acid+heptafluoro-"+"fluorad+fc+23"+"
h0024"+"nsc+820"+@TERM+@rn+375-22-4("pfba"+"fc+23"+"hfba"))+(
fluorocarbon*+
fluorotelomer*+polyfluoro*+perfluoro*+perfluorinated+fluorinated+pfas+pfo
s+pfoa)+@RANGE+yr+2017+2018
2017-2018
TSCATS
Search
terms
375-22-4[rn] AND tscats[org]
No date
limit-7/20/2017
a Yearly spring updates are conducted following release of the draft for public comment; see also the docket
("EPA-HQ-QRD-2020-0675-0022") for studies identified after the last formal update preceding public release.
B-3
-------
Supplemental Information ofPFBA and Related Salts
Table B-2. Title/abstract-level screening criteria for the initial literature
searches
Inclusion criteria
Exclusion criteria
Populations
Humans
Standard mammalian animal models, including rat,
mouse, rabbit, guinea pig, hamster, monkey, dog
Alternative animal models in standard laboratory
conditions (e.g., Xenopus, zebrafish, minipig)
Human or animal cells, tissues, or organs (not whole
animals); bacteria, nonmammalian eukaryotes; other
nonmammalian laboratory species
Ecological species
Exposures
Exposure is to PFBA
Study population is not exposed to PFBA
Exposure via oral, inhalation, dermal, intraperitoneal,
or intravenous injection routes
Exposure is to a mixture only
Exposure is measured in air, dust, drinking water,
diet, gavage, injection or via a biomarker of exposure
(PFBA levels in whole blood, serum, plasma, or
breastmilk)
Outcomes
Studies that include a measure of one or more health
effect endpoints, including but not limited to, effects
on reproduction, development, developmental
neurotoxicity, liver, thyroid, immune system, nervous
system, genotoxicity, and cancer
In vivo or in vitro studies related to toxicity
mechanisms, physiological effects/adverse outcomes,
and studies useful for elucidating toxic modes of
action (MOAs)
Qualitative or quantitative description of absorption,
distribution, metabolism, excretion, toxicokinetic or
toxicodynamic models (e.g., PBPK, PBTK, PBTK/TD)
Studies addressing risks to infants, children, pregnant
women, occupational workers, the elderly, and any
other susceptible or differentially exposed
populations
B-4
-------
Supplemental Information ofPFBA and Related Salts
Inclusion criteria
Exclusion criteria
Other
Structure and physiochemical properties
Reviews and regulatory documents
Not on topic, including:
Abstract only, inadequately reported
abstract, or no abstract and not
considered further because study was
not potentially relevant
Bioremediation, biodegradation, or
chemical or physical treatment of PFBA,
including evaluation of wastewater
treatment technologies and methods for
remediation of contaminated water and
soil
Ecosystem effects
Studies of environmental fate and
transport of PFBA in environmental
media
Analytical methods for
detecting/measuring PFAS compounds in
environmental media and use in sample
preparations and assays
Studies describing the manufacture and
use of PFBA
Not chemical specific (studies that do
not involve testing of PFBA)
Studies that describe measures of
exposure to PFBA without data on
associated health effects
MOA = mode of action; PBPK = physiologically based pharmacokinetic; PBTK = physiologically based toxicokinetic;
TD = toxicodynamic.
B-5
-------
Supplemental Information ofPFBA and Related Salts
Table B-3. Example DistillerSR form questions to be used for title/abstract-level and full text-level screening for
literature search updates from 2019
Used in title/abstract and full-text screening
Used in full text screening only
Source of
If meets PECO and
study if not
endocrine
identified
Does the
If meets PECO,
Which PFAS
outcome, which
from database
article meet
what type of
If supplemental, what
did the study
If meets PECO, which
endocrine tags
Question
search?
PECO criteria?
evidence?
type of information?
report?
health outcome(s) apply?
apply?
Answer
options
(can select
multiple
options)
Source other
Yes
Human
In vivo mechanistic or
PFBA
General toxicity, including
Adrenal
than HERO
No
Unclear
Tag as
potentially
relevant
supplemental
Animal (mam-
malian
models)
In vitro or
MOA studies,
PFHxA
PFHxS
PFNA
body weight, mortality,
Sex hormones
(e.g., androgen;
estrogen;
progesterone)
database
search
including non-PECO
routes of exposure
(e.g., injection) and
populations
and survival
Cancer
Cardiovascular, including
in silico
genotoxicity
(e.g., nonmammalian)
In vitro or in silico
PFDA
serum lipids
Endocrine (hormone)
Neuroendocrine
Pituitary
information
PBPKor PK
model
studies
(nongenotoxicity)
ADME/toxicokinetic
(excluding models)
Exposure assessment
or characterization (no
health outcome)
PFAS mixture study
(no individual PFAS
comparisons)
Gastrointestinal
Genotoxicity
Growth (early life) and
development
Hematological, including
nonimmune/hepatic/
renal clinical chemistry
measures
Hepatic, including liver
Steroidogenesis
Thyroid
Human case reports or
case series
measures and serum
markers (e.g., ALT; AST)
Ecotoxicity studies
Immune/
inflammation
B-6
-------
Supplemental Information ofPFBA and Related Salts
Used in title/abstract and full-text screening
Used in full text screening only
Question
Source of
study if not
identified
from database
search?
Does the
article meet
PECO criteria?
If meets PECO,
what type of
evidence?
If supplemental, what
type of information?
Which PFAS
did the study
report?
If meets PECO, which
health outcome(s) apply?
If meets PECO and
endocrine
outcome, which
endocrine tags
apply?
Environmental fate or
occurrence (including
food)
Manufacture,
engineering, use,
treatment,
remediation, or
laboratory methods
Other assessments or
records with no
original data
(e.g., reviews,
editorials,
commentaries)
Musculoskeletal
Nervous system,
including behavior and
sensory function
Nutrition and metabolic
Ocular
PBPK or PK model
Renal, including urinary
measures (e.g., protein)
Reproductive
Respiratory
Skin and connective
tissue effects
ADME = absorption, distribution, metabolism, and excretion; ALT = alanine aminotransferase; AST = aspartate aminotransferase; HERO = Health and
Environmental Research Online; MOA = mode of action; PBPK = physiologically based pharmacokinetic; PECO = Populations, Exposures, Comparators, and
Outcomes; PFAS = per- and polyfluoroalkyl substance; PFBA = perfluorobutanoic acid; PFDA = perfluorodecanoic acid; PFHxA = perfluorohexanoic acid;
PFHxS = perfluorohexanesulfonate; PFNA = perfluorononanoic acid; PK = pharmacokinetic.
B-7
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX C. ADDITIONAL TOXICOKINETIC
INFORMATION IN SUPPORT OF DOSE-RESPONSE
ANALYSIS
C.l. USE OF HALF-LIVES OF EXCRETION FOR DOSIMETRIC ADJUSTMENTS
The pharmacokinetics ofPFBA have only been measured after direct administration of
PFBA in single-exposure/single-day studies in animals fChang etal.. 20081. For the mouse, Chang et
al. f20081 performed 24-hour toxicokinetic studies after 10, 30, and 100 mg/kg oral doses. Based
on the area-under-the-concentration-curve (AUC) and maximum concentration (Cmax), the data also
appear approximately linear below 30 mg/kg but show some saturation above that dose rate (see
Figure C-l, Figure C-2).
9000
SOOO
7000
6000
E
i"
5000
4000
D
<
3000
2000
1000
0
* Male
+ Fan ale
0
i
+
Mouse 24-h AUC
after oral doses of
}
20
40 60
Dose (mg/kg)
SO
100
Figure C-l. Mouse AUC after oral doses of PFBA.
C-l
-------
Supplemental Information ofPFBA and Related Salts
350
300
250
•|2t>0
| 150
u
• Male
+ Fa^iale
100
5
I
Mouse Cmax after
oral doses of PFBA
}
}
50
0
0
20
40 60
Dose (mg/kg)
SO
100
Figure C-2. Mouse Cmax after oral doses of PFBA.
Chang etal. f20081 reported serum and liver concentrations in male rats and serum
concentrations in female rats given a 3-300 mg/kg oral dose ofPFBA at 24 hours after dosing.
Although the time point for these measurements is not ideal given the short half-life of PFBA, the
data indicate that the dosimetry is approximately linear up to 100 mg/kg in male rats and up to
30 mg/kg in female rats (see Figure C-3, Figure C-4). Tissue levels then appear to saturate or
decline; this might be due to incomplete absorption at higher doses, saturable renal resorption, or
both, whereby excretion is more rapid for concentrations above the level of saturable resorption in
the kidney. With the half-life in female rats being ~3 hours, the female serum 24-hour data are
particularly subject to experimental noise, but at least provide an indication that use of the half-life
measured using a 30 mg/kg dose is applicable to BMD levels from bioassays at or below this dose
rate.
C-2
-------
Supplemental Information ofPFBA and Related Salts
72
to 60
° 48
E
36
s
o
£ 24
c
OJ
u
312
• Serum
+ Lr/er
I i
0 *
0
¦i x
5
Male rat tissue
concentrations 24 h after
an oral dose of PFBA
50 100 150 200
Dose (mg/kg)
250
300
Figure C-3. Rat AUC after oral doses of PFBA.
C-3
-------
Supplemental Information ofPFBA and Related Salts
E
1
e
o
%
ai
0.4
0.35
0.3
0.25
0.2
0.15
o
0.1
0.05
0
Female rat serum
concentration 24 h after
j
an oral dose of PFBA
i
L
!
-1
«
r
-
*
0 50 100 150 200
Dose (mg/kg)
250
300
Figure C-4. Rat Cmax after oral doses of PFBA.
For the human data analyzed by Chang et al. f2008I detailed toxicokinetic parameters are
not available, but one can evaluate the relationship between the initial concentration and ti/2. Here
only data for subjects in which the final concentration is greater than the limit of quantification is
considered to avoid statistical artifacts due to limited observational data. Although the lower
half-life of the subject with the highest initial concentration indicates a possible negative trend, the
half-life is in the range of subjects with lower initial concentrations. Thus, these data do not show a
clear dose dependence for half-life and are interpreted as only showing interindividual variation
(see Figure C-5). The human data appear consistent with first-order clearance across the range of
concentrations observed.
C-4
-------
Supplemental Information ofPFBA and Related Salts
160
140
120
i 100
Ol
= SO
<•1
« 60
40
20
0
t
Human half-life vs. initial concentration
*
• •
•
•
•
•
•
} 20 40 60 SO
Initial concentration (ng/ml)
Figure C-5. Estimated human half-lives versus initial serum concentrations.
Chang etal. f20081 only evaluated one PFBA dose in monkeys, so determining whether the
biphasic clearance pattern is due to the classical distinction between distribution and excretion
phases or a nonlinearity in clearance is not possible. The data show linear clearance from 1-7 or
10 days after the i.v. dose was given, however, when serum concentrations were below 100 ng/mL.
Thus, interpreting these data as showing linear kinetics for serum concentrations below 100 ng/mL
under long-term exposure conditions seem reasonable. Because the highest initial condition of the
human subjects in Chang etal. (20081 was 72 ng/mL, to the extent that kinetics in monkeys can be
extrapolated to humans, the results for monkeys confirm the conclusion that human kinetics are
also reasonably assumed linear below ~100 ng/mL. This is approximately 1,000-fold below the
range of linearity in mice and rats, however, so uncertainty exists as to whether the range of linear
kinetics in humans and monkeys extends into the range of rodent-based points of departure.
Russell etal. (20151 attempted to evaluate the kinetics ofPFBA as a metabolite of
6:2 fluorotelomer alcohol (FTOH) during a 1-day inhalation study (6-hour exposure, 24-hour
observation) and at the end of 23 days of exposure. The half-life of PFBA, however, could not be
estimated from the single-day data for male rats and could be estimated only for the high-level
exposure in female rats, with yields of PFBA 0.2% in males and not detectable or 0.02% in females.
Also, three metabolic intermediates occur between 6:2 FTOH and PFBA, but the model appears to
have assumed direct, instantaneous transformation through the first two steps. Assumptions about
the volume of distribution were made by (Russell etal.. 20151. These simplifications in the model
likely explain the large discrepancy between the PFBA half-life determined from the single-day
exposure 6:2 FTOH for female rats (19 hours) and the half-life obtained for direct exposure to PFBA
(1.4-hour average) by f Chang etal.. 20081. Russell etal. f20151 used only male rats in the 23-day
C-5
-------
Supplemental Information ofPFBA and Related Salts
6:2 FTOH inhalation study, from which they estimated a half-life of 27.7 hours, over three times
higher than the average obtained by (Chang etal.. 20081. The discrepancy also could be due to an
underestimation of the metabolic yield from the 1-day experiments. In summary, whereas Russell
etal. f20151 described measurements ofPFBA in male rats from 23 days of exposure to 6:2 FTOH,
the results for female rats after a single exposure are completely inconsistent with the results of
f Chang etal.. 20081. Therefore, the conclusions from the multiday study are considered too
unreliable to be used.
The other long-term data available on internal dosimetry are from the bioassays (Butenhoff
etal.. 2012: Das etal.. 2008: van Otterdiik. 2007bl. Serum concentrations in nonpregnant female
mice after 17 days of exposure (24 hours after the last dose) are 2.0 ± 1.0 and 2.4 ± 1.7 [ig/mL, and
for pregnant mice are 3.8 ± 1.0 and 4.4 ± 0.7 [ig/mL, for the 35- and 175-mg/kg dose groups,
respectively fDas etal.. 20081. For female mice dosed with 30- and 100-mg/kg PFBA, Chang et al.
(20081 reported 4.1 ± 1.7 and 6.4 ± 3.9 |J.g/mL in serum 24 hours after the dose; using linear
extrapolation based on the difference in dose, one might expect 4.8 and 11.2 |J.g/mL at 24 hours
after doses of 35 and 175 mg/kg, given these data. Although the concentrations in the Das et al.
(20081 study are somewhat lower than these projections, the difference, especially at the low dose,
is within the range of uncertainty and precision expected for PK analysis.
Of note is that, given an average clearance of 28 mL/kg-hour obtained by Chang et al.
f20081 after 10- and 30-mg/kg doses, the predicted average serum concentrations for a 35-mg/kg
dose is 52 [ig/mL. This average concentration reflects the much higher concentrations expected in
the first few hours after each dose.
For male rats, Butenhoff et al. (20121 measured end-of-treatment serum levels of 38 ± 23
and 52 ± 25 |J.g/mL after 28 and 90 days, respectively, at 30 mg/kg-day; EPA presumes these
measurements were made 24 hours after the last dose. The corresponding values reported by
Chang etal. f20081 for a 30-mg/kg oral dose in the dose-range and time-course studies are 16 ± 3
and 29 ± 13 [ig/mL, respectively. Although again, some discrepancy is found between the
short-term PK data and the bioassay measurements, the difference is that it is roughly within a
factor of 2, which is acceptable for PK analysis and does not indicate a strong time dependence in
the PK. One should keep in mind that the estimated clearance and half-life values are based on
multiple time points at which the serum concentration is measured, while the comparisons above
use only a single time point, 24 hours after dosing, when the result will be sensitive to experimental
variation.
Given these data and results, the half-life or clearance ofPFBA measured in single-day
exposures by Chang etal. (20081 will be assumed to predict dosimetry after repeated exposures
that occur in bioassays. This is a common assumption for chemicals with relatively short half-lives
because pharmacokinetic studies are typically confined to a single day or less. Clearance in rats and
mice might include a slower beta phase, like that observed in monkeys. If a slow clearance phase
exists, internal dose from long-term exposure will be higher than is effectively estimated using the
C-6
-------
Supplemental Information ofPFBA and Related Salts
clearance rate determined from single-day exposures, which would increase the HED compared
with the current prediction. Using an animal-human ratio of clearance values to estimate the HED
relies only on the assumptions that the average serum concentration (Cavg) is predictive of systemic
effects in adults and that the relationship between Cavg and dose rate is linear with the
proportionality determined by the clearance values estimated here (i.e., the clearance from
single-day experiments is predictive of bioassay conditions).
The human half-life estimates were from subjects who had been occupationally exposed to
PFBA, with the duration of the PK observation 7-10 days. Thus, those results are reasonably
expected to represent clearance under (subsequent to) chronic exposure conditions. The primary
uncertainty in predicting human clearance comes from assuming a volume of distribution equal to
that estimated for monkeys, which is thought modest given the physiological similarity between
monkeys and humans. Thus, the overall uncertainty from using the animal-human clearance ratio
to predict the HED for systemic effects in adults appears modest, especially compared to the case
where PK data such as used here are not available.
Because developmental effects are usually presumed to depend on peak concentration
rather than average concentration, it must be noted that use of the clearance ratio to estimate HEDs
for those endpoints also involves an assumption that the absorption rate in humans is similar to
that of animals. For PFBA, the absorption rate in mice and rats is fairly rapid, with the peak
concentration occurring 0.6-4 hours after bolus oral doses fChang etal.. 20081. That absorption in
humans would be faster than in rodents seems unlikely, and exposures are more likely spread out
over the day than in the animal bioassays. Therefore, the most likely case is that the peak
concentration in humans exposed at the HED will be lower than the peak concentration in mice or
rats at the corresponding dose rate. Thus, although this assumption creates uncertainty in the dose
extrapolation, the result is not expected to underpredict human health risks.
C.2. MIXED MODELING TO ESTIMATE HALF-LIFE IN HUMANS
A linear mixed-effects model was additionally used to estimate a ti/2 for PFBA according to
methods described in (Li etal.. 2018). Briefly, linear mixed-effect models are extensions of simple
linear models that use the best linear unbiased prediction estimator to estimate random and fixed
effects for clustered data. One important consequence of clustering is that measurements of serum
PFBA units within the same person (cluster) are more similar than measurements on serum PFBA
in different people (i.e., other clusters). Failure to account for the intracluster correlation would
result in misleading inferences. Each individual in Chang etal. (2008) was assumed to have been
selected randomly from a larger population. Below is the mixed model formula used for estimating
the half-life of serum PFBA:
ln(PFBAjy) = (apop + at) + (/cpop + kt) x ty + (C-l)
C-7
-------
Supplemental Information ofPFBA and Related Salts
where ln(PFBA,y) is the natural logarithm of the serum PFBA concentrations measured at the jth
time point for the ith subject, crpop is the population mean (also known as the fixed intercept for the
population); ai ~ N (0, o2a) is a random intercept for the ith subject; /fpop is the fixed slope for the
population (also known as the average excretion rate constant for serum PFBA for the whole
population); ki ~ N (0, a2k) is the random slope for the ith subject that allows the excretion rate to
vary by individuals; t,y represents the observation time for the jth measurement of serum PFBA for
ith subject; and £,y ~ N (0, g2£) is the random-error effect (residual) for jth measurement of ith
subject Of note, the small sample sizes (due to the exclusion of the only two subjects identified as
females) limited our ability to draw clear conclusions in gender-stratified comparisons.
The subjects from Chang etal. f20081 used in this analysis and the half-lives estimated by
Chang etal. f20081 are listed in the following table. As explained in section 5.2.1, Approach for
Animal-Human Extrapolation of Perfluorobutanoic Acid (PFBA) Dosimetry, subjects whose second
concentration measurement was below the lower limit of quantitation (LLOQ) were excluded from
analysis because the half-life for these subjects is highly uncertain. This choice is expected to bias
the analysis towards higher half-lives but given the small number of human subjects for which data
are available, and that variability in clearance among the human population is expected, this is
considered a reasonably health-protective choice.
Subject
Sex
Reported half-life (h)
Cottage Grove Subject 1
NS
105.3
Cottage Grove Subject 1
NS
109.7
Cordova Subject 2
M
53
Cordova Subject 3
M
72
Cordova Subject 4
M
44
Cordova Subject 6
M
152
Cordova Subject 8
M
63
Cordova Subject 9
M
47
The half-life of serum PFBA for the study population (ti/2,pop) then was estimated as:
t
1/2,pop
lll(2)
'•pop
(C-2)
The mixed-effects model estimated /fpop to be -0.010, therefore resulting in an estimated ti/2
of 67.9 hours. Of the estimated half-lives reported by Chang etal. (2008). including those excluded
from analysis by EPA, five values (42% of the population) were greater than 67.9 hours and seven
(58% of the population) were below. This value also matches very closely to the median value
calculated when not taking clustering into account, and therefore was considered a reasonable
estimate of the population mean and was used in estimation of clearance in humans.
C-8
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX D. BENCHMARK DOSE MODELING
RESULTS
D.l. BENCHMARK DOSE MODELING APPROACHES
As discussed in Section 5 of the body of the Toxicological Review, the endpoints selected for
benchmark dose (BMD) modeling were relative liver weight, liver hypertrophy, total T4, and
thyroid follicular hypertrophy incidence from Butenhoff et al. f20121 and relative liver weight, full
litter resorption, delayed eye opening delayed vaginal opening, and delayed preputial separation
from (Das etal.. 20081. The animal doses in the study were used in the BMD modeling and then
converted to human equivalent doses (HEDs) using the ratio of animal-to-human clearance values;
the modeling results are presented in this appendix.
D.l.l. Modeling Procedure for Dichotomous Noncancer Data
BMD modeling of dichotomous noncancer data was conducted using EPA's Benchmark Dose
Software (BMDS, version 3.1.2). For these data, the Gamma, Logistic, Log-Logistic, Log-Probit,
Multistage, Probit, Weibull, and Dichotomous Hill models available within the software were fit
using a benchmark response (BMR) of 10% extra risk (see Toxicological Review, Section 4.2.1 for
justification of selected BMRs). The Multistage model is run for all polynomial degrees up to n - 2,
where n is the number of dose groups including control. Adequacy of model fit was judged on the
basis of x2 goodness-of-fit p-value (p >0.1), scaled residuals at the data point (except the control)
closest to the predefined benchmark response (absolute value <2.0), and visual inspection of the
model fit. In the cases where no best model was found to fit to the data, a reduced data set without
the high-dose group was further attempted for modeling and the result presented with that of the
full data set. In cases where a model with several parameters equal to the number of dose groups
was fit to the data set, all parameters were estimated, and no p-value was calculated, that model
was not considered for estimating a point of departure (POD) unless no other model provided
adequate fit. Among all models providing adequate fit, the benchmark dose lower confidence limit
(BMDL) from the model with the lowest Akaike's information criterion (AIC) was selected as a
potential POD when BMDL values were sufficiently close (within threefold). Otherwise, the lowest
BMDL was selected as a potential POD.
D.l.2. Modeling Procedure for Continuous Noncancer Data
BMD modeling of continuous noncancer data was conducted using EPA's Benchmark Dose
Software (BMDS, version 3.1.2). For these data, the Exponential, Hill, Polynomial, and Power models
available within the software are fit using a BMR of 1 standard deviation (SD) when no toxicological
D-l
-------
Supplemental Information ofPFBA and Related Salts
information was available to determine an adverse level of response. When toxicological
information was available, the BMR was based on relative deviation, as outlined in the Benchmark
Dose Technical Guidance fU.S. EPA. 20121 (see Toxicological Review, Section 5.2.1 for justification
for using BMRs); when a BMR based on relative deviation was used, modeling results using BMRs
based on SD are included for reference. An adequate fit is judged on the basis of x2 goodness-of-fit
p-value (p > 0.1), scaled residuals at the data point (except the control) closest to the predefined
benchmark response (absolute value <2.0), and visual inspection of the model fit In addition to
these three criteria for judging adequacy of model fit, a determination is made on whether the
variance across dose groups is homogeneous. If a homogeneous variance model is deemed
appropriate based on the statistical test provided by BMDS (i.e., Test 2), the final BMD results are
estimated from a homogeneous variance model. If the test for homogeneity of variance is rejected
(p < 0.05), the model is run again while modeling the variance as a power function of the mean to
account for this nonhomogeneous variance. If this nonhomogeneous variance model does not
adequately fit the data (i.e., Test 3; p < 0.05), alternative approaches were assessed on a case-by-
case basis. For example, in cases where neither variance model fit, or constant variance did not fit
(with adequate Test-4 p-value) and nonconstant variance did fit (with inadequate Test-4 p-value),
the log-normal distribution was attempted.
In cases where a model with several parameters equal to the number of dose groups was fit
to the data set, all parameters were estimated, and no p-value was calculated, that model was not
considered for estimating a POD unless no other model provided adequate fit. Among all models
providing adequate fit, the BMDL from the model with the lowest AIC was selected as a potential
POD when BMDL estimates differed by less than threefold. When BMDL estimates differed by
greater than threefold, the model with the lowest BMDL was selected to account for model
uncertainty.
In situations where there are multiple, related continuous endpoints in an organ system, or
endpoints that inform a generalized effect to the exposed organism (e.g., developmental delays in
multiple organ systems), modeling a combined endpoint of "total affected" animals was not
pursued. Such a systematic multi-endpoint modeling approach is not currently available in BMDS
other than the MS-Combo model that requires an assumption that the endpoints are independent;
such an assumption is likely not valid with respect to continuous non-cancer endpoints in the PFBA
toxicity database. Further, with respect to combining multiple continuous and dichotomous
endpoints outside of a specific multi-endpoint model, first a cut-off value would need to be
established for continuous endpoints to determine when an animal has "responded" (i.e., the
continuous data would need to be "dichotomized"). Such dichotomization of continuous data
results in a loss of precision and is recommended against in the BMD Technical Guidance (U.S. EPA,
2012).
D-2
-------
Supplemental Information ofPFBA and Related Salts
D.1.3. Modeling Procedure for Continuous Noncancer Developmental Toxicity Data
For continuous developmental toxicity data, data for individual animals were requested
from the study authors when possible. The use of individual animal data allows for the correct
measure of variance to be calculated. When a biological rationale for selecting a benchmark
response level is lacking, a BMR equal to 0.5 SD was used. The use of 1 SD for the BMR for
continuous endpoints is based on the observation that shifting the distribution of the control group
by 1 SD results in ~10% of the animal data points falling beyond an adversity cutoff defined at the
~1.5 percentile (Crump. 19951. This approximates the 10% extra risk commonly used as the BMR
for dichotomous endpoints. Thus, the use of 0.5 SD for continuous developmental toxicity endpoints
approximates the extra risk commonly used for dichotomous developmental toxicity endpoints.
D.1.4. Modeling Procedure for Dichotomous Noncancer Developmental Toxicity Data
For dichotomous developmental toxicity data, data for individual animals were requested
from the study authors when possible. This allowed the use of the nested logistic model, which
statistically accounts for intralitter similarity (the propensity of littermates to respond more like
one another than pups from another litter) by estimating intralitter correlation and using
litter-specific covariates. Other models (Rai and van Ryzin, NCTR) that also account for intralitter
similarity were not considered in modeling dichotomous developmental toxicity data as they are
not currently implemented in BMDS 3.2. Judging model fit for this model is identical to the
procedure used for regular dichotomous models.
D.1.5. Data Used for Modeling
The source of the data used for modeling is provided in Table D-l. For endpoints from the
Das etal. f20081 study, the study authors kindly provided individual dam-level data to facilitate
modeling and to provide corrected data where needed. These data also are included in full in the
tables below.
Table D-l. Sources of data used in benchmark dose modeling of PFBA
endpoints
Endpoint/Reference
Reference
Location
HAWC link
Relative liver weight
Butenhoff et al.
Appendix 1, page 37 (van
Otterdiik, 2007b)
https://hawcprd. eoa.gov/ani/endDoint/
(2012)
100507453/
Relative liver weight
Das et al. (2008)
Figure 2, page 175
https://hawcprd. eoa.gov/ani/endDoint/
100507508/
Liver hypertrophy
Butenhoff et al.
Table 9, page 523
httos://hawcord. eoa.gov/ani/endooint/
(2012)
100507383/
Total T4
Butenhoff et al.
Table 8, page 522
httos://hawcord. eoa.gov/ani/endooint/
(2012)
100507375/
Full litter resorption
Das et al. (2008)
Table D-2
D-3
-------
Supplemental Information ofPFBA and Related Salts
Endpoint/Reference
Reference
Location
HAWC link
Fetal mortality (full
litter resorptions
combined with fetal
death from litters
without full litter
resorptions)
Das et al. (2008)
Table D-3
Eyes opening
Das et al. (2008)
Table D-4
Vaginal opening
Das et al. (2008)
Table D-5
Preputial separation
Das et al. (2008)
Table D-6
Table D-2. Data received from study authors for Das et al. (2008)
on full litter resorptions (FLR)
Dose (mg/kg-d)
Number of implants FLR
0
8
0
18
35
2
175
2
175
2
175
9
175
5
350
3
350
2
350
13
350
13
350
3
350
14
350
13
Table D-3. Data received from study authors for Das et al. (2008) on fetal
death (litters without full litter resorptions) combined with full litter
resorptions
Dose (mg/kg-d)
Number of implants
Number of dead
Dam weight on GDI (litter-
specific covariate)
0
16
1
30
0
16
1
28.2
0
11
2
27.7
D-4
-------
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
Supplemental Information ofPFBA and Related Salts
Number of implants
11
12
11
15
14
12
14
16
13
17
14
13
11
18
15
13
13
14
15
13
12
13
14
16
13
15
13
14
Number of dead
18
Dam weight on GDI (litter-
specific covariate)
27.4
25.9
24.1
29.2
28
27.1
26.8
26.6
25.1
30.1
29
27.5
28.1
26.9
26.7
23.3
25.8
31.4
28.1
29.3
27.4
27
26.9
25.7
31.6
29.2
27.7
27.5
28.1
25.5
30.3
27.5
28.1
D-5
-------
35
35
35
35
35
35
35
35
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
Supplemental Information ofPFBA and Related Salts
Number of implants
13
11
10
13
13
13
12
14
15
14
14
15
14
15
16
11
16
11
13
11
15
14
13
12
16
11
14
Number of dead
Dam weight on GDI (litter-
specific covariate)
27.9
26.4
27.4
27.9
26.1
24.8
24.8
23.1
28.1
27.5
27.4
27.5
29.4
27.5
26
26.2
23.4
29.1
28.2
25.8
26.8
26.9
25
26.7
25.5
25.4
29
25
29.2
26.3
27.4
25.1
25.3
D-6
-------
Supplemental Information ofPFBA and Related Salts
Dose (mg/kg-d)
Number of implants
Number of dead
Dam weight on GDI (litter-
specific covariate)
350
12
1
29.5
350
16
2
28.8
350
17
2
26.2
350
12
2
26.2
350
16
0
27.3
350
9
3
27.6
350
13
0
27.7
350
13
0
27.4
350
13
1
26.4
350
7
1
24.6
350
3
3
21.5
350
2
2
23
350
13
13
25.8
350
13
13
24.6
350
3
3
25.1
350
14
14
28.2
350
13
13
29.2
350
1
1
25.4
Table D-4. Data received from study authors for Das et al. (2008)on delayed
eye opening
Dose (mg/kg-d)
Average day of eye opening
0
16.27
0
15.57
0
15.22
0
15.27
0
14.55
0
14.91
0
17.64
0
15.69
0
15.00
0
17.57
0
17.71
D-7
-------
0
0
0
0
0
0
0
0
0
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
Supplemental Information ofPFBA and Related Salts
Average day of eye opening
14.91
16.50
17.58
16.50
16.25
15.20
17.25
18.00
18.00
16.00
17.31
18.00
17.23
17.23
16.82
18.78
17.31
17.57
17.53
18.00
15.25
17.00
17.82
18.09
17.70
16.11
18.29
17.50
17.55
17.60
17.78
17.69
17.67
D-8
-------
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
350
350
350
350
350
350
350
350
350
350
350
350
350
350
350
Supplemental Information ofPFBA and Related Salts
Average day of eye opening
15.71
17.77
16.91
18.00
17.69
17.27
17.17
17.64
18.00
18.00
18.09
18.88
18.00
18.00
18.20
15.00
18.64
17.85
17.64
18.00
17.36
17.85
17.93
18.00
18.00
18.00
18.60
18.00
18.09
18.00
D-9
-------
Supplemental Information ofPFBA and Related Salts
Table D-5. Data received from study authors for Das et al. (2008) on delayed
vaginal opening
Dose (mg/kg-d)
Average day of vaginal opening
0
32.40
0
27.00
0
30.80
0
30.20
0
34.17
0
33.67
0
30.33
0
28.00
0
30.14
0
33.67
0
28.00
0
31.90
0
32.50
0
34.00
0
29.25
0
28.00
0
29.33
0
35.57
0
34.83
35
28.20
35
34.00
35
37.25
35
34.00
35
31.00
35
31.20
35
35.67
35
34.25
35
35.38
35
30.00
35
31.50
35
31.20
D-10
-------
35
35
35
35
35
35
35
35
35
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
350
350
350
350
350
350
350
Supplemental Information ofPFBA and Related Salts
Average day of vaginal opening
33.50
32.50
37.67
35.00
35.20
33.00
34.50
38.50
34.30
31.60
29.40
33.67
31.67
34.20
34.50
37.00
32.22
38.00
34.50
34.33
34.67
37.86
33.00
36.50
35.33
39.25
35.00
36.00
33.80
33.00
32.00
31.17
33.57
D-ll
-------
Supplemental Information ofPFBA and Related Salts
Dose (mg/kg-d)
Average day of vaginal opening
350
34.10
350
33.33
350
38.70
350
36.33
350
36.00
350
37.25
350
35.00
350
38.50
Table D-6. Data received from study authors for Das et al. (2008) on delayed
preputial separation
Dose (mg/kg-d)
Average day of preputial separation
0
29.00
0
28.20
0
28.20
0
28.00
0
31.80
0
29.20
0
28.71
0
30.00
0
31.00
0
28.29
0
30.00
0
29.80
0
31.00
0
29.50
0
29.00
0
31.00
0
29.67
35
27.40
35
33.40
35
28.20
35
31.80
D-12
-------
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
Supplemental Information ofPFBA and Related Salts
Average day of preputial separation
30.00
31.33
35.50
30.22
33.17
30.00
29.00
30.14
30.29
29.80
30.43
30.00
27.50
28.20
28.57
29.25
30.17
26.60
28.80
30.50
31.71
31.11
32.33
28.00
31.00
35.00
30.60
30.13
29.50
30.00
31.60
31.00
30.17
D-13
-------
Supplemental Information ofPFBA and Related Salts
Dose (mg/kg-d)
Average day of preputial separation
175
31.50
350
28.00
350
31.80
350
31.50
350
32.40
350
31.83
350
30.80
350
31.17
350
33.80
350
34.00
350
30.33
350
30.00
350
33.17
350
32.00
350
32.80
D.2. RELATIVE LIVER WEIGHT-MALE RATS EXPOSED 90 DAYS
fBUTENHOFF ET AL.. 2012: VAN OTTERDIIK. 2007B11
Table D-7. Dose-response data for relative liver weight in male rats following
90 day exposure fButenhoff et al.. 2012: van Otterdiik. 2007bl
Dose (mg/kg-d)
n
Mean
SD
0
10
2.11
0.13
1.2
10
2.29
0.14
6
10
2.26
0.16
30
10
2.8
0.32
'Throughout this document, if a model was selected as appropriately fitting the modeled data, that model's
entries in the tables are in green shaded cells and the text is bolded.
D-14
-------
Supplemental Information ofPFBA and Related Salts
Table D-8. Benchmark dose results for relative liver weight in male rats
exposed 90 days —constant variance, BMR = 10% relative deviation
fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models
Restriction3
10% Relative
deviation
p-Value
AIC
BMDS
classification13
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
11.3634
9.4685
0.1720
-8.8244
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Modeled control response
SD >| 1.51 actual response
SD
Exponential 3
(CV—normal)
Restricted
11.3634
9.4572
0.1720
-8.8244
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Modeled control response
SD >| 1.51 actual response
SD
Exponential 4
(CV—normal)
Restricted
10.4110
4.8569
0.0584
-6.7628
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Goodness-of-fit
p-value <0.1
Modeled control response
SD >| 1.51 actual response
SD
Exponential 5
(CV—normal)
Restricted
10.4033
4.8563
0.0584
-6.7621
Questionable
Constant variance test
failed (Test 2 p-
value < 0.05)
Goodness-of-fit
p-value <0.1
Modeled control response
SD >| 1.51 actual response
SD
Hill
(CV—normal)
Restricted
6.6152
6.0656
NA
-4.1913
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Modeled control response
SD >| 1.51 actual response
SD
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
D-15
-------
Supplemental Information ofPFBA and Related Salts
Models
Restriction3
10% Relative
deviation
p-Value
AIC
BMDS
classification13
BMDS notes
BMD
BMDL
Constant variance
Polynomial
(3 degree)
(CV—normal)
Restricted
12.8952
8.4671
0.0624
-6.8714
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Goodness-of-fit
p-value <0.1
Modeled control response
SD >| 1.51 actual response
SD
Polynomial
(2 degree)
(CV—normal)
Restricted
12.1463
8.4560
0.0611
-6.8370
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Goodness-of-fit
p-value <0.1
Modeled control response
SD >| 1.51 actual response
SD
Power
(CV—normal)
Restricted
10.4151
8.4328
0.1668
-8.7631
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Modeled control response
SD >| 1.51 actual response
SD
Linear
(CV—normal)
Unrestricted
10.4151
8.4328
0.1668
-8.7631
Questionable
Constant variance test
failed (Test 2
p-value < 0.05)
Modeled control response
SD >| 1.51 actual response
SD
a"Restriction" column denotes the restriction status of applied models.
^'Classification" column denotes whether a model can be considered for model selection purposes. See BMDS
User Guide: https://www.epa.gov/bmds.
D-16
-------
Supplemental Information ofPFBA and Related Salts
Table D-9. Benchmark dose results for relative liver weight in male rats
exposed 90 days—nonconstant variance, BMR = 10% relative deviation
fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Exponential 2
(NCV—normal)
Restricted
11.3982
9.0908
0.0362
-15.2001
Questionable
Goodness-of-fit
p-value <0.1
Exponential 3
(NCV—normal)
Restricted
11.3962
9.0911
0.0362
-15.2001
Questionable
Goodness-of-fit
p-value <0.1
Exponential 4
(NCV—normal)
Restricted
10.5179
5.2058
0.0096
-13.1325
Questionable
Goodness-of-fit
p-value <0.1
Exponential 5
(NCV—normal)
Restricted
10.5091
5.2055
0.0096
-13.1313
Questionable
Goodness-of-fit
p-value <0.1
Hill
(NCV—normal)
Restricted
11.1854
7.9783
0.0090
-13.0126
Questionable
Goodness-of-fit
p-value <0.1
Polynomial
(3 degree)
(NCV—normal)
Restricted
12.7313
8.1751
0.0104
-13.2674
Questionable
Goodness-of-fit
p-value <0.1
Polynomial
(2 degree)
(NCV—normal)
Restricted
11.9089
8.1513
0.0100
-13.2065
Questionable
Goodness-of-fit
p-value <0.1
Power
(NCV—normal)
Restricted
10.5174
8.1228
0.0350
-15.1326
Questionable
Goodness-of-fit
p-value <0.1
Linear
(NCV—normal)
Unrestricted
10.5179
8.1236
0.0350
-15.1326
Questionable
Goodness-of-fit
p-value <0.1
D-17
-------
Supplemental Information ofPFBA and Related Salts
Table D-10. Benchmark dose results for relative liver weight in male rats
exposed 90 days—log-normal distribution, constant variance, BMR = 10%
relative deviation fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models3
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Log-normal distribution, constant variance
Exponential 2
(CV— log-normal)
Restricted
11.5672
9.5455
0.1004
-14.1752
Viable-
Alternate
Modeled control
response SD >| 1.51
actual response SD
Exponential 3
(CV—log-normal)
Restricted
11.5672
9.6019
0.1004
-14.1752
Viable-
Recommended
Lowest AIC
Modeled control
response SD > 11.51
actual response SD
Exponential 4
(CV—log-normal)
Restricted
10.6449
5.1404
0.0311
-12.1242
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Exponential 5
(CV—log-normal)
Restricted
10.6419
5.1401
0.0311
-12.1239
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Hill
(CV—log-normal)
Restricted
10.5728
4.9799
0.0976
-14.1178
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Polynomial
(3 degree)
(CV—log-normal)
Restricted
12.6948
8.5635
0.0328
-12.2144
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Polynomial
(2 degree)
(CV—log-normal)
Restricted
11.9903
8.5515
0.0321
-12.1783
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
D-18
-------
Supplemental Information ofPFBA and Related Salts
Models3
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Log-normal distribution, constant variance
Power
(CV— log-normal)
Restricted
10.6452
8.5334
0.0979
-14.1242
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Linear
(CV—log-normal)
Unrestricted
10.6452
8.5334
0.0979
-14.1242
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Frequentist Exponential Degree 3 Model with BMR of 0.1 Rel. Dev. for the BMD and 0.95
Lower Confidence Limit for the BMDL
3.1
Dose
Figure D-l. Dose-response curve for the Exponential M3 model fit to relative
liver weight in male rats exposed 90 days (Butenhoff et al.. 2012: van
Otterdijk. 2007b).
D-19
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Exponential degree 3 vl.l
Dataset Name
Butenhoff_90_Lweight_rel
User notes
[Add user notes here]
Dose-Response Model
M[dose] = a * exp(±l * (b * dose)Ad)
Variance Model
Var[i] = alpha
Model Options
BMR Type
Rel. Dev.
BMRF
0.1
Tail Probability
-
Confidence Level
0.95
Distribution Type
Log-normal
Variance Type
Constant
Model Data
Dependent Variable
[Dose]
Independent Variable
[Mean]
Total # of Observations
4
Adverse Direction
Automatic
D-20
-------
Supplemental Information ofPFBA and Related Salts
Model Results
Benchmark Dose
BMD
11.56718731
BMDL
9.60187006
BMDU
14.67526197
AIC
-14.17517344
Test 4 P-value
0.100441772
D.O.F.
2
Model Parameters
# of Parameters
4
Va ria ble
Estimate
a
2.171112769
b
0.0082397
d
Bounded
log-alpha
-5.045994496
Goodness of Fit
Dose
Size
Estimated
Median
Calc'd
Median
Observed
Mean
Estimated
GSD
Calc'd GSD
Observed
SD
Scaled
Residual
0
10
2.171112769
2.10600663
2.11
1.08352413
1.063487
0.13
-0.17835832
1.2
10
2.192686432
2.28573248
2.29
1.08352413
1.062982
0.14
0.284010771
6
10
2.281146197
2.25435749
2.26
1.08352413
1.073268
0.16
-0.061715421
30
10
2.779944166
2.78189148
2.8
1.08352413
1.120657
0.32
0.058533184
Likelihoods of Interest
# of
Model
Log Likelihood*
Parameters
AIC
A1
12.38576382
5
-14.771528
A2
15.32442666
8
-14.648853
A3
12.38576382
5
-14.771528
fitted
10.08758672
3
-14.175173
R
-8.71328445
2
21.4265689
* Includes additive constant of -70.8323. This constant was not included in the LL derivation prior to BMDS 3.0.
Tests of Interest
-2*Log(Likelihood
Test
Ratio)
Test df
p-value
1
48.07542222
6
<0.0001
2
5.877325671
3
0.11773355
3
5.877325671
3
0.11773355
4
4.596354207
2
0.10044177
D-21
-------
Supplemental Information ofPFBA and Related Salts
Table D-ll. Benchmark dose results for relative liver weight in male rats
exposed 90 days—log-normal distribution, constant variance, BMR = 1
standard deviation fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Log-normal distribution, constant variance
Exponential 2
(CV— log-normal)
Restricted
9.7357
7.6047
0.1004
-14.1752
Viable-
Alternate
Modeled control
response SD >| 1.51
actual response SD
Exponential 3
(CV—log-normal)
Restricted
9.7356
7.6049
0.1004
-14.1752
Viable-
Recommended
Lowest AIC
Modeled control
response SD > 11.51
actual response SD
Exponential 4
(CV—log-normal)
Restricted
8.8962
0.0000
0.0311
-12.1242
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Exponential 5
(CV—log-normal)
Restricted
8.8943
6.9746
0.0311
-12.1239
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Hill
(CV—log-normal)
Restricted
8.8323
4.0523
0.0976
-14.1178
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Polynomial
(3 degree)
(CV—log-normal)
Restricted
10.7197
6.8148
0.0328
-12.2144
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Polynomial
(2 degree)
(CV—log-normal)
Restricted
10.1369
6.8036
0.0321
-12.1783
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Power
(CV—log-normal)
Restricted
8.8972
6.7871
0.0979
-14.1242
Questionable
Goodness-of-fit
p-value < 0.1
Modeled control
response SD >| 1.51
actual response SD
Linear
(CV—log-normal)
Unrestricted
8.8972
6.7871
0.0979
-14.1242
Questionable
Goodness-of-fit
p-value < 0.1
D-22
-------
Supplemental Information ofPFBA and Related Salts
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Log-normal distribution, constant variance
Modeled control
response SD >| 1.51
actual response SD
D.3. RELATIVE LIVER WEIGHT-Po MICE fDAS ET AL.. 20081
Table D-12. Dose-response data for relative liver weight in pregnant mice
fDas et al.. 20081
Dose (mg/kg-d)
n
Mean
SD
0
6
8.04
0.66
35
6
8.76
1.37
175
7
10.28
0.75
350
6
10.65
0.62
D-23
-------
Supplemental Information ofPFBA and Related Salts
Table D-13. Benchmark dose results for relative liver weight in pregnant
mice—constant variance, BMR = 10% relative deviation fDas et al.. 20081
Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
130.2877
98.9543
0.0486
73.1479
Questionable
Goodness-of-fit
p-value < 0.1
Exponential 3
(CV—normal)
Restricted
130.2877
99.1362
0.0486
73.1479
Questionable
Goodness-of-fit
p-value < 0.1
Exponential 4
(CV—normal)
Restricted
36.1911
15.1545
0.8612
69.1285
Viable-
recommended
Lowest AIC
Exponential 5
(CV—normal)
Restricted
39.4346
15.2398
NA
71.0979
Questionable
df = 0, saturated
model
(goodness-of-fit p-
value cannot be
calculated)
Hill
(CV—normal)
Restricted
38.7873
12.3846
NA
71.0979
Questionable
df = 0, saturated
model
(goodness-of-fit p-
value cannot be
calculated)
Polynomial (3
degree)
(CV—normal)
Restricted
115.5880
84.4884
0.0736
72.3159
Questionable
Goodness-of-fit
p-value < 0.1
Polynomial (2
degree)
(CV—normal)
Restricted
115.5878
84.4883
0.0736
72.3159
Questionable
Goodness-of-fit
p-value < 0.1
Power
(CV—normal)
Restricted
115.5870
84.4876
0.0736
72.3159
Questionable
Goodness-of-fit
p-value < 0.1
Linear
(CV—normal)
Unrestricted
115.5882
84.4875
0.0736
72.3159
Questionable
Goodness-of-fit
p-value < 0.1
D-24
-------
Supplemental Information ofPFBA and Related Salts
Frequentist Exponential Degree 4 Model with BMR of 0.1 Rel. Dev. for the BMD and 0.95
Lower Confidence Limit for the BMDL
i CS—1
Estimated Probability
Response at BMD
O Data
BMD
BMDL
Dose
Figure D-2. Dose-response curve for the Exponential M4 model fit to relative
liver weight in pregnant mice (Das etal.. 2008).
D-25
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Exponential degree 4 vl.l
Dataset Name
Das_p_Lweight_rel
User notes
[Add user notes here]
Dose-Response Model
M[dose] = a * [c-(c-l) * exp(-b * dose)]
Variance Model
Var[i] = alpha
Model Options
BMR Type
Rel. Dev.
BMRF
0.1
Tail Probability
-
Confidence Level
0.95
Distribution Type
Normal
Variance Type
Constant
Model Data
Dependent Variable
[Dose]
Independent Variable
[Mean]
Total # of Observations
4
Adverse Direction
Automatic
D-26
-------
Supplemental Information ofPFBA and Related Salts
#NAME?
Benchmark Dose
BMD
36.19110286
BMDL
15.15446485
BMDU
87.70968183
AIC
69.12846157
Test 4 P-value
0.861196136
D.O.F.
1
Model Parameters
# of Parameters
4
Variable
Estimate
a
8.018710905
b
0.009531749
c
1.342753894
log-alpha
-0.39273843
Goodness of Fit
Dose
Size
Estimated
Median
Calc'd
Median
Observed
Mean
Estimated
SD
Calc'd SD
Observed
SD
Scaled
Residual
0
6
8.018710905
8.04
8.04
0.82170879
0.66
0.66
0.063462168
35
6
8.798356028
8.76
8.76
0.82170879
1.37
1.37
-0.114338192
175
7
10.24876199
10.28
10.28
0.82170879
0.75
0.75
0.100580637
350
6
10.66937939
10.65
10.65
0.82170879
0.62
0.62
-0.057769406
Likelihoods of Interest
# of
Model
Log Likelihood*
Parameters
AIC
A1
-30.54894422
5
71.0978884
A2
-27.8068244
8
71.6136488
A3
-30.54894422
5
71.0978884
fitted
-30.56423079
4
69.1284616
R
-42.8486201
2
89.6972402
* Includes additive constant of -22.97346. This constant was not included in the LL derivation prior to BMDS 3.0.
Tests of Interest
-2*Log(Likelihood
Test
Ratio)
Test df
p-value
1
30.08359139
6
<0.0001
2
5.484239634
3
0.13958431
3
5.484239634
3
0.13958431
4
0.030573129
1
0.86119614
D-27
-------
Supplemental Information ofPFBA and Related Salts
Table D-14. Benchmark dose results for relative liver weight in pregnant
mice—constant variance, BMR = 1 standard deviation fDas et al.. 20081
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
141.5518
104.9937
0.0524
73.6332
Questionable
Goodness-of-fit
p-value < 0.1
Exponential 3
(CV—normal)
Restricted
141.5511
104.9942
0.0524
73.6331
Questionable
Goodness-of-fit
p-value < 0.1
Exponential 4
(CV—normal)
Restricted
37.2658
16.6945
0.5517
70.0879
Viable-
recommended
Lowest AIC
Exponential 5
(CV—normal)
Restricted
40.3641
16.7699
NA
71.7337
Questionable
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Hill
(CV—normal)
Restricted
39.5789
13.8731
NA
71.7337
Questionable
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Polynomial
(3 degree)
(CV—normal)
Restricted
124.9178
90.1236
0.0725
72.9822
Questionable
Goodness-of-fit
p-value < 0.1
Polynomial
(2 degree)
(CV—normal)
Restricted
124.9176
90.1235
0.0725
72.9822
Questionable
Goodness-of-fit
p-value < 0.1
Power
(CV—normal)
Restricted
124.9169
90.1256
0.0725
72.9822
Questionable
Goodness-of-fit
p-value < 0.1
Linear
(CV—normal)
Unrestricted
124.9180
90.1238
0.0725
72.9822
Questionable
Goodness-of-fit
p-value < 0.1
D-28
-------
Supplemental Information ofPFBA and Related Salts
DA. LIVER HYPERTROPHY-MALE RAT rBUTENHOFF ET AL.. 2012: VAN
OTTERDIIK. 2007B1
Table D-15. Dose-response data liver hypertrophy in male rats
(Butenhoff et al.. 2012: van Otterdijk. 2007b)
Dose (mg/kg-d)
n
Incidence
0
10
0
1.2
10
0
6
10
0
30
10
9
Table D-16. Benchmark dose results for liver hypertrophy in
rats—BMR = 10% extra risk fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models
Restriction
10% Extra risk
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Gamma
Restricted
16.2946
5.3859
1.0000
8.5017
Viable—alternate
Log-logistic
Restricted
23.5001
5.4486
1.0000
10.5017
Viable—alternate
Multistage 3rd
Restricted
10.8404
5.0184
0.9796
8.8673
Viable—alternate
Multistage 2nd
Restricted
6.8934
3.6966
0.8078
10.2814
Viable—alternate
Multistage 1st
Restricted
2.4428
1.4091
0.0817
18.5672
Questionable
Goodness-of-fit
p-value < 0.1
Weibull
Restricted
25.2757
5.3801
1.0000
8.5017
Viable-
recommended
Lowest AIC
Dichotomous Hill
Unrestricted
23.4994
5.8336
0.9995
12.5017
Viable—alternate
Logistic
Unrestricted
23.4727
8.4278
1.0000
8.5017
Viable—alternate
Log-probit
Unrestricted
20.1374
5.4722
1.0000
10.5017
Viable—alternate
Probit
Unrestricted
21.2661
7.6123
1.0000
10.5017
Viable—alternate
D-29
-------
Supplemental Information ofPFBA and Related Salts
Frequentist Weibull Model with BMR of 10% Extra Risk for the BMD and 0.95 Lower
Confidence Limit for the BMDL
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0®-
e-
-e-
15
Dose
Estimated Probability
Response at BMD
O Data
BMD
BMDL
Figure D-3. Dose-response curve for the Weibull model fit to liver
hypertrophy in male rats (Butenhoff et al.. 2012: van Otterdijk. 2007b).
D-30
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Weibull vl.l
Dataset Name
Butenhoff_90_Lhypertrophy
User notes
[Add user notes here]
Dose-Response Model
P[dose] = g + (l-g)*[l-exp(-b*doseAa)]
Model Options
RiskType
Extra Risk
BMR
0.1
Confidence Level
0.95
Background
Estimated
Model Data
Dependent Variable
Dose
Independent Va riable
Incidence
Total # of Observations
4
Model Results
Benchmark Dose
BMD
25.27565904
BMDL
5.380065202
BMDU
26.31774355
AiC
8.501660382
P-value
1
D.O.F.
3
Chi2
4.56905E-07
Model Parameters
# of Parameters
3
Variable
Estimate
g
Bounded
a
Bounded
b
5.94337E-27
Goodness of Fit
Dose
Estimated
Probability
Expected
Observed
Size
Scaled
Residual
0
1.523E-08
1.523E-07
0
10
-0.00039
1.2
1.523E-08
1.523E-07
0
10
-0.00039
6
1.52306E-08
1.52306E-07
0
10
-0.00039
30
0.899999999
8.999999992
9
10
8.003E-09
Analysis of Deviance
Model
Log Likelihood
# of Parameters
Deviance
Test d.f.
P Value
Full Model
-3.250829734
4
-
-
-
Fitted Model
-3.250830191
1
9.1381E-07
3
1
Reduced Model
-21.32655363
1
36.1514478
3
<0.0001
D-31
-------
Supplemental Information ofPFBA and Related Salts
Table D-17. Dose-response data for liver hypertrophy (slight severity lesions)
in male rats fButenhoff et al.. 2012: van Otterdiik. 2007bl
Dose (mg/kg-d)
n
Incidence
0
10
0
1.2
10
0
6
10
0
30
10
4
Table D-18. Benchmark dose results for liver hypertrophy (slight severity
lesions) in male rats—BMR = 10% extra risk fButenhoff et al.. 2012: van
Otterdijk. 2007b)
Models
Restriction
10% Extra risk
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Gamma
Restricted
23.1357
5.6717
1.0000
15.4602
Viable—alternate
Log-logistic
Restricted
27.1575
5.5461
1.0000
17.4602
Viable—alternate
Multistage 3rd
Restricted
17.7871
5.5407
0.9978
15.5422
Viable—alternate
Multistage 2nd
Restricted
13.9892
5.1121
0.8984
17.8741
Viable—alternate
Multistage 1st
Restricted
8.1158
3.9098
0.5376
19.5942
Viable-
recommended
Lowest BMDL
Weibull
Restricted
27.4811
5.6718
1.0000
17.4602
Viable—alternate
Dichotomous
Hill
Unrestricted
27.1562
5.2830
0.9995
19.4602
Viable—alternate
BMD:BMDL ratio > 5
Logistic
Unrestricted
26.9449
13.6106
1.0000
15.4602
Viable—alternate
Log-Probit
Unrestricted
24.8237
5.3131
1.0000
17.4602
Viable—alternate
Probit
Unrestricted
25.5166
12.1561
1.0000
17.4602
Viable—alternate
D.5. TOTAL T4—MALE RAT fBUTENHOFF ET AL.. 2012: VAN OTTERDIIK.
2007B1
Table D-19. Dose-response data for total T4 levels in male rats
(Butenhoff et al.. 2012: van Otterdijk. 2007b)
Dose (mg/kg-d)
n
Mean
SD
0
10
5.27
0.71
1.2
10
5.97
1.08
6
9
4.46
0.88
30
9
3.23
0.55
D-32
-------
Supplemental Information ofPFBA and Related Salts
Table D-20. Benchmark dose results for total T4 levels in male rats—constant
variance, BMR = 1 standard deviation fButenhoff et al.. 2012: van Otterdiik.
2007hl
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
9.2322
6.5166
0.0138
104.3816
Questionable
Goodness-of-fit
p-value <0.1
Exponential 3
(CV—normal)
Restricted
9.2324
6.5166
0.0138
104.3816
Questionable
Goodness-of-fit
p-value <0.1
Exponential 4
(CV—normal)
Restricted
4.9496
2.5239
0.0075
104.9572
Questionable
Goodness-of-fit
p-value <0.1
Exponential 5
(CV—normal)
Restricted
5.7655
3.5138
NA
103.5642
Questionable
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Hill
(CV—normal)
Restricted
5.5394
3.2999
NA
103.5644
Questionable
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Polynomial
(3 degree)
(CV—normal)
Restricted
11.5906
8.7704
0.0090
105.2374
Questionable
Goodness-of-fit
p-value <0.1
Polynomial
(2 degree)
(CV—normal)
Restricted
11.5906
8.7704
0.0090
105.2374
Questionable
Goodness-of-fit
p-value <0.1
Power
(CV—normal)
Restricted
11.5906
8.7706
0.0090
105.2374
Questionable
Goodness-of-fit
p-value <0.1
Linear
(CV—normal)
Unrestricted
11.5906
8.7704
0.0090
105.2374
Questionable
Goodness-of-fit
p-value <0.1
D-33
-------
Supplemental Information ofPFBA and Related Salts
Table D-21. Benchmark dose results for total T4 levels in male
rats—nonconstant variance, BMR = 1 standard deviation fButenhoff et al..
2012: van Otterdiik. 2007bl
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Exponential 2
(NCV—normal)
Restricted
11.3786
7.8978
0.0182
102.5921
Questionable
Goodness-of-fit
p-value <0.1
Exponential 3
(NCV—normal)
Restricted
11.3789
7.8977
0.0182
102.5921
Questionable
Goodness-of-fit
p-value <0.1
Exponential 4
(NCV—normal)
Restricted
5.8707
2.9606
0.0104
103.1558
Questionable
Goodness-of-fit
p-value <0.1
Exponential 5
(NCV—normal)
Restricted
5.8297
3.9098
NA
102.1810
Questionable
df= 0,saturated
model (goodness-of-fit
p-value cannot be
calculated)
Hill
(NCV—normal)
Restricted
5.8562
3.7033
NA
102.1809
Questionable
df= 0,saturated
model (goodness-of-fit
p-value cannot be
calculated)
Polynomial
(3 degree)
(NCV—normal)
Restricted
13.7327
10.1890
0.0130
103.2666
Questionable
Goodness-of-fit
p-value <0.1
Polynomial
(2 degree)
(NCV—normal)
Restricted
13.7329
10.1889
0.0130
103.2666
Questionable
Goodness-of-fit
p-value <0.1
Power
(NCV—normal)
Restricted
13.7325
10.1890
0.0130
103.2666
Questionable
Goodness-of-fit
p-value <0.1
Linear
(NCV—normal)
Unrestricted
13.7332
10.1889
0.0130
103.2666
Questionable
Goodness-of-fit
p-value <0.1
D-34
-------
Supplemental Information ofPFBA and Related Salts
Table D-22. Benchmark dose results for total T4 levels in male
rats—log-normal distribution, constant variance, BMR = 1 standard deviation
fButenhoff et al.. 2012: van Otterdiik. 2007bl
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Log-normal distribution, constant variance
Exponential 2
(CV— log-normal)
Restricted
12.0074
7.6347
0.0223
98.5676
Questionable
Goodness-of-fit
p-value <0.1
Exponential 3
(CV—log-normal)
Restricted
12.0074
7.6347
0.0223
98.5676
Questionable
Goodness-of-fit
p-value <0.1
Exponential 4
(CV—log-normal)
Restricted
5.7060
2.5325
0.0200
98.3698
Questionable
Goodness-of-fit
p-value <0.1
Exponential 5
(CV—log-normal)
Restricted
5.9263
3.4425
NA
97.5382
Questionable
df= 0,saturated
model (goodness-of-fit
p-value cannot be
calculated)
Hill
(CV—log-normal)
Restricted
Questionable
df= 0,saturated
model (goodness-of-fit
p-value cannot be
calculated)
Polynomial
(3 degree)
(CV—log-normal)
Restricted
Questionable
Goodness-of-fit
p-value <0.1
Polynomial
(2 degree)
(CV—log-normal)
Restricted
Questionable
Goodness-of-fit
p-value <0.1
Power
(CV—log-normal)
Restricted
-
-
-
-
Questionable
Goodness-of-fit
p-value <0.1
Linear
(CV—log-normal)
Unrestricted
-
-
-
-
Questionable
Goodness-of-fit
p-value <0.1
D-35
-------
Supplemental Information ofPFBA and Related Salts
D.6. INCREASED FETAL MORTALITY - MALE AND FEMALE Fi MICE fDAS
ETAL.. 20081
Table D-23. Dose-response data for increased fetal mortality (Das et al.. 2008)
Dose (mg/kg-d)
n (No. of implants)
No. of dead fetuses/neonates by
PND21
Litter-specific covariate
(Maternal weight on GDI)
0
16
1
30
0
16
1
28.2
0
11
2
27.7
0
11
0
27.4
0
12
3
25.9
0
11
0
24.1
0
15
0
29.2
0
14
1
28
0
12
3
27.1
0
14
0
26.8
0
16
1
26.6
0
13
2
25.1
0
17
3
30.1
0
14
0
29
0
6
0
27.5
0
9
2
28.1
0
6
0
26.9
0
13
1
26.7
0
11
0
23.3
0
8
8
25.8
0
18
18
31.4
35
15
3
28.1
35
13
0
29.3
35
13
0
27.4
35
14
1
27
35
15
2
26.9
35
13
2
25.7
35
12
4
31.6
35
13
0
29.2
35
14
1
27.7
D-36
-------
35
35
35
35
35
35
35
35
35
35
35
35
35
35
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
175
Supplemental Information ofPFBA and Related Salts
n (No. of implants)
16
13
15
13
14
13
11
10
13
13
13
12
14
15
14
14
15
14
15
16
11
16
11
13
11
15
14
13
. of dead fetuses/neonates by
PND21
0
Litter-specific covariate
(Maternal weight on GDI)
27.5
28.1
25.5
30.3
27.5
28.1
27.9
26.4
27.4
27.9
26.1
24.8
24.8
23.1
28.1
27.5
27.4
27.5
29.4
27.5
26
26.2
23.4
29.1
28.2
25.8
26.8
26.9
25
26.7
25.5
25.4
29
25
D-37
-------
Supplemental Information ofPFBA and Related Salts
Dose (mg/kg-d)
n (No. of implants)
No. of dead fetuses/neonates by
PND21
Litter-specific covariate
(Maternal weight on GDI)
350
7
2
29.2
350
12
1
26.3
350
16
3
27.4
350
11
0
25.1
350
14
2
25.3
350
12
1
29.5
350
16
2
28.8
350
17
2
26.2
350
12
2
26.2
350
16
0
27.3
350
9
3
27.6
350
13
0
27.7
350
13
0
27.4
350
13
1
26.4
350
7
1
24.6
350
3
3
21.5
350
2
2
23
350
13
13
25.8
350
13
13
24.6
350
3
3
25.1
350
14
14
28.2
350
13
13
29.2
Table D-24. Benchmark dose results for increased fetal mortality (male and
female mice)—BMR = 1% extra risk fDas etal.. 20081
Models
Restriction
1% Extra risk
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nested logistic
(lsc+ilc+)
Restricted
19.5989
5.7383
Infinity
0.2633
Viable-
Recommended
Lowest AIC
BMDL 3x lower
than lowest non-
zero dose
Nested logistic
(Isc+ilc-)
Restricted
326.9633
170.7455
Infinity
<0.0001
Questionable
Goodness of fit p-
value < 0.1
D-38
-------
Supplemental Information ofPFBA and Related Salts
Models
Restriction
1% Extra risk
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nested logistic
(lsc-ilc+)
Restricted
50.4014
10.1822
Infinity
0.0833
Questionable
Goodness of fit p-
value < 0.1
BMDL 3x lower
than lowest non-
zero dose
Nested logistic
(Isc-ilc-)
Restricted
191.2272
81.9934
Infinity
<0.0001
Questionable
Goodness of fit p-
value < 0.1
Frequentist Nested Logistic Model with BMR of 0.01 Std. Dev. for the BMDand 0.95
Lower Confidence Limit for the BMDL
0.6
Dose
Figure D-4. Dose-response curve for the Nested-Logistic model fit to increased
fetal mortality in male and female mice fDas et al.. 20081.
D-39
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Nested Logistic_lsc+ilc+_ v2.2
Dataset Name
Das FLR Fetal Death
User notes
[Add user notes here]
Dose-Response Model
P[dose] = alpha + thetal*Rij + [1 - alpha -
thetal*Rij]/[l+exp(-beta-theta2*Rij-rho*log(dose))]
Model Options
Risk Type
Extra Risk
BMR
0.01
Confidence Level
0.95
Litter Specific Covariate
Overall Mean
Intralitter Correlation
Estimate
Background
Estimate
Model Data
Dependent Variable
Dose
Independent Variable
Incidence
Total # of Observations
87
D-40
-------
Supplemental Information ofPFBA and Related Salts
Model Results
Benchmark Dose
BMD
19.59891366
BMDL
5.738265629
BMDU
-
AIC
688.92042
P-value
0.263333333
D.O.F.
78
Chi2
96.74138773
Model Parameters
# of Parameters
9
Va riable
Estimate
alpha
-0.5312932
beta
16.8290783
thetal
0.024711312
theta2
-0.913645475
rho
1.08467654
phil
0.368252856
phi2
0.135465621
phi3
0.509745798
phi4
0.576861839
Bootstrap Results
# Iterations
1000
Bootstrap Seed
1599045577
Log-likelihood
-335.46021
Observed Chi-square
96.74138773
Combined P-value
0.263333333
Bootstrap Runs
Run
Bootstrap Chi-square Percentiles
P-Value
50th
90th
95th
99th
1
0.285
85.65851617
109.848694
117.6
134.18995
2
0.258
85.12942257
110.722914
119.0851
131.33939
3
0.247
85.05473338
108.751327
115.9296
137.48443
Combined
0.263333333
85.30000651
109.644757
117.9646
135.17128
Scaled Residuals
Minimum scaled residual for dose group nearest the BMD
-0.50395
Minimum ABS(scaled residual) for dose group nearest the BMD
0.503952
Average Scaled residual for dose group nearest the BMD
-0.50395
Average ABS(scaled residual) for dose group nearest the BMD
0.503952
Maximum scaled residual for dose group nearest the BMD
-0.50395
Maximum ABS(scaled residual) for dose group nearest the BMD
0.503952
D-41
-------
Supplemental Information ofPFBA and Related Salts
Litter Data
Dose
Lit. Spec. Cov.
Est. Prob.
Litter Size
Expected
Observed
Scaled Residua
0
23.3
0.044480361
11
0.489284
0
-0.330689547
0
24.1
0.06424941
11
0.706744
0
-0.401615664
0
25.1
0.088960722
13
1.156489
2
0.353012617
0
25.8
0.10625864
8
0.850069
8
4.33673202
0
25.9
0.108729771
12
1.304757
3
0.699492477
0
26.6
0.126027689
16
2.016443
1
-0.299772055
0
26.7
0.12849882
13
1.670485
1
-0.238711127
0
26.8
0.130969952
14
1.833579
0
-0.603802545
0
26.9
0.133441083
6
0.800646
0
-0.570250323
0
27.1
0.138383345
12
1.6606
3
0.498243205
0
27.4
0.145796738
11
1.603764
0
-0.633212408
0
27.5
0.14826787
6
0.889607
0
-0.606305933
0
27.7
0.153210132
11
1.685311
2
0.121734257
0
28
0.160623525
14
2.248729
1
-0.377819018
0
28.1
0.163094657
9
1.467852
2
0.241698202
0
28.2
0.165565788
16
2.649053
1
-0.434253259
0
29
0.185334837
14
2.594688
0
-0.741848906
0
29.2
0.190277099
15
2.854156
0
-0.756724162
0
30
0.210046149
16
3.360738
1
-0.567256611
0
30.1
0.21251728
17
3.612794
3
-0.138387912
0
31.4
0.244641985
18
4.403556
18
2.76675012
35
23.1
0.420439493
2
0.840879
2
1.558208765
35
24.8
0.193667312
12
2.324008
1
-0.612920728
35
24.8
0.193667312
13
2.517675
1
-0.657368032
35
25.5
0.160429819
7
1.123009
3
1.435714172
35
25.7
0.15539473
13
2.020131
2
-0.009511434
35
26.1
0.149721705
13
1.946382
0
-0.933727752
35
26.4
0.148530963
11
1.633841
0
-0.902727468
35
26.9
0.150776303
15
2.261645
2
-0.110930327
35
27
0.151716296
14
2.124028
1
-0.503951928
35
27.4
0.156698447
13
2.03708
0
-0.959178264
35
27.4
0.156698447
10
1.566984
1
-0.331093052
35
27.5
0.158199979
16
2.5312
0
-0.995853056
35
27.5
0.158199979
13
2.0566
0
-0.964622031
35
27.7
0.16145156
14
2.260322
1
-0.550928027
35
27.9
0.164988496
13
2.14485
1
-0.527947549
35
27.9
0.164988496
13
2.14485
1
-0.527947549
35
28.1
0.168763344
15
2.53145
3
0.189788804
35
28.1
0.168763344
14
2.362687
1
-0.585185094
35
28.1
0.168763344
13
2.193923
2
-0.088622513
35
29.2
0.192293335
13
2.499813
0
-1.08570969
35
29.3
0.194583201
13
2.529582
0
-1.093706422
35
30.3
0.218173919
15
3.272609
1
-0.834803748
35
31.6
0.249793258
12
2.997519
4
0.423637452
175
23.4
0.753292803
11
8.286221
0
-2.346999161
175
25
0.450913899
14
6.312795
1
-1.033293673
175
25
0.450913899
5
2.254569
5
1.415449135
175
25.4
0.381299381
2
0.762599
2
1.466122666
175
25.5
0.365523168
2
0.731046
2
1.516399081
175
25.8
0.322690467
13
4.194976
0
-0.932876612
175
26
0.298046899
15
4.470703
0
-0.884741991
175
26.2
0.276550941
16
4.424815
2
-0.460908554
175
26.7
0.235805287
13
3.065469
1
-0.505849379
175
26.8
0.229690589
11
2.526596
2
-0.152863219
175
26.9
0.2241858
15
3.362787
1
-0.512836334
175
27.4
0.204707536
14
2.865906
0
-0.687383811
175
27.5
0.202204115
14
2.830858
1
-0.441145101
175
27.5
0.202204115
14
2.830858
1
-0.441145101
175
27.5
0.202204115
15
3.033062
0
-0.683561326
175
28.1
0.194568014
14
2.723952
1
-0.421446879
175
28.2
0.194300585
11
2.137306
0
-0.659587572
175
29
0.199087513
9
1.791788
9
2.670218904
175
29.1
0.200338227
16
3.205412
3
-0.043633258
175
29.4
0.204695235
15
3.070429
2
-0.240145327
350
21.5
0.971795484
3
2.915386
3
0.201065485
350
23
0.901250165
2
1.8025
2
0.372789514
350
24.6
0.695111669
13
9.036452
13
0.848375256
350
24.6
0.695111669
7
4.865782
1
-1.502678856
350
25.1
0.601246578
3
1.80374
3
0.961150815
350
25.1
0.601246578
11
6.613712
0
-1.565382323
350
25.3
0.562306321
14
7.872288
2
-1.085132125
350
25.4
0.542866281
1
0.542866
1
0.91764604
350
25.8
0.467290016
13
6.07477
13
1.367719528
350
26.2
0.398842995
12
4.786116
2
-0.606043119
350
26.2
0.398842995
17
6.780331
2
-0.740295677
350
26.3
0.383300844
12
4.59961
1
-0.788584398
350
26.4
0.368470315
13
4.790114
1
-0.774204472
350
27.3
0.269132881
16
4.306126
0
-0.781258285
350
27.4
0.261775084
13
3.403076
0
-0.762807027
350
27.4
0.261775084
16
4.188401
3
-0.217527974
350
27.6
0.249017755
9
2.24116
3
0.246847242
350
27.7
0.243559594
13
3.166275
0
-0.726875687
350
28.2
0.224244443
14
3.139422
14
2.387132337
350
28.8
0.214724836
16
3.435597
2
-0.281313821
350
29.2
0.214117934
13
2.783533
13
2.454127084
350
29.2
0.214117934
7
1.498826
2
0.218630258
350
29.5
0.215777472
12
2.58933
1
-0.411519634
D-42
-------
Supplemental Information ofPFBA and Related Salts
D.7. DELAYED EYE OPENING-Fi MALE AND FEMALE MICE fDAS ET AL..
20081
Table D-25. Dose-response data for delayed eye opening in male
and female mice (Das et al.. 2008)
Dose (mg/kg-d)
n
Mean
SD
0
20
16.28
1.19
35
22
17.38
0.79
175
17
17.69
0.68
350
15
17.8
0.83
D-43
-------
Supplemental Information ofPFBA and Related Salts
Table D-26. Benchmark dose results for delayed eye opening in male and
female mice—constant variance, BMR = 5% relative deviation fDas etal..
20081
Models
Restriction
5% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
252.3387
178.6688
0.0008
211.1176
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Exponential 3
(CV—normal)
Restricted
252.3380
178.7347
0.0008
211.1176
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Exponential 4
(CV—normal)
Restricted
20.4436
0.0000
0.7270
198.8811
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
Exponential 5
(CV—normal)
Restricted
175.5239
0.0000
NA
215.6060
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
| Residual at control | >2
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Hill
(CV—normal)
Restricted
16.1508
4.8878
0.8659
198.7878
Viable-
recommended
Lowest AIC
BMDL 3x lower than
lowest nonzero dose
Polynomial
(3 degree)
(CV—normal)
Restricted
247.2477
172.9292
0.0008
210.9441
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Polynomial
(2 degree)
(CV—normal)
Restricted
247.2476
172.9292
0.0008
210.9441
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Power
(CV—normal)
Restricted
247.2483
172.9366
0.0008
210.9441
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Linear
(CV—normal)
Unrestricted
247.2471
172.9288
0.0008
210.9441
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
D-44
-------
Supplemental Information ofPFBA and Related Salts
Frequentist Hill Model with BMR of 0.05 Rel. Dev. for the BMD and 0.95 Lower
Confidence Limit for the BMDL
Dose
Figure D-5. Dose response curve for the Hill model fit to delayed eye opening
in male and female mice (Das etal.. 2008).
D-45
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Hill vl.l
Dataset Name
Das EO litter SDs
User notes
[Add user notes here]
Dose-Response Model
M[dose] = g + v*doseAn/(kAn + doseAn)
Variance Model
Var[i] = alpha
Model Options
BMR Type
Rel. Dev.
BMRF
0.05
Tail Probability
-
Confidence Level
0.95
Distribution Type
Normal
Variance Type
Constant
Model Data
Dependent Variable
[Dose]
Independent Variable
[Mean]
Total # of Observations
4
Adverse Direction
Automatic
D-46
-------
Supplemental Information ofPFBA and Related Salts
Model Results
Benchmark Dose
BMD
16.15084927
BMDL
4.88775303
BMDU
58.67497527
AIC
198.7877861
Test 4 P-value
0.865852068
D.O.F.
1
Model Parameters
# of Parameters
5
Variable
Estimate
g
16.28027637
V
1.557732828
k
14.75612987
n
Bounded
alpha
0.771309051
Goodness of Fit
Dose
Size
Estimated
Median
Calc'd
Median
Observed
Mean
Estimated
SD
Calc'd SD
Observed
SD
Scaled
Residual
0
20
16.28027637
16.28
16.28
0.87824202
1.19
1.19
-0.001407337
35
22
17.3760338
17.38
17.38
0.87824202
0.79
0.79
0.021182211
175
17
17.71687421
17.69
17.69
0.87824202
0.68
0.68
-0.126167037
350
15
17.77499146
17.8
17.8
0.87824202
0.83
0.83
0.110285841
Likelihoods of Interest
# of
Model
Log Likelihood*
Parameters
AIC
A1
-95.37962446
5
200.759249
A2
-91.88601151
8
199.772023
A3
-95.37962446
5
200.759249
fitted
-95.39389305
4
198.787786
R
-109.7197233
2
223.439447
* Includes additive constant of -68.00145. This constant was not included in the LL derivation prior to BMDS 3.0.
Tests of Interest
-2*Log(Likelihood
Test
Ratio)
Test df
p-value
1
35.6674235
6
<0.0001
2
6.987225901
3
0.07230604
3
6.987225901
3
0.07230604
4
0.028537187
1
0.86585207
D-47
-------
Supplemental Information ofPFBA and Related Salts
Table D-27. Benchmark dose results for delayed eye opening in male and
female mice—constant variance, BMR = 1 standard deviation fDas etal.. 20081
Models
Restriction
1 Standard deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
289.0417
204.0632
0.0008
211.1176
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Exponential 3
(CV—normal)
Restricted
289.0397
204.0631
0.0008
211.1176
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Exponential 4
(CV—normal)
Restricted
23.0895
12.5328
0.7270
198.8811
Viable-
recommended
Lowest AIC
Exponential 5
(CV—normal)
Restricted
-9,999.0000
0.0000
NA
215.6060
Unusable
BMD computation
failed
BMD not estimated
BMDL not estimated
| Residual at control |
>2
df= 0,saturated
model (goodness-of-fit
p-value cannot be
calculated)
Hill
(CV—normal)
Restricted
19.0723
0.0000
0.8659
198.7878
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
Polynomial
(3 degree)
(CV—normal)
Restricted
284.0211
198.2059
0.0008
210.9441
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Polynomial
(2 degree)
(CV—normal)
Restricted
284.0211
198.2059
0.0008
210.9441
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Power
(CV—normal)
Restricted
284.0218
198.2009
0.0008
210.9441
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Linear
(CV—normal)
Unrestricted
284.0204
198.2054
0.0008
210.9441
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
D-48
-------
Supplemental Information ofPFBA and Related Salts
D.8. VAGINAL OPENING-Fi FEMALE MICE fDAS ET AL.. 20081
Table D-28. Dose response data for delayed vaginal opening in
female mice fDas et al.. 20081
Dose (mg/kg-d)
n
Mean
SD
0
83
31.59
5.386
35
97
33.598
5.715
175
89
34.292
5.714
350
87
35.023
5.188
Table D-29. Benchmark dose results for delayed vaginal opening in female
mice—constant variance, 5% relative deviation (Das et al.. 2008)
Models
Restriction
5% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
199.6149
137.1410
0.0106
348.8761
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
Exponential 3
(CV—normal)
Restricted
199.6216
137.1431
0.0106
348.8761
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
Exponential 4
(CV—normal)
Restricted
17.1139
0.0000
0.6944
341.9320
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
Exponential 5
(CV—normal)
Restricted
30.5201
0.0000
NA
343.9392
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Hill
(CV—normal)
Restricted
13.5161
3.7929
0.8401
341.8184
Viable-
recommended
Lowest AIC
BMDL 3x lower than
lowest nonzero dose
Polynomial (3
degree)
(CV—normal)
Restricted
193.4400
130.5619
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
D-49
-------
Supplemental Information ofPFBA and Related Salts
Models
Restriction
5% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Polynomial (2
degree)
(CV—normal)
Restricted
193.4443
130.5615
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
Power
(CV—normal)
Restricted
193.4434
130.5626
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
Linear
(CV—normal)
Unrestricted
193.4436
130.5610
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control |
>2
Figure D-6. Dose response curve for the Hill model fit to delayed vaginal
opening in female mice fDas et al.. 20081.
D-50
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Hill vl.l
Dataset Name
Das VO litter SDs
User notes
[Add user notes here]
Dose-Response Model
M[dose] = g + v*doseAn/(kAn + doseAn)
Variance Model
Var[i] = alpha
Model Options
BMR Type
Rel. Dev.
BMRF
0.05
Tail Probability
-
Confidence Level
0.95
Distribution Type
Normal
Variance Type
Constant
Model Data
Dependent Variable
[Dose]
Independent Variable
[Mean]
Total # of Observations
4
Adverse Direction
Automatic
D-51
-------
Supplemental Information ofPFBA and Related Salts
Model Results
Benchmark Dose
BMD
13.51609885
BMDL
3.792905489
BMDU
58.81907947
AIC
341.8183924
Test 4 P-value
0.840124836
D.O.F.
1
Model Parameters
# of Parameters
5
Variable
Estimate
g
31.25160173
V
3.782877454
k
19.2052612
n
Bounded
alpha
6.040525655
Goodness of Fit
Dose
Size
Estimated
Median
Calc'd
Median
Observed
Mean
Estimated
SD
Calc'd SD
Observed
SD
Scaled
Residual
0
19
31.25160173
31.25
31.25
2.45774809
2.62
2.62
-0.002840717
35
21
33.69418217
33.71
33.71
2.45774809
2.59
2.59
0.029493016
175
17
34.66038453
34.57
34.57
2.45774809
2.59
2.59
-0.151628625
350
15
34.83770206
34.92
34.92
2.45774809
2.23
2.23
0.129687238
Likelihoods of Interest
# of
Model
Log Likelihood*
Parameters
AIC
A1
-166.8888479
5
343.777696
A2
-166.5982185
8
349.196437
A3
-166.8888479
5
343.777696
fitted
-166.9091962
4
341.818392
R
-177.364099
2
358.728198
* Includes additive constant of -66.16357. This constant was not included in the LL derivation prior to BMDS 3.0.
Tests of Interest
-2*Log(Likelihood
Test
Ratio)
Test df
p-value
1
21.53176107
6
0.00147157
2
0.581258883
3
0.900709
3
0.581258883
3
0.900709
4
0.040696527
1
0.84012484
D-52
-------
Supplemental Information ofPFBA and Related Salts
Table D-30. Benchmark dose results for delayed vaginal opening in female
mice—constant variance, 1 standard deviation fDas etal.. 20081
Models
Restriction
1 Standard
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
316.9350
218.4320
0.0106
348.8761
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Exponential 3
(CV—normal)
Restricted
316.9457
218.4320
0.0106
348.8761
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Exponential 4
(CV—normal)
Restricted
35.1705
15.4720
0.6944
341.9320
Viable-
recommended
Lowest AIC
Exponential 5
(CV—normal)
Restricted
34.9991
15.4632
NA
343.9392
Questionable
df = 0, saturated model
(goodness-of-fit p-value
cannot be calculated)
Hill
(CV—normal)
Restricted
35.6204
0.0000
0.8401
341.8184
Unusable
BMD computation
failed; lower limit
includes zero
BMDL not estimated
Polynomial
(3 degree)
(CV—normal)
Restricted
311.4806
211.1287
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Polynomial
(2 degree)
(CV—normal)
Restricted
311.4877
211.1313
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Power
(CV—normal)
Restricted
311.4864
211.1303
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
Linear
(CV—normal)
Unrestricted
311.4866
211.1307
0.0115
348.7113
Questionable
Goodness-of-fit
p-value < 0.1
| Residual at control | >2
D-53
-------
Supplemental Information ofPFBA and Related Salts
D.9. PREPUTIAL SEPARATION—Fi MALE MICE fDAS ET AL.. 20081
Table D-31. Dose-response data for delayed preputial separation
in male mice fDas etal.. 20081
Dose (mg/kg-d)
n
Mean
SD
0
17
29.55
1.14
35
21
30.21
1.99
175
17
30.56
1.84
350
15
31.88
1.72
Table D-32. Benchmark dose results for delayed preputial separation in male
mice—constant variance, BMR = 5% relative deviation (Das et al.. 2008)
Models
Restriction
5% Relative
deviation
p-Value
AIC
BMDS classification
BMDS
notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
254.8183
179.1436
0.6004
277.5960
Viable—alternate
Exponential 3
(CV—normal)
Restricted
254.8005
179.1431
0.6004
277.5960
Viable—recommended
Lowest AIC
Exponential 4
(CV—normal)
Restricted
252.8480
102.0115
0.3080
279.6149
Viable—alternate
Exponential 5
(CV—normal)
Restricted
252.5410
101.9527
0.3076
279.6166
Viable—alternate
Hill
(CV—normal)
Restricted
194.2094
175.4639
0.2286
280.0252
Viable—alternate
Polynomial
(3 degree)
(CV—normal)
Restricted
276.4524
176.5648
0.3427
279.4759
Viable—alternate
Polynomial
(2 degree)
(CV—normal)
Restricted
269.5337
175.9153
0.3268
279.5372
Viable—alternate
Power
(CV—normal)
Restricted
252.7648
175.1179
0.5950
277.6140
Viable—alternate
Linear
(CV—normal)
Unrestricted
252.7653
175.1182
0.5950
277.6140
Viable—alternate
D-54
-------
Supplemental Information ofPFBA and Related Salts
Frequentist Exponential Degree 3 Model with BMR of 0.05 Rel, Dev. for the BMD and
0.95 Lower Confidence Limit for the BMDL
34
33
32
— tstimated Probability
Response at BMD
O Data
— BMD
BMDL
Figure D-7. Dose response curve for the Exponential 3 model fit to delayed
preputial separation in male mice fDas et al.. 20081.
D-55
-------
Supplemental Information ofPFBA and Related Salts
User Input
Info
Model
frequentist Exponential degree 3 vl.l
Dataset Name
Das PS litter SDs
User notes
[Add user notes here]
Dose-Response Model
M[dose] = a * exp(±l * (b * dose)Ad)
Variance Model
Var[i] = alpha
Model Options
BMR Type
Rel. Dev.
BMRF
0.05
Tail Probability
-
Confidence Level
0.95
Distribution Type
Normal
Variance Type
Constant
Model Data
Dependent Variable
[Dose]
Independent Variable
[Mean]
Total # of Observations
4
Adverse Direction
Automatic
D-56
-------
Supplemental Information ofPFBA and Related Salts
Model Results
Benchmark Dose
BMD
254.8005164
BMDL
179.1431485
BMDU
443.2041287
AIC
277.5960319
Test 4 P-value
0.600364435
D.O.F.
2
Model Parameters
# of Parameters
4
Variable
Estimate
a
29.74458616
b
0.000191484
d
Bounded
log-alpha
1.042066246
Goodness of Fit
Dose
Size
Estimated
Median
Calc'd
Median
Observed
Mean
Estimated
SD
Calc'd SD
Observed
SD
Scaled
Residual
0
17
29.74458616
29.55
29.55
1.68376629
1.14
1.14
-0.47649088
35
21
29.94460185
30.21
30.21
1.68376629
1.99
1.99
0.722313504
175
17
30.75820529
30.56
30.56
1.68376629
1.84
1.84
-0.485353184
350
15
31.80636595
31.88
31.88
1.68376629
1.72
1.72
0.169372344
Likelihoods of Interest
# of
Model
Log Likelihood*
Parameters
AIC
A1
-135.2877975
5
280.575595
A2
-132.4445224
8
280.889045
A3
-135.2877975
5
280.575595
fitted
-135.7980159
3
277.596032
R
-142.6419354
2
289.283871
* Includes additive constant of -64.3257. This constant was not included in the LL derivation prior to BMDS 3.0.
Tests of Interest
-2*Log(Likelihood
Test
Ratio)
Test df
p-value
1
20.39482594
6
0.00235492
2
5.686550161
3
0.12789698
3
5.686550161
3
0.12789698
4
1.020436835
2
0.60036443
D-57
-------
Supplemental Information ofPFBA and Related Salts
Table D-33. Benchmark dose results for delayed preputial separation in male
mice—constant variance, BMR = 1 standard deviation fDas et al.. 20081
Models
Restriction
1 Standard deviation
p-Value
AIC
BMDS classification
BMDS
notes
BMD
BMDL
Constant variance
Exponential 2
(CV—normal)
Restricted
287.5467
201.6707
0.6004
277.5960
Viable—alternate
Exponential 3
(CV—normal)
Restricted
287.5612
201.6697
0.6004
277.5960
Viable—recommended
Lowest AIC
Exponential 4
(CV—normal)
Restricted
286.3951
198.7931
0.3080
279.6149
Viable—alternate
Exponential 5
(CV—normal)
Restricted
286.1679
197.6553
0.3076
279.6166
Viable—alternate
Hill
(CV—normal)
Restricted
201.3711
94.7311
0.2286
280.0252
Viable—alternate
Polynomial
(3 degree)
(CV—normal)
Restricted
302.3780
199.5688
0.3427
279.4759
Viable—alternate
Polynomial
(2 degree)
(CV—normal)
Restricted
297.6581
198.8516
0.3268
279.5372
Viable—alternate
Power
(CV—normal)
Restricted
286.2526
197.9759
0.5950
277.6140
Viable—alternate
Linear
(CV—normal)
Unrestricted
286.2531
197.9763
0.5950
277.6140
Viable—alternate
D.10. RELATIVE LIVER WEIGHT-MALE HUMANIZED PPARA MICE
fFOREMAN ET AL.. 20091
Table D-34. Dose-response data for relative liver weight in male
humanized PPARa mice fForeman et al.. 20091
Dose (mg/kg-d)
n
Mean
SD
0
10
4.07
0.261
35
10
5.62
0.719
175
10
6.65
0.784
350
10
7.38
0.719
D-58
-------
Supplemental Information ofPFBA and Related Salts
Table D-35. Benchmark dose results for relative liver weight in male
humanized PPARa mice —nonconstant variance, BMR = 10% relative
deviation fForeman et al.. 20091
Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Exponential 2
(NCV—normal)
Restricted
77.3820
62.7400
<0.0001
107.4138
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
Exponential 3
(NCV—normal)
Restricted
77.3912
62.7399
<0.0001
107.4138
Questionable
Goodness-of-fit
p-value <0.1
| Residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
Exponential 4
(NCV—normal)
Restricted
6.7656
4.8076
0.0951
80.0462
Questionable
Goodness-of-fit
p-value <0.1
BMD 3x lower than
lowest nonzero dose
BMDL 3x lower than
lowest nonzero dose
Exponential 5
(NCV—normal)
Restricted
6.7678
4.8076
0.0951
80.0462
Questionable
Goodness-of-fit
p-value <0.1
BMD 3x lower than
lowest nonzero dose
BMDL 3x lower than
lowest nonzero dose
Hill
(NCV—normal)
Restricted
5.4945
4.4070
0.2883
78.3878
Viable-
recommended
Lowest AIC
BMD 3x lower than
lowest nonzero dose
BMDL 3x lower than
lowest nonzero dose
Polynomial
(3 degree)
(NCV—normal)
Restricted
59.5695
46.0032
<0.0001
104.4698
Questionable
Goodness-of-fit
p-value <0.1
| residual for dose
group near BMD| >2
| residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
D-59
-------
Supplemental Information ofPFBA and Related Salts
Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Polynomial
(2 degree)
(NCV—normal)
Restricted
59.5723
46.0033
<0.0001
104.4698
Questionable
Goodness-of-fit
p-value <0.1
| residual for dose
group near BMD| >2
| residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
Power
(NCV—normal)
Restricted
59.5691
46.0034
<0.0001
104.4698
Questionable
Goodness-of-fit
p-value <0.1
| residual for dose
group near BMD| >2
| residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
Linear
(NCV—normal)
Unrestricted
59.5725
46.0031
<0.0001
104.4698
Questionable
Goodness-of-fit
p-value <0.1
| residual for dose
group near BMD| >2
| residual at control |
>2
Modeled control
response SD >| 1.51
actual response SD
D-60
-------
Supplemental Information ofPFBA and Related Salts
D.ll. RELATIVE LIVER WEIGHT-MALE RATS EXPOSED 28 DAYS
fBUTENHOFF ET AL.. 2012: VAN OTTERDIIK. 2007B1
Table D-36. Dose-response data for relative liver weight in male rats
following 28 day exposure (Butenhoff et al.. 2012: van Otterdijk. 2007b)
Dose (mg/kg-d)
n
Mean
SD
0
10
2.42
0.17
1.2
10
2.55
0.25
6
10
3
0.33
30
10
3.59
0.46
Table D-37. Benchmark dose results for relative liver weight in male
rats exposed 28 days—nonconstant variance, BMR = 10% relative deviation
(Butenhoff et al.. 2012: van Otterdijk. 2007b)
(Das et al..
2008) Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Exponential 2
(NCV—normal)
Restricted
39.0522
30.9899
0.0010
30.9052
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
Modeled control
response std.
dev. >| 1.51 actual
response std. dev.
Exponential 3
(NCV—normal)
Restricted
39.0519
30.9899
0.0010
30.9052
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
Modeled control
response std.
dev. >| 1.51 actual
response std. dev.
Exponential 4
(NCV—normal)
Restricted
9.9467
6.3433
0.9596
19.1475
Viable -
Recommended
Lowest AIC
Exponential 5
(NCV—normal)
Restricted
10.1350
6.3447
NA
21.1450
Questionable
d.f.=0, saturated
model (Goodness of fit
test cannot be
calculated)
D-61
-------
Supplemental Information ofPFBA and Related Salts
(Das et al.,
2008) Models
Restriction
10% Relative
deviation
p-Value
AIC
BMDS
classification
BMDS notes
BMD
BMDL
Nonconstant variance
Hill
(NCV—normal)
Restricted
9.9219
5.3433
NA
21.1450
Questionable
d.f.=0, saturated
model (Goodness of fit
test cannot be
calculated)
Polynomial
(3 degree)
(NCV—normal)
Restricted
31.8784
23.5467
0.0028
28.8760
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
Polynomial
(2 degree)
(NCV—normal)
Restricted
31.8784
23.5468
0.0028
28.8760
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
Power
(NCV—normal)
Restricted
31.8784
23.5470
0.0028
28.8760
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
Linear
(NCV—normal)
Unrestricted
31.8784
23.5468
0.0028
28.8760
Questionable
Goodness of fit p-
value < 0.1
| Residual for Dose
Group Near BMD| >2
D-62
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX E. SUMMARY OF PUBLIC AND
EXTERNAL PEER REVIEW COMMENTS AND EPA'S
DISPOSITION
The Toxicological Review of Perfluorobutanoic Acid and Related Salts was released for
public comment in August 2021. Public comments on the assessment were submitted to the U.S.
Environmental Protection Agency (EPA) by November 8, 2021. The Toxicological Review has also
undergone a formal external peer review in accordance with U.S. Environmental Protection Agency
(EPA) guidance on peer review fU.S. EPA. 20151. A public, external peer-review meeting was held
February 22 and 23, 2022, which included another opportunity for public comment. The external
peer reviewers were tasked with providing written answers to general questions on the overall
assessment approach, key conclusions, and areas of scientific controversy or uncertainty. A
summary of comments made by the external peer reviewers and public commenters, as well as
EPA's responses to these comments, are arranged by charge question as follows. In many cases, the
comments of the individual external reviewers have been synthesized and paraphrased for brevity
(please consult the final peer review report for the full text of the panel's comments: Peer Review
Report! External Peer Reviewers were asked to prioritize their comments to indicate their relative
importance. The prioritization instructions are duplicated below from the IRIS PFBA charge
questions to the peer reviewers, which can be found in the public EPA docket: EPA-HO-QRD-2020-
0675:
Tier 1: Recommended Revisions - Key major recommendations necessary for strengthening
the scientific basis for the Toxicological Review ofPFBA. The implication of such key Tier 1
recommendations is that the assessment conclusions are not adequately supported without
addressing the recommendations and need to be reconsidered or better substantiated. For
Tier 1 recommendations, please describe the specific revisions necessary to modify or
better substantiate the most scientifically appropriate assessment conclusions.
Tier 2: Suggestions - Recommendations that are encouraged to strengthen the scientific
analyses and conclusions in the Toxicological Review ofPFBA. That other factor
(e.g., timeliness) also may also be considered before deciding to address or incorporate
Tier 2 suggestions is understood. For Tier 2 recommendations, please provide specific
suggestions to strengthen the scientific basis for assessment conclusions or improve the
clarity of the analyses and presentation.
Tier 3: Future Considerations - Scientific exploration that might inform future work. These
recommendations are outside the immediate scope or needs of the current document under
review but could inform future toxicological reviews or research efforts
E-l
-------
Supplemental Information ofPFBA and Related Salts
Appendix E lists all Tier 1 recommendations and Tier 2 Suggestions from the external peer
reviewers organized by charge question. For Tier 3 Considerations, please refer to the external
peer review report linked above. Where public comments were made on topics raised by the
external peer reviewers, they are noted alongside the external peer review comments. All Tier 1
recommendations were implemented in this revised assessment, either through revision or
addition to the peer reviewed analyses or text. Tier 2 suggestions were considered in light of
the extent to which those suggestions would impact the conclusions or quantitative analyses of
the assessment, consistency across panelists in raising the suggestion, and the level of effort to
implement For this assessment, all Tier 2 suggestions deemed to be impactful to the toxicity
value conclusions were implemented in this revised assessment Additional public comments
not raised by the peer reviewers are included in a separate section at the end of each charge
question section. Where possible, the public comments have been reproduced in this Appendix
as they were submitted, but in some cases have synthesized and paraphrased for brevity. A
summary document collating all public comments was provided as a courtesy to the external
peer review panel. Please see docket EPA-HO-QRD-2020-0675 for both this summary
document and the full text of the submitted public comments.
External peer reviewer and public comments regarding requests for additions of clarifying text
or editorial or grammatical corrections have been made throughout the assessment as
appropriate; these comments and responses have not been tracked in this Appendix.
E.l. CHARGE QUESTION 1 - SYSTEMATIC REVIEW
The Toxicological Review describes and applies a systematic review process for identifying and
screening pertinent studies that is described in detail in Section 1.2.1 (Literature Search and
Screening) and Appendix A (Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA
IRIS Assessments). Please comment on whether the search strategy and screening criteria for PFBA
are appropriate and clearly described. Please identify additional peer-reviewed studies ofPFBA that
the assessment should incorporate2.
E.l.l. Overarching External Peer Reviewer Comments on Systematic Review
"All reviewers agreed that the literature search was well done, noting that it was
comprehensive and that the methods used were appropriate and clearly described. They also
stressed how challenging it is to conduct a thorough literature search in such a rapidly evolving
field, where information may be out of date in a matter of months or even weeks."
2 Newly identified studies (i.e., studies identified by EPA or the public that meet PECO criteria but were not
addressed in the external review draft, for example due to recent publication) will be characterized by EPA and
presented to the peer review panel. This characterization will focus on EPA's judgment of whether the studies
would have a material impact on the conclusions (i.e., identified hazards or toxicity values) in the external review
draft. The peer review panel is asked to review EPA's characterization and provide tiered recommendations to EPA
regarding which studies, if any, to incorporate into the assessment before finalizing.
E-2
-------
Supplemental Information ofPFBA and Related Salts
E.1.2. Tier 1 Recommendations
Comment: EPA should clarify when and how papers identified from the related systematic
reviews for the other PFAS compounds were included in the PFBA toxicological review. EPA
could provide this clarification by adding a small section and/or a table describing how the
health effects text in the PFBA report was similar or was supported by the application of
information from the review of related PFAS compounds.
EPA Response: Section 1.2.1 "Literature Search and Screening" describes the identification
of studies during problem formulation, scoping and title/abstract screening for other PFAS
that are relevant to the PFBA toxicological review. Specifically, some studies relevant to
PFBA were identified by searches focused on the other four PFAS currently being assessed
by the Integrated Risk Information System (IRIS) Program (i.e., PFHxA, PFHxS, PFNA, and
PFDA) or from other authoritative reviews (e.g., final EPA reviews). This mostly applied to
epidemiological studies as animal and mechanistic studies on specific PFAS are better
indexed by specific PFAS. In addition, Table 4-2 has been added to Section 4.1 to
demonstrate similarities and differences (contingent on the availability of data) in the
health effects observed across the EPA PFAS human health assessments published at the
time of finalization of the PFBA assessment As EPA finalizes more PFAS assessments, this
table will be expanded in subsequent IRIS assessments.
Comment: Multiple reviewers recommended EPA update the literature search to include
the most up-to-date set of studies. Specifically, one reviewer recommended that EPA
incorporate the Weatherly etal. (2021) study before finalizing the Toxicological Review.
Multiple public comments were also received recommending that EPA update its literature
search and incorporate relevant studies (including Weatherly et al., 2021). Public comments
also recommended that EPA explicitly state the date of the last literature search used for the
Toxicological Review.
EPA Response: The date of the last literature search used for the Toxicological Review
(April 2021) was added to Section 2.1. Updates to the literature incorporated into the public
comment draft (after the last literature update) are reflected in a separate document posted
to the docket ("EPA-HO-ORD-2020-0675-0022") and provided to the peer reviewers. This
document describes the consideration of the studies deemed relevant based on the methods
laid out in the protocol and documents the justification for the subset of those incorporated
into the revised assessment. A specific charge question was posed to the peer reviewers on
these decisions and no disagreements were noted in the panel's final comments. The
Weatherly et al. (2021) study was added to the assessment given it was specifically
identified by the peer review panel (see Sections 3.2.2, 3.2.5, and 5.2.1).
E-3
-------
Supplemental Information ofPFBA and Related Salts
Comment: To improve access to the studies identified, one reviewer recommended that
EPA (1) develop a simple table explicitly listing all the in vitro, in silico, or nonmammalian
model "supplemental material" studies that were considered and selected, and (2) develop
a simple table listing all the studies that were considered but not selected, that also briefly
identifies the reasons for rejecting each of these studies. Public comments were also
received requesting EPA make available the lists of included and supplemental studies and
to ensure that the list of studies in HAWC are accurate.
EPA Response: The included, excluded and supplemental studies can be found in HERO
(https://hero.epa.gov/hero/index.cfm/proiect/page/search/true/isws/false/proiect id/26
32 /1. With respect to inclusion or exclusion, studies are excluded if they do not meet all
PECO criteria. During screening, most studies are excluded because they do not meet any or
only meet a few of the PECO criteria. Thus, a single screened out study typically has multiple
reasons for exclusion which is unwieldy to document, especially at the title and abstract
level when screening may be needed for thousands of studies. The annotation used in the
assessment is consistent with the convention in systematic review (Page etal.. 20211. Note
also that multiple tags may be applied to a single study (e.g., tagged "supplemental" during
title/abstract and "in vitro" during full-text screening) which results in potential
discrepancies when cross referencing numbers between HERO and the literature flow
diagrams or HAWC study evaluation heatmaps.
E.1.3. Tier 2 Suggestions
Comment: EPA should add an introductory preview on how it is approaching PFBA in
relation to other forms of PFAS that have been much more extensively studied.
EPA Response: Text was added to Section 1.2.1 "Literature Search and Screening"
explaining that relevant literature on PFBA was identified by searches focused on the other
four PFAS currently being assessed by the Integrated Risk Information System (IRIS)
Program (i.e., PFHxA, PFHxS, PFNA, and PFDA) or from other authoritative reviews (e.g.,
final EPA reviews). Table 4-2 has been added to Section 4.1 to demonstrate similarities and
differences (contingent on the availability of data) in the health effects observed across the
EPA PFAS human health risk assessments published at the time of finalization of the PFBA
assessment As EPA finalizes more PFAS assessments, this table will be expanded in
subsequent IRIS assessments.
Comment: One reviewer suggested, as an enhancement to the added table (listing all the in
vitro, in silico, or nonmammalian model "supplemental material" studies) recommended in
the Tier 1 recommendation above, EPA should incorporate columns (a) summarizing
qualitatively the confidence (low, medium, high) associated with the information presented
E-4
-------
Supplemental Information ofPFBA and Related Salts
in each study, and (b) listing potential outcomes associated with the "supplemental
material" studies.
EPA Response: Confidence in studies meeting PECO criteria is documented in HAWC
fhttps: //hawc.epa.gov/study/assessment/100500073/1. and several additions were made
in this revised assessment to ensure that the health outcome-specific confidence for these
studies is conveyed in the figures or text of the relevant synthesis sections. Studies that are
identified as potentially relevant supplemental material (e.g., mechanistic,
pharmacokinetic) can be found in HERO.
Comment: Numerical inconsistencies in the number of studies listed in Figures 2-1, 2-2,
and 2-3, the number of studies discussed in the text of the Toxicological Review, and the
number of studies listed in HAWC should be corrected (a Tier 1 recommendation on this
topic was also provided under Charge Question 2 below). Public comments were also
received recommending that EPA ensure that all numbers of studies are properly reported
within the document, figures, tables, and associated meta-data.
EPA Response: Eight epidemiology studies and seven animal studies were identified as
meeting the PECO criteria following full-text review. Nine animal studies are listed in Figure
2-1 which includes the Butenhoff et al. (20121 study which reported the findings of two
unpublished industry reports: a 28-day and 90-day gavage study fully reported in (van
Otterdiik. 2007a. b). These industry reports were conducted at the same facility and largely
by the same staff but independently of one another and at different times: July 26, 2006,
through September 15, 2006, for the 28-day study and April 5, 2007, through August 6,
2007, for the 90-day study. Throughout the Toxicological Review, both fButenhoff et al..
20121 and the relevant industry report are cited when discussing effects observed in these
reports. However, only one study evaluation was performed for this group of citations in
HAWC (see Figure 2-2), the overall confidence level of high applies to both the 28-day and
90-day reports and grouping of these studies accounts for the discrepancy between the
number of animal studies in Figures 2-1 and 2-2.
Comment: EPA could add a statement about what kind of information would be required
to change the overall analysis/conclusions, with a clearer description of when updates will
be made.
EPA Response: With respect to information to change overall analysis or conclusions, this
is implicit in the evidence synthesis and integration analysis. For example, conclusions of
"evidence inadequate" are reached after describing specific limitations to the evidence base.
These limitations can be translated by the research community into information gaps that, if
filled, could potentially change an overall analysis or conclusion. Presentation of uncertainty
E-5
-------
Supplemental Information ofPFBA and Related Salts
factors is another area in the IRIS assessment that provides an indication of information
that could be impactful to change an overall analysis/conclusion. Once the IRIS
Toxicological Review of PFBA is finalized, the IRIS program has no immediate plans on
updating the assessment. Given the finite resources of the IRIS Program, IRIS assessment
activities are based on the priority needs of EPA National Program and Regional Offices
identified through a structured internal (to EPA) nomination process.
Comment: EPA could consider providing a brief discussion of what is known (and not
known) to help inform animal-to-human extrapolation. For example, the relevance of PPAR
as a mode of action is an important point and the degree to which it is or is not relevant to
humans could be mentioned at the outset Similarly, the dramatic sex differences in some
rodents are clearly not applicable to humans for other forms of PFAS and presumably not
for PFBA either.
EPA Response: The human relevance of the animal data is explicitly addressed within the
context of the evidence available to inform each individual hazard. However, without
specific evidence to the contrary, effects in animals are presumed relevant to humans (U.S.
EPA. 2005.1998.1991). Once a determination is made that an effect is considered relevant
to humans using the currently available evidence, the quantitative implications of the
remaining uncertainties in extrapolation are addressed through dosimetric adjustment and
application of the UFa during dose-response analysis.
Comment: EPA could consider adding a section that discusses available information on
PFBA's potential immunomodulation (immunosuppression) effects. Existing studies most
probably cannot support derivation of relevant reference values, but compilation and
evaluation of the available information can provide an initial framework for addressing this
challenge in future revisions.
EPA Response: A discussion of potential immunomodulation was added to Section 3.2.5
"Other Non-Health Effects."
Comment: PFAS information submitted to IRIS should be available to all EPA programs
and vice versa.
EPA Response: IRIS assessments rely on publicly available information in the published
literature and can potentially include information submitted to EPA programs (e.g., TSCA), if
those data can be made publicly available.
E-6
-------
Supplemental Information ofPFBA and Related Salts
Comment: EPA should correct the PECO element Exposures statement that incorrectly
suggests that the 6:2fluorotelomer is metabolized into two analytes; it is in fact multiple
analytes.
EPA Response: The IRIS PFAS protocol states (text of interest underlined) the following:
"[Note: although while these PFAS are not metabolized or transformed in the body, there
are precursor compounds known to be biotransformed to a PFAS of interest; for example,
6:2 fluorotelomer alcohol is metabolized to PFHxA and PFBA (Russell etal.. 20151. Thus,
studies of precursor PFAS that identify and quantify a PFAS of interest will be tracked as
potential supplemental material (e.g., for ADME analyses or interpretations)]". This text
does not preclude that this compound can be metabolized to other analytes; it is simply
emphasizing those analytes of interest to the protocol, namely PFBA, PFHxA, PFHxS, PFNA,
or PFDA. However, a small editorial change to the text was made for clarity.
E.1.4. Public Comments
Comment: EPA improperly excluded its own relevant studies (i.e., Das et al., 2008) in
developing the draft IRIS review. In the Draft IRIS Review, EPA made numerous
comparisons between PFBA, a four carbon perfluoroalkyl carboxylate, and
perfluorobutanesulfonate (PFBS), a four-carbon perfluoroalkyl sulfonate congener. While
both PFBA and PFBS have generally been considered short-chain PFAS compounds given
the relatively short serum elimination half-lives in the species evaluated (rodents, non-
human primates, and humans), EPA leaned too heavily on this similarity and ignored
relevant data available for PFBA itself.
EPA Response: This comment implies that the draft assessment did not include the Das et
al. f20081 mouse developmental toxicity study in the PBFA assessment This is incorrect.
Section 5 indicates that multiple effects from Das etal. f20081 were modeled and
considered for RfD derivation.
Additionally, the consideration of data regarding the toxicological effects of PFBS in the
PFBA Toxicological Review is consistent with methods described in Appendix A (Protocol)
supporting the consideration of data on similar chemicals to inform PFBA-specific data
gaps. Thus, PFBS data is described in the PFBA assessment in cases where data was lacking
for PFBA or when drawing parallels between chemicals was useful in discussing potentially
consistent toxicological effects across PFAS. That said, PFBS data were not used in the PFBA
assessment for quantitative purposes or toxicity value derivation, nor were they necessary
to draw the evidence integration conclusions regarding PFBA, which were sufficiently
supported by PFBA-specific data.
E-7
-------
Supplemental Information ofPFBA and Related Salts
Comment: In analyzing the potential health effects of PFBA, the draft assessment makes
several comparisons to data available for other PFAS. While such a "read-across" approach
can be a useful in qualitatively assessing the potential for a compound to impact health
endpoints, it is important that the comparison be made to compounds that are as
structurally similar to the compound of interest as possible and, when appropriate, to
indicate how structural differences may impact ability to compare the toxicokinetics, target
organs, critical effects, potential molecular targets, and shapes of the dose-response curves.
In the case ofPFBA, other short-chain (<8 carbons) carboxylates are the most appropriate
for comparison. Consequently, comparison to perfluorohexanoic acid (PFHxA) would be
more appropriate than to a sulfonic acid like perfluorobutyl sulfonate (PFBS). In this regard,
the lack of observed significant developmental effects associated with PFHxA is noted.
EPA Response: As noted above, the IRIS PFBA assessment considers the data available on
other PFAS to inform PFBA-specific data gaps, with an emphasis on and increased use of
data and judgments on PFAS for which EPA has available a final assessment at the time of
developing the PFBA assessment. It is important to emphasize that EPA has not reached a
final conclusion on whether the evidence supports a "lack of observed significant
developmental effects associated with PFHxA." EPA's IRIS PFHxA assessment is still under
development
Comment: EPA should provide further clarification and better reporting when multiple
publications of the same data are included. For example, the studies reported as van
Otterdijk 2007c and van Otterdijk 2007d are industry documents available in EPA's HERO
database but have also been published in the peer reviewed literature in the study by
Butenhoff etal. 2012. That these studies contain overlapping and duplicative data, should
be more clearly noted in the literature flow diagram (Figure 2-1) and the discussion of
Study Evaluation Results in Section 2.2.
EPA Response: This is noted in Section 3.2.1 "Non-Cancer Evidence Synthesis and
Integration" (footnote 11). Additional clarification was also added to Study Evaluation
Results in Section 2.2: "The studies meeting PECO criteria at the full-text level included
eight epidemiological studies, nine animal studies (including one published study
[Butenhoff et al. (2012)] that reported on the same data in two unpublished industry
reports [fvan Otterdiik. 2007al and fvan Otterdiik. 2007bl]."
E.2. CHARGE QUESTION 2 - STUDY EVALUATION
The Toxicological Review describes the results of the evaluations of individual studies in Section 2.2
(Study Evaluation Results) and presents and analyzes the findings from those studies deemed
informative in the relevant health effect-specific synthesis sections.
E-8
-------
Supplemental Information ofPFBA and Related Salts
a. Please comment on whether the study confidence conclusions for the PFBA studies are
scientifically justifiedgiving appropriate consideration to important methodological features
of the assessed outcomes. Please specify any study confidence conclusions that are not justified
and explain any alternative study evaluation decisions.
b. Results from individual PFBA studies are presented and synthesized in the health system-
specific sections. Please comment on whether the presentation and analysis of study results is
clear, appropriate, and effective to allow for scientifically supported syntheses of the findings
across sets of studies.
E.2.1. Overarching External Peer Reviewer Comments on Study Evaluation
"Reviewers agreed that the confidence conclusions were scientifically justified, and that
Section 2.2 was well done. Reviewers noted that the scientific justification presented was clear and
effective, and found that the interactive visualizations provided a convenient overview. [One
reviewer] also found it very beneficial that EPA presented the logic for giving more or less attention
to particular studies and specific outcomes."
E.2.2. Tier 1 Recommendations
Comment: EPA should correct numerical inconsistencies in the number of studies listed in
the Figures 2-1, 2-2, and 2-3, the number of studies discussed in the text of the Toxicological
Review, and the number of studies listed in HAWC. (a Tier 2 suggestion on this topic was
also provided under Charge Question 1 above).
EPA Response: See EPA Response to Charge Question 1 above and referenced in footnote
11 of the Toxicological Review which states "Eight epidemiology studies and seven animal
studies were identified as PECO relevant following full-text review. Nine animal studies are
listed in Figures 2-1 which includes the Butenhoff et al. f20121 study which reported the
findings of two unpublished industry reports: a 28-day and 90-day gavage study fully
reported in fvan Otterdiik. 2007a. b). These industry reports were conducted at the same
facility and largely by the same staff but independently of one another and at different
times: July 26, 2006, through September 15, 2006, for the 28-day study and April 5, 2007,
through August 6, 2007, for the 90-day study. Throughout the Toxicological Review, both
Butenhoff et al. f20121 and the relevant industry report are cited when discussing effects
observed in these reports. However, only one study evaluation was performed for this
group of citations in HAWC (see Figure 2-2), the overall confidence level of high applies to
both the 28-day and 90-day reports and accounts for the discrepancy between nine animal
studies in Figures 2-1 and 2-2."
E.2.3. Tier 2 Suggestions
Comment: Comparison of effects in males and females is a common theme in the document
and sometimes tentatively related to the pharmacokinetic differences ofPFBA in females
E-9
-------
Supplemental Information ofPFBA and Related Salts
(more rapid clearance) than males (slower clearance). EPA could select an administered
dose in males that is nearly equivalent to females based on pharmacokinetic dose metrics
and compare toxicity outcomes, as it may provide a clearer picture of the role that
pharmacokinetics plays in toxicity in male and female lab animals.
EPA Response: A brief comparison and discussion of pharmacokinetic parameters and
relative responses for liver and thyroid endpoints at a constant dose has been added to
Section 4.3 (Conclusions Regarding Susceptible Populations and Lifestages).
Comment: While the information on the available epidemiologic studies is provided in the
HAWC table, EPA could provide text that briefly notes the nature of the studies in general,
and since these studies are so few in number, a sentence or two about each. It could be as
little as 2-3 sentences that gives a synopsis of what was done (the PECO attributes), key
methodologic limitations, and overall assignment regarding its quality. This is editorial in
nature but would help make this a more transparent, user-friendly document Those who
want the full details could look to HAWC and appendices as needed.
EPA Response: Confidence in studies meeting PECO criteria is documented in HAWC
fhttps://hawc.epa.gov/assessment/100500073/). and several additions were made in this
revised assessment to ensure that the health outcome-specific confidence for these studies
is conveyed in the figures or text of the relevant synthesis sections. Because study
evaluations in IRIS are outcome-specific, the revised draft assessment does not include text
summarizing the studies and their strengths/limitations in an outcome-nonspecific manner.
Further, the health outcome-specific syntheses attempt to distill the available
(epidemiological or other) evidence to those aspects of the studies (e.g., design; confidence)
most pertinent to drawing hazard judgments; they intentionally try to avoid study-by-study
summaries. This focus on developing concise IRIS assessments is based on feedback from
reviewers over many years.
Comment: The Toxicological Review states that most of the animal studies evaluated for
study confidence were adequate, but the specific rationale for the analyses was unclear
because the interactive HAWC link did not work. Ultimately it was unclear how the study
confidence conclusions were determined, other than professional judgement. This chapter
could be improved by writing more about the overall confidence conclusions in the chapter,
even if the information is found in links.
EPA Response: The HAWC site is now public
fhttps: //hawc.epa.gov/assessment/100500073 /1. Links were checked and fixed
throughout the revised assessment, and the study confidence conclusions are detailed
there. Please see responses above about text additions relating to study confidence ratings.
E-10
-------
Supplemental Information ofPFBA and Related Salts
E.3. CHARGE QUESTIONS 3 AND 4 - HEPATIC EFFECTS
For each health effect considered in the assessment and outlined below, please comment on whether
the available data have been clearly and appropriately synthesized to describe the strengths and
limitations. For each, please also comment on whether the weight-of-evidence decisions for hazard
identification have been clearly described and scientifically justified.
• For hepatic effects; the Toxicological Review concludes that the available evidence indicates
PFBA exposure is likely to cause hepatic effects in humans given relevant exposure
circumstances; on the basis of a series of short-term, subchronic, and developmental studies in
rats and mice demonstrating consistent and coherent effects with a clear biological gradient.
Although the available mechanistic information indicates the effects in rodents are relevant to
humans, some uncertainty remains regarding potential differences in sensitivity across species
due to evidence for the involvement of both PPARa-dependent and PPARa-independent
pathways in these effects (see Charge Question 4 requesting input specific to this latter
uncertainty).
Appendix A (Systematic Review Protocol for the PFBA PFHxA, PFHxS, PFNA, and PFDA IRIS
Assessments) identifies the human relevance of hepatic effects in animals that involve peroxisome
proliferator-activated receptor alpha (PPARa) receptors as a key science issue. To the extent
supported by the PFBA literature (and to a lesser extent, literature for other PFAS), the
Toxicological Review evaluates the evidence relevant to the potential involvement of PPARa and
non-PPARa pathways with respect to the reported hepatic effects. The Toxicological Review
ultimately concludes evidence from in vivo and in vitro studies support that multiple modes of
action (MOA) are operant in the induction of hepatic effects by PFBA exposure and the relative
contribution of these different MOAs cannot be concluded with confidence from the available data.
Please comment on whether the available animal and mechanistic studies support this conclusion
and whether the analysis presented in the Toxicological Review is clearly documented.
E.3.1. Overarching External Peer Reviewer Comments on Hepatic Effects
"All reviewers agreed that the PFBA document clearly and appropriately synthesizes
available data to describe the strengths and limitations of hepatic effects. [One reviewer]
commented that "The evidence integration section is well done and supported by a great summary
in Table 3-8." All reviewers also found the document supported the conclusion that MOAs are
operant in the induction of hepatic effects by PFBA exposures, and the relative contribution of these
different MOAs cannot be concluded with confidence. [The same reviewer] commented that "EPA
did a great job in describing the complications" of the data, and another stated that "the available
animal and mechanistic studies are clearly documented, and the conclusion is supported.""
E.3.2. Tier 1 Recommendations
Reviewers had no Tier 1 recommendations.
E.3.3. Tier 2 Suggestions
Comment: EPA could consider adding tables and/or figures that would help readers
visualize important EPA conclusions, such as coherence of liver histopathology with liver
E-ll
-------
Supplemental Information ofPFBA and Related Salts
weight effects (cited in Table 3-8), since these results are only presented in separate tables
within the document (i.e., Table 3-7/Figure 3-6 for histopathology and Table 3-6/Figure 3-5
for liver weight changes).
EPA Response: A data pivot chart has been created in HAWC displaying liver
histopathology and liver weight effects together for ease of assessing the consistency and
coherence of liver effects across species, sexes, durations of exposure, and study types. A
link to this chart has been added to Section 3.2.2 of the assessment.
Comment: The EPA should resolve an apparent discrepancy where statements regarding
serum biomarker data are incoherent. Observations for serum biomarkers of altered liver
function or injury appear as a factor that decreases certainty in Table 3-8 (p. 3-40) which
contradicts statements in Section 3 that state that the inconsistent serum biomarker results
did not influence the evidence integration judgements.
EPA Response: The statement in Table 3-8 those incoherent observations across serum
biomarkers (e.g., increased ALP but decreased bilirubin) decreased certainty is in reference
to the certainty in the evidence for the serum biomarkers specifically. For the overall
evidence integration judgment about hepatic effects, the noted incoherence across some of
the biomarkers findings was not influential, but rather this judgment was based on the
strong evidence of consistent and coherent effects on liver weights and liver histopathology.
The text in Section 3 has been revised to provide more clarity on exactly how the serum
biomarker evidence was used in drawing the overall evidence integration judgment
Comment: EPA should explicitly state the meaning of "consistent effects" in the sentence:
"The available animal evidence for effects on the liver includes multiple high and medium
confidence studies with consistent effects across multiple species, sexes, exposure
durations, and study designs...". This phrase could have several meanings (e.g., all the same
effects occurred at the same or similar doses across multiple species, sexes, exposure
durations, and study designs; one or more hepatic effects were consistently found at he
some dose across studies though the specific effects observed may vary across studies) and
different readers can/will interpret it differently.
EPA Response: As described in the protocol, 'consistency' generally relates to findings for
a given outcome (e.g., liver weight), while 'coherence' reflects the observed findings across
related (e.g., through biological understanding or MOA) outcomes, such as effect on liver
weight, and separately, histopathology, within individual studies or across multiple studies.
Further, consistency is judged based on the pattern of findings across studies and
comparisons and can range from consistency in the direction of the response to something
more specific such as consistency in the magnitude of change in response at a given
E-12
-------
Supplemental Information ofPFBA and Related Salts
exposure level (while the latter can provide stronger evidence of consistency, it is neither
expected nor required). In the context noted in the comment, the summary statement about
consistency in the evidence integration section reflects the analyses presented in detail
within the preceding synthesis sections on individual liver outcomes. An edit was made to
clarify consistency refers to the findings of increased liver weight and, separately, increased
liver histopathology incidence across the available studies.
Comment: EPA should consider differences in metabolic pathways between species when
comparing rodent to human exposures. The data on exposure obtained from the mouse
used a humanized mouse model. Caution is always required when using data from a
"humanized model" when the humanized model is limited to a single gene replacement As
noted in the public comments, only a single nuclear receptor was humanized (PPARa) and
that there are other nuclear receptors exist in the rat that can be induced and may lead to
hepatocellular hypertrophy. This would not necessarily lead to a similar effect in humans at
similar doses. This Tier 2 comment was also provided under Charge Question 6; responses
to both instances of this Tier 2 comment are provided here for brevity's sake.
EPA Response: Due to reported cross-species differences in PPARa signaling potency and
dynamics, the potential human relevance of some hepatic effects has been questioned.
Thus, the Foreman etal. f20091 study is informative in providing evidence on the relative
contribution that PPARa has on PFBA-induced liver effects. While true that only PPARa was
humanized in the Foreman etal. (20091 study, given the response in the humanized mice,
including mouse PPARa-independent increases in hepatocellular hypertrophy and
vacuolation, Foreman etal. (20091 provides evidence that rodent PPARa is not necessary
for PFBA to induce some liver effects. The assessment indicates the data from Foreman et al.
f20091 is largely used in qualitative analyses; thus, no claim is made that similar effects,
whether in type or magnitude, would be observed in humans as in rats at similar doses. The
mechanistic section discusses the activation and human relevance of other nuclear
receptors (PXR, CAR) that might also contribute to the hepatic effects ofPFBA.
Comment: The one issue that may call for more comment concerns the absence of evidence
that PFBA affects liver enzymes (ALT, etc.) because that is the one human health endpoint
that has consistently been found to be associated with other forms of PFAS. EPA should
consider whether that contrast between an absence of an effect for PFBA in rodents and an
apparent effect for PFOA and PFOS for humans has relevance to the final judgment
EPA Response: The IRIS assessment does not conclude there is no evidence that PFBA
exposure affects liver enzymes. The currently available data for PBFA on this endpoint are
inconclusive. Specifically, only one epidemiology study was available and, although this
study did not find any associations between serum biomarker levels and PFBA exposure, it
E-13
-------
Supplemental Information ofPFBA and Related Salts
was considered deficient with respect to sensitivity and likely biased towards the null.
Additionally, while the two animal studies were also ultimately inconclusive with regard to
ALT and other enzymes, some effects on related biomarkers were observed. Specifically,
while no effects were observed after 28 days of exposure, changes in several serum
biomarkers potentially indicative of liver injury (e.g., increased ALP) were observed after 90
days ofPFBA exposure, along with some incoherent findings (e.g., changes in the opposite
direction as expected for bilirubin). Although uncertain, the serum biomarker data
specifically were inconclusive and thus not necessarily in contrast to the findings for PFOA
and PFOS. Regardless, in the case of PFBA, there is sufficient evidence on liver weight
increases and increased histopathological lesions, as well as data informing mode-of-action,
that the serum biomarker data, including if one were to more explicitly consider the serum
biomarker data for other PFAS, would not change the overall evidence integration judgment
of "evidence indicates (likely)" for PFBA.
Comment: Because the entire Toxicological Report rests on animal-to-human
extrapolation, to the extent that there are mechanisms in animals known not to apply to
humans, this should be explained and factored into the report.
EPA Response: A primary purpose of the Mechanistic Evidence and Supplemental
Information section for hepatic effects is to evaluate the currently available mechanistic
evidence informing whether the hepatic effects observed in rodents are relevant to human
health (and thus suitable as the basis for reference value derivations). It is unclear what
mechanisms in animals the commenter is referring to as "known" not to apply to humans,
but given the focus on PPARa activation for PFAS, this is the mechanism of interest this
response assumes. However, to clarify, activation of PPARa by other compounds can
contribute to toxicity in humans and furthermore PFBA appears to be a ligand for PPARa in
humans, given the data from the humanized PPARa mice and the observed interaction
between PFBA and the human receptor in vitro. The concern is rather one of a differential
response magnitude based on the presumption that hepatic responses to PPARa activation
are exaggerated in rodents as compared to humans and thus may generally be more difficult
to interpret as relating to a significant change in hepatic effects in most human exposure
scenarios. Focusing on this concern, a key science issue addressed by the section noted
above is whether the available data are sufficient to support dependence on PPARa
activation for the observed hepatic effects of PFBA, or whether it is possible that the hepatic
effects can also be mediated through non-PPARa modes-of-action. The other potential
mechanisms discussed in this section are also considered in light of the available evidence
on their human relevance, with a general assumption that animal mechanisms are relevant
to humans without data to the contrary. The final conclusion, which is also supported in the
peer review comments, is that PFBA's liver effects appear to be mediated through both
E-14
-------
Supplemental Information ofPFBA and Related Salts
PPARa-dependent and -independent pathways and that the observed liver effects are
relevant to human health. As this is already extensively discussed and explained (similarly
to the explanations in the above response) in the peer reviewed draft, additional discussion
on this topic was not added to this revised assessment
E.3.4. Public Comments
Comment: EPA improperly characterizes liver hypertrophy in rats as an adverse effect of
PFBA exposure. This conclusion is not consistent with EPA's own guidance, which states
that reported liver hypertrophy is not an adverse effect As noted in the EPA Office of
Pesticide Programs Health Effects Division Guidance Document# G2002.01 on
Hepatocellular Hypertrophy (U.S.EPA, 2002), liver hypertrophy does not necessarily
represent liver toxicity, nor is it necessarily a precursor to a particular outcome of toxicity.
In fact, Guidance Document # G2002.01 recommends a weight-of-evidence approach for
determining whether a liver effect should be characterized as adverse. This approach
includes evaluation of other findings, such as: (1) type and severity of observed effects; (2)
onset, duration, and progression of effects; (3) study method and design; and (4) other
relevant effects and data. This guidance states that liver size or weight changes may be
"indicative of adaptation which, by itself, is not necessarily adverse."
EPA Response: The HED guidance document does not state that "liver hypertrophy is not
an adverse effect" but rather, as the comment itself points out "[hepatocellular
hypertrophy (and its corresponding increased liver size/weight) may be indicative of
adaptation which, by itself, is not necessarily adverse. However¦, it might be associated with
other more severe changes [emphasis added]. These changes are usually accompanied by
alterations in relevant clinical chemistry parameters and/or histopathology." The HED
guidance document recommends a weight-of-evidence approach to determine whether
hypertrophy and related liver weight changes are considered to reflect an adverse hepatic
response. Consistent with the protocol, to judge the adversity of the observed liver effects,
the PFBA IRIS assessment considered the panel recommendations outlined by Hall et al.
(2012) and the HED 2002 Guidance document Ultimately, using this paradigm, the
assessment concluded "application of the recommendations from Hall etal. (2012) supports
the conclusion that the multiple and interconnected effects observed in the livers of
exposed animals meet the criteria for adversity" (i.e., the conclusion of adversity is not
based on liver hypertrophy alone). This conclusion was unanimously supported by the
external peer review panel as noted in the final Peer Review Report: "[a] 11 reviewers agreed
that the PFBA document clearly and appropriately synthesizes available data to describe the
strengths and limitations of hepatic effects."
E-15
-------
Supplemental Information ofPFBA and Related Salts
Comment: The reported liver hypertrophy in rats with PFBA exposure is an adaptive
response occurring through increased activation of PPARa and CAR nuclear receptors. The
activation of PPARa and CAR can lead to expansion of the smooth endoplasmic reticulum in
the liver cell, and this added intracellular mass is reflected in the overall liver weight
macroscopically and hypertrophy microscopically. Another consequence of PPARa and
CAR/PXR activation in rodents is the potential stimulation of cell division (hyperplasia] and
a decrease in the normal process of removal of worn-out cells (apoptosis). These processes
also increase liver mass and can potentially lead to tumor formation in rodents. However, it
is important to note that the hyperplastic processes are not present in mice with humanized
PPARa expression (Foreman et al., 2009). Therefore, the processes that lead to tumor
formation in rats as a result of PFBA exposure are not applicable to humans. In addition,
Foreman et al. also demonstrated that activation of the human form of PPARa with PFBA
does not produce frank liver toxicity. Moreover, Bjork and Wallace (2009) demonstrated
that human hepatocytes did not respond to PFBA-induced PPARa activation at
concentrations up to 200 |j.M (42,600 ppb); rat hepatocytes responded at PFBA
concentration of 25 |j.M (5,325 ppb) and above.
EPA Response: Please see responses above relating to the adversity and human relevance
of the hepatic changes observed in rodents following PFBA exposure, noting that the
assessment does not draw a conclusion regarding the potential for PFBA exposure to lead to
tumor formation (i.e., the available carcinogenicity evidence was judged as inadequate).
E.4. CHARGE QUESTION 3 - THYROID EFFECTS
For each health effect considered in the assessment and outlined below, please comment on whether
the available data have been clearly and appropriately synthesized to describe the strengths and
limitations. For each, please also comment on whether the weight-of-evidence decisions for hazard
identification have been clearly described and scientifically justified.
• For thyroid effects; the Toxicological Review concludes that the available evidence
indicates PFBA exposure is likely to cause thyroid toxicity in humans given relevant
exposure circumstances; primarily on the basis of short-term and subchronic studies in
male rats reporting a consistent and coherent pattern of thyroid effects following PFBA
exposure, but also drawing from the consistency of effects when considering evidence from
structurally related PFAS. The Toxicological Review concludes the thyroid effects are
considered relevant to humans in the absence of evidence to suggest otherwise.
E.4.1. Overarching External Peer Reviewer Comments on Thyroid Effects
"Six of the seven reviewers found the conclusions for thyroid effects to be clearly described
and scientifically justified. For example, [One reviewer] wrote: "The data leading to the conclusion
E-16
-------
Supplemental Information ofPFBA and Related Salts
that the thyroid effects are considered relevant to humans were appropriately synthesized and
their strengths and limitations were adequately described; the weight-of-evidence decisions for
hazard identification have been adequately described; the weight-of-evidence decisions for hazard
identification have been adequately described and justified." The seventh reviewer commented that
there was a "lack of clarity on human exposures.""
E.4.2. Tier 1 Recommendations
Comment: One reviewer recommended that EPA provide a table presenting lab animal and
human citations and data that show connections between a percent decrease in serum T4
and adverse outcome, relating percent change with adversity in humans especially. This
reviewer noted that, "This argument needs to be solid and presenting data will help" to
relate thyroid effects in rats and humans quantitatively.
EPA Response: To our knowledge, there is only limited information from human studies
that demonstrate what percent decrease in T4 leads to adverse outcomes such as
neurodevelopmental outcomes (note: these outcomes are the focus of this response and
discussions in the assessment since these associations are the best studied). This is mainly
due to the nature of epidemiological studies, typically with representative samples analyzed
post hoc; many also bin data by "hypothyroid, euthyroid, hypothyroxinemic" based on
reference ranges, and then correlate to adverse outcomes. There are a few human studies
Tansen etal. (20191: Levie etal. (20181: Korevaar etal. (20161 where the sample sizes are
large enough to capture a wide range of TSH and/or T4 values, which were then correlated
to various neurodevelopmental outcomes that could be quantified. However, these studies
still do not make direct comparisons from a percent decrease in hormones that would lead
to an adverse effect; rather, they stratify their hormone samples by standard deviation to
the mean/median, quartiles, etc. Therefore, it's difficult to make a conclusion in humans
regarding what percent of hormone dysfunction is adverse, as those kinds of data are not
generated. Additionally, there are no conclusive values from animal studies regarding to
what degree of T4 reduction is adverse. This is due to several factors, including the
existence of multiple thyroid-dependent processes in the brain, which likely have differing
spatiotemporal sensitivities. But there are studies that show how graded reductions in T4
can lead to neuronal heterotopia Gilbert etal. f20141. synaptic transmission defects Gilbert
and Sui f20081 and differential gene expression. When unable to estimate the BMR based on
an evidentiary approach, the EPA takes a default approach as outlined in the Benchmark
Dose Technical Guidance (U.S. EPA. 20121. Given the limited and inconsistent information
available from human and animal studies, the assessment uses a standard deviation
definition for the BMR (see Toxicological Review, Section 5.2.1 for justification for using
BMRs). The Mechanistic Support and Supplemental Information section for thyroid effects
E-17
-------
Supplemental Information ofPFBA and Related Salts
has been revised to more thoroughly discuss the nature of the available human and animal
evidence.
E.4.3. Tier 2 Suggestions
Comment: Related to the Tier 1 recommendation above, the same reviewer re-iterated
that the thyroid findings in rats are relevant to humans but proposing how to relate rat
thyroid effects to humans in a quantitative manner (dose-response) is the challenge. This
reviewer suggested a brief discussion on this topic would be useful.
EPA Response: Please see the responses provided to the Tier 1 Recommendation above as
well as the public comment on this topic below.
Comment: One reviewer noted that, although Butenhoff et al. (2012) do not report
statistically significant changes in serum TSH related to PFBA exposure, the coefficient of
variation for TSH measurements in controls ranges from 40% to 72% in the two studies this
group reports. The reviewer suggested that this means that detecting relatively small
changes in serum TSH is difficult with this assay assembled from the reagents provided by
the NIH Pituitary Program and that it is more likely, as pointed out in the Agency report,
that the effect of PFBA exposure on thyroid histology reflects an increase in serum TSH that
was not detected in this assay. This reviewer suggested the EPA consider evaluating the
issue of the TSH assay using the reagents characterized in the Butenhoff studies and rather
than emphasize the comparison with "hypothyroxinemia" (a clinical term that doesn't
translate perfectly to animal studies), describe the situation as one in which serum T4 is low
and TSH levels have increased.
EPA Response: The description of PFBA exposure on thyroid effects now includes
discussion of the potential lack of detectable increases in serum TSH levels with
corresponding decreased T4. Lack of detectable increase in serum TSH levels may be
related to the specific assay utilized in the Butenhoff et al. f20121 study, and text has been
added to the assessment noting this possibility.
Comment: One reviewer suggested that it is important to discriminate between "thyroid
function" and "thyroid hormone action." This reviewer noted that evidence presented in
Butenhoff et al. (2012) shows that PFBA decreases both serum total and free T4 while also
showing an increase in thyroid hormone action in the liver. This reviewer further notes that
this scenario is highly reminiscent of the effects of PCB and PBDE exposure on TH signaling
in which measures of TH action indicate an increase in TH action in the liver despite a
reduction in serum total and free T4 (Giera et al., 2011; Bansal et al., 2014). In the case of
PCBs, this pattern of effects on serum hormones and gene expression in the liver was
coincident with a complex pattern of effects on thyroid hormone action in brain (Zoeller et
E-18
-------
Supplemental Information ofPFBA and Related Salts
al., 2000; Bansal and Zoeller, 2008). Finally, this reviewer noted that Butenhoff etal. 2012
reports thatPFBA exposure also decreases serum cholesterol (Table 7), and thyroid
hormone is known to reduce serum cholesterol in rodents and humans (Mullur et al., 2014).
Ultimately this reviewer suggested that "[although paradoxical ... that PFBA reduces serum
total and free T4 while the liver appears to be responding to increased thyroid hormone
action, it is recommended that the Agency incorporate all measured endpoints of thyroid
hormone action in their analysis."
EPA Response: A brief description of reduced thyroid function (decreased serum total and
free T4) and increased thyroid hormone action in the liver (decreased serum cholesterol)
has been added to Section 3.2.1 (Thyroid Effects).
Comment: One reviewer agreed that decreases in serum total and free T4 are very clearly
associated with developmental and physiological deficits: "In both humans and
experimental animals, low TH is related to permanent neural and cognitive deficits (e.g.,
(Zoeller and Rovet, 2004; Rovet, 2014; Stagnaro-Green and Rovet, 2016a). This is likely to
be true for several (if not all) organs including heart, bone, lung and intestine (Bizzarro and
Gross, 2004; Bassett et al., 2007; Mochizuki etal., 2007; Wexler and Sharretts, 2007). Much
of the experimental literature on this topic makes use of models of severe hypothyroidism
(Crofton et al., 2005; Crofton and Zoeller, 2005). However, graded effects of thyroid
hormone insufficiency have become the focus of increased attention in experimental
systems (Gilbert et al., 2020). This experimental interest reflects the degree to which
subclinical hypothyroidism should be viewed as a disease state (Cooper, 2001). Thus, a
preponderance of information indicates that even a small degree of thyroid hormone
insufficiency is associated with cognitive deficits in children (Haddow et al 1999; Rose et al.,
2006; Nakamizo etal., 2007; Oerbecketal., 2007; Korevaar etal., 2016)." This reviewer
suggested that the EPA document more fully the sensitivity of the human brain to thyroid
hormone insufficiency as it may strengthen the support for this choice.
EPA Response: Additional references which support an association between subclinical
hypothyroidism and cognitive deficits have been added to Section 3.2.1 (Thyroid Effects).
Comment: One reviewer suggested that the EPA could consider a broader context of PFBA
impacts on the thyroid system, including considerations on genetic deficits in specific
proteins related to the thyroid hormone system.
EPA Response: Text was added to Section 3.2.1 (Thyroid Effects) which includes a brief
discussion ofPFBA effects and genetic differences in thyroid hormone system-related
proteins.
E-19
-------
Supplemental Information ofPFBA and Related Salts
E.4.4. Public Comments
Comment: One commenter noted that the significance of changes in T4 levels in rodents to
human risk assessment has been questioned by the National Academy of Sciences (NAS)
and others because of the significant differences in binding proteins and affinities among
species. These differences in binding proteins, binding affinities of the proteins for the
hormones, turnover rates of the hormones, and thyroid stimulation lead to important
quantitative differences between rats and humans. As a result, NAS concluded that "rats are
much more sensitive to agents that disturb thyroid function than are humans, so the
relevance of rat studies in quantitative terms to humans is limited." NAS further noted that
"[t]he committee does not agree that transient changes in serum thyroid hormone or TSH
concentrations are adverse health effects; they are simply biochemical changes that might
precede adverse effects.
EPA Response: As noted above, six of the seven peer reviewers indicated the T4 effects are
significant and relevant to humans. One of the external peer reviewers further specified that
decreases in serum total and free T4 are very clearly associated with developmental and
physiological deficits: "In both humans and experimental animals, low TH is related to
permanent neural and cognitive deficits (e.g., Stagnaro-Green and Rovet f20161: Zoeller and
Rovet (200411. This is likely to be true for several (if not all) organs including heart, bone,
lung, and intestine (Bassett et al. (20071: Mochizuki et al. (20071: Wexler and Sharretts
(20071: Bizzarro and Gross (200411. Much of the experimental literature on this topic makes
use of models of severe hypothyroidism (Croftonand Zoeller. 2005: Crofton etal.. 20051.
However, graded effects of thyroid hormone insufficiency have become the focus of
increased attention in experimental systems f Gilbert etal.. 20201. This experimental
interest reflects the degree to which subclinical hypothyroidism should be viewed as a
disease state fCooper. 20011. Thus, a preponderance of information indicates that even a
small degree of thyroid hormone insufficiency is associated with cognitive deficits in
children (Korevaar etal.. 2016: Nakamizo etal.. 2007: Oerbeck. 2007: Haddowetal.. 19991.
The majority of the panel concluded that the conclusions for the thyroid effects to be clearly
described and scientifically, with another panel member stating that "the data leading to the
conclusion that the thyroid effects are considered relevant to humans were appropriately
synthesized and their strengths and limitations were adequately described." In response to
this comment, however, the assessment has been revised to include a more detailed
discussion on the relationship between thyroid hormone status and neurotoxicological
effects to further support assessment conclusions.
Regarding the NAS (2005) report on the Health Implications of Perchlorate Ingestion, the
conclusions of that report should be considered narrowly in the context of health effects
due to exposure to perchlorate and not generally as broad statements on the human
E-20
-------
Supplemental Information ofPFBA and Related Salts
relevance of thyroid effects observed in animals. The statement that the "committee does
not agree that the transient changes in serum thyroid hormone or TSH concentrations are
adverse health effects; they are simply biochemical changes that might precede adverse
effects" is made regarding what endpoints to base the reference dose value on for
perchlorate given the presumed chemical-specific mode of action. This mode of action is not
pertinent to PFBA or other PFAS.
Comment: EPA failed to consider studies where quantitative histomorphometric analysis
on thyroid function after PFBA exposure did not report statistically significant changes.
Histomorphometric analyses of thyroid follicles provide a more quantitative indication of
thyroid response than histopathological assessments. To conduct a thorough and defensible
risk assessment for PFBA, EPA must consider this and similar studies.
EPA Response: The quantitative histomorphometric analyses from Butenhoff et al. T20121
are discussed in Section 3.2.1 (Thyroid Effects).
E.5. CHARGE QUESTION 3 - DEVELOPMENTAL EFFECTS
For each health effect considered in the assessment and outlined below, please comment on whether
the available data have been clearly and appropriately synthesized to describe the strengths and
limitations. For each, please also comment on whether the weight-of-evidence decisions for hazard
identification have been clearly described and scientifically justified.
• For developmental effects¦, the Toxicological Review concludes that the available evidence
indicates PFBA exposure is likely to cause developmental effects in humans given relevant
exposure circumstances; on the basis of a coherent pattern of delays in acquisition of three
different developmental milestones in a single study in mice, with the findings presumed
relevant to humans in the absence of evidence to suggest otherwise. The assessment
discusses similar effects observed for structurally related PFAS.
E.5.1. Overarching External Peer Reviewer Comments on Developmental Effects
"Most reviewers agreed that the available data were clearly and appropriately synthesized
and concurred with EPA's conclusion. [One reviewer] commented that "in several but not all ways"
the available data on developmental effects were clearly and appropriately synthesized to describe
the strengths and limitations but noted that the document lacked a discussion of "relevant
mechanistic information or information on the conserved biological processes causing the
developmental effects observed in mice to be considered relevant to humans." [The same reviewer]
noted that such discussions "would inevitably lead to a more complete and transparent description
of the strengths and limitations of the available data relevant to developmental effects.""
E-21
-------
Supplemental Information ofPFBA and Related Salts
E.5.2. Tier 1 Recommendations
Comment: One reviewer stated that EPA should add information supporting the human
relevance of the developmental effects to the assessment (e.g., the Evidence Integration
Summary) in order to "... more fully and clearly support applicability of the weight-of-
evidence decision." This reviewer noted that, currently, the text of the assessment lacks
sufficient discussion of "...the conserved biological processes between mice and humans
that the EPA considers relevant to the observed developmental effects (e.g., for delayed
vaginal opening and preputial separation), whether the mouse has been shown to be a good
laboratory animal model for assessing potential human developmental effects, or what
human developmental endpoints (e.g., delayed onset of puberty) may be presumed to be
correlates of some of the PFBA-induced developmental effects observed in the single mouse
study (e.g., delays in vaginal opening and preputial separation in Das et al. 2008)." This
reviewer further noted that some of this information could found later in the document (i.e.,
Section 5.2.1) and this information could be included and expanded on in Section 3. Public
comments were also received regarding whether delays in vaginal opening and preputial
separation whether "these endpoints accurately reflect pubertal development" and stating
that "the biological basis for this assumption is lacking and reviews suggest that these
measurements accurately reflect pubertal development in the rat but not the mouse."
Public comments further stated that "mouse "puberty" has been used for POD calculations
in some of the PFAS documents and I question that this is biologically correct, and the
Agency may want to reconsider this. Clearly these are valid developmental landmarks in the
mouse like eye-opening and etc. but they may not be valid indices of puberty."
EPA Response: An expanded discussion of the conserved biological processes between
mice and humans that is considered relevant to the observed developmental effects (e.g., for
delayed vaginal opening and preputial separation) has been included in Section 3 and
updated in Section 5.2.1. to include supporting references. "The onset of puberty in humans
is driven by surges in the levels of estrogen in females and testosterone in males, so the
timing of puberty can be altered by exposure to endocrine disrupting chemicals that mimic
or antagonize these hormones In female rodents, pubertal markers include vaginal opening
(indicative of the first ovulation in rats, but not mice) and the subsequent first estrus and
onset of regular estrous cyclicity (rats and mice) [Prevot, V., Puberty in Mice and Rats, in
Knobil and Neill's Physiology of Reproduction, T.M. Plant and A.J. Zeleznik, Editors. 2015 p.
1395-1439], Since Das etal. (2008). found delayed vaginal opening in mice (not rats), this is
not a direct correlate to puberty in humans. However, the Reproductive Guidelines state
that both accelerations and delays in the timing of reproductive milestones can be
considered adverse.
E-22
-------
Supplemental Information ofPFBA and Related Salts
Comment: One reviewer recommended that references to the developmental toxicity
effects of other PFAS should be documented in a chart and/or link to a chart/table.
EPA Response: This information can be found in three recently published reports Carlson
etal. (2022}: Radke etal. f20221.and Pelch etal. f20221. and that outline the available
references of developmental effects following hundreds of other PFAS. Additionally, Table
4-2 has been added to the assessment (see Section 4.1) to facilitate comparisons of
developmental toxicity hazard conclusions across EPA PFAS assessments.
Comment: One reviewer recommended emphasizing statements on the observation of
fetal effects in the absence of maternal toxicity and data gaps regarding information on the
thyroid and nervous system following gestational exposures. EPA Response: This data gap
is now discussed in the evidence integration summary of Section 3.2.3. (Developmental
Effects); specifically, Section 3.2.3 notes that developmental delays consistent with delayed
sexual maturation are observed in the absence of body weight or maternal effects, thus
strengthening the certainty that the observed effects are adverse fetal effects.
E.5.3. Tier 2 Suggestions
Comment: One reviewer suggested that the EPA could opine on a thyroid-mediated mode-
of-action in young animals based on evidence in adult rats (i.e., thyroid histopathology and
serum T4 decreases in adult animals).
EPA Response: There is insufficient available data supporting a direct link between a
thyroid-mediated mode-of-action in young animals based on evidence in adult rats.
Comment: One reviewer commented that "[w]hile the rationale [for developmental effects]
is well-explained, the tone of the report implies notably weaker support than for thyroid or
hepatic effects. It seems odd to put them into the same ultimate bin when there is such a
sharp contrast in the evidence and even in the way that the evidence is interpreted and
expressed." This reviewer suggested further explanation in the report why developmental
effects "ended up in the same level" despite a weaker evidence base.
EPA Response: Additional references have been incorporated in Section 3.2.3 to further
support the rationale for the evidence integration judgment for developmental effects.
Please note that the methods for drawing the evidence integration judgments and
determining the levels of certainty regarding potential hazard are outlined in Appendix A.
These methods do allow for a range of evidence scenarios with some variation in "strength"
to ultimately lead to the same overall judgments.
E-23
-------
Supplemental Information ofPFBA and Related Salts
E.5.4. Public Comments
Comment: The discussion of the effects of PFOA on timing of vaginal opening and preputial
separation in mice reported by Lau et al. (2006) is incomplete and appears to be
inappropriate. Lau et al. (2006) reported a non-monotonic dose-response for both day of
vaginal opening and day of preputial separation in mouse offspring from dams treated with
PFOA on gestational day (GD) 1-17. Vaginal opening was accelerated at the lowest dose (1
mg/kg/day) and delayed, at higher doses (3-20 mg/kg/day), with a greater delay with
increasing dose. More notably, preputial separation was accelerated, rather than delayed, at
all doses (1, 3, 5, and 10 mg/kg/day) except the highest dose (20 mg/kg/day). The
acceleration of preputial separation was greatest at the lowest dose (1 mg/kg/day) and the
magnitude of this effect decreased with increasing dose from 1-10 mg/kg/day. At 20
mg/kg/day, the only dose at which preputial separation was delayed, there was severe
toxicity including full litter resorptions in 88% of dams and approximatelyl0% survival
(i.e., ~90% mortality) of offspring at postnatal day (PND) 23. As such, the statement that the
observations in male offspring treated with PFOA from Lau et al. (2006) are part of a
"consistent pattern of delayed pubertal milestones...folio wing exposure to related PFAS"
does not appear to be supportable.
EPA Response: The relevant text in the Toxicological Review has been updated to indicate,
where appropriate, markers of sexual maturation were delayed or accelerated. Specifically,
to the comment that the dose-response for vaginal opening observed in Lau etal. (2006)
was non-monotonic: while the mean value of time to vaginal opening was slightly lower in
the lowest exposure group (compared to controls), this difference was not statistically
significant and is more appropriate characterized as "no effect" rather than "accelerated".
Comment: The Reproductive Toxicity Guidelines (U.S. EPA,1996) state that alterations in
the age at puberty should be considered as adverse effects. Citing this, the Toxicology
Review of PFBA uses delays in vaginal opening and preputial separation in the mouse for
RfD, assuming that these endpoints accurately reflect pubertal development. However, the
biological basis for this assumption is lacking and reviews suggest that these measurements
accurately reflect pubertal development in the rat but not the mouse. This species
difference is described in Chapter 30 "Puberty in Rats and Mice" by Vincent Prevot (Knobil
and Neil's Physiology of Reproduction. Elsevier Science and Technology, 2015) and in a
publication in Scientific Reports (2017) by Gaytan etal. (DOI:10.1038/srep46381).
Related Comment: In rats, but not mice, vaginal opening is normally associated with
maturation of the HP axis, ovulation and the initiation of reproductive cycles. Mouse
"puberty" has been used for POD calculations in some of the PFAS documents and this may
not be biologically correct and the Agency may want to reconsider this. Clearly these are
E-24
-------
Supplemental Information ofPFBA and Related Salts
valid developmental landmarks in the mouse like eye-opening but they may not be valid
indices of puberty.
EPA Response: As noted by the commenter, alterations in the timing of reproductive
development (whether delayed or accelerated) are considered adverse in accordance with
the EPA Reproductive Guidelines. While vaginal opening in mice may not be a direct
correlate to puberty in humans (as it is in rats), alterations in reaching reproductive or
developmental milestones are considered adverse and relevant to human health. However,
the linkage of these delays in acquisition of these milestones to "puberty" has been removed
in this revised assessment
E.6. CHARGE QUESTION 3 - REPRODUCTIVE AND OTHER EFFECTS
For each health effect considered in the assessment and outlined below, please comment on whether
the available data have been clearly and appropriately synthesized to describe the strengths and
limitations. For each, please also comment on whether the weight-of-evidence decisions for hazard
identification have been clearly described and scientifically justified.
• For reproductive effects and other noncancer effects (i.e., cardiometabolic effects, renal
effects, ocular effects, body weight), the Toxicological Review concludes there is
inadequate evidence to determine whether PFBA exposure has the potential to cause these
effects in humans on the basis of the sparsity of available evidence.
E.6.1. Overarching External Peer Reviewer Comments on Reproductive and Other Effects
"All reviewers concurred with EPA's conclusion that there is insufficient evidence to
determine if PFBA can cause reproductive or other noncancer effects."
E.6.2. Tier 1 Recommendations
Reviewers had no Tier 1 Recommendations.
E.6.3. Tier 2 Suggestions
Comment: One reviewer suggested that EPA consider adding a recent study by Ou et al.
(2021) that indicates some PFAS may increase the risk of heart defects.
EPA Response: While the Ou et al. (2021) study does suggest that some PFAS are
associated with increased odds of septal defects, for PFBA specifically, odds ratios were not
statistically significantly increased or decreased for septal defects, conotruncal defects, and
all congenital heart defects. Hence, given that no defects were observed to be associated
with PFBA exposure, discussion of this study was not added to the assessment given that
consideration of its observations would not materially change the evidence integration
conclusion for reproductive or developmental effects.
E-25
-------
Supplemental Information ofPFBA and Related Salts
Comment: One reviewer suggested that EPA consider including supporting mechanistic
evidence or supplemental information for reproductive and other noncancer effects for a
fuller description of the strength and limitations of the available information for
reproductive and noncancer effects.
EPA Response: There is inadequate evidence to determine whether PFBA exposure has
the potential to cause human reproductive effects (aside from the delays in sexual and/or
reproductive development discussed in Section 3.2.3: Developmental Effects) or the "other
noncancer effects" discussed in the draft assessment. In general, and as in this case here
where there is only limited (or none) supplemental information on these outcomes, there is
not clear added value in delineating the potentially relevant mechanistic information on
health effects for which toxicological evidence is generally lacking. Therefore, no additional
discussions of mechanistic or supporting evidence for reproductive and other noncancer
effects were added to this revised assessment.
E.7. CHARGE QUESTION 5 - CANCER HAZARD
The draft assessment concludes there is inadequate evidence to assess carcinogenic potential for PFBA
and that this descriptor applies to oral and inhalation routes of human exposure. Please comment on
whether the available animal and mechanistic studies, and the analysis presented in the Toxicological
Reviewsupport this conclusion.
E.7.1. Overarching External Peer Reviewer Comments on Cancer Hazard
"All reviewers concurred with EPA's conclusion that there is inadequate evidence to assess
carcinogenic potential for PFBA for either oral or oral inhalation exposure. [One reviewer] noted
that the "evidence that PFBA is carcinogenic is sparse and continues to be sparse according to a
recent review of PFAS and cancer (Steenland and Winquist 2021)."
E.7.2. Tier 1 Recommendations
Comment: One reviewer commented unlike the other sections with limited data, no
discussion was included in the carcinogenicity section about how other studies of PFBA-
related compounds could or could not inform the data gaps for carcinogenicity or
genotoxicity. This reviewer recommended that EPA should include a short section to
address this missing component in a manner similar to the other sections of the report
EPA Response: A short explanation that the evidence for carcinogenicity due to PFBA
exposure is non-existent and is limited for other related PFAS has been added to Section 3.3.
In addition, the carcinogenicity conclusions from other EPA PFAS analyses that have been
finalized have been included in Table 4.2, although they are not influential to the
carcinogenicity judgment for PFBA. Conclusions from other ongoing IRIS assessments have
E-26
-------
Supplemental Information ofPFBA and Related Salts
not been added because they have not yet been finalized. Although not currently available,
future efforts on PFAS (see PFAS Strategic Roadmap {U.S. EPA, 2021,10002133@@author-
year), potentially including additional studies on PFBA, may help to inform this data gap.
E.7.3. Tier 2 Suggestions
Reviewers had no Tier 2 suggestions.
E.8. CHARGE QUESTION 6 - NONCANCER TOXICITY VALUE DATA
SELECTION
For PFBA, no RfC was derived. The Butenhoff et al. (2012) 90-day rat study was the study chosen for
use in deriving the RfD on the basis of an increased incidence of hepatocellular hyperplasia and
decreased total T4 in male rats. Is the selection of this study and these effects for use in deriving the
RfD for PFBA scientifically justified?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effect(s) that should be used to support the
derivation of the RfD and detail the rationale for use of such an alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the effects
selected are appropriate for use in deriving the RfD, including considerations regarding
adversity (or appropriateness in representing an adverse change) and the scientific support
for their selection. More specifically, Appendix A identifies interpreting the adversity of certain
outcomes observed in rodents, including some hepatic effects, as a key science issue. Please
consider in your recommendation the narrative in the Toxicological Review related to the
decision that the observed hepatocellular hypertrophy, when considered within the broader
constellation of effects, is representative of an adverse change in the organ.
d. Given the lack of studies on inhalation exposure to PFBA, no reference concentration (RfC) is
derived. Please comment on this decision.
E.8.1. Overarching External Peer Reviewer Comments on Noncancer Toxicity Value Data
Selection
"All reviewers concurred: (1) with the selection of the Butenhoff et al. (2012) study as
scientifically justified for derivation of an RfD for PFBA; (2) that the critical effects selected were
appropriate for use in deriving the RfD; and (3) that the decision to not derive a reference
concentration (RfC) was justified, given the lack of studies on inhalation exposures to PFBA."
E.8.2. Tier 1 Recommendations
Comment: One reviewer recommended adding a discussion of other carcinogenicity
studies across the structurally related perfluorinated compounds to Section 5.2.
E-27
-------
Supplemental Information ofPFBA and Related Salts
EPA Response: A short explanation that the evidence for carcinogenicity due to PFBA
exposure is non-existent and is limited for other related PFAS has been added to Section 3.3.
In addition, the carcinogenicity conclusions from other EPA PFAS analyses that have been
finalized have been included in Table 4.2.
E.8.3. Tier 2 Suggestions
Comment: One reviewer commented that identification of hepatic effects observed in
rodents as a key scientific issue is a very conservative approach and that the conclusion is in
contradiction with Hall etal. (2012), stating "The results of this workshop concluded that
hepatomegaly as a consequence of hepatocellular hypertrophy without histologic or clinical
pathology alterations indicative of liver toxicity was considered an adaptive and a non-
adverse reaction." This reviewer stated that this conclusion should be reached by an
integrative weight of evidence approach.
EPA Response: An integrative weight of evidence approach is the approach taken in
Section 3.2.2. The conclusions in Section 3.2.2 are entirely in-line with the
recommendations of {Hall, 2012, 2718645@@author-year} that coincident histological
evidence of liver injury/damage can be used to support the conclusion that liver weight and
hypertrophic changes are adverse. Specifically, for PFBA, liver weight changes and
hypertrophic lesions are accompanied by other lesions such as necrosis and vacuolation,
supporting their adversity along a progression to more severe effects. Thus, no revision was
made. See related responses to comments on charge question 3.
E.8.4. Public Comments
Comment: The Das study was not considered as the critical study because it did not
measure serum thyroid hormones. This decision does not withstand scrutiny from the
scientific community. Serum thyroid hormones are subject to a great degree of variability
due to the assay issues, diurnal variations, and even husbandry conditions, and thus are not
suitable grounds upon which to exclude a relevant study. The Das study evaluated
important key functional aspects of pregnancy outcomes as well as neonatal development
into young adults and should have been considered as the critical study in the Draft IRIS
Review for PFBA. Further, given the current Draft IRIS Review's emphasis on
developmental toxicity with exposure to PFBA and the availability of relevant studies, EPA
should rely on its own developmental study for PFBA as the critical study to develop the
reference value for PFBA.
EPA Response: It is not accurate to state that the Das study was not considered as the
critical study because it did not measure serum thyroid hormones. Endpoints from both Das
etal. (20081 and Butenhoff etal. (20121 were considered for POD derivation. Ultimately, as
E-28
-------
Supplemental Information ofPFBA and Related Salts
liver and thyroid effects from (Butenhoff etal.. 20121 were interpreted with the highest
confidence and resulted in lower PODs, the candidate values for these endpoints were
selected for the RfD. Additionally, the candidate value for developmental delays from Das et
al. f20081 was selected for the subchronic RfD (see below).
Comment: It is suggested that the rationale for not considering hepatic effects in male mice
exposed to PFBA for 28 days in Foreman et al. (2009) as the basis for Reference Dose (RfD)
development be reconsidered. It (i.e., the assessment) is stated that the subchronic (90 day)
study in rats (Butenhoff et al., 2012b) and the developmental study in mice with 17 days of
exposure (GD 1-17; Das etal., 2008) were considered for RfD development because these
study designs can "estimate potential effects of lifetime exposure, as compared to short-
term [i.e., 28 day] or acute studies." However, the preference for developmental studies
over short term (i.e., 28 day) studies does not appear to be supportable for RfDs based on
systemic effects such as increased relative liver weight or histopathological changes in the
liver. Specifically, Table 3-5 shows that both wild-type and humanized PPAR-alpha male
mice exposed to PFBA for 28 days (Foreman et al., 2009) are more sensitive to increased
relative liver weight (i.e., a greater increase at the same or similar dose) than pregnant and
non-pregnant female mice exposed for 17 days (Das et al., 2008) and are also more sensitive
than male rats exposed for 28 and 90 days (Butenhoff et al., 2012b).
EPA Response: For the purpose of deriving chronic non-cancer reference values,
subchronic exposure studies are preferred over short-term studies when chronic studies
are lacking in the toxicity database. Also, developmental toxicity studies are useful for
evaluating the potential for increased susceptibility in pregnant animals or their offspring
during this sensitive lifestage, when short periods of exposure during critical windows of
development can be considered more relevant to identifying sensitive health effects from a
lifetime of exposure than subchronic or chronic exposure durations. This latter
consideration does not necessarily apply to all health endpoints, likely including the hepatic
effects of PFBA. A greater amount of uncertainty exists in extrapolating from short-term to
chronic durations and thus, given the availability of preferred studies in the PFBA database,
the results of short-term studies are only considered as supporting information for toxicity
value derivation.
E.9. CHARGE QUESTION 7 - SUBCHRONIC REFERENCE DOSE
In addition, for PFBA, an RfD for less-than-lifetime ("subchronic") exposures is derived. No
"subchronic" RfC was derived. The study chosen for use in deriving the subchronic RfD is the
gestational exposure mouse study by Das et al. (2008) with the RfD based on delayed acquisition of
developmental milestones¦, as indicated by delayed time to vaginal opening, eye opening, and preputial
E-29
-------
Supplemental Information ofPFBA and Related Salts
separation in exposed male and female offspring. Is the selection of this study and these effects for the
derivation of the subchronic RfD for PFBA scientifically justified?
a. If so, please provide an explanation.
b. If not, please provide an alternative study(ies) or effect(s) that should be used to support the
derivation of the subchronic RfD and detail the rationale for use of such an alternative.
c. As part of the recommendations in "a" or "b" above, please comment on whether the effects
selected are appropriate for use in deriving the RfD, including considerations regarding
adversity (or appropriateness in representing an adverse change) and the scientific support
for their selection.
d. Given the lack of studies on inhalation exposure to PFBA, no "subchronic" RfC is derived. Please
comment on this decision.
E.9.1. Overarching External Peer Reviewer Comments on the Subchronic Reference Dose
"All reviewers concurred that selection of the Das et al. (2008) study is scientifically
justified for derivation of the subchronic RfD. [One reviewer] noted that a subchronic RfD based on
developmental effects resulting from a shorter exposure duration "will provide a useful risk
assessment complement to the chronic RfD, furthering risk assessment and risk communication."
"Most reviewers concurred that the selection of effects were justified for derivation of the
RfD, although [one reviewer] commented that EPA should consider improving the scientific
justification to the extent possible for certain effects."
"All reviewers concurred that the decision to not derive a subchronic RfC was justified."
E.9.2. Tier 1 Recommendations
Comment: One reviewer recommended that a discussion be included regarding how
inclusion of delayed eye opening would or would not change the RfD.
EPA Response: The candidate value for delayed eye opening is included in the endpoints
considered for final RfD selection in Table 5-7 of the assessment. As can be seen from the
values presented in this table, selection of this endpoint would result in an RfD 33% higher
than currently selected (8 x 10-3 mg/kg-day vs. 6 * 10:i mg/kg-day) and was not interpreted
with greater confidence, and thus would be inadequately protective of human health. A brief
discussion to this effect has been added.
Comment: One reviewer commented that they were surprised that the cumulative
endpoint of "total affected implants" was not reported or evaluated. This reviewer
E-30
-------
Supplemental Information ofPFBA and Related Salts
recommended that the Toxicological Review add a paragraph discussing further several of
the decisions to not look at the endpoint of all affected, why the Rai and Van Ryzin model
was not used and discuss the importance of looking at this cumulative endpoint This
paragraph should specifically discuss whether these suggestions would have changed the
calculation of the subchronic RfD.
EPA Response: A short discussion of this topic, including an explanation for why the
cumulative endpoint was not modeled, has been added to the assessment in Appendix D.1.2
and as a footnote to Table 5-1. The Rai and van Ryzin model are a nested dichotomous
model that can account for intralitter similarity (via estimation of intralitter correlation and
use of litter specific covariate parameters). Functionally, this model has the same
capabilities of the nested Logistic model, which is currently implemented in BMDS 3.2 and
was used to model the embryo/fetal mortality endpoint How modeling "total affected
implants" would ultimately impact the final RfD derived is difficult to characterize but given
that fetuses that lived to parturition and experienced some delay in developmental
milestones would be counted as "responding" alongside fetuses that died in utero, the total
incidence of "affected fetuses" would be greater per dose group than for the individual
endpoints, thus resulting in a lower BMD and BMDL. This would result in a lower POD.
Comment: One reviewer commented on the need to add a paragraph that expands the
discussion of what the lack of a functional reproductive or neurodevelopmental assessment
means for the application of the current uncertainty factors in this incomplete toxicological
assessment package. The signals from the Das et al 2008 study should raise concern of this
lack and require additional assessment of what this widespread environmentally relevant
compound and metabolic common breakdown product means to the overall IRIS
assessment report across the structurally related perfluorinated compounds.
EPA Response: The lack of a functional reproductive or neurodevelopmental study in
relation to selecting the uncertainty factors is discussed in Section 5.2.1. Briefly, the lack of a
functional reproductive study was not considered a key quantitative data gap in the PFBA
database, in part given the general lack of evidence for sensitive reproductive effects for
some other similar PFAS (see Table 4-2). Regarding the lack of a functional
neurodevelopmental study, while this is identified as a key data gap in the PFBA evidence
base, concerns over developmental neurotoxicological effects presumed to result from
thyroid hormone insufficiency are mitigated given evidence from PFBS (i.e., PODs for effects
in dams and offspring are almost identical for thyroid hormones). However, concern over
developmental neurotoxicity independent of a thyroid hormone-mediated mechanism
remains an important uncertainty accounted for via application of a UFd = 3.
E-31
-------
Supplemental Information ofPFBA and Related Salts
Comment: One reviewer reiterated their previous Tier 1 Recommendation (see Section
E.5) that EPA should consider improving scientific justification (e.g., expanded to other
developmental effects) to the extent possible, particularly for the critical effect. Specifically,
this reviewer requested that EPA expand discussion on the conserved biological processes
between mice and humans that the EPA considers relevant to the observed developmental
effects (e.g., for delayed vaginal opening and preputial separation), whether the mouse has
been shown to be a good laboratory animal model for assessing potential human
developmental effects, or what human developmental endpoints (e.g., delayed onset of
puberty) may be presumed to be correlates of some of the PFBA-induced developmental
effects observed in the single mouse study (e.g., delays in vaginal opening and preputial
separation in Das etal. (2008)).
EPA Response: A discussion on the known similarities and differences in the development
of these outcomes between rodents and humans has been added to the assessment (see
Section 3.2.3). It is important to emphasize, as described in EPA guidelines, that it is not
expected that the effects that manifest in animal studies will manifest similarly in humans,
although outcomes that are strongly correlated across species can provide strong evidence.
Likewise, in the absence of evidence to the contrary, effects observed in animal models are
considered relevant to humans.
Comment: One reviewer noted that it appeared to them that the "...document supports use
of delayed eye opening and embryo/fetal mortality more strongly than other developmental
effects. While these effects provide candidate subchronic RfD values similar to that selected
for the proposed final subchronic RfD, delays in vaginal opening observed in Das et al.
(2008) apparently provides the most specific basis for thatvalue (Table 5-10, p. 5-24), the
adversity of which does not appear to be fully addressed in the document." This reviewer
recommended that relevant information to support the adversity of the critical effect
ultimately selected for the determination of the subchronic RfD be added to the assessment.
EPA Response: A discussion regarding the adversity of the observed developmental delays
has been added to the assessment in Section 3.2.3.
E.9.3. Tier 2 Suggestions
Reviewers had no Tier 2 suggestions.
E.10. CHARGE QUESTION 8 - NONCANCER TOXICITY VALUE DOSE-
RESPONSE MODELING
EPA used benchmark dose modeling (U.S. EPA, 2012) to identify points-of-departure (PODs) for oral
exposure to PFBA. Are the modeling approaches used, selection and justification of benchmark
E-32
-------
Supplemental Information ofPFBA and Related Salts
response levels, and the selected models used to identify each POD for toxicity value derivation
scientifically justified?
E.10.1. Overarching External Peer Reviewer Comments on Noncancer Toxicity Value Dose-
Response Modeling
"The reviewers who responded to this charge question supported the modeling approach
used and the justification provided, with some reviewers noting enthusiastic support for the
approach."
"Several reviewers noted that benchmark dose modeling is not their area of expertise and declined
to comment."
E.10.2. Tier 1 Recommendations
Reviewers had no Tier 1 recommendations.
E.10.3. Tier 2 Suggestions
Comment: One reviewer suggested that EPA attempt to strengthen the benchmark
response (BMR) justification for vaginal opening delays beyond historical precedence (to
the extent possible) given that it is ultimately the critical effect used in determining the
sub chronic RfD.
EPA Response: The selection of a BMR = 5% relative deviation for delayed vaginal opening
is not based on historical precedence, but rather is based on a determination of what level
of change in this effect has been judged to be a relevant response level (i.e., a minimally
biologically significant response) in the EPA Endocrine Disruptor Screening Program. In
addition, this approach in defining a BMR for continuous endpoints based on biological
information is consistent with the recommendations of the BMD Technical Guidance fU.S.
EPA. 20121. A full discussion of the selection of the BMR for developmental delays is
included in Table 5-2.
Comment: One reviewer commented that the model selected for liver hypertrophy should
possibly be the log-logistic model vs the currently selected Weibull model based on equal
Akaike Information Criterion values and a BMDL value that appears to better agree with the
actual study data.
EPA Response: After consideration of the shape of the dose-response curve for this
dataset, it was determined that the minimum-maximum characteristic of the data (i.e.,
response going from 0/10 to 9/10 between the mid- and high-dose groups precludes
modeling this dataset (consistent with the BMD Technical Guidance (2012) document (U.S.
EPA. 2012)). Thus, the NOAEL/LOAEL approach was used instead for this endpoint This
E-33
-------
Supplemental Information ofPFBA and Related Salts
change in POD derivation methodology does not have a large impact on the assessment as
the NOAEL (6 mg/kg-day) is very close to the BMDL (5.4 mg/kg-day).
E.ll. CHARGE QUESTION 9 - TOXICOKINETICS
Appendix A identifies the potential for toxicokinetic differences across species and sexes as a key
science issue and lays out a hierarchy for using relevant toxicokinetic data in extrapolating doses
between laboratory animals and humans. Given what is known and not known about the potential
interspecies differences in toxicokinetics of PFBA, EPA used the ratio of human-to-animal serum
clearance values to adjust the POD to estimate a human equivalent dose in the derivation of the
respective RfDs.
a. Is applying the ratio of human-to-animal serum clearance values for PFBA scientifically
justified? If not, please provide an explanation and detail on a more appropriate approach.
b. Do the methods used to derive toxicity values for PFBA appropriately account for uncertainties
in evaluating the toxicokinetic differences between the experimental animal data and
humans?
E.ll.l. Overarching External Peer Reviewer Comments on Toxicokinetics
"The reviewers concurred that the application of the ratio of human-to-animal serum
clearance values for PFBA was scientifically justified. [One reviewer] commented that applying the
ratio is a more appropriate choice than scaling doses allometrically using body weight (BW)3/4
methods."
E.11.2. Tier 1 Recommendations
Comment: One reviewer suggested that another approach to help justify or strengthen the
animal to human extrapolation is to use kidney filtration (GFR) and that comparison of
excretion ratios (i.e., clearance/GFR) could be compared to other PFASs where data exist
This reviewer stated that these analyses may be insightful for characterizing the degree of
renal/hepatic reabsorption and ultimately increase the confidence in the calculated human
CL.
EPA Response: Two paragraphs were added to Section 3.1.5 comparing the clearance
values to species-specific average GFR values for mice, rats, and humans, with or without
including the impact of serum binding (i.e., GFR*free). A comparison across PFAS was not
made in the revised PFBA assessment (a preliminary analysis of results across draft and
final assessments indicates that there is not a consistent, predictable pattern, so any such
discussion would not be straightforward or brief]. However, this is taken as a useful
research note for future applications.
E-34
-------
Supplemental Information ofPFBA and Related Salts
Comment: One reviewer recommended that EPA clarify sections on PFBA toxicokinetics
with respect to the consistency of the linearity assumptions with the presence of saturation
processes. This reviewer specifically noted that EPA stated that dosimetric adjustments are
made "assuming the exposure being evaluated is low enough to be in the linear (or first-
order) range of clearance" in one part of Section 5, but in another part stated that "results
for both male and female mice [from Chang et al., 2008] show a dose-dependent increase in
clearance across all dose levels, consistent with the hypothesis of saturable renal
resorption."
EPA Response: Given the limited PK data available, an extensive analysis of this issue is
not possible. However, a paragraph has been added to Section 3.1.5 explaining that the
mouse data appear to be reasonably consistent with constant CL at < 30 mg/kg-day, and the
plotted rat data for the same dose in Chang etal. f20081 appear likewise, so linearity will be
assumed valid for dose levels below that, not above. A statement was also added in Section
5, where the approach for extrapolation is described.
E.11.3. Tier 2 Suggestions
Comment: One reviewer suggested that EPA could use dosimetrically adjusted doses
rather than a ratio of clearance values, and that this could be considered to see if such
calculations would make significant differences.
EPA Response: If the dosimetric adjustment still involves use of a single clearance value,
then changing the order of dosimetric adjustment and dose-response analysis will not
change the outcome. Such an approach could make a significant difference if the adjustment
is dose-dependent However, such an approach would significantly complicate the analysis.
Clearance will not be a function of dose directly, but rather a function of blood
concentration or internal dose, which in turn both depends on dose and time. Thus, a
nonlinear PK model would need to be developed, validated, and applied to properly account
for the time- and dose-dependence in a bioassay. The comment suggests consideration of a
simpler approach, with clearance just assumed to be a nonlinear function of dose, but this
would require picking a function form and then using it to interpolate between and
extrapolate above doses used in the mouse PK studies, which has its own uncertainties.
Further, since the PK study for rats only used a single dose level, the approach can't be used
directly for that species. Given these considerations, development and application of a
nonlinear PK model was not attempted.
Comment: One reviewer noted that, while the Chang et al. (2008) occupational exposure
data provide important insights into PFBA pharmacokinetic behavior, these data are not
sufficiently robust for calculating pharmacokinetic parameters, and that EPA uses a half-life
E-35
-------
Supplemental Information ofPFBA and Related Salts
that is sufficiently long to encompass most adults from the study. This reviewer suggested
showing how the half-lives were determined with the data and the logic for selecting a
specific value.
EPA Response: Details of the analysis were previously provided in Appendix C.2. The
rationale for excluding subjects for whom the second measurement was below the LLOQ
were provided in Section 5.2.1 but are now restated in Appendix C.2. A table listing the
subjects used in EPA's analysis has been added to Appendix C.2. Of those eight subjects, four
had half-lives greater than the estimated half-life, 67.9 hours, and four were below. Of the
four excluded subjects, one had an estimated half-life greater than 67.9 hours. Hence, while
it may be true that 7 of the 12 total subjects (i.e., "most") had half-lives lower than EPA's
estimate, this is only 58% of the study population. Therefore, the assessment concludes that
the estimated mean in this estimate is reasonable.
While it is recognized that there is uncertainty inherent in the use of these data, but they are
nonetheless human elimination data. The alternative to use of these human elimination data
is to estimate the human half-life based on BWy> scaling, which leads to predicted half-lives
of 38 hours in men and 7 hours in women, which are within a factor of 2 of the value from
EPA's analysis (67.9 hours) for men and within a factor of 10 of that estimates for women.
Given that PFAS are generally known to be subject to renal resorption, use of 67.9 hours is
therefore considered both biologically plausible and modestly health-protective compared
to use of BW% scaling. The presumption is that it provides an average elimination, hence
clearance rate for humans and that uncertainty in this value is addressed by the portion of
UF_H assigned (i.e., factor of 3).
Comment: One reviewer commented that, under steady-state exposure to humans (e.g.,
drinking contaminated water) the bioaccumulation would depend on the half-life but no
matter how quickly it's cleared, it is steadily replaced. This reviewer further stated that
humans don't receive a single dose and stop being exposed, so slow or rapid clearance
would be less influential than in the typical laboratory situation. This reviewer suggested
that this could be explained in the report, indicating how the steady-state exposure to
humans bears on the evaluation.
EPA Response: If humans are exposed to a regular (daily) dose, D, then use of the
estimated human clearance (CLh) leads to a prediction of an ongoing blood concentration
equal to D/CLh; i.e., that is the steady-state or average blood concentration given the daily
dose, D. Hence, this evaluation assumes that the steady-state level increases or decreases in
direct proportion to D, with 1/CLh being the proportionality constant. This is now stated in
the section on dose extrapolation.
E-36
-------
Supplemental Information ofPFBA and Related Salts
E.11.4. Public Comments
Comment: The assessment uses a worst-case estimate for the serum excretion half-life
(tl/2) of PFBA in humans which is confounded by potential co-exposure to other PFAS
molecules. The draft IRIS assessment assumes a tl/2 of 67.9 hours based on a study of
workers exposed to other PFAS believed to be metabolized to PFBA by Chang et al. (2008).
The authors [i.e., Chang et al., 2008] note, however, thatthe data need to be interpreted
cautiously, because the workers were not exposed directly to PFBA and were exposed to
materials that, given their chemical structures, likely are metabolized to PFBA via oxidation
or hydrolysis. Moreover, the draft assessment rejects the default body weight (BW0.75)
approach which would generate a half-life of 37.8 hours - within a factor of two of that
derived from Chang etal., 2008.
EPA Response: Given the lack of controlled PK studies of PFBA in humans, the only
empirical data one can use are those potentially confounded with possible ongoing
exposures or metabolic production from precursors. Environmental epidemiological
analyses, in which observed blood concentrations are correlated with estimated exposure
levels, have an inherent uncertainty in that the exposure is not exactly known. Given that
any estimate of human elimination has uncertainty, the EPA considers the use of a health-
protective estimate to be reasonable, while noting that 67.9 hours is in fact not an upper
bound of the Chang etal. (20081 data: five of the 12 subjects had estimated half-lives
greater than that, as noted above. It is also noted that for subjects 1 and 2 of the Cottage
Grove group in Chang, for whom multiple time-points are shown in Figure 6, the decline is
very close to log-linear. If there was significant ongoing exposure or metabolic production,
the decline would asymptotically approach a plateau, the steady-state level given that
exposure. Further, the tl/2 values for these two subjects are above the average obtained. In
summary, the results do not represent a worst-case scenario, which would involve use of an
upper-bound half-life estimate from these data. While it is possible that the true average is
lower, there are no specific data to support this possibility, for example that PFBA
elimination is reduced in this population due to co-exposures to other PFAS. Further, EPA
guidelines U.S. EPA (20111 support the use of such empirical data over default BW% scaling
when such data exist.
Comment: The approach taken in the draft assessment for calculating a dose adjustment
factor (DAF) for the animal data overestimates serum concentrations which leads to an over
prediction of the toxicity ofPFBA. Although using the ratio of clearance rates to generate the
DAF is preferable to using the ratio based on half-life, the uncertainty around the tl/2
estimate raises significant questions about the decision to use it to derive the human
equivalent dose (HED). Given this uncertainty, using the default body-weight scaling
E-37
-------
Supplemental Information ofPFBA and Related Salts
method to develop the DAF is more appropriate. Failing that, the assessment should include
additional discussion of the uncertainty in the clearance model that is chosen.
EPA Response: As noted above, EPA guidelines U.S. EPA f20111 state that extrapolation
based on chemical-specific data is preferred over default scaling. Discussion of the
uncertainty in the human half-life estimate has been augmented in the revised assessment
(see Sections 3.1.4 and 5.2.1), including discussion of the potential for ongoing exposure or
metabolic production ofPFBA, although noting that there is no clear indication of this
occurring. The 2-fold difference between the estimated clearance and that predicted from
BWA% is well within the range of uncertainty that would be expected for either value.
Comment: The body weight values for the animals in the study that is being evaluated, if
available, should be used instead of default body weight values to derive dosimetric
adjustment factors (DAFs) based on body weight3/4.
EPA Response: In general, study-specific body weight (BW) should be used when
extrapolating dosimetry for a given toxicological observation. However, use of a DAF based
on clearance builds in a BW adjustment, in that clearance is a rate per kg BW. While some
intra-species variation in CL may occur with BW, the data available are not sufficient to
demonstrate such variation, such variation will not have a significant effect on the outcome,
and as addressed above, BW3/4 scaling was not determined to be the best approach for
PFBA.
Comment: A recent paper, Abraham et al. (2021), that investigated the distribution of
PFBA in human tissues and came to conclusions that differ from those of Perez et al. (2013).
should be cited. See: Abraham etal. (2021). Perfluorobutanoic acid (PFBA): No high-level
accumulation in human lung and kidney tissue. International Journal of Hygiene and
Environmental Health, 237,113830.
EPA Response: First, the term "accumulation" is interpreted as an increase in tissue levels
over time, given ongoing exposure. The determination of whether or not accumulation in
that sense occurs requires longitudinal samples of the same individual over time while
carefully monitoring exposure. Both Perez etal. (2013) and Abraham etal. (2021) only
reported single time-point samples; hence, neither paper contains data that can be used to
demonstrate or refute accumulation (e.g., a high tissue concentration can occur due to high
exposure without accumulation). Second, the subjects of Perez etal. f20131 were from
T arragona County (Catalonia, Spain), while those analyzed by Abraham etal. f20211 were
from France, collected 3-6 years later than Perez, so there were likely differences in their
exposure levels. That being said, the enormous differences in reported tissue levels
certainly raises questions about the respective analytic methods, and so the results of
E-38
-------
Supplemental Information ofPFBA and Related Salts
Abraham are now also provided in the revised assessment (see Section 3.1.2), noting the
discrepancy given these caveats about unknown differences in exposure. A recently
published paper by EPA authors Bangma etal. f20211 did identify an endogenous
compound in placenta that is a likely analytic interferent, and this has been added to the
document The interfering compound identified by Bangma etal. f20211 would have to be
present in human tissues but not in the pig tissues used for QA by Perez etal. T20131 in
order to explain the discrepancy.
Importantly, as noted in the revised assessment, Abraham etal. (20211 paper does not
report some necessary methodological information including whether matrix-matched
calibration curves were used or what QA/QC measures were taken (i.e., duplicates, method
blanks, continuous calibration verification, etc.). Perez etal. f20131 state that matrix
matched calibration was used for each tissue type using pig brain, liver, bone, and kidney;
that reagent blanks, sample blanks, a repeated measures were conducted. The Perez etal.
(2013) study also states that their method was validated, which is not reported by
(Abraham etal.. 2021). Hence, based on details provided in the published papers, EPA's
evaluation yields much higher confidence in (Perez etal.. 2013) than (Abraham etal.. 2021).
Since the results of Perez etal. f20131 are not used in the quantitative assessment, this
notation will only have a qualitative impact on the Toxicological Review.
Comment: The toxicokinetic differences between wild-type, PPAR-alpha null, and
humanized PPAR-alpha mice reported by Foreman et al. (2009) are unlikely to result from
differences in PPAR-alpha status and are potentially relevant to interpretation of
differences in susceptibility to toxicity among these strains. Suggest adding serum and liver
concentration data from Foreman et al. (2009) to Table 3-1.
EPA Response: While some of the differences between strains reported by Foreman etal.
f20091 are not easily attributable to PPRAR-alpha status, this is not true for all of the
differences. For example, the lower liver concentrations in null mice compared to wild-type
is explainable as the lack of binding to PPAR-alpha in the null mouse liver. Other differences
may be secondary to differences in hepatotoxicity (i.e., that this toxicity may be more severe
in PPAR-alpha-carrying mice, but not in null mice). That both serum and liver
concentrations in the humanized mice at 175 mg/kg were lower than at 35 mg/kg is more
difficult to explain, as it suggests significantly higher clearance in this strain at that dose
compared to lower doses. The data indicate, though, that PPAR-alphas status has a minimal
effect on liver:serum distribution. This now stated and the data have been added to Table 3-
1.
E-39
-------
Supplemental Information ofPFBA and Related Salts
E.12. CHARGE QUESTION 10 - UNCERTAINTY FACTOR APPLICATION
EPA has evaluated and applied where appropriate uncertainty factors to account for intraspecies
variability (UFh), interspecies differences (UFa), database limitations (UFd), duration (UFs), and
LOAEL-to-NOAEL extrapolation (UFl) for PFBA.
a. Has uncertainty been adequately accounted for in the derivation of the toxicity values? Please
describe and provide suggestions¦, if needed.
b. For uncertainty in interspecies differences (UFA), a value of 3 is applied to extrapolate
between effects in laboratory animals and in humans. Although PPARa dependence might
support a value of UFA = 1 if that were the sole mode of action, evidence for non-PPARa MOAs
is available in the PFBA (and larger PFAS) database. Thus, uncertainty remains regarding the
potential differences in sensitivity across species due to the involvement of both PPARa-
dependent and PPARa-independent mechanisms. Further, data are lacking to determine with
confidence the relative contribution of these competing MOAs. As such, the Toxicological
Review concludes the available data are not adequate to determine if humans are likely to be
equally or less sensitive than laboratory animals with respect to the observed hepatic effects
and that a value ofUFA=3 is warranted to account for the residual uncertainty in
toxicodynamic differences across species. Please comment on whether the available animal
and mechanistic studies support this conclusion and whether the analysis presented in the
Toxicological Review is clearly documented.
c. For uncertainty in extrapolating from subchronic to chronic exposure scenarios (UFS), a
default value of 10 is applied. The assessment concludes there is conflicting evidence on
whether effects manifest at lower exposure levels or are more severe at equivalent exposure
levels when comparing findings across short-term and subchronic exposure durations. Thus, to
account for the potential for some effects to worsen with longer durations of exposure
(subchronic vs. short-term) and the lack of data on whether effects from subchronic exposures
might worsen in a chronic exposure scenario, a UFS=10 is applied in the Toxicological Review.
Does the provided scientific rationale support this decision? Please explain.
d. To inform uncertainty in intraspecies variability (UFH), the assessment evaluates and
considers the available evidence on potential susceptibility to PFBA within different
populations or lifestages, including any potential human health impacts from early life
exposure. Are the available information and data appropriately considered and the resultant
UFH values scientifically justified and clearly described?
e. Does the provided scientific rationale support the application of the remaining uncertainty
factors (UFL, UFD)? Please explain.
E.12.1. Tier 1 Recommendations
Reviewers had no Tier 1 recommendations.
E.12.2. Tier 2 Suggestions
Comment: One reviewer suggested that the selection of a value of 3 for UFa should be
evaluated and considered as an alternative to a value of 10 (rather than a value of 1) as our
E-40
-------
Supplemental Information ofPFBA and Related Salts
current understanding of interspecies differences in PFAS toxicokinetics and
toxicodynamics has very significant gaps.
EPA Response: A value of UFa = 10 is not supported given that interspecies differences in
toxicokinetics are accounted for in the application of the dosimetric adjustment factor
(DAF). In the case of the PFBA assessment, this DAF explicitly accounts for differences in the
toxicokinetics (i.e., serum clearance values) between rodents and humans. However, the
data gaps in understanding ofPFBA toxicokinetics remain highlighted in the revised
assessment as an area deserving of additional research.
Comment: One reviewer noted that, while hepatocellular hypertrophy exhibited exposure
duration dependence when comparing 28- and 90-day results, there were no apparent
increased sensitivity with longer exposure durations for liver weight or thyroid hormone
measures. This reviewer commented that a UFs of 10 for liver weight and thyroid hormones
may then not be justified and their corresponding candidate RfDs may be unjustifiably low.
This reviewer suggested that the EPA consider additional justification for the selection of
the UFs for these endpoints.
EPA Response: Although no increase of effect was noted in liver weight endpoints or
thyroid hormone levels when comparing short-term (28-day) and subchronic (90-day
exposures), this is not sufficient evidence to conclude that effects would not worsen at the
same exposure level or become evident at lower exposure levels with chronic exposure. It
should be noted that the increase in exposure duration is approximately 8-fold when
comparing chronic exposures to subchronic exposures and only 3-fold when comparing
subchronic exposures to short-term exposures. A short discussion of this matter has been
added to Section 5.2.1 of the assessment and a value of UFs = 10 is retained for all non-
developmental toxicity endpoints.
Comment: One reviewer commented that the UFa of 3 did not account for interspecies
differences in thyroid toxicity or developmental effects related to thyroid hormone
insufficiency. This reviewer suggested that the EPA consider increasing the UFa from 3 to
10.
EPA Response: As noted in Table 5-5, after application of the chemical-specific DAF (based
explicitly on differences in rodent and human serum clearance values), residual uncertainty
regarding toxicodynamics remains and is accounted for in the application of a UFa= 3. EPA
is unaware of any chemical-specific data that would support increasing the UFa to 10 to
account for differences in toxicodynamics regarding PFBA-induced thyroid or
developmental effects.
E-41
-------
Supplemental Information ofPFBA and Related Salts
E.12.3. Public Comments
Comment: The draft assessment applies a total uncertainty factor of 1,000 to generate the
chronic reference dose (RfD). This includes a 10-fold adjustment for a subchronic-to-
chronic exposure adjustment (UFS), and a 3-fold adjustment for database uncertainty
(UFD). The data from the study by Butenhoff et al. suggest, however, that PFBA levels have
reached steady state conditions in rat livers after 28 days. As a result, the draft assessment
notes that "[i]ncreased duration of exposure might not elicit increased effects in the target
tissue." In the absence of a chronic study, however, the draft concludes that liver effects may
increase with prolonged exposure and that includes a UFS of 10. EPA's conclusion is based
on evidence that hepatocellular hypertrophy was observed at lower doses after 90 days
when compared to the 28-day results, despite the fact that the lowest observed adverse
effect level (LOAEL) for increased liver weight (likely resulting from hypertrophy) was the
same in both the 28 and 90 day study and the clinical chemistry was inconsistent across
endpoints and durations of exposure. EPA's conclusion is also not supported by the studies
with perfluorooctanoic acid (PFOA), a structurally similar compound. While hypertrophy
was observed in chronic studies of male rats exposed to PFOA, the concentrations required
were 10-fold higher than those eliciting a response in the subchronic studies. Consequently,
the addition of a UFS of 10 for liver effects is not supported by the available evidence. While
a study of chronic exposure to PFBA is not available, the chronic data from studies with
PFOA indicate that a UFS of 1 is more appropriate.
EPA Response: For PFBA, there is chemical-specific data that demonstrates, for some
endpoints (i.e., hepatocellular hypertrophy), an increase in duration results in observation
of effects at lower doses. While this data is missing for other effects (liver weight, thyroid
hormone levels), it is consistent with EPA guidelines U.S. EPA f20021 to apply a UFs = 10 to
account for the possibility of increased effects at lower doses when considering chronic
exposures. Application of a UFs = 10 for liver effects in adult animals is consistent with other
recent EPA assessments of PFAS, including GenX U.S. EPA (2021). with the justification for
this decision documented in Section 5.2.1 of the revised assessment.
Comment: For the candidate RfDs for both chronic and subchronic effects, the draft
assessment includes a UFD of 3 based on concern for neurodevelopmental effects
independent of a thyroid hormone-related mechanism. Agency guidance explains, however,
that a database uncertainty factor is applied when reproductive and developmental toxicity
studies are missing since they have been found to provide useful information for
establishing the lowest no adverse effect level. The guidance notes that, for a reference dose
(RfD) based on animal data, a factor of 3 is often applied if either a prenatal toxicity study or
a two-generation reproduction study is missing, or a factor of 10 may be applied if both are
missing. In deciding whether to apply an UFD, EPA advises that the assessor should
E-42
-------
Supplemental Information ofPFBA and Related Salts
consider both the data lacking and the data available for a particular organ system as well
as life stages. As noted in the draft assessment, a high confidence developmental study is
available for PFBA; the draft also notes that the lack of a multigenerational reproductive
study "is not considered a major concern." Therefore, EPA's proposal to add an uncertainty
factor to address concern about neurodevelopmental effects is not supported by its own
analysis.
EPA Response: EPA's A Review of the Reference Dose and Reference Concentration Processes
(U.S. EPA. 20021 states that the "database UF is intended to account for the potential for
deriving an underprotective RfD/RfC as a result of an incomplete characterization of the
chemical's toxicity." This comment itself acknowledges that the RfD/RfC document
recommends "...the assessor should consider both the data lacking and the data available
for particular organ systems as well as life stages" when determining the value of the UFd.
Therefore, it is wrong to conclude that the recommendations for application of the UFd state
that this uncertainty factor is intended to only account for the lack of developmental or
reproductive studies. Given residual uncertainties regarding the potential for reproductive,
developmental neurotoxicity, immunotoxicity, or mammary gland effects, a UFd = 3 is
applied and this is consistent with EPA's RfD/RfC recommendations.
Comment: Uncertainties are not appropriately accounted for. In the recent Human Health
Toxicity Values derivation for PFBS, EPA included a database uncertainty factor of 10, citing
a lack of chronic studies and neurodevelopmental and immunotoxicity studies as well as a
lack of mammary gland studies. The same deficits were noted by EPA for PFBA. It is
therefore unclear why EPA drew a different conclusion in the draft toxicological review of
PFBA, deciding to only apply a partial database uncertainty factor of 3.
EPA Response: The rationale for the selected UFd = 3 in the PFBA assessment is
extensively discussed in Section 5.2.1 and the considerations that inform the final selection
of UFd = 3 (lack of developmental neurotoxicity, immunotoxicity, and mammary gland
toxicity studies) are consistent with the rationale for a higher UFd provided in the PFBS
assessment The PFBS assessment additionally considers the lack of a chronic study in the
final selection of a UFd = 10; this is considered and, as necessary, addressed by the UFs in the
PFBA assessment For PFBA, the selection of UFd = 3 was supported by the external peer
reviewers and is retained in the revised assessment
Comment: Biomonitoring studies demonstrate that Americans have chronic exposure to
multiple PFAS chemicals throughout their lifetimes. Therefore, it is impossible to be
exposed to PFBA and no other PFAS chemicals. CDC's NHANES studies reveal that nearly
every American has detectable concentrations of four PFAS chemicals in their bloodstream
(PFOS, PFOA, PFHxS and PFNA). Multiple other PFAS have been detected in NHANES and
E-43
-------
Supplemental Information ofPFBA and Related Salts
state biomonitoring programs. Toxicity assessment (i.e, the PFBA assessment) should
account for simultaneous exposure to other PFAS chemicals that impact the same target
organs. EPA must promote similar assessments for other PFAS related health outcomes
with potential for additive toxicity, including kidney and liver toxicity, lipid metabolism,
birth outcomes, immunotoxicity and developmental effects. At the very least, EPA should
add an additional uncertainty factor to account for the high likelihood of additive effects
with other PFAS.
EPA Response: The PFBA assessment derives organ or system-specific reference doses
and states "... these toxicity values might be useful in some contexts (e.g., when assessing
the potential cumulative effects of multiple chemical exposures occurring simultaneously)."
Therefore, when assessing cumulative risk, the values presented in the PFBA assessment
can be used by risk assessors in conjunction with values in other PFAS human health risk
assessments (and separately conducted exposure assessments) to account for exposures to
multiple PFAS simultaneously.
E.13. CHARGE QUESTION 11 - CANCER TOXICITY VALUES
Given the conclusion there was inadequate evidence to assess carcinogenic potential for PFBA (Charge
Question 5), the Toxicological Review does not derive quantitative estimates for cancer effects for oral
or inhalation exposures. Is this decision scientifically justified?
E.13.1. Overarching External Peer Reviewer Comments on Cancer Toxicity Values
"All reviewers concurred with the decision to not derive quantitative estimates for cancer
effects due to inadequate evidence to assess carcinogenic potential for PFBA. Additionally, [one
reviewer] commented that "no robust scientific foundation has been laid, critically reviewed and
broadly accepted by the scientific community for the use of any surrogate PFAS with
carcinogenicity data (e.g., PFOA) for this purpose.""
E.13.2. Tier 1 Recommendations
Reviewers had no Tier 1 recommendations.
E.13.3. Tier 2 Suggestions
Reviewers had no Tier 2 suggestions.
E-44
-------
Supplemental Information ofPFBA and Related Salts
APPENDIX F. QUALITY ASSURANCE
FOR THE IRIS
TOXICOLOGICAL REVIEW OF
PERFLUOROBUTANOIC ACID AND
RELATED
COMPOUND AMMONIUM
PERFLUOROBUTANOATE
This assessment was prepared under the auspices of the U.S. Environmental Protection
Agency's (EPA's) Integrated Risk Information System (IRIS) Program. The IRIS Program is housed
within the Office of Research and Development (ORD) in the Center for Public Health and
Environmental Assessment (CPHEA). EPA has an agency-wide quality assurance policy and that
policy is outlined in the EPA Quality Manual for Environmental Programs (see CIO 2105-P-01.11 and
follows the specifications outlined in EPA Order CIO 2105.1.
As required by CIO 2105.1, ORD maintains a Quality Management Program, which is
documented in an internal Quality Management Plan (QMP). The latest version was developed in
2013 using Guidance for Developing Quality Systems for Environmental Programs (OA/G-11. An
NCEA/CPHEA-specific QMP also was developed in 2013 as an appendix to the ORD QMP. Quality
assurance for products developed within CPHEA is managed under the ORD QMP and applicable
appendices.
The IRIS Toxicological Review ofPerfluorobutanoicAcid and Related Salts has been
designated as Influential Scientific Information (ISI) and is classified as QA Category A. Category A
designations require reporting of all critical QA activities, including audits. IRIS assessments are
developed through a seven-step process. Documentation of this process is available on the IRIS
website: https://www.epa.gov/iris/basic-information-about-integrated-risk-information-
svstem#process.
Specific management of quality assurance within the IRIS Program is documented in a
Programmatic Quality Assurance Project Plan (PQAPP). A PQAPP was developed using the EPA
Guidance for Quality Assurance Project Plans (OA/G-51. and the latest approved version is dated June
2022. All IRIS assessments follow the IRIS PQAPP, and all assessment leads and team members are
required to receive QA training on the IRIS PQAPP. During assessment development, additional
QAPPs may be applied for quality assurance management. They include:
F-l
-------
Supplemental Information ofPFBA and Related Salts
Title
Document Number
Date
Program Quality Assurance Project
Plan (PQAPP) for the Integrated Risk
Information System (IRIS) Program
L-CPAD-0030729-QP-1-5
June 2022
An Umbrella Quality Assurance
Project Plan (QAPP) for Dosimetry
and Mechanism-Based Models
(PBPK)
L-CPAD-0032188-QP-1-2
December 2020
Quality Assurance Project Plan
(QAPP) for Enhancements to
Benchmark Dose Software (BMDS)
L-HEEAD-0032189-QP-1-2
September 2020
Umbrella Quality Assurance Project
Plan for CPHEA PFAS Toxicity
Assessments
L-CPAD-0031652-QP-1-4
October 2021
During assessment development, this project underwent four quality audits during
assessment development including:
Date
Type of audit
Major findings
Actions taken
August 2022
Technical System Audit
No findings
None
July 2021
Technical System Audit
No findings
None
August 2020
Technical System Audit
No findings
None
August 2019
Technical System Audit
No findings
None
During Step 3 of the IRIS Process, the IRIS Toxicological Review was subjected to external
reviews by other federal agency partners including the Executive Offices of the White House.
Comments during these IRIS Process steps are available in the Docket EPA-HQ-ORD-2020-0675 on
http:/www. regulations.gov.
During Step 4 assessment development, the IRIS Toxicological Review of Perfluorobutanoic
Acid and Related underwent public comment from August 23, 2021, to November 8, 2021.
Following this comment period, the toxicological review underwent external peer review by a
contractor-led panel performed by ERG from October 2021 to June 2022. The peer-review report is
available on the peer review website. All public and peer-review comments are available in the
docket EPA-HQ-ORD-2020-0675.
Prior to release (Step 7 of the IRIS process), the final toxicological review is submitted to
management and QA clearance. During this step the CPHEA QA Director and QA Managers review
the project QA documentation and ensure that EPA QA requirements are met.
F-2
-------
Supplemental Information ofPFBA and Related Salts
REFERENCES
Abraham. K: El-Khatib. AH: Schwerdtle. T: Monien. BH. (2021). Perfluorobutanoic acid (PFBA): No
high-level accumulation in human lung and kidney tissue. Int J Hyg Environ Health 237:
113830. http://dx.doi.org/10.1016/i.iiheh.2021.113830.
Bangma. IT: Reiner. 1: Fry. RC: Manuck. T: Mccord. 1: Strvnar. Ml. (2021). Identification of an
Analytical Method Interference for Perfluorobutanoic Acid in Biological Samples. Environ
SciTechnol Lett 8: 1085-1090. http://dx.doi.org/10.1021/acs.estlett.lc00901.
Bassett. THD: Clement-Lacroix. P: Samarut. 1: Chassande. 0: Williams. GR: O'Shea. PI:
Sriskantharaiah. S: Rabier. B: Bovde. A: Howell. PGT: Weiss. R. ovE: Roux. TP: Malaval. L. uc.
(2007). Thyroid hormone excess rather than thyrotropin deficiency induces osteoporosis in
hyperthyroidism. Mol Endocrinoll095-1107. http://dx.doi.org/10.1210/me.2007-0033.
Bizzarro. Ml: Gross. I. (2004). Effects of hormones on fetal lung development [Review], Obstet
Gynecol Clin North Am 31: 949-961, xii. http://dx.doi.Org/10.1016/i.ogc.2004.08.001.
Butenhoff. TL: Biork. TA: Chang. SC: Ehresman. DT: Parker. GA: Das. K: Lau. C: Lieder. PH: van
Otterdiik. FM: Wallace. KB. (2012). Toxicological evaluation of ammonium
perfluorobutyrate in rats: twenty-eight-day and ninety-day oral gavage studies. Reprod
Toxicol 33: 513-530. http: / /dx. do i. or g /10.1016 /i. r ep r otox. 2 011.0 8.0 0 4.
Carlson. LA: Angrish. M: Shirke. AY: Radke. EG: Schulz. B: Kraft. A: Tudson. R: Patlewicz. G: Blain. R:
Lin. C: Vetter. N: Lemeris. C: Hartman. P: Hubbard. H: Arzuaga. X: Davis. A: Dishaw. LV:
Druwe. IL: Hollinger. H: Tones. R: Kaiser. TP: Lizarraga. L: Noves. PD: Taylor. M: Shapiro. AT:
Williams. AT: Thayer. KA. (2022). Systematic evidence map for over one hundred and fifty
per- and polyfluoroalkyl substances (PFAS) [Review], Environ Health Perspect 130: 56001.
http: //dx.doi.org/10.1289 /EHP10343.
Chang. S: Das. K: Ehresman. DT: Ellefson. ME: Gorman. GS: Hart. TA: Noker. PE: Tan. Y: Lieder. PH:
Lau. C: Olsen. GW: Butenhoff. TL. (2008). Comparative pharmacokinetics of
perfluorobutyrate in rats, mice, monkeys, and humans and relevance to human exposure via
drinking water. Toxicol Sci 104: 40-53. http://dx.doi.org/10.1093/toxsci/kfn057.
Cooper. PS. (2001). Clinical practice. Subclinical hypothyroidism [Review], N Engl J Med 345: 260-
265. http://dx.doi.Org/10.1056/NETM200107263450406.
Crofton. KM: Craft. ES: Hedge. TM: Gennings. C: Simmons. IE: Carchman. RA: Carter. WH. Tr: Devito.
MT. (2005). Thyroid-hormone-disrupting chemicals: Evidence for dose-dependent additivity
or synergism. Environ Health Perspect 113: 1549-1554.
http: / /dx. doi. o r g /10.12 89 /ehp. 819 5.
Crofton. KM: Zoeller. RT. (2005). Mode of action: neurotoxicity induced by thyroid hormone
disruption during development-hearing loss resulting from exposure to PHAHs [Review],
Crit Rev Toxicol 35: 757-769. http://dx.doi.org/10.1080/10408440591007304.
Crump. KS. (1995). Calculation of benchmark doses from continuous data. Risk Anal 15: 79-89.
http: / /dx.doi.org/10.1111 /i.l 539-6924.1995.tb00095.x.
Das. KP: Grey. BE: Zehr. RD: Wood. CR: Butenhoff. TL: Chang. SC: Ehresman. DT: Tan. YM: Lau. C.
(2008). Effects of perfluorobutyrate exposure during pregnancy in the mouse. Toxicol Sci
105: 173-181. http://dx.doi.org/10.1093/toxsci/kfn099.
Foreman. IE: Chang. SC: Ehresman. DT: Butenhoff. TL: Anderson. CR: Palkar. PS: Kang. BH: Gonzalez.
FT: Peters. TM. (2009). Differential Hepatic Effects of Perfluorobutyrate Mediated by Mouse
and Human PPAR-alpha. Toxicol Sci 110: 204-211.
http: / /dx. doi. o r g /10.10 9 3 /toxsci /kfp077.
R-l
-------
Supplemental Information ofPFBA and Related Salts
Gilbert. ME: O'Shaughnessv. KL: Axelstad. M. (2020). Regulation of Thyroid-disrupting Chemicals to
Protect the Developing Brain [Review], Endocrinology 161.
http://dx.doi.org/10.1210/endocr/bqaal06.
Gilbert. ME: Ramos. RL: Mccloskev. DP: Goodman. TH. (2014). Subcortical band heterotopia in rat
offspring following maternal hypothyroxinaemia: Structural and functional characteristics. J
Neuroendocrinol 26: 528-541. http://dx.doi.org/10.Ill 1 /ine. 12169.
Gilbert. ME: Sui. L. (2008). Developmental exposure to perchlorate alters synaptic transmission in
hippocampus of the adult rat. Environ Health Perspect 116: 752-760.
http://dx.d0i.0rg/l 0.12 89 /ehp. 11089.
Haddow. IE: Palomaki. GE: Allan. WC: Williams. TR: Knight. GT: Gagnon. 1: O'Heir. CE: Mitchell. ML:
Hermos. RT: Waisbren. SE: Faix. ID: Klein. RZ. (1999). Maternal thyroid deficiency during
pregnancy and subsequent neuropsychological development of the child. N Engl J Med 341:
549-555. http://dx.d0i.0rg/l 0.1056/NETM199908193410801.
Hall. AP: Elcombe. CR: Foster. TR: Harada. T: Kaufmann. W: Knippel. A: Kiittler. K: Malarkev. DE:
Maronpot. RR: Nishikawa. A: Nolte. T: Schulte. A: Strauss. V: York. Ml. (2012). Liver
hypertrophy: a review of adaptive (adverse and non-adverse) changes-conclusions from
the 3rd International ESTP Expert Workshop [Review], Toxicol Pathol 40: 971-994.
http://dx.doi.org/10.1177/0192623312448935.
Tansen. TA: Korevaar. TIM: Mulder. TA: White. T: Muetzel. RL: Peeters. RP: Tiemeier. H. (2019).
Maternal thyroid function during pregnancy and child brain morphology: a time window-
specific analysis of a prospective cohort. Lancet Diabetes Endocrinol 7: 629-637.
http://dx.doi.org/10.1016/S2213-8587ri9130153-6.
Korevaar. TIM: Muetzel. R: Medici. M: Chaker. L: Taddoe. VWV: de Riike. YB: Steegers. EAP: Visser.
TT: White. T: Tiemeier. H: Peeters. RP. (2016). Association of maternal thyroid function
during early pregnancy with offspring IQ and brain morphology in childhood: A population-
based prospective cohort study. Lancet Diabetes Endocrinol 4: 35-43.
http://dx.d0i.0rg/l 0.1016/S2213-8587C15100327-7.
Lau. C: Thibodeaux. TR: Hanson. RG: Narotskv. MG: Rogers. TM: Lindstrom. AB: Strvnar. Ml. (2006).
Effects of perfluorooctanoic acid exposure during pregnancy in the mouse. Toxicol Sci 90:
510-518. http: //dx.doi.org/10.1093 /toxsci/kfi 105.
Levie. D: Korevaar. TIM: Bath. SC: Dalmau-Bueno. A: Murcia. M: Espada. M: Dineva. M: Ibarluzea. TM:
Sunver. T: Tiemeier. H: Rebagliato. M: Ravman. MP: Peeters. RP: Guxens. M. (2018). Thyroid
Function in Early Pregnancy, Child IQ, and Autistic Traits: A Meta-Analysis of Individual
Participant Data. J Clin Endocrinol Metab 103: 2967-2979.
http://dx.doi.0rg/lO.i2iO/ic.2Ol8-OO224.
Li. Y: Fletcher. T: Mucs. D: Scott. K: Lindh. CH: Tallying. P: Takobsson. K. (2018). Half-lives of PFOS,
PFHxS and PFOA after end of exposure to contaminated drinking water. Occup Environ Med
75: 46-51. http://dx.doi.org/10.1136/oemed-2017-104651.
Mochizuki. K: Yagi. E: Sakaguchi. N: Mochizuki. H: Takabe. S: Kuranuki. S: Suzuki. T: Shimada. M:
Goda. T. (2007). The critical period for thyroid hormone responsiveness through thyroid
hormone receptor isoform alpha in the postnatal small intestine. Biochim Biophys Acta
1770: 609-616. http://dx.d0i.0rg/l0.1016/i.bbagen.2006.12.011.
Nakamizo. M: Tovabe. S: Asami. T: Akazawa. K. (2007). Mental development of infants with
congenital hypothyroidism: a longitudinal study. Clin Pediatr (Phila) 46: 53-58.
http://dx.doi.Org/10.1177/0009922806289326.
Oerbeck. B. (2007). Young adults with severe congenital hypothroidism: Cognitive event related
potentials (ERPs) and the significance of an early start of thyroxine treatment Scand J
Psychol 48: 61-67. http://dx.doi.Org/10.llll/i.1467-9450.2006.00545.x.
R-2
-------
Supplemental Information ofPFBA and Related Salts
Page. M: McKenzie. T: Bossuvt. P: Boutron. I: Hoffmann. T: Mulrow. Cea. (2021). The PRISMA 2020
statement: an updated guideline for reporting systematic reviews. BMJ Acad Ed 372: 1-9.
http://dx.doi.org/10.1136/bmi.n71.
Pelch. KE: Reade. A: Kwiatkowski. CF: Merced-Nieves. FM: Cavalier. H: Schultz. K: Wolffe. T:
Varshavskv. 1. (2022). The PFAS-Tox Database: A systematic evidence map of health studies
on 29 per- and polyfluoroalkyl substances. Environ Int 167: 107408.
http://dx.doi.Org/10.1016/i.envint.2022.107408.
Perez. F: Nadal. M: Navarro-Ortega. A: Fabrega. F: Domingo. 1L: Barcelo. D: Farre. M. (2013).
Accumulation of perfluoroalkyl substances in human tissues. Environ Int 59: 354-362.
http://dx.doi.Org/10.1016/i.envint.2013.06.004.
Radke. E: Wright. MI: Christensen. K: Lin. CI: Goldstone. AE: Glenn. B: Thayer. K. (2022).
Epidemiology evidence for health effects of 150 per- and polyfluoroalkyl substances: A
systematic evidence map [Review], Environ Health Perspect 130: 96003.
http://dx.doi.org/10.1289/EHP11185.
Russell. MH: Himmelstein. MW: Buck. RC. (2015). Inhalation and oral toxicokinetics of 6:2 FTOH
and its metabolites in mammals. Chemosphere 120:328-335.
http://dx.doi.Org/10.1016/i.chemosphere.2014.07.092.
Stagnaro-Green. A: Rovet. 1. (2016). Pregnancy: Maternal thyroid function in pregnancy - A tale of
two tails [Comment], Nat Rev Endocrinol 12: 10-11.
http://dx.doi.org/10.1038/nrendo.2015.212.
U.S. EPA (U.S. Environmental Protection Agency). (1991). Guidelines for developmental toxicity risk
assessment Fed Reg 56: 63798-63826.
U.S. EPA (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicity risk
assessment [EPA Report] (pp. 1-89). (ISSN 0097-6326
EISSN 2167-2520
EPA/630/R-95/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment
Forum, http://www.epa.gov/risk/guidelines-neurotoxicity-risk-assessment.
U.S. EPA (U.S. Environmental Protection Agency). (2002). A review of the reference dose and
reference concentration processes. (EPA630P02002F). Washington, DC.
https://www.epa.gov/sites/production/files/2014-12/documents/rfd-final.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2005). Guidelines for carcinogen risk assessment
[EPA Report], (EPA630P03001F). Washington, DC.
https://www.epa.gov/sites/production/files/2013-
09/documents/cancer guidelines final 3-25-05.pdf.
U.S. EPA fU.S. Environmental Protection Agency). (2011). Recommended use ofbody weight 3/4 as
the default method in derivation of the oral reference dose. (EPA100R110001). Washington,
DC. https://www.epa.gov/sites/production/files/2013-09/documents/recommended-use-
of-bw34.pdf.
U.S. EPA (U.S. Environmental Protection Agency). (2012). Benchmark dose technical guidance [EPA
Report], (EPA100R12001). Washington, DC: U.S. Environmental Protection Agency, Risk
Assessment Forum, https: //www.epa.gov/risk/benchmark-dose-technical-guidance.
U.S. EPA (U.S. Environmental Protection Agency). (2015). Peer review handbook [EPA Report] (4th
ed.). (EPA/100/B-15/001). Washington, DC: U.S. Environmental Protection Agency, Science
Policy Council, https: //www.epa.gov/osa/peer-review-handbook-4th-edition-2015.
U.S. EPA (U.S. Environmental Protection Agency). (2021). Human health toxicity values for
hexafluoropropylene oxide (HFPO) dimer acid and its ammonium salt (CASRN 13252-13-6
and CASRN 62037-80-3). Also known as "GenX chemicals." Final report [EPA Report], (EPA-
822R-21-010). Washington, DC: U.S. Environmental Protection Agency, Office of Water.
R-3
-------
Supplemental Information ofPFBA and Related Salts
van Otterdiik. FM. (2007a). Repeated dose 28-day oral toxicity study with MTDID-8391 by daily
gavage in the rat, followed by a 21-day recovery period. (Study Number 06-226).
Maple wood, MN: 3M.
van Otterdiik. FM. (2007b). Repeated dose 90-day oral toxicity study with MTDID 8391 by daily
gavage in the rat followed by a 3-week recovery period. (Study Number 06-398).
Maple wood, MN: 3M.
Weatherlv. LM: Shane. HL: Lukomska. E: Baur. R: Anderson. SE. (2021). Systemic toxicity induced
by topical application of heptafluorobutyric acid (PFBA) in a murine model. Food Chem
Toxicol 156: 112528. http: //dx.doi.org/10.1016/i.fct.2021.112528.
Wexler. TA: Sharretts. 1. (2007). Thyroid and bone [Review], Endocrinol Metab Clin North Am 36:
673-705, vi. http://dx.doi.Org/10.1016/i.ecl.2007.04.005.
Zoeller. RT: Rovet. 1. (2004). Timing of thyroid hormone action in the developing brain: Clinical
observations and experimental findings [Review], J Neuroendocrinol 16: 809-818.
http://dx.doi.Org/10.llll/j.1365-2826.2004.01243.
------- |