c EDA
WtrM
April 2024
EPA Document No. 815R24009
FINAL
APPENDIX: Human Health Toxicity Assessment for
Perfluorooctane Sulfonic Acid (PFOS) and Related Salts
-------
APRIL 2024
FINAL
APPENDIX: Human Health Toxicity Assessment for Perfluorooctane Sulfonic
Acid (PFOS) and Related Salts
Prepared by:
U.S. Environmental Protection Agency
Office of Water (4304T)
Health and Ecological Criteria Division
Washington, DC 20460
EPA Document Number: 815R24009
April 2024
-------
APRIL 2024
Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection Agency
(EPA) policy and approved for publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
i
-------
APRIL 2024
Contents
Disclaimer i
Contents ii
Figures vi
Tables xii
Acronyms and Abbreviations xxiii
Appendix A. Systematic Review Protocol for Updated PFOS Toxicity Assessment A-l
A. 1 Overview of Background Information and Systematic Review Protocol A-2
A. 1.1 Summary of Background Information A-2
A. 1.2 Problem Formulation A-3
A. 1.3 Overall Objective and Specific Aims A-5
A. 1.4 Populations, Exposures, Comparators, and Outcomes (PECO) Criteria A-7
A. 1.5 Literature Search A-10
A. 1.6 Literature Screening Process to Target Dose-Response Studies and PK
Models A-21
A. 1.7 Study Quality Evaluation Overview A-52
A. 1.8 Data Extraction for Epidemiological Studies A-98
A. 1.9 Data Extraction for Animal Toxicological Studies A-107
A. 1.10 Evidence Synthesis and Integration A-l 12
A. 1.11 Dose-Response Assessment: Selecting Studies and Quantitative
Analysis A-l 17
A.2 Meta-Analysis Table A-128
A.3 Studies Identified In Supplemental Literature Search Assessment A-134
A.4 Studies Identified After Assessment Literature Searches A-150
Appendix B. Detailed Toxicokinetics B-l
B.l Absorption B-l
B.l.l Cellular Uptake B-l
B.1.2 Oral Exposure B-2
B. 1.3 Inhalation Exposure B-2
B.1.4 Dermal Exposure B-2
B.1.5 Developmental Exposure B-2
B. 1.6 Bioavailability B-2
B.2 Distribution B-3
B.2.1 Protein Binding B-4
B.2.2 Tissue Distribution B-6
B.2.3 Distribution During Reproduction and Development B-17
B.2.4 Volume of Distribution B-41
B.3 Metabolism B-47
B.4 Excretion B-48
ii
-------
APRIL 2024
B.4.1 Urinary and Fecal Excretion B-48
B.4.2 Physiological and Mechanistic Factors Impacting Excretion B-50
B.4.3 Maternal Elimination Through Lactation and Fetal Partitioning B-52
B.4.4 Other Routes of Elimination B-54
B.4.5 Half-life Data B-56
Appendix C. Nonpriority Health Systems Evidence Synthesis and Integration C-l
C.l Reproductive C-l
C.l.l Human Evidence Study Quality Evaluation and Synthesis C-l
C.1.2 Animal Evidence Study Quality Evaluation and Synthesis C-l3
C.1.3 Mechanistic Evidence C-22
C.1.4 Evidence Integration C-23
C.2 Endocrine C-42
C.2.1 Human Evidence Study Quality Evaluation and Synthesis C-42
C.2.2 Animal Evidence Study Quality Evaluation and Synthesis C-50
C.2.3 Mechanistic Evidence C-70
C.2.4 Evidence Integration C-70
C.3 Metabolic/Systemic C-76
C.3.1 Human Evidence Study Quality Evaluation and Synthesis C-76
C.3.2 Animal Evidence Study Quality Evaluation and Synthesis C-93
C.3.3 Mechanistic Evidence C-102
C.3.4 Evidence Integration C-103
C.4 Nervous C-l 11
C.4.1 Human Evidence Study Quality Evaluation and Synthesis C-l 11
C.4.2 Animal Evidence Study Quality Evaluation and Synthesis C-l 19
C.4.3 Mechanistic Evidence C-128
C.4.4 Evidence Integration C-128
C.5 Renal C-l38
C.5.1 Human Evidence Study Quality Evaluation and Synthesis C-138
C.5.2 Animal Evidence Study Quality Evaluation and Synthesis C-143
C.5.3 Mechanistic Evidence C-146
C.5.4 Evidence Integration C-147
C.6 Hematological C-l 53
C.6.1 Human Evidence Study Quality Evaluation and Synthesis C-153
C.6.2 Animal Evidence Study Quality Evaluation and Synthesis C-156
C.6.3 Mechanistic Evidence C-158
C.6.4 Evidence Integration C-159
C.l Respiratory C-l63
C.7.1 Human Evidence Study Quality Evaluation and Synthesis C-163
C.7.2 Animal Evidence Study Quality Evaluation and Synthesis C-165
C.l3 Mechanistic Evidence C-168
C.7.4 Evidence Integration C-168
-------
APRIL 2024
C.8 Musculoskeletal C-172
C.8.1 Human Evidence Study Quality Evaluation and Synthesis C-172
C.8.2 Animal Evidence Study Quality Evaluation and Synthesis C-175
C.8.3 Mechanistic Evidence C-176
C.8.4 Evidence Integration C-177
C.9 Gastrointestinal C-181
C.9.1 Human Evidence Study Quality Evaluation and Synthesis C-181
C.9.2 Animal Evidence Study Quality Evaluation and Synthesis C-183
C.9.3 Mechanistic Evidence C-185
C.9.4 Evidence Integration C-185
C. 10 Dental C-188
C.10.1 Human Evidence Study Quality Evaluation and Synthesis C-188
C.10.2 Animal Evidence Study Quality Evaluation and Synthesis C-190
C.10.3 Mechanistic Evidence C-190
C.10.4 Evidence Integration C-190
C.ll Ocular C-192
C.ll.l Human Evidence Study Quality Evaluation and Synthesis C-192
C.11.2 Animal Evidence Study Quality Evaluation and Synthesis C-193
C.11.3 Mechanistic Evidence C-194
C. 11.4 Evidence Integration C-195
C. 12 Dermal C-197
C.12.1 Human Evidence Study Quality Evaluation and Synthesis C-197
C.12.2 Animal Evidence Study Quality Evaluation and Synthesis C-198
C.12.3 Mechanistic Evidence C-199
C.12.4 Evidence Integration C-199
Appendix D. Detailed Information from Epidemiology Studies D-l
D.l Developmental D-l
D.2 Reproductive D-50
D.2.1 Male D-50
D.2.2 Female D-61
D.3 Hepatic D-71
D.4 Immune D-79
D.5 Cardiovascular D-120
D.5.1 Cardiovascular Endpoints D-120
D.5.2 Serum Lipids D-136
D.6 Endocrine D-l64
D.7 Metabolic/Systemic D-l73
D.8 Nervous D-191
D.9 Renal D-214
D.10 Hematological D-223
D. 11 Respiratory D-225
iv
-------
APRIL 2024
D.12 Musculoskeletal D-227
D.13 Gastrointestinal D-232
D. 14 Dental D-234
D.15 Ocular D-234
D.16 Dermal D-235
D.17 Cancer D-236
Appendix E. Benchmark Dose Modeling E-l
E.l Epidemiology Studies E-l
E. 1.1 Modeling Results for Immunotoxicity E-2
E.l.2 Modeling Results for Decreased Birthweight E-24
E.l.3 Modeling Results for Liver Toxicity E-32
E.l.4 Modeling Results for Increased Cholesterol E-43
E.2 Toxicology Studies E-53
E.2.1 Butenhoff et al. (2012)/Thorn lord (2002) E-53
11.2.2 I.eeetal. (2015) 11-69
11.2.3 Luebker et al. (2005b) 11-70
E.2.4NTP (2019) 11-76
11.2.5 Zhong et al. (2016) 11-80
11.2.6 Lau et al. (2003) 11-82
E.2.7 Luebker et al. (2005a) E-84
Appendix F. Pharmacokinetic Modeling F-l
F.l Comparison of Fits to Training Datasets Used in Wambaugh et al. (2013) F-l
F.2 Visual Inspection of Test Datasets not Used for Initial Fitting F-4
F.3 Human Model Validation F-7
Appendix G. Relative Source Contribution G-l
G. 1 Background G-l
G.2 Literature Review G-2
G.2.1 Systematic Review G-2
G.2.2 Evidence Mapping G-3
G.3 Summary of Potential PFOS Sources G-3
G.3.1 Dietary Sources G-4
G.3.2 Consumer Product Uses G-9
G.3.3 Indoor Dust G-10
G.3.4 Ambient Air G-10
G.3.5 Other Possible Exposure Sources G-l 1
G.4 Recommended RSC G-l 1
References R-l
v
-------
APRIL 2024
Figures
Figure A-l. Overview of Study Quality Evaluation Approach A-52
Figure B-l. Summary ofPFOS Absorption Studies B-l
Figure B-2. Summary ofPFOS Distribution Studies B-3
Figure B-3. Summary ofPFOS Metabolism Studies B-47
Figure B-4. Summary ofPFOS Excretion Studies B-48
Figure C-l. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Male Reproductive Effects C-3
Figure C-2. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Female Reproductive Effects C-8
Figure C-3. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Female Reproductive Effects (Continued) C-9
Figure C-4. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies ofPFOS Exposure and Reproductive Effects C-14
Figure C-5. Gestation Length in Rats Following Exposure to PFOS C-15
Figure C-6. Sperm Parameters in Male Rodents Following Exposure to PFOS C-16
Figure C-l Percent Change in Testosterone Levels Relative to Controls in Male Rodents
and Non-Human Primates Following Exposure to PFOS C-17
Figure C-8. Percent Change in Estradiol Levels Relative to Controls in Male Rodent and
Non-Human Primates Following Exposure to PFOS C-18
Figure C-9. Percent Change in LH and Prolactin Levels Relative to Controls in Male Rats
Following Exposure to PFOS C-19
Figure C-10. Percent Change in Prolactin-Family Hormone Levels Relative to Controls in
Female Mice Following Exposure to PFOS C-19
Figure C-l 1. Percent Change in Estradiol and Testosterone Levels Relative to Controls in
Female Rodents and Non-Human Primates Following Exposure to PFOS C-20
Figure C-12. Summary of Mechanistic Studies ofPFOS and Reproductive Effects C-23
Figure C-13. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Endocrine Effects C-45
Figure C-14. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Endocrine Effects (Continued) C-46
Figure C-15. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies ofPFOS Exposure and Endocrine Effects C-51
vi
-------
APRIL 2024
Figure C-16. Percent Change in Adrenal Hormones Relative to Controls in Rodents
Following Exposure to PFOSa'b C-67
Figure C-17. Summary of Mechanistic Studies of PFOS and Endocrine Effects C-70
Figure C-18. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects C-79
Figure C-19. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects (Continued) C-80
Figure C-20. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects (Continued) C-81
Figure C-21. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Metabolic Effects C-94
Figure C-22. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Systemic Effects21 C-97
Figure C-23. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Systemic Effects (Continued)21 C-98
Figure C-24. Effects on Body Weight in Rodents and Non-Human Primates Following
Exposure to PFOS (Logarithmic Scale) C-101
Figure C-25. Summary of Mechanistic Studies of PFOS and Metabolic Effects C-102
Figure C-26. Summary of Mechanistic Studies of PFOS and Systemic Effects C-103
Figure C-27. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Neurological Effects C-l 13
Figure C-28. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Neurological Effects (Continued) C-l 14
Figure C-29. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Nervous Effects C-120
Figure C-30. Summary of Mechanistic Studies of PFOS and Nervous Effects C-128
Figure C-31. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Renal Effects C-140
Figure C-32. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Renal Effects C-144
Figure C-33. Summary of Mechanistic Studies of PFOS and Renal Effects C-147
Figure C-34. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Hematological Effects C-155
Figure C-35. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Hematological Effects C-157
vii
-------
APRIL 2024
Figure C-36. Summary of Mechanistic Studies of PFOS and Hematological Effects C-158
Figure C-37. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Respiratory Effects C-164
Figure C-38. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Respiratory Effects C-165
Figure C-39. Summary of Mechanistic Studies of PFOS and Respiratory Effects C-168
Figure C-40. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Musculoskeletal Effects C-174
Figure C-41. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Musculoskeletal Effects C-176
Figure C-42. Summary of Mechanistic Studies of PFOS and Musculoskeletal Effects C-177
Figure C-43. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Gastrointestinal Effects C-183
Figure C-44. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Gastrointestinal Effects C-184
Figure C-45. Summary of Mechanistic Studies of PFOS and Gastrointestinal Effects C-185
Figure C-46. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Dental Effects C-189
Figure C-47. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Ocular Effects C-193
Figure C-48. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Ocular Effects C-194
Figure C-49. Summary of Mechanistic Studies of PFOS and Ocular Effects C-195
Figure C-50. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Dermal Effects C-198
Figure C-51. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Dermal Effects C-199
Figure E-l. Difference in Population Tail Probabilities Resulting from a One Standard
Deviation Shift in the Mean from a Standard Normal Distribution, Illustrating
the Theoretical Basis for a Baseline BMR of 1 SD E-5
Figure E-2. Difference in Population Tail Probabilities Resulting from a V2 Standard
Deviation Shift in the Mean from an Estimation of the Distribution of
Log2(Tetanus Antibody Concentrations at Age 7 Years) E-7
Figure E-3. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Male Rats Following
viii
-------
APRIL 2024
Exposure to PFOS, for Number of Animals Per Group at Start of Study
(Butenhoff et al., 2012; Thomford, 2002) E-55
Figure E-4. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Male Rats Following
Exposure to PFOS, for Number of Animals Per Group at Time of First Tumor....E-55
Figure E-5. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Incidence of Islet Cell Carcinomas in Male Rats Following
Exposure to PFOS, for Number of Animals Per Group at Start of Study
(Butenhoff et al., 2012; Thomford, 2002) E-58
Figure E-6. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Incidence of Islet Cell Carcinomas in Male Rats Following
Exposure to PFOS, for Number of Animals Per Group at Time of First Tumor
(Butenhoff et al., 2012; Thomford, 2002) E-58
Figure E-7. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Combined Incidence of Islet Cell Adenomas and
Carcinomas in Male Rats Following Exposure to PFOS, for Number of
Animals Per Group at Start of Study (Butenhoff et al., 2012; Thomford, 2002)....E-61
Figure E-8. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Combined Incidence of Islet Cell Adenomas and
Carcinomas in Male Rats Following Exposure to PFOS, for Number of
Animals Per Group at Time of First Tumor (Butenhoff et al., 2012; Thomford,
2002) E-61
Figure E-9. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Female Rats Following
Exposure to PFOS, for Number of Animals Per Group at Start of Study
(Butenhoff et al., 2012; Thomford, 2002) E-64
Figure E-10. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Female Rats Following
Exposure to PFOS, for Number of Animals Per Group at Time of First Tumor
(Butenhoff et al., 2012; Thomford, 2002) E-64
Figure E-l 1. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas and Carcinomas in Female Rats
Following Exposure to PFOS, for Number of Animals Per Group at Start of
Study (Butenhoff et al., 2012; Thomford, 2002) E-66
Figure E-12. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas and Carcinomas in Female Rats
Following Exposure to PFOS, for Number of Animals Per Group at Time of
First Tumor (Butenhoff et al., 2012; Thomford, 2002) E-67
Figure E-13. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Log-
Logistic Model for Individual Cell Necrosis in the Liver in Female Sprague-
ix
-------
APRIL 2024
Dawley Crl:CD(SD)IGS BR Rats Following Exposure to PFOS (Butenhoff et
aL 2012; Thomford, 2002) E-69
Figure E-14. Plot of Mean Response by Dose (Including Highest Dose) with Fitted Curve
for the Polynomial 6 Model for Pup Body Weight Relative to the Litter at LD 5
in Fi Male and Female Sprague-Dawley Rats Following Exposure to PFOS
(Luebker et al., 2005b) E-73
Figure E-15. Plot of Mean Response by Dose with Fitted Curve for the Selected
Exponential 5 Model for Pup Body Weight Relative to the Litter at LD 5 in Fi
Male and Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker
et al., 2005b) E-73
Figure E-16. Plot of Mean Response by Dose with Fitted Curve for the Selected
Exponential 3 Model for Pup Body Weight Relative to the Litter at LD 1 in Fi
Male and Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker
et al., 2005b) E-76
Figure E-17. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Logistic
Model for Extramedullary Hematopoiesis in the Spleen in Male Sprague-
Dawley Rats Following Exposure to PFOS (NTP, 2019) E-78
Figure E-18. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Extramedullary Hematopoiesis in the Spleen in Female
Sprague-Dawley Rats Following Exposure to PFOS (NTP, 2019) E-80
Figure E-19. Plot of Mean Response by Dose with Fitted Curve for the Selected Hill Model
for PFC Response of Splenic Cells in Fi Male C57BL/6 Mice at PNW 4
Following Exposure to PFOS (Zhong et al., 2016) E-82
Figure E-20. Plot of Mean Response by Dose with Fitted Curve for the Selected
Exponential 4 Model for Pup Body Weight Relative to the Litter at LD 1 in Fi
Male and Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker
et al., 2005a) E-87
Figure F-l. Experimentally Observed Serum Concentrations (Chang et al., 2012) and
Median Prediction for a Single Oral Dose of 1 or 20 mg/kg PFOS to Female
CD1 Mice F-l
Figure F-2. Experimentally Observed Serum Concentrations (Chang et al., 2012) and
Median Prediction for a Single Oral Dose of 1 or 20 mg/kg PFOS to Male CD1
Mice F-2
Figure F-3. Experimentally Observed Serum Concentrations (Chang et al., 2012) and
Median Prediction for a Single IV Dose of 2 mg/kg or a Single Oral Dose of 2
or 15 mg/kg PFOS to Male Sprague-Dawley Rats F-2
Figure F-4. Model Prediction Summary for PFOS Training Data F-3
Figure F-5. PFOS Sensitivity Coefficients of the Adult Model and Developmental Model F-4
x
-------
APRIL 2024
Figure F-6. mentally Observed Serum Concentrations (Huang et al., 2021) and Median
Predictions for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 or 20 mg/kg
PFOS to Male Sprague-Dawley Rats F-5
Figure F-7. Experimentally Observed Serum Concentrations (Huang et al., 2021) and
Median Predictions for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 or 20
mg/kg PFOS to Female Sprague-Dawley Rats F-5
Figure F-8. Experimentally Observed Serum Concentrations (Kim et al., 2016b) and
Median Prediction for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 mg/kg
PFOS to Male Sprague-Dawley Rats F-6
Figure F-9.Experimentally Observed Serum Concentrations (Kim et al., 2016b) and
Median Prediction for a Single IV Dose of 2 mg/kg an Oral Dose of 2 mg/kg
PFOS to Female Sprague-Dawley Rats F-6
Figure F-10. Model Prediction Summary for PFOS Test Data F-7
Figure F-ll. Model Comparison F-8
Figure F-12. Predicted Child Serum Levels Compared to Reported Values F-9
Figure F-13. Comparison of Predicted and Observed Child Serum Concentration F-9
FigureF-14. Sensitivity Coefficients F-ll
Figure F-15. Predicted Child Serum Levels Compared to Reported Values with Increased
Volume of Distribution in Children as was Implemented in the Minnesota
Department of Health Model F-12
Figure F-16. Predicted Child Serum Levels Compared to Reported Values with Constant
Volume of Distribution and Variable Exposure Based on Drinking Water Intake. F-13
Figure G-l. Application of the Exposure Decision Tree (U.S. EPA, 2000) for PFOS G-13
xi
-------
APRIL 2024
Tables
Table A-l. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for a
Systematic Review on the Health Effects from Exposure to PFOA and PFOS A-7
Table A-2. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for
Absorption, Distribution, Metabolism, and/or Excretion (ADME) Studies A-8
Table A-3. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for
Mechanistic Studies A-9
Table A-4. Search String for April 2019 Database Searches A-l 1
Table A-5. Search String for September 2020, February 2022, and February 2023 Database
Searches A-13
Table A-6. Key Epidemiological Studies of Priority Health Outcomes Identified from 2016
PFOS Health Effects Support Document A-16
Table A-7. Key Toxicological Animal Toxicological Studies Identified from 2016 PFOS
Health Effects Support Document A-19
Table A-8. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for a
Systematic Review on the Health Effects from Exposure to PFOA and PFOS A-21
Table A-9. DistillerSR Form for Title/Abstract Screening A-24
Table A-10. SWIFT-Active Form for Title/Abstract Screening A-25
Table A-l 1. Supplemental Tags for Title/Abstract and Full-Text Screening A-25
Table A-12. Mechanistic Study Categories Considered as Supplemental A-26
Table A-13. DistillerSR Form for Full-Text Screening A-28
Table A-14. Health Effect Categories Considered for Epidemiological Studies A-31
Table A-15. Litstream Form for ADME Screening and Light Data Extraction A-34
Table A-16. Litstream Form for Mechanistic Screening and Light Data Extraction A-44
Table A-17. Possible Domain Scores for Study Quality Evaluation A-53
Table A-l8. Overall Study Confidence Classifications A-53
Table A-19. Study Quality Evaluation Considerations for Participant Selection A-55
Table A-20. Study Quality Evaluation Considerations for Exposure Measurement A-57
Table A-21. Criteria for Evaluating Exposure Measurement in Epidemiology Studies of
PFAS and Health Effects A-60
Table A-22. Study Quality Evaluation Considerations for Outcome Ascertainment A-62
Table A-23. Study Quality Evaluation Considerations for Confounding A-64
xii
-------
APRIL 2024
Table A-24. Study Quality Evaluation Considerations for Analysis A-66
Table A-25. Study Quality Evaluation Considerations for Selective Reporting A-68
Table A-26. Study Quality Evaluation Considerations for Study Sensitivity A-69
Table A-27. Evaluation Considerations for Overall Study Confidence - Overall
Confidence, Epidemiological Studies A-70
Table A-28. Study Quality Evaluation Considerations for Reporting Quality A-72
Table A-29. Study Quality Evaluation Considerations for Selection and Performance -
Allocation A-75
Table A-30. Study Quality Evaluation Considerations for Selection and Performance -
Observational Bias/Blinding A-77
Table A-31. Study Quality Evaluation Considerations for Confounding/Variable Control A-81
Table A-32. Study Quality Evaluation Considerations for Selective Reporting and Attrition
- Reporting and Attrition Bias A-83
Table A-33. Study Quality Evaluation Considerations for Exposure Methods Sensitivity -
Chemical Administration and Characterization A-85
Table A-34. Study Quality Evaluation Considerations for Exposure Methods Sensitivity -
Exposure Timing, Frequency, and Duration A-88
Table A-35. Study Quality Evaluation Considerations for Outcome Measures and Results
Display - Endpoint Sensitivity and Specificity A-90
Table A-36. Study Quality Evaluation Considerations for Outcome Measures and Results
Display - Results Presentation A-93
Table A-37. Evaluation Considerations for Overall Study Confidence - Overall
Confidence, Animal Toxicological Studies A-95
Table A-38. DistillerSR Form for Data Extraction of Epidemiological Studies A-98
Table A-39. Epidemiological Study Design Definitions A-105
Table A-40. HAWC Form Fields for Data Extraction of Animal Toxicological Studies A-107
Table A-41. Framework for Strength-of-Evidence Judgments for Epidemiological Studies21 A-l 13
Table A-42. Framework for Strength-of-Evidence Judgments for Animal Toxicological
Studies21 A-l 14
Table A-43. Evidence Integration Judgments for Characterizing Potential Human Health
Effects in the Evidence Integration21 A-l 15
Table A-44. Attributes used to evaluate studies for derivation of toxicity values (adapted
from ORD Staff Handbook for Developing IRIS Assessments Table 7-2) A-120
Table A-45. Epidemiologic Meta-Analysis Studies Identified from Literature Review A-128
xiii
-------
APRIL 2024
Table A-46. Studies Identified After Updated Literature Review (Published or Identified
After February 2022) A-136
Table A-47. Animal Studies Identified After Updated Literature Review (Published or
Identified After February 2022) A-147
Table A-48. Human Studies Identified After 2023 Updated Literature Search (Published or
Identified After February 2023) A-150
Table B-l. Cellular Accumulation and Retention Relative to Lipophilicity and
Phospholipophilicity as Reported by Sanchez Garcia et al. (2018) B-l
Table B-2. PFOS Parameters From Toxicokinetic Studies Informing Bioavailability in
Sprague-Dawley Rats B-3
Table B-3. Concentrations of PFOS in Various Tissues of Male Sprague-Dawley Rats
Exposed to PFOS by Gavage for 28 Days as Reported by Cui et al. (2009) B-l 1
Table B-4. Concentrations of PFOS in Various Tissues of Male and Female Sprague-
Dawley Rats Exposed to PFOS by Feed for 28 Days as Reported by Curran et
al. (2008) B-l2
Table B-5. Distribution of PFOS in Male Wistar Rats Exposed via Drinking Water for 1 or
3 Months as Reported by Iwabuchi et al. (2017) B-l3
Table B-6. PFOS Levels in the Serum and Liver of Male and Female Sprague-Dawley Rats
Exposed to PFOS in Feed for 2 Years as Reported by Thomford (2002) B-13
Table B-7. PFOS Concentrations in Human Cord Blood, Maternal Blood, and
Transplacental Transfer Ratios (RCM) B-21
Table B-8. Summary of PFOS Concentrations in Human Maternal Blood, Cord Blood,
Placenta and Amniotic Fluid Studies B-26
Table B-9. Summary of Human PFOS Concentrations in Maternal Serum, Breast Milk, and
Infant Serum B-32
Table B-10. Percent Change in PFOS Ratios in Human Maternal Serum and Breast Milk
and Breast Milk and Infant Serum by Infant Age as Reported by Mondal et al.
(2014) B-34
Table B-l 1. Percent Change in Human PFOS Serum Concentration by Exclusive, Mixed or
No Breastfeeding Per Month as Reported by Mogensen et al. (2015b) B-34
Table B-12. Liver, Serum, Urine, and Feces PFOS Concentrations in Pregnant Sprague-
Dawley Dams and Fetuses (Luebker et al., 2005a) B-35
Table B-13. Serum, Liver, and Brain Tissue PFOS Concentrations of Sprague-Dawley
Dams and Offspring as Reported by Chang et al. (2009) B-37
Table B-14. Serum, Hippocampus, and Cortex PFOS Concentrations of Sprague-Dawley
Rat Pups as Reported by Zeng et al. (2011) B-38
XIV
-------
APRIL 2024
Table B-15. Serum and Lung PFOS Concentration of Sprague-Dawley Rat Pups (Chen et
aL 2012b) B-39
Table B-16. Concentration Ratios of 35S-PFOS Maternal Serum to Various Organs of
C57BL/6 Mouse Dams, Fetuses, and Pups (Lai et al., 2017b) B-40
Table B-17. Percent Distribution of PFOS in Male and Female KM Mice After 50 mg/kg
Subcutaneous Injection (Liu et al., 2009) B-40
Table B-18. Summary of PFOS Volume of Distribution Values Assigned in Human Studies .B-42
Table B-19. Summary of PFOS Volume of Distribution in Rats B-44
Table B-20. Pharmacokinetic Parameters After Acute PFOS Exposure in Cynomolgus
Monkeysa (Chang et al., 2017) B-47
Table B-21. Enterohepatic Transporters of PFOS B-52
Table B-22. Estimated Percentage of the Sum of PFOS, PFNA, and PFOA in Excreta and
Serum of Male and Female Wistar Ratsa as Reported by Gao et al. (2015) B-55
Table B-23. Summary of PFOS Concentration in Blood and Urine in Relation to Half-life
Values in Humans B-60
Table B-24. Summary of Human PFOS Half-Life Values B-64
Table B-25. Summary of Animal PFOS Half-Life Values Identified in the Literature
Review B-68
Table C-l. Evidence Profile Table for PFOS Reproductive Effects in Males C-26
Table C-2. Evidence Profile Table for PFOS Reproductive Effects in Females C-35
Table C-3. Summary of Results for Thyroid and Thyroid-Related Hormones in
Toxicological Studies Following Exposure to PFOS C-55
Table C-4. Associations Between PFOS Exposure and Endocrine Organ Weights in
Rodents and Non-human Primates C-68
Table C-5. Evidence Profile Table for PFOS Endocrine Effects C-72
Table C-6. Evidence Profile Table for PFOS Systemic and Metabolic Effects C-105
Table C-l. Associations Between PFOS Exposure and Neurobehavioral Effects in Rodents C-123
Table C-8. Associations Between PFOS Exposure and Neurotransmitters in Rodents C-126
Table C-9. Evidence Profile Table for PFOS Nervous System Effect C-131
Table C-10. Evidence Profile Table for PFOS Renal Effects C-149
Table C-l 1. Evidence Profile Table for PFOS Hematological Effects C-160
Table C-12. Evidence Profile Table for PFOS Respiratory Effects C-170
Table C-13. Evidence Profile Table for PFOS Musculoskeletal Effects C-179
xv
-------
APRIL 2024
Table C-14. Evidence Profile Table for PFOS Gastrointestinal Effects C-186
Table C-15. Evidence Profile Table for PFOS Dental Effects C-191
Table C-16. Evidence profile table for PFOS Ocular effects C-196
Table C-17. Evidence Profile Table for PFOS Dermal Effects C-201
Table D-l. Associations Between PFOS Exposure and Developmental Effects in Recent
Epidemiological Studies D-l
Table D-2. Associations Between PFOS Exposure and Male Reproductive Effects in
Recent Epidemiologic Studies D-50
Table D-3. Associations Between PFOS Exposure and Female Reproductive Effects in
Female Children and Adolescents D-61
Table D-4. Associations Between PFOS Exposure and Female Reproductive Health Effects
in Pregnant Women D-65
Table D-5. Associations Between PFOS Exposure and Female Reproductive Health Effects
in Non-Pregnant Adult Women D-69
Table D-6. Associations Between PFOS Exposure and Hepatic Effects in Epidemiologic
Studies D-71
Table D-7. Associations Between PFOS Exposure and Vaccine Response in Recent
Epidemiological Studies D-79
Table D-8. Associations Between PFOS Exposure and Infectious Disease in Recent
Epidemiological Studies D-94
Table D-9. Associations Between PFOS Exposure and Asthma in Recent Epidemiologic
Studies D-101
Table D-10. Associations Between PFOS Exposure and Allergies in Recent Epidemiologic
Studies D-l 10
Table D-l 1. Associations Between PFOS Exposure and Eczema in Recent Epidemiologic
Studies D-l 15
Table D-12. Associations Between PFOS Exposure and Autoimmune Health Effects in
Recent Epidemiologic Studies D-l 18
Table D-13. Associations Between PFOS Exposure and Cardiovascular Effects in Recent
Epidemiological Studies D-120
Table D-14. Associations Between PFOS Exposure and Serum Lipid Effects in Recent
Epidemiologic Studies D-l36
Table D-15. Associations Between PFOS Exposure and Endocrine Effects in Recent
Epidemiologic Studies D-l64
xvi
-------
APRIL 2024
Table D-16. Associations Between PFOS Exposure and Metabolic Effects in Recent
Epidemiologic Studies D-173
Table D-17. Associations Between PFOS Exposure and Neurological Effects in Recent
Epidemiologic Studies D-191
Table D-18. Associations Between PFOS Exposure and Renal Effects in the General
Population D-214
Table D-19. Associations Between PFOS Exposure and Hematological Effects in Recent
Epidemiologic Studies D-223
Table D-20. Associations Between PFOS Exposure and Respiratory Effects in Recent
Epidemiologic Studies D-225
Table D-21. Associations Between PFOS Exposure and Musculoskeletal Effects in Recent
Epidemiologic Studies D-227
Table D-22. Associations Between PFOS Exposure and Gastrointestinal Effects in Recent
Epidemiologic Studies D-232
Table D-23. Associations Between PFOS Exposure and Dental Effects in Recent
Epidemiologic Studies D-234
Table D-24. Associations Between PFOS Exposure and Ocular Effects in Recent
Epidemiologic Studies D-234
Table D-25. Associations Between PFOS Exposure and Dermal Health Effects in Recent
Epidemiologic Studies D-23 5
Table D-26. Associations Between PFOS Exposure and Cancer in Recent Epidemiologic
Studies D-23 6
Table E-l. Summary of Modeling Approaches for POD Derivation from Epidemiological
Studies E-l
Table E-2. Results Specific to the Slope from the Linear Analyses of PFOS Measured at
Age 5 Years and Log2(Tetanus Antibody Concentrations) Measured at Age
7 Years from Table 1 in Budtz-j0rgensen and Grandjean (2018a) in a Single-
PFAS Modela and in a Multi-PFAS Modelb E-3
Table E-3. BMDs and BMDLs for Effect of PFOS at Age 5 Years on Anti-Tetanus
Antibody Concentrations at age 7 Years (Budtz-j0rgensen and Grandjean,
2018a) Using a BMR of V2 SD Change in Log2(Tetanus Antibodies
Concentration) and a BMR of 1 SD Change in Log2(Tetanus Antibodies
Concentration) E-7
Table E-4. Results of the Linear Analyses of PFOS Measured Perinatally and Tetanus
Antibodies Measured at Age 5 Years from Budtz-j0rgensen and Grandjean
(2018b) in a Single-PFAS Model and in a Multi-PFAS Model E-8
xvii
-------
APRIL 2024
Table E-5. BMDs and BMDLs for Effect of PFOS Measured Perinatally and Anti-Tetanus
Antibody Concentrations at Age 5 Years (Budtz-j0rgensen and Grandjean,
2018a) I >10
Table E-6. BMDs and BMDLs for Effect of PFOS on Anti-Tetanus Antibody
Concentrations (Timmermann et al., 2021) Using a BMR of V2 SD Change in
Logio(Tetanus Antibodies Concentration) and a BMR of 1 SD Change in
Logio(Tetanus Antibodies Concentration) E-l 1
Table E-7. BMDLs for Effect of PFOS on Anti-Tetanus Antibody Concentrations Using a
BMR of V2 SD (Timmermann et al., 2021) E-l2
Table E-8. Results Specific to the Slope from the Linear Analyses of PFOS Measured at
Age 5 Years and Log2(Diphtheria Antibodies) Measured at Age 7 Years from
Table 1 in Budtz-j0rgensen and Grandjean (2018a) in a Single-PFAS Modela
and in a Multi-PFAS Modelb E-l2
Table E-9. BMDs and BMDLs for Effect of PFOS at Age 5 Years on Anti-Diphtheria
Antibody Concentrations at Age 7 Years (Budtz-j0rgensen and Grandjean,
2018a) Using a BMR of V2 SD Change in Log2(Diphtheria Antibodies
Concentration) and a BMR of 1 SD Log2(Diphtheria Antibodies Concentration) .E-l4
Table E-10. Results of the Linear Analyses of PFOS Measured Perinatally and Diphtheria
Antibodies Measured at age 5 Years from Budtz-j0rgensen and Grandjean
(2018b) in a Single-PFAS Modela and in a Multi-PFAS Modelb E-l6
Table E-l 1. BMDs and BMDLs for Effect of PFOS Measured Perinatally and Anti-
Diphtheria Antibody Concentrations at age 5 Years (Budtz-j0rgensen and
Grandj ean, 2018a) E-17
Table E-12. BMDs and BMDLs for Effect of PFOS on Anti-Diphtheria Antibody
Concentrations (Timmermann et al., 2021) Using a BMR of V2 SD Change in
Logio(Tetanus Antibodies Concentration) and a BMR of 1 SD Change in
Logio(Tetanus Antibodies Concentration) E-18
Table E-13. BMDLs for Effect of PFOS on Anti-Diphtheria Antibody Concentrations
Using a BMR of V2 SD (Timmermann et al., 2021) E-l9
Table E-14. Levels of Rubella Vaccine-Induced Antibodies at the Age of 3 Years (Adapted
from Table 3 in Granum et al. (2013)) E-20
Table E-15. BMDs and BMDLs for Effect of Maternal Serum PFOS on Anti-Rubella
Antibody Concentrations in Children Using a BMR of V2 SD Change in Rubella
Antibodies Concentration and a BMR of 1 SD Change in Rubella Antibodies
Concentration (Granum et al., 2013) E-22
Table E-16. BMDs and BMDLs for Effect of PFOS on Anti-Rubella Antibody
Concentrations in Adolescents (Zhang et al., 2023c) Using a BMR of V2 SD
Change in Ln(Rubella Antibodies Concentration) and a BMR of 1 SD Change
in Ln(Rubella Antibodies Concentration) E-23
-------
APRIL 2024
Table E-17. BMDs and BMDLs in ng/mL for Effect of PFOS on Anti-Rubella Antibody
Concentrations E-24
Table E-18. BMDLs for Effect of PFOS on Anti-Rubella Antibody Concentrations Using a
BMR of 5% E-24
Table E-19. BMDs and BMDLs in ng/mL for Effect of PFOS on Decreased Birth Weight,
by Using the Exact Percentage (8.27%) of Live Births Falling Below the Public
Health Definition of Low Birth Weight, or Alternative Study-Specific Tail E-30
Table E-20. BMDs and BMDLs for Effect of PFOS on Decreased Birth Weight by
Background Exposure, Using the Exact Percentage of the Population (8.27%)
of Live Births Falling Below the Public Health Definition of Low Birth Weight,
or Alternative Tail Probability E-31
Table E-21. BMD and BMDL for Effect of PFOS (ng/mL) on Increased ALT in Nian et al.
(2019), for 5% and 10% Extra Risk 11-33
Table E-22. NHANES Mean and Standard Deviation of Ln(ALT) (In IU/L) and Mean
PFOS (Ln ng/mL) 11-34
Table E-23. Prevalence of Elevated ALT E-35
Table E-24. BMD and BMDL for Effect of PFOS (ng/mL) on Increased ALT in Gallo et al.
(2012) E-3 6
Table E-25. Odds Ratios for Elevated ALT by Decile of PFOS Serum Concentrations
(ng/mL) from Gallo et al. (2012) E-37
Table E-26. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et
al. (2012) Using the Unadjusted Mean PFOS Serum Concentration E-38
Table E-27. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et
al. (2012) Using the Adjusted, No Intercept Mean PFOS Serum Concentration ...E-39
Table E-28. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et
al. (2012) Using the Unadjusted, Median PFOS Serum Concentration E-40
Table E-29. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et
al. (2012) Using the Adjusted, No Intercept Median PFOS Serum
Concentration E-41
Table E-30. BMDs and BMDLs in ng/mL for Effect of PFOS on Serum Ln(ALT) in
Females E-42
Table E-31. BMDLs for Effect of PFOS on Serum ALT Using a BMR of 5% E-43
Table E-32. NHANES Mean and Standard Deviation of Total Cholesterol (mg/dL) and
Mean PFOS (ng/mL) 11-44
Table E-33. BMDs and BMDLs for Effect of PFOS on Increased Cholesterol in Dong et al.
(2019) 11-46
xix
-------
APRIL 2024
Table E-34. NHANES Mean and Standard Deviation of Ln(TC) (Ln(mg/dL)) and Mean
I.n(PI OS) (Ln(ng/mL)) I>46
Table E-35. BMDs and BMDLs for Effect of PFOS on Increased Cholesterol in Steenland
et al. (2009) E-47
Table E-36. Regression Results for Serum Total Cholesterol by Deciles of Serum PFOS
from Steenland et al. (2009) E-47
Table E-37. Summary of Benchmark Dose Modeling Results for Increase Mean Serum
Total Cholesterol in Steenland et al. (2009) E-48
Table E-38. Odds Ratios for Elevated Serum Total Cholesterol by Quartiles of Serum
PFOS from Steenland et al. (2009) E-49
Table E-39. Summary of Benchmark Dose Modeling Results for Elevated Total
Cholesterol in Steenland et al. (2009) E-50
Table E-40. Adjusted Mean Differences in Serum Total Cholesterol by Quartiles of Serum
PFOS (ng/mL) from Lin et al. (2019) E-51
Table E-41. Summary of Benchmark Dose Modeling Results for Increase Mean Serum
Total Cholesterol Lin et al. (2019) E-52
Table E-42. BMDs and BMDLs in ng/mL for Effect of PFOS on Serum Total Cholesterol.... E-52
Table E-43. BMDLs for Effect of PFOS on Serum Total Cholesterol Using a BMR of 5% ....E-53
Table E-44. Dose-Response Modeling Data for Hepatocellular Adenomas in Male Rats
Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002) E-53
Table E-45. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas in Male Rats Following Exposure to PFOS (Butenhoff et al., 2012;
Thomford, 2002) 11-54
Table E-46. Dose-Response Modeling Data for Incidence of Islet Cell Carcinomas in Male
Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002) E-56
Table E-47. Summary of Benchmark Dose Modeling Results for Incidence of Islet Cell
Carcinomas in Male Rats Following Exposure to PFOS (Butenhoff et al., 2012;
Thomford, 2002) E-56
Table E-48. Dose-Response Modeling Data for Combined Incidence of Islet Cell
Adenomas and Carcinomas in Male Rats Following Exposure to PFOS
(Butenhoff et al., 2012; Thomford, 2002) E-59
Table E-49. Summary of Benchmark Dose Modeling Results for Combined Incidence of
Islet Cell Adenomas and Carcinomas in Male Rats Following Exposure to
PFOS (Butenhoff et al., 2012; Thomford, 2002) 11-60
Table E-50. Dose-Response Modeling Data for Hepatocellular Adenomas in Female Rats
Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002) E-62
xx
-------
APRIL 2024
Table E-51. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas in Female Rats Following Exposure to PFOS (Butenhoff et al.,
2012; Thomford, 2002) E-62
Table E-52. Dose-Response Modeling Data for Hepatocellular Adenomas and Carcinomas
in Female Rats Following Exposure to PFOS (Butenhoff et al., 2012;
Thomford, 2002) E-65
Table E-53. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas and Carcinomas in Female Rats Following Exposure to PFOS
(Butenhoff et al., 2012; Thomford, 2002) E-65
Table E-54. Dose-Response Modeling Data for Individual Cell Necrosis in the Liver in
Female Sprague-Dawley Crl:CD(SD)IGS BR Rats Following Exposure to
PFOS (Butenhoff et al., 2012; Thomford, 2002) E-67
Table E-55. Summary of Benchmark Dose Modeling Results for Individual Cell Necrosis
in the Liver in Female Sprague-Dawley Crl:CD(SD)IGS BR Rats Following
Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002) E-68
Table E-56. Dose-Response Modeling Data for Fetal Body Weight in Fi Male and Female
CD-I Mice Following Exposure to PFOS (Lee et al., 2015) E-69
Table E-57. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 5) in Fi Male and Female Sprague-Dawley Rats Following Exposure to
PFOS (Luebker et al., 2005b) E-70
Table E-58. Summary of Benchmark Dose Modeling Results for Pup Body Weight
Relative to the Litter (LD 5) in Fi Male and Female Sprague-Dawley Rats
Following Exposure to PFOS (Constant Variance) (Luebker et al., 2005b) E-72
Table E-59. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 1) in Fi Male and Female Sprague-Dawley Rats Following Exposure to
PFOS (Luebker et al., 2005b) E-74
Table E-60. Summary of Benchmark Dose Modeling Results for Pup Body Weight
Relative to the Litter (LD 1) in Fi Male and Female Sprague-Dawley Rats
Following Exposure to PFOS (Nonconstant Variance) (Luebker et al., 2005b) ....E-75
Table E-61. Dose-Response Modeling Data for Extramedullary Hematopoiesis in Male
Sprague-Dawley Rats Following Exposure to PFOS (NTP, 2019) E-76
Table E-62. Summary of Benchmark Dose Modeling Results for Extramedullary
Hematopoiesis in Male Sprague-Dawley Rats Following Exposure to PFOS
(NTP, 2019) E-77
Table E-63. Dose-Response Modeling Data for Extramedullary Hematopoiesis in the
Spleen in Female Sprague-Dawley Rats Following Exposure to PFOS (NTP,
2019) E-78
xxi
-------
APRIL 2024
Table E-64. Summary of Benchmark Dose Modeling Results for Extramedullar
Hematopoiesis in the Spleen in Female Sprague-Dawley Rats Following
Exposure to PFOS (M P. 2019) E-79
Table E-65. Dose-Response Modeling Data for PFC Response of Splenic Cells in Fi Male
C57BL/6 Mice at PNW 4 Following Exposure to PFOS (Zhong et al., 2016) E-81
Table E-66. Summary of Benchmark Dose Modeling Results for Plaque-Forming Cell
Response of Splenic Cells in Fi Male C57BL/6 Mice at PNW 4 Following
Exposure to PFOS (Constant Variance) (Zhong et al., 2016) E-81
Table E-67. Dose-Response Modeling Data for Pup Survival at PND 5 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Lau et al., 2003) ....E-83
Table E-68. Dose-Response Modeling Data for Pup Survival at PND 22 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Lau et al., 2003) ....E-83
Table E-69. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 1) in Fi Male and Female Sprague-Dawley Rats Following Exposure to
PFOS (Luebker et al., 2005a) E-84
Table E-70. Summary of Benchmark Dose Modeling Results for Pup Body Weight
Relative to the Litter at LD 1 in Fi Male and Female Sprague-Dawley Rats
Following Exposure to PFOS (Nonconstant Variance) (Luebker et al., 2005a) E-86
Table F-l. Root mean squared error comparison between the baseline model (as applied in
the main risk assessment) and alternative models with features inspired by the
MDH model F-l 2
Table G-l. Summary of EPA national fish tissue monitoring results for PFOS G-6
xxii
-------
APRIL 2024
Acronyms and Abbreviations
17-OHP
17-hy droxyprogesterone
BALF
bronchoalveolar lavage
ABC
ATP-binding cassette
fluid
transporter
BBB
blood-brain barrier
aBMD
areal bone mineral
BCERP
Breast Cancer and
density
Environment Research
ACD
anterior chamber depth
Program
ACE
America's Children and
BCRP
breast cancer resistance
the Environment
protein
ACTH
adrenocorticotropic
BD
bolus dose
hormone
BDI
Beck Depression
ADHD
attention deficit
Inventory
hyperactivity disorder
BDI-II
Beck Depression
ADME
absorption, distribution,
Inventory-II
metabolism, and
BMC
bone mineral content
excretion
BMD
benchmark dose
AF:CB
amniotic fluid and cord
BMDL
lower limit of benchmark
blood ratio
dose
AFFF
aqueous film forming
BMDLo.5sd
lower bound on the dose
foam
level corresponding to the
AGD
anogenital distance
95% lower confidence
AIC
Akaike information
limit for a change in the
criterion
mean equal to 0.5
ALSPAC
Avon Longitudinal Study
standard deviation from
of Parents and Children
the control mean
ALT
alanine aminotransferase
BMDLisd
lower bound on the dose
AMH
anti-Mullerian hormone
level corresponding to the
95% lower confidence
ASBT
apical sodium-dependent
limit for a change in the
bile acid transporter
mean equal to one
ASD
autism spectrum disorder
standard deviation from
ASQ
Ages and Stages
the control mean
Questionnaire
BMDLs
lower bound on the dose
ATP III
Adult Treatment Panel III
level corresponding to the
AT SDR
Agency for Toxic
95% lower confidence
Substances and Disease
limit for a 5% response
Registry
level
AUC
area under the curve
BMDL io
lower bound on the dose
AUMC
area under the first
level corresponding to the
moment curve
95% lower confidence
AZI
azithromycin-dihydrate
limit for a 10% change
P
regression coefficients
BMDS
Benchmark Dose
Software
xxiii
-------
APRIL 2024
BMI
body mass index
BMR
benchmark response
BRIEF
Behavior Rating
Inventory of Executive
Function
BUN
blood urea nitrogen
BW
body weight
C7,avg
average concentration
over final week of study
CalEPA
California Environmental
Protection Agency
Cavg,pup,gest
area under the curve
normalized per day
during gestation
Cavg,pup,gest,lact
area under the curve
normalized dose per day
during gestation/lactation
Cavg,pup,lact
area under the curve
normalized per day
during lactation
Cavg-pup-diet
average concentration
during the post-weaning
phase
CDI
Comprehensive
Developmental Inventory
C-F
carbon-fluorine
CHCA
a-Cyano-4-
hydroxycinnamic acid
CHECK
Children's Health and
Environmental Chemicals
in Korea
CHEF
Children's Health and the
Environment in the
Faroes
CHO
Chinese hamster ovary
CI
confidence interval
CKD
chronic kidney disease
Cmax
maximum blood
concentration
Cmax,dam
maximum maternal
concentration during
gestation
Cmax,pup,gest
maximum fetal
concentration during
gestation
Cmax,pup,lact
maximum pup
concentration during
lactation
CNS
central nervous system
CRH
corticotropin releasing
hormone
CSF
cancer slope factor
CSM
cholestyramine
CTX
type I collagen
CVD
cardiovascular disease
DFI
deoxyribonucleic acid
fragmentation index
DHEA
dehydroepiandrosterone
DHEAS
dehydroepiandrosterone
sulfate
DNA
deoxyribonucleic acid
DNBC
Danish National Birth
Cohort
DPP
Diabetes Prevention
Program
dU
diurnal urinary
E2
estradiol
EFSA
European Food Safety
Authority
GLP
good laboratory practice
eGFR
estimated glomerular
filtration rate
eNT
equilibrative nucleoside
transporter
EPA
U.S. Environmental
Protection Agency
EYHS
European Youth Heart
Study
Fi
first generation
FDA
U.S. Food and Drug
Administration
FEV1
forced expiratory volume
in one second
FR
folate receptor
xxiv
-------
APRIL 2024
FSH
follicle stimulating
IQR
hormone
IRIS
FT3
free triiodothyronine
FT4
free thyroxine
IUFD
FTI
free thyroxine index
IV
FVC
forced vital capacity
Kd
GD
gestation day
Kmem/w
GI
gastrointestinal
GM
geometric mean
KpS
Hb
hemoglobin
LC50
HDL
high-density-lipoprotein
HED
human equivalent dose
LD
HEK293
human embryonic kidney
cells
LDL
HERO
Health and
Environmental Research
L-FABP
Online
LH
HESD
health effects support
document
LIFE
HHRA
human health risk
assessment
HOMA-B
Homeostatic Model
Assessment of Beta-Cell
LINC
Function
LLOQ
HOMA-IR
Homeostatic Model
Assessment for Insulin
LOAEL
Resistance
HOME
Health Outcome
LOD
Measures of the
LOQ
Environment
MALDI
HUMIS
Norwegian Human Milk
Study
MCDI
IBD
inflammatory bowel
disease
ICso
median inhibiting
concentration
MCLG
ID
intellectual disability
IMS
imaging mass
spectrometry
MDH
INUENDO
Biopersistent
Organochlorines in Diet
MDI
and Human Fertility
MDL
IQ
intelligence quotient
MDR1
interquartile range
Integrated Risk
Information System
intrauterine fetal death
intravenous
disassociation constant
membrane/water partition
coefficients
tissue-to-plasma partition
coefficients
median lethal
concentration
lactation day
low-density lipoprotein
liver fatty acid binding
protein
luteinizing hormone
Longitudinal
Investigation of Fertility
and the Environment
Study
Linking Maternal
Nutrition to Child Health
lower limit of
quantification
lowest-observed-adverse-
effect level
limit of detection
limit of quantification
matrix-assisted laser
desorption/ionization
MacArthur
Communicative
Development Inventories
for Infants
Maximum Contaminant
Level Goal
Minnesota Department of
Health
Mental Development
Index
minimum detection limit
p-glycoprotein
xxv
-------
APRIL 2024
MeSH
medical subject headings
Mg/kg-day
milligrams per kilogram
per day
MIREC
Maternal-Infant Research
on Environmental
Chemicals
MLR
mixed linear regression
MPAH
N-methyl-PFOSA
mPL-II
mouse placental lactogen
mPLP
prolactin-like protein
MRL
minimum reporting level
mRNA
messenger ribonucleic
acid
MRP
multi-drug resistance-
associated protein
MOA
mode of action
MoBA
Norwegian Mother,
Father, and Child Cohort
Study
MPAH
2-(N-methyl-PFOSA)
acetate
NHANES
National Health and
Examination Survey
NICHD
U.S. National Institute of
Child Health and Human
Development
NJDEP
New Jersey Department
of Environmental
Protection
NO A A
National Oceanic and
Atmospheric
Administration
NOAEL
no-ob served-adverse-
effect level
NOAEC
no observed adverse
effect concentration
NPDWR
national primary drinking
water regulation
NTCP
sodium-taurocholate
cotransporting
polypeptide
NTP
National Toxicology
Program
OATs
organic anion transporters
OATPs
organic anion
transporting polypeptides
OCC
Odense Child Cohort
OCT
organic cation/carnitine
transporter
OECD
Organisation for
Economic Co-operation
and Development
OR
Odds Ratio
ORD
Office of Research and
Development
OVA
ovalbumin
Po
parental generation
PBPK
physiologically-based
pharmacokinetic
Pc
partition coefficient
PC
phosphatidylcholine
PCOS
polycystic ovary
syndrome
PDI
Psychomotor
Development Index
PECO
Populations, Exposures,
Comparator, and
Outcomes
PEF
peak expiratory flow rate
PFAA
perfluorinated alkyl acid
PFAS
per- and polyfluoroalkyl
substances
PFBA
perfluorobutanoic acid
PFBS
perfluorobutane sulfonate
PFC
plaque forming cell
PFCA
perfluorocarboxylates
PFDA
perfluorodecanoic acid
PFHxA
perfluorohexanoic acid
PFHxS
perfluorohexane sulfonate
PFOA
perfluorooctanoic acid
PFOS
perfluorooctane sulfonic
acid
PFNA
perfluorononanoic acid
PFSA
perfluoroalkanesulfonic
acid
xxvi
-------
PHQ-9
Patient Health
Questionnaire
Pion
passive anionic
permeability
PFUnDA
perfluoroundecanoic acid
PK
pharmacokinetic
PND
postnatal day
PNW
postnatal week
POD
point-of-departure
PODhed
point-of-departure human
equivalent dose
POI
premature ovarian
insufficiency
POPUP
Persistent Organic
Pollutants in Uppsala
Primiparas
PPARa
proliferator-activated
receptor alpha
Qi
quantile 1
Q2
quantile 2
Q3
quantile 3
Q4
quantile 4
QA
quality assurance
QUICKI
Quantitative Insulin
Sensitivity Check Index
RBC
red blood cell
RCM
ratio of cord blood to
maternal blood
concentrations
RFC
reduce folate carrier
RfD
reference dose
RIS
Research Information
System
ROBINS-I
Risk of Bias in
Nonrandomized Studies
of Interventions
Rpm
ratio of
placental :maternal
concentrations
RSC
relative source
contribution
rT3
reverse triiodothyronine
SAB
Science Advisory Board
APRIL 2024
SE
standard errors
SERT
serotonin transporter
SES
socioeconomic status
SD
standard deviation
SDQ
Strengths and Difficulties
Questionnaire
SDWA
Safe Drinking Water Act
SHBG
sex hormone binding
globulin
SMBCS
Shanghai Minhang Birth
Cohort Study
SWAN
Study of Women's Health
Across the Nation
T3
triiodothyronine
14
thyroxine
TA
thyroid antibody
TC
total cholesterol
TDS
Total Diet Study
TgAB
thyroblobulin antibodies
TiAb
title-abstract
Tmax
maximum plasma
concentration
TPO
anti-thyroid peroxidase
TPoAb
thyroid peroxidase
antibody
TSH
thyroid stimulating
hormone
TT3
total triiodothyronine
TTE
transplacental transfer
efficiencies
UCMR3
third Unregulated
Contaminant Monitoring
Rule
Vi
volume of central
distribution
V2
volume of peripheral
distribution
Vd
volume of distribution
VI
visual impairment
VLDL
very low-density
lipoproteins
xxvii
-------
APRIL 2024
VMWM Virtual Morris Water
Maze
WBHGB whole blood hemoglobin
WHO World Health
Organization
ww wet weight
xxviii
-------
APRIL 2024
Appendix A. Systematic Review Protocol for
Updated PFOS Toxicity Assessment
Per- and polyfluoroalkyl substances (PFAS) refers to a large group of fluorinated anthropogenic
chemicals that includes perfluorooctanoic acid (PFOA), perfluorooctane sulfonic acid (PFOS),
and thousands of other chemicals. The universe of environmentally relevant PFAS, including
parent chemicals, metabolites, and degradants, is greater than 12,000 compounds
(https://comptox.epa.eov/dashboard/chemical-lists/PFASMASTER). The Organisation for
Economic Co-operation and Development (OECD) New Comprehensive Global Database of
Per- and Polyfluoroalkyl Substances (PFASs) includes over 4,700 PFAS (OECD, 2018). The
number of PFAS used globally in commercial products at the time of the drafting of this
document is approximately 250 substances (Buck et al., 2021).
PFAS have been manufactured and used in a wide variety of industries around the world,
including in the United States since the 1950's. PFAS have strong, stable, carbon-fluorine (C-F)
bonds, making them resistant to hydrolysis, photolysis, microbial degradation, and metabolism
(Ahrens, 2011; Buck et al., 2011; Beach et al., 2006). There are many families or classes of
PFAS, each containing many individual structural homologues that can exist as either branched-
chain or straight-chain isomers (Buck et al., 2011). The chemical structures of PFAS enable them
to repel water and oil, remain chemically and thermally stable, and exhibit surfactant properties;
these properties make PFAS useful for commercial and industrial applications and make some
PFAS extremely persistent in the human body and the environment (Calafat et al., 2019; Calafat
et al., 2007). Because of their widespread use, physicochemical properties, persistence, and
bioaccumulation potential, many different PFAS co-occur in environmental media (e.g., air,
water, ice, sediment) and in tissues and blood of aquatic and terrestrial organisms, including
humans.
To understand and address the complexities associated with PFAS, the U.S. Environmental
Protection Agency (EPA) is developing human health toxicity assessments for individual PFAS,
in addition to other components of the broader PFAS action plan underway at EPA
(https://www.epa.eov/pfas/epas-pfas-action-plan). The updated toxicity assessment that was
developed for PFOS according to the scope and methods outlined in this protocol builds upon
several other assessments, including the Health Effects Support Document for PFOS (U. S. EPA,
2016c) (hereafter referred to as the 2016 PFOS HESD) and Proposed Approaches to the
Derivation of a Draft Maximum Contaminant Level Goal for Perfluorooctane Sulfonic Acid
(PFOS) (CASRN335-67-1) in Drinking Water (U.S. EPA, 2021c), which was released to the
public for review by the Science Advisory Board (SAB) in November 2021.
This protocol describes the methods used for conducting the systematic review and dose-
response analyses for the assessment of PFOS {Human Health Toxicity Assessment for
Perfluorooctane Sulfonic Acid (PFOS) and Related Salts) and has been updated to address
comments from the SAB. It should be noted that PFOA and PFOS underwent some steps of
systematic review (e.g., literature searches) concurrently.
A-l
-------
APRIL 2024
A.l Overview of Background Information and Systematic
Review Protocol
The methods used to conduct the systematic review for PFOS are consistent with the methods
described in the draft and final EPA ORD Staff Handbook for Developing IRIS Assessments (U.S.
EPA, 2022b, 2020a) (hereafter referred to as the Integrated Risk Information System (IRIS)
Handbook) and a companion publication (Thayer et al., 2022). EPA's IRIS Handbook has
incorporated feedback from the National Academy of Sciences (NAS) at workshops held in 2018
and 2019 and was well regarded by the NAS review panel for reflecting "significant
improvements made by EPA to the IRIS assessment process, including systematic review
methods for identifying chemical hazards" (NASEM, 2021). Furthermore, EPA's IRIS program
has used the IRIS Handbook to develop toxicological reviews for numerous chemicals, including
some PFAS (U.S. EPA, 2023b, 2022a). Though the IRIS Handbook was finalized concurrently
with the development of this assessment, the revisions in the final IRIS Handbook compared to
the draft version do not conflict with the methods used in this assessment. The assessment team
concluded that implementing these minor changes in study quality evaluation between the draft
and final IRIS Handbook versions would not change the assessment conclusions. Therefore, EPA
considers the methods described herein to be consistent with the final IRIS Handbook and cites
this version accordingly. Additionally, the methods used to conduct the systematic review are
also consistent with and largely mirror the Systematic Review Protocol for the PFBA, PFHxA,
PFHxS, PFNA, and PFDA (anionic and acidforms) IRIS Assessments (U.S. EPA, 2020b).
The Safe Drinking Water Act (SDWA) regulatory process enables EPA to receive comments and
feedback on the systematic review protocol, including through SAB early input and via the
public comment period associated with rule proposal. This protocol has been updated based on
SAB recommendations to improve the clarity and transparency of the methods descriptions. It
now includes information about additional data sources and how they were evaluated and
expands the application of systematic review through dose-response analysis.
A.l.l Summary of Background Information
This section summarizes more detailed sections on these topics from the Human Health Toxicity
Assessment for Perfluorooctane Sulfonic Acid (PFOS) and Related Salts (hereafter referred to as
the Toxicity Assessment, (U.S. EPA, 2024)) and is provided for context. Please refer to the
Toxicity Assessment (U.S. EPA, 2024) for more detailed information about chemical identity,
physical-chemical properties, and occurrence.
A.l.l.1 Chemical Identity
PFOS is a PFAA that was used as an aqueous dispersion agent and emulsifier in a variety of
water-, oil-, and stain-repellent products (e.g., agricultural chemicals, alkaline cleaners, carpets,
firefighting foam, floor polish, textiles) (NLM, 2022). It can exist in linear- or branched-chain
isomeric form. PFOS is a strong acid that is generally present as the sulfonate anion at typical
environmental pH values. Therefore, this assessment applies to all isomers of PFOS, as well as
nonmetal salts of PFOS that would be expected to dissociate in aqueous solutions of pH ranging
from 4 to 9 (e.g., in the human body). PFOS is stable in environmental media because it is
resistant to environmental degradation processes such as biodegradation, photolysis, and
A-2
-------
APRIL 2024
hydrolysis. In water, no natural degradation has been demonstrated, and dissipation is by
advection, dispersion, and sorption to particulate matter.
A.1.1.2 Occurrence Summary
Key PFOS occurrence information is summarized below. More detail is provided in Chapter 1 of
the Toxicity Assessment (U.S. EPA, 2024).
A.l.1.2.1 Biomonitoring
The CDC NHANES has measured blood serum concentrations of several PFAS in the general
U.S. population since 1999. PFOS has been detected in up to 98% of serum samples taken in
biomonitoring studies that are representative of the U.S. general population; however, blood
levels have dropped 60% to 80% between 1999 and 2014, presumably due to restrictions on its
commercial usage in the United States.
A.l.1.2.2 Occurrence in Water
PFOS is one of the dominant PFAS compounds detected in ambient water, along with PFOA
(Remucal, 2019; Dinglasan-Panlilio et al., 2014; Zareitalabad et al., 2013; Benskin et al., 2012;
Ahrens, 2011; Nakayama et al., 2007).
Data from the third Unregulated Contaminant Monitoring Rule (UCMR 3), collected from 2013-
2015, are currently the best available national occurrence data for PFOA and PFOS (U.S. EPA,
2023a, 2021a, 2017). Under UCMR 3, 36,972 samples from 4,920 PWSs were analyzed for
PFOA and PFOS. The minimum reporting level (MRL)1 for PFOA was 0.02 |ig/L and the MRL
for PFOS was 0.04 (J,g/L. A total of 1.37% of samples had reported detections (>MRL) of at least
one of the two compounds.
A.1.2 Problem Formulation
As described in the Toxicity Assessment (U.S. EPA, 2024), EPA conducted this updated
assessment of PFOS (including all isomers as well as nonmetal salts of PFOS that would be
expected to dissociate in aqueous solutions of pH ranging from 4 to 9 (e.g., in the human body))
to support derivation of chronic cancer and noncancer toxicity values for PFOS. This problem
formulation section will describe the key considerations and scope of the assessment, which were
informed in part by EPA's past human health assessments of PFOS (2016 PFOS HESD and 2021
Proposed Approaches to the Derivation of a Draft Maximum Contaminant Level Goal for
Perfluorooctane Sulfonic Acid (PFOS) (CASRN 335-67-1) in Drinking Water) (U.S. EPA,
2021c) as well as ongoing and final EPA assessments of other PFAS (e.g., perfluorobutanoic
acid (PFBA) and draft perfluorohexanoic acid (PFHxA), perfluorohexane sulfonate (PFHxS),
perfluorononanoic acid (PFNA), and perfluorodecanoic acid (PFDA) IRIS assessments (U.S.
EPA, 2020b)).
The 2016 PFOS HESD identified several adverse health outcomes associated with PFOS
exposure based on results from animal toxicological and epidemiological studies, including
1 The reporting level is the threshold at or above which a contaminant's presence or concentration is officially quantitated. In the
case of many of EPA's nationwide drinking water studies, the selected reporting level is known officially as the MRL. The MRL
for each contaminant in each study is set at a level that EPA believes can be achieved with specified confidence by a broad
spectrum of capable laboratories across the nation (U.S. EPA, 202 Id).
A-3
-------
APRIL 2024
developmental effects (e.g., decreased birth weight, accelerated puberty, skeletal variations),
cancer (e.g., liver), liver effects (e.g., tissue damage), immune effects (e.g., antibody production
and immunity), thyroid effects (e.g., hypothyroidism), and other effects (e.g., cholesterol
changes). It concluded that there was "suggestive evidence of carcinogenic potential" for PFOS.
EPA's 2021 Proposed Approaches to the Derivation of a Draft Maximum Contaminant Level
Goal for Perfluorooctane Sulfonic Acid (PFOS) (CASRN 335-67-1) in Drinking Water) (U.S.
EPA, 2021c) evaluated associations between PFOS and all cancer and noncancer health
outcomes. After reviewing that draft scoping assessment, the SAB recommended that the scope
be narrowed to focus on the five priority health outcomes that have the strongest weight of
evidence (immune, developmental, hepatic, cardiovascular, and cancer), most of which were also
identified in the conclusions from the 2016 PFOS HESD. Therefore, the current assessment
provides a comprehensive systematic review of all health effects literature published through
February 2022 for these five health outcomes. Mechanistic data for these health outcomes were
also synthesized. For other health outcomes beyond the five priority outcomes, the current
assessment summarizes the health effects literature published prior to 2016 and includes a
systematic review of the health effects literature published from 2016-2020.
The Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA (anionic and
acidforms) IRIS Assessments outlines key science issues related to PFAS in general (U.S. EPA,
2020b), many of which are relevant to PFOS. They include: toxicokinetic differences across
species and sexes; human relevance of effects in animals that involve peroxisome proliferator-
activated receptor alpha (PPARa); potential confounding by other PFAS exposures in
epidemiology studies; and toxicological relevance of changes in certain hepatic endpoints in
rodents. Differences in PFOS toxicokinetics across species and sexes were accounted for in the
PFOS-specific animal and human pharmacokinetic models (see Toxicity Assessment, (U.S. EPA,
2024)). The human relevance of effects in animals that involve PPARa was investigated in the
mechanistic syntheses of the five priority health outcomes (see Toxicity Assessment, (U.S. EPA,
2024)). Potential confounding by other PFAS (and other co-occurring contaminants) in
epidemiology studies was considered as part of the confounding domain during study quality
evaluations. Specifically, if a study did not account for potential confounding with other co-
occurring PFAS in its statistical analyses, then the maximum quality rating this domain could
receive was adequate. Concerns about potential confounding by other PFAS were limited when
there was evidence that exposure was predominantly PFOS-based (such as in certain
occupational or high-exposure studies) and the potential for co-exposure was minimal, or the
correlations between co-exposures were small. The toxicological relevance of changes in certain
hepatic endpoints in rodents was accounted for by incorporating the Hall (2012) criteria into the
animal hepatic synthesis and hazard conclusions.
The Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA (anionic and
acidforms) IRIS Assessments outlines key science issues related to PFAS in general (U.S. EPA,
2020b), many of which are relevant to PFOS. They include: toxicokinetic differences across
species; human relevance of effects in animals that involve peroxisome proliferator-activated
receptor alpha (PPARa); potential confounding by other PFAS exposures in epidemiology
studies; and toxicological relevance of changes in certain hepatic endpoints in rodents.
Differences in PFOS toxicokinetics across species were accounted for in the PFOS-specific
animal and human toxicokinetic pharmacokinetic models (see Toxicity Assessment, (U.S. EPA,
2024)). The human relevance of effects in animals that involve PPARa was investigated in the
A-4
-------
APRIL 2024
mechanistic syntheses of the five priority health outcomes (see Toxicity Assessment, (U.S. EPA,
2024)). Potential confounding by other PFAS (and other co-occurring contaminants) in
epidemiology studies was considered as part of the confounding domain during study quality
evaluations and is further discussed in Section 5 of the Toxicity Assessment (U.S. EPA, 2024).
Specifically, if a study did not account for potential confounding with other co-occurring PFAS
in its statistical analyses, then the maximum quality rating this domain could receive was
adequate. Concerns about potential confounding by other PFAS were limited when there was
evidence that exposure was predominantly PFOS-based (such as in certain occupational or high-
exposure studies) and the potential for co-exposure was minimal, or the correlations between co-
exposures were small. The toxicological relevance of changes in certain hepatic endpoints in
rodents was accounted for by incorporating the Hall (2012) criteria into the animal hepatic
synthesis and hazard conclusions.
An additional key science issue that EPA has encountered for PFAS toxicity assessments is a
general lack of data on human and ecological toxicity. For PFOS, this is less of an issue as there
has been substantial research and publication of both epidemiological and animal toxicological
studies.
A.1.3 Overall Objective and Specific Aims
A. 1.3.1 Objective
The primary objective of this toxicity assessment for PFOS is to support derivation of chronic
cancer and noncancer toxicity values for PFOS, as well as update the cancer descriptor for
PFOS, if warranted. EPA also considered potential pathways of exposure and derived a relative
source contribution (RSC) specific to the final RfD for PFOS. The toxicity values, cancer
classification, and RSC derived in this assessment build upon the work completed in the
Proposed Approaches to the Derivation of a Draft Maximum Contaminant Level Goal for
Perfluorooctane Sulfonic Acid (PFOS) (CASRN1763-23-1) in Drinking Water (U.S. EPA,
2021c) and in the 2016 PFOS HESD (U.S. EPA, 2016c).
A.l.3.2 Specific Aims
The specific aims of the PFOS toxicity assessment document are to:
• Describe and document transparently the literature searches conducted and systematic
review methods used to identify health effects information (epidemiological and animal
toxicological studies and physiologically based pharmacokinetic models) in the literature
(Sections 2 and 3 of the Toxicity Assessment (U.S. EPA, 2024); Appendix A and
Appendix B).
• Describe and document literature screening methods, including use of the Populations,
Exposures, Comparators, and Outcomes (PECO) criteria and the process for tracking
studies throughout the literature screening (Section 2 of the Toxicity Assessment (U.S.
EPA, 2024); Appendix A).
• Identify epidemiological and animal toxicological literature that reports health effects
after exposure to PFOS (and its related salts) as outlined in the PECO criteria (Section 3
of the Toxicity Assessment (U.S. EPA, 2024)).
A-5
-------
APRIL 2024
• Describe and document the study quality evaluations conducted on epidemiological and
animal toxicological studies considered potentially useful for point-of-departure (POD)
derivation (Section 3 of the Toxicity Assessment (U.S. EPA, 2024)).
• Describe and document the data from all epidemiological studies and animal toxicological
studies that were considered for POD derivation (Section 3 of the Toxicity Assessment
(U.S. EPA, 2024)).
• Synthesize and document the adverse health effects evidence across studies. The
assessment focuses on synthesizing the available evidence for five priority health
outcomes that were found to have the strongest weight of evidence, as recommended by
the SAB - developmental, hepatic, immune, and cardiovascular effects, and cancer
(Section 3 of the Toxicity Assessment (U.S. EPA, 2024)) - and also provides
supplemental syntheses of evidence for dermal, endocrine, gastrointestinal, hematologic,
metabolic, musculoskeletal, nervous, ocular, renal, and respiratory effects, reproductive
effects in males or females, and general toxicity (Appendix C).
• Evaluate and document the available mechanistic information (including toxicokinetic
understanding) associated with PFOS exposure to inform interpretation of findings related
to potential health effects in studies of humans and animals, with a focus on five priority
health outcomes (developmental, hepatic, immune, and cardiovascular effects, and cancer)
(Section 3 of the Toxicity Assessment (U.S. EPA, 2024)).
• Develop and document strength of evidence judgments across studies (or subsets of
studies) separately for epidemiological, animal toxicological, and mechanistic lines of
evidence for the five priority health outcomes (Section 3 of the Toxicity Assessment (U.S.
EPA, 2024)).
• Develop and document integrated expert judgments across evidence streams (i.e.,
epidemiological, animal toxicological, and mechanistic streams) as to whether and to what
extent the evidence supports that exposure to PFOS has the potential to be hazardous to
humans (Section 3 of the Toxicity Assessment (U.S. EPA, 2024)).
• Determine the cancer classification for PFOS using a weight-of-evidence approach
(Section 3 of the Toxicity Assessment (U.S. EPA, 2024)).
• Describe and document the attributes used to evaluate and select studies for derivation of
toxicity values. These attributes are considered in addition to the study confidence
evaluation domains and enable extrapolation to relevant exposure levels (e.g., studies with
exposure levels near the range of typical environmental human exposures, broad exposure
range, or multiple exposure levels) (Section 4 of the Toxicity Assessment (U.S. EPA,
2024)).
• Describe and document the dose-response analyses conducted on the studies identified for
POD derivation (Section 4 of the Toxicity Assessment (U.S. EPA, 2024)).
• Derive candidate RfDs (Section 4.1 (U.S. EPA, 2024)) and CSFs (Section 4.2 of the
Toxicity Assessment (U.S. EPA, 2024)), select the final RfD (Section 4.1.6 of the
Toxicity Assessment (U.S. EPA, 2024)) and CSF (Section 4.2.3 of the Toxicity
Assessment (U.S. EPA, 2024)) for PFOS, and describe the rationale.
• Characterize hazards (e.g., uncertainties, data gaps) (Sections 3, 4, and 5 of the Toxicity
Assessment (U.S. EPA, 2024)).
A-6
-------
APRIL 2024
A.1.4 Populations, Exposures, Comparators, and Outcomes
(PECO) Criteria
This section describes the PECO criteria that were developed and used for this assessment.2 As
described in the IRIS Handbook (U.S. EPA, 2022c), the PECO criteria provide the framework
for literature search strategies and are the inclusion/exclusion criteria by which literature search
results will be screened for relevancy to identify epidemiological and animal toxicological
evidence that addresses the aims of the assessment. For the PFOS assessment, the PECO criteria
were used to screen results of the literature searches to identify and prioritize the dose-response
literature and studies containing pharmacokinetic (PK) or PBPK models. For studies captured in
the 2019 and 2020 literature searches, the PECO criteria were used to screen and categorize
("tag") studies of PFOS absorption, distribution, metabolism, and excretion (ADME) and studies
with mechanistic data for further evaluation using ADME- and mechanistic-specific PECO
criteria. ADME, mechanistic, and other supplemental studies captured in the 2022 and 2023
literature searches were not tagged or considered further in this assessment.
Table A-l describes the PECO criteria used to screen the results of the literature search (the
literature search is described in Section A. 1.5 of this appendix). ADME- and mechanistic-
specific PECO criteria are outlined in Table A-2 and Table A-3, respectively.
Table A-l. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for a
Systematic Review on the Health Effects From Exposure to PFOA and PFOS
PECO Element Inclusion Criteria
Population Human: Any population and lifestage (occupational or general population, including children
and other sensitive populations).
Animal: Nonhuman mammalian animal species (whole organism) of any lifestage (including
preconception, in utero, lactation, peripubertal, and adult stages).
In vitro/cell studies or in silico/modeling toxicity studies should be tagged as supplemental.
Exposure Any exposure to PFOA, PFOS, and/or the salts of PFOA/PFOS, including but not limited
to:
PFOA {Chemical Abstracts Service (CAS) number 335-67-1).
Other names: perfluorooctanoate; perfluorooctanoic acid; 2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
pentadecafluorooctanoic acid; pentadecafluoro-l-octanoic acid; pentadecafluoro-n-octanoic
acid; perfluorocaprylic acid; pentadecafluorooctanoic acid; perfluoroheptanecarboxylic acid;
octanoic-acid, pentadecafluoro-
Relevant Salts of PFOA: ammonium perfluorooctanoate (APFO), sodium perfluorooctanoate,
potassium perfluorooctanoate
PFOS {CAS number 1763-23-1).
Other names: perfluorooctane sulfonate, perfluorooctanesulfonic acid, perfluorooctane sulfonic
acid, perfluorooctane sulphonate, perfluorooctanyl sulfonate, heptadecafluorooctane-1-
sulphonic, heptadecafluoro-l-octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
heptadecafluoro-l-octanesulfonic acid
Relevant Salts of PFOS: lithium perfluorooctanesulfonate, potassium perfluorooctanesulfonate
(K+PFOS), ammonium perfluorooctanesulfonate, sodium perfluorooctanesulfonate
Human: Any exposure to PFOA or PFOS via oral routes. Other exposure routes, including
inhalation, dermal, or unknown/multiple routes will be tracked during title and abstract
screening and tagged as "potentially relevant supplemental information."
2 Note: Although this appendix and its accompanying Toxicity Assessment (U.S. EPA, 2024) pertain to PFOS, the PECO criteria
also cover PFOA because the literature searching and screening were performed concurrently for PFOA and PFOS.
A-7
-------
APRIL 2024
PECO Element
Inclusion Criteria
Animal: Any exposure to PFOA or PFOS via oral routes. Other exposure routes, including
inhalation, dermal, injection or unknown/multiple routes, will be tracked during title and
abstract screening and tagged as "potentially relevant supplemental information." Studies
involving exposures to mixtures will be included only if they include exposure to PFOA or
PFOS alone. Studies with less than 28 d of dosing, with the exception of reproductive,
developmental, immune and neurological health outcome studies, should be tagged as
supplemental.
Human: A comparison or referent population exposed to lower levels (or no exposure/exposure
below detection limits) of PFOA or PFOS, or exposure to PFOA or PFOS for shorter periods of
time. Case reports and case series will be tracked as "potentially relevant supplemental
information."
Animal: A concurrent control group exposed to vehicle-only treatment or untreated control.
All health outcomes (both cancer and noncancer).
Studies describing physiologically based pharmacokinetic (PBPK) models will be included.
Comparator
Outcome
PBPK Models
Epidemiological, animal toxicological, and in vitro studies tagged as containing potentially
relevant ADME data were further screened using ADME-focused PECO criteria (Table A-2).
Key information from each study meeting the ADME-focused PECO criteria was extracted using
ICF's litstream™ software.
Table A-2. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for
Absorption, Distribution, Metabolism, and/or Excretion (ADME) Studies
PECO
Element
Inclusion Criteria
Population Human: Any population and lifestage (occupational or general population, including children and
other sensitive populations): whole organism, tissues, individual cells, or biomolecules.
Animal: Select nonhuman mammalian animal species: only nonhuman primates, rats, and mice
(whole organism, tissues, individual cells, or biomolecules) of any lifestage (preconception, in
utero, lactation, peripubertal, and adult stages).
Exposure Any exposure to PFOA, PFOS, and/or the salts of PFOA/PFOS, including in vitro, in vivo (by
various routes of exposure), and ex vivo. In silico studies will also be included if the model system
can be linked to a PECO-relevant species.
PFOA (CAS number 335-67-1).
Other names: perfluorooctanoate, perfluorooctanoic acid, perfluoroctanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic acid, pentadecafluoro-l-octanoic acid,
pentadecafluoro-n-octanoic acid, octanoic acid, pentadecafluoro-, perfluorocaprylic acid,
pentadecafluorooctanoic acid, perfluoroheptanecarboxylic acid, octanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-, ammonium perfluorooctanoate (APFO), sodium
perfluorooctanoate, potassium perfluoroctanoate
PFOS (CAS number 1763-23-1).
Other names: perfluorooctane sulfonate, perfluorooctanesulfonic acid, perfluorooctane sulfonic
acid, perfluorooctane sulphonate, perfluorooctane sulfonate, perfluorooctanyl sulfonate,
heptadecafluorooctane-l-sulphonic, heptadecafluoro-l-octanesulfonic acid,
1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-l-octanesulfonic acid,
heptadecafluorooctanesulfonic acid, lithium perfluorooctanesulfonate, potassium
perfluorooctanesulfonate, ammonium perfluorooctanesulfonate, sodium perfluorooctanesulfonate
Comparator Any comparison that informs PFOA or PFOS (1) absorption by the oral, inhalation, or dermal route
of exposure, (2) distribution across biological compartments, (3) metabolism, and/or (4) excretion.
Outcome Any examination of PFOA and/or PFOS (1) absorption of dose through gastrointestinal (GI) tract,
lungs, or skin, (2) distribution across biological compartments, (3) metabolism, and/or (4)
excretion. Studies describing PK models for PFOA and/or PFOS will be included.
A-8
-------
APRIL 2024
PECO
Element
Inclusion Criteria
Information and terms that are typically found in relevant ADME/PK modeling studies include the
following:
Absorption: Bioavailability; absorption rate(s); uptake rates; tissue location of absorption (e.g.,
stomach versus intestine, nasal versus lung); blood:air partition coefficient (PC); irritant/respiratory
depression; overall mass transfer coefficient; gas-phase diffusivity; gas-phase mass transfer
coefficient; liquid- (or tissue-) phase mass transfer coefficient; deposition fraction; retained
fractions; computational fluid (airway) dynamics.
Distribution: Volume of distribution (Vd) and parameters that determine Vd, including blood:
tissue PCs (especially for the target or a surrogate tissue) or lipophilicity; tissue burdens; storage
tissues or tissue components (e.g., serum binding proteins) and the binding coefficients;
transporters (active and passive).
Note: PFOA/PFOS are not metabolized so we are not expecting studies that focus on metabolites.
The terms below are general terms associated with metabolism.
Metabolism: Metabolic/biotransformation pathway(s); enzymes involved; metabolic rate;
maximum rate of transport (Vmax), Michaelis constant (Km);; metabolic induction; metabolic
inhibition, Ki; metabolic saturation/non-linearity; key organs involved in metabolism; key
metabolites (if any)/pathways; metabolites measured; species-, inter-individual-, and/or age-related
differences in enzyme activity or expression ("ontogeny"); site-specific activation (may be
toxicologically significant, but little systemic impact); cofactor (e.g., glutathione) depletion.
Excretion: Route(s)/pathway(s) of excretion for parent and metabolites; urine, fecal, exhalation,
hair, sweat, lactation; elimination rate(s); mechanism(s) of excretion (e.g., passive diffusion, active
transport).
Notes: CAS = Chemical Abstracts Service; PK = pharmacokinetic ADME = absorption, distribution, metabolism, and/or
excretion.
Epidemiological and animal toxicological studies that were tagged as containing potentially
relevant mechanistic data were further screened using mechanistic-focused PECO criteria (Table
A-3). Studies meeting the mechanistic-focused PECO criteria underwent a light extraction of key
study information using ICF's litstream™ software.
Table A-3. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for
Mechanistic Studies
PECO
Element
Evidence
Population Human: Any population and lifestage (occupational or general population, including children and
other sensitive populations).
Animal: Select mammals (i.e., nonhuman primates and rodents (i.e., rats, mice, rabbits, guinea
pigs, other rodent models) and fish (i.e., zebrafish) of any lifestage (preconception, in utero,
lactation, peripubertal, and adult stages).
Ex vivo, in vitro, in silico: Cultures of human or animal cells from relevant animal models
(primary, immortalized, transformed), organ slices, organotypic culture, in vitro molecular or
biochemical assay systems. In silico modeling data if it informs PFOA/PFOS MOA.
Exposure Any exposure to PFOA, PFOS, and/or the salts of PFOA/PFOS, including in vitro, in vivo (by
various routes of exposure), and ex vivo. In silico studies will also be included if the model
system can be linked to a PECO-relevant species.
PFOA (CAS number 335-67-1).
Other names: perfluorooctanoate, perfluorooctanoic acid, perfluoroctanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic acid, pentadecafluoro-l-octanoic acid,
pentadecafluoro-n-octanoic acid, octanoic acid, pentadecafluoro-, perfluorocaprylic acid,
pentadecafluorooctanoic acid, perfluoroheptanecarboxylic acid, octanoic acid,
A-9
-------
APRIL 2024
PECO
Element
Comparator
Outcome
Evidence
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-, ammonium perfluorooctanoate (APFO), sodium
perfluorooctanoate, potassium perfluoroctanoate
PFOS (CAS number 1763-23-1).
Other names: perfluorooctane sulfonate, perfluorooctanesulfonic acid, perfluorooctane sulfonic
acid, perfluorooctane sulphonate, perfluorooctane sulfonate, perfluorooctanyl sulfonate,
heptadecafluorooctane-l-sulphonic, heptadecafluoro-l-octanesulfonic acid,
1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-l-octanesulfonic acid,
heptadecafluorooctanesulfonic acid, lithium perfluorooctanesulfonate, potassium
perfluorooctanesulfonate, ammonium perfluorooctanesulfonate, sodium perfluorooctanesulfonate
Human: Comparison to group with no exposure or lower exposure.
Animal, ex vivo, in vitro, in silico: Comparison to an appropriate vehicle or no treatment control.
Any mechanistic data related to the MOA of PFOA/PFOS toxicity. This may include molecular
initiating events with PFOA/PFOS or downstream key events that inform the MOA or adverse
outcome pathway linking PFOA/PFOS exposure to disease.
Notes: MOA = mode of action; CAS = Chemical Abstracts Service.
A.1.5 Literature Search
EPA assembled a database of epidemiological, animal toxicological, mechanistic, and
toxicokinetic studies for this updated toxicity assessment based on three data streams: 1)
literature published from 2013 through 2019 and then updated in the course of this review (i.e.,
through February 6, 2023) identified via literature searches of a variety of publicly available
scientific literature databases, 2) literature identified via other sources (e.g., searches of the gray
literature, studies shared with EPA by the SAB, and studies submitted through public comment),
and 3) literature identified in EPA's 2016 PFOA and PFOS HESDs, which captured literature
through 2013 (U.S. EPA, 2016c, d).
A. 1.5.1 Literature Search Strategies
The following sections describe literature search strategies used for databases and for additional
sources. They also describe methods used to incorporate studies from the 2016 PFOS HESD and
other sources into the literature database. The literature search strategy included searches within
core literature databases (e.g., PubMed®, Web of Science™) as well as relevant domestic and
international non-periodical "gray" literature, such as technical reports, monographs, and
conference and symposium proceedings prepared by select committees or bodies (e.g., those
convened by the National Academy of Sciences or the World Health Organization (WHO)).
A.1.5.2 Database Searches
The database literature searches for this updated assessment focused only on the chemical name
(PFOS and related salts) with no limitations on lines of evidence (i.e., human/epidemiological,
animal, in vitro, in silico) or health outcomes. These searches comprised all literature related to
health effects resulting from acute, subchronic, and chronic exposure durations, and from
inhalation, oral, dermal, and injection exposure studies. Epidemiological, animal toxicological,
and in vitro studies that provide MOA information were included, and data specifically useful for
addressing risks to children and other susceptible populations (e.g., the elderly, pregnant or
lactating women, genetically susceptible populations) were identified. The searches likewise
included ADME studies and models useful for dose-response assessment, such as dosimetry and
A-10
-------
APRIL 2024
PBPK models. The initial database search covered from January 2013 through April 11, 2019
(the 2019 literature search). That was subsequently updated by a search covering April 2019
through September 3, 2020 (2020 literature search), another covering September 2020 through
February 3, 2022 (2022 literature search), and a final supplemental search covering February
2022 through February 6, 2023 (described in Section A.3 below). The date field tag used for
these searches may reflect either the date the article was published in print or e-published which
may result in small amounts of literature being captured in a literature search despite being
published prior to the start date. At the recommendation of SAB peer reviewers, the 2022
literature search and supplemental 2023 literature search focused on the five priority health
outcomes that have been concluded to have the strongest evidence (developmental, hepatic,
immune, and cardiovascular effects, and cancer). EPA considered mechanistic and toxicokinetic
data identified through the September 2020 literature search, as well as any more recent studies
recommended by the SAB.
The databases listed below were searched for literature containing the search strings identified in
Table A-4 and Table A-5:
• Web of Science™ (Thomson Reuters),
• PubMed® (National Library of Medicine),
• ToxLine (incorporated into PubMed post 2019), and
• TSCATS (Toxic Substances Control Act Test Submissions).
Table A-4. Search String for April 2019 Database Searches
Database Search String Date Run
WoS ((TS="perfluorooctanoic acid" OR TS="perfluorooctane sulfonic acid") AND 4/10/2019
PY=(2013-2019) OR (TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-
Octanoic acid" OR TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic
acid" OR TS="3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-Hexanoyl fluoride" OR
TS="3,3,4,4,5,5,6,6,6-nonafluoro-2-oxohexanoyl fluoride" OR TS="Hexanoyl
fluoride, 3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-" OR TS="Octanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-" OR TS="Pentadecafluoro-l-
octanoic acid" OR TS="Pentadecafluoro-n-octanoic acid" OR
TS="Pentadecafluorooctanoic acid" OR TS="Perfluorocaprylic acid" OR
TS="Perfluoroctanoic acid" OR TS="Perfluoroheptanecarboxylic acid" OR
TS="perfluorooctanyl sulfonate" OR TS="Perfluorooctanoic acid" OR
TS="Octanoic acid, pentadecafluoro-" OR TS="Perfluorooctanoate" OR
TS="perfluorooctane sulfonate" OR TS="A 5717" OR TS="EF 201" OR
TS="Eftop EF 201" OR TS="Perfluoro-l-heptanecarboxylic acid" OR
TS="l,l,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-Heptadecafluoro-l-octanesulfonic acid"
OR TS=" 1-Octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
heptadecafluoro-" OR TS="l-Perfluorooctanesulfonic acid" OR TS="EF 101"
OR TS="Eftop EF 101" OR TS="Heptadecafluoro-l-octanesulfonic acid" OR
TS="Heptadecafluorooctane-l-sulphonic acid" OR TS="Perfluorooctane
sulfonate" OR TS="perfluorooctane sulfonate" OR TS="Perfluorooctane
sulfonic acid" OR TS="Perfluorooctanesulfonic acid" OR
TS="Perfluorooctylsulfonic acid" OR TS="perfluorooctane sulphonate" OR
TS="perfluorooctane sulfonate" OR TS="l-Octanesulfonic acid,
heptadecafluoro-"OR TS="Heptadecafluorooctanesulfonic acid" OR
TS="Perfluoro-n-octanesulfonic acid" OR TS="Perfluorooctane Sulphonic
Acid" OR TS="Perfluorooctanesulfonate" OR TS="Perfluorooctylsulfonate"
OR ((TS="PFOA" OR TS="PFOS") AND (TS="fluorocarbon*" OR
A-ll
-------
APRIL 2024
Database Search String Date Run
TS="fluorotelomer*" OR TS="polyfluoro*" OR TS="perfluoro-*" OR
TS="perfluoroa*" OR TS="perfluorob*" OR TS="perfluoroc*" OR
TS="perfluorod*" OR TS="perfluoroe*" OR TS="perfluoroh*" OR
TS="perfluoron*" OR TS="perfluoroo*" OR TS="perfluorop*" OR
TS="perfluoros*" OR TS= "perfluorou*" OR TS="perfluorinated" OR
TS="fluorinated" OR TS="PFAS"))) AND PY=(2013-2019))
PubMed (335-67-1 [rn] OR 1763-23-l[rn] OR 45298-90-6[rn] OR "perfluorooctanoic 4/10/2019
acid"[nm] OR "perfluorooctane sulfonic acid"[nm]) AND
(2013/01/01:3000[pdat] OR 2013/01/01:3000[mhda] OR
2013/01/01:3000[edat] OR 2013/01/01:3000[crdt]) OR
(("2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-Octanoic acid"[tw] OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic acid"[tw] OR
"3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-Hexanoyl fluoride"[tw] OR
"3,3,4,4,5,5,6,6,6-nonafluoro-2-oxohexanoyl fluoride"[tw] OR "Hexanoyl
fluoride, 3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-"[tw] OR "Octanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-"[tw] OR "Pentadecafluoro-1-
octanoic acid"[tw] OR "Pentadecafluoro-n-octanoic acid"[tw] OR
"Pentadecafluorooctanoic acid"[tw] OR "Perfluorocaprylic acid"[tw] OR
"Perfluoroctanoic acid"[tw] OR "Perfluoroheptanecarboxylic acid"[tw] OR
"perfluorooctanyl sulfonate" [tw] OR "Perfluorooctanoic acid"[tw] OR
"Octanoic acid, pentadecafluoro-"[tw] OR "Perfluorooctanoate"[tw] OR
"perfluorooctane sulfonate"[tw] OR "A 5717"[tw] OR "EF 201"[tw] OR "Eftop
EF 201"[tw] OR "Perfluoro-l-heptanecarboxylic acid"[tw] OR
"1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-Heptadecafluoro-l-octanesulfonic acid"[tw]
OR "1-Octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-
"[tw] OR "1-Perfluorooctanesulfonic acid"[tw] OR "EF 101"[tw] OR "Eftop
EF 101"[tw] OR "Heptadecafluoro-l-octanesulfonic acid"[tw] OR
"Heptadecafluorooctane-l-sulphonic acid"[tw] OR "Perfluorooctane
sulfonate" [tw] OR "perfluorooctane sulfonate "[tw] OR "Perfluorooctane
sulfonic acid"[tw] OR "Perfluorooctanesulfonic acid"[tw] OR
"Perfluorooctylsulfonic acid"[tw] OR "perfluorooctane sulphonate" [tw] OR
"perfluorooctane sulfonate" [tw] OR "1-Octanesulfonic acid, heptadecafluoro-
"[tw] OR "Heptadecafluorooctanesulfonic acid"[tw] OR "Perfluoro-n-
octanesulfonic acid"[tw] OR "Perfluorooctane Sulphonic Acid"[tw] OR
"Perfluorooctanesulfonate"[tw] OR "Perfluorooctylsulfonate"[tw] OR
(("PFOA"[tw] OR "PFOS"[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] ORpolyfluoro*[tw] ORperfluoro-*[tw] OR
perfluoroa*[tw] ORperfluorob*[tw] ORperfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] ORperfluoroh*[tw] ORperfluoron*[tw] OR
perfluoroo*[tw] ORperfluorop*[tw] ORperfluoros*[tw] OR perfluorou* [tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw]))) AND
(2013/01/01:3000[pdat] OR 2013/01/01:3000[mhda] OR
2013/01/01:3000[edat] OR 2013/01/01:3000[crdt]))
Toxline @AND+@OR+("perfluorooctane sulfonate"+"pfos"+"perfluorooctanesulfonic 4/11/2019
acid"+"perfluorooctane sulfonic acid"+"perfluorooctane
sulphonate"+"perfluorooctane sulfonate"+"perfluorooctanyl
sulfonate"+"Heptadecafluorooctane- l-sulphonic"+"Heptadecafluoro-1 -
octanesulfonic acid"+"l,l,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-l-
octanesulfonic acid"+"perfluorooctanoate"+"perfluorooctanoic
acid"+"perfluoroctanoic acid"+"pfoa"+"2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
pentadecafluorooctanoic acid"+"Pentadecafluoro-l-octanoic
acid"+"Pentadecafluoro-n-octanoic acid"+"Octanoic acid, pentadecafluoro-
"+"Perfluorocaprylic acid"+"Pentadecafluorooctanoic
acid"+"perfluoroheptanecarboxylic acid"+@TERM+@rn+335-67-
A-12
-------
APRIL 2024
Database
Search String
Date Run
l+@TERM+@rn+1763-23 - l+@TERM+@rn+45298-90-
6)+@NOT+@org+pubmed+@AND+@RAN GE+y r+2013+2019
TSCATS
@ AND+@OR+@rn+" 335-67-
l"+@AND+@org+TSCATS+@NOT+@org+pubmed
@AND+@OR+@rn+" 1763-23-
l"+@AND+@org+TSCATS+@NOT+@org+pubmed
4/11/2019
Table A-5. Search String for September 2020, February 2022, and February 2023 Database
Searches
Database
Search String
Date Run
PubMed (335-67-1 [rn] OR 1763-23-1 [rn] OR 45298-90-6[rn] OR "perfluorooctanoic 9/3/2020, 2/2/2022,
Batch IDs: acid"[nm] OR "perfluorooctane sulfonic acid"[nm] OR 2/6/2023
39678, 46137 "2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-Octanoic acid"[tw] OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic acid"[tw] OR
"3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-Hexanoyl fluoride"[tw] OR
"3,3,4,4,5,5,6,6,6-nonafluoro-2-oxohexanoyl fluoride"[tw] OR "Hexanoyl
fluoride, 3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-"[tw] OR "Octanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-"[tw] OR "Pentadecafluoro-1-
octanoic acid"[tw] OR "Pentadecafluoro-n-octanoic acid"[tw] OR
"Pentadecafluorooctanoic acid"[tw] OR "Perfluorocaprylic acid"[tw] OR
"Perfluoroctanoic acid"[tw] OR "Perfluoroheptanecarboxylic acid"[tw] OR
"perfluorooctanyl sulfonate" [tw] OR "Perfluorooctanoic acid"[tw] OR
"Octanoic acid, pentadecafluoro-"[tw] OR "Perfluorooctanoate"[tw] OR
"perfluorooctane sulfonate"[tw] OR "A 5717"[tw] OR "EF 201"[tw] OR
"Eftop EF 201"[tw] OR "Perfluoro-l-heptanecarboxylic acid"[tw] OR
"1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-Heptadecafluoro-l-octanesulfonic acid"[tw]
OR "1-Octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-
"[tw] OR "1-Perfluorooctanesulfonic acid"[tw] OR "EF 101"[tw] OR "Eftop
EF 101"[tw] OR "Heptadecafluoro-l-octanesulfonic acid"[tw] OR
"Heptadecafluorooctane-l-sulphonic acid"[tw] OR "Perfluorooctane
sulfonate"[tw] OR "perfluorooctane sulfonate"[tw] OR "Perfluorooctane
sulfonic acid"[tw] OR "Perfluorooctanesulfonic acid"[tw] OR
"Perfluorooctylsulfonic acid"[tw] OR "perfluorooctane sulphonate" [tw] OR
"perfluorooctane sulfonate" [tw] OR "1-Octanesulfonic acid, heptadecafluoro-
"[tw] OR "Heptadecafluorooctanesulfonic acid"[tw] OR "Perfluoro-n-
octanesulfonic acid"[tw] OR "Perfluorooctane Sulphonic Acid"[tw] OR
"Perfluorooctanesulfonate"[tw] OR "Perfluorooctylsulfonate"[tw] OR
(("PFOA"[tw] OR "PFOS"[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] ORpolyfluoro*[tw] ORperfluoro-*[tw] OR
perfluoroa*[tw] ORperfluorob*[tw] ORperfluoroc*[tw] ORperfluorod*[tw]
ORperfluoroe*[tw] ORperfluoroh*[tw] ORperfluoron*[tw] OR
perfluoroo*[tw] ORperfluorop*[tw] ORperfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw]))) AND
(2020/09/03:3000[dp])
(TS="perfluorooctanoic acid" OR TS="perfluorooctane sulfonic acid" OR 9/3/2020, 2/3/2022,
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-Octanoic acid" OR 2/6/2023
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluorooctanoic acid" OR
TS="3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-Hexanoyl fluoride" OR
TS="3,3,4,4,5,5,6,6,6-nonafluoro-2-oxohexanoyl fluoride" OR TS="Hexanoyl
fluoride, 3,3,4,4,5,5,6,6,6-nonafluoro-2-oxo-" OR TS="Octanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-pentadecafluoro-" OR TS="Pentadecafluoro-l-
Web of
Science
Batch IDs:
39681,46144
A-13
-------
APRIL 2024
Database Search String Date Run
octanoic acid" OR TS="Pentadecafluoro-n-octanoic acid" OR
TS="Pentadecafluorooctanoic acid" OR TS="Perfluorocaprylic acid" OR
TS="Perfluoroctanoic acid" OR TS="Perfluoroheptanecarboxylic acid" OR
TS="perfluorooctanyl sulfonate" OR TS="Perfluorooctanoic acid" OR
TS="Octanoic acid, pentadecafluoro-" OR TS="Perfluorooctanoate" OR
TS="perfluorooctane sulfonate" OR TS="A 5717" OR TS="EF 201" OR
TS="Eftop EF 201" OR TS="Perfluoro-l-heptanecarboxylic acid" OR
TS="l,l,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-Heptadecafluoro-l-octanesulfonic acid"
OR TS="l-Octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
heptadecafluoro-" OR TS="l-Perfluorooctanesulfonic acid" OR TS="EF 101"
OR TS="Eftop EF 101" OR TS="Heptadecafluoro-l-octanesulfonic acid" OR
TS="Heptadecafluorooctane-l-sulphonic acid" OR TS="Perfluorooctane
sulfonate" OR TS="perfluorooctane sulfonate" OR TS="Perfluorooctane
sulfonic acid" OR TS="Perfluorooctanesulfonic acid" OR
TS="Perfluorooctylsulfonic acid" OR TS="perfluorooctane sulphonate" OR
TS="perfluorooctane sulfonate" OR TS="l-Octanesulfonic acid,
heptadecafluoro-"OR TS="Heptadecafluorooctanesulfonic acid" OR
TS="Perfluoro-n-octanesulfonic acid" OR TS="Perfluorooctane Sulphonic
Acid" OR TS="Perfluorooctanesulfonate" OR TS="Perfluorooctylsulfonate"
OR ((TS="PFOA" OR TS="PFOS") AND (TS="fluorocarbon*" OR
TS="fluorotelomer*" OR TS="polyfluoro*" OR TS="perfluoro-*" OR
TS="perfluoroa*" OR TS="perfluorob*" OR TS="perfluoroc*" OR
TS="perfluorod*" OR TS="perfluoroe*" OR TS="perfluoroh*" OR
TS="perfluoron*" OR TS="perfluoroo*" OR TS="perfluorop*" OR
TS="perfluoros*" OR TS= "perfluorou*" OR TS="perfluorinated" OR
TS="fluorinated" OR TS="PFAS"))) AND PY=(2020-2022)
TOXLINE TOXLINE taken down, cannot search. -
TSCATS Incorporated into PubMed post 2019. -
The database searches were conducted by EPA and/or contractor information specialists and
librarians on April 11, 2019, September 3, 2020, February 2 and 3, 2022, and February 6, 2023
and all search results were stored in the Health and Environmental Research Online (HERO)
database (https://hero.epa.eov/hero/index.cfm/proiect/paee/proiect id/2608). After deduplication
(i.e., removal of duplicate results) in HERO, the database search results were imported into
SWIFT Review software for filtering/prioritization. SWIFT Review identifies those references
most likely to be applicable to human health risk assessment (https://www.sciome.com/swift-
review/; see also (Howard et al., 2016)). In brief, SWIFT Review has preset literature search
strategies ("filters") developed and applied by information specialists to identify and prioritize
studies that are most likely to be useful for identifying human health content from those that
likely are not (e.g., studies on analytical methods). The filters function like a typical search
strategy in which studies are tagged as belonging to a certain category if the terms in the filter
literature search strategy appear in title, abstract, keyword, and/or medical subject headings
(MeSH) fields content. The applied SWIFT Review filters focused on the following evidence
types: human (epidemiology), animal models for human health, and in vitro studies. The details
of the search strategies that underlie the filters are available online
(https://hawcprd.epa.eov/media/attachment/SWIFT-Review Search Strateeies.pdf). The use of
SWIFT Review is consistent with the IRIS Handbook (U.S. EPA, 2022b) and the Systematic
Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA (anionic and acidforms)
IRIS Assessments (U.S. EPA, 2020b)
A-14
-------
APRIL 2024
For all literature searches, the evidence stream filters used were human, animal (all), animal
(human health model), [no tag], epidemiological quantitative analysis, and in vitro (with one
exception - for the 2022 and 2023 literature searches, the in vitro evidence stream filter was not
used because the goal of those literature search was to identify studies relevant to dose-response
only). Studies not captured using these filters were not considered further. Studies that were
captured with these SWIFT Review evidence stream filters were exported as a RIS (Research
Information System) file for title and abstract screening using either DistillerSR or SWIFT
ActiveScreener software (described in subsequent sections of this appendix).
A.1.5.3 Additional Sources
The literature search strategies used were designed to be broad; however, like any search
strategy, studies may be missed (e.g., if the chemical of interest is not mentioned in title, abstract,
or keyword content; or if gray literature is not indexed in the databases that were searched).
Thus, additional sources were reviewed to identify studies that could have been missed in the
database searches. Reviews of additional sources included the following:
1. Review of studies cited in assessments published by other U.S. federal agencies, as well
as international, and U.S. state-level agencies (including Agency for Toxic Substances
and Disease Registry (ATSDR) and California Environmental Protection Agency
(CalEPA) assessments that were ongoing at the time of searching).
• Manual review of the reference list from ATSDR's Toxicological Profile for
Perfluoroalkyls (ATSDR, 2021) (not date limited).
• Manual review of the reference list from CalEPA's First Public Review Draft of
Proposed Public Health Goals for Perfluorooctanoic Acid and Perfluorooctane
Sulfonic Acid in Drinking Water (CalEPA, 2021) (not date limited).
• Manual review of National Toxicology Program (NTP) publications
(https://ntp.niehs.nih.eov/data/index.html). In 2020, the NTP website was searched
for PFOS toxicity study final reports that could provide relevant health effects
information.
• Manual review of PFAS toxicity studies identified by the New Jersey Department of
Environmental Protection (NJDEP).
2. Review of studies identified during mechanistic or toxicokinetic evidence synthesis:
• Manual review of the reference lists of studies screened as PECO-relevant after full-
text review were reviewed at the title level for potential relevance (backward citation
search).
• Manual review of other EPA PFAS assessments or literature searches under
development by IRIS.
3. Review of studies identified by the SAB PFAS Panel peer reviewers in their final report
(published in August 2022).
A-15
-------
APRIL 2024
4. Review of studies submitted through public comment by May 2023
(https://www.regulations.gov/docket/EPA-HQ-OW-2022-0114).
A.1.5.4 Incorporation of Data from the 2016 PFOS Health Effects
Support Document
The 2016 PFOS HESD contained a comprehensive summary of relevant literature based on
searches conducted through 2013. The 2016 PFOS HESD underwent a public comment period in
February 2014, and an independent external public panel peer review in August 2014. EPA
incorporated key studies from the 2016 PFOS HESD that addressed one or more of the five
priority health outcomes into this updated PFOS assessment, as described below.
Over 140 epidemiological studies were captured in the 2016 PFOS HESD. The 2016 PFOS
HESD did not use the epidemiological data quantitatively. For this updated assessment, EPA
reviewed the epidemiological studies that were included in the 2016 PFOS HESD summary
tables and identified those that were relevant to one or more of the five priority health outcomes
(i.e., developmental, immune, hepatic, cardiovascular, and cancer). A total of 47 epidemiological
studies were included and are listed in Table A-6 (studies relevant to more than one health
outcome are listed under each applicable category in the table).
Table A-6. Key Epidemiological Studies of Priority Health Outcomes Identified From 2016
PFOS Health Effects Support Document
HERO ID
Reference
Title
Cancer
4727072
Alexander and Olsen
(2007)
Bladder cancer in perfluorooctanesulfonyl fluoride manufacturing workers
1291101
Alexander et al.
(2003)
Mortality of employees of a perfluorooctanesulphonyl fluoride
manufacturing facility
2150988
Boncfcld-Jorgcnscn
Perfluorinated compounds are related to breast cancer risk in Greenlandic
et al. (2011)
Inuit: a case-control study
2851186
Boncfcld-Jorgcnscn
et al. (2014)
Breast cancer risk after exposure to perfluorinated compounds in Danish
women: a case-control study nested in the Danish National Birth Cohort
2919344
Eriksen et al. (2009)
Perfluorooctanoate and perfluorooctanesulfonate plasma levels and risk of
cancer in the general Danish population
4930271
Grice et al. (2007)
Self-reported medical conditions in perfluorooctanesulfonyl fluoride
manufacturing workers
2968084
Hardell et al. (2014)
Case-control study on perfluorinated alkyl acids (PFAAs) and the risk of
prostate cancer
Cardiovascular
1291101
Alexander et al.
(2003)
Mortality of employees of a perfluorooctanesulphonyl fluoride
manufacturing facility
2919285
Chateau-Degat et al.
(2010)
Effects of perfluorooctanesulfonate exposure on plasma lipid levels in the
Inuit population of Nunavik (Northern Quebec)
2919150
Eriksen etal. (2013)
Association between plasma PFOA and PFOS levels and total cholesterol
in a middle-aged Danish population
2850962
Fitz-Simon et al.
(2013)
Reductions in serum lipids with a 4-year decline in serum perfluorooctanoic
acid and perfluorooctanesulfonic acid
1430763
Frisbee et al. (2010)
Perfluorooctanoic acid, perfluorooctanesulfonate, and serum lipids in
children and adolescents: results from the C8 Health Project
3749193
Fu et al. (2014)
Associations between serum concentrations of perfluoroalkyl acids and
serum lipid levels in a Chinese population
A-16
-------
APRIL 2024
HERO ID
Reference
Title
2850925
Geigeretal. (2014a)
The association between PFOA, PFOS and serum lipid levels in adolescents
2851286
Geigeretal. (2014b)
No association between perfluoroalkyl chemicals and hypertension in
children
1290820
Lin et al. (2009)
Association among serum perfluoroalkyl chemicals, glucose homeostasis,
and metabolic syndrome in adolescents and adults
1291110
Nelson et al. (2010)
Exposure to polyfluoroalkyl chemicals and cholesterol, body weight, and
insulin resistance in the general US population
1290020
Olsen et al. (2003a)
Epidemiologic assessment of worker serum perfluorooctanesulfonate
(PFOS) and perfluorooctanoate (PFOA) concentrations and medical
surveillance examinations
10228462
Olsenetal. (2001)
A longitudinal analysis of serum perfluorooctane sulfonate (PFOS) and
perfluorooctanoate (PFOA) levels in relation to lipid and hepatic clinical
chemistry test results from male employee participants of the 1994/95, 1997
and 2000 fluorochemical medical surveillance program. Final report.
2850928
Starling et al.
(2014b)
Perfluoroalkyl substances and lipid concentrations in plasma during
pregnancy among women in the Norwegian Mother and Child Cohort Study
1290816
Stein et al. (2009)
Serum levels of perfluorooctanoic acid and perfluorooctane sulfonate and
pregnancy outcome
2850370
Timmermann et al.
(2014)
Adiposity and glycemic control in children exposed to perfluorinated
compounds
Developmental
1429893
Andersen et al.
(2010)
Prenatal exposures to perfluorinated chemicals and anthropometric
measures in infancy
1290833
Apelberg et al.
(2007b)
Cord serum concentrations of perfluorooctane sulfonate (PFOS) and
perfluorooctanoate (PFOA) in relation to weight and size at birth
1290900
Apelberg et al.
(2007a)
Determinants of fetal exposure to polyfluoroalkyl compounds in Baltimore,
Maryland
1332466
Chen et al. (2012a)
Perfluorinated compounds in umbilical cord blood and adverse birth
outcomes
2850274
Darrow et al. (2014)
PFOA and PFOS serum levels and miscarriage risk
2850966
Darrow et al. (2013)
Serum perfluorooctanoic acid and perfluorooctane sulfonate concentrations
in relation to birth outcomes in the Mid-Ohio Valley, 2005-2010
1005775
Fei et al. (2007)
Perfluorinated chemicals and fetal growth: A study within the Danish
National Birth Cohort
1290822
Fei et al. (2008a)
Prenatal exposure to perfluorooctanoate (PFOA) and
perfluorooctanesulfonate (PFOS) and maternally reported developmental
milestones in infancy
2349574
Fei et al. (2008b)
Fetal growth indicators and perfluorinated chemicals: a study in the Danish
National Birth Cohort
1290814
Hamm et al. (2010)
Maternal exposure to perfluorinated acids and fetal growth
1332465
Maisonet et al.
(2012)
Maternal concentrations of polyfluoroalkyl compounds during pregnancy
and fetal and postnatal growth in British girls
2349575
Monroy et al. (2008)
Serum levels of perfluoroalkyl compounds in human maternal and
umbilical cord blood samples
1290816
Stein et al. (2009)
Serum levels of perfluorooctanoic acid and perfluorooctane sulfonate and
pregnancy outcome
1291133
Washino et al. (2009)
Correlations between prenatal exposure to perfluorinated chemicals and
reduced fetal growth
Hepatic
1291101
Alexander et al.
(2003)
Mortality of employees of a perfluorooctanesulphonyl fluoride
manufacturing facility
A-17
-------
APRIL 2024
HERO ID
Reference
Title
1276142
Galloetal. (2012)
Serum perfluorooctanoate (PFOA) and perfluorooctane sulfonate (PFOS)
concentrations and liver function biomarkers in a population with elevated
PFOA exposure
4930271
Grice et al. (2007)
Self-reported medical conditions in perfluorooctanesulfonyl fluoride
manufacturing workers
1291111
Lin et al. (2010)
Investigation of the Associations Between Low-Dose Serum Perfluorinated
Chemicals and Liver Enzymes in US Adults
1290020
Olsen et al. (2003a)
Epidemiologic assessment of worker serum perfluorooctanesulfonate
(PFOS) and perfluorooctanoate (PFOA) concentrations and medical
surveillance examinations
10228462
Olsenetal. (2001)
A longitudinal analysis of serum perfluorooctane sulfonate (PFOS) and
perfluorooctanoate (PFOA) levels in relation to lipid and hepatic clinical
chemistry test results from male employee participants of the 1994/95, 1997
and 2000 fluorochemical medical surveillance program. Final report.
Immune
1937230
Dong et al. (2013)
Serum polyfluoroalkyl concentrations, asthma outcomes, and
immunological markers in a case-control study of Taiwanese children
1290805
Fei et al. (2010)
Prenatal exposure to PFOA and PFOS and risk of hospitalization for
infectious diseases in early childhood
1248827
Grandjean et al.
(2012)
Serum vaccine antibody concentrations in children exposed to
perfluorinated compounds
1937228
Granum et al. (2013)
Pre-natal exposure to perfluoroalkyl substances may be associated with
altered vaccine antibody levels and immune-related health outcomes in
early childhood
2851240
Humblet et al. (2014)
Perfluoroalkyl chemicals and asthma among children 12-19 yr of age:
NHANES (1999-2008)
2850913
Looker etal. (2014)
Influenza vaccine response in adults exposed to perfluorooctanoate and
perfluorooctanesulfonate
1424977
Wang et al. (2011)
The effect of prenatal perfluorinated chemicals exposures on pediatric
atopy
Serum Lipids
2919285
Chateau-Degat et al.
(2010)
Effects of perfluorooctanesulfonate exposure on plasma lipid levels in the
Inuit population of Nunavik (Northern Quebec)
2919150
Eriksenetal. (2013)
Association between plasma PFOA and PFOS levels and total cholesterol
in a middle-aged Danish population
2919156
Fisher etal. (2013)
Do perfluoroalkyl substances affect metabolic function and plasma lipids? -
Analysis of the 2007-2009, Canadian Health Measures Survey (CHMS)
Cycle 1
2850962
Fitz-Simon et al.
(2013)
Reductions in serum lipids with a 4-year decline in serum perfluorooctanoic
acid and perfluorooctanesulfonic acid
1430763
Frisbee et al. (2010)
Perfluorooctanoic acid, perfluorooctanesulfonate, and serum lipids in
children and adolescents: results from the C8 Health Project
3749193
Fu et al. (2014)
Associations between serum concentrations of perfluoroalkyl acids and
serum lipid levels in a Chinese population
2850925
Geigeretal. (2014a)
The association between PFOA, PFOS and serum lipid levels in adolescents
1290820
Lin et al. (2009)
Association among serum perfluoroalkyl chemicals, glucose homeostasis,
and metabolic syndrome in adolescents and adults
3981585
Maisonet et al.
(2015b)
Prenatal exposures to perfluoroalkyl acids and serum lipids at ages 7 and 15
in females
1291110
Nelson et al. (2010)
Exposure to Polyfluoroalkyl Chemicals and Cholesterol, Body Weight, and
Insulin Resistance in the General US Population
A-18
-------
APRIL 2024
HERO ID
Reference
Title
1290020
Olsen et al. (2003a)
Epidemiologic assessment of worker serum perfluorooctanesulfonate
(PFOS) and perfluorooctanoate (PFOA) concentrations and medical
surveillance examinations
10228462
Olsenetal. (2001)
A longitudinal analysis of serum perfluorooctane sulfonate (PFOS) and
perfluorooctanoate (PFOA) levels in relation to lipid and hepatic clinical
chemistry test results from male employee participants of the 1994/95, 1997
and 2000 fluorochemical medical surveillance program. Final report.
2850928
Starling et al.
(2014b)
Perfluoroalkyl substances and lipid concentrations in plasma during
pregnancy among women in the Norwegian Mother and Child Cohort Study
1291109
Steenland et al.
(2009)
Association of perfluorooctanoic acid and perfluorooctane sulfonate with
serum lipids among adults living near a chemical plant
2850370
Timmermann et al.
(2014)
Adiposity and glycemic control in children exposed to perfluorinated
compounds
Notes: NHANES = National Health and Examination Survey.
EPA also reviewed the animal toxicological studies in the 2016 PFOS HESD summary tables
that were identified as relevant for the five priority health outcomes. A total of nine "key" animal
toxicological studies that were either considered quantitatively in the 2016 PFOS HESD or
provided data that may quantitatively impact the assessment conclusions were included and
listed in Table A-7 (studies relevant to more than one health outcome are listed under each
applicable category in the table).
Table A-7. Key Toxicological Animal Toxicological Studies Identified From 2016 PFOS
Health Effects Support Document
HERO ID
Reference
Title
Cardiovascular
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Endocrine
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
757854
Lau et al. (2003)
Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II:
Postnatal evaluation
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Developmental
757873
Butenhoff et al. (2009)
Gestational and lactational exposure to potassium perfluorooctanesulfonate
(K+PFOS) in rats: Developmental neurotoxicity
757854
Lau et al. (2003)
Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II:
Postnatal evaluation
A-19
-------
APRIL 2024
HERO ID
Reference
Title
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1276160
Luebker et al. (2005a)
Two-generation reproduction and cross-foster studies of
perfluorooctanesulfonate (PFOS) in rats
Hematologic
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Hepatic
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
2919266
Kawamoto et al.
(2011)
Ultrasonic-induced tonic convulsion in rats after subchronic exposure to
perfluorooctane sulfonate (PFOS)
757854
Lau et al. (2003)
Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II:
Postnatal evaluation
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Immune
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Metabolic
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
Nervous
757873
Butenhoff et al. (2009)
Gestational and lactational exposure to potassium perfluorooctanesulfonate
(K+PFOS) in rats: Developmental neurotoxicity
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
1276160
Luebker et al. (2005a)
Two-generation reproduction and cross-foster studies of
perfluorooctanesulfonate (PFOS) in rats
Renal
1276144
Butenhoff et al. (2012)
Chronic dietary toxicity and carcinogenicity study with potassium
perfluorooctanesulfonate in Sprague-Dawley rats
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
A-20
-------
APRIL 2024
HERO ID
Reference
Title
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Reproductive
757873
Butenhoff et al. (2009)
Gestational and lactational exposure to potassium perfluorooctanesulfonate
(K+PFOS) in rats: Developmental neurotoxicity
757854
Lau et al. (2003)
Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II:
Postnatal evaluation
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1276160
Luebker et al. (2005a)
Two-generation reproduction and cross-foster studies of
perfluorooctanesulfonate (PFOS) in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
Systemic
757873
Butenhoff et al. (2009)
Gestational and lactational exposure to potassium perfluorooctanesulfonate
(K+PFOS) in rats: Developmental neurotoxicity
757871
Curran et al. (2008)
Altered fatty acid homeostasis and related toxicologic sequelae in rats
exposed to dietary potassium perfluorooctanesulfonate (PFOS)
757857
Luebker et al. (2005b)
Neonatal mortality from in utero exposure to perfluorooctanesulfonate
(PFOS) in Sprague-Dawley rats: dose-response, and biochemical and
pharamacokinetic parameters
1276160
Luebker et al. (2005a)
Two-generation reproduction and cross-foster studies of
perfluorooctanesulfonate (PFOS) in rats
1290852
Seacat et al. (2003)
Subchronic dietary toxicity of potassium perfluorooctanesulfonate in rats
757853
Seacat et al. (2002)
Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in
cynomolgus monkeys
A.1.6 Literature Screening Process to Target Dose-Response
Studies and PK Models
This section summarizes the methods used to screen the literature search results against the
PECO criteria to identify relevant studies potentially suitable for use in dose-response analyses
and studies featuring PK models. Literature search results were screened at both title/abstract and
full-text levels. These screening steps are described further below.
The PECO criteria used to screen the literature search results are the same as those used to frame
the initial literature search (Table A-l) and are outlined again in Table A-8 below.
Table A-8. Populations, Exposures, Comparators, and Outcomes (PECO) Criteria for a
Systematic Review on the Health Effects From Exposure to PFOA and PFOS
PECO
Element
Inclusion Criteria
Population Human: Any population and lifestage (occupational or general population, including children and
other sensitive populations).
Animal: Nonhuman mammalian animal species (whole organism) of any lifestage (including
preconception, in utero, lactation, peripubertal, and adult stages).
A-21
-------
APRIL 2024
PECO
Element
Inclusion Criteria
Exposure
In vitro/cell studies or in silico/modeling toxicity studies should be tagged as supplemental.
Any exposure to PFOA, PFOS, and/or the salts of PFOA/PFOS, including but not limited to:
PFOA {Chemical Abstracts Service (CAS) number 335-67-1).
Other names: perfluorooctanoate; perfluorooctanoic acid; 2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-
pentadecafluorooctanoic acid; pentadecafluoro-l-octanoic acid; pentadecafluoro-n-octanoic acid;
perfluorocaprylic acid; pentadecafluorooctanoic acid; perfluoroheptanecarboxylic acid; octanoic-
acid, pentadecafluoro-
Relevant Salts of PFOA: ammonium perfluorooctanoate (APFO), sodium perfluorooctanoate,
potassium perfluorooctanoate
PFOS {CAS number 1763-23-1).
Other names: perfluorooctane sulfonate, perfluorooctanesulfonic acid, perfluorooctane sulfonic
acid, perfluorooctane sulphonate, perfluorooctanyl sulfonate, heptadecafluorooctane-l-sulphonic,
Heptadecafluoro-l-octanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,8-heptadecafluoro-l-
octanesulfonic acid
Relevant salts of PFOS: lithium perfluorooctanesulfonate, potassium perfluorooctanesulfonate
(K+PFOS), ammonium perfluorooctanesulfonate, sodium perfluorooctanesulfonate
Human: Any exposure to PFOA or PFOS via oral routes. Other exposure routes, including
inhalation, dermal, or unknown/multiple routes will be tracked during title and abstract screening
and tagged as "potentially relevant supplemental information."
Animal: Any exposure to PFOA or PFOS via oral routes. Other exposure routes, including
inhalation, dermal, injection or unknown/multiple routes, will be tracked during title and abstract
screening and tagged as "potentially relevant supplemental information." Studies involving
exposures to mixtures will be included only if they include exposure to PFOA or PFOS alone.
Studies with less than 28 d of dosing, with the exception of reproductive, developmental, immune,
and neurological health outcome studies, should be tagged as supplemental.
Human: A comparison or referent population exposed to lower levels (or no exposure/exposure
below detection limits) of PFOA or PFOS, or exposure to PFOA or PFOS for shorter periods of
time. Case reports and case series will be tracked as "potentially relevant supplemental
information."
Animal: A concurrent control group exposed to vehicle-only treatment or untreated control.
All health outcomes (both cancer and noncancer).
Comparator
Outcome
PBPK Models Studies describing PBPK models will be included.
Notes: PBPK = physiologically based pharmacokinetic.
Following SWIFT Review filtering (see Section A. 1.5.2), literature search results were imported
into either DistillerSR (Evidence Partners;
https://www.evidencepartners.com/prodiicts/distillersr-systematic-review-software) or SWIFT
ActiveScreener (Sciome; https://www.sciome.com/swift-activescreener/) software and were
screened against the PECO criteria at the title and abstract level to identify PECO-relevant
studies that could influence the derivation of an oral RfD and/or CSF. Studies that met the PECO
criteria were tagged as having relevant human data, relevant animal data (in a mammalian
model), or a PBPK model. Studies that did not meet the PECO criteria as determined by
title/abstract screening but did appear to include potentially important supplemental information
were categorized according to the type of supplemental information they provided (described
below). Following completion of title/abstract screening (described further in Sections A. 1.6.3
and A. 1.6.4), the literature search results were re-screened at the full-text level (described further
in Section A. 1.6.5).
A-22
-------
APRIL 2024
The title/abstract and full-text level screenings were performed by two independent reviewers
using structured forms in DistillerSR, with a process for conflict resolution that included
discussion of conflicts with the screening team. During full-text screening, literature inventories
identifying evidence types and health effect systems were created for PECO-relevant studies and
studies tagged as containing potentially relevant supplemental material to facilitate review of
studies by topic-specific experts. These procedures are consistent with those outlined in the IRIS
Handbook (U.S. EPA, 2022c).
Studies that did not meet the PECO criteria but contained potentially relevant supplemental
information were inventoried during the literature screening process. Potentially relevant
supplemental materials included the following (see Table A-l 1 for full list):
• Mechanistic data (including in vitro/ex vivo/in silico studies),
• Studies in non-mammalian or transgenic mammalian model systems,
• Non-oral routes of administration (for animal toxicological studies),
• ADME and toxicokinetic studies (including the application of existing PBPK models),
• Exposure assessment or characterization studies (no health outcome assessment),
• Mixture studies (animal toxicological studies on mixtures of PFOS and other substances
or epidemiological studies that only report associations based on sum or total PFAS),
• Human case reports (n = 1-3 cases per report),
• Records or other assessments with no original data (e.g., reviews, editorials,
commentaries),
• Conference abstracts, and
• Non-English language studies.
Following title/abstract and full-text level screening, studies tagged as containing potentially
relevant mechanistic, ADME, or toxicokinetic data underwent additional screening and data
extraction steps that were separate from steps followed for PECO-relevant studies. Additionally,
studies that were tagged as containing relevant PBPK models were sent to the modeling technical
experts for scientific and technical review. Details on the screening and data extraction methods
for ADME and mechanistic studies are described below.
A.1.6.1 Screening ADME Studies
Studies identified as containing potentially relevant supplemental ADME data during
title/abstract and/or full-text screening underwent further screening against the ADME-specific
PECO criteria outlined in Table A-2. For studies that met the ADME-specific PECO criteria (see
Table A-2), key study information was extracted using litstream™ software. Methods for this
ADME screening and extraction of some key study information into litstream is described
further in Section A. 1.6.7.
A. 1.6.2 Screening Mechanistic Studies
Studies identified as containing potentially relevant supplemental mechanistic data during full-
text screening underwent further screening against the mechanistic-specific PECO criteria
outlined in Table A-3. Studies that met the mechanistic-specific PECO criteria were extracted
into litstream™. Methods for this mechanistic information screening and extraction of some key
study information into litstream is described further in Section A. 1.6.8.
A-23
-------
APRIL 2024
A. 1.6.3 Title/Abstract Screening Questions - DistillerSR
Studies identified from the 2016 PFOS HESD and recent systematic literature search and review
efforts (searches through 2020) were imported into DistillerSR software for title/abstract
screening. For each study, the screeners reviewed the title and abstract and responded to a series
of prompts within structured DistillerSR forms to assess PECO relevance and identify evidence
stream(s). Table A-9 below lists the prompts within the DistillerSR forms used for title/abstract
screening and the response options for each prompt.
Table A-9. DistillerSR Form for Title/Abstract Screening
Question/Prompt
Response Options
1 Does the article meet PECO criteria?
[Select one]
• Yesa
• No
• Tag as potentially relevant supplemental material
• Unclear
If "Yes" to Question #1:
2a What type of evidence?
[Select all that apply]
• Human
• Animal (mammalian models)
• PBPK model
If "Tag as potentially relevant supplemental male rial" to Question #1:
2b What kind of supplemental
material?b
[Select all that apply]
• Mechanistic0
• Non-mammalian model
• ADME/toxicokinetic
• Acute/short-term duration exposures
• Non-oral route of administration
• Exposure characteristics (no health outcome)
• Susceptible population (no health outcome)
• Environmental fate or occurrence (including food)
• Mixture study
• Case study or case series
• Other assessments or records with no original data (e.g., reviews,
editorials, commentaries)
• Conference abstract
• Bioaccumulation data in fish
Notes: PBPK = physiologically based pharmacokinetic.
a Errata, corrections, and corrigenda were tagged to the original study and not considered a separate relevant record.
b Refer to list of supplemental tags in Appendix A. 1.6.4.1.
c Refer to list of mechanistic information in Appendix A. 1.6.4.2.
A. 1.6.4 Title/Abstract Screening Questions - SWIFT-Active
Studies identified from the most recent literature search (2020-2022) were imported into
SWIFT-Active Screener software for title/abstract screening. For each study, screeners reviewed
the title and abstract and responded to a set of prompts designed to ascertain PECO relevance
and identify evidence stream(s). Table A-10 below lists the prompts within SWIFT-Active that
were used for title/abstract screening and the response options for each prompt.
A-24
-------
APRIL 2024
Table A-10. SWIFT-Active Form for Title/Abstract Screening
Question/Prompt
Response Options
1
Include this reference?
Select "Yes, include the reference" if unsure.
[Select one]
• Yes, include the reference3
• No, exclude the reference
i r
"Yes" to Question #1:
2a
Identify the Type of Evidence
[Select all that apply]
• Human/Epidemiological
• Animal
• Unsure
If
"No. exclude the reference" to Question #1:
2b
Not Relevant or Supplemental?1"
Select whether the reference is not relevant to
PECO and should be excluded or if the
reference contains supplemental information.
[Select all that apply]
• Exclude/Not Relevant
• Supplemental
Notes:
a Errata, corrections, and corrigenda were tagged to the original study and not considered a separate relevant record.
b Refer to the list of supplemental tags in Appendix A. 1.6.4.1.
A.l.6.4.1 Supplemental Tags
The categories shown in Table A-l 1 were considered supplemental throughout the title/abstract
and full-text screening processes. With the exception of studies tagged as containing mechanistic
or ADME/TK information, which were further considered as described in Section A. 1.6.7 and
Section A. 1.6.8 of this appendix, studies identified as not PECO-relevant but containing
potentially useful supplemental material were not considered for the subsequent steps of the
systematic review process.
Table A-ll. Supplemental Tags for Title/Abstract and Full-Text Screening
Category Evidence
Mechanistic Studies Studies reporting measurements related to a health outcome that inform the biological
or chemical events associated with phenotypic effects, in both mammalian and non-
mammalian model systems, including in vitro, in vivo (by various routes of
exposure), ex vivo, and in silico studies. When possible, mechanistic studies will be
sub-tagged as pertinent to cancer, noncancer, or unclear/unknown.
Non-Mammalian Model Studies in non-mammalian model systems, e.g., fish, birds, C. elegans.
Systems
ADME and Toxicokinetic Studies designed to capture information regarding absorption, distribution,
metabolism, and excretion, including toxicokinetic studies. Such information may be
helpful in updating or revising the parameters used in existing PBPK models.
Acute/Short-Term Duration Animal studies of less than 28 d (unless the study is a developmental/reproductive,
Exposures neurological, or immune study).
Only One Exposure Group Animal studies with only one exposure group, e.g., control and 1 mg/kg/day PFOA/S.
Non-Oral Routes of Exposure Studies not addressing routes of exposure that fall outside the PECO scope, include
inhalation and dermal exposure routes.
A-25
-------
APRIL 2024
Category Evidence
Exposure Characteristics Exposure characteristic studies include data that are unrelated to toxicological
(No Health Outcome) endpoints, but which provide information on exposure sources or measurement
properties of the environmental agent (e.g., demonstrate abiomarker of exposure).
Susceptible Populations Studies that identify potentially susceptible subgroups; for example, studies that focus
(No Health Outcome) on a specific demographic, lifestage, or genotype.
Environmental Fate or Studies that focus on describing where the chemical will end up after it is used and
Occurrence (Including Food) released into the environment.
Mixture Studies Mixture studies that are not considered PECO-relevant because they do not contain
an exposure or treatment group assessing only the chemical of interest.
Case Studies or Case Series Case reports and case series will be tracked as potentially relevant supplemental
information.
Records With No Original Records that do not contain original data, such as other agency assessments,
Data informative scientific literature reviews, editorials, or commentaries.
Other Assessments or Records Secondary studies (e.g., reviews, editorials, commentaries, assessments) that do not
with No Original Data (e.g., provide any primary research/results.
Reviews, Editorials,
Commentaries)
Conference Abstracts Records that do not contain sufficient documentation to support study evaluation and
data extraction.
Bioaccumulation in Fish Retained records relevant to other EPA projects mentioned in the PFAS Action Plan.
Non-English Reports Studies not reported in English.
Notes: PK = pharmacokinetic; PBPK = physiologically based pharmacokinetic; ADME = absorption, distribution, metabolism,
and/or excretion; C. elegans = Caenorhabditis elegans.
A.l.6.4.2 Mechanistic Study Categories and Keywords
The following categories were considered mechanistic throughout the title/abstract and full-text
screening (Table A-12). Studies tagged as containing potentially relevant supplemental
mechanistic information were further considered as described in Section A. 1.6.8 of this
appendix.
Table A-12. Mechanistic Study Categories Considered as Supplemental
Category
Examples of Keywords
Genotoxicity, micronuclei, DNA strand break, sister chromatid exchange, aneuploidy,
genomic instability, gene amplification, epigenomics, DNA methylation, DNA
methyltransferase, histone, DNA repair, base excision repair, nucleotide excision
repair, DNA mismatch repair
Chromosome or DNA
structure, function, repair, or
integrity
Gene expression and
transcription
Individual genes, pathway-related genes, transcriptomics, epigenetics, transcription
factors, microRNAs, noncoding RNAs
Protein synthesis, folding,
function, transport,
localization, or degradation
Proteomics, translation, ribosomes, chaperones, heat shock proteins, ubiquitin,
proteasome, ER stress, UPR, PERK
Anabolic or catabolic pathways for lipids, carbohydrates, amino acids, nucleotides;
energy metabolism; biochemical pathways; metabolomics; lipidomics; enzyme or
coenzyme activity or function
Metabolism
A-26
-------
APRIL 2024
Category
Examples of Keywords
Ligand interactions with membrane, cytoplasmic and nuclear receptors (e.g., AHR,
ER, AR, CAR, RAR, neurotransmitter receptors, insulin receptor, G-protein coupled
receptors), tyrosine kinases, phosphatase, phospholipases, GTPase, second
messengers (calcium, diacylglycerol, ceramide, NO), signaling pathways (NF-RB.
MAPK/ERK, AKT, mTOR, IP3/DAG, cAMP-dependent, Wnt, (3-catenin, TGF(3,
etc.)
Cell signaling or signal
transduction pathway
Membrane integrity, cell scaffolding, cytoskeleton, actin, microtubules, ER, Golgi,
mitochondria, lysosome, endosome, phagosome, nucleus, chemotaxis, atrophy,
hypertrophy
Cell or organelle structure,
motility, integrity
Extracellular matrix or
molecules
ECM proteins (collagens, elastins, fibronectins and laminins), proteoglycans, matrix
metalloproteinases (MMPs)
Cell cycle (Gl, S, G2, M), cyclins, CDKs, p53, p27, Rb, E2F stem cell, progenitor,
apoptosis, Annexin V, TUNEL, necrosis, blebbing, pyknosis, Bax, Bcl-2,
hyperplasia, dysplasia
Cell growth, differentiation,
proliferation, or viability
Activation of intrinsic cell Cytokines, chemokines, caspases, MHC/HLA molecules, pattern recognition
defense molecules or systems receptors (PRRs), NLR, proteasomes, autophagy
Oxidative stress
Hormone function
Biomarkers of cerebral
function
Reactive oxygen species (ROS), oxidative stress, hydroxyl radical, hydrogen
peroxide, reactive nitrogen species, superoxide anion, peroxyl radicals, antioxidant
response, catalase, superoxide dismutase, EROD, glutathione (GSH), GSH
peroxidase, glutathione-S-transferase, 8-OHdG
GnRH, CRF, ADH/vasopressin, FSH, LH, ACTH, GH, TH, TSH, PTH, Cortisol,
epinephrine/norepinephrine, melatonin, oxytocin, estrogen, testosterone, adiponectin,
leptin, insulin, glucagon
Apoptotic neurodegeneration protein markers, cerebral glucose metabolism, brain
glucose levels
Please provide specific details regarding reason for supplemental tag in the notes
section.
Other (provide details)
Notes: DNA = deoxyribonucleic acid; microRNA = micro ribonucleic acid; RNA = ribonucleic acid; ER = estrogen receptor;
UPR = unfolded protein response; PERK = protein kinase R-like endoplasmic reticulum kinase; AHR = aryl hydrocarbon
receptor; CAR = constitutive androstane receptor; RAR = retinoic acid receptor; GTPase = guanosine triphosphate; NO = nitric
oxide; NF-RB = nuclear factor kappa B; mTOR = rapamycin; DAG = diacylglycerol; TGFp = transforming growth factor beta;
ECM = extracellular matrix;; CDK = cyclin-dependent kinase; Bcl-2 = B-cell lymphoma 2; TUNEL = terminal
deoxynucleotidyl transferase-mediated dUTP nick end labeling; MHC/NHLA = major histocompatibility complex/human
leukocyte antigen; NLR = nucleotide-binding oligomerization domain-like receptors; EROD = ethoxyresorufin-O-dealkylase; 8-
OHdG = 8-hydroxy-2'-deoxyguanosine; GnRH = gonadotropin-releasing hormone; CRF = corticotropin-releasing factor;
ADH = Antidiuretic hormone; FSH = follicle stimulating hormone; LH = luteinizing hormone; ACTH = adrenocorticotropic
hormone; GH = growth hormone; TH = thyroid hormone; PTH = parathyroid hormone.
A-27
-------
APRIL 2024
A. 1.6.5 Full-Text Screening Questions
All studies identified as PECO-relevant from title/abstract screening advanced to full-text screening - which was performed in
DistillerSR. Screeners reviewed each full study report and any supplemental study materials to respond to prompts pertaining to
PECO relevance, evidence stream, and health outcome(s), and whether PFOS and/or PFOA was evaluated (some screening efforts for
PFOA and PFOS were performed concurrently). Table A-13 below lists the prompts and response options that were used for full-text
screening.
Table A-13. DistillerSR Form for Full-Text Screening
Question/Prompt
Response Options
1 Source of study if not identified from database
search.
[Select one]
• Source other than HERO database search
2 Does the article meet PECO criteria?
[Select one]
• Yes
• No
• Tag as potentially relevant supplemental material
• Unclear
If "Yes" to Question #1:
3a If meets PECO, what type of evidence?
[Select all that apply]
• Human
• Animal (mammalian models)
• PBPK model
4a If meets PECO, which health outcome(s) apply?3
[Select all that apply]
• General toxicity, including body weight, mortality, and survival
• Cancer
• Cardiovascular, including serum lipids
• Endocrine (hormone)
• Gastrointestinal
• Genotoxicity
• Growth (early life) and developmental
• Hematological, including non-immune/hepatic/renal clinical chemistry measures
• Hepatic, including liver measures and serum markers (e.g., ALT, AST)
• Immune/inflammation
• Musculoskeletal
• Nervous system, including behavior and sensory function
• Nutrition and metabolic
• Ocular
A-28
-------
APRIL 2024
Question/Prompt Response Options
• PBPK or PK model
• Renal, including urinary measures (e.g., protein)
• Reproductive
• Respiratory
• Skin and connective tissue effects
• Dermal
• Unsure
• Other
If meets PECO and endocrine outcome, which endocrine tags apply?
[Select all that apply]
• Adrenal
• Sex hormones (e.g., androgen, estrogen, progesterone)
• Neuroendocrine
• Pituitary
• Steroidogenesis
• Thyroid
If "Unsure" or "Other" is selected for health outcome, write reasoning in the respective
free text-box.
[Free-text]
If"Tag as potentially relevant supplemental male rial" to Question #1:
3b
If supplemental, what type of information?b c
• Mechanistic
[Select all that apply]
• Non-mammalian model
• ADME/toxicokinetic
• Acute/short-term duration exposures'1
• Non-oral route of administration
• Exposure characteristics (no health outcome)
• Susceptible population (no health outcome)
• Environmental fate or occurrence (including food)
• Mixture study
• Case study or case series
• Other assessments or records with no original data (e.g., reviews, editorials, commentaries)
• Conference abstract
• Bioaccumulation data in fish
4b If "Acute," which health outcome(s) apply?
[Select all that apply]
A-29
-------
APRIL 2024
Question/Prompt
Response Options
• General toxicity, including body weight, mortality, and survival
• Cancer
• Cardiovascular, including serum lipids
• Endocrine (hormone)
• Gastrointestinal
• Genotoxicity
• Growth (early life) and developmental
• Hematological, including non-immune/hepatic/renal clinical chemistry measures
• Hepatic, including liver measures and serum markers (e.g., ALT, AST)
• Immune/inflammation
• 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
• Dermal
• Unsure
If "Yes." "Tag as potentially relevant supplemental material."
" or "Unclear" to Question #1:
5 Which PFAS did the study report?
• PFOA
[Select all that apply]
• PFOS
• Other PFAS
Notes: PBPK = physiologically based pharmacokinetic; ALT = alanine transaminase; AST = aspartate aminotransferase; PK = pharmacokinetic; ADME = absorption, distribution,
metabolism, and/or excretion.
a Refer to list of health outcomes and examples in Appendix A. 1.6.5.1.
b Refer to list of supplemental tags in Appendix A. 1.6.4.1.
c Refer to list of mechanistic information in Appendix A. 1.6.4.2.
dReferto definition of acute/short-term duration exposures in Appendix A. 1.6.6.
A-30
-------
APRIL 2024
A. 1.6.5.1 Health Effect Categories and Example Outcomes for Epidemiological
Studies
The following health effects categories were considered throughout the full-text screening and
subsequent steps of the systematic review process for epidemiological studies (Table A-14).
Table A-14. Health Effect Categories Considered for Epidemiological Studies
Health Effect Category
Example Health Outcomes
Notes
Cancer
• Tumors
• Precancerous lesions (e.g., dysplasia)
—
Cardiovascular
• Serum lipids (e.g., cholesterol, LDL,
HDL, triglycerides)
• Blood pressure
• Hypertension
• Atherosclerosis
• Coronary heart disease
• Other cardiovascular disease
Dermal
• Skin sensitivity
-
Developmental
• Birth size (birth weight; birth length;
small for gestational age)
• Preterm birth
• Sex ratio
• Postnatal growth
• Markers of development specific to
other systems are organ/system-specific
(e.g., tests of sensory maturation are
under Nervous System)
• Pubertal development is under
Reproductive.
Endocrine
• Thyroid hormones (e.g., T3, T4, TSH)
• Thyroid weight and histopathology
• Hormonal measures in any tissue or
blood (non-reproductive)
• Reproductive hormones (e.g., estrogen,
progesterone, testosterone) are under
Reproductive.
Gastrointestinal
• Symptoms of the stomach and intestines-
(e.g., diarrhea, nausea, vomiting,
abdominal pain, and cramps)
Hematologic
• Blood count
• Red blood cells
• Blood Hematocrit or hemoglobin
• Corpuscular volume
• Blood Platelets or reticulocytes
• Blood biochemical measures (e.g.,
sodium, calcium, phosphorus)
• White blood cell counts and globulin
are under Immune.
• Serum lipids are under
Cardiovascular.
• Serum liver markers are under Hepatic.
Hepatic
• Liver enzymes (e.g., ALT; AST; ALP)
• Liver disease
• Liver-specific serum biochemistry (e.g..
albumin)
• Serum lipids are under
Cardiovascular.
, • Biochemical markers, such as albumin,
are under Hepatic. Liver tissue
cytokines are under Immune.
• Globulin is under Immune.
• Serum glucose is under Metabolic.
Immune
• Asthma
• Allergy
• Atopic dermatitis/eczema
• Vaccine response
• IgE
• Autoimmune or infectious disease
• Hypersensitivity
• Red blood cells are under
Hematological.
• Non-immune measures of pulmonary
function are under Respiratory.
• Interleukin 6 (IL-6) is considered a
Mechanistic outcome.
A-31
-------
APRIL 2024
Health Effect Category
Example Health Outcomes
Notes
• General immune assays (e.g., white
blood cell counts)
• Immune responses in the respiratory
system
• Stress-related factors in blood (e.g.,
glucocorticoids or other adrenal
markers)
Metabolic/Systemic
• Obesity
• BMI
• Adiposity
• Diabetes (including gestational
diabetes)
• Insulin resistance
• Blood glucose
• Allostatic load
• Metabolic syndrome
• Waist circumference, ponderal index,
BMI SDS, BMI z-scores, are all
included here.
• Gestational weight gain, adult weight
change also included here.
Musculoskeletal/Connective
Tissue
• Bone health
• Osteoporosis
• Bone density
Nervous
• Cognition
• Behavior
• Autism
• Attention (ADHD)
• Depression
• Communication
• Motor
Ocular
• Vison changes
• Eye irritation
—
Reproductive, female
• Reproductive hormones
• Breastfeeding
• Fecundity
• PCOS
• Spontaneous abortion
• Menopause
• Endometriosis
• Pubertal development
• Menstrual cycle characteristics
• Anogenital distance (females)
• If data indicate altered birth parameters
are likely attributable to female fertility,
these data may be discussed under
Female Reproductive.
Reproductive, male
• Reproductive hormones
• Semen parameters
• Sperm DNA damage
• Pubertal development
• Anogenital distance (males)
Respiratory
• Non-immune measures of pulmonary
(lung) function (e.g., FEV1, FVC, lung
capacity)
• Asthma, wheeze, lower/upper
respiratory trat infections are Immune.
Renal
• GFR
• Uric acid
• Creatinine
• Renal function
A-32
-------
APRIL 2024
Health Effect Category
Example Health Outcomes
Notes
• Urinary measures (e.g., protein;
volume; pH; specific gravity)
Other
• Select this category if the outcome does
-
not fit in any of the above categories.
Notes: LDL = low-density lipoprotein; HDL = high-density lipoprotein; T3 = triiodothyronine; T4 = thyroxine; TSH = thyroid
stimulating hormone; ALT = alanine transaminase; AST = aspartate aminotransferase; ALP = alkaline phosphatase;
IgE = immunoglobulin E; BMI = body mass index; ADHD = attention deficit hyperactivity disorder; PCOS = polycystic ovary
syndrome; DNA = deoxyribonucleic acid; FEV1 = forced expiratory volume in one second; FVC = forced vital capacity;
GFR = glomerular filtration rate.
A. 1.6.6 Animal Toxicological Study Design Definitions
The following definitions were used throughout full-text screening and data extraction for animal
toxicological studies:
• Acute/short-term: Exposure duration between 1-28 days.
• Subchronic: Exposure duration between 28-90 days.
• Chronic: Exposure duration greater than 90 days.
• Developmental: Exposure occurs during gestation and dams are sacrificed prior to birth.
These studies are typically focused on the pups and evaluate viability, developmental
milestones, and other growth and developmental effects in pups.
• Reproductive: Exposure begins prior to mating and may continue through birth and, in
some cases, through a second generation. These studies will typically evaluate
reproductive outcomes in the dams (e.g., copulation and fertility indices, numbers of
corpora lutea and implantation sites, pre- and post-implantation loss).
A-33
-------
APRIL 2024
A. 1.6.7 ADMEScreening and Light Data Extraction
All studies identified as containing ADME data during title/abstract or full-text screening were imported into litstream and underwent
additional screening. Studies that met certain criteria (e.g., PECO relevant and evaluated multiple timepoints, tissues, and/or dose
levels) underwent light data extraction. For each study, at least two reviewers (one primary screener/extractor and one quality
assurance (QA) reviewer) reviewed the full study and any supplemental study materials to respond to prompts pertaining to key study
elements (e.g., tested species or population, tissues evaluated, dose levels tested, ADME endpoints measured). Table A-15 below
describes the prompts and response options that were used for ADME screening of epidemiological or animal toxicological studies.
Table A-15. Litstream Form for ADME Screening and Light Data Extraction
Question/Prompt
Response Options
Suggested Considerations
1 General Questions
1.1
Does the article meet PECO
criteria?
[Select one]
• Yes
• No
• Use ADME-specific PECO statement (See Toxicity
Assessment, (U.S. EPA, 2024)) and "Draft EPA IRIS
Handbook: Principles and Procedures for Integrated
Risk Information System (IRIS) Toxicological
Reviews" to inform the answer.
• Examples of exclusions may include abstract-only,
foreign language, secondary data sources, exposure
studies, physical-chemical properties, and species that
aren't relevant.
• If "No" is selected, do not move forward with the light
extraction. Finish filling out Section 1 - General
Questions (if applicable) and add a note in Section 5 -
Notes under "Notes from Initial Extractor to QA/QC
team" briefly explaining why the study does not meet
PECO.
1.2
What PFAS did the study report?
[Select all that apply]
• PFOA
• PFOS
-
1.3
Does this study contain multiple
time points, multiple tissues,
and/or multiple doses?
[Select one]
• Yes
• No
• If "No" is selected, do not move forward with the light
extraction. Finish filling out Section 1 - General
Questions (if applicable) and add a note in Section 5 -
Notes under "Notes from Initial Extractor to QA/QC
team" briefly explaining why the study meets PECO
but does not contain multiple time points, multiple
tissues, and/or multiple doses.
A-34
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
1.4
Does this study contain supporting
epidemiological information?
[Select one]
• Yes
• No
• Supporting epidemiological information includes
studies that compare PFAS levels in women of
different parity status or weeks of breastfeeding as well
as studies that compare PFAS levels across multiple
age groups or multiple time points even if it is not the
same individuals who are being followed over time
(e.g., a cross-sectional study that enrolls people of
various ages and compares PFOS/PFOA levels in a
specific tissue in children vs. older adults).
1.5
Indicate if there is supplemental
data for this study.
[Select all that apply; Free-text]
• MO A/Mechanistic
• Exposure Study
• Use the free text field below to provide a brief
description of the type of MO A/mechanistic (refer to
Appendix A. 1.6.4.2 for examples) and/or exposure
information that is available.
• Examples of exposure information include studies of
PFAS levels in environmental media not directly linked
to human exposure (e.g., soil, sediment, microbes,
water [except drinking water], birds, or fish [except
those typically consumed by humans]).
1.6
Identify the species, system, or
model.
[Select all that apply]
• Human
• Nonhuman primate
• Rat
• Mouse
• Mammalian cells (in vitro studies)
• PBPK/TK models (or in silico studies)
• If a study only contains PBPK/TK models, do not
move forward with the light extraction. Finish filling
out Section 1 - General Questions (if applicable) and
add a note in Section 5 - Notes under "Notes from
Initial Extractor to QA/QC team" briefly describing the
model.
2
Human Studies Sub-Form
If the study docs not contain a human study, skip this section and move on to Section 3
- Animal Studies Sub-Form.
2.1
Population Name
[Free-Text]
• Name a population (e.g., Females - pregnant, PFOS)
• Separate populations should be made for each
chemical, population sex, lifestage where ADME data
was collected, exposure route, etc. combination.
2.2
Select whether the study looks at
absorption, distribution,
metabolism, and/or excretion.
[Select all that apply]
• Absorption
• Distribution
• Metabolism
• Excretion
• Note: PFOA and PFOS are not metabolized so
"metabolism" is an unlikely selection.
2.3
List the specific ADME endpoints
addressed.
-
• List all the ADME endpoints analyzed for this
population.
A-35
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
[Free-text]
2.4
Exposure Category
Use the free text field if additional
information is needed (e.g., it is a
unique exposure, occupational
setting, etc.).
[Select one; Free-text]
• General environmental
• Poisoning
• Occupational
• Developmental
2.5
Identify the Exposure Route
[Select one; Free-text]
• Inhalation
• Oral
• Dermal
• Lactational transfer
• In utero/placental transfer
• Other (e.g., intraperitoneal, intramuscular, intranasal)
• If "other" option is selected, use the free text field to
describe exposure route.
• If the study population is exposed through more than
one route (e.g., oral and dermal), select one route from
the list and use the free text field to describe the other
exposure routes listed in the paper.
• If the study population is offspring that were exposed
"in utero/placental" AND by "lactational transfer,"
select "in utero/placental" and use the free text field to
note that lactational transfer also occurred.
• If exposure route is unknown, select "other" option and
write in "Unknown" in the free text field.
• If the route is unspecified or multiple routes were
suspected based on the exposure vehicle, select "other"
and write in suspected exposure route in the free text
field.
2.6
What is the exposure vehicle?
[Select one]
• Drinking water
• Diet
• Breast milk
• In utero/placental transfer
• Occupational
• Unknown
• Other
• If "other" option is selected, use the free text field to
describe exposure vehicle.
• If the study population is offspring that were exposed
"in utero/placental" AND by "breast milk," select "in
utero/placental" and use the free text field to note that
lactational transfer also occurred via breast milk.
• If "occupational" option is selected, use the free text
field to describe exposure vehicle.
2.7
What is the sex of the population?
[Select one]
• Male
• Female
• Unspecified
• If results are given separately for each sex, separate
sub-forms should be used for each population.
2.8
Number of Subjects
-
• Example: Total number of subjects = 428
A-36
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
Use the free text field to add
additional details on number of
subjects if they are broken up by
groups or quartiles.
[Free-text]
2.9 What is the lifestage when the
ADME data was collected?
Use the free text field to provide
additional lifestage notes.
[Select one; Free-text]
• Prenatal: conception to birth
• Infancy: 0-12 mo
• Childhood: 13 mo to 11 yr
• Adolescence: 12 to 20 yr
• Adult: 21 to 65 yr
• Elderly: >65 yr
• If there is more than one lifestage when ADME data
was collected, add an additional population in another
form.
2.10 Exposure Levels
Use the free text field to enter the
numeric exposure levels (if
known/estimated in an environmental
medium such as air, water, dust,
food, breast milk, etc.).
[Free-text]
< Do not report levels in serum or urine for this question.
2.11 Exposure Units
Use the free text field to report the
exposure units as presented in the
paper.
[Free-text]
• Examples: mg/kg-d; mg/m3; ppm
• Use "Not Reported" if appropriate
2.12 Exposure Duration
Use the free text field to enter the
details of the exposure duration if
known.
[Free-text]
• Use abbreviations (h, d, wk, mon, y).
oExamples: 28 d; 13 wk; 2 y
• Use "Not Reported" if appropriate.
2.13 Time Points Analyzed
Use the free text field to enter the
time points data were analyzed.
[Free-text]
• Use abbreviations (h, d, wk, mon, y).
oExamples: 28 d; 13 wk; 2 y
• Use "Not Reported" if appropriate.
2.14 Measured Tissues
Use the free text field to enter the
tissues measured in the study (e.g.,
plasma, breast milk, cord blood).
[Free-text]
A-37
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
Animal Studies
If the studv docs not contain an animal study, skip this section and move on to Section 4 - Mammalian Cells/In vitro.
3.1 Population Name
[Free-text]
• Name a population (e.g., Females dams, PFOS).
• Separate populations should be made for each
chemical, species, population sex, lifestage where
ADME data was collected, exposure route, etc.
combination.
3.2 Select whether the study looks at
• Absorption
• PFOA and PFOS are not metabolized, so "metabolism"
absorption, distribution,
• Distribution
is an unlikely selection.
metabolism, and/or excretion.
• Metabolism
[Select all that apply]
• Excretion
3 .3 List the specific ADME Endpoints
addressed.
Use the free text field below to list all
the ADME endpoints analyzed for
this population.
[Free-text]
3.4 Identify the Exposure Route
[Select one]
< Inhalation (nose only)
> Inhalation (whole head exposure)
> Inhalation (whole body exposure)
> Oral (diet)
> Oral (drinking water)
> Oral (gavage)
1 Dermal
> Lactational transfer
> In utero/placental transfer
> Other (e.g., intraperitoneal, intramuscular, intravenous,
intranasal)
• If "other" option is selected, use the free text field
below to describe exposure route
• If the study population is offspring that were exposed
"in utero/placental" AND by "lactational transfer,"
select "in utero/placental" and use the free text field to
note that lactational transfer also occurred
• If there is more than one exposure route identified, add
an additional population in another form.
3.5 What is the exposure vehicle?
[Select one]
• Diet
• Water
• Breast milk
• In utero/placental transfer
• Corn oil
• Filtered air
• Olive oil
• Ethanol
• If "other" option is selected, use the free text field
below to describe exposure vehicle
• If the study population is offspring that were exposed
"in utero/placental" AND by "breast milk," select "in
utero/placental" and use the free text field to note that
lactational transfer also occurred via breast milk.
A-38
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
• DMSO
• Mineral oil
• Corn oil:acetone
• Other
3 .6 What is the strain?
-
• If there is more than one species studied, add an
Use the free text field to list the strain
(e.g., Sprague-Dawley).
[Free-text]
additional population in another form.
3 .7 What is the sex?
• Male
• If results are given separately for each sex, add an
[Select one]
• Female
• Male and Female
additional population in another form.
3.8 What is the lifestage when the
• Prenatal
• Prenatal
animal was dosed?
[Select all that apply]
• Weaning
• Adolescent
o Nonhuman primates: conception to birth
o Rodents: GDO to birth
• Adult
• Elderly
• Weaning
o Nonhuman primates: 1-130 d (0.35 yr)
o Rodents: PND 1-21
• Adolescent
o Nonhuman primates: 130-1,825 d (0.35-5 yr)
o Rodents: 21-50 d (3-7 wk)
• Adult
o Nonhuman primates: 5-35 yr
o Rodents: >50 d (>7 wk)
• Elderly
o Nonhuman primates: >35 yr
3 .9 What is the reported average age
of the animals when dosing began?
[Free-text]
• Use "Not Reported" if appropriate.
3.10 What is the average initial body
weight of the animals when dosing
began?
[Free-text]
• Use "Not Reported" if appropriate.
3 .11 What is the lifestage when the
• Prenatal
• Prenatal
ADME data was collected?
[Select all that apply; Free-text]
• Weaning
• Adolescent
o Nonhuman primates: conception to birth
o Rodents: GD 0 to birth
• Adult
• Weaning
A-39
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
• Elderly
oNonhuman primates: 1-130 d (0.35 yr)
o Rodents: PND 1-21
• Adolescent
oNonhuman primates: 130-1,825 d (0.35-5 yr)
o Rodents: 21-50 d (3-7 wk)
• Adult
oNonhuman primates: 5-35 yr
o Rodents: >50 d (>7 wk)
• Elderly
o Nonhuman primates: >35 yr; use the free text field
to provide additional lifestage notes.
• If there is more than one lifestage when ADME data
were collected, add an additional population in another
form.
3 .12 What is the number of animals per
dosing group?
Use the free text field to report the
number of animals per dosing group.
[Free-text]
• Example: Control = 10, low dose = 20, high dose = 20;
All groups = 20
• Use "Not Reported" if appropriate.
3 .13 Dose Levels
Use the free text field to enter the
numeric dose levels.
[Free-text]
• Example: 0, 450, 900
3 .14 Dose Units
Use the free text field to report the
dosage units as presented in the
paper.
[Free-text]
• Examples: mg/kg-d; mg/m3; ppm
• Use "Not Reported" if appropriate.
3 .15 Dose Duration
Use the free text field to enter the
details of the dose duration if known.
[Free-text]
• Use abbreviations (h, d, wk, mo, y).
• For reproductive and developmental studies, where
possible instead include abbreviated age descriptions
such as "GD 1-10" or "GD 2-PND 10"
oExamples: 14 d, 13 w (6 h/d x 5 d/wk); GD 2-
PND 10
• Use "Not Reported" if appropriate.
3 .16 Time Points Analyzed
-
• Use abbreviations (h, d, wk, mo, y)
oExamples: 14 or 28 d; 13 wk; 2 y
A-40
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
Use the free text field to enter the
• Use "Not Reported" if appropriate.
time points data were analyzed.
[Free-text]
3.17 Measured Tissues
Use the free text field to enter the
tissues measured in the study (e.g.,
plasma, liver, adipose).
[Free-text]
4
Mammalian Cells/In vitro
If the study docs not contain an in vitro component, skip this section and move on to Section 5
- Notes.
4.1
Population Name
[Free-text]
• Name a population (e.g., Primary Human Hepatic,
PFOA; A549, PFOS)
• Separate populations should be made for each
chemical, population sex, lifestage where ADME data
was collected, exposure route, etc. combination. Use
the "Clone" button to copy forms/information for easier
extraction if the study populations are similar.
4.2
Select whether the study looks at
absorption, distribution,
metabolism, and/or excretion.
[Select all that apply]
• Absorption
• Distribution
• Metabolism
• Excretion
• PFOA and PFOS are not metabolized so "metabolism"
is an unlikely selection.
4.3
List the specific ADME Endpoints
addressed.
Use the free text field below to list all
the ADME endpoints analyzed for
this population.
[Free-text]
4.4
Does the study present data on
protein binding?
[Select one; Free-text]
• Yes
• No
• If "Yes" option is selected, use the free text field to list
the binding proteins.
4.5
Does the study present data on
active transport?
[Select one; Free-text]
• Yes
• No
• If "Yes" option is selected, use the free text field to list
the transporters.
4.6
Cell Line Name or Tissue Source
Use the free text field to list the cell
line name or tissue source the cells
were derived from.
• Examples: A549; liver tissue from adult Sprague-
Dawley female rats
A-41
-------
APRIL 2024
Question/Prompt Response Options Suggested Considerations
[Free-text]
• If there is more than one cell line name or tissue source
studied, add an additional population in another form.
4.7 In vitro System • Mammalian cells
• If "other" option is selected, use the free text field
[Select one; Free-text] • Cell-free system
below to describe the in vitro system.
• In silico system
• If there is more than one in vitro source studied, add an
• Other
additional population in another form.
4.8 Select all study design elements • Multiple time points
that apply. • Multiple cell/tissue types
[Select all that apply] . Multiple dose levels
4.9 Exposure Design
Use the free text field to describe the
-
exposure design, be as succinct as
possible.
[Free-text]
4.10 What is the exposure vehicle?
Use the free text field to describe the
- -
exposure vehicle, be as succinct as
possible
[Free-text]
4.11 Dose Levels
Use the free text field to enter the
- • Example: 0, 450, 900
numeric dose levels.
[Free-text]
4.12 Dose Units
Use the free text field to report the
dosage units as presented in the
• Examples: ppm; mg/mL
• Use "Not Reported" if appropriate.
paper.
[Free-text]
4.13 Dose Duration
• Use abbreviations (h, d, wk, mon, y)
Use the free text field to enter the
oExamples: 28 d; 13 wk; 2 y
details of the exposure duration.
[Free-text]
• Use "Not Reported" if appropriate.
4.14 Time Points Analyzed
• Use abbreviations (h, d, wk, mon, y)
Use the free text field to enter the
oExamples: 28 d; 13 wk; 2 y
time points data were analyzed.
[Free-text]
• Use "Not Reported" if appropriate.
A-42
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
5 • Notes
5.1 General Study Notes
[Free-text]
Use the free text field to add any
general study notes not captured
above that may be of interest to the
QC reviewer or PBPK modelers
• Please indicate whether the study contains information
on PFOA/PFOS that is broken up by linear/branched
isomers. Use the following phrase: "Contains
linear/branched isomer information"
5.2 Notes from Initial Extractor to - -
QA/QC Team
Use the free text field to add any
general study notes not captured
above that may be of interest to the
QC reviewer.
[Free-text]
5.3 Notes from QA/QC Team
Use the free text field to add any
general study notes not captured
above that may be of interest to the
PBPK modelers.
[Free-text]
Notes: ADME = absorption, distribution, metabolism, and/or excretion; QA/QC = quality assurance/quality control; MOA = mode of action; PBPK = physiologically based
pharmacokinetic; TK = toxicokinetic; GD = gestational day; PND = postnatal day; ppm = parts per million.
A-43
-------
APRIL 2024
A. 1.6.8 Mechanistic Screening and Light Data Extraction
All studies identified as mechanistic in title/abstract or full-text screening were imported into litstream and underwent additional
screening. Studies that were confirmed to be PECO relevant underwent light data extraction. For each study, at least two reviewers
(one primary screener/extractor and one QA reviewer) reviewed the full study and any study materials to respond to prompts
pertaining to key study elements (e.g., tested species or population, mechanistic endpoint(s) evaluated, lifestage(s) at which
evaluations were performed). Table A-16 below describes the prompts and response options that were used for screening studies with
mechanistic evidence.
Table A-16. Litstream Form for Mechanistic Screening and Light Data Extraction
Question
Options
Suggested Considerations
1 General Questions
1.1
Does the article meet PECO
criteria?
[Select one]
• Yes
• No
1.2
What PFAS did the study report?
[Select all that apply]
• PFOA
• PFOS
-
1.3
Publication Type
[Select one]
• Primary research
• Review article
-
1.4
Indicate if there is hazard ID or
supplemental data for this study.
[Select all that apply; Free-text]
• Animal tox
• Epi
• ADME
• Use free text field to provide an explanation.
2
Human Studies Sub-Form
If the study docs not contain a human study, skip this section and move on to Section 3
- Animal Studies Sub-Form.
2.1
Population/Study Group Name
[Free-text]
-
-
2.2
Exposure Category
[Select one; Free-text]
• General environmental
• Poisoning
• Occupational
• Developmental
• Controlled experimental
• Free text field if additional information is needed.
2.3
Identify the Exposure Route
[Select all that apply]
• Inhalation
• Oral
• Dermal
• Lactational transfer
• Free text field to elaborate on "other" and "unknown"
options.
A-44
-------
APRIL 2024
Question
Options
Suggested Considerations
• In utero/placental transfer
• Other (e.g., intraperitoneal, intramuscular, intranasal)
• Unknown
2.4
What is the exposure vehicle?
• Drinking water
• Free text field to elaborate on "other" and "unknown"
[Select one]
• Diet
options.
• Breast milk
• In utero/placental transfer
• Occupational
• Unknown
• Other
2.5
What is the lifestage when the
• Prenatal
• Free text for lifestage notes.
mechanistic data was collected?
• Infancy
[Select one; Free-text]
• Childhood
• Adolescence
• Adult
• Elderly
2.6
What is the corresponding health
• Cancer
• Free field for "other" option, includes endpoints that do
outcome system?
• Cardiovascular
not fit neatly into any one health outcome system.
[Select one]
• Dental
• Dermal
• Developmental
• Endocrine
• Gastrointestinal
• Hematologic
• Hepatic
• Immune
• Lymphatic
• Metabolic
• Musculoskeletal/connective tissue
• Nervous
• Ocular
• Renal
• Reproductive
• Respiratory
• Systemic/whole body
• Other
A-45
-------
APRIL 2024
Question
Options
Suggested Considerations
2.7
Mechanistic Category
[Select all that apply; Free-text]
• Epigenetics
• Chromosome/DNA structure, function, repair or
integrity
• Gene expression and transcription
• Protein expression, synthesis, folding, function,
transport, localization, or degradation
• Metabolomics
• Cell or organelle structure, motility, or integrity
• Structure, Morphology, or Morphometry
• Other
• Free text field for "other" option.
2.8
Mechanistic Pathway
[Select all that apply; Free-text]
• Angiogenic, antiangiogenic, vascular tissue remodeling
• Atherogenesis and clot formation
• Big data, non-targeted analysis
• Cell growth, differentiation, proliferation, or viability
• Cell signaling or signal transduction
• Extracellular matrix or molecules; Fatty acid synthesis,
metabolism, storage, transport, binding, (3-oxidation
• Hormone function
• Inflammation and Immune Response
• Oxidative stress
• Renal dysfunction
• Vasoconstriction/vasodilation
• Xenobiotic metabolism
• Other
• Free text field for "other" option.
2.9
Mechanistic Endpoints
[Free-text]
-
• Free text field to list mechanistic endpoints.
3
• Animal Studies Sub-Form
• If the study docs not contain an animal study, skip this section and move on to Section 4 - In vitro Sub-Form.
3.1
Population/Study Group Name
[Free-text]
-
-
3.2
What is the species?
[Select one; Free-text]
• Nonhuman primate
• Zebrafish
• Rat
• Mouse
• Free text field to list species for "other rodent model"
option.
• Rabbit
• Guinea pig
A-46
-------
APRIL 2024
Question
Options
Suggested Considerations
• Other rodent model
3.3
What is the strain?
[Free-text]
-
-
3.4
Identify the Exposure Route
[Select one]
• Inhalation (nose only)
• Inhalation (whole head exposure)
• Inhalation (whole body exposure)
• Oral (diet)
• Oral (drinking water)
• Oral (gavage)
• Dermal
• Lactational transfer
• In utero/placental transfer
• Other (e.g., intraperitoneal, intramuscular, intravenous,
intranasal)
• Free text field for "other" option.
3.5
What is the exposure vehicle?
[Select one]
• Diet
• Water
• Breast milk
• In utero/placental transfer
• Corn oil
• Filtered air
• Olive oil
• Ethanol
• DMSO
• Mineral oil
• Corn oil: acetone
• Other
• Free text field for other "other" option.
3.6
What is the lifestage when the
animal was dosed?
[Select one; Free-text]
• Prenatal
• Weaning
• Adolescent
• Adult
• Elderly
• Free text field for lifestage notes.
3.7
What is the lifestage when the
mechanistic data was collected?
[Select one; Free-text]
• Prenatal
• Weaning
• Adolescent
• Adult
• Free text field for lifestage notes.
A-47
-------
APRIL 2024
Question
Options Suggested Considerations
• Elderly
3.8 What is the corresponding health
• Cancer • Free text field for "other" option, includes endpoints
outcome system?
• Cardiovascular that do not fit neatly into any one health outcome
[Select all that apply; Free-text]
• Dental system.
• Dermal
• Developmental
• Endocrine
• Gastrointestinal
• Hematologic
• Hepatic
• Immune
• Lymphatic
• Metabolic
• Musculoskeletal/connective tissue
• Nervous
• Ocular
• Renal
• Reproductive
• Respiratory
• Systemic/whole body
• Other
3 .9 Mechanistic Category
• Epigenetics chromosome/DNA structure, function, • Free text field for "other" option.
[Select all that apply; Free-text]
repair, or integrity
• Gene expression and transcription
• Protein expression, synthesis, folding, function,
transport, localization, or degradation
• Metabolomics
• Cell or organelle structure, motility, or integrity
• Structure, Morphology, or Morphometry
• Other
3.10 Mechanistic Pathway
• Angiogenic, antiangiogenic, vascular tissue remodeling • Free text field for "other" option.
[Select all that apply; Free-text]
• Atherogenesis and clot formation
• Big data, non-targeted analysis
• Cell growth, differentiation, proliferation, or viability
• Cell signaling or signal transduction
A-48
-------
APRIL 2024
Question Options Suggested Considerations
• Extracellular matrix or molecules
• Fatty acid synthesis, metabolism, storage, transport,
binding, (3-oxidation
• Hormone function
• Inflammation and Immune Response
• Oxidative stress
• Renal dysfunction
• Vasoconstriction/vasodilation
• Xenobiotic metabolism
• Other
3.11 Mechanistic Endpoints
[Free-text]
• Free text field to list mechanistic endpoints
4 In vitro Sub-Form
If the study docs not contain an in vitro component, skip this section and move on to Section 5 - Notes.
4.1 Population/Study Group Name
[Free-text]
4.2 Does the study present data on
protein binding?
[Select one; Free-text]
• Yes
• No
• Free text field if "Yes" to list binding proteins.
4.3 Does the study present data on
active transport?
[Select one; Free-text]
• Yes
• No
• Free text field if "Yes" to list transporters.
4.4 In vitro System
[Select one; Free-text]
• Mammalian cells
• Cell-free system
• In silico system
• Other
• Free text field for "other" option.
4.5 If a cellular model is used, is it a
cell line or primary cells?
[Select one]
• Cell line
• Primary cell
4.6 Cell Or Tissue Source for In
vitro/Ex Vivo Studies
[Select one; Free-text]
• Human
• Zebrafish
• Nonhuman primate
• Rat
• Mouse
• Rabbit
• Guinea pig
• Free text field to list "other rodent model" option.
A-49
-------
APRIL 2024
Question
Options
Suggested Considerations
• Other rodent model
4.7 What is the corresponding health
• Cancer
• Free text field for "other" option, includes endpoints
outcome system?
• Cardiovascular
that do not fit neatly into any one health outcome
[Select all that apply; Free-text]
• Dental
system.
• Dermal
• Developmental
• Endocrine
• Gastrointestinal
• Hematologic
• Hepatic
• Immune
• Lymphatic
• Metabolic
• Musculoskeletal/connective tissue
• Nervous
• Ocular
• Renal
• Reproductive
• Respiratory
• Systemic/whole body
• Other
4.8 Mechanistic Category
• Epigenetics chromosome/DNA structure, function,
• Free text field for "other" option.
[Select all that apply; Free-text]
repair, or integrity
• Gene expression and transcription
• Protein expression, synthesis, folding, function,
transport, localization, or degradation
• Metabolomics
Cell or organelle structure, motility, or integrity
Structure, morphology, or morphometry
Other
Angiogenic, antiangiogenic, vascular tissue remodeling • Free text field for "other" option.
Atherogenesis and clot formation
Big data, non-targeted analysis
Cell growth, differentiation, proliferation, or viability
Cell signaling or signal transduction
4.9 Mechanistic Pathway
[Select all that apply; Free-text]
A-50
-------
APRIL 2024
Question Options Suggested Considerations
• Extracellular matrix or molecules
• Fatty acid synthesis, metabolism, storage, transport,
binding, (3-oxidation
• Hormone function
• Inflammation and immune response
• Oxidative stress
• Renal dysfunction
• Vasoconstriction/vasodilation
• Xenobiotic metabolism
• Other
4.10 Mechanistic Endpoints - -
[Free-text]
5 • Notes
5.1 General Study Notes
Use the free text field to add any
general study notes not captured
above that may be of interest to the
QC reviewer or PBPK modelers.
[Free-text]
5.2 Notes from Initial Extractor to
QA/QC Team
Use the free text field to add any
general study notes not captured
above that may be of interest to the
QC reviewer.
[Free-text]
5.3 Notes from QA/QC Team
Use the free text field to add any
general study notes not captured
above that may be of interest to the
PBPK modelers.
[Free-text]
Notes: ADME = absorption, distribution, metabolism, and/or excretion; DNA = deoxyribonucleic acid; DMSO = dimethyl sulfoxide, PBPK = physiologically based
pharmacokinetic; QA/QC = quality assurance/quality control.
• Please indicate whether the study contains information
on PFOA/PFOS that is broken up by linear/branched
isomers. Use the following phrase: "Contains
linear/branched isomer information"
A-51
-------
APRIL 2024
A. 1.7 Study Quality Evaluation Overview
After literature search results were screened and inventoried, epidemiological and animal
toxicological studies that met PECO criteria underwent study quality evaluation to assess each
study's validity and utility. As outlined in the IRIS Handbook (U.S. EPA, 2022c), the key
concerns during the review of epidemiological and animal toxicological studies are potential bias
(factors that affect the magni tude or directi on of an effect in either directi on) and insensitivity
(factors that limit the ability of a study to detect a true effect; low sensitivity is a bias toward the
null when an effect exists). Study quality evaluations produce overall judgments about
confidence in the reliability of study results. The general approach for study quality evaluation is
outlined in Figure A-l, which has been adapted from Figure 4-1 in the IRIS Handbook (U.S.
EPA, 2022c) (previously Figure 6-1 in the draft IRIS Handbook (U.S. EPA, 2020a)). Study
quality evaluations were performed using the structured platform for study evaluation housed
within EPA's Health Assessment Workplace Collaborative (HAWC).
(a)
Develop assessment-
specific considerations
Pilot testing (and possible
refinement)
Independent evaluation
by two reviewers
Conflict resolution
(b)
Study Quality Evaluation Domains
Animal Toxicological Studies
Epidemiological Studies
Reporting quality ¦ Allocation ¦ Observational bias'blinding
Confounding variable control ¦ Selective reporting and attrition
Chemical administration and characterization
Exposure timing, frequency, & duration
Endpoint sensitivity and specificity ¦ Results presentation
Exposure measurement ¦ Outcome ascertainment
Participant selection ¦ Potential confounding
Analysis ¦ Sensitivity ¦ Selective reporting
Domain Scores
I Good (Appropriate study conduct relating to die domain; minor deficiencies not expected to influence results)
Adequate (Some limitations relating to die domain, but not likely to be severe or to have a notable impact on results)
Deficient (Identified biases or deficiencies interpreted as likely to have had a notable impact on die result or prevent
reliable interpretation of study findings)
I Critically Deficient (Serious flaws that make observed effects uninterpretable)
*
Finalization of domain
judgements and overall
ratings
Overall Confidence Ratings
High (No notable deficiencies or concerns identified; potential for bias unlikely or minimal: sensitive methodology)
Medium (Possible deficiencies or concerns noted, but resulting bias or lack of sensitivity is unlikely to be of a notable degree)
Low (Deficiencies or concerns were noted, and the potential for substantive bias or inadequate sensitivity could have a
significant impact on the stud)' results or their interpretation^
Uninformafive (Serious flaws make study results unusable for hazard identification or dose-response)
Figure A-l. Overview of Study Quality Evaluation Approach
(a) An overview of the study quality evaluation process; (b) Evaluation domains and ratings definitions (i.e., domain scores and
overall confidence ratings, performed on an outcome-specific basis as applicable).
The overall aims of study quality evaluation are the same for both epidemiological and animal
toxicological studies, but some aspects of the approaches are different. Therefore, study quality
evaluation procedures for epidemiological and animal toxicological studies are described
separately in the following sections. In brief, at least two primary reviewers independently
judged the reliability of the study results according to multiple study quality evaluation domains
presented in the IRIS Handbook. Domain-specific core and prompting questions are provided to
guide the reviewer in assessing different aspects of study design and conduct related to reporting,
risk of bias, and study sensitivity. For each domain, each reviewer assigned a rating of good,
adequate, deficient (or "not reported," which carried the same functional interpretation as
A-52
-------
APRIL 2024
deficient), or critically deficient (see Figure A-l and Table A-17). A QA reviewer (in accordance
with protocols outlined in the IRIS Handbook) engaged in conflict resolution with the two
independent reviewers as needed and made a final determination (reflected as study confidence
ratings; see Figure A-l and Table A-18) regarding each health outcome or outcome grouping of
interest; thus, different "judgments" were possible for different health outcomes within the same
study. The overall confidence rating should, to the extent possible, reflect interpretations of the
potential influence on the results across all domains. The rationale supporting the overall
confidence rating is documented clearly and consistently and includes a brief description of any
important study strengths and/or limitations and their potential impact on the overall confidence.
The specific study limitations identified during study quality evaluation were carried forward to
inform the synthesis of findings within each body of evidence for a given health effect (i.e.,
study confidence determinations were not used to inform "judgments" in isolation).
Studies containing PBPK, mechanistic or ADME data did not undergo study quality evaluation,
as study quality domains for these types of studies are not currently available (U.S. EPA, 2022b).
Table A-17. Possible Domain Scores for Study Quality Evaluation
Good
Intended to represent a judgment that there was appropriate study conduct relating to the
domain (as defined by consideration of the criteria listed below), and any minor deficiencies
that were noted would not be expected to influence interpretation of the study findings.
Adequate
Indicates a judgment that there were study design limitations relating to the domain (as
defined by consideration of the criteria listed below), but that those limitations are not likely
to be severe and are expected to have minimal impact on interpretation of the study findings.
Deficient
Denotes identified biases or limitations that are interpreted as likely to have had a substantial
impact on the results or that prevent reliable interpretation of the study findings.
Note: Not reported indicates that the information necessary to evaluate the domain was not
available in the study. Generally, this term carries the same functional interpretation as
Deficient for the purposes of the study confidence classification.
Critically Deficient
Reflects a judgment that the study design limitations relating to the domain introduced a flaw
so serious that the study should not be used without exceptional justification (e.g., it is the
only study of its kind and may highlight possible research gaps). This judgment should only
be used if there is an interpretation that the limitation(s) would be the primary driver of any
observed effect(s), or if it makes the study findings uninterpretable.
Table A-18. Overall Study Confidence Classifications
High Confidence
No notable concerns were identified (e.g., most or all domains rated Good).
Medium Confidence
Some concerns are identified but expected to have minimal impact on the interpretation of
the results (e.g., most domains rated Adequate or Good; may include studies with
Deficient ratings if concerns are not expected to strongly impact the magnitude or
direction of the results). Any important concerns should be carried forward to evidence
synthesis.
Low Confidence
Identified concerns are expected to significantly impact the study results or their
interpretation (e.g., generally. Deficient ratings for one or more domains). The concerns
leading to this confidence judgment must be carried forward to evidence synthesis.
A-53
-------
APRIL 2024
Serious flaw(s) make the study results unusable for informing hazard identification (e.g.,
generally. Critically Deficient rating in any domain; many Deficient ratings).
Uninformative studies are not considered further in the synthesis and integration of
evidence.
A. 1.7.1 Study Quality Evaluation for Epidemiological Studies
Study quality evaluation domains for assessing risk of bias and sensitivity in epidemiology
studies of health effects are: participant selection, exposure measurement, outcome
ascertainment, potential confounding, analysis, study sensitivity, and selective reporting. As
noted in the IRIS Handbook, this framework is adapted from the Risk Of Bias in Nonrandomized
Studies of Interventions (ROBINS-I) tool (https://methods.cochrane.org/methods-
cochrane/robins-i-tooO, modified by IRIS for use with the types of studies more typically
encountered in EPA's work. As outlined in Section A. 1.7 of this appendix, study quality
evaluations are performed for a set of established domains, and core and prompting questions are
provided for each domain to guide the reviewer. Each domain is assigned a score of Good,
Adequate, Deficient, Not Reported or Critically Deficient, and rationales to support the scores
are developed. Once all domains are evaluated, a confidence rating of High, Medium, or Low
confidence or Uninformative is assigned.
The tables presented in the following sections describe the epidemiological study quality
evaluation domains and the prompting questions and considerations for assessing study quality in
relation to each domain.
A-54
-------
APRIL 2024
A.l.7.1.1 Participant Selection
The aim of study quality evaluation for this domain is to ascertain whether the reported information indicates that selection in or out of
the study (or analysis sample) and participation was not likely to be biased (i.e., the exposure-outcome distribution of the participants
is likely representative of the exposure-outcome distribution in the overall population of eligible persons) (Table A-19).
Table A-19. Study Quality Evaluation Considerations for Participant Selection
Core Question: Is there evidence that selection into or out of the study (or analysis sample) was jointly related to exposure and to outcome?
Prompting Questions
Follow-Up Questions
Suggested Considerations
For longitudinal cohort:
Did participants volunteer for the cohort based on
knowledge of exposure and/or preclinical disease
symptoms? Was entry into the cohort or
continuation in the cohort related to exposure and
outcome?
For occupational cohort:
Did entry into the cohort begin with the start of
the exposure?
Was follow-up or outcome assessment
incomplete, and if so, was follow-up related to
both exposure and outcome status?
Could exposure produce symptoms that would
result in a change in work assignment/work status
("healthy worker survivor effect")?
For case-control study:
Were controls representative of population and
time periods from which cases were drawn?
Were differences in
participant enrollment and
follow-up evaluated to
assess the potential for bias?
If there is a concern about
the potential for bias, what
is the predicted direction or
distortion of the bias on the
effect estimate (if there is
enough information)?
Were appropriate analyses
performed to address
changing exposures over
time in relation to
symptoms?
Is there a comparison of
participants and
• Minimal concern for selection bias based on description of
recruitment process (e.g., selection of comparison
population, population-based random sample selection,
recruitment from sampling frame including current and
previous employees) such that study participants were
unlikely to differ from a larger cohort based on recruitment
or enrollment methods (or data provided to confirm a lack
of difference)
• Exclusion and inclusion criteria specified and would not be
likely to induce bias.
• Participation rate is reported at all steps of study (e.g.,
initial enrollment, follow-up, selection into analysis
sample). If rate is not high, there is appropriate rationale for
why it is unlikely to be related to exposure (e.g.,
comparison between participants and nonparticipants or
other available information indicates differential selection
is not likely).
• Comparison groups are similar with respect to factors
expected to influence exposure-outcome relationship
(confounders. effect measure modifiers).
A-55
-------
APRIL 2024
Core Question: Is there evidence that selection into or out of the study (or analysis sample) was jointly related to exposure and to outcome?
Are hospital controls selected from a group
whose reason for admission is independent of
exposure?
Could recruitment strategies, eligibility criteria,
or participation rates result in differential
participation relating to both disease and
exposure?
For population based-survey:
Was recruitment based on advertisement to
people with knowledge of exposure, outcome,
and hypothesis?
nonparticipants to address
whether differential
selection is likely?
Adequate • Enough of a description of the recruitment process (i.e.,
recruitment strategy, participant selection or case
ascertainment) to be comfortable that there is no serious
risk of bias.
• Inclusion and exclusion criteria specified and would not
induce bias.
• Participation rate is incompletely reported for some steps of
the study, but available information indicates participation
is unlikely to be related to exposure.
• Comparison groups are largely similar with respect to
factors expected to influence exposure-outcome
relationship (confounders, effect measure modifiers) or
these are mostly accounted for in the study analysis.
Deficient • Little information on recruitment process, selection
strategy, sampling framework and/or participation OR
aspects of these processes raises the likelihood of bias (e.g.,
healthy worker effect, survivor bias).
Example: Enrollment of "cases" from a specific clinic
setting (e.g., diagnosed autism), which could be biased by
referral practices and ser\>ices availability, without
consideration of similar selection forces affecting
recruitment of controls.
Critically
Deficient
' Aspects of the processes for recruitment, selection strategy,
sampling framework, or participation result in concern that
the likelihood of selection bias is high (e.g., convenience
sample with no information about recruitment and
selection, cases and controls are recruited from different
sources with different likelihood of exposure, recruitment
materials stated outcome of interest and potential
participants are aware of or are concerned about specific
exposures).
¦ Convenience sample, and recruitment and selection not
described.
' Case report, case series, or other study designs lacking a
comparison group (these should be excluded if they do not
meet assessment PECO criteria).
A-56
-------
APRIL 2024
A.l.7.1.2 Exposure Measurement
This domain may need to be evaluated multiple times for a single study if more than one measurement of exposure is assessed.
Therefore, different sets of criteria may be applied for different exposure assessments in the same study. Table A-20 outlines criteria
that apply across exposure assessments (first row), and specific additional criteria for specific types of exposure assessments (e.g.,
biomarkers, occupational) in subsequent rows.
Table A-20. Study Quality Evaluation Considerations for Exposure Measurement
Core Question: Does the exposure measure reliably distinguish between levels of exposure in a time window considered most relevant for a causal effect
with respect to the development of the outcome?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Does the exposure measure capture the variability
in exposure among the participants, considering
intensity, frequency, and duration of exposure?
Is the degree of exposure
misclassification likely to
vary by exposure level?
Does the exposure measure reflect a relevant time If the correlation between
window? If not, can the relationship between
measures in this time and the relevant time window
be estimated reliably?
Was the exposure measurement likely to be
affected by a knowledge of the outcome?
Was the exposure measurement likely to be
affected by the presence of the outcome (i.e..
reverse causality)?
exposure measurements is of Adequate
concern is there an adequate
statistical approach to
ameliorate variability in
measurements?
If there is a concern about
the potential for bias, what is
the predicted direction or
distortion of the bias on the
effect estimate (if there is
enough information)?
Valid exposure assessment methods used, which represent
the etiologically relevant time period for reported effects
(e.g., exposure during a critical developmental window or
exposure preceding the evaluation of the outcome).
Exposure misclassification is expected to be minimal.
Valid exposure assessment methods used, which represent
the etiologically relevant time period of interest.
Exposure misclassification may exist but is not expected to
greatly impact the effect estimate.
Deficient • Specific knowledge about the exposure and outcome raise
concerns about reverse causality, but there is uncertainty
whether it is influencing the effect estimate.
» Exposed groups are expected to contain a notable
proportion of unexposed or minimally exposed individuals,
the method did not capture important temporal or spatial
variation, or there is other evidence of exposure
misclassification that would be expected to notably change
the effect estimate.
Critically
Deficient
• Exposure measurement does not characterize the
etiologically relevant time period of exposure or is not
valid.
• There is evidence that reverse causality is very likely to
account for the observed association.
• Exposure measurement was not independent of outcome
status.
A-57
-------
APRIL 2024
Core Question: Does the exposure measure reliably distinguish between levels of exposure in a time window considered most relevant for a causal effect
with respect to the development of the outcome?
Additional prompting questions for biomarkers of
exposure:
Is a standard assay used? What are the intra- and
inter-assay coefficients of variation? Is the assay
likely to be affected by contamination? Are values
less than the limit of detection dealt with
adequately?
What exposure time period is reflected by the
biomarker? If the half-life is short, what is the
correlation between serial measurements of
exposure?
Additional suggested considerations for biomarkers of exposure (should be
evaluated in addition to the general considerations above):
Good
z
Adequate
Use of appropriate analytic method such as [specific gold
standard exposure assessment method for the exposure of
interest].
Use of appropriate (but not gold standard) analytic method.
Deficient • Did not identify analytical methods used to measure
exposure.
• Failure to report LOD, percentage less than LOD, and
methods used to account for values below the LOD.
• Failure to report QA/QC measures and results.
Critically
Deficient
Use of inappropriate analytical method or use of an
appropriate method with measurement issues that are likely
to impact the interpretation of results.
Additional prompting questions for case-control
studies of occupational exposures:
Is exposure based on a comprehensive job history
describing tasks, setting, time period, and use of
specific materials?
Additional suggested considerations for occupational exposures (should be
evaluated in addition to the general considerations above):
• Describes the use of personal protective equipment.
• Confirmed contrast in exposure between groups using
biomarker measurements.
• Expert assessment method based on a detailed lifetime
occupational history and using a high-quality, validated job
exposure matrix (JEM) or a JEM that incorporates industry,
time period, population/country, tasks, and material used.
• Describes the use of personal protective equipment.
• Confirmed contrast in exposure between groups using
biomarker measurements.
Deficient • Expert assessment method based on incomplete
occupational history information (lacking job titles,
employers, industries, start and finish years, number of
hours worked per day, number of days worked per week,
tasks performed, or materials used) - may be Critically
Deficient, depending on severity of this limitation.
A-58
-------
APRIL 2024
Core Question: Does the exposure measure reliably distinguish between levels of exposure in a time window considered most relevant for a causal effect
with respect to the development of the outcome?
Critically
Deficient
> JEM with data indicating it cannot differentiate between
exposure levels over time, area, or between individuals.
A-59
-------
APRIL 2024
PFAS-Specific Exposure Measurement Study Quality Evaluation Criteria
Standard analytical methods of individual PFAS in serum or whole blood using quantitative
techniques, such as liquid chromatography triple quadrupole mass spectrometry, are considered
well-established methods (Table A-21).
Table A-21. Criteria for Evaluating Exposure Measurement in Epidemiology Studies of
PFAS and Health Effects
Rating Criteria
Good
• Evidence that exposure was consistently assessed using well-established analytical methods
that directly measure exposure (e.g., measurement of PFAS in blood, serum, or plasma).
OR
• Exposure was assessed using less established methods (e.g., measurement of PFAS in breast
milk) or methods that indirectly measure exposure (e.g., drinking water concentrations and
residential location/history, questionnaire or occupational exposure assessment by a certified
industrial hygienist) that are supported by well-established methods (i.e., inter-methods
validation: one method vs. another) in the target population of interest.
And all the following:
• Exposure was assessed in a relevant time-window (i.e., temporality is established, and
sufficient latency occurred prior to disease onset) for development of the outcome based on
current biological understanding.
• There is evidence that sufficient exposure data measurements are above the limit of
quantification for the assay.
• The laboratory analysis included data on standard quality control measures with demonstrated
precision and accuracy.
Adequate
• Exposure was assessed using less established methods or indirect measures that are validated
but not in the target population of interest.
OR
• Evidence that exposure was consistently assessed using methods described in Good, but there
were some concerns about quality control measures or other potential for nondifferential
misclassification.
And all the following:
• Exposure was assessed in a relevant time-window for development of the outcome.
• There is evidence that sufficient exposure data measurements are above the limit of
quantification for the assay.
• The laboratory analysis included some data on standard quality control measures with
demonstrated precision and accuracy.
Deficient
Any of the following:
• Some concern, but no direct evidence, that the exposure was assessed using methods that have
not been validated or empirically shown to be consistent with methods that directly measure
exposure.
• Exposure was assessed in a relevant time window(s) for development of the outcome, but there
could be some concern about the potential for bias due to reverse causality3 between exposure
and outcome, yet no direct evidence that it is present; or has somehow been mitigated by the
design, etc.
Critically
Deficient
Any of the following:
• Exposure was assessed in a time window that is unknown or not relevant for development of
the outcome. This could be due to clear evidence of bias from reverse causality between
exposure and outcome, or other concerns such as the lack of temporal ordering of exposure and
disease onset, insufficient latency, or having exposure measurements that are not reliable
measures of exposure during the etiologic window(s).
• Direct evidence that bias was likely because the exposure was assessed using methods with
poor validity.
A-60
-------
APRIL 2024
Rating Criteria
I • Evidence of differential exposure misclassification (e.g., differential recall of self-reported
exposure).
I • There is evidence that an insufficient number of the exposure data measurements were above
the limit of quantification for the assay.
Notes:
a Reverse causality refers to a situation where an observed association between exposure and outcome is not due to causality from
exposure to outcome, but rather due to the outcome of interest causing a change in the measured exposure.
A-61
-------
APRIL 2024
A.1.7.1.3 Outcome Ascertainment
This domain may need to be evaluated multiple times for a single study if more than one PECO-relevant outcome is reported.
Therefore, outcome-specific criteria (Radke et al., 2019) may be applied for each outcome measured in a study. Table A-22 presents
criteria that apply across outcomes.
Table A-22. Study Quality Evaluation Considerations for Outcome Ascertainment
Core Question: Does the outcome measure reliably distinguish the presence or absence (or degree of severity) of the outcome?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Is outcome ascertainment likely to be affected by Is there a concern that any Good
knowledge of, or presence of, exposure
(e.g., consider access to healthcare, if based on
self-reported history of diagnosis)?
For case-control studies:
Is the comparison group without the outcome
(e.g., controls in a case-control study) based on
objective criteria with little or no likelihood of
inclusion of people with the disease?
For mortality measures:
How well does cause of death data reflect
occurrence of the disease in an individual? How
well do mortality data reflect incidence of the
disease?
For diagnosis of disease measures:
Is the diagnosis based on standard clinical
criteria? If it is based on self-report of the
diagnosis, what is the validity of this measure?
For laboratory-based measures (e.g., hormone
levels):
Is a standard assay used? Does the assay have an
acceptable level of inter-assay variability? Is the
sensitivity of the assay appropriate for the
outcome reclassification
is nondifferential,
differential, or both?
What is the predicted
direction or distortion of
the bias on the effect
estimate (if there is enough
information)?
• High certainty in the outcome definition (i.e., specificity
and sensitivity), minimal concerns with respect to
misclassification.
• Assessment instrument was validated in a population
comparable to the one from which the study group was
selected.
A-62
-------
APRIL 2024
Core Question: Does the outcome measure reliably distinguish the presence or absence (or degree of severity) of the outcome?
outcome measure in this study population? Were
QA/QC measures and results reported?
Adequate • Moderate confidence that outcome definition was specific
and sensitive, some uncertainty with respect to
misclassification but not expected to greatly change the
effect estimate.
• Assessment instrument was validated but not necessarily
in a population comparable to the study group.
Deficient • Outcome definition was not specific or sensitive.
• Uncertainty regarding validity of assessment instrument.
Critically
• Invalid/insensitive marker of outcome.
Deficient
• Outcome ascertainment is very likely to be affected by
knowledge of, or presence of, exposure.
Note: Lack of blinding should not be automatically
construed to be Critically Deficient.
Notes: QA/QC = quality assurance/quality control.
A-63
-------
APRIL 2024
A.l.7.1.4 Potential Confounding
The aim of evaluating this domain is to ascertain whether confounding of the relationship between the exposure and health outcome of
interest is likely to exist, and if so, whether it was considered in the design and/or analysis of the study (Table A-23). Co-exposures to
other PFAS were considered in this domain.
Table A-23. Study Quality Evaluation Considerations for Confounding
Core Question: Is confounding of the effect of the exposure likely?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Is confounding adequately addressed by
considerations in:
• Participant selection (matching or restriction)?
• Accurate information on potential confounders
and statistical adjustment procedures?
• Lack of association between confounder and
outcome, or confounder and exposure in the
study?
• Information from other sources?
Is the assessment of confounders based on a
thoughtful review of published literature,
potential relationships (e.g., as can be gained
through directed acyclic graphing), and
minimizing potential overcontrol (e.g., inclusion
of a variable on the pathway between exposure
and outcome)?
If there is a concern about Good
the potential for bias, what is I
the predicted direction or
distortion of the bias on the
effect estimate (if there is
enough information)?
• Conveys strategy for identifying key confounders. This
may include: a priori biological considerations, published
literature, causal diagrams, or statistical analyses; with
recognition that not all "risk factors" are confounders.
• Inclusion of potential confounders in statistical models not
based solely on statistical significance criteria
(e.g., p < 0.05 from stepwise regression).
• Does not include variables in the models that are likely to
be influential colliders or intermediates on the causal
pathway.
• Key confounders are evaluated appropriately and
considered to be unlikely sources of substantial
confounding. This often will include:
o Presenting the distribution of potential confounders by
levels of the exposure of interest and/or the outcomes
of interest (with amount of missing data noted);
o Consideration that potential confounders were rare
among the study population, or were expected to be
poorly correlated with exposure of interest;
o Consideration of the most relevant functional forms of
potential confounders;
o Examination of the potential impact of measurement
error or missing data on confounder adjustment;
o Presenting a progression of model results with
adjustments for different potential confounders, if
warranted.
A-64
-------
APRIL 2024
Core Question: Is confounding of the effect of the exposure likely?
Adequate • Similar to Good but may not have considered all potential
confounders (though all key confounders were considered),
or less detail may be available on the evaluation of
confounders (e.g., sub-bullets in Good). It is possible that
residual confounding could explain part of the observed
effect, but concern is minimal.
Deficient • All key confounders were not considered by design or in
the statistical analysis.
• Assessed an outcome based on report of medical diagnosis
that would have required access to a health professional
(e.g., autism, ADHD, depression) and failed to consider
some marker of socioeconomic status (e.g., maternal
education, household income, marital status, crowding,
poverty, job status) as a potential confounder.
• Does not include variables in the models that are likely to
be influential colliders or intermediates on the causal
pathway.
And any of the following:
• The potential for bias to explain some of the results is high
based on an inability to rule out residual confounding, such
as a lack of demonstration that key confounders of the
exposure-outcome relationships were considered;
• Descriptive information on key confounders (e.g., their
relationship relative to the outcomes and exposure levels) is
not presented; or
• Strategy of evaluating confounding is unclear or is not
recommended (e.g., only based on statistical significance
criteria or stepwise regression [forward or backward
elimination]).
Critically J • Includes variables in the models that are colliders and/or
intermediates in the causal pathway, indicating that
substantial bias is likely from this adjustment; or
• Substantial confounding is likely present and not accounted
for, such that all of the results were most likely due to bias.
• If confounders not considered by design or in the analysis
(e.g.. only simple correlations presented).
Notes: ADHD = attention deficit hyperactivity disorder.
A-65
-------
APRIL 2024
A.l.7.1.5 Analysis
Information relevant to evaluation of analysis includes, but is not limited to: the extent (and if applicable, treatment) of missing data
for exposure, outcome, and confounders; approach to modeling; classification of exposure and outcome variables (continuous vs.
categorical); testing of assumptions; sample size for specific analyses; and relevant sensitivity analyses (Table A-24).
Table A-24. Study Quality Evaluation Considerations for Analysis
Core Question: Does the analysis strategy and presentation convey the necessary familiarity with the data and assumptions?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Are missing outcome, exposure, and covariate
data recognized, and if necessary, accounted for
in the analysis?
Does the analysis appropriately consider variable
distributions and modeling assumptions?
Does the analysis appropriately consider
subgroups of interest (e.g., based on variability in
exposure level or duration or susceptibility)?
Is an appropriate analysis used for the study
design?
Is effect modification considered, based on
considerations developed a priori?
Does the study include additional analyses
addressing potential biases or limitations
(i.e., sensitivity analyses)?
If there is a concern about
the potential for bias, what
is the predicted direction or
distortion of the bias on the
effect estimate (if there is
enough information)?
Use of an optimal characterization of the outcome variable.
> Quantitative results presented (effect estimates and
confidence limits or variability in estimates (e.g., standard
error, standard deviation); i.e., not presented only as a
p-value or "significant'T'not significant").
> Descriptive information about outcome and exposure
provided (where applicable).
> Amount of missing data noted and addressed appropriately
(discussion of selection issues - missing at random vs.
differential).
> Where applicable, for exposure, includes LOD (and
percentage below the LOD), and decision to use log
transformation.
> Includes analyses that address robustness of findings,
e.g., examination of exposure-response (explicit
consideration of nonlinear possibilities, quadratic, spline, or
threshold/ceiling effects included, when feasible); relevant
sensitivity analyses; effect modification examined based
only on a priori rationale with sufficient numbers.
> No deficiencies in analysis evident. Discussion of some
details may be absent (e.g.. examination of outliers).
> Same as Good, except:
> Descriptive information about exposure provided (where
applicable) but may be incomplete; might not have
discussed missing data, cut points, or shape of distribution.
> Includes analyses that address robustness of findings
(examples in Good), but some important analyses are not
performed.
A-66
-------
APRIL 2024
Core Question: Does the analysis strategy and presentation convey the necessary familiarity with the data and assumptions?
Deficient • Descriptive information about exposure levels not provided
(where applicable).
• Effect estimate and p-value presented, without standard
error or confidence interval (where applicable).
• Results presented as statistically "significant'T'not
significant."
• Results of analyses of effect modification examined without
clear a priori rationale and without providing main/principal
effects (e.g., presentation only of statistically significant
interactions that were not hypothesis driven).
• Analysis methods are not appropriate for design or data of
the study.
Notes: LOD = limit of detection.
Critically
Deficient
A-67
-------
APRIL 2024
A.l.7.1.6 Selective Reporting
This domain concerns the potential for misleading results that can arise from selective reporting (e.g., of only a subset of the measures
or analyses that were conducted). The concept of selective reporting involves the selection of results from among multiple outcome
measures, multiple analyses, or different subgroups, based on the direction or magnitude of these results (e.g., presenting "positive"
results) (Table A-25).
Table A-25. Study Quality Evaluation Considerations for Selective Reporting
Core Question: Is there reason to be concerned about selective reporting?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Were results provided for all the primary
analyses described in the methods section?
Is there appropriate justification for restricting
the amount and type of results that are shown?
Are only statistically significant results
presented?
If there is a concern about
the potential for bias, what
is the predicted direction or
distortion of the bias on the
effect estimate (if there is
enough information)?
Adequate • The results reported by study authors are consistent with
the primary and secondary analyses described in a
registered protocol or methods paper
OR
• The authors described their primary (and secondary)
analyses in the methods section and results were reported
for all primary analyses.
Deficient • Concerns were raised based on previous publications, a
methods paper, or a registered protocol indicating that
analyses were planned or conducted that were not reported,
or that hypotheses originally considered to be secondary
were represented as primary in the reviewed paper.
• Only subgroup analyses were reported; results for the entire
group were omitted without any justification (e.g., to
address effect measure modification).
• Of the PECO-relevant outcomes examined, only
statistically significant results were reported.
A-68
-------
APRIL 2024
A.l.7.1.7 Study Sensitivity
The aim of evaluation of this domain is to determine if there are features of the study that affect its ability to detect a true association
(Table A-26). Some of the study features that can affect study sensitivity may have already been included in the outcome, exposure, or
other categories, such as the validity of a method used to ascertain an outcome, the ability to characterize exposure in a relevant time
period for the outcome under consideration, selection of affected individuals out of the study population, or inappropriate inclusion of
intermediaries in a model.
Other features may not have been addressed, and so should be included here. Examples include the exposure range (e.g., the contrast
between the "low" and "high" exposure groups within a study), the level or duration of exposure, and the length of follow-up. In some
cases (for very rare outcomes), sample size or number of observed cases may also be considered within this "sensitivity" category.
Table A-26. Study Quality Evaluation Considerations for Study Sensitivity
Core Question: Is there a concern that sensitivity of the study is not adequate to detect an effect?
Prompting Questions
Follow-Up Questions
Suggested Considerations
Is the exposure range/contrast adequate to detect
associations that are present?
Was the appropriate (at risk) population included?
Was the length of follow-up adequate? Is the
time/age of outcome ascertainment optimal given
the interval of exposure and the health outcome?
Are there other aspects related to risk of bias or
otherwise that raise concerns about sensitivity?
Adequate • The range of exposure levels provides adequate variability
to evaluate primary hypotheses in study.
• The population was exposed to levels expected to have an
impact on response.
• The study population was sensitive to the development of
the outcomes of interest (e.g., ages, lifestage, sex).
• The timing of outcome ascertainment was appropriate
given expected latency for outcome development
(i.e., adequate follow-up interval).
• The main effects and stratified analyses were fairly precise
(relatively small confidence bounds)
• The study was adequately powered to observe an effect.
Consider sample size, precision (e.g., width of confidence
intervals), anticipated power, exposure ranges and
contrasts.
• No other concerns raised regarding study sensitivity.
Deficient • Concerns were raised about the issues described for
Adequate that are expected to notably decrease the
sensitivity of the study to detect associations for the
outcome.
A-69
-------
APRIL 2024
A.l.7.1.8 Overall Confidence
Table A-27. Evaluation Considerations for Overall Study Confidence - Overall Confidence, Epidemiological Studies
Provide judgement and rationale for each endpoint or groups of endpoints. The overall confidence rating considers the likely impact of the noted concerns
(i.e., limitations or uncertainties) in reporting, bias and sensitivity on the results. Evaluation Core Question: Considering the identified strengths and
limitations, what is the overall confidence rating for the endpoint(s)/outcome(s) of interest?
Prompting Questions
Suggested Considerations
High
confidence
For each endpoint/outcome or grouping of
endpoints/outcomes in a study:
Were concerns (i.e., limitations or uncertainties)
related to the reporting quality, risk of bias, or
sensitivity identified?
If yes, what is their expected impact on the overall
interpretation of the reliability and validity of the
study results, including (when possible) Low confidence
interpretations of impacts on the magnitude or
direction of the reported effects?
• No notable concerns are identified (e.g., most or all domains rated Good).
NOTE: Reviewers should mark studies that are
rated lower than high confidence only due to low
sensitivity (i.e., bias towards the null) for
additional consideration during evidence
synthesis. If the study is otherwise well-conducted
and an effect is obser\>ed, the confidence may be
increased.
Medium • Some concerns are identified but expected to have minimal impact on the interpretation
confidence of the results, (e.g., most domains rated Adequate or Good; may include studies with
Deficient ratings if concerns are not expected to strongly impact the magnitude or
direction of the results). Any important concerns should be carried forward to evidence
synthesis.
• Identified concerns are expected to significantly impact on the study results or their
interpretation (e.g., generally. Deficient ratings for one or more domains). The concerns
leading to this confidence judgment must be carried forward to evidence synthesis (see
note).
Uninformative
• Serious flaw(s) that make the study results unusable for informing hazard identification
(e.g., generally. Critically Deficient rating in any domain; many Deficient ratings).
Uninformative studies are not considered further in the synthesis and integration of
evidence.
A-70
-------
APRIL 2024
A. 1.7.2 Study Quality Evaluation for Animal Toxicological Studies
As noted in the IRIS Handbook, the approach to evaluating study quality for animal
toxicological studies considers study design and experimental conduct in the context of reporting
quality, risk of bias, and study sensitivity. As outlined in Section A. 1.7 of this appendix, study
quality evaluations are performed for a set of established domains, and core and prompting
questions are provided for each domain to guide the reviewer. Each domain is assigned a score
of Good, Adequate, Deficient, Not Reported or Critically Deficient, and rationales to support
the scores are developed. Once all domains are evaluated, a confidence rating of High, Medium,
or Low confidence or Uninformative is assigned for each endpoint/outcome from the study.
The tables in the following sections describe the core and prompting questions and
considerations for assessing each domain during animal toxicological study quality evaluation.
Tables within each section also provide example evaluations for each domain.
A-71
-------
APRIL 2024
A.l.7.2.1 Reporting Quality
Evaluation of this domain is focused on ascertaining whether the study reports enough information to enable evaluation of the study
(Table A-28).
Table A-28. Study Quality Evaluation Considerations for Reporting Quality
Core Question: Does the study report information for evaluating the design and conduct of the study for the endpoint(s)/outcome(s) of interest?
Prompting Questions
Suggested Considerations Example Answers
Does the study report the following?
Good
• Minimal concern for selection bias based • Good. Important information is provided
on description of recruitment process (e.g., for test species, strain, sex, age, exposure
Critical information necessary to perform
selection of comparison population, methods, experimental design, endpoint
study evaluation:
population-based random sample evaluations and the presentation of results.
• Species; test article name; levels and duration of
selection, recruitment from sampling • The authors report that "the study was
exposure; route (e.g., oral; inhalation);
frame including current and previous conducted in compliance with the OECD
qualitative or quantitative results for at least one
employees) such that study participants guidelines for Good Laboratory Practice
endpoint of interest
were unlikely to differ from a larger cohort [c(81) 30 (Final)]."
based on recruitment or enrollment
Imnortant information for evaluating the studv
methods (or data provided to confirm a
methods:
lack of difference)
• Test animal: strain, sex, source, and general
• Exclusion and inclusion criteria specified
husbandry procedures
and would not be likely to induce bias.
• Exposure methods: source, purity, method of
• Participation rate is reported at all steps of
administration
study (e.g., initial enrollment, follow-up.
• Experimental design: frequency of exposure.
selection into analysis sample). If rate is
animal age and lifestage during exposure and at
not high, there is appropriate rationale for
endpoint/outcome evaluation
why it is unlikely to be related to exposure
• Endpoint evaluation methods: assays or
(e.g., comparison between participants and
procedures used to measure the
nonparticipants or other available
endpoints/outcomes of interest
information indicates differential selection
is not likely).
• Comparison groups are similar with
respect to factors expected to influence
exposure-outcome relationship
(confounders, effect measure modifiers).
A-72
-------
APRIL 2024
Core Question: Does the study report information for evaluating the design and conduct of the study for the endpoint(s)/outcome(s) of interest?
Note:
• Reviewers should reach out to authors to obtain
missing information when studies are
considered key for hazard evaluation and/or
dose-response.
• This domain is limited to reporting. Other
aspects of the exposure methods, experimental
design, and endpoint evaluation methods are
evaluated using the domains related to risk of
bias and study sensitivity.
Adequate • Enough of a description of the recruitment
process (i.e., recruitment strategy,
participant selection or case ascertainment)
to be comfortable that there is no serious
risk of bias.
• Inclusion and exclusion criteria specified
and would not induce bias.
• Participation rate is incompletely reported
for some steps of the study, but available
information indicates participation is
unlikely to be related to exposure.
• Comparison groups are largely similar
with respect to factors expected to
influence exposure-outcome relationship
(confounders, effect measure modifiers) or
these are mostly accounted for in the study
analysis.
> Adequate. All critical information is
reported but some important information
is missing. Specifically, it is unclear what
strain of rats was used.
Deficient • Little information on recruitment process,
selection strategy, sampling framework
and/or participation OR aspects of these
processes raises the likelihood of bias
(e.g., healthy worker effect, survivor bias).
Example: Enrollment of "cases" from a
specific clinic setting (e.g., diagnosed
autism), which could be biased by referral
practices and ser\>ices availability, without
consideration of similar selection forces
affecting recruitment of controls.
• Deficient. All critical information is
reported, but some important information
is missing that makes additional study
evaluation and interpretation of the results
difficult. Specifically, it is not reported
(and cannot be inferred) what age/lifestage
the animals were at outcome evaluation.
Critically ¦ • Aspects of the processes for recruitment.
Deficient selection strategy, sampling framework, or
participation result in concern that the
likelihood of selection bias is high
(e.g., convenience sample with no
information about recruitment and
selection, cases and controls are recruited
from different sources with different
likelihood of exposure, recruitment
materials stated outcome of interest and
• Example 1: Critically Deficient. Critical
information is missing. Authors did not
report the duration of the exposure or the
results (qualitative or quantitative).
• Example 2: Critically Deficient. Critical
information is missing. The study reports
animals were exposed to per-and
polyfluoroalkyl substances (PFAS), but
the specific chemicals tested were not
provided.
A-73
-------
APRIL 2024
Core Question: Does the study report information for evaluating the design and conduct of the study for the endpoint(s)/outcome(s) of interest?
¦ potential participants are aware of or are
concerned about specific exposures).
• Convenience sample, and recruitment and
selection not described.
• Case report, case series, or other study
designs lacking a comparison group (these
should be excluded if they do not meet
assessment PECO criteria).
Notes: OECD = Organisation for Economic Co-operation and Development.
For the Reporting domain, the Deficient rating was used as a flag to potentially reach out to study authors to obtain missing critical information (e.g., blinding, randomization) that
may impact the overall confidence rating of the study (e.g., from b confidence to Medium confidence). A Deficient rating does not necessarily relegate the study to Low
confidence, but it is an indicator that obtaining information from the study authors may change the overall confidence rating. EPA could then judge if it was necessary to contact
the study authors. If the study received a Deficient rating for this domain and correspondence with the study authors could potentially increase the confidence, a statement was
added to indicate that obtaining information from the study authors could impact the confidence.
If EPA followed up with authors to obtain missing information, the study details page was updated to note that the authors were contacted and provided the corresponding details.
A-74
-------
APRIL 2024
A.l.7.2.2 Selection and Performance - Allocation
Table A-29. Study Quality Evaluation Considerations for Selection and Performance - Allocation
Core Question: Were animals assigned to experimental groups using a method that minimizes selection bias?
Prompting Questions
Suggested Considerations
Example Answers
For each study:
Did each animal or litter have an equal chance of
being assigned to any experimental group (i.e.,
random allocation)?
Is the allocation method described?
Aside from randomization, were any steps taken
to balance variables across experimental groups
during allocation?
> Experimental groups were randomized and
any specific randomization procedure was
described or inferable (e.g., computer-
generated scheme). [Note that
normalization is not the same as
randomization (see response for
Adequate').]
> Good. The study authors report that "Fifty
males and fifty females were randomly
assigned to groups by a computer-
generated weight-ordered distribution such
that individual body weights did not
exceed + 20% of the mean weight for each
sex."
• Authors report that groups were
randomized but do not describe the
specific procedure used (e.g., 'animals
were randomized'). Alternatively, authors
used a non-random method to control for
important modifying factors across
experimental groups (e.g., body weight
normalization).
• Example 1: Adequate. Randomization
was not performed. However,
normalization procedures that balance
important variables across groups were
performed. Specifically, the authors state
that animals were "allocated into groups
with similar distributions in body weight."
• Example 2: Adequate. The study authors
state that "animals were randomly
distributed to exposure groups." However,
the specific randomization method used
was not described.
• Example 3: Adequate. Randomization
was not explicitly reported. However, the
study was performed according to OECD
416 and EPA OPPT 870.3800 guidelines
which both specify randomization,
although the specific methods of
randomization used in the current study
could not be inferred. OECD 416
guidelines state "animals should be
randomly assigned to the control and
A-75
-------
APRIL 2024
Core Question: Were animals assigned to experimental groups using a method that minimizes selection bias?
treated groups (stratification by body
weight is recommended)." The EPA OPPT
870.3800 guidelines state "animals should
be randomly assigned to the control and
treatment groups, in a manner which
results in comparable mean body weight
values among all groups."
• Example 4: Adequate. The study authors
state that "Animals were randomized by
weight into treatment groups," and do not
present the specific randomization
procedural details.
Not Reported
• No indication of randomization of groups • Not reported (interpreted as Deficient).
(Interpreted as
or other methods (e.g., normalization) to The authors did not indicate
Deficient)
control for important modifying factors randomization or other normalization
across experimental groups. procedures for balancing important
variables across groups.
Critically
• Bias in the animal allocations was reported • Critically Deficient. There is direct
Deficient
or inferable. evidence that animals were allocated to
treatment groups in a subjective way.
involving the judgment of the investigator.
Specifically, the study authors report "the
heavier dams were assigned to the higher
dose groups to reduce toxicity from
[chemical]"; dam weight is an important
variable for these developmental
outcomes.
Notes: OECD = Organisation for Economic Co-operation and Development; OPPT = Office of Pollution Prevention and Toxics.
A-76
-------
APRIL 2024
A.l.7.2.3 Selection and Performance - Observational Bias/Blinding
Table A-30. Study Quality Evaluation Considerations for Selection and Performance - Observational Bias/Blinding
Core Question: Did the study implement measures to reduce observational bias?
Prompting Questions
Suggested Considerations Example Answers
For each endpoint/outcome or grouping of
Good
• Measures to reduce observational bias • Examnlc 1: Good. Histonatholoev:
endpoints/outcomes in a study:
were described (e.g., blinding to conceal Although the study did not indicate
treatment groups during endpoint blinding, blinding during the initial
Does the study report blinding or other
evaluation; consensus-based evaluations of evaluation of tissues for initial or non-
methods/procedures for reducing observational
histopathology lesions3). targeted evaluations is generally not
bias?
recommended as masked evaluation can
make the task of separating treatment-
If not, did the study use a design or approach for
related changes from normal variation
which such procedures can be inferred?
more difficult and may result in subtle
lesions being overlooked (Crissman et al..
What is the expected impact of failure to
2004). The study did include a secondary
implement (or report implementation) of these
evaluation by a pathology working group
methods/procedures on results?
(PWG) review on coded pathology slides
which minimized the potential for
observational bias.
• Examiile 2: Good. Orsan weishts. FOB.
motor activity, swim maze and
histooatholoev: Authors reported that the
investigators were blinded to the animal
treatment group during evaluation for all
outcome measures. Although blinding is
not recommended for initial or non-
targeted evaluations (Crissman et al..
2004), this study evaluated prespecified
outcomes in targeted evaluations for
which blinding is appropriate (cell counts
in the CA3 region of the hippocampus).
Adequate • Methods for reducing observational bias • Adequate. Histopathology measures:
(e.g., blinding) can be inferred or were Authors report "lesions were counted by 2
reported but described incompletely. observers in a blinded fashion" although it
should be noted that blinding during the
initial evaluation of tissues is generally not
recommended for initial or non-targeted
A-77
-------
APRIL 2024
Core Question: Did the study implement measures to reduce observational bias?
evaluations as masked evaluation can
make the task of separating treatment-
related changes from normal variation
more difficult and may result in subtle
lesions being overlooked (Crissman et al..
2004).
Not Reported
• Measures to reduce observational bias • Example 1: Not reported (interpreted as
(Interpreted as
were not described. Adeauate). Bodv and orsan weiehts.
Adequate)
• The potential concern for bias was developmental landmarks, and hormone
mitigated based on use of measures: Authors did not indicate
automated/computer driven systems, whether investigators were blinded during
standard laboratory kits, relatively simple, outcome assessment. Potential concern for
objective measures (e.g., body or tissue bias was mitigated for these endpoints
weight), or screening-level evaluations of which were measured using
histopathology. automated/computer driven systems.
standard laboratory kits, relatively simple.
objective measures (e.g., body or tissue
weight).
• Example 2: Not reported (interpreted as
Adeauate). Histonatholoev: Blindins
during the initial evaluation of tissues is
generally not recommended as masked
evaluation can make the task of separating
treatment-related changes from normal
variation more difficult and may result in
subtle lesions being overlooked (Crissman
et al., 2004). Histopathology was
evaluated by an independent laboratory
(Toxicology Pathology Associates Little
Rock, Arkansas, John Pletcher, D.V.M.,
DACPV). No subsequent steps to
minimize the potential for observational
bias were reported (i.e., conducting a
secondary targeted blinded review.
independent prospective or retrospective
peer-review, formation of a pathology
working group).
A-78
-------
APRIL 2024
Core Question: Did the study implement measures to reduce observational bias?
• Example 3: Not reported (interpreted as
Adequate). Fetal evaluation for
malformations: Blinding during initial
evaluation of fetuses is typically not
conducted as masked evaluation can make
the task of separating treatment-related
changes from normal developmental
variation more difficult and may result in
subtle developmental anomalies being
overlooked. Fetal evaluations were
conducted in accordance with regulatory
test guideline recommendations, using
standardized nomenclature. No subsequent
steps to minimize the potential for
observational bias were reported (e.g.,
conducting a secondary targeted blinded
review, or an independent prospective or
retrospective peer-review).
• Not reported (interpreted as Deficient).
Neurobehavior (auditory and visual
sensory reactivity): Procedural methods
addressing observational bias were not
described for these endpoints, which were
measured using highly subjective methods
(i.e., it appears that investigators measured
reactivity using manually operated timers).
Critically
• Strone evidence for observational bias that • Critically Deficient. Neurobehavior after
Deficient
could have impacted results. restraint stress: There is direct evidence of
observational bias in testing methods.
Specifically, the study reported that, to
minimize stress from changing
investigators across trials, one investigator
consistently stressed control mice each
day for 30 minutes and subsequently
tested behaviors, while a separate
investigator conducted stress and
behavioral testing in treated mice. There
A-79
Not Reported • Measures to reduce observational bias
(Interpreted as were not described.
Deficient) • The potential impact on the results is
major (e.g., outcome measures are highly
subjective).
-------
APRIL 2024
Core Question: Did the study implement measures to reduce observational bias?
was no mention of blinding of
investigators.
Notes: FOB = functional observed battery.
aFor non-targeted or screening-level histopathology outcomes often used in guideline studies, blinding during the initial evaluation of tissues is generally not recommended as
masked evaluation can make 'the task of separating treatment-related changes from normal variation more difficult' and 'there is concern that masked review during the initial
evaluation may result in missing subtle lesions.' Generally, blinded evaluations are recommended for targeted secondary review of specific tissues or in instances when there is a
pre-defmed set of outcomes that is known or predicted to occur (Crissman et al., 2004).
A-80
-------
APRIL 2024
A. 1.7.2.4 Confounding/Variable Control
Table A-31. Study Quality Evaluation Considerations for Confounding/Variable Control
Core Question: Are variables with the potential to confound or modify results controlled for and consistent across all experimental groups?
Prompting Questions
Suggested Considerations
Example Answers
For each study:
Are there differences across the treatment groups
(e.g., co-exposures, vehicle, diet, palatability,
husbandry, health status, etc.) that could bias the
results?
If differences are identified, to what extent are
they expected to impact the results?
» Outside of the exposure of interest,
variables that are likely to confound or
modify results appear to be controlled for
and consistent across experimental groups.
» Good. On the basis of the study report,
vehicle (deionized water with 2% Tween
80) and husbandry practices were inferred
to be the same in controls and treatment
groups. The experimental conditions
described provided no indication of
concern for uncontrolled variables or
different practices across groups.
» Some concern that variables that were
likely to confound or modify results were
uncontrolled or inconsistent across groups
but are expected to have a minimal impact
on the results.
1 Example 1 (oral): Adequate. Honnone
measurements: Authors did not use a soy-
free diet. Soy-based rodent feeds contain
phytoestrogens that may act as a
confounder for endocrine-related
measures. Since this study includes
relatively high doses (100 and
1500 mg/kg-d) the concern is minimal.
Example 2 (inhalation): Adequate.
Behavior, immunological responses, and
hormonal changes: control rats did not
appear to receive chamber air exposures
(they were left in their home cages). As
this might introduce a difference in
stressors across groups, this difference is
interpreted as a possible confounder for
measures shown to be sensitive to stress,
although the impact of this limitation on
the results is expected to be minimal.
A-81
-------
APRIL 2024
Core Question: Are variables with the potential to confound or modify results controlled for and consistent across all experimental groups?
Deficient • Notable concern that potentially • Deficient. Dams in the medium and high
confounding variables were uncontrolled exposure groups (1500 and 15,000 ppm,
or inconsistent across groups and are respectively) showed significantly lower
expected to substantially impact the consumption of the treated food
results. throughout the exposure period (gestation)
that increased to control levels after the
exposure ended. Addition of the test
chemical may have affected the
palatability of the food and reduced food
intake during gestation may have
significantly impacted the developmental
outcomes in the pups.
• Critically Deficient. The study did not
include a vehicle-only control group, and,
given the high concentration of DMSO
required to solubilize the test article in
other experiments using a similar
exposure design, this is interpreted as
likely to be a significant driver of any
observed effects.
Notes: ppm = parts per million; DMSO = dimethyl sulfoxide.
Critically
Deficient
• Confounding variables were presumed to
be uncontrolled or inconsistent across
groups, and are expected to be a primary
driver of the results.
A-82
-------
APRIL 2024
A.l.7.2.5 Reporting and Attrition Bios
Table A-32. Study Quality Evaluation Considerations for Selective Reporting and Attrition - Reporting and Attrition Bias
Core Question: Did the study report results for all prespecified outcomes and tested animals?
Prompting Questions
Suggested Considerations
Example Answers
For each study:
Selective reporting bias:
Are all results presented for endpoints/outcomes
described in the methods (see note)?
Attrition bias:
Are all animals accounted for in the results?
If there are discrepancies, do authors provide an
explanation (e.g., death or unscheduled sacrifice
during the study)?
If unexplained results omissions and/or attrition
are identified, what is the expected impact on the
interpretation of the results?
NOTE: This domain does not consider the
appropriateness of the analysis/results
presentation. This aspect of study quality is
evaluated in another domain.
' Quantitative or qualitative results were
reported for all prespecified outcomes
(explicitly stated or inferred), exposure
groups and evaluation timepoints. Data not
reported in the primary article is available
from supplemental material. If results
omissions or animal attrition are identified,
the authors provide an explanation and
these are not expected to impact the
interpretation of the results.
• Good. Animal loss was reported (the
authors treated 10 rats/sex/dose group and
noted one death in a high-dose male rat at
day 85 of study). All endpoints described
in methods were reported qualitatively or
quantitatively.
• Quantitative or qualitative results are
reported for most prespecified outcomes
(explicitly stated or inferred), exposure
groups and evaluation
timepoints. Omissions and/or attrition are
not explained but are not expected to
significantly impact the interpretation of
the results.
• Adequate. Animal loss occurred and was
reported (see below), but these are not
expected to significantly impact the
interpretation of the results. All endpoints
described in methods were reported
qualitatively or quantitatively.
• "In the high dose (1000 mg/kg-day) group
no male animals were able to complete the
entire study; whereas all male rats exposed
at other doses completed the 4-week
experiment. In the female group, 1 rat was
removed in the 250 mg/kg-day group at
day 25, 1 rat in the 500 mg/kg-day was
removed at day 21 and 8 rats in the
1000 mg/kg/day group were removed
between days 16 and 27 of the
experiment." Justification for removals
was provided by the study authors.
A-83
-------
APRIL 2024
Core Question: Did the study report results for all prespecified outcomes and tested animals?
Deficient
• Quantitative or qualitative results are • Example 1: Deficient. Unaccounted for
missing for many prespecified outcomes loss of animals was difficult to assess
(explicitly stated or inferred), exposure because the study authors do not provide a
groups and evaluation timepoints and/or clear description of the number of animals
high animal attrition; omissions and/or per exposure group or the selection of
attrition are not explained and may animals for outcome analysis. Table 1
significantly impact the interpretation of states there were 8 animals used in
the results. experiment 1 and 6 animals used in
experiments 2 and 3. The figures and
tables report data for varying numbers of
animals (from 4 to 8), but the authors do
not provide a description of the approach
used to sample animals for each outcome.
• Example 2: Deficient. Although the
authors indicated that "the liver, kidneys.
and spleen were weighed and processed
for routine histopathology at study
termination," qualitative or quantitative
findings were not reported for liver or
kidney weights, nor for liver, kidney, or
spleen histopathology ("spleen weights"
were described as unchanged during the
description of changes in cultured splenic
immune cells).
Critically
• Extensive results omission and/or animal • Critically Deficient. None of the animals
Deficient
attrition are identified and prevents in the high and medium dose groups
comparisons of results across treatment survived and there was high mortality
groups. (>75%) in the low-dose group.
A-84
-------
APRIL 2024
A.l.7.2.6 Exposure Methods Sensitivity - Chemical Administration and Characterization
Table A-33. Study Quality Evaluation Considerations for Exposure Methods Sensitivity - Chemical Administration and
Characterization
Core Question: Did the study adequately characterize exposure to the chemical of interest and the exposure administration methods?
Prompting Questions
Suggested Considerations
Example Answers
For each study:
Does the study report the source and purity and/or
composition (e.g., identity and percent distribution!
of different isomers) of the chemical? If not, can
the purity and/or composition be obtained from
the supplier (e.g., as reported on the website)
Was independent analytical verification of the test
article purity and composition performed?
Did the authors take steps to ensure the reported
exposure levels were accurate?
For inhalation studies: were target concentrations
confirmed using reliable analytical measurements
in chamber air?
For oral studies: if necessary, based on
consideration of chemical-specific knowledge
(e.g., instability in solution; volatility) and/or
exposure design (e.g., the frequency and duration
of exposure), were chemical concentrations in the
dosing solutions or diet analytically confirmed?
Are there concerns about the methods used to
administer the chemical (e.g., inhalation chamber
type, gavage volume, etc.)?
NOTE: Consideration of the appropriateness of
the route of exposure is not evaluated at the
• Chemical administration and
characterization are complete (i.e., source,
purity, and analytical verification of the
test article are provided). There are no
concerns about the composition, stability,
or purity of the administered chemical, or
the specific methods of administration. For
inhalation studies, chemical concentrations
in the exposure chambers are verified
using reliable analytical methods.
» Example 1 (oral): Good. Source (3M)
and purity (98%) are described, and the
authors provided verification using
analytical methods (GC/MS). Addressing
concerns about known instability in
solution for this chemical, the authors
verified the dosing solutions twice weekly
over the course of the experiment.
Animals were exposed via gavage with all
dose groups receiving the same volume.
» Example 2 (inhalation): Good. Source
(3M) and purity (98%) of the test article
are described. All animals were transferred
to dynamic inhalation exposure chambers
for the exposures. The concentration of the
test chemical in the air was continuously
monitored from the animals' breathing
zone throughout the 6-hour exposure
periods and mean daily average
concentrations and variability were
reported.
» Some uncertainties in the chemical
administration and characterization are
identified but these are expected to have
minimal impact on interpretation of the
results (e.g., source and vendor- reported
purity are presented, but not independently
verified; purity of the test article is sub-
optimal but not concerning; For inhalation
studies, actual exposure concentrations are
» Example 1 (oral): Adequate. Purity
(98%) is described, but source is missing.
Purity is assumed to be vendor reported
because independent analytical
verification of the purity is not described.
Authors were contacted to try to obtain the
vendor information however they did not
respond. Stability assessments were not
necessary because fresh dosing solutions
were prepared daily.
A-85
-------
APRIL 2024
Core Question: Did the study adequately characterize exposure to the chemical of interest and the exposure administration methods?
individual study level. Relevance and utility of the missing or verified with less reliable
routes of exposure are considered in the PECO methods). • Example 2 (inhalation): Adequate.
criteria for study inclusion and during evidence Source (3M) and purity (98%) of the test
synthesis. article are described. All animals were
transferred to dynamic inhalation exposure
chambers for the exposures. The
nominal/target concentrations of the test
chemical were not verified by analytical
measurements of the chamber air.
• Example 1 (oral): Deficient. Test
chemical supplied by the chemical
manufacturer. Purity and isomeric
composition are not described and could
not be obtained from manufacturer's
website. Analytical verification of the test
article's purity and composition was not
provided, and the stability of chemical in
the diet across the 1-year exposure period
does not appear to have been assessed.
• Example 2 (inhalation): Deficient.
Source (3M) and vendor-reported purity
are described, although these were not
independently verified. The animals
appear to have been exposed in static (i.e.,
without dynamic airflow) chambers; this is
not interpreted as a critical deficiency due
to the relatively short (2-hour) durations of
daily exposure.
• Example 1 (oral): Critically Deficient.
The test article contains large amounts of a
known impurity [specify] that has
previously been shown to cause the
outcome(s) of interest. On the basis of the
doses tested (and inferences regarding the
administered doses of the impurity), this is
likely to be a significant driver of any
observed effects.
• Uncertainties in the exposure
characterization are identified and
expected to substantially impact the results
(e.g., source of the test article is not
reported; levels of impurities are
substantial or concerning; deficient
administration methods, such as use of
static inhalation chambers or a gavage
volume considered too large for the
species and/or lifestage at exposure).
Critically
Deficient
• Uncertainties in the exposure
characterization are identified and there is
reasonable certainty that the results are
largely attributable to factors other than
exposure to the chemical of interest (e.g.,
identified impurities are expected to be a
primary driver of the results).
A-86
-------
APRIL 2024
Core Question: Did the study adequately characterize exposure to the chemical of interest and the exposure administration methods?
Notes: GC/MS = gas chromatography mass spectrometry.
Example 2 (inhalation): Critically
Deficient. Dams were exposed in static
chambers during gestation, and there was
evidence of overt toxicity (i.e., gasping)
throughout the 12-hr daily exposures at all
tested concentrations. This is likely to be a
substantial driver of any observed
developmental effects.
A-87
-------
APRIL 2024
A.l.7.2.7 Exposure Methods Sensitivity - Exposure Timing, Frequency, and Duration
Table A-34. Study Quality Evaluation Considerations for Exposure Methods Sensitivity - Exposure Timing, Frequency, and
Duration
Core Question: Was the timing, frequency, and duration of exposure sensitive for the endpoint(s)/outcome(s) of interest?
Prompting Questions Suggested Considerations Example Answers
For each endpoint/outcome or grouping of
Good
• The duration and frequency of the • Example 1: Good. Study uses a standard
endpoints/outcomes in a study:
exposure was sensitive and the exposure OECD short-term (28-day) study design to
included the critical window of sensitivity examine toxicological effects that are
Does the exposure period include the critical
(if known). routinely evaluated in this testing
window of sensitivity?
guideline.
Was the duration and frequency of exposure
• Example 2: Good. experimental
sensitive for detecting the endpoint of interest?
potential
male
effects. The experiment was designed to
evaluate
recommendations
guidelines.
Adequate
• The duration and frequency of the • Adequate. The study does not include the
exposure was sensitive and the exposure full developmental window of exposure
covered most of the critical window of most informative to evaluating potential
sensitivity (if known). effects on androgen-dependent
development of male reproductive organs.
Specifically, the study exposed rats from
GD 18-GD 21, whereas the critical
window for the development of these
endpoints (i.e., cryptorchidism; testes and
seminal vesicle weights; and male
reproductive organ histopathology) begins
on GD 15, and peaks around GD 17 (NRC
2008 [635834]; Scott et al 2009 [673313])
in rats. The incomplete coverage of this
critical window in this study is expected to
result in a minor bias towards the null.
A-88
-------
APRIL 2024
Core Question: Was the timing, frequency, and duration of exposure sensitive for the endpoint(s)/outcome(s) of interest?
Deficient • The duration and/or frequency of the
exposure is not sensitive and did not
include the majority of the critical window
of sensitivity (if known). These limitations
are expected to bias the results towards the
null.
Critically
Deficient
> Deficient. The experimental design is not
considered appropriate for evaluation of
male fertility. Male rats were exposed for
chemical X for 1 wk and fertility was
assessed on wk 2 of the study. This design
is considered deficient because in most
rodent species "damage to spennatogonial
stem cells will not appear in samples from
the cauda epididymis or in ejaculates for 8
to 14 wk" (U.S. EPA, 1996).
> The exposure design was not sensitive and
is expected to strongly bias the results
towards the null. The rationale should
indicate the specific concern(s).
1 Critically Deficient. The experimental
design is not appropriate for evaluation of
cancer endpoints. Animals were
necropsied and tissues evaluated for the
presence of tumors and/or neoplasms 4 wk
after only a 28-day exposure period.
Notably, because this critical deficiency is
due to insensitivity, depending on other
identified limitations, the utility of this
study will depend on whether effects were
observed in the study (i.e., if tumors were
observed, this study could be adjusted to a
higher rating).
Notes: OECD = Organisation for Economic Co-operation and Development; OPPT = Office of Pollution Prevention and Toxics; GD = gestation day.
A-89
-------
APRIL 2024
A. 1.7.2.8 Outcome Measures and Results Display - End point Sensitivity and Specificity
Table A-35. Study Quality Evaluation Considerations for Outcome Measures and Results Display - Endpoint Sensitivity and
Specificity
Core Question: Are the procedures sensitive and specific for evaluating the endpoint(s)/outcome(s) of interest?
Prompting Questions Suggested Considerations Example Answers
For each endpoint/outcome or grouping of
Good
• Example 1: Good. Lipid/Lipoproteins:
endpoints/outcomes in a study:
There are no notable concerns about
aspects of the procedures, or for the timing
Are there concerns regarding the specificity and
of these evaluations. Study authors used
validity of the protocols?
standard methodology (i.e., commercial
kits) appropriate for use in adult liver
Are there serious concerns regarding the sample
tissue samples.
size (see note)?
• Example 2: Good. Orsan weisht. bodv
Are there concerns regarding the timing of the
weights, and hormone measures: no
endpoint assessment?
concerns regarding the specificity and
validity of the protocols and measures
NOTE: Sample size alone is not a reason to
were identified. Study authors used
conclude an individual study is critically deficient.
standard methodology for evaluating
organ and body weights. Thyroid
hormones were measured using
commercial electrochemiluminescence-
immunoassay methods, and the known
diurnal variation in these measures was
accounted for during blood collection.
Adequate
• Examiile 1: Adeauate. Histooatholoev:
Tissues were fixed in 10% neutral
buffered formalin, trimmed, sectioned (5
microns) and embedded and stained with
H&E. Evaluations included 12 tissues
from all animals in the control and highest
dose groups. Although not explicitly
stated, it is inferred that tissues from
animals in the low- and mid-dose groups
would have been evaluated if significant
increases in lesion incidence were
observed at the highest dose. This practice
A-90
-------
APRIL 2024
Core Question: Are the procedures sensitive and specific for evaluating the endpoint(s)/outcome(s) of interest?
is consistent with NTP pathology
guidelines (ref) and is expected to be of
minimal concern unless effects are
observed at the high dose. Additionally,
the report did not provide information on
sampling (e.g., # sections evaluated/tissue,
sections evaluated at x micron or section
intervals). Together, the missing study
details introduce some concern for
potential insensitivity.
• Example 2: Adequate. Clinical chemistry:
Some concern was raised regarding the
procedural methods, as no information
was provided on the diagnostic kits and,
for some of the specific measures (i.e.,
those without specific data reported), it is
unclear whether serum or plasma was
analyzed.
Deficient - • Example 1: Deficient. Histopathology
(testis): Concerns regarding the method
used to preserve testis for histological
analysis: 10% formalin. For evaluation of
histopathological effects in the testis,
conventional immersion fixation in
buffered formalin is not recommended as
it gives very poor penetration of fixative
and may result in artifacts (Haschek (ed)
et al 2009 [3987435]; Foley et al 2001
[PMID: 11215684]).
• Example 2: Deficient. Nipple retention:
Concerns for insensitivity were raised due
to the timing of endpoint evaluation.
Specifically, the authors examined nipple
retention in rats at PND 9, whereas this
endpoint is more appropriately evaluated
around PNDs 12-14.
A-91
-------
APRIL 2024
Core Question: Are the procedures sensitive and specific for evaluating the endpoint(s)/outcome(s) of interest?
• Exami)le 3: Deficient. Motor activity:
Concerns were raised regarding the small
sample sizes used to evaluate these
outcomes. Specifically, the authors tested
4 animals (sex not specified, but assumed
males) per group. Ideally, it is preferable
to have more than 10 animals/sex/group
for this type of evaluation, according to
OECD guidelines.
Critically
Deficient
- • Critically Deficient. fEndooint namel:
[Assay X] has been shown to be unreliable
for evaluating [endpoint of interest].
Currently best practice is to use [Assay Y]
for this endpoint.
Notes: NTP = National Toxicology Program; PND = postnatal day; OECD = Organisation for Economic Co-operation and Development.
A-92
-------
APRIL 2024
A.l.7.2.9 Outcome Measures and Results Display - Results Presentation
Table A-36. Study Quality Evaluation Considerations for Outcome Measures and Results Display - Results Presentation
Core Question: Are the results presented in a way that makes the data usable and transparent?
Prompting Questions
Suggested Considerations
Example Answers
For each endpoint/outcome or grouping of
endpoints/outcomes in a study:
Good
• Good. There are no notable concerns
about the way the results are analyzed or
presented.
Does the level of detail allow for an informed
interpretation of the results?
Adequate
-
• Examiile 1: Adeauate. Reproductive
orsan weiehts. hormone measures: results
Are the data analyzed, compared, or presented in a
way that is inappropriate or misleading?
are presented graphically; however, the
authors do not clarify whether error bars
correspond to SD or SE.
• Example 2: Adequate. Developmental
effects: the study failed to report
information on potential maternal toxicity;
however, all tested doses other than the
highest dose are not expected to cause
overt toxicity in adults, reducing the level
of concern.
• Example 3: Adequate. Anogenital
distance (AGP): The authors reported
AGD without adjusting for body weight,
which is preferred (Daston 1998
[3393032]). However, because the study
also provided body weight data,
approximation was possible, limiting
concern.
Deficient - • Example 1: Deficient. Histopathology:
Incidence and severity of individual
effects was unclear, as only scores across
multiple, disparate pathological endpoints
were reported.
• Example 2: Deficient. Behavior
(neuromuscular function and dexterity):
A-93
-------
APRIL 2024
Core Question: Are the results presented in a way that makes the data usable and transparent?
Performance on the rotarod was presented
as incidence of falling off the rod within
an arbitrary time, rather than as time spent
on the rod (the preferred metric). This
dichotomization of continuous data
without sound justification is expected to
strongly bias the results towards observing
an effect.
• Exami)le 3: Deficient. Brain weisht:
Authors presented only relative brain
weights, and absolute weights could not be
calculated. The adult CNS is highly
protected, and absolute brain weight data
are preferred [include reference].
• Exami)le 4: Deficient. Birth outcomes:
Data on pup viability, weights, and
malformations were reported as pup
averages, without addressing potential
litter effects.
Critically
Deficient
- • Critically Deficient. Endooint name: The
study presents the results for this endpoint
in both a table and figure; however, the
data do not match (e.g., mean ± SE
reported for the control group is 2.3 ± 0.5
in the table and 1.9 ± 0.2 in the figure).
This reporting discrepancy could not be
resolved from the information provided in
the study and study authors did not
respond to queries for clarification.
A-94
-------
APRIL 2024
A.l.7.2.10 Overall Confidence
The overall confidence rating considers the likely impact of the noted concerns (i.e., limitations or uncertainties) in reporting, bias and
sensitivity on the results (Table A-37).
Table A-37. Evaluation Considerations for Overall Study Confidence - Overall Confidence, Animal Toxicological Studies
Provide judgement and rationale for each endpoint or groups of endpoints. The overall confidence rating considers the likely impact of the noted
concerns (i.e., limitations or uncertainties) in reporting, bias and sensitivity on the results. Evaluation Core Question: Considering the identified
strengths and limitations, what is the overall confidence rating for the endpoint(s)/outcome(s) of interest?
Prompting Questions
Suggested Considerations
Example Answers
For each endpoint/outcome or grouping of High
endpoints/outcomes in a study: confidence
Were concerns (i.e., limitations or uncertainties)
related to the reporting quality, risk of bias, or
sensitivity identified?
If yes, what is their expected impact on the overall I
interpretation of the reliability and validity of the
study results, including (when possible)
interpretations of impacts on the magnitude or
direction of the reported effects?
NOTE: Reviewers should mark studies that are
rated lower than high confidence only due to low
sensitivity (i.e., bias towards the null) for
additional consideration during evidence
synthesis. If the study is otherwise well-conducted
and an effect is obsen'ed, the confidence may be
increased.
• No notable concerns are identified (e.g.,
most or all domains rated Good).
> High confidence. Reproductive and
developmental effects other than behavior:
The study was well-designed for the
evaluation of reproductive and
developmental toxicity induced by
chemical exposure. The study applied
established approaches, recommendations,
and best practices, and employed an
appropriate exposure design for these
endpoints. Evidence was presented clearly
and transparently.
Medium • Some concerns are identified but expected
confidence to have minimal impact on the
interpretation of the results, (e.g., most
domains rated Adequate or Good; may
include studies with Deficient ratings if
concerns are not expected to strongly
impact the magnitude or direction of the
results). Any important concerns should be
carried forward to evidence synthesis.
• Example 1: Medium confidence.
Developmental effects: The study was
adequately designed for the evaluation of
developmental toxicity. Although the
authors failed to describe randomized
allocation of animals to exposure groups
and some concerns were raised regarding
the sensitivity (i.e., timing) and sample
sizes (i.e., n = 6 litters/group) used for the
evaluation of potential effects on male
reproductive system development with
gestational exposure, these limitations are
expected to have a minimal impact on the
results.
A-95
-------
APRIL 2024
Provide judgement and rationale for each endpoint or groups of endpoints. The overall confidence rating considers the likely impact of the noted
concerns (i.e., limitations or uncertainties) in reporting, bias and sensitivity on the results. Evaluation Core Question: Considering the identified
strengths and limitations, what is the overall confidence rating for the endpoint(s)/outcome(s) of interest?
• Example 2: Medium confidence.
Histopathologv: The study authors did not
report information on the severity of
histological effects for which this is
routinely provided. The authors also failed
to describe use of methods to reduce
potential observational bias.
• Example 1: Low confidence.
Developmental effects: Substantial
concerns were raised regarding
quantitative analyses without addressing
potential litter effects. Other significant
limitations included incomplete data
presentation (sample sizes for outcome
assessment were unclear; no information
on maternal toxicity was provided), and
methods for selection of animals for
outcome assessment.
• Example 2: Low confidence. Behavioral
measures: The cursory cage-side
observations of activity are considered
insensitive and nonspecific methods for
detecting motor effects, with a strong bias
towards the null.
• Example 1: Uninformative. Critical
information was not reported. Specifically,
the study authors did not report the
duration of the exposure or the results
(qualitative or quantitative). Given this
critical deficiency, the other domains were
not evaluated.
• Example 2: Uninformative. Concerns
were raised over the lack of information
on test animal strain and allocation, and
Low confidence • Identified concerns are expected to
significantly impact on the study results or
their interpretation (e.g., generally.
Deficient ratings for one or more
domains). The concerns leading to this
confidence judgment must be carried
forward to evidence synthesis (see note).
Uninformative
> Serious flaw(s) that make the study results
unusable for informing hazard
identification (e.g., generally. Critically
Deficient rating in any domain; many
Deficient ratings). Uninformative studies
are not considered further in the synthesis
and integration of evidence.
A-96
-------
APRIL 2024
Provide judgement and rationale for each endpoint or groups of endpoints. The overall confidence rating considers the likely impact of the noted
concerns (i.e., limitations or uncertainties) in reporting, bias and sensitivity on the results. Evaluation Core Question: Considering the identified
strengths and limitations, what is the overall confidence rating for the endpoint(s)/outcome(s) of interest?
chemical source/purity. The lack of
information on blinding or other methods
to reduce observational blinding is also of
significant concern for the endpoints of
interest (i.e., follicle counts, ova counts,
and evaluation of estrous cyclicity).
Finally, concerns were also raised over the
apparent self-plagiarism in similar
chromium studies published in 1996 by
this group of authors. Taken together, this
combination of limitations resulted in an
interpretation that the results were
unreliable.
1 Example 3: Uninformative. Sperm
Measures: Issues were identified with the
methods used to prepare samples for
analysis, which are likely to introduce
artifacts. Concerns were also raised
regarding results presentation (i.e., lack of
group variability), missing information on
sample sizes and loss of animals, and a
lack of information on the timing of these
evaluations. Taken together, the evaluation
of this endpoint was considered
uninformative.
A-97
-------
APRIL 2024
A. 1.8 Data Extraction for Epidemiological Studies
All epidemiological studies identified as PECO-relevant after full-text screening were considered
eligible for data extraction. As noted in the IRIS Handbook (U.S. EPA, 2022c), during data
extraction, relevant results from each study are extracted to facilitate organization, visualization,
comparison, and analysis of findings and results. Data from PECO-relevant epidemiological
studies were extracted if they received a medium or high confidence study quality evaluation
rating. In cases where data were limited (e.g., thyroid cancer) or when there was a notable effect,
results from low confidence studies were extracted. Data extracted from low confidence studies
was considered qualitatively only (e.g., in the evidence synthesis and integration). Studies
evaluated as being uninformative were not considered further and therefore did not undergo data
extraction. Extraction was targeted towards the five priority health outcomes recommended by
the SAB (i.e., cancer, cardiovascular, developmental, hepatic, and immune). Results from main
analyses were extracted, and age- and sex-stratified analyses were extracted if available. Results
from other stratified and sensitivity analyses were extracted when deemed appropriate for a
given outcome (e.g., medication use status for cardiovascular outcomes).
Data extraction of epidemiological studies was carried out using a set of structured forms in
DistillerSR. Studies slated for extraction were pre-screened by an expert epidemiologist who
identified the relevant results to be extracted. Data extraction was performed by one reviewer
and then independently verified by at least one other reviewer for quality control. Any conflicts
or discrepancies related to data extraction were resolved by discussion and confirmation within
the extraction team.
Table A-38 outlines the content of the DistillerSR forms that were populated during data
extraction of epidemiological studies, including the extraction questions or prompts and response
options.
Table A-38. DistillerSR Form for Data Extraction of Epidemiological Studies
Question/Prompt Response Options Suggested Considerations
1
Has this study been
QC'd?
[Select one]
• No (select if doing data
extraction)
• Yes, no corrections needed
• Yes, corrections were
needed and completed
during QC (please list any
major revisions, e.g.,
incomplete responses,
NOEL/LOEL incorrect,
etc.)
• Study is not PECO-relevant
(please specify why)
2
Reference (short form)
e.g., Smith et al. (1978)
[Free-text]
• Enter author information; use the format
specified in the Distiller form.
3
Population
[Select one]
• General population, adults
and children
• General population, adults
• Do not select "pregnant women" if pregnant
women are only included as part of a general
population sample.
A-98
-------
APRIL 2024
Question/Prompt Response Options Suggested Considerations
• General population, • When exposure is measured in cord blood and
children and outcome in children, the study population
adolescents <18 yr would be "children."
• Occupational
• Pregnant women
• Occupational/general
population, adults
• Other
4 Population Summary - • Briefly describe the study population (e.g.,
[Free-text] women undergoing fertility treatment,
NHANES adults 18+). Try to capture anything
outside a typical general population sample.
Keep it brief - does not need to be in full
sentences.
• For studies of mother-child cohorts, when
exposure is in maternal blood and outcome is
evaluated in children, use "pregnant women
and their children."
• For example, if any of these (non-exhaustive')
scenarios apply, capture them in this field:
• Known potential for PFAS exposure (e.g.,
contamination event/lawsuit)
• Follow-up timing
• Participants are drawn from a specific
population, such as people with a specific
health condition, narrow age range within
"adults" and "children" (e.g., infants, seniors),
specific environments (e.g., assisted living
facility, daycare, farmers), etc.
5
Study Design
[Select one]
• Cohort
• Case-control
• Cross-sectional
• Ecological
• Controlled trial
• Other
• Nested caste-control
• Cross-sectional and
prospective analyses
• Cohort and cross-sectional
• Case-control and cross-
sectional
• See Appendix A. 1.8.1 for different types of
study design.
• Note: Third trimester samples with outcome
measured at birth should be classified as
cohort studies.
• Cohort studies reporting prospective and
cross-sectional analyses should be classified as
Cohort and cross-sectional.
• Case-control studies reporting cross-sectional
analyses among the whole study population or
within cases or controls should be classified as
Case-control and cross-sectional.
6
Study Name (if
applicable)
[Free-text]
• Only use the name of an official study or
cohort. Leave blank if there is no name.
7
Country (or Countries)
[Free-text]
-
• Use full names such as "United States" (not
US).
8
Year of Data
List which year(s) the data
came from.
[Free-text]
• For prospective cohort studies that only state
the period the population was recruited (e.g.,
2012-2015) and mention the outcomes were
assessed at follow-up (e.g., state "5 yr later"
but do not provide dates), extract "recruitment
A-99
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
2012-2015, outcome assessed at 5-year
follow-up."
9 Exposure Measurement
• Biomonitoring
-
[Select all that apply]
• Air
• Food
• Drinking water
• Occupational (use in cases
where exposure is based on
factors such as job function,
place in building where
people worked, job
exposure matrices)
• Modeled
• Questionnaire
• Direct administration - oral
• Direct administration -
inhalation
• Other
10 If "biomonitoring" was
• Blood
• For biomonitoring matrix, if PFAS is
selected, indicate the
• Serum
measured in serum, select serum (and not also
matrix.
• Plasma
blood). Only select blood if something more
[Select all that apply]
• Maternal blood
specific is not specified (e.g., cord blood,
• Cord blood
maternal blood, plasma, serum).
• Urine
• Feces
• Breast milk
• Hair
• Saliva
• Nails
• Teeth
• Semen
• Cerebrospinal fluid
• Exhaled breath
• Other
• Glucose
• Maternal serum
• Amniotic fluid
• Maternal Plasma
11 Quantitative Data Extraction (Sub-Forms)
11.1 Health Effect Category
• Cancer
• See Appendix A.1.6.5.1 for what kind of
[Select one]
• Cardiovascular
health outcomes are grouped under which
• Dermal
health effect category. Please create a separate
• Developmental
form for each outcome.
• Endocrine
• Gastrointestinal
• Hematologic
• Hepatic
• Immune
• Metabolic
A-100
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
• Musculoskeletal/Connective
Tissue
• Nervous
• Ocular
• Reproductive, female
• Reproductive, male
• Respiratory
• Renal
• Other
11.2 Measured
• Describe the measured outcome/endpoint and
Outcome/Endpoint
start with most relevant word (e.g., "glucose
[Free-text]
concentration in serum" preferred to "serum
glucose").
• Provide units in parentheses if relevant and
readily available.
• If the outcome is log transformed, please note
it here:
o Weight (ln-grams)
o Triglyceride (logio-mg/dL)
• Some outcomes are dichotomous (e.g., high
blood pressure, high cholesterol, etc.), indicate
the outcome definition in parentheses. For
example:
oHigh cholesterol (>5.0 mg/dL)
11.3 If developmental, when
• <2 yr of age
-
was the outcome
• 2-5 yr of age
measured?
• >5 yr of age
[Select all that apply]
11.4 PFAS
• PFOA
-
[Select one]
• PFOS
11.5 For neurodevelopmental
outcomes, when was
PFAS exposure
measured?
[Select all that apply]
Participants were <6 mo of
age
• Participants were >6 mo of
age
11.6 Sub-population
[Free-text]
• If relevant, specify subgroup within the study
(e.g., sex, age group, age at outcome and/or
exposure measurement).
• Leave blank if not applicable.
11.7 N
[Free-text]
> N should be for everyone in the analysis, not
just one exposure/comparison group.
However, if extracting results for specific
population subgroups (age category, gender-
specific) and if reported, the N should reflect
the number of participants in that specific
subgroup (e.g., number of boys in the male-
specific result extracted).
11.8 Exposure Levels
[Free-text]
• Exposure level should be for everyone in the
analysis, not just one comparison group.
• Ideally extract median and the 25th-75th
percentile range for PFAS being extracted.
The following format is preferred:
A-101
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
median = xx (units) (25th-75th percentile: xx-
xx).
• Provide labels and units (e.g.,
median = xx (units) (range: min-max: xx-
xx)).
o If median is not available, please extract
other measures of distribution, such as
mean or geometric mean, range, other
percentiles.
• Extract levels for the overall study
population. If only available by subgroups,
specify which subgroup.
• Example: Males: median = 6.4 ng/mL (25th-
75th percentile: 3.6-9.2 ng/mL); Females:
median = 5.8 ng/mL (25th-75th percentile:
3.1-8.3 ng/mL)
• Note: sometimes manuscripts will incorrectly
use IQR rather than 25th-75th percentile. The
IQR is the difference between the 75th and the
25th percentile, so it should be a single
number, not a range. If a range is labeled IQR,
please use "25th—75th percentile."
11.9 % with Negligible
Exposure (e.g., below the
LOD)
[Free-text]
• Number of samples below LOD/LOQ; do not
include the percent sign.
• Leave blank if not reported.
11.10 Description of the Effect
Estimate, including
Comparison Group if
applicable
[Free-text]
• Describe the effect estimate, including
comparison group if applicable.
• Brief description of the effect estimate:
describe the comparison being made (e.g., beta
regression coefficient for IQR increase; OR for
Q2 vs. Ql). Make sure to specify unit change
for continuous measures (e.g., 1 ln-unit, IQR
change, SD increase).
• Use ln() over log() for natural log
transformations. If not In, specify log (base)
(e.g., logioor log(10)).
Good Examples/Formatting:
• regression coefficient (per l-log2 ng/mL
increase in PFOA)
• OR (per 1-ln ng/mL increase in estimated
plasma PFOS)
• OR (for Q2 vs. Ql)
• OR [for Q2 (0.83-1.4 ng/mL) vs. Ql
(<0.83 ng/mL)]
• OR [for T2 (0.83-1.4 ng/mL) vs. T1
(<0.83 ng/mL)]
Bad Examples/Formatting:
• beta coefficient
A-102
-------
APRIL 2024
Question/Prompt
Response Options
Suggested Considerations
• linear regression coefficient (standard error)
with one unit increase in log-PFC in adults
11.11 Rank this Comparison
Group by Exposure
[Free-text]
• For standalone result of unit change, leave
blank.
• If results are presented for quantiles of
exposure, the comparison group for Q2 to Q1
would be ranked as 1, while Q3 to Q1 would
be ranked as 2.
11.12 Effect Estimate Type
[Select one]
• Odds Ratio (OR)
• Relative Risk Ratio (RR)
• Absolute Risk %
• Beta Coefficient (b)
• Beta Coefficient
(standardized)
• Standardized Mortality
Ratio (SMR)
• Standardized Incidence
Ratio (SIR)
• Incidence Risk Ratio (IRR)
• Absolute Risk
Reduction/Risk Difference
(ARR or RD)
• Hazard Ratio (HR)
• Comparison of Means
• Incidence Rate Ratio
• Comparison of Means
• Spearman's Correlation
Coefficient
• Correlation Coefficient
• Percent Incidence
• Regression Coefficient
• Proportionate Mortality
Ratio (PMR)
• Mean Difference
• Percent Difference
• Percent Change
• Benchmark Dose (BMD)
• Mean
• Geometric Mean
• Least Square Means (LSM)
• Geometric Mean Ratio
• Fecundability Ratio
• Adjusted r2
• Mean Ratio
• Prevalence Ratio (PR)
• If the effect estimate is a regression coefficient
(a beta or (3), select from the menu
"Regression Coefficient" rather than "Beta
Coefficient."
• If PFOS/PFOA was the outcome of interest
(e.g., study looked at the impact of a disease
on PFOS/PFOA level), please still extract the
data but make a note under the Results
Comments (11.19).
11.13 Effect Estimate
[Free-text]
• Only report the effect estimate from the
adjusted model. If there are multiple
adjustment sets, use the final model.
• Do not extract the reference group (1) for
results comparing exposure levels (i.e.,
A-103
-------
APRIL 2024
Question/Prompt Response Options Suggested Considerations
extract OR (for Q2 vs. Ql), but don't extract
the OR of 1 for the reference group Ql).
11.14 CI LCL: Confidence -
-
Interval - Lower
Confidence Limit
[Free-text]
11.15 CI UCL: Confidence
-
Internal - Upper
Confidence Limit
[Free-text]
11.16 SD or SE
• Enter the SD or SE if reported for the effect
[Free-text]
estimate.
• Leave blank if not reported.
11.17 p-value
• Enter the quantitative p-value if available (e.g.,
[Free-text]
"0.0001" or "<0.001")
o If the study/table only indicates that p-
value is not significant, enter "ns" for not
significant.
o If the p-value is not reported or does not
apply to the estimate being reported, leave
blank.
o If table footnote mentioned "*p < 0.05"
for the results with *, then enter <0.05. If
results do not have a * and no p-value was
reported, then leave blank.
o If the p-value is not reported and
text/methods mention significance level is
0.05, and:
¦ the text mentioned the specific result
is statistically significant, then
enter <0.05 (and make a note in the
Results Comments (11.19) which page
is this from).
¦ the text mentioned a result as not
statistically significant, then enter "ns"
(and make a note in the Results
Comments (11.19) which page is this
from).
• Make sure the p-value reported corresponds to
the regression coefficient being extracted.
Authors will occasionally report p-values for
other things such as the model fit.
• Other types of p-values such as interaction p-
values or trend p-values are reported, these can
be placed in Results Comments (11.19).
11.18 Covariates in Model
• If there are multiple adjustment sets, list
[Free-text]
covariates in the final model, but make a note
in the comment field on the main form (14).
that additional adjustment sets were available
for sensitivity analyses.
• List just the covariates, no need to add
"adjusted for..."
• Example: age. gender, race. SES
A-104
-------
APRIL 2024
Question/Prompt Response Options Suggested Considerations
11.19 Results Comments - • Enter the location of the extracted data (e.g.,
[Free-text] "Table 3" or "in-text p. 650").
• Enter any relevant p-values, such as
interaction p-values or trend p-values.
• Enter any additional details on the outcome
measurement or definition.
12 Select PFOS orPFOAif • PFOS
it was measured in the • PFOA
study but not analyzed
with health effects.
13 Correlations across the • Yes
• Note whether the main manuscript or the
included PFAS presented • No
supplemental material present a table or text
in paper or supplement.
describing the (Spearman) correlation
[Select one]
coefficients between concentrations of PFAS
included in the paper.
14 Comments -
• Briefly mention if effect modification is
Include brief description of
analyzed but not extracted (e.g., stratified
results provided in
analyses by race, by BMI categories, etc.).
supplemental materials but
Note: Stratification by sex and age should
not extracted (e.g.,
always be extracted.
stratified analyses,
• Do not need to specify how values below the
sensitivity analyses).
LOD were handled.
[Free-text]
• If data is presented by subgroup/strata (e.g.,
race) in the supplemental material, just note
that here. Note: Stratification by sex and
age should always be extracted.
• Briefly, describe any other supplemental
results (e.g., sensitivity analyses, etc.) here;
no need to list all confounders other models
adjusted for.
• Any outcome definitions if study-specific
(e.g., how was elevated ALT defined in a
study reporting ORs of elevated ALT).
Notes: QC = quality control; NOEL = no-observed-effect level; LOEL = lowest-observed-effect level; PECO = populations,
exposures, comparators, and outcomes; NHANES = National Health and Nutrition Examination Survey; PFAS = perfluoroalkyl
substances; PFOA = perfluorooctanoate acid; PFOS = perfluorooctane sulfonic acid; IQR = interquartile range; LOD = limit of
detection; LOQ = limit of quantification; Q2 = quarter 2; Q1 = quarter 1; In = natural log; SD = standard deviation; T2 = tertile
2S; T1 = tertile 1; PFC = ; Q3 = quarter 3; CI = confidence interval; SE = standard error; ns = not significant;
SES = socioeconomic status; BMt = body mass index; ALT = alanine transaminase.
A. 1.8.1 Epidemiological Study Design Definitions
Epidemiological studies with cross-sectional, cohort, case-control, ecological, or controlled trial
study designs were included. The study design definitions shown in Table A-39 were used
throughout full-text screening and data extraction for epidemiological studies.
Table A-39. Epidemiological Study Design Definitions
Study Design
Description
Cross-sectional
Exposure and outcome are examined at the same point in time in a defined study
population. Cannot determine if exposure came before or after outcome.
Cohort
A group of people is examined over time to observe a health outcome. Everyone
belongs to the same population (e.g., general U.S. population; an occupational group;
A-105
-------
APRIL 2024
Study Design Description
cancer survivors). All cohort studies (prospective or retrospective) consider exposure
data from before the occurrence of the health outcome.
Case-control Cases (people with the health outcome) and controls (people without the health
outcome) are selected at the start of a study. Exposure is determined and compared
between the two groups. A case-control study can be nested within a cohort.
Ecological The unit of observation is at the group level (e.g., zip code; census tract), rather than the
individual level. Ecological studies are often used to measure prevalence and incidence
of disease. Cannot make inferences about an individual's risk based on an ecological
study.
Controlled Trial Exposure is assigned to subject and then outcome is measured.
A-106
-------
APRIL 2024
A.1.9 Data Extraction for Animal Toxicological Studies
All animal toxicological studies identified as PECO-relevant after full-text screening in DistillerSR were eligible for data extraction.
As noted in the IRIS Handbook (U.S. EPA, 2022c), during data extraction, relevant results from each study are extracted to facilitate
organization, visualization, comparison, and analysis of findings and results. PECO-relevant animal toxicological studies that received
a medium or high confidence study quality evaluation rating were extracted.
Data extraction was carried out using a set of structured forms in HAWC (Table A-40). Studies slated for extraction were pre-screened
by an expert toxicologist who identified the relevant results. Extraction was performed by one reviewer and then independently
verified by at least one other reviewer for quality control. Any conflicts or discrepancies were resolved by discussion and confirmation
with a third reviewer.
Table A-40. HAWC Form Fields for Data Extraction of Animal Toxicological Studies
Questions/Prompts and Options
Suggested Considerations
1
Experiment
1.1
Name Field
[Free-text]
• Name should be short and simple. For example, '28-Day Oral' '2-Year Drinking Water', '1-Week Inhalation'.
• Reproductive/developmental if appropriate, then route of exposure (oral/inhalation), not number of generations
or acute/short-term/sub-chronic/chronic.
• If a study includes multiple experiments (e.g., multiple species, varied exposure durations), create separate
experiments for each.
1.2
Type Field
[Select one]
• For reproductive and/or developmental studies, select 'reproductive' or 'developmental' as appropriate
(recognizing that a study may contain both reproductive and developmental endpoints, but is typically defined
as one or the other based on design).
• In general, use reproductive when the study begins treatment prior to mating and continues through birth and
in some cases through a second generation. These studies will typically evaluate reproductive outcomes in the
dams (e.g., copulation and fertility indices, numbers of corpora lutea and implantation sites, pre- and post-
implantation loss). Use developmental when the exposure occurs during gestation and dams are sacrificed
prior to birth. These studies are typically focused on the pups and evaluate viability, developmental milestones,
and other growth and developmental effects in pups and primarily they are looking for abnormalities in the
pups.
• If reproductive or developmental are selected, indicate if there are data for more than one generation.
1.3
Chemical Name Field
[Free-text]
• Enter the preferred name of the chemical (i.e., PFOA or PFOS).
• Refer to the PECO statement in for a list of synonyms for each chemical.
1.4
Chemical Identifier (CAS) Field
[Free-text]
• Be sure to include the dashes in the CAS number.
• The CAS number for the chemical can be found in the PECO statement if they are not listed in the paper.
A-107
-------
APRIL 2024
Questions/Prompts and Options
Suggested Considerations
1.5
Chemical Source Field
[Free-text]
• If the chemical source is not provided by the authors, add in "Not Reported" to this field.
1.6
Chemical Purity Fields
[Checkbox]
• As a default, the 'Chemical purity available?' box will be checked. If the box is checked, entries for 'Purity
qualifier' and 'Chemical purity (%)' are required.
• Uncheck this box if chemical purity information is not available.
2
Animal Group
2.1
Name Field
[Free-text]
• Name should include sex, common strain name, and species (e.g., Male Sprague-Dawley Rats).
• For reproductive or developmental studies, include the generation before sex in title (e.g., F1 Male Sprague-
Dawley Rats or P0 Female C57 Mice).
• If a study combines male and female subjects into one group, use "Male and Female" (e.g., Male and Female
Sprague-Dawley Rats).
• If gender is unclear, do not mention (e.g., Sprague-Dawley Rats).
• Use the plural form for species (e.g., Rats, Mice).
2.2
Animal Source and Husbandry Field
[Free-text]
• Copy and paste details directly from the paper using quotation marks.
• If the authors do not provide the animal source, add in "Not Reported" to this field.
• For multigenerational reproductive or developmental studies, the animal group dosed might be the parental (or
P0) group. For example, a P0 female rat may be dosed during pregnancy and/or lactation, and developmental
effects are then measured in offspring - or Fi animals.
• For a multigenerational study, specify the 'Generation'.
3
Add Dosing Regime
3.1
Exposure Duration (Days) Field
[Free-text]
• Decimals are allowed, so a 4 h single day study can be represented as 0.17 d. However, decimals are likely not
needed for the PFOA/PFOS project since acute studies are not PECO relevant.
3.2
Exposure Duration (Text) Field
[Free-text]
• For all time units, use the following abbreviations: year = yr; month = mo; week = wk; day = d; hour = hr;
minute = min; second = sec.
• Eliminate unnecessary space between length of time and unit (i.e., "2wk" instead of "2 wk").
3.3
Description Field
[Free-text]
• Include dosing description from materials and methods. Be sure to use quotation marks around all text directly
copied/pasted from the paper.
• Include any information on how dosing solutions were prepared.
• Summarize any results the authors present on analytical work conducted to confirm dose, stability, and purity.
3.4
Dose-Groups Field
[Free-text]
• Dose groups should be listed lowest to highest (dose group 1=0 mg/kg-d).
• For visualization purposes dose units need to be in mg/kg-d. For studies that provide the units, please use those
for extraction purposes.
• For dietary or drinking water studies, if they provide BOTH concentration of the dose formulation (e.g., ppm)
AND doses as mg/kg-d, please extract both.
A-108
-------
APRIL 2024
Questions/Prompts and Options Suggested Considerations
• For dietary or drinking water studies that ONLY provide the dose concentration, enter the dose concentrations
as reported in the study and then utilize the conversions spreadsheet to convert the dosage into mg/kg-day
(note that mg/kg body weight/day is the same as mg/kg-d so you just need to use the mg/kg-d).
• If PFOA/PFOS are administered as salts and the doses are presented as salts of PFOA/PFOS, please contact
senior-level extractors before using the conversion spreadsheet.
• If converting doses, add in "Data extractor calculated [PFOS/PFOA] equivalent doses for mg/kg-day" into the
"Description" box.
• When defining the dosing regime for a multigenerational experiment, creating a new dosing regime may not be
needed; instead specify the existing dosing regime of the Po (dosed during gestation and/or lactation).
• A new dosing regime may be needed if offspring were exposed after weaning and, if applicable, acknowledge
parental exposure in the 'Description' field on the 'Dosing regime' page.
• If the authors provide internal measurements of PFOS/PFOA in any tissue, add this information in as an
additional dose group using the mean tissue levels as the value and the tissue as part of the dose units (e.g.,
mg/kg bone, ppm brain).
4
Endpoints (General)
4.1
Endpoint Name Field
[Free-text]
• Name should not include descriptive information captured in other fields within HAWC such as sex, strain,
species, duration, route, etc.
• Include common abbreviation in parenthesis if applicable.
• Endpoint detail should be added after main endpoint, ex. "Body Weight, Fetal" NOT "Fetal Body Weight."
• In general, specific endpoint names are used except for general categories such as 'Clinical Observations' or
histopathology (e.g., 'Kidney Histopathology'), which may comprise a number of observational endpoints.
• Examples: Liver Weight, Relative; Triiodothyronine (T3)
4.2
System Field
[Free-text]
• Represents the appropriate system for the endpoint.
• Examples: Hepatic; Endocrine
4.3
Organ (and Tissue) Field
[Free-text]
• Represents the appropriate organ or tissue for the endpoint.
• Examples: Liver; Thyroid
4.4
Effect and Effect Subtype Fields
[Free-text]
• Represents the appropriate system for the endpoint.
• Examples: Hepatic; Endocrine
4.5 Observation Time Fields • The 'Observation time' text field is included in visualizations and should be filled in; the 'Observation time'
[Free-text] numeric field and 'Observation time units' can be left blank.
• For all time units, use the following abbreviations: year = yr; month = mo; week = wk; day = d; hour = hr
• Eliminate unnecessary space between length of time and unit (i.e., "2wk" instead of "2 wk").
• Example: 2yr; 6hr; 45d; 90 min
• For developmental and reproductive studies, specify observation time in terms of development (e.g., GD 16,
PNDo).
A-109
-------
APRIL 2024
Questions/Prompts and Options
Suggested Considerations
4.6
Values Estimated Field
[Free-text]
• If data was extracted from a figure into HAWC using a measured ruler, check this box.
• For data requiring a digital ruler, use the WebPlotDigitizer tool: https://apps.automeris.io/wpd/.
• If there are multiple time points, extract only the latest time point (i.e., end of treatment) or if the last time
point is not significant and an earlier time point is, extract the earlier time point (this information should be
provided in the data to extract instructions, but this is the general rule in case there are no instructions
provided).
• Provide additional information in the results comment box to make note of what happened at other timepoints
that were not extracted.
4.7
Litter Effects Field
[Free-text]
• If the experiment type has been identified as either 'reproductive' or 'developmental', the 'Litter effects' will
be required, and a choice other than 'not applicable' must be selected.
4.8
Dataset Type Field
[Free-text]
• Select the appropriate dataset type for the endpoint. In general, 'Dataset type' is continuous except for
incidence data, which is dichotomous.
4.9
NOAEL and LOAEL Fields
[Free-text]
• Be sure to enter the significance level (e.g., 0.05) for significant results as well as NOAEL/LOAEL.
• The NOAEL is the highest dose at which there was not an observed toxic or adverse effect. If the LOAEL is
the lowest (non-control) dose, then NOAEL should be , not 0.
• The LOAEL is the lowest dose at which there was an observed toxic or adverse effect. These fields are critical
to the visualizations. If there is no LOAEL, leave as .
• In cases where the study authors did not conduct statistical tests, use the study authors conclusions to indicate
where effects occur. Just make sure to note in the results comments that these were based on author
conclusions and no statistical testing was conducted.
4.10 Statistical Test Field
[Free-text]
• If the statistical test is not provided in the study, add "Not Reported" to the text field.
4.11
Results Notes Field
[Free-text]
• If needed, copy and paste details into this field using quotation marks. Although the methods text field can
describe all methods used, results comments should be more endpoint specific.
5
Endpoint (Dummy Variables)
Data to be extracted using dummy
variables for the following reasons:
• Results that are qualitatively discussed
in the text, but actual data are not
provided.
• For instances where study authors
specify that only the significant effects
are described - and certain endpoints
are then not discussed - assume that no
change occurred in these endpoints.
Create dummy variables for all
endpoints stated to be measured with
• For endpoints for which no quantitative data are provided, create the endpoint as described above with the
exceptions below.
• 'Dataset type' is dichotomous or continuous based on the data type if there were data available.
• For 'Response units,' use whatever units correspond to the effect for which you are creating the dummy
variable (e.g., 'incidence' for histopathology observations, 'grams' for body weight)
• Under 'Dose-response data', fill in with a dummy variable. Use 0 to indicate no change from control, a 1 to
indicate an increase from control and a -1 to indicate a decrease from the control.
• 'Significance Level' should be populated if the author indicates significance. Otherwise, 'Significance Level'
is left blank.
• Multiple clinical observations can be grouped together into a single endpoint.
• Example: create an endpoint for clinical observations and add dummy variables to indicate no effect.
A-110
-------
APRIL 2024
Questions/Prompts and Options
Suggested Considerations
the assumption if they are not
discussed they were not significant and
make sure to document this in the
results comments field.
• If an endpoint is discussed in the
methods, but there is no mention at all
in the results (even to indicate that
only significant effects were reported),
then create an endpoint only and do
not extract any data. In this case,
uncheck the 'data reported' and 'data
extracted' boxes on the endpoint page.
• Organs/tissues that were examined for
histopathological changes, but no
changes were noted.
• Clinical observations in which
multiple clinical signs or general
observations are grouped together.
' If a single endpoint called "Clinical Observation," create the dummy variables above using all 0 with nothing
tagged as significant.
¦ Or if there was an effect, still create a single endpoint called "Clinical Observation" and then put a 1 at the
dose where the effects were observed and then in the results comment field indicate the effects that were
observed. This would be common in reproductive and developmental studies; indicate if there were "Clinical
Observations in Dams" and where they occurred but didn't want to have a separate endpoint for each
observation.
' Example: for any organ listed but not specified any lesions to extract, create a histopathology endpoint and
create a dummy variable to indicate no treatment-related effect.
' Create an endpoint for each organ (e.g., Liver Histopathology, Kidney Histopathology, Uterus
Histopathology), and create the dummy variables described above using all 0 with nothing tagged as
significant.
¦ Whenever using dummy variables instead of actual data, make sure to note in the results comment text box
that the data are dummy variables using the standard language given in the instructions in HAWC under the
'Results notes' box.
Notes: NOAEL = no-observed-adverse-effect level; LOAEL = lowest-observed-adverse-effect level; CAS = Chemical Abstracts Service.
A-lll
-------
APRIL 2024
A. 1.10 Evidence Synthesis and Integration
For the purposes of this assessment, evidence synthesis and integration are considered distinct
but related processes. For each assessed health effect, the evidence syntheses provide a summary
discussion of each body of evidence considered in the review, considering the conclusions from
the individual study quality evaluations. Syntheses of the evidence for human and animal health
effects are based primarily on studies of high and medium confidence; low confidence results
were given less weight compared with high or medium confidence results during evidence
synthesis and integration. However, in certain instances (i.e., for health outcomes for which few
or no studies with higher confidence are available), low confidence studies might be used to help
evaluate consistency or if the study designs of the low confidence studies address notable
uncertainties in the set of high or medium confidence studies on a given health effect.
The available human and animal evidence pertaining to the potential health effects of PFOS were
synthesized separately, and a summary discussion of the available evidence was developed for
each evidence stream. For the five priority health outcomes, mechanistic evidence was also
considered in the development of each synthesis. Strength-of-evidence judgments were made for
each health outcome within each evidence stream (i.e., human or animal) using standard
terminology (i.e., robust, moderate, slight, indeterminate) and definitions according to the
framework described in the IRIS Handbook and outlined in Table A-41 and Table A-42.
Following evidence synthesis, the evidence for humans and animals was integrated for each
health outcome. The evidence integration was conducted following the guidance outlined in the
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA (anionic and acid
forms) IRIS Assessments (U.S. EPA, 2020b). Integrated judgements were drawn across all lines
of evidence for each assessed health outcome as to whether, and to what extent, the evidence
supports that exposure to PFOS has the potential to be hazardous to humans. The evidence
integration provided a summary of the causal interpretations from the available studies, as well
as mechanistic evidence for the five priority health outcomes. Mechanistic evidence was
organized by signaling pathway or other categories (e.g., key characteristics of carcinogens) as
relevant to each outcome. The integrated judgments are developed through structured review of
the evidence against an established set of considerations for causality. These considerations
include risk of bias, sensitivity, consistency, strength (effect magnitude) and precision, biological
gradient/dose-response, coherence, and mechanistic evidence related to biological plausibility.
The evidence integration involved an overall judgment on whether there was sufficient evidence
or insufficient evidence for each potential human health effect and an evidence basis rationale.
During evidence integration, a structured and documented process was used, as follows:
• Summarize human and animal health effect studies in parallel but separately, using the set
of considerations for causality first introduced by Austin Bradford Hill (Hill, 1965) and
relevant mechanistic evidence (or mode of action (MOA) understanding).
• Identify strength of the human and animal health evidence in light of inferences across
evidence streams.
• Summarize judgment as to whether the available evidence base for each potential health
outcome as a whole indicates that PFOS exposure has the potential to cause adverse health
effects in humans (see Table A-43) ("evidence demonstrates," "evidence indicates
(likely)," "evidence suggests," "evidence is inadequate," or "strong evidence supports no
effect").
A-112
-------
APRIL 2024
The decision points within the structured evidence integration process are summarized in an
evidence profile table for each assessed health effect.
Table A-41. Framework for Strength-of-Evidence Judgments for Epidemiological Studies3
Strength-of-
Evidence
Judgment
Description
Robust A set of high- or medium-confidence studies reporting an association between the exposure and
(©©©) the health outcome, with reasonable confidence that alternative explanations, including chance,
bias, and confounding, can be ruled out across studies. The set of studies is primarily consistent,
with reasonable explanations when results differ; and an exposure response gradient is
demonstrated. Supporting evidence, such as associations with biologically related endpoints in
human studies (coherence) or large estimates of risk or severity of the response, may help to rule
out alternative explanations. Similarly, mechanistic evidence from exposed humans may serve to
address uncertainties relating to exposure-response, temporality, coherence, and biological
plausibility (i.e., providing evidence consistent with an explanation for how exposure could cause
the health effect based on current biological knowledge) such that the totality of human evidence
supports this judgment.
Moderate • Multiple studies showing generally consistent findings, including at least one high or medium
(©©O) confidence study and supporting evidence, but with some residual uncertainty due to potential
chance, bias, or confounding (e.g., effect estimates of low magnitude or small effect sizes given
what is known about the endpoint; uninterpretable patterns with respect to exposure levels).
Associations with related endpoints, including mechanistic evidence from exposed humans, can
address uncertainties relating to exposure response, temporality, coherence, and biological
plausibility, and any conflicting evidence is not from a comparable body of higher confidence,
sensitive studies
• A single high- or medium-confidence study demonstrating an effect with one or more factors
that increase evidence strength, such as: a large magnitude or severity of the effect, a dose-
response gradient, unique exposure or outcome scenarios (e.g., a natural experiment), or
supporting coherent evidence, including mechanistic evidence from exposed humans. There are
no comparable studies of similar confidence and sensitivity providing conflicting evidence, or if
there are, the differences can be reasonably explained (e.g., by the population or exposure levels
studied)
Slight One or more studies reporting an association between exposure and the health outcome, where
(©OO) considerable uncertainty exists:
• A body of evidence, including scenarios with one or more high- or medium-confidence studies
reporting an association between exposure and the health outcome, where either (1) conflicting
evidence exists in studies of similar confidence and sensitivity (including mechanistic evidence
contradicting the biological plausibility of the reported effects), a (2) a single study without a
factor that increases evidence strength (factors described in moderate), OR (3) considerable
methodological uncertainties remain across the body of evidence (typically related to exposure
or outcome ascertainment, including temporality), AND there is no supporting coherent
evidence that increases the overall evidence strength.
• A set of only low confidence studies that are largely consistent.
• Strong mechanistic evidence in well-conducted studies of exposed humans (medium or high
confidence) or human cells, in the absence of other substantive data, where an informed
evaluation has determined that the data are reliable for assessing the health effect of interest and
the mechanistic events have been reasonably linked to the development of that health effect.
Indeterminate • No studies in humans or well-conducted studies of human cells.
(OOO) • Situations when the evidence is highly inconsistent and primarily of low confidence.
• May include situations with medium or high confidence studies, but unexplained heterogeneity
exists (in studies of similar confidence and sensitivity), and there are additional outstanding
concerns such as effect estimates of low magnitude, uninterpretable patterns with respect to
A-113
-------
APRIL 2024
Strength-of-
Evidence Description
Judgment
exposure levels, or uncertainties or methodological limitations that result in an inability to
discern effects from exposure.
• A set of largely null studies that does not meet the criteria for compelling evidence of no effect,
including evidence bases with inadequate testing of susceptible populations and lifestages.
Several ///"//-confidence studies showing null results (for example, an odds ratio of 1.0), ruling
out alternative explanations including chance, bias, and confounding with reasonable confidence.
Each of the studies should have used an optimal outcome and exposure assessment and adequate
sample size (specifically for higher exposure groups and for susceptible populations). The set as a
whole should include the full range of levels of exposures that human beings are known to
encounter, an evaluation of an exposure response gradient, and an examination of at-risk
populations and lifestages.
Notes:
aTable slightly adapted from Table 11-3 in the IRIS Handbook.
Table A-42. Framework for Strength-of-Evidence Judgments for Animal Toxicological
Studies3
Compelling
evidence of no
effect ( )
Strength-of-
Evidence
Judgment
Description
Robust A set of high- or /;/t'c//'z«/-confidcncc studies with consistent findings of adverse or
(©©©) toxicologically significant effects across multiple laboratories, exposure routes, experimental
designs (e.g., a subchronic study and a two-generation study), or species; and the experiments
reasonably rule out the potential for nonspecific effects to have caused the effects of interest. Any
inconsistent evidence (evidence that cannot be reasonably explained based on study design or
differences in animal model) is from a set of experiments of lower confidence or sensitivity. To
reasonably rule out alternative explanations, multiple additional factors in the set of experiments
exist, such as: coherent effects across biologically related endpoints; an unusual magnitude of
effect, rarity, age at onset, or severity; a strong dose-response relationship; or consistent
observations across animal lifestages, sexes, or strains. Similarly, mechanistic evidence (e.g.,
precursor events linked to adverse outcomes) in animal models may exist to address uncertainties
in the evidence base such that the totality of animal evidence supports this judgment.
Moderate • At least one high- or medium-confidence study with supporting information increasing the
(©©O) strength of the evidence. Although the results are largely consistent, notable uncertainties
remain. However, in scenarios when inconsistent evidence or evidence indicating nonspecific
effects exist, it is not judged to reduce or discount the level of concern regarding the positive
findings, or it is not from a comparable body of higher confidence, sensitive studies. The
additional support provided includes either consistent effects across laboratories or species;
coherent effects across multiple related endpoints; an unusual magnitude of effect, rarity, age at
onset, or severity; a strong dose-response relationship; or consistent observations across
exposure scenarios (e.g., route, timing, duration), sexes, or animal strains. Mechanistic evidence
in animals may serve to provide this support or otherwise address residual uncertainties.
• A single high or medium confidence experiment demonstrating an effect in the absence of
comparable experiment(s) of similar confidence and sensitivity providing conflicting evidence,
namely evidence that cannot be reasonably explained (e.g., by respective study designs or
differences in animal model).
Slight • Scenarios in which there is a signal of a possible effect, but the evidence is conflicting or weak:
(©OO) • A body of evidence, including scenarios with one or more high or medium confidence
experiments reporting effects but without supporting or coherent evidence (see description in
moderate) that increases the overall evidence strength, where conflicting evidence exists from a
A-114
-------
APRIL 2024
Strength-of-
Evidence
Judgment
Description
Indeterminate
(OOO)
set of sensitive experiments of similar or higher confidence (including mechanistic evidence
contradicting the biological plausibility of the reported effects).
• A set of only low confidence experiments that are largely consistent.
• Strong mechanistic evidence in well-conducted studies of animals or animal cells, in the
absence of other substantive data, where an informed evaluation has determined the assays are
reliable for assessing the health effect of interest and the mechanistic events have been
reasonably linked to the development of that health effect.
• No animal studies or well-conducted studies of animal cells.
• The available models (not considering human relevance) or endpoints are not informative to the
hazard question under evaluation.
• The evidence is inconsistent and primarily of low confidence.
• May include situations with medium or high confidence studies, but there is unexplained
heterogeneity and additional concerns such as small effect sizes (given what is known about the
endpoint) or a lack of dose-dependence.
• A set of largely null studies that does not meet the criteria for compelling evidence of no effect.
A set of high confidence experiments examining a reasonable spectrum of endpoints relevant to a
type of toxicity that demonstrate a lack of biologically significant effects across multiple species,
both sexes, and a broad range of exposure levels. The data are compelling in that the experiments
have examined the range of scenarios across which health effects in animals could be observed,
and an alternative explanation (e.g., inadequately controlled features of the studies' experimental
designs; inadequate sample sizes) for the observed lack of effects is not available. The
experiments were designed to specifically test for effects of interest, including suitable exposure
timing and duration, post exposure latency, and endpoint evaluation procedures, and to address
potentially susceptible populations and lifestages. Mechanistic data in animals (in vivo or in
vitro) that address the above considerations or that provide information supporting the lack of an
association between exposure and effect with reasonable confidence may provide additional
support such that the totality of evidence supports this judgment.
Compelling
evidence of no
effect ( )
Notes:
aTable slightly adapted from Table 11-4 in the IRIS Handbook.
Table A-43. Evidence Integration Judgments for Characterizing Potential Human Health
Effects in the Evidence Integration3
Evidence
integration
judgment level
Explanation and example scenarios
Evidence
demonstrates
Evidence
indicates
(likely)
A strong evidence base demonstrating that [chemical] exposure causes [health effect] in humans
• For when there is robust human evidence supporting an effect
• Could also be used when there is moderate human evidence and robust animal evidence if there
is strong mechanistic evidence that MOA(s) or key precursors identified in animals are
expected to occur and progress in humans
An evidence base that indicates that [chemical] exposure likely causes [health effect] in humans,
although there may be outstanding questions or limitations.
• Used if there is robust animal evidence supporting an effect and slight or indeterminate human
evidence, or with moderate human evidence when strong mechanistic evidence is lacking
• Could also be used with moderate human evidence supporting an effect and slight or
indeterminate animal evidence, or with moderate animal evidence supporting an effect and
slight or indeterminate human evidence. In these scenarios, any uncertainties in the moderate
evidence are not sufficient to substantially reduce confidence in the reliability of the evidence,
A-115
-------
APRIL 2024
Evidence
integration
judgment level
Explanation and example scenarios
Evidence
suggests
or mechanistic evidence in the slight or indeterminate evidence base (e.g., precursors) exists to
increase confidence in the reliability of the moderate evidence
A decision between "evidence indicates" and "evidence suggests" considers the extent to which
findings are coherent or biologically consistent across lines of evidence streams, and may
incorporate other supplemental evidence (e.g., structure-activity data; chemical class information)
An evidence base that suggests that [chemical] exposure may cause [health effect] in humans, but
there are very few studies that contributed to the evaluation, the evidence is weak or conflicting,
and/or the methodological conduct of the studies is poor.
• Used if there is slight human evidence and indeterminate or slight animal evidence
• Used with slight animal evidence and indeterminate or slight human evidence
• Could also be used with moderate human evidence and slight or indeterminate animal evidence,
or with moderate animal evidence and slight or indeterminate human evidence. In these
scenarios, there are outstanding issues regarding the moderate evidence that substantially
reduced confidence in the reliability of the evidence, or mechanistic evidence in the slight or
indeterminate evidence base (e.g., null results in well-conducted evaluations of precursors)
exists to decrease confidence in the reliability of the moderate evidence
• When there is general scientific understanding of mechanistic events that result in a health
effect, this judgment level could also be used if there is strong mechanistic evidence that is
sufficient to highlight potential human toxicity in the absence of informative conventional
studies in humans or in animals
This conveys either a lack of information or an inability to interpret the available evidence for
[health effect]. On an assessment-specific basis, a single use of this "evidence inadequate"
judgment might be used to characterize the evidence for multiple health effect categories.
• Used if there is indeterminate human and animal evidence
• Used if there is slight animal evidence and compelling evidence of no effect human evidence
• Could also be used with slight or robust animal evidence and indeterminate human evidence if
strong mechanistic information indicated that the animal evidence is unlikely to be relevant to
humans
Extensive evidence across a range of populations and exposure levels has identified no
effects/associations. This scenario requires a high degree of confidence in the conduct of
individual studies, including consideration of study sensitivity, and comprehensive assessments
of the endpoints and lifestages of exposure potentially relevant to the heath effect of interest.
• Used if there is compelling evidence of no effect in human studies and compelling evidence of
no effect or indeterminate animal evidence
• Also used if there is indeterminate human evidence and compelling evidence of no effect
animal evidence in models judged as relevant to humans
• Could also be used with compelling evidence of no effect in human studies and moderate or
robust animal evidence if strong mechanistic information indicated that the animal evidence is
unlikely to be relevant to humans
Evidence
inadequate13
Strong evidence
supports no
effect
Notes: MOA = mode of action.
aTable adapted from Table 11-5 in the IRIS Handbook.
b An "evidence inadequate" judgment is not a determination that the chemical does not cause the indicated human health
effect(s), but rather an indication that the available evidence is insufficient to reach a judgment.
A.1.10.1 Epidemiological Studies Included From the 2016 PFOS HESD
For the five priority health outcomes (i.e., developmental, immune, hepatic, cardiovascular and
cancer), epidemiological studies identified and reviewed in the 2016 PFOS HESD were included
in the evidence synthesis, including discussion of study quality considerations, according to the
recommendations from the SAB. Inferences drawn from included studies from the 2016 PFOS
A-116
-------
APRIL 2024
HESD were considered in drawing health effects conclusions and were incorporated into the
evidence profile tables
For all non-priority health outcomes, epidemiological studies identified and reviewed in the 2016
PFOS HESD were included in the evidence syntheses in summary paragraphs describing
previously reached conclusions for each health outcome. Study quality was considered, but
domain-based, structured study quality evaluations were not performed for 2016 PFOS HESD
studies. Inferences drawn from evidence in the current literature search were compared with the
results described from 2016 studies.
A. 1.10.2 Epidemiological Studies Excluded From Synthesis
Some epidemiological studies were not included in the evidence synthesis narrative if they
included overlapping results (e.g., overlapping NHANES studies). Studies reporting results from
the same cohort with the same health outcome were considered overlapping evidence, and these
studies were not discussed in the synthesis narrative to avoid duplication or overrepresentation of
results from the same group of participants. When participants from the same cohort were
included in more than one eligible study, the study with the largest number of participants was
included in the evidence synthesis narrative. In general, to best gauge consistency and magnitude
of reported associations, EPA focused on the most accurate and most prevalent measures. In
some cases, such as developmental outcomes, studies on the same population providing more
accurate outcome measures (e.g., birthweight and birth length for fetal growth restriction) were
given preference over studies providing less accurate outcome measures (e.g., ponderal index for
fetal growth restriction). Overlapping studies were included in study quality figures.
Meta-analyses were considered during evidence integration as support of consistent effects
across studies. Details of the identified meta-analyses and assessment implications are
summarized in Section A.2.
A. 1.11 Dose-Response Assessment: Selecting Studies and
Quantitative Analysis
As noted in the IRIS Handbook, selection of studies and endpoints for dose-response assessment
involves considerations of the data that build from "judgments" and decisions made during
earlier steps of the systematic review and assessment process. EPA guidance and support
documents that describe data requirements and other considerations for dose-response modeling
include EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012), Review of the Reference
Dose and Reference Concentration Processes (U.S. EPA, 2002), Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a), and Supplemental Guidance for Assessing Susceptibility from
Early-Life Exposure to Carcinogens (U.S. EPA, 2005b).
Dose-response assessments are performed for both noncancer and cancer oral health hazards, if
supported by existing data. For noncancer hazards, an oral RfD will be derived when possible.
An RfD is an estimate, with uncertainty spanning perhaps an order of magnitude, of an exposure
to the human population (including susceptible subgroups) that is likely to be without an
appreciable risk of deleterious health effects over a lifetime (U.S. EPA, 2002). Reference values
are not predictive risk values; that is, they provide no information about risks at higher or lower
exposure levels.
A-117
-------
APRIL 2024
For cancer hazards, a CSF will be derived to estimate human cancer risk when low-dose linear
extrapolation for cancer effects is supported. A CSF is a plausible upper bound lifetime cancer
risk from chronic ingestion of a chemical per unit of mass consumed per unit body weight per
day (mg/kg-day). In contrast to RfDs, CSFs can be used in conjunction with exposure
information to predict cancer risk at a given dose.
The derivation of reference values will depend on the conclusions drawn during previous steps of
this protocol. Specifically, EPA will attempt dose-response assessments for noncancer outcomes
when the evidence integration judgements indicate stronger evidence of hazard (i.e., evidence
demonstrates and evidence indicates integration judgements). Quantitative analyses are generally
not attempted for other evidence integration conclusions. Similarly, EPA will attempt dose-
response assessments for cancer outcomes for chemicals that are classified as Carcinogenic or
Likely to be Carcinogenic to Humans. When there is Suggestive Evidence of Carcinogenic
Potential to Humans, EPA generally does not conduct dose-response assessment unless a well-
conducted study is available and a quantitative analysis is deemed useful.
A.1.11.1 Study Selection
Evidence synthesis and integration enabled identification of the health outcomes with the
strongest weight of evidence supporting causal relationships between PFOS exposure and
adverse health effects, as well as the most sensitive cancer and noncancer endpoints within those
health outcomes. Dose-response modeling was performed for endpoints within health outcomes
with data warranting evidence integration conclusions of evidence demonstrates and evidence
indicates (likely) for noncancer endpoints and carcinogenicity descriptors of Carcinogenic to
Humans and Likely to be Carcinogenic to Humans. Human epidemiological and animal
toxicological studies that were consistent with the overall weight of evidence for a specific
endpoint were considered for dose-response. Additionally, for human evidence, all high or
medium confidence studies pertaining to a specific endpoint were considered; for animal
evidence, only animal toxicological studies with at least two PFOS exposure groups that were of
high or medium confidence were considered. Relevance of the endpoint or species reported by
animal toxicological studies to human health effects was also considered. When multiple
endpoints for a health outcome are available, endpoints are selected for dose-response analysis
based on rationale describing how the endpoint is representative of the broader health outcome
(U.S. EPA, 2022c). Studies were evaluated for use in POD derivation following considerations
described in Table 7-2 (Table A-44) of the IRIS Handbook (U.S. EPA, 2022c). These attributes
support a more complete characterization of the shape of the exposure-response curve and
decrease the uncertainty in the associated exposure-response metric (e.g., RfD) by reducing
statistical uncertainty in the POD and minimizing the need for low-dose extrapolation. Some
important considerations include:
• human data are preferred over animal data to eliminate interspecies extrapolation
uncertainties,
• animal species known to respond similarly to humans are preferred over studies of other
species,
• high or medium confidence studies are preferred over low confidence studies,
• chronic or subchronic studies, or studies encompassing a sensitive lifestage (i.e.,
gestational) are preferred for the derivation of chronic toxicity values over acute studies,
A-118
-------
APRIL 2024
• studies with a design or analysis that addresses relevant confounding for a given outcome
are preferred,
• human studies providing the most updated data on a population are preferred over prior
publications,
• and studies reporting all necessary data (e.g., total population or quartile exposure
concentrations) for dose-response analysis are preferred.
The number of studies considered for toxicity value derivation was reduced based on these
considerations and others described in EPA (U.S. EPA, 2022c, 2012).
A-119
-------
APRIL 2024
Table A-44. Attributes used to evaluate studies for derivation of toxicity values (adapted from ORD Staff Handbook for
Developing IRIS Assessments Table 7-2)
Study Attributes
Considerations
Human studies
Animal studies
Study confidence
Rationale for choice of
species
High or medium confidence studies are highly preferred over low confidence studies. The selection of low confidence studies should
include an additional explanatory justification (e.g., only low confidence studies had adequate data for toxicity value derivation). The
available high and medium confidence studies are further differentiated on the basis of the study attributes below, as well as a
reconsideration of the specific limitations identified and their potential impact on dose-response analyses.
Human data are preferred over animal data to eliminate
interspecies extrapolation uncertainties (e.g., in
pharmacodynamics, dose-response pattern in relevant dose
range, relevance of specific health outcomes to humans).
Animal studies provide supporting evidence when adequate human
studies are available, and they are considered the studies of primary
interest when adequate human studies are not available. For some
hazards, studies of particular animal species known to respond
similarly to humans would be preferred over studies of other species.
Relevance Exposure
of exposure route
paradigm
Exposure
durations
Exposure
levels
Subject selection
Studies involving human environmental exposures (oral,
inhalation).
Studies by a route of administration relevant to human environmental
exposure are preferred. A validated pharmacokinetic model can also
be used to extrapolate across exposure routes.
When developing a chronic toxicity value, chronic or subchronic studies are preferred over studies of acute exposure durations.
Exceptions exist, such as when a susceptible population or lifestage is more sensitive in a particular time window (e.g., developmental
exposure).
Exposures near the range of typical environmental human exposures are preferred. Studies with a broad exposure range and multiple
exposure levels are preferred to the extent that they can provide information about the shape of the exposure-response relationship (see
the EPA Benchmark Dose Technical Guidance, §2.1.1) and facilitate extrapolation to more relevant (generally lower) exposures.
Studies that provide risk estimates in the most susceptible groups are preferred.
Controls for possible
confounding
Studies with a design (e.g., matching procedures, blocking) or analysis (e.g., covariates or other procedures for statistical adjustment)
that adequately address the relevant sources of potential critical confounding for a given outcome are preferred.
Measurement of
exposure
Health outcome(s)
Studies that can reliably distinguish between levels of Studies providing actual measurements of exposure (e.g., analytical
exposure in a time window considered most relevant for inhalation concentrations vs. target concentrations) are preferred,
development of a causal effect are preferred. Exposure Relevant internal dose measures might facilitate extrapolation to
assessment methods that provide measurements at the level of humans, as would availability of a suitable animal PBPK model in
the individual and that reduce measurement error are preferred, conjunction with an animal study reported in terms of administered
Measurements of exposure should not be influenced by exposure.
knowledge of health outcome status.
Studies that can reliably distinguish the presence or absence (or degree of severity) of the outcome are preferred. Outcome
ascertainment methods using generally accepted or standardized approaches are preferred.
Studies with individual data are preferred in general. For example, individual data allow you to characterize experimental variability
more realistically and to characterize overall incidence of individuals affected by related outcomes (e.g., phthalate syndrome).
A-120
-------
APRIL 2024
Considerations
Study Attributes
Human studies Animal studies
Among several relevant health outcomes, preference is generally given to those outcomes with less concerns for indirectness or with
greater biological significance.
Study size and design Preference is given to studies using designs reasonably expected to have power to detect responses of suitable magnitude This does
not mean that studies with substantial responses, but low power would be ignored, but that they should be interpreted in light of a
confidence interval or variance for the response. Studies that address changes in the number at risk (through decreased survival, loss to
follow-up) are preferred.
Notes: PBPK = physiologically based pharmacokinetic.
A-121
-------
APRIL 2024
A. 1.11.2 Approach to POD and Candidate RfD Derivation for Noncancer
Health Outcomes
The current recommended EPA human health risk assessment approach for noncancer POD
derivation described in EPA's A Review of the Reference Dose and Reference Concentration
Processes includes selection of a benchmark response (BMR), analysis of dose and response
within the observed dose range, followed by extrapolation to lower exposure levels (U.S. EPA,
2002). For noncancer health outcomes, EPA performed dose-response assessments to define
PODs, including low-dose extrapolation, when feasible, and applied uncertainty factors (UFs) to
those PODs to derive candidate RfDs. An RfD is an estimate, with uncertainty spanning perhaps
an order of magnitude, of an exposure to the human population (including susceptible subgroups)
that is likely to be without an appreciable risk of deleterious health effects over a lifetime (U.S.
EPA, 2002). For PFOS, multiple candidate RfDs were derived within a health outcome as
described in Section 4 of the Toxicity Assessment (U.S. EPA, 2024).
Considerations for BMR selection are discussed in detail in EPA's Benchmark Dose Technical
Guidance (U.S. EPA, 2012). For the derivation of RfDs, the BMR selected should correspond to
a low or minimal level of response in a population for the outcome of interest and is generally
the same across assessments, though the BMR could change over time based on new data or
developments. The following general recommendations for BMR selection were considered for
this assessment:
• For dichotomous data (e.g., presence or absence), a BMR of 10% extra risk is generally
used for minimally adverse effects. Lower BMRs (5% or lower) can be selected for severe
or frank effects. For example, developmental effects are relatively serious effects, and
BMDs derived for these effects could use a 5% extra risk BMR. Developmental
malformations considered severe enough to lead to early mortality could use an even
lower BMR (U.S. EPA, 2022c, 2012).
• For continuous data, a BMR is ideally based on an established definition of biologic
significance in the effect of interest. In the absence of such a definition, a difference of
one standard deviation (SD) from the mean response of the control mean is often used and
one-half the standard deviation is used for more severe effects. Note that the standard
deviation used should reflect underlying variability in the outcome to the extent possible
separate from variability attributable to laboratory procedures, etc. (U.S. EPA, 2022c,
2012).
Deviations of these recommendations, if any, are described in Section 4 of the Toxicity
Assessment (U.S. EPA, 2024).
For PFOS animal toxicological studies, EPA attempted benchmark dose (BMD) modeling on all
studies considered for dose-response to refine the POD. BMD modeling was performed after
converting the administered dose reported by the study to an internal dose using a
pharmacokinetic model (see Toxicity Assessment, (U.S. EPA, 2024).). This approach resulted in
dose levels corresponding to specific response levels near the low end of the observable range of
the data and identified the lower limits of the BMDs (BMDLs) which serve as potential PODs
(U.S. EPA, 2012). EPA used the publicly available Benchmark Dose Software (BMDS) program
developed and maintained by EPA (https://www.epa.eov/bmds). BMDS fits mathematical
A-122
-------
APRIL 2024
models to the data and determines the dose (i.e., BMD) that corresponds to a predetermined level
of response (i.e., benchmark response or BMR). For dichotomous data, the BMR is typically set
at either 5% or 10% above the background or the response of the control group. For continuous
data, a BMR of one-half or one standard deviation from the control mean is typically used when
there are no outcome-specific data to indicate what level of response is biologically significant
(U.S. EPA, 2012). For dose-response data for which BMD modeling did not produce an adequate
model fit, a no-observed-adverse-effect level (NOAEL) or lowest-observed-adverse-effect level
(LOAEL) was used as the POD. However, a POD derived using a BMD approach typically
provides a higher level of confidence in the conclusions for any individual case, as the BMDL
takes into account all the data from the dose-response curve, incorporates the evaluation of the
uncertainty in the BMD, and is related to a known and predefined potential effect size (i.e., the
BMR) (U.S. EPA, 2022b, 2012). For noncancer endpoints, there were several factors considered
when selecting the final model and BMD/BMDL, including the type of measured response
variable (i.e., dichotomous or continuous), experimental design, and covariates (U.S. EPA,
2012). However, as there is currently no prescriptive hierarchy, selection of model types was
often based on the goodness-of-fit and was judged based on the x2 goodness-of-fit p-value (p >
0.1), magnitude of the scaled residuals in the vicinity of the BMR, and visual inspection of the
model fit. Thq Benchmark Dose Technical Guidance provides a "BMD Decision Tree" to assist
in model selection (U.S. EPA, 2012). See Appendix E for additional details on the study-specific
modeling.
For the epidemiological studies considered for dose-response assessment, EPA used multiple
modeling approaches to determine PODs, depending upon the health outcome and the data
provided in the studies. For the developmental, hepatic, and serum lipid dose-response studies,
EPA used a hybrid modeling approach that involves estimating the incidence of individuals
above or below a level considered to be adverse and determining the probability of responses at
specified exposure levels above the control (U.S. EPA, 2012) because the EPA was able to
define a level considered clinically adverse for these outcomes (see Appendix E). As sensitivity
analyses for comparison purposes, EPA also performed BMD modeling and provided study
LOAELs/NOAELs as PODs for the epidemiological hepatic and serum lipid dose-response
studies. For the immune studies, for which a clinically defined adverse level is not established,
EPA used multivariate models provided in the studies and determined a BMR according to EPA
guidance to calculate BMDs and BMDLs (U.S. EPA, 2012). See Appendix E for additional
details on the study-specific modeling.
After POD derivation, EPA used a pharmacokinetic model for human dosimetry to estimate
human equivalent doses (HEDs) from both animal and epidemiological studies. A
pharmacokinetic model for human dosimetry is used to simulate the HED from the animal PODs
and is also used to simulate selected epidemiological studies to obtain a chronic dose that would
result in the internal dose POD obtained from dose-response modeling (see Toxicity Assessment,
(U.S. EPA, 2024).). Based on the available data, a serum PFOS concentration was identified as a
suitable internal dosimetry target for the human and animal endpoints of interest.
Next, reference values are estimated by applying relevant adjustments to the point-of-departure
human equivalent doses (PODheds) to account for five possible areas of uncertainty and
variability For each noncancer dataset analyzed for dose-response, reference values are estimated
by applying relevant adjustments to the point-of-departure human equivalent doses (PODheds) to
A-123
-------
APRIL 2024
account for five possible areas of uncertainty and variability: extrapolation from animals to
humans, human variation, the type of POD being used for reference value derivation,
extrapolation to chronic exposure duration, and extrapolation to a minimal level of risk (if not
observed in the dataset). The particular value for these adjustments is usually 10, 3, or 1, but
different values may be applied based on chemical-specific information if sufficient information
exists in the chemical database. The assessment discusses the scientific bases for estimating these
data-based adjustments and uncertainty factors (UFs). UFs used in this assessment were applied
according to methods described in EPA's Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA, 2002).
• Animal-to-human extrapolation: If animal results are used to make inferences about
humans, the toxicity value incorporates cross-species differences, which may arise from
differences in toxicokinetics or toxicodynamics. If a biologically based model adjusts
fully for toxicokinetic and toxicodynamic differences across species, this factor is not
used. Otherwise, if the POD is standardized to equivalent human terms or is based on
toxicokinetic or dosimetry modeling, a factor of 101/2 (rounded to 3) is applied to account
for the remaining uncertainty involving toxicokinetic and toxicodynamic differences.
• Human variation: The assessment accounts for variation in susceptibility across the human
population and the possibility that the available data may not be representative of
individuals who are most susceptible to the effect. If population-based data for the effect
or for characterizing the internal dose are available, the potential for data-based
adjustments for toxicodynamics or toxicokinetics is considered. Further, "when sufficient
data are available, an intraspecies UF either less than or greater than 10x may be justified
(U.S. EPA, 2002). However, a reduction from the default (10) is only considered in cases
when there are dose-response data for the most susceptible population" (U.S. EPA, 2002).
This factor is reduced only if the POD is derived or adjusted specifically for susceptible
individuals (not for a general population that includes both susceptible and non-
susceptible individuals) (U.S. EPA, 2002, 1991). Otherwise, a factor of 10 is generally
used to account for this variation.
• LOAEL to NOAEL: If a POD is based on a LOAEL or a BMDL associated with an
adverse effect level, the assessment must infer an exposure level where such effects are
not expected. This can be a matter of great uncertainty if there is no evidence available at
lower exposures. A factor of up to 10 is generally applied to extrapolate to a lower
exposure expected to be without appreciable effects. A factor other than 10 may be used
depending on the magnitude and nature of the response and the shape of the dose-response
curve.
• Sub chronic-to-chronic exposure: If a chronic reference value is being developed and a
POD is based on subchronic evidence, the assessment considers whether lifetime exposure
could have effects at lower levels of exposure. A factor of up to 10 is applied when using
subchronic studies to make inferences about lifetime exposure. A factor other than 10 may
be used, depending on the duration of the studies and the nature of the response. This
factor may also be applied, albeit rarely, for developmental or reproductive effects if
exposure covered less than the full critical period.
• In addition to the adjustments above, if database deficiencies raise concern that further
studies might identify a more sensitive effect, organ system, or lifestage, the assessment
may apply a database UF (U.S. EPA, 2002, 1991). The size of the factor depends on the
nature of the database deficiency. For example, EPA typically follows the suggestion that
A-124
-------
APRIL 2024
a factor of 10 be applied if a prenatal toxicity study and a two-generation reproduction
study are both missing, and a factor of 101/2 (rounded to 3) if either one or the other is
missing. A database UF would still be applied if this type of study were available but
considered to be a low confidence study.
The POD for a particular RfD is divided by the product of these factors. The RfD review
recommends that any composite factor that exceeds 3,000 represents excessive uncertainty and
recommends against relying on the associated RfD.
A. 1.11.3 Cancer Assessment
A.l.11.3.1 Approach for Cancer Classification
In accordance with EPA's 2005 Guidelines for Carcinogen Risk Assessment, a descriptive
weight of evidence expert judgment is made, based on all available animal, human, and
mechanistic data, as to the likelihood that a contaminant is a human carcinogen and the
conditions under which the carcinogenic effects may be expressed (U.S. EPA, 2005a). A
narrative is developed to provide a complete description of the weight of evidence and conditions
of carcinogenicity. The potential carcinogenicity descriptors (presented in the 2005 guidelines)
are:
• Carcinogenic to Humans
• Likely to Be Carcinogenic to Humans
• Suggestive Evidence of Carcinogenic Potential
• Inadequate Information to Assess Carcinogenic Potential
• Not Likely to Be Carcinogenic to Humans
More than one carcinogenicity descriptor can be applied if a chemical's carcinogenic effects
differ by dose, exposure route, or mode of action (MOA)3. For example, a chemical may be
carcinogenic to humans above but not below a specific dose level if a key event in tumor
formation does not occur below that dose. MOA information informs both the qualitative and
quantitative aspects of the assessment, including the human relevance of tumors observed in
animals. The MOA analysis must be conducted separately for each target organ/tissue type (U.S.
EPA, 2005a).
A.1.11.3.2 Derivation of Candidate Cancer Slope Factors
EPA's 2005 Guidelines for Carcinogen Risk Assessment recommends a two-step process for the
quantitation of cancer risk as a CSF. A CSF is a plausible upper bound lifetime cancer risk from
chronic ingestion of a chemical per unit of mass consumed per unit body weight per day (mg/kg-
day) (U.S. EPA, 2005a). First, a model is used to fit a dose-response curve to the data, based on
the doses and associated tumors observed (U.S. EPA, 2005a). In the second step of quantitation,
the POD is extrapolated to the low-dose region of interest for environmental exposures. The
approach for extrapolation depends on the MOA for carcinogenesis (i.e., linear or nonlinear).
When evidence indicates that a chemical causes cancer through a mutagenic MOA (i.e., mutation
of deoxyribonucleic acid (DNA)) or the MOA for carcinogenicity is not known, the linear
3MOA is defined as a sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through
operational and anatomical changes, and resulting in cancer formation. It is contrasted with "mechanism of action," which
implies a more detailed understanding and description of events.
A-125
-------
APRIL 2024
approach is used, and the extrapolation is performed by drawing a line (on a graph of dose vs.
response) from the POD to the origin (zero dose, zero tumors). The slope of the line
(Aresponse/Adose) gives rise to the CSF, which can be interpreted as the risk per mg/kg/day.
For animal toxicological studies, EPA used the publicly available Benchmark Dose Software
(BMDS) program developed and maintained by EPA (https://www.epa.eov/bmds). First, a PK
model converted the administered dose reported by the study to an internal dose (see Toxicity
Assessment, (U.S. EPA, 2024)). Then, BMDS fits multistage models, the preferred model type
(U.S. EPA, 2012), to the data and the model is used to identify a POD for extrapolation to the
low-dose region based on the BMD associated with a significant increase in tumor incidence
above the control. According to the 2005 guidelines, the POD is the lowest dose that is
adequately supported by the data. The BMDio (the dose corresponding to a 10% increase in
tumors) and the BMDLio (the 95% lower confidence limit for that dose) are also reported and are
often used as the POD. Similar to noncancer PODs, selection of model types is often based on
the goodness-of-fit (U.S. EPA, 2012). For PFOS, after a POD was determined, a PK model was
used to calculate the HED for animal oral exposures (PODhed). The CSF is derived by dividing
the BMR by the PODhed. See Appendix E for additional details on the study-specific modeling.
In addition, according to EPA's Supplemental Guidance for Assessing Susceptibility from Early-
Life Exposure to Carcinogens (U.S. EPA, 2005b), affirmative determination of a mutagenic
MOA (as opposed to defaulting to a mutagenic MOA based on insufficient data or limited data
indicating potential mutagenicity) indicates the potential for higher cancer risks from an early-
life exposure compared to the same exposure during adulthood, and so requires that the
application of age-dependent adjustment factors (ADAFs) be considered in the quantification of
risk to account for additional sensitivity of children. The ADAFs are 10- and 3-fold adjustments
that are combined with age specific exposure estimates when estimating cancer risks from early
life (<16 years of age) exposure to a mutagenic chemical.
In cases for which a chemical is shown to cause cancer via an MOA that is not linear at low
doses, and the chemical does not demonstrate mutagenic or other activity consistent with
linearity at low doses, a nonlinear extrapolation is conducted. EPA's 2005 Guidelines for
Carcinogen Risk Assessment state that "where tumors arise through a nonlinear MOA, an oral
RfD or inhalation reference concentration, or both, should be developed in accordance with
EPA's established practice of developing such values, taking into consideration the factors
summarized in the characterization of the POD" (U.S. EPA, 2005a). In these cases, an RfD-like
value is calculated based on the key event4 for carcinogenesis or the tumor response.
A. 1.11.4 Selecting Health Outcome-Specific and Overall Toxicity Values
The next step is to select a health outcome-specific toxicity value for each hazard (cancer and
noncancer) identified in the assessment. This selection can be based on the study confidence
considerations, the most sensitive outcome, a clustering of values, or a combination of such
factors; the rationale for the selection is presented in the assessment. Key considerations for
candidate value selection are described in the IRIS Handbook (U.S. EPA, 2022c) and include: 1)
the weight of evidence for the specific effect or health outcome; 2) study confidence; 3)
4The key event is defined as an empirically observed precursor step that is itself a necessary element of the MOA or is a
biologically based marker for such an element.
A-126
-------
APRIL 2024
sensitivity and basis of the POD; and 4) uncertainties in modeling or extrapolations. The value
selected as the organ/system-specific toxicity value is discussed in the assessment.
The selection of overall toxicity values for noncancer and cancer effects involves the study
preferences described above, consideration of overall toxicity, study confidence, and confidence
in each value, including the strength of various dose-response analyses and the possibility of
basing a more robust result on multiple datasets. The values selected as the overall RfD and CSF
are discussed in the assessment.
A-127
-------
APRIL 2024
Meta-Analysis Table
Studies identified in title/abstract and full-text screening as assessments or records with no original data were considered supplemental
material. Meta-analysis studies were included among those secondary studies. Consideration of meta-analyses alongside original
epidemiology studies could lead to duplication of results and give greater weight to studies included in meta-analyses; therefore, meta-
analysis studies were summarized separately. For PFOS, 21 epidemiological meta-analysis studies were identified and summarized
below (Table A-45).
Table A-45. Epidemiologic Meta-Analysis Studies Identified from Literature Review
Reference
Number of
Studies
Countries
Health Outcome
Results/Conclusions11
Meta-Analysis Studies Identified before February 2022
Verner et al.
(2015)
Canada, Denmark,
Japan, Norway,
Taiwan, United
Kingdom, United
States
Developmental Birthweight:
• Pooled (3 per 1 ng/mL increase of PFOS in maternal or cord blood (6
studies) = -5.0 g (-8.9, -1.1)
• Physiologically based pharmacokinetic model simulations suggest that the
association between PFAS levels and birthweight may be confounded by
changes in glomerular filtration rate and due to blood draw timing
Negri et al.
(2017)
13
Canada, China,
Denmark, Germany,
Greenland, Japan,
Norway, Poland,
South Korea,
Taiwan, Ukraine,
United Kingdom,
United States
Developmental Birthweight:
• Pooled (3 per 1 ng/mL increase in PFOS in maternal or cord blood (8
studies) = -0.92 g (-3.4, 1.6), I2 = 74%
• Pooled (3 per 1-ln ng/mL increase in PFOS in maternal or cord blood (8
studies) = -46.1 g (-80.3, -11.9), I2 = 25%
Dzierlenga et al. 29
(2020)
Australia, Belgium,
Canada, China,
Denmark,
Greenland, Japan,
Norway, Poland,
South Korea, Spain,
Sweden, Taiwan,
Ukraine, United
Kingdom, United
States
Developmental Birthweight:
• Pooled (3 per 1 ng/mL increase in PFOS in maternal or cord blood (29
studies) = -3.22 g (-5.11, -1.33), I2 = 58.3%
• Pooled (3 per 1 ng/mL in PFOS sampled before or in early pregnancy (8
studies) = -1.35 g (-2.33, -0.37), I2 = 5%
• Pooled (3 per 1 ng/mL in PFOS sampled in later pregnancy (21
studies) = -7.15 g (-10.93, -3.41), I2 = 55%
• Meta-regression modeling for timing of blood draw (early vs. late) showed
that when drawn from before or early pregnancy, there was no significant
relationship between birthweight and PFOS: 0.59 g/ng/mL (-1.94, 3.11)
A-128
-------
APRIL 2024
Reference
Number of
Studies
Countries
Health Outcome
Results/Conclusions"
Caoetal. (2021) 5
South Korea, Spain,
Taiwan, United
States
Developmental LBW:
• Pooled OR for PFOS in maternal blood (5 studies) = 1.32 (1.09, 1.55),
I2 = 0.00%
• Stratified by region: positive association in United States (2 studies), pooled
OR = 1.44 (1.15, 1.72)
Dejietal. (2021) 21
Brazil, Canada,
China, Denmark,
Norway, Spain,
United States
Developmental,
Female Reproductive
PTBC:
• Pooled OR (16 studies) = 1.20 (1.04, 1.38), I2 = 54.3%
• Pooled OR (6 studies in in North America) = 1.09 (1.01, 1.19); I2 = 0%
Miscarriage:
» Pooled OR (6 studies): 1.01, 95% CI: 0.92, 1.10; I2 = 35.9%
Gaoetal. (2021) 29
Brazil, Canada,
China, Denmark,
Norway, Spain,
Sweden, United
States
Developmental,
Female Reproductive
Preeclampsia:
• Pooled OR per 1-log increase inPFOS (4 studies) = 1.27 (1.06, 1.51)
PTBC:
• Pooled OR per 1 ng/mL increase in PFOS (8 studies): 1.01 (1.00-1.02)
GDM (7 studies), miscarriage (2 studies), pregnancy-induced hypertension (2
studies), SGA (6 studies), LBW (2 studies): Associations not statistically
significant
Yang et al. 22
(2022b)
Belgium, Canada,
China, Denmark,
Netherlands,
Norway, Slovakia,
Spain, Sweden,
United States
Developmental PTBC:
• Pooled OR (14 studies): 1.54 (1.20, 1.98), I2 = 63.4%
o Significant associations between PFOS and PTB in America [5 studies,
pooled OR = 1.44(1.19, 1.76), I2 = 2.1%]
o Significant associations for PFOS in maternal blood sampled in lst-2nd
trimester and in 3rd trimester to delivery, and for maternal blood sample
type overall
Miscarriage:
• Pooled OR (5 studies) = 1.10 (0.93, 1.32), I2 = 0%
SGA:
• Pooled OR (9 studies) = 1.22 (0.92, 1.61), I2 = 74.3%
o Significant associations for PFOS in cord blood at delivery [2 studies,
pooled OR = 2.51 (1.45, 4.34), I2 = 0.00%]
• Pooled OR (7 studies): 1.52 (1.19, 1.94), I2 = 19.1%
LBW:
• Pooled OR (2 studies, U.S. only): 1.71 (1.19, 2.47), I2 = 0%
• Pooled OR for PFOS in maternal blood (6 studies): 1.48 (1.16, 1.90),
I2 = 22.9%
A-129
-------
APRIL 2024
Reference
Number of
Studies
Countries
Health Outcome
Results/Conclusions"
Costello et al.
(2022)
25
Asia (NOS), Europe Hepatic
(NOS), United
States
ALT:
• PFOS was associated with higher ALT levels in adults and adolescents
o Cross-sectional (6 studies) weighted z-score = 3.55, p < 0.001
o Longitudinal (1 study) reported a positive association
o No associations for PFOS and ALT in children less than 12 yr of age or
other liver enzymes
GGT, AST, and other liver enzymes:
• Associations for PFOS not statistically significant
Abdullah Soheimi 29
et al. (2021)
Canada, China,
Denmark, Italy,
Norway, Spain,
Sweden, Taiwan,
United States
Cardiovascular (16
studies)
Serum Lipids (10
studies)
CVD:
• Strong evidence of association between serum PFOS and CVD risk (14
studies); z = 3.87, p < 0.0001,12 = 60.13%
CIMT:
• Inconsistent associations between serum PFOS and CIMT (2 studies)
• Consistent associations between serum PFOS and increased serum TC, LDL,
and TG levels
Metabolic (3 studies) GDM:
• Inconsistent associations between serum PFOS and increased GDM in
pregnant mothers compared with non-pregnant mothers
Kim etal. (2018) 12
Canada, China, Endocrine
South Korea, Japan,
Norway, Taiwan,
United States
Free T4:
• Pooled z-value (9 studies): 0.05 (0.03, 0.08), I2 = 0%
• More pronounced correlation between blood PFOS and free T4 in
intermediate exposure group (8-16 ng/mL): 0.07 (0.02, 0.11), I2 = 0%
• Association not statistically significant among subgroup of pregnant women
• Total T4 (8 studies), Total T3 (8 studies), TSH (12 studies): Associations
not statistically significant
• Sensitivity analyses removed outlier for total T4 and total T3; total T4 z-
value = -0.04 (-0.07, -0.01), I2 = 5%; total T3 z-value = -0.06 (-0.09,
-0.03), I2 = 31%
Zare Jeddi et al. 7
(2021b)
Canada, China,
Croatia, Italy,
United States
Metabolic
Metabolic syndrome:
• Pooled OR: 0.94 (0.79, 1.10), I2 = 78.7%
Stratakis et al.
(2022)
21
China, Demark,
Faroe Islands,
Greenland,
Netherlands,
Norway, Spain,
Metabolic
BMI z-score:
• In infancy (3 studies): Pooled (3 per unit increase in prenatal PFOS: -0.007
(-0.012,-0.003), I2 = 0%
• In childhood period (2-9 years) (10 studies): Pooled (3 per unit increase in
prenatal PFOS = 0.00 (-0.01, 0.01), I2 = 42.9%
A-130
-------
APRIL 2024
Reference
Number of
Studies
Countries
Health Outcome
Results/Conclusions"
Sweden, Taiwan,
Waist circumference:
Ukraine, United
• In childhood (4 studies): Pooled (3 per unit increase in prenatal PFOS = -0.06
Kingdom, United
(-0.19, 0.07), I2 = 20.5%
States
• Inconsistent associations between PFOA exposure and fat mass, overweight
risk
Quetal. (2021) 8
Denmark,
Greenland, Norway,
Poland, Sweden,
Ukraine, United
States
Neurodevelopmental ADHD:
• Pooled OR: 1.01 (0.88, 1.14), I2 = 54.7%
• Subgroup analysis between children's blood and prevalence rate of ADHD (2
studies), pooled OR = 1.05 (1.02, 1.08), I2 = 48.7%
• Subgroup analysis between PFOS exposure and prevalence rate of ADHD in
the United States (2 studies), OR = 1.05 (1.02, 1.08), I2 = 48.7%
Meta-Analyses Studies Identified after February 2022
Jiang et al. (2022) 8
China, Denmark,
France, Japan, The
Philippines, United
States
Cancer
Breast cancer:
• PFOS had no association with breast cancer risk (pooled OR= 1.01 [0.87,
1.17], I2 = 99.8%)
• Pooled OR (8 studies) =1.01 (0.87, 1.17), I2 = 99.8%
• Serious methodological limitations warrant cautious interpretation of results
from this publication.
Gui et al. (2022a) 46
Australia, Brazil,
Canada, China,
Denmark, England,
Faroe Islands,
Germany,
Greenland, Japan,
Norway, Poland,
South Korea, Spain,
Sweden, Taiwan,
Ukraine, United
States
Developmental Meta-analysis of 23 studies, pooled change in birthweight per 1-ln ng/mL
increase in PFOS (unadjusted for gestational age/unstandardized birth weight).
Significant effects observed for birth weight, birth length, ponderal index, and
head circumference. No significant associations observed for preterm birth, low
birth weight or small for gestational age. Subgroup analyses were included, by
fetal sex, time of blood sample collection, blood sample type and whether
adjusted for GA/parity, study design, and geographic region. Described
assessment of risk of bias for studies included in the meta-analyses.
Birth weight:
• Pooled (3 per 1 ln(ng/mL) increase in PFOS (23 studies) = -34.88 g (-52.53,
-17.24), I2 = 66.2%
Birth length:
• Pooled (3 per 1 ng/mL increase in PFOS (3 studies) = - 0.034 cm (-0.062,
-0.005), I2 = 0.0%
Ponderal index:
• Pooled (3 per 1 ng/mL increase in PFOS (2 studies) = -0.355 g/cm3 (-0.702,
-0.008), I2 = 0.0%
Head circumference:
A-131
-------
APRIL 2024
Reference
Number of
Studies
Countries
Health Outcome
Results/Conclusions"
• Pooled (3 per 1 ng/mL increase in PFOS (2 studies) = -0.021 cm (-0.038,
-0.004), I2 = 0.0%
PTBC:
• Pooled OR for the highest vs. lowest PFOS exposure (7 studies) = 1.46 (0.97,
2.18)
LBW:
• Pooled OR per 1 ln(ng/mL) increase in PFOS (3 studies) = 1.23 (0.96, 1.57)
SGA:
• Associations for PFOS not statistically significant.
Zhang et al. 9
(2022b)
Faroe Islands,
Germany,
Greenland, Guinea-
Bissau, Norway,
United States
Immune
Vaccine antibody production in children:
Tetanus antibodies:
• Pooled effect estimate (3 studies, 5 results) = -10.04 (-19.12, -0.96), p-value
for heterogeneity = 0.546
• Unclear what the effect estimate measures reported are and what units were
used for PFOS exposure.
Diphtheria antibodies:
• No association for PFOS exposure.
Gui et al. (2022b) 22
China, Norway,
Sweden, South
Korea, Taiwan,
United States
Metabolic
Diabetes:
• Case-control studies (number of studies not reported): OR of T2DM
incidence for high vs. low PFOS exposure = 1.80, (1.09, 2.97), I2 = 5%; OR
per ln-ng/mL increase in PFOS = 0.12 (0.07, 0.20), I2 = 0%
• Cohort studies (6 studies): HR per ln-ng/mL increase in PFOS = 1.40 (1.15,
1.69), I2 = 47%
• No association with PFOS in case-control and cross-sectional studies
combined.
Wang et al.
(2022a)
China, Denmark,
Faroe Islands,
Greenland, Poland,
Ukraine, United
States
Male Reproductive
Semen quality:
• No association with any of the six semen parameters
Pan etal. (2023) 11
China, Italy,
Norway, Sweden,
United States
Cardiovascular Hypertension:
Pooled OR (11 studies, 12 results) = 1.19 (1.06-1.34), 12 = 87.8%
Unit change in PFOS associated with pooled OR not reported.
Serious methodological limitations warrant cautious interpretations of results
from this publication. These include missing studies, inclusion of studies with
overlapping populations, lack of effect estimate with common unit change in
exposure.
A-132
-------
APRIL 2024
Reference Number of
Countries Health Outcome
Results/Conclusions"
Studies
Yu et al. (2023) 9
NR Renal
Hyperuricemia:
Pooled OR (6 studies) = 1.23 (1.01, 1.50), I2 = 58%
Change in PFOA associated with pooled OR not reported.
Zhang et al. ,,
(2023a)
Canada, China,
Denmark, Norway,
Spain, South Korea, Endocrine
Taiwan, United
States,
TSH during pregnancy:
Pooled (3 perng/mL increase in PFOS (13 studies) = 0.010 (0.009, 0.011),
I2 = 26.0%
No significant associations with other thyroid hormones (e.g., total T3, total T4,
free T3, free T4)
Notes: PFOS = perfluorooctane sulfonic acid; PFAS = perfluoroalkyl substances; In = natural log; OR = odds ratio; LBW = low birth weight; PTB = preterm birth;
GDM = gestational diabetes mellitus; SGA = small for gestational age; ALT = alanine aminotransferase; GGT = y-glutamyltransferase; AST = aspartate aminotransferase;
CVD = cardiovascular disease; CIMT = carotid artery intima-media thickness (mm); TC = total cholesterol; LDL = low-density lipoproteins; TG = triglyceride; T4 = thyroxine;
T3 = triiodothyronine; TSH = thyroid stimulating hormone; BMI = body mass index; ADHD = attention deficit hyperactivity disorder; GA = gestational age; T2DM = type 2
diabetes mellitus; HR = hazard ratio; NR = not rated.
¦Results reported as effect estimate and 95% confidence interval (CI) unless otherwise stated.
bToxicological study data included in these publications were not subject to meta-analysis.
'Preterm birth was defined as birth < 37 weeks of gestation.
A-133
-------
APRIL 2024
A3 Studies Identified In Supplemental Literature Search
Assessment
The EPA conducted a supplemental literature search in 2023. Consistent with the final IRIS
handbook (U.S. EPA, 2022c), the studies identified after February 3, 2022, including studies
recommended via public comment, were "considered for inclusion only if they [were] directly
relevant to the assessment PECO criteria and [were] expected to potentially impact assessment
conclusions or address key uncertainties" (U.S. EPA, 2022b). For the purposes of this
assessment, EPA defined impacts on the assessment conclusions as data from a study (or studies)
that, if incorporated into the assessment, have the potential to significantly affect (i.e., by an
order of magnitude or more) the final toxicity values (i.e., RfDs and CSFs) for PFOS or alter the
cancer classification.
The EPA has defined a systematic process for assessing the potential for a quantitative impact
that is consistent with the final IRIS Handbook. First, EPA reviewed studies against two broad
inclusion criteria for new relevant health effects studies: 1) the study met the pre-defined PECO
criteria and 2) following the SAB PFAS Review Panel's recommendation, the health
effect/endpoint described in the study was within the one of the five health outcomes determined
to have the strongest weight of evidence (i.e., developmental, hepatic, immune, cardiovascular,
and cancer) (U.S. EPA, 2022c). Second, for studies that met these two inclusion criteria, two or
more subject matter experts (e.g., epidemiologists and/or toxicologists) independently reviewed
the studies to determine whether the study conclusions potentially impacted assessment
conclusions. Subject matter experts considered a variety of factors to determine this, including,
but not limited to, whether the publication provided 1) information on a health effect (within the
five priority health outcomes) that was not previously quantitatively considered for dose
response; 2) information on health effects that were previously considered quantitatively and
potentially indicated effects at lower doses than the critical studies selected for the draft points of
departure (PODs), RfD, or CSF; or 3) information on health effects that were previously
considered quantitively and may have improved study design or data analyses compared with
those that were selected for POD, RfD, or CSF derivation. If the subject matter experts disagreed
about a study's potential quantitative impact, an additional expert independently reviewed the
rationale and made a final decision. The EPA provides the rationales for study inclusion
decisions in Table A-46 and Table A-47. For PFOS, 52 epidemiological and 4 animal toxicity
studies were identified after the updated literature search in 2022 and underwent title/abstract
and full-text screening according to Section A. 1.6. These studies are summarized below (Table
A-46 and Table A-47). Studies that were selected for inclusion proceeded to study quality
evaluation and were incorporated into the relevant evidence synthesis and dose-response analysis
when the study was determined to be medium or high confidence.
Numerous studies identified in the supplemental literature search examined associations between
elevated exposure to PFOS and the primary health outcomes described in the Toxicity
Assessment (U.S. EPA, 2024) (i.e., cancer, hepatic, immune, cardiovascular, and
developmental). Specifically, there were six studies examining the effects of exposure to PFOS
on cancer, 11 studies examining the effects of exposure to PFOS on serum lipids, nine studies
examining the effects of exposure to PFOS on birth weight, one study examining the effect of
exposure to PFOS on antibody response in children, and four studies examined the effect of
exposure to PFOS on ALT concentrations. Summaries of these studies and their potential impact
A-134
-------
APRIL 2024
to the evidence base, as well as additional studies examining outcomes belonging to the five
priority health outcomes, are provided in Table A-46.
Two studies (Cao et al., 2022; Goodrich et al., 2022) identified in the supplemental literature
search evaluated the risk of liver cancer with elevated exposure to PFOS, and both studies
reported increased risks of liver cancer. In the 2022 updated evidence base, there were no studies
(0/2) that reported significantly increased risk of liver cancer. Considering both studies identified
in the 2023 supplemental literature search reported significant positive associations, there are
altogether two studies reporting significantly increased risk of liver cancer (2/4). Both studies
went through study quality evaluation, extraction, were considered for deriving PODs for PFOS,
and were moved forward and integrated into the PFOS MCLG syntheses for cancer. One study
(Zhang et al., 2023c) examining immune effects was determined to impact assessment
conclusions and proceeded through systematic review steps, including study quality evaluation,
extraction, incorporation into the evidence synthesis, and considered for dose-response analysis.
The study reported a decreased antibody response to rubella in adolescents associated with
elevated exposure to PFOA. This effect was consistent with other studies reporting decreased
antibody response to other pathogens (i.e., tetanus and diphtheria), but provided additional
evidence for a different pathogen.
A-135
-------
APRIL 2024
Table A-46. Studies Identified After Updated Literature Review (Published or Identified After February 2022)
Reference
Major Findings
Assessment Implications
Cancer
Cao et al. (2022)
Case-control study conducted in Zhejiang, China of 203 liver cancer cases and
203 healthy controls. The odds of liver cancer incidence were significantly
elevated with increasing PFOS exposure (OR = 2.609, 95% CI: 1.179, 4.029, p
for trend = 0.001).
Goodrich et al. Nested case-control study within the MEC Study, including incident, non-viral
(2022) HCC cases (n = 50) and healthy controls (n = 50). Significant increase in risk in in
those with high exposure (>85th percentile; >54.9 |ig/L) r.v. low exposure (<85th
percentile; <54.9 (ig/L) (OR = 4.50, 95% CI: 1.20, 16.00).
Liver Cancer: Exposure to PFOS may be
associated with increased risk of liver cancer in
adults. In the updated evidence base, there were
_no studies (0/2) that reported significantly
increased risk of liver cancer. Considering both
studies post-dating the 2022 literature search
which reported significant positive associations,
there are altogether two studies reporting
significantly increased risk of liver cancer (2/4).
Both studies were considered for deriving
PODs for PFOS and were moved forward and
integrated into the MCLG synthesis for cancer
(see Toxicity Assessment, (U.S. EPA, 2024)).
Cao et al. (2022) was determined to be low
confidence during study quality evaluation and
was not modeled. For Goodrich et al. (2022),
the study had a limited number of cases
(n = 11) and controls (n = 4) in the highest
exposure group and was not modeled due to
low sensitivity.
Feng et al.
(2022b)
Li et al. (2022)
Case-cohort study within the Dongfeng-Tongji cohort, including incident breast
cancer cases (n = 226) and a random sub-cohort (n = 990). No association with
PFOS or with summed PFSAs.
Case-control study of incident Chinese breast cancer cases (n = 373) and healthy
controls (n = 657). An inverse relationship was observed between increasing
PFOS exposure and incident breast cancer.
Velarde et al. Case-control study of 150 Filipino women (75 breast cancer cases and 75
(2022) controls). Serum PFOS levels were significantly higher in cases than in controls.
PFOS was positively but not statistically significantly associated with breast
cancer risk across quartiles of exposure after adjusting for potential confounders.
Positive significant association observed in crude models only in the highest
quartile of PFOS.
Breast Cancer: Exposure to PFOS may be
associated with increased risk of breast cancer.
_Evidence for breast cancer was mixed in the
updated evidence base with four studies
reporting an increased risk (4/10). Significant
_increases in risk were only observed in some
subpopulations (e.g., stratified by genotype)
and for some specific types of breast cancer
(e.g., ER- and PR- breast cancers). A recent
meta-analysis reported a nonsignificant positive
association for breast cancer, although there
were methodological imitations that warrant
cautions interpretations of results (Jiang et al.,
2022). Considering the studies post-dating the
2022 updated literature review, one study (1/3)
A-136
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
reported an indication of increased risk of
breast cancer. Altogether, five studies (5/13)
reported an increased risk of breast cancer,
which provides an indication of a potentially
increased risk of breast cancer with increasing
PFOS exposure, but the overall evidence
remains mixed. The studies were judged to not
quantitatively impact assessment conclusions
and were not moved forward.
Wen et al.
(2022)
Population-based cohort study of 11,747 participants from 1999-2014 NHANES
followed up to December 2015. PFOS was statistically significantly associated
with an increased risk in cancer mortality (OR = 1.75; 95% CI: 1.10, 2.83), but
only in the highest tertile (>17.1 ng/mL) compared with the lowest tertile
(<7.9 ng/mL).
All-cause Cancer: There was concern for the
lack of specificity of all-cause cancer in this
study. Neither study examining all-cause cancer
(0/2) from the updated evidence base reported a
significantly increased risk. Considering the
study post-dating the 2022 updated literature
did not observe associations, there was
altogether mixed evidence for all-cause cancer
(1/3). The studies were judged to not
quantitatively impact assessment conclusions
and were not moved forward.
Cardiovascular
Batzella et al. Cross-sectional study of residents (n = 36,517; aged 20-64) of the Veneto Region,
(2022b) Italy, a high-exposure community. In single-pollutant models, PFOS was
significantly associated with increased TC ((3 per 1-ln ng/mL increase in
PFOS = 5.14, 95% CI: 4.56, 5.72), HDL-C ((3 = 1.34, 95% CI: 1.12, 1.56), and
LDL-C ((3 = 4.11, 95% CI: 3.60, 4.62). Significant positive associations were
observed for all three lipid measurements in PFAS mixture analyses (WQS), with
PFOS identified as the primary contributor to the association between increased
PFAS exposure and elevated TC (weight: 0.43), HDL-C (weight: 0.65), and LDL-
C (0.61) in the overall population. Similar results were observed in BKMR and Q-
Gcomp analyses.
Batzella et al. Cross-sectional occupational study of retired and former male workers (n = 232)
(2022a) at a PFAS production plant located in Veneto, Italy (2018-2020). TC, LDL-C,
and SBP were significantly elevated in the highest quartile of PFOS exposure
compared with the lowest (TC: (3 = 17.04, 95% CI: 2.8, 31.27; LDL-C: (3 = 16.79,
95% CI: 3.37, 30.21; SBP (3 = 4.51, 95% CI: 0.09, 8.93), and in analyses of
continuous exposure (TC (3 per 1-ln-ng/mL increase in PFOS = 7.26, 95% CI:
2.04, 12.48; LDL-C (3 = 5.90, 95% CI: 0.97, 10.83; SBP (3 = 2.58, 95% CI: 0.97,
Total Cholesterol: Eleven studies identified after the
2022 updated literature search evaluated changes in
TC, and eight (8/11) reported significant increases
in TC with elevated exposure to PFOS (Batzella et
al., 2022b; Batzella et al., 2022a; Cakmak et al.,
2022; Cheng et al., 2022; Maranhao Neto et al.,
2022; Nilsson et al., 2022; Rosen et al., 2022;
Schillemans et al., 2022). In the updated evidence
base, there was evidence of increases in TC (18/23)
associated with elevated PFOS exposure in studies
of adults (see Toxicity Assessment, (U.S. EPA,
2024)). Considering the updated evidence base and
studies post-dating the 2022 literature search
together, there were 23 of 33 general population
adult studies reporting positive associations for TC.
Overall, these studies support EPA's conclusion of
evidence indicates that elevated exposures to PFOS
A-137
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Cheng et al. (2022)
Cakmak et al.
(2022)
Maranhao Neto et
al. (2022)
Rosen et al. (2022)
Schillemans et al.
(2022)
4.18). SBP was also significantly elevated in WQS regression analyses of PFAS
mixture, with PFOS identified as a main contributor (PFOS weight: 0.56). No
significant associations observed for HDL-C and DBP.
Cross-sectional study of 98 patients recruited from Shiyan Renmin Hospital
(Hubei, China), 2018-2019. Plasma PFOS was significantly associated with
increased TC ((3 per 1-ln-ng/mL increase in PFOS = 4.761, 95% CI: 0.863, 8.809)
and LDL-C ((3 = 6.206, 95% CI: 1.832, 10.767). No significant associations were
observed for HDL-C or TG. Associations were partly mediated by methylation of
genes related to lipid metabolism.
Population-based cross-sectional study (Canadian Health Measures Survey) of
6,768 participants aged 3-79 yr old. Increases in PFOS were significantly
associated with increased TC (percent change per GM [5.3 jxg/L] increase in
PFOS: 3.3, 95% CI: 0.7, 5.9) and the TC/HDL ratio (2.6, 95% CI: 0.8, 4.4). No
significant associations observed for LDL-C, HDL-C, or TG.
Cross-sectional study of 479 adult participants (aged 25-89) from the Kardiovize
study, Czech Republic. Serum PFOS was significantly associated with increased
SBP ((3 per 1-ln-ng/mL increase inPFOS = 1.18, SE = 0.48), TC ((3 = 0.13,
SE = 0.05), and HDL-C ((3 = 0.04, SE = 0.01). No significant associations
observed for DBP, LDL-C, or TG.
Cross-sectional study of 326 participants in the GenX Exposure Study (2017-
2018) in Wilmington, North Carolina. Serum PFOS was positively associated
with total non-HDL-C ((3 per quartile increase in PFOS = 4.89, 95% CI: 0.10,
9.68) and withTC ((3 = 5.71, 95% CI: 0.38, 11.04). Associations for non-HDL
cholesterol and TC were strongest among older participants aged 63-86 yr. No
associations were observed between serum PFOS and other serum lipid outcomes
(HDL-C, LDL-C, and TG).
Population-based nested case-control study of Swedish adults (n = 1,528) within
two cohorts, the SMC-C and the Cohort of 60YO, including the first incident
myocardial infarction (n = 345) and stroke (n = 354) cases. In cross-sectional
analyses among 631 controls, baseline plasma PFOS was associated with
increased TC ((3 per 1-SD-ln-ng/mL PFOS = 0.14, 95% CI: 0.06, 0.22), increased
LDL-C ((3 = 0.13, 95% CI: 0.06, 0.20), increased HDL-C ((3 = 0.05, 95% CI: 0.01,
0.07), increased apolipoprotein Al ((3 = 0.04, 95% CI: 0.02, 0.08), and decreased
TG ((3 = -0.11, 95% CI: -0.17, -0.05). No significant association was observed
for apolipoprotein B. In prospective analyses of the pooled cohorts, there were no
significant associations between baseline PFOS and subsequent incidence of
myocardial infarction, stroke, or CVD.
are associated with adverse cardiovascular effects,
specifically serum lipids, as well as EPA's selection
of increased total cholesterol in adults for dose-
response modeling.
LDL-C, HDL-C, and TG: Nine studies identified
after the 2022 updated literature search evaluated
_changes in LDL-C, and four studies (4/9) reported
significant increases in LDL-C with elevated
exposure to PFOS (Batzella et al., 2022b; Batzella
et al., 2022a; Cheng et al., 2022; Nilsson et al.,
2022). In the updated evidence base, thirteen
_general population adult studies (13/18) reported
positive associations for LDL-C. Considering the
updated evidence base and studies post-dating the
2022 literature search together, there were 17
general population adult studies (17/27) reporting
positive associations for LDL-C. The findings for
HDL-C and TG in these ten studies were mixed,
similar to results provided in the updated evidence
base. Overall, the studies were judged to not
quantitatively impact assessment conclusions and
were not moved forward; however, these studies
support EPA's conclusion of evidence indicates that
elevated exposures to PFOS are associated with
adverse cardiovascular effects, specifically serum
lipids.
Blood Pressure: Measures of blood pressure and
hypertension were examined in six studies identified
after the updated 2022 literature search, and three
studies (3/6) reported significant increases in
systolic blood pressure or increased risk of
hypertension (Batzella et al., 2022a; Ding et al.,
2022; Maranhao Neto et al., 2022). One meta-
A-138
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Study of Norwegian adults ages 24-72 yr old (n = 127) in the EuroMix study.
Serum PFOS was associated with significantly higher day 1 and day 2 HDL-C,
apolipoprotein Al, and apolipoprotein A2 in women after adjustment for false
discovery rate (HDL-C % change per IQR increase in PFOS = 10%, 95% CI: 4%,
18%; apolipoprotein A1 = 7%, 95% CI: 3%, 12%; apolipoprotein A2 = 8%, 95%
CI: 3%, 12%). No significant associations were observed for TC, LDL-C, or TG.
Prospective occupational study of Australian firefighters who had used AFFF
reporting cross-sectional (n = 783) and longitudinal (n = 130) analyses. PFOS was
significantly associated with increased TC for those in the highest exposure
quartile ((3 per doubling of PFOS = 0.273, 95% CI: 0.027, 0.52) and LDL-C
((3 = 0.100, 95% CI: 0.029, 0.17) in cross-sectional analyses. No significant
associations were observed for serum lipids in longitudinal analyses.
Cross-sectional study of 7,242 NHANES participants (cycles 2003-2016). Serum
PFOS was positively associated with ln-TC and the magnitude of the association
was not substantially altered by additional adjustment for energy intake-adjusted
fiber.
Dunder et al. Prospective cohort study PIVUS of seniors at age 70 (n = 864), followed up at age
(2022) 75 (n = 614) and age 80 (n = 404). Increases in PFOS over the 10-year follow-up
were significantly associated with increases in HDL-C ((3 = 0.04, 95% CI: 0.02,
0.06, p = 0.001). No significant association between changes in PFOS and TC,
LDL-C, or TG.
Papadopoulou et al.
(2022)
Nilsson et al.
(2022)
Linakis et al.
(2022)
Ding et al. (2022)
Yang et al. (2022a)
Tian et al. (2023)
Cohort study of 1,058 women (ages 42-52) with no hypertension from the
multiethnic and multiracial SWAN. There was significantly increased risk of
hypertension per doubling of PFOS (HR = 1.18, 95% CI: 1.09, 1.28), and across
tertiles of baseline serum PFOS (p-trend < 0.0001). In the mixture analysis,
women in the highest tertile of PFAS concentrations had a significantly higher
risk of hypertension compared with those in the lowest tertile (HR = 1.71,95%
CI: 1.15, 2.54; p-trend = 0.008).
Prospective study of 826 pregnant women from the Jiashan Birth Cohort
(enrollment 2016-2018), Jiashan, Zhejiang, China. Plasma PFOS measured within
16 wk gestation was inversely associated with gestational hypertension (OR per 1-
ln ng/mL increase in PFOS = 0.62, 95% CI: 0.39, 0.99), and with SBP in the third
trimester ((3 = -1.14, 95% CI: -2.10, -0.18). Similar associations were observed
across PFOS quartiles. PFOS was not associated with SBP in other trimesters, or
with DBP in any trimester.
Case-control study of pregnant women from Hangzhou, China, with (n = 82) and
without (n = 169) preeclampsia. PFOS exposure measured 1-2 d before delivery
was not significantly associated with SBP or DBP in pregnant women.
analysis post-dating the 2022 literature search
reported a significantly increased risk of
hypertension in adults, but there were some
methodological limitations which warrant cautious
interpretations of results (Pan et al., 2023). In the
_updated evidence base, there was evidence of
increases in systolic (7/9) and diastolic blood
pressure (7/8), and increased risk of hypertension
(4/7) in adults. Considering the updated evidence
base and studies post-dating the 2022 literature
search together, there were nine general population
_adult studies (9/14) reporting increases in systolic
blood pressure and diastolic blood pressure (9/17);
and five (5/9) general population adult studies
reporting increases in risk for hypertension.
_Evidence for changes in blood pressure and
increases in risk for hypertension were supportive of
a conclusion of moderate evidence for
cardiovascular effects, specifically serum lipids,
although blood pressure and hypertension were not
selected as outcomes for modeling.
Cardiovascular disease: A variety of cardiovascular
diseases, including heart arrythmia, myocardial
infarction, stroke, angina, heart disease, and acute
coronary syndrome were examined in four studies
identified after the 2022 updated literature search,
and three studies (3/4) reported significant increases
in risk for at least one type of cardiovascular disease
(Li et al., 2023; Feng et al., 2022a; Wen et al.,
2022). In the updated evidence base, evidence was
limited for cardiovascular diseases with one study
reporting increased risk of microvascular disease
(1/1), myocardial infarction (1/1), and all
"cardiovascular disease (1/4). Other cardiovascular
diseases were examined in single studies, and no
associations were observed. Considering the
A-139
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Zhang et al. (2022e) Prospective study of 1,080 participants in the Dongfeng-Tongji cohort of retired
workers in China established in 2008 and followed for approximately 5 yr.
Baseline serum PFOS concentrations were significantly inversely associated with
risk of incident hypertension (RR per 1-loglO-ng/mL increase in PFOS = 0.94,
95% CI: 0.88, 0.99) and with changes in SBP over the follow-up period
((3 = -1.48, 95% CI: -2.56, -0.51). Significantly inverse associations for risk of
incident hypertension (p for trend = 0.016) and changes in SBP (p for
trend = 0.032) persisted across PFOS quartiles. Associations with hypertension
risk were observed among females but not males (p-value for interaction -0.44).
Baseline serum PFOS concentrations were not significantly associated with
changes in DBP.
Feng et al. (2022a)
Lin et al. (2022)
Li et al. (2023)
Wen et al. (2022)
Population-based cross-sectional study (NHANES, 2003-2012) of 7,904 adults. In
males, there was a significantly increased odds of heart attack (OR per 1-log-
ng/mL increase in PFOS = 1.01, 95% CI: 1.00, 1.01, p = 0.040) and stroke (OR
per 1-log-ng/mL increase in PFOS = 1.01, 95% CI: 1.00, 1.01, p = 0.008). No
associations were observed between PFOS exposure and heart failure, coronary
heart disease, angina, or total CVD in either males or females.
Cross-sectional study of participants from two prior studies in Taiwan: controls
(aged 22-63; n = 601) from a CVD study (2008-2011) and participants (aged 12-
30, n = 886) from the YOTA cohort (2006-2008). Serum PFOS was associated
with significantly increased mean CIMT ((3 per 1-ln ng/mL increase = 9.240, SE:
2.077). Significantly increased CIMT was also observed when examining specific
measurements such as the left and right common carotid artery, and the left and
right carotid bulb. No associations were observed for the left and right internal
carotid arteries.
Hospital-based case-control study of adults with and without ACS (355 cases, 355
age- and sex-matched controls) recruited in 2022 in Shijiazhuang, Hebei, China.
In single PFAS models, plasma PFOS was significantly associated with ACS (OR
per 1-lm-ng/mL increase in PFOS = 1.65, 95% CI: 1.14, 2.38). The association
between PFOS and ACS remained significant in multiple-PFAS models. No
significant associations were observed with PFAS mixtures.
Population-based cohort study of 11,747 participants from 1999-2014 NHANES
followed up to December 2015. PFOS was statistically significantly associated
with an increased risk in heart disease mortality (OR = 1.75; 95% CI: 1.10, 2.83),
but only in the highest tertile (>17.1 ng/mL) compared with the lowest tertile
(<7.9 ng/mL).
updated evidence base and studies post-dating the
2022 literature search together, evidence was mixed
for any cardiovascular disease (4/8). Overall, the
studies were judged to not quantitatively impact
assessment conclusions and were not moved
forward.
Atherosclerotic changes: One study identified after
the updated 2022 literature search examined
_atherosclerotic changes in young adults, and the
study reported significantly increased CIMT (Lin et
al., 2022). In the updated evidence base, two studies
in children and adolescents (2/3) observed
significant changes in CIMT across exposure
groups. Considering the updated evidence base and
_studies post-dating the 2022 literature search
together, there were three studies (3/4) reporting
changes to CIMT in children and adolescents.
Overall, the studies were judged to not
quantitatively impact assessment conclusions and
were not moved forward.
Developmental
A-140
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Sevelsted et al.
(2022)
Tian et al. (2023)
Jia et al. (2023)
Hall et al. (2022)
Prospective study of 738 maternal-child pairs enrolled in a population-based birth
cohort study (COPSAC-2010) in Zealand, Denmark (2008-2010). Maternal
plasma PFOS (measured at 24 wk GA and 1 wk postpartum) was associated with
significantly lower birth BMI z-score ((3 per 1-ng/mL increase = -0.04, 95% CI:
-0.08, -0.01), decreased birth weight z-score ((3 = -0.04, 95% CI: -0.07, -0.01),
and decreased birth weight percentile for sex and GA ((3 = -1.07, 95% CI: -1.96,
-0.19).
Case-control study of pregnant women from Hangzhou, China, with (n = 82) and
without (n = 169) preeclampsia. PFOS exposure measured 1-2 d before delivery
was significantly associated with decrease in birth weight ((3 per 1-loglO-unit
increase in PFOS = -20.3, 95% CI: -33.2, -7.54).
Cross-sectional study of 66 infants born to women at a maternity hospital in
Shijiazhuang, Hebei, China in 2022. Umbilical cord serum PFOS was inversely
correlated with birth weight (Spearman correlation coefficient = -0.319,
p < 0.05).
Prospective birth cohort study of 120 mother-child pairs enrolled in the HPHB
cohort in Durham, North Carolina (enrollment 2010-2011). The highest tertile of
placental PFOS exposure was significantly associated with decreased birth weight
percentile in male infants (% change compared with lowest tertile = -13%, 95%
CI: -23%, -1.6%), and significantly associated with increased birth weight in
female infants (% change = 11%, 95% CI: 2.8%, 19%). No other associations with
gestational age or birth weight for gestational age were observed.
Prospective study of 506 maternal-child pairs enrolled in a birth cohort study in
Hangzhou, China (2020-2021). No significant associations were observed
between maternal serum PFOS (GA at assessment not specified) and birth weight,
Apgar scores, or preterm birth after adjustment for confounders.
Prospective study of 180 maternal-child pairs enrolled in a birth cohort study in
Tangshan City, Hebei province, China, 2013-2014. No associations were
observed between placental PFOS and birth outcomes (birth weight, birth length,
head circumference, and ponderal index).
Prospective study of 1,405 maternal-child pairs enrolled in the Shanghai Birth
Cohort in Shanghai, China (2013-2016). No significant associations were
observed between first trimester PFOS and birth weight z-score in children of
women with low or high fasting plasma third trimester glucose levels.
Peterson et al. (2022) Prospective study of pregnant women and their fetuses (n = 335 mother-fetus
pairs) from the Maternal and Developmental Risks from Environmental and
Social Stressors (MADRES) pregnancy cohort. No significant associations were
observed between maternal serum PFOS measured during pregnancy
Shen et al. (2022)
Wang et al. (2023a)
Wang et al. (2023b)
Birth weight: Nine studies identified after the
updated literature search evaluated changes in birth
weight (i.e., birth weight and birth weight for sex
and GA), and four studies reported significant
decreases. Studies reporting significant results
examined changes in birth weight in relation to
_PFOS concentrations measured in later pregnancy
(Jia et al., 2023; Tian et al., 2023; Hall et al., 2022;
Sevelsted et al., 2022). Other studies not observing
decreases in birth weight were generally smaller
_(i.e., <200 participants) (Wang et al., 2023a; Zhang
et al., 2023a). In the updated evidence base, there
were 27 studies reporting deficits in birth weight
(27/39). Considering the updated evidence base and
studies post-dating the 2022 literature search
together, deficits in birth weight were observed in
33 studies (31/48). Overall, these studies support
EPA's conclusion of evidence indicates that
elevated exposures to PFOS are associated with
adverse developmental effects, as well as EPA's
selection of decreased birth weight for dose-
response modeling.
Other FGR: Regarding other fetal growth restriction
outcomes, three studies identified after the updated
literature search evaluated changes in other
"measures of fetal growth restriction (e.g., birth
length, head circumference, and ponderal index) and
no associations were observed. In the updated
evidence base, there was some evidence of adverse
"effects for birth length (15/28) and head
circumference (13/23), but the evidence was
generally mixed. Overall, the studies were judged to
not quantitatively impact assessment conclusions
"and were not moved forward.
Gestational duration and PTB: Preterm birth was
examined in three studies, and two studies reported
A-141
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Yu et al. (2022)
(median =19 wk, range = 5.7-38.3 wk GA) and fetal growth parameters (head
circumference, biparietal diameter, femur bone length, abdominal circumference,
and estimated fetal weight).
Zhang et al. (2022a) Cohort study of pregnant women and their children (n = 94 mother-child pairs)
living near an e-waste recycling facility in Guangdong, China (2016). No
significant associations observed between maternal serum PFOS and birth
outcomes (i.e., birth weight, birth length, and head circumference).
Liao et al. (2022) Prospective study of 1,341 maternal-child pairs enrolled in the Guangxi Zhuang
Birth Cohort study in Guangxi, China, 2015-2019. In single PFAS models, first
trimester serum PFOS was associated with increased risk of preterm birth (RR per
loglO-ng/mL increase = 2.251, 95% CI: 1.307, 3.874). There was a significant
trend across PFOS quartiles. PFAS mixture was associated with increased risk of
preterm birth, with PFOS identified as one of the main contributors (weight:
31.8%).
Prospective study of 836 maternal-child pairs enrolled in the Maoming Cohort
Study in Maoming, China, 2015-2018. Maternal third trimester serum PFOS was
positively associated with preterm birth (OR per ln-ng/mL increase = 2.07, 95%
CI: 1.70, 2.52); paternal serum PFOS was inversely associated with preterm birth
(OR = 0.44, 95% CI: 0.36, 0.54). No association was observed with neonatal
PFOS.
Case-control study of women with and without unexplained recurrent spontaneous
abortion (URSA) (464 cases, 440 controls) in Shandong and Zhejiang provinces,
China (2014-2016). No association was observed between prepregnancy plasma
PFOS and URSA.
Nested case-control study of women with and without early pregnancy loss (41
cases, 47 controls) in Beijing, China (2018-2020). No association was observed
between prenatal PFOS (GA at measurement not specified) and early pregnancy
loss.
Romano et al. (2022) Prospective study of 481 maternal-child pairs enrolled in the NHBCS with at least
four child anthropometric measurements in the first year of life, (2009-2018).
Among girls, maternal second trimester PFOS was associated with an increased
chance of following a growth trajectory in which BMI increases gradually over
the first year of life compared with a growth trajectory in which BMI increases
gradually and plateaus around 3 mo (relative risk ratio per doubling of
PFOS = 2.5, 95% CI: 1.0, 6.1). Among girls, PFOS was also associated with an
increased chance of following a growth trajectory in which BMI steeply increases
in mo 1-3 of life (relative risk ratio per doubling = 2.8, 95% CI: 1.0, 7.6). At
Nian et al. (2022)
Mi et al. (2022)
significantly increased risks (Liao et al., 2022; Yu et
al., 2022). Exposure sample timing differed between
_the two studies, with one cohort study colleting
maternal samples in the first trimester (Liao et al.,
2022) and one study colleting maternal samples in
the third trimester (Yu et al., 2022). In the updated
_evidence base, there are ten studies (10/20)
reporting increased risk of preterm birth. Overall,
the studies were judged to not quantitatively impact
assessment conclusions and were not moved
forward.
Pregnancy loss: Pregnancy loss was examined in
two studies, and neither study reported significantly
increased risks (Mi et al., 2022; Nian et al., 2022).
Timing of exposure sample collection was reported
in one case-control study analyzing pre-pregnancy
plasma samples (Nian et al., 2022) and one nested
case-control study did not report exposure sample
timing (Mi et al., 2022). In the updated evidence
base, there are four studies (4/7) reporting increased
risk of pregnancy loss. Considering the updated
evidence base and studies post-dating the 2022
literature search together, there are four studies
reporting increased risk of pregnancy loss (4/9).
"Overall, the studies were judged to not
quantitatively impact assessment conclusions and
were not moved forward.
~Postnatal growth: Five studies examined postnatal
growth in early childhood, and two studies reported
an increased risk of following adverse BMI growth
trajectories in early childhood (Zeng et al., 2023;
Romano et al., 2022). No significant associations
were reported from other studies examining
postnatal growth from studies on birth cohorts such
as the Shanghai Birth Cohort (Zhang et al., 2022d),
the Flemish Environmental Health Study (Cai et al.,
A-142
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
Zeng et al. (2023)
6 mo, the estimated mean difference in BMI was significantly higher in girls in
the highest tertile of PFOS exposure compared with the lowest ((3 = 0.54, 95% CI:
0.05, 1.03). No associations were observed with other growth trajectory groups or
with BMI at other timepoints.
Prospective study of mother-child pairs (n = 1,671) from the Shanghai Birth
Cohort in China (2013-2016). Child anthropometric measures were taken at 6, 12,
24, and 48 mo. Maternal serum PFOS measured in early pregnancy (9-16 weeks
GA) was associated with significantly increased odds of following a BMI-for-age
z-score trajectory which increases steadily for the first 12 mo followed by steeper
increases up to 40 mo compared with a trajectory with a steep increase in the first
12 mo but progressively reversed to a stable trajectory at 40 mo (OR = 2.36, 95%
CI: 1.27,4.40).
2023), or the Danish National Birth Cohort (Luo et
al., 2022). In the updated evidence base, increased
risk for adverse changes in postnatal weight changes
in infancy were observed in eight (8/10) studies.
Considering the updated evidence base and studies
post-dating the 2022 literature search together, there
are ten studies reporting increased risk adverse
effects on postnatal growth (12/15). Overall, the
studies were judged to not quantitatively impact
assessment conclusions and were not moved
forward.
Cai et al. (2023)
Prospective study of 207 mother-child pairs from two birth cohorts from the
FLEHS: FLEHS I (2002-2004) and FLEHS II (2008-2009). No statistically
significant associations were observed between cord blood PFOS and infant
growth in single- or multi-pollutant models.
Zhang et al. (2022d) Prospective cohort study (the Shanghai Birth Cohort) of 2,395 mother-infant
pairs. Prenatal PFOS exposure measured in early pregnancy (median, 15
gestational wk) was not associated with infant length, weight, and head
circumference at birth, 42 d, 6 mo, and 12 mo.
Luo et al. (2022)
Prospective study in the DNBC, 656 children. Prenatal exposure to PFOS was not
associated with facial features (measures of palpebral fissure length, philtrum
groove, and upper-lip thickness) in children at age 5.
Immune
Zhang et al. (2023c) Population-based cross-sectional study of adolescents aged 12-19 with detectable
serum rubella and measles antibody levels (n = 819) from the NHANES 2009-
2010 and 2013-2014 cycles. The study population was stratified in two groups of
lower (n = 552) and upper (n = 267) folate levels based on the <66th percentile.
Significant inverse associations were observed for rubella antibody response in
the whole study population (% change per 2.7-fold increase in serum
PFOS = -8.16, 95% CI: -13.67, -2.31) and in the lower folate group (%
change = -11.00, 95% CI: -18.08, -3.31). No significant associations for rubella
antibodies in the higher folate group, or for mumps and measles antibodies.
Kaur et al. (2023)
Cross-sectional study of pregnant participants with past SARS-CoV-2 infection
(n = 72) from the Generation C Study. No significant association was observed
between maternal plasma PFOS and SARS-CoV-2 anti-spike IgG titers. In WQS
regression analysis of a PFAS mixture index, maternal SARS-CoV-2 anti-spike
Vaccine response: Three studies identified after the
updated literature search evaluated antibody
responses to multiple pathogens in different
populations, and two studies observed an effect
(Zhang et al., 2023c; Porter et al., 2022). The only
study examining rubella antibody response observed
a significant decrease (Zhang et al., 2023c). In the
updated evidence base, there was one study (1/2)
which reported significant decreases in rubella
antibody response in children and adolescents.
Considering the updated evidence base and studies
post-dating the 2022 literature search together, there
were two studies (2/3) in children and adolescents
A-143
-------
APRIL 2024
Reference Major Findings Assessment Implications
IgG titers were significantly decreased ((3 = -0.35, 95% CI: -0.52, -0.17, p- reporting significantly decreased rubella antibody
value = 0.0003), with PFOS accounting for greater than 10% of the effect. response. Zhang et al. (2023c) was considered for
Porter et al. (2022) Longitudinal study of current and retired workers (n = 415; 757 observations) of deriving PODs for PFOS and was moved forward
3M facilities in Decatur, Alabama and Menomonie, Wisconsin (Spring 2021). and integrated into the MCLG synthesis for immune
Serum PFOS was associated with decreases in SARS-CoV-2 anti-spike IgG effects (see Toxicity Assessment, (U.S. EPA,
antibody and SARS-CoV-2 neutralizing antibody response after adjustment for 2024)). One study (1/2) examining SARS-CoV-2
age, gender, race, BMI, location, smoking, immunocompromising conditions or antibody response reported significant inverse
recent corticosteroid use, and time since antigenic stimulus. Associations were not associations (Porter et al., 2022). There were a
significant after further adjustment for the antigenic stimulus group. limited number of studies examining SARS-CoV-2
Jones et al. (2022) Cross-sectional analysis of infants (n = 3,448) from the Upstate KIDS Study Birth in the studies captured in updated 2022 evidence
Cohort (2008-2010). PFOS and immunoglobulins were both quantified in infant base, but these studies post-dating the 2022 updated
heel stick blood spots. No significant associations were observed for IgA, IgE, literature search suggest there may be an association
IgGi, IgGz, IgG3, IgG4, or IgM. between exposure to PFOS and decreased SARS-
Zhang et al. (2022c) Population-based cross-sectional study of children aged 3-11 (n = 517) and CoV-2 antibody response, coherent with decreases
adolescents aged 12-19 (n = 2,732) from the NHANES 2013-2014 cycle and in the antibody response for other pathogens.
2003-2016 cycles, respectively. The odds of a recent common cold (i.e., past Overall, these studies provide additional evidence
30 d) was significantly elevated in adolescents per doubling in serum PFOS after for decreased antibody response for multiple
mutual adjustment for other PFAS (OR per l-log2 increase = 1.26, 95% CI: 1.01, pathogens, including in populations located in the
1.56). The association was not significant in single-pollutant models, and no United States, and support EPA's conclusion of
associations were observed in children. evidence indicates that elevated exposures to PFOS
Qu et al. (2022) Case-control study from the Second Affiliated Hospital of Zhejiang University arc associated with immunological effects in
School of Medicine (2019-2020), including rheumatoid arthritis patients (n = 156) humans, as well as EPA's selection of decreased
and healthy controls (n =156). The odds of rheumatoid arthritis were non- vaccine response in children for dose-response
significantly elevated with increasing serum PFOS (OR = 1.381, 95% CI: 0.972, modeling.
1.658, p = 0.06).
Zhao et al. (2022b) Case-control study of rheumatoid arthritis (RA) patients (n = 294) and volunteer Infectious disease: One study identified after the
controls (n = 280) in Hangzhou, China from January 2018-December 2020. A updated 2022 literature search examined infectious
significant positive association was observed between serum PFOS and RF, an disease in children and reported a significantly
indicator of RA ((3 per 1-ln ng/mL increase = 0.52, 95% CI: 0.28, 0.77), and increased odds of a recent common cold (Zhang et
ACPA ((3 = 0.48, 95% CI: 0.23, 0.73). A significant inverse association was ^ 2022c). In the updated evidence base, results
observed for IgM ((3 = -0.24, 95% CI: -0.64, 0.15). No significant associations were mixed for infectious disease in children, with
observed for C-RP, IgA, IgG, C4, C3, KAP, and LAM. five studies (5/12) reporting positive associations or
Zhao et al. (2022a) Case-control study from the Second Affiliated Hospital of Zhejiang University increased risk. Considering the updated evidence
School of Medicine (2019-2020), including RA patients (n = 155) and healthy ^ase arl<^ studies post-dating the 2022 literature
controls (n = 145). Serum PFOS concentrations were higher in cases than in search together, there were five studies (6/13)
controls (p < 0.0001). In a cross-sectional analysis of cases only, cases were reporting positive associations or increased risk of
categorized by their DAS28; inactivity, low activity, moderate activity, and high infectious disease in children. Overall, the study
activity). Significant differences (p = 0.0001) in median serum PFOS was judged to not quantitatively impact assessment
conclusions and were not moved forward.
A-144
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
concentrations were observed between the four groups of DAS28, with the highest
median PFOS concentrations observed among cases with the highest DAS28
score (>5.1). Comparing cases categorized as with leukopenia
(WBC < 4.0 x 109/L) to those without leukopenia (WBC > 4.0 x 109/L), serum
PFOS levels were higher in the non-leukopenia group. No significant associations
were observed between PFOA exposure and interstitial lung disease in cases.
Immunoglobulins: Two studies identified after the
updated 2022 literature search examined
immunoglobulins, and one study (1/2) observed an
effect (Zhang et al., 2022c). In the updated evidence
base, four studies examined immunoglobulins in a
variety of populations, with mixed evidence.
Overall, the studies were judged to not
quantitatively impact assessment conclusions and
were not moved forward.
Autoimmune disease: Three studies examining RA
were identified after the updated 2022 literature
search, and two studies (2/3) observed significantly
increased RA biomarkers (Zhao et al., 2022b) and
increased RA severity scores (Zhao et al., 2022a).
While both studies observed increases in risk or
evidence of increased biomarkers related to RA, the
methods of examination differed between the
studies, limiting comparability of the results. No
studies in the updated evidence base examined RA.
Evidence for other autoimmune diseases in the
updated evidence base was mixed and limited to a
small number of studies. Overall, the studies were
judged to not quantitatively impact assessment
conclusions and were not moved forward.
Hepatic
Liu et al. (2022) Community-based cross-sectional study of adults (n = 1,303) living in
Guangzhou, China. Positive dose-response relationships between PFOS and liver
enzymes, except for ALP. Significant associations were observed for the 50th
compared with the 25th percentile of PFOs for liver function biomarkers
(percentage differences): ALB (4.80, 95% CI: 4.47, 5.13), ALT (7.01, 95% CI:
4.69, 9.37), AST (2.76, 95% CI: 1.29, 4.25), GGT (6.74, 95% CI: 4.01, 9.55), and
DBIL (3.72, 95% CI: 5.41). Associations remained significant for other
comparisons (75th percentile vs. 25th percentile and 95th percentile vs. 25th
percentile). No significant association observed for ALP.
Borghese et al. Population-based cross-sectional study of adults (n = 4,657) from three cycles of
(2022) the CHMS. A twofold increase in serum PFOS was associated with significantly
ALT: Four studies identified after the updated 2022
literature search examined ALT, and one (1/4)
reported a significant increase (Liu et al., 2022). In
the updated evidence base, there were six medium
confidence studies (6/8) reporting increased ALT in
adults. Considering the updated evidence base and
studies post-dating the 2022 literature search
together, there were eleven (7/12) studies reporting
_increases in ALT in adults. Overall, the studies
support EPA's conclusion that evidence indicates
that PFOS exposure is likely to cause hepatotoxicity
A-145
-------
APRIL 2024
Reference
Major Findings
Assessment Implications
elevated liver enzymes (percent difference): GGT (7.6, 95% CI: 3.0, 12.4), ALP
(4.1, 95% CI: 2.4, 5.9), and AST in men (7.6, 95% CI: 4.6, 10.8) and women (3.3,
1.2, 5.5). No significant associations observed for ALT and total bilirubin.
Cakmak et al. (2022) Population-based cross-sectional study (CHMS) of 6,768 participants aged 3-
79 yr old. Increases in PFOS were significantly associated with decreased ALP
(percent change per GM [5.3 jxg/L] increase in PFOS: -2.1, 95% CI: -3.7, -0.5).
Significant increases were observed for GGT (11.6, 95% CI: 1.8, 22.3), and
bilirubin (4.7, 95% CI: 3.8, 5.6). No significant associations observed for AST or
ALT.
Zhang et al. (2022a) Cohort study of pregnant women and their children (n = 94 mother-child pairs)
living near an e-waste recycling facility in Guangdong, China (2016). Cross-
sectional analyses of maternal liver enzymes observed significantly decreased
AST ((3 per 1-ln ng/mL increase in PFOS = -0.236, 95% CI: -0.429, -0.043), but
no association for ALT.
Nilsson et al. (2022) Cross-sectional occupational study of Australian firefighters who had used AFFF
(n = 783). No significant associations were observed for ALT or self-reported
liver problems.
E et al. (2023) Population-based cross-sectional study of adults (n = 3,464) from NHANES
(2005-2018). The relative risk of NAFLD was decreased in men (RR per 1-log
ng/mL increase in PFOS = 0.878, 95% CI: 0.778, 0.991). No significant
associations were observed in all participants or in women only. No significant
monotonic trends across qualities of PFOS were observed.
in humans, specifically increased ALT in adults;
however, the studies were judged to not
_quantitatively impact assessment conclusions and
were not moved forward.
Other liver enzymes: Three studies identified after
the updated 2022 literature search examined liver
enzymes besides ALT, and all three studies (3/3)
_observed effects (Borghese et al., 2022; Cakmak et
al., 2022; Liu et al., 2022). In the updated evidence
base, there were three studies (3/11) reporting
increases in GGT in adults. Results for other liver
enzymes in adults were generally mixed.
_Considering the updated evidence base and studies
post-dating the 2022 literature search together, there
were six studies (6/14) reporting increases in GGT
in adults. Overall, the studies support EPA's
conclusion that evidence indicates that PFOS
exposure is likely to cause hepatotoxicity in
humans, specifically increased ALT in adults;
however, the studies were judged to not
quantitatively impact assessment conclusions and
were not moved forward.
Liver disease: Two studies identified after the
updated 2022 literature search examined liver
disease, and none reported increased risk of any
liver disease (0/2). In the updated evidence base,
there was one study examining all liver disease and
did not observe an association (0/1). Overall, the
studies were judged to not quantitatively impact
assessment conclusions and were not moved
forward.
Notes: OR = odds ratio; CI = confidence interval; PFOS = perfluorooctane sulfonic acid; POD = point of departure; MEC = Multiethnic Cohort Study; HCC = hepatocellular
carcinoma; PFSAs = perfluorinated sulfonic acids; NHANES = National Health and Nutrition Examination Survey; TC = total cholesterol; HDL-C = high-density lipoprotein
cholesterol; LDL-C = low-density lipoprotein cholesterol; PFAS = perfluoroalkyl substances; WQS = weighted quantile sum; BKMR = Bayesian kernel machine regression; Q-
Gcomp = quantile-g computation; In = natural log; SBP = systolic blood pressure; DBP = diastolic blood pressure; TG = triglycerides; GM = geometric mean; SE = standard
error; SMC-C = Swedish Mammography Cohort-Clinical; 60YO = 60-year-olds; SD = standard deviation; CVD = cardiovascular disease; MI = myocardial infarction;
IQR = interquartile range; AFFF = aqueous film forming foams; PIVUS = Prospective Investigation of the Vasculature in Uppsala Seniors Study; SWAN = Study of Women's
A-146
-------
APRIL 2024
Health Across the Nation; HR = hazard ratio; GA = gestational age; BMI = body mass index; RR = relative risk; YOTA = Young Taiwanese Cohort Study; CIMT = carotid artery
intima-media thickness; ACS = acute coronary syndrome; COPSAC-2010 = Copenhagen Prospective Studies of Asthma in Childhood 2010; HPHB = Healthy Pregnancy,
Healthy Baby; MADRES = Maternal and Developmental Risks from Environmental and Social Stressors; URSA = unexplained recurrent spontaneous abortion; NHBCS = New
Hampshire Birth Cohort Study; FLEHS = Flemish Environment and Health Studies; DNBC = Danish National Birth Cohort; SARS-CoV-2 = severe acute respiratory syndrome
coronavirus 2; IgG = immunoglobulin G; IgA = immunoglobulin A; IgE = immunoglobulin E; IgGi = immunoglobulin G subclass 1; IgG2 = immunoglobulin G subclass 2;
IgG3 = immunoglobulin G subclass 3; IgG4 = immunoglobulin G subclass 4; IgM = immunoglobulin M; RA = rheumatoid arthritis; RF = rheumatoid factor; ACPA = anti-
citrullinated protein antibodies; C-RP = c-reactive protein; C4 = complement 4; C3 = complement 3; KAP = light chain kappa isotype; LAM = light chain lambda isotype;
DAS28 = Disease Activity Score28; WBC = white blood cell; ALP = alkaline phosphatase; ALB = albumin; ALT = alanine transaminase; GGT = gamma-glutamyl transferase;
DBIL = direct bilirubin; CHMS = Canadian Health Measures Survey; AST = aspartate transaminase; NAFLD = non-alcoholic fatty liver disease.
Table A-47. Animal Studies Identified After Updated Literature Review (Published or Identified After February 2022)
Reference
Health Outcome(s)
Major Findings
Assessment Implications
Narizzano et al.
(2022)
Cardiovascular,
Developmental,
Hepatic, Immune
PFOS (0.2, 1, or 5 mg/kg/day) was administered via
oral gavage for 28 d prior to gestation and continued
throughout gestation and weaning (until PND25) to
parental white-footed mice (Peromyscus leucopus).
PFOS exposure led to neonatal mortality and total
litter loss at high doses. Both sexes of parental
animals exhibited increased livenbody weight,
decreased serum thyroxine, and increased hepatocyte
cytoplasmic vacuolization.
General Notes: Study uses a slightly lower dose level
(0.2 mg/kg/day) than many other studies and a new rodent
model (Peromyscus, a wild species with observable
differences to traditional laboratory strains).
Cardiovascular: No change in heart weight in parental
generation, database currently has mixed results, and
decreases seen only at levels around highest dose of this
study.
Developmental: Effects on developmental endpoints (e.g.,
stillbirth, live pups born) are consistent with results from
Luebker et al. (2005a), which used similar dose range in
rats with a similar study design but more dose groups.
There was a lack of effect on pup and fetal weight not
consistent with other data but could have been confounded
by fetal mortality. Developmental delays are not apparent
but could also be confounded by increased fetal mortality.
This animal model has a small litter size compared with
traditional laboratory mouse models.
Hepatic: Increased organ weight in parental generation
consistent with current database. Effects are not observed
at the lowest dose. The study does demonstrate
histopathological evidence of cytoplasmic vacuolization.
Immune: The study measures spleen and thymus weights
only.
Overall Assessment Conclusion: Effects are generally
consistent with current database, using generally similar
dose levels (though low dose is lower than many studies).
Endpoints measured for hepatic, developmental,
A-147
-------
APRIL 2024
Reference
Health Outcome(s)
Major Findings
Assessment Implications
cardiovascular, and immune were not endpoints that were
brought forward previously for dose-response. Generally,
data are supportive of endpoints currently modeled but
would not impact conclusions of assessment. Study will
not move forward for further evaluation.
Dangudubiyyam et al. Cardiovascular,
(2022) Developmental
PFOS exposure (0.005, 0.05, 0.5, 5, 10, or 50 (ig/mL)
via drinking water during gestation in rats
(GD4-GD20) increased vascular resistance and
increased blood pressure in dams. Decreased fetal and
placental weight were also observed. No effect
(qualitative) was observed on the number of living
fetuses.
Cardiovascular: Increased maternal mean arterial blood
pressure on GD 20 (0.5-50 (ig/mL); however, this
endpoint was not previously modeled.
Developmental: Decreased fetal body weight (0.5-
50 (ig/mL) and placental weight (10-50 (ig/mL) is
consistent with PFOS database.
Overall Assessment Conclusion: Effects are consistent
with current database. Fetal body weight, an endpoint
previously brought forward for dose-response modeling,
was decreased with increased dose levels. Study moved
forward to study quality evaluation. Maternal body weight
could significantly impact fetal body weights and be a
main driver of the results. Correspondence with the author
did not provide additional information on maternal
toxicity. Doses provided in drinking water in (ig/mL were
difficult to accurately equate to mg/kg/day without
maternal body weight and water consumption data. Not
modeled due to low confidence study quality evaluation
for developmental endpoints.
Conley et al. (2022a)
Developmental, Pregnant rats were exposed to PFOS (0.1,0.3, 1, 2, or
Hepatic 5 mg/kg/day) via oral gavage from GD8 to PND2.
Maternal gestational and postnatal body weights were
decreased, and relative liver weight was increased.
Offspring effects were also observed and included
decreased pup body weight, survival, absolute and
relative liver weight, and liver glycogen.
General Notes: A large number of dose groups at levels
that are relatively low compared with other studies in the
PFOS database.
Developmental: Decreased maternal body weight in the
high dose group is consistent with PFOS database;
however, other studies found this effect at lower dose
levels. Decreased pup weight in the two highest dose
groups is consistent with PFOS database; however, other
studies such as the Luebker et al. (2005a) critical study
found this effect at lower dose levels. Pup survival
decrease in the high dose group is consistent with PFOS
database; however, other studies found this effect at lower
dose levels.
A-148
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
Hepatic: Maternal liver weight was increased in the high
dose group while pup liver weight displayed a non-
monotonic decrease. Serum enzyme levels were all
nonsignificant.
Overall Assessment Conclusion: Effects are generally
consistent with current database, although other studies
investigating developmental endpoints tended to use lower
dose levels. Endpoints measured for hepatic (e.g., liver
weight, serum enzymes, bilirubin) were not endpoints that
were brought forward previously for dose-response (e.g.,
individual cell necrosis in the liver). Generally, data are
supportive of endpoints currently modeled but would not
impact conclusions of assessment. Study will not move
forward for further evaluation.
General Notes: Only two dose groups, both of which are
relatively high dose levels compared with other studies in
the PFOS database.
Developmental: Reduced fetal body weight is consistent
with PFOS database, although other studies use lower dose
levels and more dose groups. Alterations in pup weight are
inconsistent depending on diet.
Hepatic: Increased relative liver weight and lipid
accumulation are consistent with PFOS database.
Overall Assessment Conclusion: Effects are generally
consistent with the current database, although other studies
tended to use lower dose levels and more dose groups.
Males were the only sex studied in this paper, and the
exposure window was not for the entirety of gestation.
Generally, data are supportive of endpoints currently
modeled but would not impact conclusions of assessment.
Study will not move forward for further evaluation.
Notes: PFOS = perfluorooctane sulfonic acid; PND = postnatal day; GD = gestational day.
Shi et al. (2022) Developmental, Gestational PFOS exposure (1 or 3 mg/kg/day) via
Hepatic oral gavage from GD4.5-17.5 in mice. At GD 17.5,
male fetuses exposed to PFOA had lower body
weights and higher relative liver weights. At PND21,
male offspring in the 3 mg/kg/day group had
increased body weights.
A-149
-------
APRIL 2024
A.4 Studies Identified After Assessment Literature Searches
Studies identified after the updated literature review (February 2023) did not undergo the systematic review protocol. Studies were
reviewed for major findings and how those findings may affect the assessment. For PFOS, 17 epidemiological studies were identified
after the updated literature search in 2023 and are summarized below (Table A-48).
Table A-48. Human Studies Identified After 2023 Updated Literature Search (Published or Identified After February 2023)
Reference
Health Outcome(s)
Major Findings
Assessment Implications
Primary Epidemiologic Studies
Purdue et al. (2023) Cancer
Nested case-control study of 530 matched pairs of U.S.
Air Force Servicemen conducted using serum samples
from the DoD Serum Repository and the DoD Cancer
Registry (1990-2018). Sera was collected as a part of
routine screening and was collected every 2 yr starting
in 2004. Using the earliest pre-diagnosis sample for all
Servicemen, a nonsignificant increase in risk of TGCT
was observed comparing the fourth quartile to the first
quartile of PFOS exposure (OR = 1.8, 95% CI: 0.9,
3.6, p-trend = 0.15), after adjustment for other PFAS.
For those with multiple PFOS samples, significant
increases in risk of TGCT were observed when
comparing the third quartile of PFOS exposure
(OR = 2.8, 95% CI: 1.1, 7.0) and the fourth quartile
(OR = 4.6, 95% CI: 1.4, 15.1) to the lowest quartile,
after adjustment for other PFAS. The trend across
quartiles was significant before and after adjustment
for other PFAS (p-trend = 0.009).
Exposure to PFOS may be associated with TGCT.
Supports determination of carcinogenicity for PFOS.
Kang et al. (2023) Cardiovascular
Prospective study of 1,130 women from the Study of
Women's Health Across the Nation 45-56 yr old at
baseline (1999-2000) followed through 2016. Serum
lipids were collected at multiple timepoints over the
course of 17 yr, and high, medium, and low trajectories
for serum lipids were identified using a latent class
growth model. Exposure to branched PFOS at baseline
was associated with an increased risk of belonging to
the high trajectory class for TC compared with the low
trajectory class (OR per doubling of Sm-PFOS = 1.20,
95% CI: 1.00, 1.44). A similar positive association was
observed for total PFOS and belonging to the high
Supports an association between exposure to PFOS and
trajectories of total cholesterol. Exposure to PFOS may
be associated with trajectories of LDL cholesterol.
A-150
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
trajectory class of TC compared with the low trajectory
class (OR = 1.21, 95% CI: 0.99, 1.49). In categorical
analyses, significant increases in risk for belonging to
the high trajectory class for TC were observed for both
the second and third tertiles of PFOS exposure
compared with the first. Serum concentrations of total
PFOS were associated with significantly increased risk
of belonging to the high trajectory class for LDL-C
compared with the low trajectory class (OR = 1.28,
95% CI: 1.04, 1.56). No associations were observed
between PFOS and risk of belonging to medium or high
trajectories classes of HDL-C or TG compared with the
low trajectory classes. In PFAS mixture analyses,
significant increases in risk were observed for
belonging to the medium or high trajectory class for TC
and LDL-C. In cross-sectional analysis of baseline
PFOS concentrations and serum lipid concentrations,
Sm-PFOS was associated with increased LDL-C ((3 per
doubling in Sm-PFOS = 2.47, 95% CI: 0.53, 4.42), and
n-PFOS was associated with decreased TG. No
associations were observed for TC, HDL-C, or TG in
cross-sectional analyses of baseline data.
Tan et al. (2023) Immune Prospective study of 425 pregnant women from the Exposure to PFAS mixture may be associated with
Atlanta African American Maternal-Child Cohort. The increased cytokine and inflammatory markers. No
association between serum PFAS mixture, collected at change.
8-14 wk gestation, and serum inflammatory biomarkers
was analyzed using mixture modeling approaches,
including quantile g-computation, BKMR, BWS, and
WQS. Serum PFAS mixture was associated with
significantly increased serum concentrations of multiple
cytokines and inflammatory markers (i.e., IFN-y IL-10,
and TNF-a) in both cross-sectional analyses (i.e., 8-
14 wk gestation) and at a later follow-up visit at 24-
30 wk gestation. PFOS was noted to be consistently the
main driver of overall mixture effects across the four
methods, and the effect was reported to be stronger for
inflammatory biomarkers measured at the 24-30 wk
visit.
A-151
-------
APRIL 2024
Reference
Health Outcome(s)
Major Findings
Assessment Implications
Andersson et al.
(2023)
Immune
Prospective study of adults (20-60 yr old) from
Ronneby, Sweden comparing a group of 309 adults
with high exposure (median PFOS
concentration = 47 ng/mL) and 47 adults with
background exposure (median PFOS
concentration = 4 ng/mL). No significant association
was observed between baseline serum PFOS
concentrations and SARS-CoV-2 anti-spike antibody
levels at 5 wk post-vaccination or 6 mo post-
vaccination. Similarly, no association was observed at
5 wk or 6 mo post-vaccination for PFAS mixture
(summed PFOA, PFOS, PFHxS, and PFNA).
No change.
Siwakoti et al. Developmental Nested case-control study of 128 preterm cases and 373 No change.
(2023) term controls from the LIFECODES cohort (2006-
2008). PFOS was measured in samples collected in
early pregnancy. No significant association was
observed for preterm birth.
Zheng etal. (2023) Developmental Cohort study of 97 pregnant women enrolled in the Supports an association between elevated PFOS
Collaborative Perinatal Project (CPP) Study (1960- exposure and reduced birth weight. No change.
1966). Sample collection timing was not reported. Birth
weight was significantly reduced for mothers above the
median PFOS exposure level compared with mothers
below the median PFOS exposure level ((3 = -0.323,
p = 0.006). No significant association for birth height or
ponderal index.
Ma et al. (2023) Hepatic Cross-sectional study of 11,794 participants from Exposure to PFOS may be associated with changes to
NHANES (2003-2016). PFOS was inversely associated ALP and bilirubin. No change,
with ALP concentrations, but the trend was not
significant. Total bilirubin was significantly increased
in participants in the highest quartile of PFOS exposure
compared with the lowest (OR = 1.57, 95% CI: 1.01,
2.43, p for trend = 0.02). No associations were observed
for ALT, AST, or GGT.
A-152
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
Gump et al. (2023) Cardiovascular Cross-sectional study of 291 children (9-11 yr old) Exposure to PFOS may be associated with increased
from the EECHO study located in upstate New York resting HR and changes to PEP reactivity. No change.
(2013-2017). Elevated exposure to PFOS was
associated with significantly increased resting HR at
baseline ((3 per ln-ng/mL = 0.17, 95% CI: 0.02, 0.32).
Blood pressure reactivity to acute stress was examined
by measuring blood pressure after three acute stress
computer tasks. Elevated exposure to PFOS was
associated with significantly decreased PEP reactivity
((3 = -0.27, 95% CI: -0.42, -0.12), which was also
significant in BKMR analyses. No associations were
observed for CIMT, cfPVW, LV mass index, resting
SBP, DBP, PEP, and PP; or SBP, DBP, HR, and PP
reactivity.
Xu et al. (2023) Cardiovascular Prospective study of 129 mother-child pairs from the Exposure to PFOS may be associated with changes in
Shanghai Birth Cohort (SBC) (recruitment: 2013- blood pressure in children. No change.
2016). Exposure to PFOS was measured in cord blood
at birth, and blood pressure was measured at a follow-
up visit at 4 yr of age (2018-2021). Elevated exposure
to PFOS was significantly associated with decreased
SBP ((3 per ln-ng/mL increase = -3.10, 95% CI: -5.20,
-0.89), decreased DBP (P = -2.15, 95% CI: -4.04, -
0.33), and decreased mean artery pressure (P = -1.96,
95% CI: -3.72, -0.24). In sex-stratified analyses, all
associations were inverse for both boys and girls, but
were only significant for one sex for SBP (male), DBP
(female), and mean artery pressure (male). Exposure to
PFAS mixture was significantly associated with
decreased SBP, DBP, and mean artery pressure in
BKMR and WQS regression analyses. No significant
association observed for pulse pressure.
A-153
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
Pumarega et al. Immune Prospective study of 240 adults from Barcelona, Spain No change.
(2023) (2016-2021). Exposure to PFOS was measured in blood
collected in 2016-2017, and SARS-CoV-2 infection
was detected in nasopharyngeal swabs or blood samples
collected in 2020-2021. No association was observed
for PFOS or PFAS mixture and SARS-CoV-2
seropositivity or COVID-19 disease.
Rhee et al. (2023) Cancer No change.
Nested case-control study of 428 matched pairs of renal
cell carcinoma (RCC) cases and healthy controls from
the Multiethnic Cohort (MEC) Study. No significant
association was observed between elevated exposure to
PFOS and increased risk of RCC.
Zhang et al. (2023b) Cancer Two individual nested case-control studies conducted No change.
on 251 matched pairs from the Alpha-Tocopherol, Beta-
Carotene Cancer Prevention Study (ATBC) and 360
matched pairs from the Prostate, Lung, Colorectal and
Ovarian Cancer Screening Trial (PLCO). No significant
association was observed between elevated exposure to
PFOS and risk of PDAC in 50-69-year-old Finnish men
from ATBC (1985-1988) or 50-74-year-old American
men and women (1993-2001) from PLCO.
van Gerwen et al. Cancer Nested case-control study of 88 matched pairs of No change.
(2023) thyroid cancer patients and healthy controls from the
BioMe Biobank, medical record-linked biobank of
participants from New York City (2008-2021).
Elevated exposure to n-PFOS was associated with a
significant increase in risk of thyroid cancer (OR per
doubling of PFOS = 1.56. 95% CI: 1.17,2.15). A
borderline significant increased risk for thyroid cancer
was observed with elevated exposure to branched
PFOS (OR = 1.32, 95% CI: 0.99, 1.81). In sensitivity
analyses, the association between elevated exposure to
n-PFOS remained significant when restricting the
A-154
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
analysis to papillary thyroid cancer (n = 74 matched
pairs). Additional sensitivity analyses stratified the
sample by longitudinal cases (i.e., diagnosed >1 yr after
sample collection; n = 31 matched pairs) and cross-
sectional cases (i.e., diagnosed <1 yr after sample
collection; n = 57 matched pairs), and significant
increases in risk for thyroid cancer were observed for n-
PFOS in for both longitudinal and cross-sectional case
analyses.
Kim et al. (2023) Hepatic Cross-sectional study of 1,404 adults from the Korean No change.
National Environmental Health Survey (KoNEHS),
Cycle 3 (2015-2017). Significant positive associations
were observed between serum PFOS concentrations and
levels of ALT, AST, and GGT in single-pollutant
models. In sex-stratified analyses, associations
remained significant for men and women for ALT and
AST. For GGT, the association was only significant in
women. In analyses stratified by BMI status, significant
positive associations were observed for all three liver
enzymes in individuals with a BMI <25. For individuals
with a BMI of 25 or greater, the association was
significant for AST only. PFAS mixture was analyzed
using quantile g-computation, and significant positive
associations were observed for ALT, AST, and GGT.
Partial effects (weights) from quantile g-computation
were reported and demonstrated PFOS contributing to
the positive effects for ALT (PFOS weight: 0.25), AST
(0.36), and GGT (0.10).
Zell-Baran et al. Immune Prospective cohort study of 145 mother-child pairs fromSupports an association between elevated exposure to
(2023) the Healthy Start cohort study (enrollment: 2009-2014) PFOS and increased risk of decreased antibody response
with antibody levels measured at a follow-up visit at a in children. No change,
mean age of 5 yr old (2015-2019). An increased risk of
having a low antibody titer for measles and mumps was
observed, including a significantly increased risk for
low antibody titer for mumps (OR per 1-ln ng/mL
increase in PFOS = 1.72, 95% CI: 1.00, 2.97). In
quantile g-computation analyses, an increased risk of
having a low antibody was observed for both measles
A-155
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
and mumps, and a positive weight was observed for
PFOS for measles (weight: 0.13) and an inverse weight
for mumps (-0.24). In linear regression analyses, no
associations were observed for varicella or rubella
antibody titers. In quantile g-computation analyses, a
positive association was observed for PFAS mixture
and rubella antibody titer, however, the weight for
PFOS was inverse (weight: -0.14). PFAS mixture was
not associated with changes in varicella antibody titers.
Winquist et al. Cancer Case-cohort study of 999 participants without cancer at No change.
(2023) enrollment and 3,762 incident cancer cases within the
American Cancer Society's prospective Cancer
Prevention Study II (CPS-II) (1998-2001). A decreased
risk of hematological malignancies was observed with
elevated PFOS exposure in females, as well as
decreased risks for B-cell non-Hodgkin
leukemia/lymphoma in females and multiple myeloma
in females, males, and both sexes in analyses of
histological subtypes. Associations for bladder, kidney,
and pancreatic cancer were all nonsignificant in
analyses of the total population and sex-stratified
analyses.
Meta-analysis and Pooled Analysis Studies
Padula et al. (2023) Developmental Pooled analysis of 3,339 mother-child pairs from 1 Supports an association between exposure to PFOS and
prospective birth cohort in the ECHO program across decreased birthweight. No change,
the United States. Prenatal PFOS concentrations were
significantly associated with decreases in birthweight-
for-gestational-age z-score ((3 per ln-ng/mL increase in
PFOS = -0.14, 95% CI: -0.28, -0.002). Results were
similar in sex-stratified analyses. Nonsignificant
associations were observed for term low birth weight
(OR per ln-ng/mL increase in PFOS = 1.21, 95% CI:
0.43, 3.39) and preterm birth (OR = 1.29, 95% CI: 0.76,
2.18), and for gestational age at birth ((3: -0.16, 95% CI:
-0.40, 0.09). Associations were stronger between
increased PFOS in the first trimester and lower
birthweight-for-gestational-age z-score and increased
risk of term low birth weight and SGA. PFAS mixture
A-156
-------
APRIL 2024
Reference Health Outcome(s) Major Findings Assessment Implications
was inversely associated with birthweight-for-
gestational-age z-score (PFOS weight: 0.12) and
gestational age at birth (PFOS weight: 0.20), and the
association was not significant for gestational age at
birth. No associations were observed for SGA or LGA.
Notes: DoD = Department of Defense; TGCT = testicular germ cell tumors; PFOS = perfluorooctane sulfonic acid; PFAS = per- and polyfluoroalkyi substances; TC = total
cholesterol; OR = odds ratio; CI = confidence interval; LDL-C = low-density lipoprotein cholesterol; HDL-C = high-density lipoprotein cholesterol; TG = triglycerides;
BKMR = Bayesian Kernel Machine Regression; BWS = Bayesian Weighted Sums; WQS = weighted quantile sum regression; IL-y = interleukin gamma; IL-10 = interleukin 10;
TNF-a = tumor necrosis factor alpha; SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2; PFOA = perfluorooctanoic acid; PFHxS = perfluorohexanesulfonic acid;
PFNA = perfluorononanoic acid; CPP = Collaborative Perinatal Project; NHANES = National Health and Nutrition Examination Survey; ALP = alkaline phosphatase;
ALT = alanine transaminase; AST = aspartate transaminase; GGT = gamma-glutamyltransferase; EECHO = Environmental Exposures and Child Health Outcomes; HR = heart
rate; PEP = pre-ejection period; CIMT = carotid intima-media thickness; cfPVW = carotid-femoral pulse wave velocity; LV = left ventricular; SBP = systolic blood pressure;
DBP = diastolic blood pressure; PP = pulse pressure; SBC = Shanghai Birth Cohort; SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2; COVID-19 = coronavirus
disease 2019; ATBC = Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study; PLCO = Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial; PDAC = pancreatic
ductal adenocarcinoma; CPS-II = Cancer Prevention Study II; ECHO = Environmental influences on Child Health Outcomes; In = natural log; CI = confidence interval;
OR = odds ratio; SGA = small for gestational age; LGA = large-for-gestational-age
A-157
-------
APRIL 2024
Appendix B. Detailed Toxicokinetics
B.l Absorption
A summary of studies that provide information on perfluorooctane sulfonic acid (PFOS)
absorption from recent systematic literature search and review efforts conducted after publication
of the 2016 PFOS Health Effects Support Document (HESD) is shown in Figure B-l.
Evide n ce Stream G ra n d Total
Animal
5
Human
2
In Vitro
1
Grand Total
8
Figure B-l. Summary of PFOS Absorption Studies
Interactive figure and additional study details available on HAWC.
a Figure does not include studies discussed in the 2016 PFOS HESD or those that solely provided background information on
toxicokinetics.
b Select reviews are included in the figure but are not discussed in the text.
B.l.l Cellular Uptake
Lipid binding may influence PFOS accumulation in various cell types relevant to absorption as
well as distribution. Sanchez Garcia et al. (2018) compared PFOA and PFOS in their ability to
accumulate and be retained in cells including lung epithelial cells (NCI-H292). Cellular
accumulation and retention of PFOS was observed in lung cells at levels higher than
azithromycin-dihydrate (AZI), a lysosomotropic cationic amphiphilic drug used as a reference
compound. In contrast, PFOA only accumulated to very low levels (Table B-l). Phospholipid
binding was assessed by measuring the relative affinity for a phosphatidylcholine (PC)-coated
column at pH 7.4 to calculate a chromatographic index (CHIIAM7.4). Lipid binding (LogD7.4)
was determined by measuring the relative affinity of compounds for a C18-coated liquid
chromatography column at pH 7.4. LogP values obtained from the PubChem database were used
as a comparative lipophilicity measure. Phospholipophilicity correlated (r2 = 0.75) to cellular
accumulation better than other lipophilicity measures. The extent to which PFOS
phospholipophilicity influences absorption through the GI tract, lungs, or skin is unknown.
Table B-l. Cellular Accumulation and Retention Relative to Lipophilicity and
Phospholipophilicity as Reported by Sanchez Garcia et al. (2018)
Cellular Accumulation and Retention
Lipophilicity
Chemical
Accumulation in Lung Retention in Lung
Epithelium (% AZI) Epithelium
Phospholipid Binding
(CHIIAM7.4)
Lipid Binding
(LogD7.4)
LogP
PFOS
313 ±101* 26 ±4
39 ± 3*
2.33 ±0.11*
5
PFOA
15 ±3 ND
29 ± 1
1.29 ±0.02
4.9
Notes: AZI = azithromycin-dihydrate; ND = not determined.
* Statistically significant at p < 0.05 from PFOA.
B-l
-------
APRIL 2024
The study by Sanchez Garcia et al. (2018) raises the possibility of passive update of PFOS
into cells. This is consistent with observations that cells transfected with vector only, could
take up PFOS, albeit at lower levels than cells transfected with PFOS-specific transporters
(discussed further in Section B.4.2.1). Ebert et al. (2020) determined membrane/water
partition coefficients (Kmem/w) for PFOS and examined passible permeation into cells by
measuring the passive anionic permeability (Pion) through planar lipid bilayers. Membrane
permeability and partition coefficients were predicted using an approach developed to model
molecules in micellar systems and biomembranes (COSMOmic and related tools, Klamt et
al., 2008). The predicted log (Kmem/w/[L/kg]) for PFOS was 4.69, similar to the
experimentally determined value of 4.89 ± 0.30. Kmem/w values increase with increasing chain
length, reflecting increased surface area for van der Waals interactions. The authors observed
that perfluoroalkanesulfonic acids (PFSAs) adsorb about 1.2 log units more strongly to the
membrane than perfluorocarboxylates (PFCAs) with the same number of perfluorinated
carbons. Permeability showed the same chain-length dependence as Kmem/w values. The
predicted anionic permeability (logPion/[cm/s]) for PFOS ranged from -4.74 to -3.58,
considered high enough to explain observed cellular uptake by passive diffusion in the
absence of active uptake processes. The extent to which passive uptake influences absorption
in vivo remains to be determined.
B.1.2 Oral Exposure
Chang et al. (2012) administered a single oral dose of 4.2 mg/kg of [14C]PFOSin solution to three
male Sprague-Dawley rats. At 48 hours after dosing, only 9.08 ± 0.51% of the total [14C]PFOS
dose was recovered across digestive tract, feces, or urine, while the carcass retained 94.2 ± 5.1%,
indicating that the PFOS was largely absorbed.
B.1.3 Inhalation Exposure
An acute median lethal concentration (LCso) study in rats indicates that PFOS absorption occurs
after inhalation exposures; however, pharmacokinetic data were not included in the published
report (Rusch, 1979). The analytical methods for measuring PFOS in animals were limited at the
time the study was conducted. More recent data on PFOS absorption following inhalation
exposure are not available.
B.1.4 Dermal Exposure
The literature contains no studies on the dermal absorption of PFOS.
B.1.5 Developmental Exposure
The literature contains no studies on PFOS absorption following developmental exposure.
Additional information on PFOS distribution during reproduction and development is found in
Section B.2.3.
B.1.6 Bioavailability
Toxicokinetic parameters informing absorption were derived by comparing oral to intravenous
(IV) dosing in rats (Kim et al., 2016b). Sprague-Dawley rats were administered 2 mg/kg by
B-2
-------
APRIL 2024
either the IV or oral route. Urine and feces were collected weekly, and blood was collected at 10
time points over the first day and then up to 70 days after exposure. In contrast to the sex
differences observed for PFOA, the time to reach the maximum PFOS plasma concentration
(Tmax) following oral exposure was similar in males and females (10.8 hours and 11.5 hours,
respectively). In a similar study (Huang et al., 2019a), male and female Sprague-Dawley rats
were administered a single dose of 2 mg/kg by IV injection or a single dose of 2 mg/kg or
20 mg/kg by oral gavage and observed from 5 minutes to 20 weeks after dosing. The maximal
plasma concentrations (Cmax) were similar for oral gavage and IV administration of 2 mg/kg,
and Tmax values were consistent with those observed by Kim and colleagues (14.3 hours and
12.2 hours in males and females, respectively).
The results from these studies are compared in Table B-2. Both studies found very high (>100%)
bioavailability in rats (calculated by dividing the dose-adjusted gavage area under the curve
(AUC) by the IV AUC). Huang and colleagues speculate that the >100% bioavailability after
oral dosing is due to enterohepatic circulation that occurs after gavage but not IV administration.
The Tmax values ranged from 10.8 to 14.3 hours and was slightly longer in the Huang study for
both males and females. Neither bioavailability nor Tmax exhibited sex-specific differences.
However, Huang et al. did observe slightly higher Cmax concentrations in females relative to
males.
Table B-2. PFOS Parameters From Toxicokinetic Studies Informing Bioavailability in
Sprague-Dawley Rats
Study
Dose (mg/kg)
Route
Sex
Cmax (ng/mL)a
Tmax (hours)b
Kim et al. (2016b)
2
Oral
Male
6.71 ±0.30
10.8 ±0.96
IV
Male
5.23 ±0.24
NA
Oral
Female
6.66 ±0.29
11.52 ± 1.2
IV
Female
5.69 ±0.33
NA
Huang et al. (2019a)
2
Oral
Male
5.00 ±5.00
14.3 ±2.7
IV
Male
5.00 ± 5.00
NA
Oral
Female
10.00 ±5.00
12.2 ±5.2
IV
Female
5.00 ± 5.00
NA
Notes: Cmax=maximum serum concentration, IV = intravenous, NA = not applicable, Tmax = time to Cmax.
a Converted published Cmax (rnM) to Cmax (|xg/mL) for Huang et al. (2019a).
b Converted published Tmax (days) to Tmax (hours) for Kim et al. (2016b).
B.2 Distribution
A summary of studies that provide information on PFOS distribution from recent systematic
literature search and review efforts conducted after publication of the 2016 PFOS HESD is
shown in Figure B-2.
Evide n ce Stream G ra n d Total
Animal
20
Human
70
In Vitro
13
Grand Total
97
Figure B-2. Summary of PFOS Distribution Studies
B-3
-------
APRIL 2024
Interactive figure and additional study details available on HAWC.
a Figure does not include studies discussed in the 2016 PFOS HESD or those that solely provided background information on
toxicokinetics.
b Select reviews are included in the figure but are not discussed in the text.
B.2.1 Protein Binding
Kerstner-Wood et al. (2003) examined the in vitro protein binding of PFOS in rat, monkey, and
human plasma at concentrations of 1 ppm to 500 ppm and found that PFOS was bound to plasma
protein in all three species. When incubated with separate human-derived plasma protein
fractions, PFOS was highly bound (99.8%) to albumin and showed affinity for low-density
lipoproteins (95.6%) with some binding to alpha-globulins (59.4%) and gamma-globulins
(24.P/o). Low levels of binding to alpha-2-macroglobulin and transferrin were measured when
the protein concentrations were approximately 10%> of physiological concentration.
Zhang et al. (2009) conducted an in vitro study using equilibrium dialysis, fluorophotometry,
isothermal titration calorimetry, and circular dichroism to characterize interactions between
PFOS with serum albumin and DNA. The authors reported that serum albumin could bind up to
45 moles of PFOS/mole of protein and 0.36 moles/base pair of DNA. The binding ratio increased
with increasing PFOS concentrations and decreasing solution pH. The authors concluded that the
interactions between serum albumin and PFOS were the results of surface electrostatic
interactions between the sulfonate functional group and the positively charged side chains of
lysine and arginine. Hydrogen binding interactions between the negative dipoles (fluorine) of the
PFOS carbon-fluorine bonds could also play a role in the noncovalent bonding of PFOS with
serum albumin.
Chen and Guo (2009) investigated the binding of PFOS to human serum albumin using site-
specific fluorescence and found that PFOS induced fluorescence quenching indicative of
binding. A binding constant of 2.2 x 104 M 1 and a binding ratio of PFOS to human albumin of
14 moles PFOS/mole albumin were calculated. Fluorescence displacement measurements were
used to study the interaction between PFOS and two high-affinity drug binding sites on human
serum albumin known as Sudlow's drug Site I and Site II. The findings indicated that PFOS has
binding sites that are similar to those identified for fatty acids.
Salvalaglio et al. (2010) used molecular modeling to determine the structure and energy of PFOS
binding sites for human serum albumin. The binding sites impacted were ones identified as
human serum albumin fatty acid binding sites. The most populated albumin binding site for
PFOS was dominated by van der Waals interactions. The PFOS binding site with the highest
energy (-8.8 kcal/mole) was located near the tip of the tryptophan 214 binding site, and the
maximum number of ligands that could bind to human serum albumin for PFOS was 11.
D'Alessandro et al. (2013) used electrospray ionization mass spectrometry to evaluate PFOS
binding to bovine serum albumin. Using this approach, the maximum number of PFOS binding
sites was estimated as 11, but the data on collision-induced PFOS removal was more consistent
with 7 binding sites. This study also showed that PFOS competes with ibuprofen for its site when
the PFOS:ibuprofen ratio is >0.5 moles: 1 mole. In addition, when the binding site is occupied by
PFOS, ibuprofen is unable to bind. Zhang et al. (2009) conducted a similar study of the impact of
PFOS on the ability of serum albumin to bind vitamin B2 (riboflavin) and found that, under
normal physiological conditions, PFOS decreased the binding ratio of serum albumin for
B-4
-------
APRIL 2024
riboflavin in vitro. These data suggest that PFOS can alter the pharmacokinetics and
pharmacodynamics of medicinal and natural substances that share a common site on albumin.
Beesoon and Martin (2015) examined differences in the binding of linear and branched chain
isomers of PFOS to calf serum albumin and human serum proteins. The linear PFOS molecule
was found to bind more strongly to calf serum albumin than the branched chain isomers. When
arranged in order of increasing binding, the order was 3m < 4m < lm < 5m < 6m (iso) < linear.
In the isomer-specific binding to spiked total human serum protein, the lm branched PFOS
isomer bound most strongly and the 4m branched PFOS isomer the least.
Liu et al. (2017) used spectroscopy, molecular modeling, and calorimetry techniques to evaluate
the mechanism by which PFOS interacts with human serum albumin through hydrogen bonds
and electrostatic interactions. PFOS binding to albumin is a spontaneous exothermic process
driven by electrostatic interactions. This study observed that the backbone and secondary
structure of albumin did not significantly change after exposure to PFOS; however, results
suggest the interaction with PFOS changed the local structure around the esterase active site. A
molecular docking study indicated that PFOS binds to the active center Arg 410 residue in
albumin. This corresponded to a 28.6% decrease in esterase activity. By examining multiple
PFAS, esterase activity of albumin was found to decrease with the shortening of the carbon chain
and the authors suggest this may correlate with toxicity.
Sheng et al. (2020) measured uptake of PFOS in human placental choriocarcinoma (JAr) cells in
the presence or absence of human serum albumin for 48 hours. PFOS concentrations in the
culture medium decreased by 21.4%, 78.1%, and 92.8% with the addition of 0.5 [xM, 10 [xM, and
200 [xM albumin, respectively. This result supports a paradigm in which binding of albumin to
PFOS in the culture medium blocked their entrance into the cells. The binding affinity (Kd) of
PFOS to human serum albumin was calculated to be 30.7 [xM. Using a limited proteolysis
technique, the authors identified the core albumin peptides that bind to PFOS as residues 189—
457.
Binding to albumin and other serum proteins may affect transfer of PFOS from maternal blood to
the fetus. Gao et al. (2019) correlated placental transfer with experimentally measured
dissociation constants (Kd) to human serum binding proteins, serum albumin, and L-FABP. For
PFOS, Kd values were calculated to be 49 ± 8 [xM for serum binding proteins, 38 ± 5 [xM for
albumin, and 81 ± 7 [xM for L-FABP. These Kd values significantly correlated with placental
transfer efficiencies measured in 132 maternal blood-cord blood pairs from subjects in Beijing,
China, suggesting serum and binding proteins, especially albumin, play an important role in
placental transfer efficiency. The authors suggested that lower cord blood albumin levels
compared with maternal blood albumin levels may set up a competition for PFOS binding on
either side of the placenta.
Since there is effectively a competition between PFOS binding in maternal serum versus cord
blood, lower cord blood albumin levels compared with maternal blood albumin levels are likely
to reduce transfer from maternal serum across the placenta. Consistent with this hypothesis, Pan
et al. (2017) found that the concentration of cord serum albumin was associated with higher
transfer efficiencies (increase of 4.1% (CI: 2.7, 5.4) per 1 g/L albumin). However, maternal
serum albumin concentration was associated with reduced transfer efficiency (decrease of 3.4%
(CI: -5.0, -1.8) per 1 g/L albumin). Because albumin cannot cross the placental barrier, the
B-5
-------
APRIL 2024
authors speculate that binding of PFOS to maternal serum albumin can reduce the free PFOS
available to move across the barrier through passive diffusion. Similarly, higher fetal albumin
levels will lead to less free PFOS in cord blood, which may facilitate the rate of placental transfer
via passive diffusion.
PFOS also binds to intracellular proteins. Luebker et al. (2002), Zhang et al. (2013a), and Yang
et al. (2020a) conducted in vitro studies that examined the binding of PFOS and other PFAS to
the liver fatty acid binding protein (L-FABP). L-FABP is an intracellular lipid carrier protein that
reversibly binds long-chain fatty acids, phospholipids, and an assortment of peroxisome
proliferators (Erol et al., 2004) and constitutes 2%-5% of the cytosolic protein in hepatocytes.
Luebker et al. (2002) evaluated the ability of perfluorinated chemicals to displace a fluorescent
substrate from L-FABP and reported that PFOA exhibited some binding to L-FABP, but its
binding potential was about 50% less than that of PFOS and far less than that of oleic acid.
Zhang et al. (2013a) cloned the human L-FABP gene and used it to produce purified protein for
evaluation of the binding of PFOA and PFOS. The median inhibiting concentration (ICso) values
for PFOA and PFOS were 9.0 ± 0.7 [j,mol and 3.3 ± 0.1 [j,mol, respectively, suggesting that
PFOA has a lower binding affinity than PFOS. PFOA was bound to the carrier protein in a 1:1
ratio, and the interaction was mediated by electrostatic interactions and hydrogen binding with
the fatty acid binding site. Using size-exclusion column coelution and nontarget analysis to
identify additional PFAS ligands from contaminated environmental sources, Yang et al. (2020a)
also found that that both polar and hydrophobic interactions are crucial for binding affinities to
L-FABP for PFOA and PFOS.
A computational modeling approach that combined molecular docking and molecular dynamics
simulation techniques was used to estimate the relative binding of affinity of PFOS for human
and rat L-FABP (Cheng and Ng, 2018). The authors found that predicted free energies correlated
well with binding affinities measured in three previous studies (Sheng et al., 2018; Zhang et al.,
2013a; Woodcroft et al., 2010). Key residues contributing to free binding energies (AGbind) for L-
FABP include ARG 122, SER 124, and ILE 52 (human) and TYR 120, ARG 122, ILE 60, and
ILE 53 (rat).
B.2.2 Tissue Distribution
B.2.2.1 Human Studies
Human blood is a known site of PFOS accumulation. A recent example measured PFAS in blood
samples from 344 Wilmington, NC residents (289 adults and 55 children) exposed to
contaminated drinking water from release of PFAS chemicals into the Cape Fear River between
1980 and 2017. The mean serum PFOS concentration was 9.4 ng/mL in adults and 5.1 ng/mL in
children (Kotlarz et al., 2020). In an analysis of Faroese children (ages 5 to 14) from three birth
cohorts, PFOS accounted for the largest fraction (54%-74%) of the PFAS in serum, followed by
PFOA (11%-24%) (Dassuncao et al., 2018). A mean serum PFOS concentration of 6.9 ng/mL
was measured in 41 Norwegian women (Haug et al., 2011). Using adjusted multiple linear
regression models, PFOS serum concentrations were significantly correlated to the number of
months since breastfeeding ended and consumption of fish, but not age or weight of participants.
PFOS accumulation in blood impacts distribution to various tissues and organs, but few studies
have examined PFOS partitioning to human blood fractions. Forsthuber et al., (2020) measured
B-6
-------
APRIL 2024
the distribution of PFOS in blood fractions including plasma, albumin, and lipoprotein fractions
(e.g., very low-density lipoproteins (VLDL), low-density lipoproteins (LDL), and high-density
lipoproteins (HDL)). Blood from four young healthy volunteers (two women, two men, 23-
31 years old) were separated into fractions using size fractionation (for proteins) and serial
ultracentrifugation. Results found that albumin was the most important carrier for PFOS with
4.3 ± 2.2 ng/mL present in this fraction. In contrast, the amount of PFOS associated with VLDL,
LDL and HDL fractions was below the limit of quantification (LOQ), 0.1 ±0.1 ng/mL, and
0.16 ± 0.06 ng/mL, respectively.
Jin et al. (2016) analyzed 60 blood samples from a Chinese population, and three whole blood
samples from an exposed Canadian family to investigate the partitioning of PFAS of different
chain lengths and their major isomers between human blood and plasma. Increasing chain length
for PFAS correlated with an increased mass fraction in human plasma (FP) from C6 (mean 0.24)
to CI 1 (0.87). The PFOS plasma:whole blood ratio in the Jin et al. (2016) study was lower
(1.5 ± 0.42) compared with the mean plasma:whole blood (2.2-2.3) (Ehresman et al., 2007) and
serum:whole blood (1.2-2.3) (Hanssen et al., 2013; Karrman et al., 2006) ratios previously
reported. Linear isomers of PFOS had lower mean FP than their corresponding total branched
isomers. In blood samples obtained from three highly exposed Canadian subjects, the highest
levels of PFOS were measured in plasma (0.14 ng/mL) compared with red blood cells (RBCs,
0.04 ng/mL) and in washed RBCs (0.04 ng/mL). The authors suggested that these values could
be used as more accurate conversion factors when converting concentrations between whole
blood and plasma.
Fractionation to blood fractions was also examined in 61 male and female participants from
Oslo, Norway in 2013-2014 (Poothong et al., 2017). The median relative PFAS compositions in
serum, plasma, and whole blood were dominated by PFOS, followed by PFOA (representing
60%-70% of blood PFAS), relative to the other 23 PFAS chemicals analyzed. Median PFOS
concentrations in plasma, serum, and whole blood were 5.24 ng/mL, 4.77 ng/mL, and
2.85 ng/mL, respectively. Similar to other studies, PFOS preferentially accumulated in plasma
relative to serum and whole blood; this result suggests that the common practice of multiplying
by a factor of 2 to convert the concentrations in whole blood to serum or plasma will not provide
accurate estimates for PFOS.
B.2.2.1.1 Distribution in Tissues
No clinical studies are available that examined tissue distribution in humans following
administration of a controlled dose of PFOS. However, samples collected in biomonitoring and
epidemiological studies provide data showing distribution of PFOS.
In humans, PFOS distributes primarily to the liver and blood. Olsen et al. (2003b) sampled both
liver and serum from cadavers for PFOS and found a good correlation between samples from the
same subject. There were no sex- or age group-specific differences in PFOS concentrations. In
another study, Karrman et al. (2010) identified PFOS in postmortem liver samples (n = 12; 6
males, 6 females, aged 27-79 years) with a mean concentration of 26.6 ng/g tissue.
Perez et al. (2013) collected tissue samples (liver, kidney, brain, lung, and bone) in the first
24 hours after death from 20 adult subjects (aged 28-83 years) who had been living in Catalonia,
Spain. PFOS was present in 90% of the samples but could be quantified in only 20% (median
1.9 ng/g). PFOS accumulated primarily in the liver (104 ng/g), kidney (75.6 ng/g), and lung
B-7
-------
APRIL 2024
(29.1 ng/g), and brain (4.9 ng/g), with levels below the limit of detection (LOD) in the bone.
Maestri et al. (2006) examined pooled postmortem tissues from five males and two females from
northern Italy ranging in age from 12 to 83 years. Of the 12 tissues analyzed, the highest PFOS
levels were detected in liver, lung, and pituitary gland (13.6, 7.9, and 7.6 ng/g, respectively) and
the lowest levels were detected in skeletal muscle, brain, and basal ganglia (1.0, 1.3, and
1.2 ng/g, respectively). Both linear and branched PFOS were observed in these tissues, with
linear:branched PFOS peak area ratios ranging from 1.6 (in blood) to 4.8 (in basal ganglia).
PFOS was also detected in cranium, rib bone, and tibia bone marrow samples from a cadaver of
a 46-year-old, and from biopsies from live subjects in a bone bank in Finland (Koskela et al.,
2017). However, PFOS was below the detection limit in other bone tissues (e.g., humerus, femur,
fibula) but was detected in soft tissues including brain, liver, and lung.
PFOS also accumulates in follicular fluid (Kang et al., 2020) and gonads (Maestri et al., 2006).
Kang et al. (Kang et al., 2020) measured a concentration of 4.82 ± 3.07 ng/mL (geometric mean)
in follicular fluid samples from 28 women undergoing oocyte retrieval for in vitro fertilization
procedures. A positive correlation was found between paired serum and follicular fluid samples
for PFOS (r2 = 0.78, p < 0.001), though PFOA correlations were even stronger (r2 = 0.93,
p < 0.001). Maestri et al. (2006) measured a mean concentration of 3.4 ng/g of the linear PFOS
isoform in pooled gonads collected postmortem from subjects in northern Italy (five males and
two females aged from 12 to 83 years). Exposure of oocytes and gonads to PFOS raises the
possibility of reproductive toxicity in humans.
Stein et al. (2012) compared PFAS levels in paired samples of maternal serum and amniotic fluid
from 28 females in their second trimester of pregnancy. PFOS was detected in all serum samples
(0.0036-0.0287 (j,g/mL) and in nine amniotic fluid samples (0.0002-0.0018 (j,g/mL). The
Spearman correlation coefficient between the serum and amniotic fluid levels was 0.76
(p = 0.01), indicating a direct relationship between PFOS levels in blood and amniotic fluid. The
median ratio of maternal serum:amniotic fluid concentration was 25.5.
Two studies examined accumulation of PFOS in cerebrospinal fluid and serum (Wang et al.,
2018b; Harada et al., 2007). In both studies, PFOS levels in cerebrospinal fluid were two orders
of magnitude lower than in the serum. These results indicate that PFOS does not easily cross the
adult blood-brain barrier.
PFOS has been detected in both umbilical cord blood and breast milk indicating that maternal
transfer occurs (von Ehrenstein et al., 2009; Volkel et al., 2008; Apelberg et al., 2007a). Karrman
et al. (2010) identified PFOS in breast milk samples from healthy females (n = 10; aged 30-
39 years), and the levels in milk (mean 0.12 ng/mL) were low compared with levels in the liver.
Balk et al. (2019) developed a one-compartment PBPK model to analyze intake in children from
1 to 10.5 years of age. Measured serum concentrations were derived from a subgroup of a
longitudinal child study (LUKAS 2) (Koponen et al., 2018). Estimated daily intakes ranged
between 0.16 and 0.55 ng/kg bw/day for low and high exposure scenarios. Measured PFOS
serum concentrations (5th-95th percentile) ranged from 1.2-4.1 ng/mL (age 6) to 0.84-
2.8 ng/mL (age 10.5). The model reconstructed median PFOS serum concentrations compared
with corresponding measured median serum concentrations and predicted that growth dilution
contributed from 63% to 77% of total PFOS loss, with elimination pathways accounting for the
remaining PFOS loss in children.
B-8
-------
APRIL 2024
B.2.2.2 Animal Studies
Studies of tissue distribution are available for several species of animals including nonhuman
primates, rats, and, to a lesser extent, mice. Studies of nonhuman primates indicate that levels of
PFOS in serum accumulate in a dose-dependent manner. While data are limited on liver
accumulation of PFOS in monkeys, PFOS accumulation in the liver appears to be similar to that
of serum, if not slightly lower. Several rodent studies identified the liver as a major site of
accumulation, and that PFOS distributes to a wide range of tissues including kidney, heart, and
lungs, and spleen. Interestingly, PFOS has been measured in moderate quantities in both the
brain and testicles of rodents, indicating that it does cross the blood-brain barrier and blood-testis
barrier. While monkeys had nearly a 1:1 liver-to-serum ratio, rodent models were observed to
contain far more PFOS in liver than serum.
B.2.2.2.1 Nonhuman Primates
Two long-term studies in monkeys examined PFOS accumulation in the serum and liver. Seacat
et al. (2002) administered 0 mg/kg/day, 0.03 mg/kg/day, 0.15 mg/kg/day, or 0.75 mg/kg/day
PFOS orally in a capsule by intragastric intubation to young-adult to adult cynomolgus monkeys
for 26 weeks. Serum and tissues were collected at necropsy. The dosing was followed by a 52-
week recovery period in two animals in the control, 0.15 mg/kg/day, and 0.75 mg/kg/day groups.
Serum PFOS measurements demonstrated a linear increase with dosing duration in the
0.03 mg/kg/day and 0.15 mg/kg/day groups and a nonlinear increase in the 0.75 mg/kg/day
group. Levels in the high-dose group appeared to plateau after about 100 days (14 weeks) but
began to decline sometime after week 37. The average percent of the cumulative dose of PFOS
in the liver at the end of treatment ranged from 4.4% to 8.7% with no difference by dose group
or sex. At the two lower doses, serum levels were comparable in the males and females, whereas
at 0.75 mg/kg/day, levels were generally elevated in the males compared with females. Only the
highest dose group appeared to reach a serum steady state at week 16. In the 0.03 mg/kg/day
groups, the serum levels continued to increase temporally until week 27 when serum sampling
stopped for that cohort. Once dosing ceased, serum levels declined in all animals that continued
in the study.
In the second study conducted in cynomolgus monkeys (Chang et al., 2017), animals were given
PFOS doses to reach target serum concentrations of 70 |ig/mL or 100 [j,g/mL that were chosen to
match levels of the medium- and high-dose groups from Seacat et al. (2002). The control group
(n = 6/sex) was dosed with vehicle, the low-dose group (n = 6/sex) received a single dose of
9 mg/kg PFOS on day 106 of the study, and the high-dose group (n = 4-6/sex) received three
separate PFOS doses (11-17.2 mg/kg) on days 43, 288, and 358. Measurements of serum PFOS
indicate that male and female monkeys reached the target dose of 70 [j,g/mL and 100 |ig/mL on
day 113 and 50, respectively. Male and female animals in the high-dose group reached peak
PFOS serum levels of 160-165 |ig/mL on day 365. Consistent with the previous study, no sex
differences were found. At the end of the experiment, the animals were reported to have a 1:1
PFOS liver:serum ratio, while the previous Seacat et al. (2002) study reported a ratio closer to
2:1. Chang et al. (2017) attributed these differences in findings to the dosing approaches and
regimens used in the two studies (gelatin capsule vs. gastric intubation).
B-9
-------
APRIL 2024
B.2.2.2.2 Rats
Numerous studies have been performed on models of PFOS distribution in rats. These studies
range from acute (hours) to longer-term studies (20 weeks) and include various levels of dosing.
Distribution is measured primarily in serum, liver, and lungs, but approaches were used to
measure brain distribution as well.
Martin et al. (2007) administered PFOS (10 mg/kg/day) to adult male Sprague-Dawley rats for 1,
3, or 5 days by gavage and determined the liver and serum levels. Mean liver PFOS levels were
83 ± 5 (J,g/g, 229 ± 10 (J,g/g, and 401 ± 21 j_ig/g after 1, 3, or 5 daily doses, respectively. Mean
serum concentrations were 23 ± 2.8 j_ig/g and 87.7 ±4.1 |ig/mL after 1 and 3 days of dosing,
respectively. Day 5 serum levels were not available through the publication. This study observed
a liver:serum ratio of nearly 3:1. Liver PFOS concentrations also exhibit a dose-dependency in
male Sprague-Dawley (SD) rats administered 1 or 10 mg/kg/day PFOS by oral gavage for
28 days with PFOS concentrations 27- and 54-fold higher than those of control rats (Han et al.,
2018a).
In another acute study performed by Yu et al. (2011), female Wistar rats were administered
doses of PFOS (0, 0.2, 1.0, or 3.0 mg/kg/day) dissolved in 0.5% Tween 20 for 5 consecutive
days. Blood and bile were collected 24 hours after the last dose was given. Data indicate that
there is a linear dose-dependent increase in both serum and bile, which likely reflects levels in
liver.
A 28-day toxicity study by NTP exemplifies patterns of PFOS accumulation in blood and liver
(NTP, 2019). Male and female Sprague-Dawley rats were administered daily doses of
0 mg/kg/day, 0.312 mg/kg/day, 0.625 mg/kg/day, 1.25 mg/kg/day, 2.5 mg/kg/day, or
5 mg/kg/day of PFOS by oral gavage. Plasma and liver concentrations were analyzed
approximately 24 hours after the last dose. A dose-dependent increase in plasma concentrations
of PFOS was observed in both males and females. In contrast to studies with PFOA, plasma
PFOS concentrations in females were generally similar to males, and dose-normalized plasma
concentrations ([iM/mmol/kg/day) in males and females were within 1.5-fold across the dose
groups. The lowest dose-normalized concentration was observed in the highest dose group in
both sexes. In males, PFOS concentrations in plasma were 23.73 ± 1.11 [j,g/mL and
318.2 ± 8.87 [j,g/mL at the lowest and highest doses, respectively. In females, these values were
30.53 ± 0.92 [j,g/mL and 413.56 ± 8.07 [j,g/mL at the lowest and highest doses, respectively.
However, there were quantifiable levels of PFOS in female controls that were 562 times lower
than the lowest dose administered and required caution in interpreting these findings.
Concentrations in livers of males increased with increasing dose, but when normalized with
dose, there was a steady decrease as dose increased. This corresponded with a decreasing
liver:plasma ratio as dose increased. Livenplasma ratios, measured only in males, were
3.76 ± 0.24 at the lowest dose and 2.74 ± 0.08 at the highest dose.
Additional studies have been performed that expand on PFOS dosing, time of treatment, and
organ distribution. Cui et al. (2009) delivered 5 or 20 mg/kg/day of PFOS via oral gavage to 3-
month-old Sprague-Dawley rats. At the end of dosing (28 days), serum and organ concentrations
were measured (Table B-3). No blood samples were available at the 20 mg/kg/day dose due to
animal deaths in this group. The liver appeared to have by far the highest concentration of PFOS
at both 5 mg/kg/day and 20 mg/kg/day. Levels in the heart were approximately half the
B-10
-------
APRIL 2024
concentration observed in liver followed by the kidney, serum, and lungs. Of the organs
examined, testicles and spleen exhibited the lowest PFOS levels. Of note was the differential
accumulation by organ and dose. For liver, kidney, and heart, 2-3-fold increases in PFOS
concentrations were observed between the low and high doses even though the high dose was 4
times higher than the low dose. Interestingly, the brain and lungs were most susceptible to the
increase in dose by accumulating 10- and 5-fold more PFOS, respectively.
Table B-3. Concentrations of PFOS in Various Tissues of Male Sprague-Dawley Rats
Exposed to PFOS by Gavage for 28 Days as Reported by Cui et al. (2009)
Tissue3
0 mg/kg/day
5 mg/kg/day
20 mg/kg/day
Blood ((ig/mL)
ND
72.0 ±25.7
No sampleb
Liver (jig/g)
ND
345 ± 40
648 ± 17
Kidney (|ig/g)
ND
93.9 ± 13.6
248 ± 26
Lung (|ig/g)
ND
46.6 ± 17.8
228 ± 122
Heart (|ig/g)
ND
168 ± 17
497 ± 64
Spleen (jig/g)
ND
38.5 ± 11.8
167 ± 64
Testicle (|ig/g)
ND
39.5 ± 10.0
127 ± 11
Brain (jig/g)
ND
13.6 ± 1.0
146 ±34
Notes: ND = not detected.
aData are presented as mean ± standard deviation.
b Animal deaths in this group precluded blood measurements.
In a similar study conducted by Curran et al. (2008), male and female Sprague-Dawley rats were
administered 0 mg/kg/day, 2 mg/kg/day, 20 mg/kg/day, 50 mg/kg/day, or 100 mg/kg/day via
feed for 28 days (Table B-4). The highest PFOS concentration was found in the liver at all doses,
accounting for 70%-80% of total distribution measured in males and 65%- 80% of total
distribution in females. The spleen and heart also contained notable levels of PFOS, however,
accumulation in the heart was approximately 25% less than the amount in spleen. PFOS in
animal livers followed a linear dose-dependent distribution between 2 mg/kg/day and
20 mg/kg/day; however, this linearity was lost between the 20 mg/kg/day, 50 mg/kg/day, and
100 mg/kg/day dose escalation. This could be due to an increase in excretion or changes in
distribution to other organs that were not measured in this study. No consistent differences
between the sexes were found, however, female rats generally had higher levels of PFOS in the
heart and spleen at all doses.
B-ll
-------
APRIL 2024
Table B-4. Concentrations of PFOS in Various Tissues of Male and Female Sprague-Dawley Rats Exposed to PFOS by Feed
for 28 Days as Reported by Curran et al. (2008)
0 mg/kg/day 2 mg/kg/day 20 mg/kg/day 50 mg/kg/day 100 mg/kg/day
raiiuueier
Males
Females
Males
Females
Males
Females
Males
Females
Males
Females
PFOS consumption
(mg/kg bw/day)
0
0
0.14 ±0.02
0.15 ±0.02
1.33 ±0.24
1.43 ±0.24
3.21 ±0.57
3.73 ±0.57
6.34 ± 1.35
7.58 ±0.68
Spleen
(MS^g)
0.27 ±0.36
2.08 ±4.17
6.07 ± 1.85
7.94 ±3.76
45.27 ±2.16
70.03 ±36.66
122.51 ±7.83
139.45 ± 15.44
230.73 ± 11.47
294.96 ±26.66
Heart
(Hg/g)
0.10 ± 0.14
1.42 ±2.91
4.67 ± 1.73
6.54 ±3.07
33.00 ±3.44
54.65 ±30.89
90.28 ±4.95
107.53 ±6.24
154.13 ± 11.78
214.45 ± 17.58
Serum
(MS^g)
0.47 ±0.27
0.95 ±0.51
0.95 ±0.13
1.50 ±0.23
13.45 ± 1.48
15.40 ± 1.56
20.93 ±2.36
31.93 ±3.59
29.88 ±3.53
43.20 ±3.95
Liver
(MS^g)
0.79 ±0.49
0.89 ±0.44
48.28 ±5.81
43.44 ±6.79
560.23 ± 104.43
716.55 ±59.15
856.90 ±353.83
596.75 ± 158.01
1030.40 ± 162.80
1008.59 ±49.41
Liver: Serum
Ratio
2.04 ± 1.39
1.30 ± 1.32
51.34 ±9.20
29.99 ±8.11
42.10 ±9.20
46.81 ±5.26
41.42 ± 16.95
20.23 ±7.50
35.23 ±8.50
23.48 ± 1.98
Notes:
aData are presented as mean ± standard deviation.
B-12
-------
APRIL 2024
Iwabuchi et al. (2017) exposed male Wistar rats to PFOS in drinking water at 0 [j,g/kg/day,
0.077 [j,g/kg/day, 0.38 [j,g/kg/day, or 1.8 |ig/kg/day for 1 or 3 months. Animals were necropsied
at the end of the 1- or 3-month study, and serum, whole blood, and organ levels of PFOS were
measured (Table B-5). Similar to previous studies, the liver was found to contain the highest
levels of PFOS; however, distribution to other organs (kidney, spleen, and heart) and serum were
remarkably lower when compared with other studies.
Table B-5. Distribution of PFOS in Male Wistar Rats Exposed via Drinking Water for 1 or
3 Months as Reported by Iwabuchi et al. (2017)
1-Month Exposure
3-Month Exposure
Tissue3
0.077
0.38
1.8
0.077
0.38
1.8
jig/kg/day
jig/kg/day
jig/kg/day
jig/kg/day
jig/kg/day
jig/kg/day
Brain (|ig/kg)
0.95
0.14
0.081
0.35
0.3
0.43
Heart (|ig/kg)
0.17
0.23
0.12
0.6
0.57
0.7
Liver (jig/kg)
44
45
25
110
100
100
Spleen (jig/kg)
0.366
0.36
0.21
0.96
0.91
1.3
Kidney (|ig/kg)
1.1
1.1
0.57
3.6
2.6
3.5
Whole Blood (|ig/L)
0.69
0.77
0.46
1.5
1.4
2.1
Serum (|ig/L)
1.1
1.3
0.73
2.7
2.5
3.1
Notes:
aData are presented as mean values.
A combined chronic toxicity/carcinogenicity good laboratory practice (GLP) study was
performed in male and female Sprague-Dawley Crl:CD (SD)IGS BR rats administered 0 ppm,
0.5 ppm, 2 ppm, 5 ppm, or 20 ppm PFOS (equivalent to 0 mg/kg/day, 0.018-0.023 mg/kg/day,
0.072-0.099 mg/kg/day, 0.184-0.247 mg/kg/day, and 0.765-1.1 mg/kg/day, respectively) for
104 weeks (Butenhoff et al., 2012; Thomford, 2002). A recovery group was administered the test
substance at 20 ppm for 52 weeks and observed until necropsy at 106 weeks. Serum and liver
samples were obtained during and at the end of the study to determine the concentration of PFOS
(Table B-6). The findings were in opposition to the Iwabuchi et al. (2017) study as dose-
dependent increases in the PFOS level in the serum and liver were observed in both male and
female rats, with values slightly higher in females after the 5 ppm and 20 ppm doses.
Table B-6. PFOS Levels in the Serum and Liver of Male and Female Sprague-Dawley Rats
Exposed to PFOS in Feed for 2 Years as Reported by Thomford (2002)
Timepoint
0 ppm
0.5 ppm
2 ppm
5 ppm
20
ppm
(weeks)
Males
Females
Males
Females
Males Females
Males
Females
Males
Females
Serum PFOS levels (ng/mL)
0
-------
APRIL 2024
Timepoint
0 ppm
0.5 ppm
2 ppm
5 ppm
20
ppm
(weeks)
Males
Females
Males
Females
Males Females
Males
Females
Males
Females
Liver PFOS levels (ng/g)
0
0.104
0.107
11.0
8.71
31.3 25.0
47.6
83.0
282
373
10
0.459
12.0
23.8
19.2
74.0 69.2
358
370
568
635
53
0.635
0.932
-
-
-
-
-
435
560
105
0.114
0.185
7.83
12.9
26.4
70.5
131
189
381
o
O
-
-
-
-
-
-
-
3.12
12.9
Notes: LOQ = limit of quantification.
aLOQ = 0.00910 pg/mL.
bLOQ = 0.0457 pg/mL.
c Samples were obtained from the recovery group administered 20 ppm for 52 weeks and then observed until necropsy at
106 weeks.
Wang et al. (2015c) compared PFOS levels in serum and brain (hippocampus) tissue in Wistar
rat dams and pups exposed continuously, only prenatally, and only postnatally. Animals were
administered either 5 or 15 mg/L PFOS in drinking water. Tissues from dams were analyzed on
PND 7 and 35, and pup tissues were analyzed on PND 1, 7, and 35. In dams, hippocampal PFOS
concentrations were lower than the respective serum PFOS concentrations, but both serum and
hippocampal levels exhibited dose- and duration-dependent increases. In serum of pups, the
highest levels were observed in pups continuously exposed for 35 days (37.8 ± 2.9 [j,g/mL and
121.0 ±7.1 [j,g/mL in the 5 and 15 g/L exposure groups). In prenatally exposed pups, serum
levels decreased over time (21.7 ±1.5 [j,g/mL on PND 7 compared with 2.7 ± 0.5 [j,g/mL on PND
35) in the high-dose group, as did levels in the hippocampus (10.8 ± 0.5 [^g/mL on PND 7
compared with 0.3 ± 0.0 [j,g/mL on PND 35). The authors suggest the lower hippocampal PFOS
concentrations in the prenatally exposed groups was primarily attributable to PFOS elimination
through feces and urine. In contrast, serum levels increased over time in postnatally exposed
pups (8.7 ±1.4 [^g/mL on PND 7 compared with 61.3 ± 1.1 [^g/mL on PND 35) and increased in
the hippocampus (3.5 ± 0.5 [j,g/mL on PND 7 compared with 5.7 ± 0.7 [j,g/mL on PND 35) in the
high-dose group. Notably, increases in PFOS levels over time in the hippocampus were not
observed in continuously exposed rats, where levels decreased from 32.30 ±1.8 [j,g/mL on PND
7 to 14.66 ±1.0 |ig/mL on PND 35 in the high-dose group. The authors suggest that this
observation may be related to maturation of the blood-brain barrier after PND 24 and/or brain
growth and PFOS redistribution. Strikingly, in the continuously exposed groups for which data
were available on PND 1, hippocampal levels exceeded serum levels (55.9 ±8.1 [j,g/mL in serum
compared with 373.4 ±1.8 [j,g/mL in the hippocampus) in the high-dose group, suggesting that
prenatal exposure poses a high risk to the neural system.
B.2.2.2.3 Mice
Few studies have evaluated PFOS exposure in mice. Findings within these studies focus
primarily on serum and liver concentrations after dosing. Lai et al. (2018) observed that
distribution from serum to liver exhibited dose-dependency after long-term (7 weeks) PFOS
administration in female CD-I mice. At the lower dose (0.3 mg/kg/day), liver and serum
concentrations were similar (32,942 ng/g and 33,781 ng/g, respectively). At the higher dose
(3 mg/kg/day) liver concentrations were higher (503,817 ng/g) than those observed in serum
(109,526 ng/g). Studies in C57BL/6 mice suggest there are limited dose-, sex-, and age-specific
B-14
-------
APRIL 2024
differences in PFOS levels in serum of mice exposed in utero. Zhong and colleagues (Zhong et
al., 2016) administered PFOS to pregnant females (0.1, 1.0, and 5.0 mg/kg/day) by gavage from
GD 1-17. PFOS serum levels were measured in male and female offspring at 4 and 8 weeks after
birth. At 4 weeks only, males had significantly higher serum PFOS levels compared with females
at the 1.0 mg/kg/day dose (47.03 ±3.23 mg/L vs. 41.81 ± 3.62 mg/L) and at the 5.0 mg/kg/day
dose (118.40 mg/L ±6.27 vs. 107.53 ±4.51 mg/L).
Bogdanska et al. (2011) performed a radioisotope distribution study in adult C57BL/6 male mice
using 35S-PFOS feed at a low and high dose for 1, 3, and 5 days. Doses were equivalent to
0.031 mg/kg/day in the low-dose group and 23 mg/kg/day in the high-dose group. At both doses
and at all timepoints, the liver contained the highest amount of PFOS. At the low dose, the liver
PFOS level relative to blood concentration increased with time, whereas at the high dose, the
ratio plateaued after 3 days. The autoradiography indicated that the distribution within the liver
did not appear to favor one area to a greater extent than any other. The liver contained 40%-50%
of the recovered PFOS at the high dose. The authors hypothesized that this could possibly reflect
high levels of binding to tissue proteins. After the liver, lungs accumulated PFOS at the next
highest level in the high-dose group. Distribution was fairly uniform with some favoring of
specific surface areas. The tissue:blood ratio for the lung was greater than that for all other
tissues except the liver. The lowest PFOS levels were in the brain and fat deposits. Levels for the
kidney roughly equaled those values observed in the blood at both concentrations and all
timepoints. For the bone measurements, a whole-body autoradiogram of a mouse 48 hours after a
single oral dose of 35S-PFOS (12.5 mg/kg) indicated that most PFOS was found in the bone
marrow and not the calcified bone.
Recently, the spatial distribution of PFOS in the kidney was investigated using imaging mass
spectrometry (IMS) based on matrix-assisted laser desorption/ionization (MALDI) (Yang et al.,
2019). This methodology can provide spatial information (defined as pixel-to-pixel) with a
unique mass to charge ratio (m/z) for a specified compound in the same tissue section without
extra labeling. The authors first determined that a-Cyano-4-hydroxycinnamic acid (CHCA) was
the optimal matrix for detection of PFOS. Next, male BALB/c mice were administered PFOS by
oral gavage at 10 mg/kg/day for 14 days, at which time kidneys were harvested and frozen.
Continued tissue sections were cut. One section was used for the analysis by MALDI-IMS while
the other two sections were homogenized and used to quantitate PFOS using HPLC-MS/MS. The
average concentration of two sections in the PFOS-exposed kidney was 2.56 ± 0.193 (J,g/mL,
almost 1,000-fold higher than the 3.25 ± 0.274 ng/mL measured in control sections. PFOS was
mainly distributed in the kidney cortex region, which was consistent with the PFOS-induced
glomerular atrophy observed in hematoxylin and eosin-stained sections. The authors conclude
that the average concentration of the whole kidney fails to reflect the spatial accumulation of
PFOS within the kidney, which can be measured and correlated to pathogenetic changes using
MALDI-IMS.
In an immunotoxicity study conducted by Qazi et al. (2009), C57BL/6 male mice were
administered diets with 0% to 0.02% PFOS for 10 days and PFOS levels in serum were
measured. The authors found that PFOS levels in the serum increased as the dietary level of
PFOS increased. While this study does not assess PFOS levels over time, it does demonstrate
dose-dependent increases in serum concentrations.
B-15
-------
APRIL 2024
Wimsatt et al. (2016) dosed male (0 mg/kg, 10 mg/kg, 50 mg/kg, or 200 mg/kg single dose) and
female (0 mg/kg, 20 mg/kg, or 250 mg/kg single dose) mice with PFOS via drinking water. After
8 weeks for males and 9 weeks for females, serum PFOS levels were found to be dose-
dependent.
Similar to rats (Cui et al., 2009), PFOS exposure is found to cross the blood-brain barrier. In Yu
et al. (2019), male ICR mice were dosed with 0 mg/kg/day, 0.25 mg/kg/day, 2.5 mg/kg/day,
25 mg/kg/day, or 50 mg/kg/day for 28 days via oral gavage, and measurements of PFOS in
serum and in brain deposits were collected. Mean serum PFOS levels were approximately
0 |ig/mL, 5 (J,g/mL, 40 (J,g/mL, 240 |ig/mL, and 300 |ig/mL and PFOS levels in the brain were
approximately 0 (J,g/g, 2 (J,g/g, 5 (J,g/g, 30 |ig/g, and 70 |ig/g for the 0 mg/kg/day, 0.25 mg/kg/day,
2.5 mg/kg/day, 25 mg/kg/day, and 50 mg/kg/day dose groups, respectively. These data indicated
that PFOS levels in serum and in brain deposits are dose-dependent and that brain levels were
much lower (100-fold less than that observed in blood and liver). These authors also conducted
in vitro studies showing that PFOS significantly decreased the expression of tight junction-
related proteins (e.g., ZO-1, Claudin-5, Claudin-11, Occludin) in endothelial cells. These
findings suggest that exposure to PFOS may also disrupt the blood-brain barrier, that in turn
could lead to increased accumulation of PFOS in brain. Qui et al. (2013) exposed ICR mice
orally to PFOS at 0 mg/kg/day, 0.25 mg/kg/day, 2.5 mg/kg/day, 25 mg/kg/day, or 50 mg/kg/day
for 28 days via gavage and examined the testicular deposition of PFOS. The study found a
positive correlation between the linear dose-dependent increases in serum concentration and
testicle deposition, indicating that PFOS can cross the blood-testis barrier in mice.
B.2.2.3 Tissue Transporters
PFOS entry from serum into tissues appears to be controlled by several families of membrane
transporters based on PFOA studies. Yu et al. (2011) administered PFOS to rats and extracted
the messenger ribonucleic acids (mRNAs) for OATpl, OATp2, and MRP2 from the liver to
determine if changes in expression of transport molecules correlated with hepatic uptake. Female
Wistar rats were administered PFOS at 0 mg/kg/day, 0.2 mg/kg/day, 1 mg/kg/day, or
3 mg/kg/day via gavage for 5 consecutive days. Blood, bile, and liver tissue were collected
24 hours after the last dose. Exposure to 3.0 mg/kg/day of PFOS increased hepatic OATp2
mRNA expression (1.43-fold) while MRP2 was increased approximately 1.80-fold and 1.69-fold
in the 1 mg/kg/day and 3 mg/kg/day groups, respectively. No effect with treatment was observed
on OATpl.
Transporters responsible for PFOS transport across the placenta are not well understood. Kummu
et al. (2015) used placentas donated from healthy mothers to investigate the role of OAT4 and
ATP-binding cassette transporter G2 (ABCG2) proteins. Using an ex vivo perfusion system, the
authors administered concentrations of PFOA and PFOS (1,000 ng/mL) by perfusing through the
maternal circulation. The fetal :maternal ratios of PFOA and PFOS were 0.20 ± 0.04 and
0.26 ± 0.09, which corresponded to transfer index percentages (TI%) of 12.9 ± 1.5% and
14.4 ± 3.9%, respectively. Immunoblot analysis of OAT4 and AGCG2 in perfused placentas
indicated a linear negative correlation between the expression of OAT4 protein (but not ABCG2)
and PFOA (r2 = 0.92, p = 0.043) and PFOS (r2 = 0.99, p = 0.007) transfer at 120 min. The
authors speculated that OAT4 may play a role in decreasing placental passage of PFAS and
intrauterine exposure to these compounds; however, the low number of placentas examined and
B-16
-------
APRIL 2024
lack of direct evidence for uptake via OAT4 indicates further studies are needed to understand
what, if any, role transporters play in placental transfer of PFOA and PFOS.
To further elucidate the role of placental transporters in facilitating the transfer of maternal PFAS
into the fetus, Li et al. (2020a) compared gene expression of selected transporters in preterm and
full-term placentas and determined whether the differences in expression could influence the
transplacental transfer efficiencies (TTEs). The authors selected nine placental genes with known
xenobiotic activity on the maternal side of the placenta: organic cation/carnitine transporter 2
(OCTN2), reduced folate carrier 1 (RFC-1), equilibrative nucleoside transporter (ENT1), folate
receptor alpha (FRa), heme carrier protein 1 (PCFT), serotonin transporter (SERT), p-
glycoprotein (MDR1), multi-drug resistance-associated protein 2 (MRP2), and breast cancer
resistance protein (BCRP). MDR1 expression levels were significantly associated with TTEs of
branched PFOS and iso-PFOS, (3 + 4 + 5)m-PFOS, but not linear PFOS or PFOA. MRP2
expression was associated with total PFOS, linear PFOS, branched PFOS, and iso-PFOS,
(3 + 4 + 5)m-PFOS, but not PFOA. BCRP expression levels did not significantly change with
PFOA or PFOS. Interestingly, the pattern of expression of MDR1, MRP2 and BCRP were only
observed in full-term placentas. Preterm placentas showed significant expression levels of ENT1,
FRa, and SERT and were associated with lm-PFOS and iso-PFOS. Thus, the expression of
transporters and TTEs appear to differ between preterm and full-term placentas. Authors noted
that the three transporters that were significantly associated with PFOS (MDR1, MRP2, and
BCRP) are also ATP-binding cassette (ABC) transporters, which play a protective role for the
placenta tissue and the fetus by effluxing xenobiotics across the placental barrier thereby
reducing exposure to PFOS. It is unclear why there were no correlations with PFOA although
this may be related to the fact that gene expression associations with TTE were not confirmed
using protein expression data of the candidate genes.
More research is needed to explain how different transporters respond to PFAS and whether
physiochemical properties such as chain length and branching may influence the substrate
binding capacity of these transplacental transporters.
B.2.3 Distribution During Reproduction and Development
The availability of distribution data from pregnant females plus animal pups and neonates is a
strength of the PFOS pharmacokinetic database because it helps to identify those tissues
receiving the highest concentration of PFOS during development. For this reason, the
information on tissue levels during reproduction and development are presented separately from
those that are representative of other lifestages.
B.2.3.1 Human Studies
Zhang et al. (2013b) recruited 32 pregnant females (aged 21-39 years; gestational period 35-
47 weeks) from Tianjin, China, for a study to examine the distribution of PFOS between
maternal blood, cord blood, the placenta, and amniotic fluid. Samples were collected at time of
delivery (31 maternal whole blood samples, 30 cord blood samples, 29 amniotic fluid samples,
and 29 placentas). The maternal blood contained variable levels of 10 PFAS, and the mean
maternal blood concentration was highest for PFOS (14.6 ng/mL), followed by PFOA
(3.35 ng/mL). In both cases, the mean was greater than the median, indicating a distribution
skewed toward the higher concentrations. PFOS was found in all fluids/tissues sampled. It was
B-17
-------
APRIL 2024
transferred to the amniotic fluid to a lesser extent than PFOA based on their relative proportions
in the maternal blood and cord blood (21% vs. 58%, respectively). Compared with the mean
PFOS value in maternal blood, the mean levels in the cord blood, placenta, and amniotic fluid
were 21%, 56%, and 0.14% of the mean levels in the mother's blood, respectively. The
correlation coefficients between the maternal PFOS blood levels and placenta, cord blood, and
amniotic fluid levels ranged from 0.7 to 0.9 (p < 0.001).
B.23.1.1 Partitioning to Placenta
The placenta serves as an important link between the mother and the growing fetus throughout
gestation. It forms a physiological barrier that facilitates the exchange of nutrients, gases,
xenobiotics, and several biological components between maternal and fetal circulation. Several
PFAS compounds including PFOA and PFOS have been identified in amniotic fluid, cord blood,
and fetal tissue, indicating that these chemicals cross the transplacental barrier and influence
PFAS distribution to the fetus and elimination during pregnancy.
The role of the placenta in facilitating the transport of PFAS compounds to the fetal
compartment during gestation is informed by the ratio of placental concentration and matched
maternal serum concentration, or RPM. RPM is a quantitative measure of the placenta's ability
to retain or accumulate compounds. To determine the transplacental transfer of PFOS, Chen et
al. (2017a, b) examined the distribution of PFAS in maternal serum, cord serum, and placentas
from 32 pregnant women and their matched infants in Wuhan, China. Mean maternal age for the
population was 27.1 years, with average pre-pregnancy BMI of 20.4 and gestational age of
38.9 weeks. In Chen et al. (2017b), mean concentrations of total PFOS in the placentas, cord
serum, and maternal serum were 2.842 ng/g, 3.668 ng/mL, and 8.670 ng/mL, respectively, and
the mean RPM was 0.330. The PFOS concentrations in all three matrices from Chen et al.
(2017a) followed a similar pattern, however, the PFOS accumulation in the placenta was
approximately 14.5% less in Chen et al. (2017a) than in Chen et al. (2017b).
Zhang et al. (2013b) (described above) recorded mean PFOS concentrations of 8.18 ng/g in the
placenta, 3.09 ng/mL in cord blood, and 14.6 ng/mL in maternal blood. These concentrations
were significantly higher than the PFOA concentrations in all three compartments. On the basis
of RPM, 59% of maternal PFOS is accumulated in the placenta. This study and the Chen et al.
(2017a, b) studies had similar maternal characteristics (sample size, geographical location
(China), gestational age, maternal age), yet placental PFOS accumulation significantly varied
across studies, ranging from 4.8% to 59%. One distinguishing characteristic that may account for
increased PFOS accumulation in Zhang et al. (2013b) is parity. About 82% of the mothers in
Zhang et al. (2013b) were primiparous whereas only 46.8% were primiparous in Chen et al.
(2017a, b), which may explain the higher PFOS concentrations in maternal serum and placenta
found in the Zhang et al. (2013b) study. Primiparous mothers also tend to have higher levels of
PFAS in breast milk than women who have had multiple children (Lee et al., 2017), adding to
the evidence that pregnancy and lactation durations are critical for PFAS distribution.
Mamsen et al. (2019) demonstrated that factors such as gestational age can affect PFOS
concentrations in maternal serum and placentas. Using a linear graph of normalized percentage
placenta accumulation as a function of gestational age, the authors observed a steady increase of
placenta accumulation of PFOS during gestation days 50 to 300, with male and female placentas
showing similar trends. However, accumulation was significantly higher in males than in
B-18
-------
APRIL 2024
females. Authors estimated a placenta PFOS accumulation rate of 0.13% increase per day during
gestation.
Zhang et al. (2015b) determined that branched PFOS makes up 18% of total PFOS in placenta,
suggesting that branched and linear PFOS accumulate in the placenta at different proportions.
Among branched isomers of the same compound, RPM seemed to differ by functional groups
and branching. Particularly, RPM of branched PFOS isomers seem to increase as the branching
points away from the sulfonate group: iso-PFOS < 4m-PFOS < (3 + 5)m-PFOS < lm-PFOS. In
contrast, the RPM of PFHxS showed a different pattern: branched PFHxS < linear PFHxS (Chen
et al., 2017b). Moreover, RPM of linear and branched PFOA (3m-PFOA) did not significantly
differ from each other. The variation in RPM between the branched isomers of PFOS, PFHxS,
PFOA and their corresponding linear isomers suggest that their capacity to accumulate in the
placenta is partly influenced by structure, functional group, and isomerization.
Umbilical cord blood is a known tissue for PFOS distribution during pregnancy. Kato et al.
(2014) collected blood samples from 71 mothers and their infants in a prospective birth cohort in
the Cincinnati, Ohio metropolitan area. They quantified PFAS in maternal blood at 16 weeks of
gestation and at delivery, evaluated the correlation between maternal PFAS levels in maternal
serum and matched cord blood. Maternal serum levels at 16 weeks of gestation and at the time of
delivery were higher for PFOS (12.7 [j,g/L and 8.50 (J,g/L, respectively) than PFOA (4.8 [j,g/L and
3.3 (J,g/L, respectively). Authors reported a positive correlation between maternal serum PFOS
levels during gestation and cord serum (correlation coefficient = 0.87). Similarly, the correlation
between maternal serum at the time of delivery and cord serum was also positive (correlation
coefficient = 0.82). A strong correlation between PFOS levels in maternal serum (collected
within 1 week of delivery) and cord serum (collected at delivery) was also observed in a cohort
of 50 mother-infant pairs from the Jiangsu province of China (correlation coefficient = 0.882, p <
0.001) (Yang et al., 2016b). In another study conducted in China, 157 paired maternal and cord
serum samples collected in Beijing around delivery (Yang et al., 2016a). PFOS, followed by
PFOA, was the dominant PFAS contaminant in these samples. Mean PFOS levels were
5.08 ± 3.26 ng/mL and 1.52 ± 1.01 ng/mL in maternal and cord serum, respectively (mean
cord:maternal serum ratio was 0.36 ± 0.35:1).
Porpora et al. (2013) quantified PFOS levels in maternal serum and cord blood from 38 mother-
infant pairs in Rome, Italy. The women were Italian Caucasian between the ages of 26 and 45
(mean age, 34.5 years). The average gestational age for participants in this study was 39 weeks.
Maternal and cord serum PFOS concentrations were 3.2 ng/g and 1.4 ng/g, respectively. A strong
positive correlation was observed between maternal and cord serum concentrations (r = 0.74,
p < 0.001). These values suggest a cord to maternal serum ratio of 0.44.
Fromme et al. (2010) measured PFOS in mothers and infants in Munich, Germany. Maternal
blood was sampled during pregnancy, at delivery, and 6 months after delivery in mothers aged
21-43 years. PFOS was also measured in cord blood and in infant blood at 6 and 19 months after
birth. Maternal PFOS serum concentrations ranged from 0.8 to 9.4 [j,g/L (38 samples) and cord
serum concentrations ranged from 0.3 to 2.8 |ig/L (33 samples). The cord to maternal serum
mean ratio was 0.3.
Wang et al. (2019c) measured the levels of 10 PFAS chemicals, including PFOS, in paired
maternal and umbilical cord serum from a prospective birth cohort in Shandong, China. PFOS
B-19
-------
APRIL 2024
was detected in all maternal and umbilical cord serum samples with a geometric mean of
4.25 ng/mL (range of 0.55-29.85 ng/mL) in maternal serum and 1.33 ng/mL (range 0.12-
5.89 ng/mL) in cord serum. PFOS concentrations in maternal serum were strongly correlated to
concentrations in cord blood (r = 0.745).
Linear and branched PFOS have been detected in both maternal and cord serum (Cai et al., 2020;
Li et al., 2020a). Branched PFOS levels in cord blood are consistently lower than linear PFOS
levels. Branched PFOS isomers contributed approximately 19.5% of total PFOS in cord blood
(Cai et al., 2020). Similarly, Li et al. (2020a) showed that branched PFOS makes up 17% of total
PFOS in cord blood from preterm births and 19.2% from full-term births (Table B-7). Together,
these studies suggest that branched PFOS is likely less accumulative in cord blood than linear
isomers. It is worth noting that other factors, such as differential binding affinities in serum and
type of chemical exposure (branched vs. linear PFOS), may also influence the proportions in
serum.
Similar to PFOA, differential TTEs were observed for linear PFOS isomers. Cai et al. (2020)
found an 8% increase in branched PFOS accumulation compared with linear PFOS isomers.
Similarly, Li et al. (2020a) showed a 6% increase in branched PFOS accumulation compared
with linear PFOS isomers. Zhao et al. (2017b) observed higher TTEs for lm, 4m, 3 + 5m, and
m2 compared with n-PFOS. Moreover, the TTEs of branched PFOS isomers increased as the
branching point moved closer to the sulfonate moiety. Together, these findings indicate that
branched isomers of PFOS transfer more efficiently from maternal blood to cord blood compared
with linear isomers.
In summary, these studies suggest that maternal serum levels of PFOS are positively correlated
with cord blood and is a direct determinant of in utero exposure regardless of gestational age or
location of exposure. Maternal serum PFOS levels are consistently higher than cord serum levels
across all studies. PFOS concentrations in both maternal and cord serum varied substantially
across studies, and factors such as exposure sources, parity, and other maternal demographics
may account for these variations. For example, in Eryasa et al. (2019), authors noted that seafood
diet (including high consumption of pilot whale) and consumer products as main sources of
exposure. This may likely explain why maternal and cord serum PFOS concentrations are higher
than all other studies listed in Table B-7. Additionally, linear PFOS are detected at higher
frequency and at higher levels in blood than branched PFOS but are less transferable across
compartments from maternal serum to cord serum.
B-20
-------
APRIL 2024
Table B-7. PFOS Concentrations in Human Cord Blood, Maternal Blood, and Transplacental Transfer Ratios (RCM)
Study
Country, Cohort
Number of
Maternal-Infant
Pairs"
Mean Gestational
Age (weeks)b
PFOS
Measurement
Cord Serum
(ng/mL)c
Maternal Serum
(ng/mL)c
Cord: Maternal
Serum Ratios
(RCM)d
Manzano-Salgado
etal. (2015)
Chenetal. (2017a)
and Chen et al.
(2017b)
53
NR
total PFOS
1.86
6.99
0.30
Sabadell and
Valencia, Spain
Note: Serum concentrations reported as p50. whereas geometric mean concentrations were used by authors to calculate cord:maternal serum
ratios. Reported concentrations from 66 maternal plasma samples, and 66 cord blood samples, and 53 maternal serum samples.
Wuhan, China
32
38.9 ± 1.6
Cariou et al.
(2015)
total PFOS
n-PFOS
iso-PFOS
(3 + 5)m-PFOS
4m-PFOS
lm-PFOS
Note: PFOS detected in 100% of maternal and cord samples except for m-PFOS in cord samples, where the detection rate of 96.87%. PFOS
isomers were reported in Chen (2017a) and total PFOS was reported in Chen (2017b).
Toulouse, France 94 NR total PFOS 1.28 3.67 0.38
Note: Concentrations represent mean values from 100 pairs. Semi-quantified values below LOD were taken into account for mean
calculation.
3.67 ±2.51
2.713
0.203
0.506
1.8
0.226
1.61 ± 5.27
6.971
0.49
0.466
0.157
0.136
0.431
0.384
0.388
0.684
0.695
0.835
Eryasa et al.
Faroese Birth
100
39.9 ± 1.3
total PFOS
9.5 (6.34-13.89)
23.8 (15.8-36.9)
0.38e
(2019)
Cohort, Denmark
(cohort 3)
n-PFOS
5.98 (3.97-8.71)
15.6 (10.5-22.96)
0.37
branched PFOS
3.50 (2.38-1.94)
8.15(5.22-12.58)
0.42
Faroese Birth
51
39.7 ± 1.1
total PFOS
3.09 (2.31—1.42)
8.82 (6.94-11.6)
0.36e
Cohort, Denmark
n-PFOS
1.89 (1.46-2.84)
5.55 (4.16-7.45)
0.35
(cohort 5)
branched PFOS
1.17 (0.88-1.73)
3.18(2.35-4.33)
0.37
Note: Cohort 3 included 100 singleton births from 1999 to 2001 and cohort 5 included 51 singleton births from 2008 to 2005. Both cohorts
had the same source of exposure and are similar in maternal characteristics. Ratios were reported as median p50. Serum concentrations
reported here geometric mean and interquartile ranges (IQR).
Cai et al. (2020)
Maoming Birth
Cohort, China
424
39.3 ± 1.1
total PFOS
linear PFOS
branched PFOS
2.66 ±4.80
2.14 ±4.42
0.52 ±0.49
6.71 ± 19.57
5.62 ± 17.33
1.09 ±2.35
0.51
0.5
0.58
B-21
-------
APRIL 2024
Study
Country, Cohort
Number of
Maternal-Infant
Pairs"
Mean Gestational
Age (weeks)b
PFOS
Measurement
Cord Serum
(ng/mL)c
Maternal Serum
(ng/mL)c
Cord: Maternal
Serum Ratios
(RCM)d
Note: Values represented as mean concentrations ± SD. Ratios were calculated from matched maternal and infant pairs for which all cord
blood samples were >LOD. Percent detect rates were 100% for total PFOS, 99.76% for linear PFOS, and 99.53% for branched PFOS.
Li et al. (2020a)
Maoming Birth
86 33.8 ±3.0
total PFOS
1.93
5.87
0.32
Cohort, China
linear PFOS
1.6
4.85
0.3
(preterm infants)
branched PFOS
0.33
1.01
0.36
iso-PFOS
0.08
0.35
0.26
(3 + 4+5)m-PFOS
0.2
0.57
0.35
lm-PFOS
0.06
0.09
0.65
Maoming Birth
187 39.5 ±1.1
total PFOS
2.6
4.44
0.58
Cohort, China
linear PFOS
2.1
3.76
0.57
(full-term infants)
branched PFOS
0.5
0.68
0.68
iso-PFOS
0.11
0.2
0.51
(3 + 4+5)m-PFOS
0.32
0.41
0.73
lm-PFOS
0.08
0.07
1.07
Note: 273 mother-infant pairs were analyzed, including 86 preterm deliveries and 187 full-term deliveries. Only PFAS substances
quantifiable in >50% of maternal and cord sera are included in generating mean concentration values.
Li et al. (2020b)
Beijing, China
112 39.0 ±1.2
total PFOS
2.31
6.74
0.482
Note: PFOA detection rate was 97.44% in maternal serum and 95.73% in cord serum. For PFOS, 112 of 117 matched cord and maternal
serum samples were used to generate Rcm
Wang et al.
Shandong, China
369 39.4 ±1.3
total PFOS
1.33
4.25
0.30
(2019c)
Note: PFOS detected in 100% of maternal and cord samples.
Pan et al. (2017)
Wuhan, China
100 39.4 ±1.3
total PFOS
4.33
12.7
0.34
Note: Maternal blood collected in third trimester (38.4 ±1.6 weeks) used for RCm
calculation and PFOS was detected in 100% of maternal
and cord samples.
Zhao et al. (2017b) People's Hospital
63 39.3 ±0.82
n-PFOS
3.86
16.8
0.21
of Hong'an
59 39.3 ±0.82
iso-PFOS
0.229
1.08
0.22
County, China
63 39.3 ±0.82
3 + 5m-PFOS
0.417
1.44
0.29
38 39.3 ±0.82
4m-PFOS
0.142
0.536
0.51
61 39.3 ±0.82
lm-PFOS
0.716
1.25
0.48
19 39.3 ±0.82
m2-PFOS
0.043
0.099
0.3
63 39.3 ±0.82
Total PFOS
5.41
21.2
0.22
B-22
-------
APRIL 2024
Study
Beeson et al.
(2011)
Country, Cohort
Number of
Maternal-Infant
Pairs"
Mean Gestational
Age (weeks)b
PFOS
Measurement
Cord Serum
(ng/mL)c
Maternal Serum
(ng/mL)c
Note: Authors reported that samples
-------
APRIL 2024
Study
Number of
„ , , -w- « . Mean Gestational PFOS
Country, Cohort Maternal-Iniant . . . ,h ,,
p . a Age (weeks) Measurement
Cord Serum
(ng/mL)c
Maternal Serum
(ng/mL)c
Cord: Maternal
Serum Ratios
(RCM)d
Note: Median serum concentrations reported. Values in parentheses are 25%-75% IQRs.
Fromme et al.
(2010)
Germany 38 maternal NR total PFOS
samples, 33 cord
samples
Note: Maternal and cord blood samples taken at time of delivery.
1
2.9
0.3
Needham et al.
(2011)
Faroe Islands 12 NR total PFOS 6.6
Note: Serum concentrations reported as median values, RCMs reported as arithmetic means.
19.7
0.34
Liu et al. (2011)
Jinhu, China 50 (all) NR total PFOS
26 (male infants) NR total PFOS
24 (female infants) NR total PFOS
Note: Maternal samples collected in the first weeks after delivery.
1.686
NR
NR
3.184
NR
NR
0.57
0.55
0.58
Midasch et al.
(2007)
NR 11 NR total PFOS 7.3
Note: Serum concentrations reported as median values, RCMs reported as arithmetic means.
13
0.6
Verneretal. NA NA NA NA NA NA 0.45
(2016) Note: Authors developed a two-compartment, two-generation pharmacokinetic model of prenatal and postnatal exposure to PFOA and
PFOS. Rcms applied in model were derived from an average of ratios reported in Aylward et al. (2014).
Notes: CHirP = Chemicals, Health and Pregnancy; IQR = interquartile range; LOD = level of detection; NA = not applicable, NR = not reported; SD = standard deviation.
a Number represents number of matched pairs used for RCM calculation unless otherwise noted in comments.
b Gestational age reported as mean ± SD, represents gestational age at the time of cord blood sampling (delivery) and may not be the same as age at the time of maternal blood
sampling.
c Concentrations in cord or maternal samples are reported as means with or without SD or IQR unless otherwise noted in comments. Note that several studies, the mean serum
concentrations may be derived from more subjects than values used for RCM calculation, which typically included only matched pairs for which both cord and maternal serum
concentrations were above the limit of detection.
dData are presented as a ratio of cord serum to maternal serum concentrations unless otherwise noted in comments.
B-24
-------
APRIL 2024
B.23.1.2 Partitioning to Amniotic Fluid
Zhang et al. (2013b) measured the levels of 11 PFAS chemicals in maternal blood, cord blood
and placenta. All 11 PFAS were detected in their respective biological tissues at different
concentrations. The mean concentration ratio between amniotic fluid and maternal blood
(AF:MB) was higher in PFOA (0.13) than in PFOS (0.0014). Similarly, the mean concentration
ratio between amniotic fluid and cord blood (AF:CB) was higher in PFOA (0.023) than in PFOS
(0.0065). Authors attributed the differences in ratios between the two compartments to the
solubility of PFOS and PFOA and their respective binding protein binding capacities in the two
matrices. PFOA is highly soluble in water relative to PFOS (solubilities of 3.4 g/L and 0.68 g/L,
respectively). Since amniotic fluid is 94% water, the solubility properties may account for the
observation that the PFOA concentration (0.044 ng/mL) was twice as much as PFOS
(0.02 ng/mL) in this matrix.
Table B-8 presents means or medians and ranges of measured and estimated PFOS
concentrations in maternal blood from recent studies (2013 to present) that also measured fetal
indicators of exposure (cord blood, placenta, and/or amniotic fluid). These studies demonstrate
the variability of PFOS accumulation in these tissues across geographic regions. Maternal serum
values ranged from 0.062 ng/mL in Rome, Italy (Porpora et al., 2013) to 183 ng/mL in Hubei,
China (Zhao et al., 2017a). Cord serum values ranged from
-------
APRIL 2024
were predicted to accumulate PFOA at higher levels than PFOS levels for most of the fetal
indicators of exposure. The authors concluded that differences in fetal exposure are not predicted
by the measurement of the maternal concentration during pregnancy.
Table B-8. Summary of PFOS Concentrations in Human Maternal Blood, Cord Blood,
Placenta and Amniotic Fluid Studies
Study (Location
Maternal Blood
Cord Blood
Infant
Placenta
Amniotic Fluid
of Study)
Blood
Porpora et al.
Maternal serum
Cord serum
NR
NR
NR
(2013)
Mean: 3.2 ng/g
Mean: 1.4 ng/g
(Rome, Italy)
Median: 2.9 ng/g
Median: 1.1
Range: 0.062-13 ng/g
Range: 0.23-3.7 ng/g
Zhang et al.
NR
NR
NR
Mean: 8.18 ng/g
Mean: 0.020 ng/mL
(2014)
Median:
Median:
-------
APRIL 2024
Study (Location
Maternal Blood
Cord Blood
Infant
Placenta
Amniotic Fluid
of Study)
Blood
Median: 13.20 ng/mL
IQR: 7.62-
20.38 ng/mL
Maternal serum
To
NR
NR
NR
1.3
Mean: 12.7 ng/mL
Median: 12.32 ng/mL
IQR: 7.61-
20.03 ng/mL
Caserta et al.
Mean: 1.54 ng/mL
Mean: 1.75 ng/mL
NR
NR
NR
(2018)
SD: 1.28 ng/mL
SD: 1.70 ng/mL
(Rome, Italy)
Range: 0.018-
Range: 0.018-
4.7 ng/mL
6.00 ng/mL
Wang et al.
Maternal serum
Cord serum
NR
NR
NR
(2019c)
GM: 4.25 ng/mL
Mean: 1.33 ng/mL
(Shandong,
Median: 4.55 ng/mL
Median: 1.39 ng/mL
China)
Range: 0.55-
Range: 0.12-
29.85 ng/mL
5.89 ng/mL
Zhao et al.
Maternal blood
Cord Blood
NR
NR
NR
(2017b)
Mean: 21.2 ng/mL
Mean: 5.41 ng/mL
(Hong'an, China) Median: 6.59 ng/mL
Median: 1.35 ng/mL
Range: 1.51—
Range: 0.346-
582 ng/mL
183 ng/mL
Brochot et al.
Group 1 mean
Mean: 2.08 ± 1.00
NR
NR
NR
(2019)
(plasma): 7.14 ± 5.35
Range: 0.53-
(INMA
(0.69-38.58) ng/mL
4.71 ng/mL
Prospective birth
Group 2 mean
cohort, Spain)b
(plasma): 5.70 ± 3.45
(0.26-25.98) ng/mL
Gao et al. (2019)
Mean: 4.64 ng/mL
Mean: 2.35 ng/mL
NR
NR
NR
(Beijing, China)
median: 4.07 ng/mL
Median: 1.8 ng/mL
range: 0.07-
Range: 0.04-
22.6 ng/mL
8.01 ng/mL
Eryasa et al.
Cohort 3
Cohort 3
NR
NR
NR
(2019)
Maternal serum
Cord serum:
(Faroese Birth
Mean: 23.8 ng/mL
Mean: 9.50 ng/mL
Cohort,
SD: 1.2 ng/mL
SD: 0.49 ng/mL
Denmark)0
IQR: 15.8-
IQR: 6.34-
36.9 ng/mL
13.89 ng/mL
Whole cord blood:
Mean: 4.90 ng/mL
SD: 0.26 ng/mL
IQR: 3.33-6.94 ng/mL
Cohort 5
Cohort 5
NR
NR
NR
mean: 8.82 ng/mL
Cord serum:
SD: 0.51 ng/mL
mean: 3.09 ng/mL
IQR: 6.94-
SD: 0.22 ng/mL
11.6 ng/mL
IQR: 2.31-1.42 ng/mL
Whole cord blood:
mean: 1.60 ng/mL
B-27
-------
APRIL 2024
Study (Location
Maternal Blood
Cord Blood
Infant
Placenta
Amniotic Fluid
of Study)
Blood
SD: 0.11 ng/mL
IQR: 1.18-2.32 ng/mL
Cai et al. (2020)
Maternal serum
Cord serum
NR
NR
NR
(Maoming Birth
Mean: 6.71 ng/mL
Mean: 2.66 ng/mL
Cohort, China)
SD: 19.57 ng/mL
SD: 4.80 ng/mL
Median: 4.32 ng/mL
Median: 1.93 ng/mL
IQR: 2.94-
IQR: 1.23-2.66 ng/mL
6.34 ng/mL
Li et al. (2020a)
Total PFOS:
Total PFOS:
NR
NR
NR
(Maoming Birth
Preterm delivery:
Preterm delivery:
Cohort, China)d
Mean: 5.87 ng/mL
Mean: 1.93 ng/mL
Median: 3.53 ng/mL
Median: 1.47 ng/mL
IQR: 2.36-5.93
IQR: 0.83-1.97
Full-term delivery:
Full-term delivery:
Mean: 4.44 ng/mL
Mean: 2.60 ng/mL
Median: 3.54 ng/mL
Median: 2.08 ng/mL
IQR 2.25-5.98
IQR 1.28-3.06
Li et al. (2020b)
Mean: 6.74 ng/mL
Mean: 2.31 ng/mL
NR
NR
NR
(Maoming Birth
(95% CI: 6.27, 8.95)
(95% CI: 2.9, 3.4)
Cohort, China)
Median: 5.99 ng/mL
Median: 1.65 ng/mL
Zhang et al.
Mean: 14.6 ng/mL
Mean: 3.09 ng/mL
NR
Mean: 8.18 ng/g
Mean: 0.020 ng/mL
(2013d)
RSD: 4.98
RSD: 1.84
RSD: 3.03
RSD: 0.032
(Tiajin, China)
Range: 7.39-
Range: 0.14-
Range: 3.25-
Range:
-------
APRIL 2024
Study (Location
Maternal Blood
Cord Blood
Infant
Placenta
Amniotic Fluid
of Study)
Blood
Mamsen et al.
Mean: 8.2 ng/g,
NR
NR
Mean: 1.3 ng/
NR
(2017)
Range: 2.5-16.7 ng/g
Range: 0.3-
(Denmark)
3.1 ng/g
Mamsen et al.
Tl serum
NR
NR
Mean: 1.43 ng/g
NR
(2019)
Mean: 8.14 ng/mL
SD: 0.63 ng/g
(Denmark)3
SD: 3.82 ng/mL
Median: 6.76 ng/mL
Range: 2.49-
16.66 ng/mL
Median:
1.35 ng/g
Range: 0.65-
3.09 ng/g
T2 serum
NR
NR
Mean: 1.23 ng/g
NR
Mean: 3.87 ng/mL
SD: 0.60 ng/g
SD: 1.99 ng/mL
Median:
Median: 3.43 ng/mL
1.08 ng/g
Range: 1.04-
Range: 0.63-
8.19 ng/mL
2.33 ng/g
T3 serum
NR
NR
Mean: 1.53 ng/g
NR
Mean: 3.58 ng/mL
SD: 0.90 ng/g
SD: 1.85 ng/mL
Median:
Median: 3.26 ng/mL
1.42 ng/g
Range: 1.07-
Range: 0.45-
9.66 ng/mL
3.87 ng/g
Kato et al. (2014) Maternal Serum at Cord serum at delivery
(Ohio, USA)f 16 wk Median: 3.50 (ig/L
Median: 12.70 |ig/L
Maternal serum at
delivery
Median: 8.50 [ig/L
Notes: AM = arithmetic mean; CI = confidence interval; GM = geometric mean; INMA = INfancia y Medio Ambiente
(Environment and Childhood) Project; IQR = interquartile range; LOD = limit of detection; LOQ = limit of quantification;
SD = standard deviation; NR = not reported; RSD = relative standard deviation; T1 = trimester 1; T2 = trimester 2;
T3 = trimester 3; USA = United States of America.
aPFOS was collected at different timepoints during gestation: first trimester (Tl), second trimester (T2) and third trimester (T3).
bBrochot et al. (2019) collected samples from women in two cohorts: Group 1 consist of 52 mother-child pairs that had available
samples of maternal blood during pregnancy and cord serum. Group 2 consists of 355 mothers who provided maternal blood
during pregnancy. Cord blood was not collected for Group 2.
cEryasa et al. (2019) collected serum and whole blood from participants in two birth cohorts: Cohort 3 (100 Singleton births from
1999 to 2001), and cohort 5 (50 singleton births from 2008 to 2005). Both cohorts had the same source of exposure and are
similar in maternal characteristics.
dLi et al. (2020a) measured PFOS in matched maternal-cord serum pairs with preterm deliveries and full-term deliveries.
eHanssen et al. (2013) collected whole blood and plasma from women in two geographical locations: Norilsk (n = 7) and
Uzbekistan (n = 10). Cord blood and cord plasma from infants born to the Norilsk mothers only.
f Kato et al. (2014) measured PFOS in 71 matched maternal and cord serum pairs. Maternal serum samples were collected at
16 weeks of gestation and at the time of delivery.
B.2.3.1.3 Distribution in Fetal Tissues
Mamsen et al. (2017) measured the concentrations of 5 PFAS chemicals in human fetuses,
placentas, and maternal plasma from a cohort of 39 pregnant women in Denmark, who legally
terminated their pregnancies before gestational week 12 for reasons other than fetal abnormality.
The samples collected included 24 maternal blood, 34 placenta, and 108 fetal organs. The
participants were healthy women ages 18-46 years with an average BMI of 22.7. About 51% of
the mothers smoked during pregnancy at an average of 10 cigarettes per day or were exposed to
B-29
-------
APRIL 2024
secondhand cigarette smoke for an average of 1.8 hours per day. Mean concentrations of PFOS
in maternal serum, placenta, and fetal organs were reported as 8.2 (2.5-16.7) ng/g, 1.0 (0.3-
2.6) ng/g, and 0.3 (0-0.7) ng/g, respectively. The concentrations of PFOS in all three matrices
were significantly higher than all four PFAS chemicals including PFOA. For 21 of the samples
where all three specimens (maternal plasma, placenta, and fetal tissues) were collected from the
same women, the concentration of PFOS decreased from maternal serum to fetal tissues as
follows: maternal serum > placenta > fetal tissues. The relative concentration of PFOS in the
placenta was 14% of the concentrations found in maternal plasma and were further reduced to
5% in fetal tissues. Although PFOS concentrations in all three matrices were higher than the
remaining PFAS chemicals, PFOS had the lowest relative concentrations in fetal tissues. In
general, a positive trend was observed between gestational age and fetal/maternal plasma ratio.
Although the gestational age reported in this study is short (37-68 days post conception), the
results suggest that PFOA and PFOS accumulate in the fetus and may potentially continue to
accumulate across gestation.
To determine whether PFOS accumulation in fetal organs changes across trimesters during
gestation, Mamsen et al. (2019) quantified PFAS levels in embryos and fetuses at gestational
weeks 7-42 and serum from their matched maternal pairs. Like Mamsen et al. (2017),
participants were similar in age (18-46 years) and BMI (22.8 (first trimester)). However, the
smoking status of the women in this study was not reported and the majority of the pregnancies
were terminated due to intrauterine fetal death (IUFD) caused by placental insufficiency and
intrauterine growth restriction (58%), and infection (13%). A total of 78 pregnant women were
enrolled in the study. Fetal tissues (placenta, liver, lung, heart, CNS, and adipose) were collected
from 38 first trimester pregnancies, 18 second trimester pregnancies, and 22 third trimester
pregnancies. In all fetal tissues examined and across trimesters, PFOS concentrations were
highest compared with other PFAS. The concentration of PFAS in fetal tissues fluctuated across
trimesters and did not follow any particular trend. For example, PFOS concentration in the liver
was higher in the second trimester compared with the third trimester, and lowest in the lung in
the second trimester compared with the first and third trimesters. Interestingly, PFOA
concentration in the liver was also highest in the second trimester compared with the first and
third trimesters. Authors attributed this phenomenon to the unique architecture of the fetal liver
during early gestation when authors attributed this phenomenon to the unique architecture of the
fetal liver during early gestation when oxygenated cord venous blood bypasses the liver into the
heart through the ductus venosus and is then delivered throughout the fetus. This pattern of blood
distribution changes between week 20 and 26 of gestation (late second trimester). The amount of
blood shunted from the liver is reduced from 60% to 30% in the second trimester Pennati et al.
(2003). This reduction results in increased flow of cord blood through the liver, thus increasing
levels of PFOA and PFOS during the second trimester. Furthermore, Mamsen et al. (2019)
observed that PFOA and PFOS levels were lowest in the CNS than any of the tissues examined,
suggesting that the CNS has less PFAS exposure and may be protected by the blood-brain
barrier. When interpreting these results, it is important to note that second and third trimester
fetal tissues were obtained from patients with IUFD and may not be comparable to normal
pregnancies as the fetus died in utero of placental insufficiency and intrauterine growth
restriction. Placental insufficiency can potentially reduce the amount of PFAS crossing the
placenta. In addition, the PFAS exposure level in this cohort may vary due to different
geographical locations of the participants. The first trimester participants were from Denmark
and the second and third trimester participants came from Sweden.
B-30
-------
APRIL 2024
B.23.1.4 Partitioning to Infants
Four studies shown in Table B-9 analyzed PFOS levels in maternal serum and levels in breast
milk and/or infant blood. Maternal and infant serum PFOS levels were substantially higher in
subjects in the United States exposed to contaminated drinking water (Mondal et al., 2014)
compared with subjects analyzed in France, Denmark (Faroe Islands), or Sweden
(Gyllenhammar et al., 2018a; Cariou et al., 2015; Mogensen et al., 2015b). In the Mondal study,
geometric mean (GM) maternal serum PFOS concentrations were lower in breastfeeding mothers
(11.63 ng/mL) versus non-breastfeeding mothers (13.48 ng/mL). Conversely, breastfed infants
had higher GM serum PFOS (13.54 ng/mL) than infants who were never breastfed
(12.65 ng/mL).
Cariou et al. (2015) reported that PFOS levels in breastmilk were approximately 66-fold lower
relative to maternal serum and the ratio between breastmilk and maternal serum PFOS was
0.38 ± 0.16 (n = 19). The authors noted that the transfer rates from serum to breastmilk of
PFAAs were lower compared with other lipophilic persistent organic pollutants such as
polychlorinated biphenyls. In this study, four PFAS compounds were analyzed (PFOA, PFOS,
PFNA, and PFHxS), and the individual patterns for these compounds exhibited important
interindividual variability. While PFOS was the main contributor in serum, PFOA and PFOS
were found to be the main contributors in breastmilk. Interestingly, while the number of
pregnancies was inversely correlated with maternal serum levels, after adjustment, the
correlation with parity did not reach significance for PFOS, although it did reach significance for
PFHxS.
Mogensen et al. (2015b) relied on maternal serum concentrations measured at 32 weeks of
pregnancy to assess prenatal exposure and measured concentrations in the serum of children at
11 and 18 months of age. They applied linear mixed models to estimate age-dependent serum
concentrations for up to 5 years after birth. The only other exposure source adjusted for in this
study was the eating whale meat by the infants. As shown in Table B-9, the increases in infant
blood PFOS concentrations over time, with the greatest increases found at the end of the
breastfeeding period, suggest that breastfeeding is the primary exposure source during infancy.
Gyllenhammar et al. (2018a) used multiple linear regression and general linear model
analysis to investigate associations between serum PFOS concentrations in 2-4-month-old
infants and maternal PFOS concentrations close to delivery, duration of in utero exposure
(gestational age at delivery), duration of breastfeeding, and other parameters. The authors
examined PFAAs of various chain lengths and observed decreased strength of association
between maternal and infant concentrations with increased PFAA carbon chain length
among breastfed infants. PFOS showed the highest median in both infants and mothers
(order among measured PFAAs was PFOS > PFOA > PFHxS > PFNA > PFDA >
PFUnDA). The infant:maternal serum ratios were similar for total, linear, and branched
PFOS (0.69 (0.14-1.5), 0.66 (0.095-1.4), and 0.72 (0.19-1.7), respectively). Despite similar
ratios, the authors observed that branched PFOS isomer concentrations increased on average
1% per day of gestational age, whereas linear isomer concentrations increased 0.75% per day
of gestational age, supporting a higher efficiency of placental transfer of branched as
opposed to linear isomers during gestation.
B-31
-------
APRIL 2024
Table B-9. Summary of Human PFOS Concentrations in Maternal Serum, Breast Milk, and Infant Serum
Study
Subjects
Maternal Blood
Breastmilk
Infant Blood
Mondal et al. (2014)
Subjects were a subcohort of the C8
Science Panel Study (exposed to
contaminated drinking water in six
water districts near Parkersburg, West
Virginia) who had a child <3.5 yr of
age and who provided blood samples
and reported detailed information on
breastfeeding at the time of survey
(633 mothers and 49 infants included).
PFAA serum concentrations were
available for all mothers and 8%
(n = 49) of the infants. Maternal and
infant serum concentrations were
regressed on duration of
breastfeeding.
Maternal serum
Breastfed and not breastfed
GM: 12.33 ng/mL
95% CI: 11.77, 12.92
Breastfed:
GM: 11.63 ng/mL
95% CI: 10.98, 12.31
Not breastfed
GM: 13.48 ng/mL
95% CI: 12.45, 14.58
NR
Infant serum
Breastfed and not breastfed
GM: 13.21 ng/mL
95% CI: 11.17, 15.61
Breastfed
GM: 13.54 ng/mL
95% CI: 10.79, 17.00
Not breastfed
GM: 12.65 ng/mL
95% CI: 9.74, 16.43
Mogensen et al. (2015b)a
80 singleton children in Faroese birth
cohort born between 1997 and 2000.
The children were breastfed
exclusively
for a median of 4.5 months, followed
by partial breastfeeding with
supplementary baby food for a median
of 4 months.
NR
NR
Birth: median: 6.0 ng/mL
(IQR 5.2,7.2)
11 months: median: 23.2 ng/mL
(IQR 14.9, 34.7)
18 months: median: 24.0 ng/mL
(IQR 20.2, 29.1)
60 months: median: 13.3 ng/mL
(IQR 10.6, 16.6)
Cariou et al. (2015)
Female volunteers hospitalized
between June 2010 and January 2013
for planned cesarean delivery in
France. Maternal blood samples
(n = 100) were collected during
cesarean delivery and breast milk
samples (61) were collected between
the 4th and 5th day after delivery.
Maternal serum
Mean: 3.67 ng/mL
Median: 3.065 ng/mL
Range: 0.316-24.5 ng/mL
Mean: 0.040 ng/mL
Median:
-------
APRIL 2024
Study
Subjects
Maternal Blood
Breastmilk
Infant Blood
Gyllenhammar et al.
(2018a)
Primaparae mother/child pairs in
1996-1999 recruited in Sweden. 101
maternal and 107 infant samples were
available for PFAA analyses. Serum
concentrations were determined in
mothers 3 weeks after delivery and in
2-4-month-old infants.
Maternal serum
Mean: 20 ng/g
SD: 8.9 ng/g
Median: 18 ng/g
Range: 7.7-61 ng/g
NR
Infant serum
Mean: 14 ng/g
SD: 6.7 ng/g
Median: 13 ng/g
Range: 2.2-44 ng/g
Haug et al. (2011)
41 female volunteers from Oslo,
Norway, of which 19 submitted breast
milk samples. The timing of serum or
milk samples obtained from
breastfeeding women was not
reported.
Maternal serum
Mean: 6.9 ng/mL
Range: 2.3-15 ng/mL
Mean: 0.093 ng/mL
Range: 0.040-
0.35 ng/mL
NR
Notes: CI = confidence interval; GM = geometric mean; IQR = interquartile range; LOD = limit of detection; LOQ = limit of quantification; PFAA = perfluoroalkyl acid; NR = not
reported; SD = standard deviation.
aNeonatal serum-PFAS concentrations was calculated based on PFAS ratios between cord and maternal pregnancy serum concentrations previously estimated for the same cohort
(0.34 for PFOA) from Needham et al. (2011).
B-33
-------
APRIL 2024
Mondal et al. (2014) also examined the change in maternal and infant PFOS levels with duration
of breastfeeding (Table B-10). Maternal serum concentrations decreased with each month of
breastfeeding (-3%; 95% CI: -5%, -2%) with the greatest decrease observed after 12 months of
breastfeeding (-39%). Correspondingly, the infant PFOS serum concentrations increased by 4%
(95%) CI: P/o, 7%>) with each month of breastfeeding. Using mixed linear model regression
(Table B-l 1), Mogensen et al. (2015b) calculated more dramatic increases in infants
during months with exclusive breastfeeding of 29.2%> and 30.2% per month at 18 and 60 months,
respectively. Increases were less striking for months with partial breastfeeding and small or none
for months without breastfeeding. The Gyllenhammar et al. (2018a) study included only five
exclusively bottle-fed infants. In this group, they observed a higher percentage of branched
PFOS compared with exclusively breastfed infants, which may be the result of the higher
efficiency of placental transfer of branched PFOS isomers versus linear isomers. Haug et al.
(2011) reported a significant positive correlation between maternal serum and breast milk
(r = 0.71, n = 19) and an average breast milk concentration of 1.4% of the corresponding serum
concentration. The mean relative proportions of branched PFOS isomers were higher in serum
(22%) compared with breast milk (17%), suggesting differential partitioning of branched isomers
between placenta and breast milk. Altogether, these findings support breastfeeding as the
primary source of infant PFOS accumulation and that distribution to the infant correlates with the
length of breastfeeding.
Table B-10. Percent Change in PFOS Ratios in Human Maternal Serum and Breast Milk
and Breast Milk and Infant Serum by Infant Age as Reported by Mondal et al. (2014)
Infant Age
Maternal Serum: Breast Milk
Breastmilk: Infant Serum
<6 months
-9% (-18%, 1%)
-31% (-53%, 1%)
7-12 months
-24% (-34%, -13%)
40% (-9%, 115%)
>12 months
-39% (-52%, -23%)
71% (9%, 167%)
Continuous (per month)
-3% (-3%, -2%)
4% (1%, 7%)
Table B-ll. Percent Change in Human PFOS Serum Concentration by Exclusive, Mixed or
No Breastfeeding Per Month as Reported by Mogensen et al. (2015b)
Breastfeeding Mixed Model up to 18 Months Mixed model up to 60 Months
Status
Percent Change P value Percent Change P value
Exclusive 29.2 (25.3,33.1) <0.0001 30.2 (26.2,34.3) <0.0001
Partial 4.4 (1.0,7.8) 0.0108 1 (-1.2,3.2) 0.3762
None 0.7 (-0.5, 1.9) 0.2693 -0.9 (-1.2,-0.6) <0.0001
The contributions of placental transfer, breastfeeding, and ingestion of PFAA-contaminated
drinking water to early life PFOS levels in children were analyzed (Gyllenhammar et al., 2019).
This study measured PFOS concentrations in children aged 4, 8, and 12 years (n = 57, 55, and
119, respectively) between 2008 and 2015 as part of the Persistent Organic Pollutants in Uppsala
Primiparas (POPUP) study in Sweden. Mixed linear regression (MLR) models were used to
ascertain associations with PFOS for these exposure sources. PFOS concentrations increased
1.3% per unit (ng/g serum) of increase in the maternal serum level at delivery. PFOS
B-34
-------
APRIL 2024
significantly increased 3.8% per month of nursing. Maternal serum and nursing duration showed
the strongest correlations in 4-year-old children. PFOS increased 0.93% per month of cumulative
drinking water exposure. The authors suggested that, in addition to exposure in utero and
through lactation, drinking water with low-to-moderate PFOS contamination is an important
source of exposure for children.
B.2.3.2 Animal Studies
B.2.3.2.1 Rots
To determine the dose-response curve for neonatal mortality in rat pups born to PFOS-exposed
dams and to investigate associated biochemical and pharmacokinetic parameters, five groups of
16 female Sprague-Dawley Crl:CD(SD)IGS VAF/Plus rats were administered 0, 0.1, 0.4, 1.6, or
3.2 mg PFOS/kg bw/day by oral gavage beginning 42 days prior to cohabitation and continuing
through gestation day (GD) 14 or GD 20 (Luebker et al., 2005a). PFOS levels were analyzed in
serum, liver, urine, and feces samples in dams and fetuses as indicated in Table B-12. The urine,
feces, and liver of the control animals all contained PFOS at small concentrations. In treated rats,
the highest concentration of PFOS was in the liver. Serum levels in the dams for each dose were
consistent between GD 1 and GD 15, indicating achievement of steady state prior to conception.
The GD 21 levels in the dams had dropped below those observed earlier in the pregnancy. Serum
levels in the GD 21 fetuses were higher than those in the dams. In contrast, PFOS levels in the
livers of dams on GD 21 were about 3 times higher than in the fetuses. Fecal excretion was
greater than urinary excretion by the dams.
Table B-12. Liver, Serum, Urine, and Feces PFOS Concentrations in Pregnant Sprague-
Dawley Dams and Fetuses (Luebker et al., 2005a)
Parameter
Dose
GD 1
GD 7
GD 15
GD 21
(mg/kg/day)
Dams
Dams
Dams
Dams
Fetuses
Serum3
0.1
8.90 ± 1.10
7.83 ± 1.11
8.81 ± 1.47
4.52 ± 1.15
9.08
0.4
40.7 ± 4.46
40.9 ±5.89
41.4 ±4.80
26.2 ± 16.1
34.3
1.6
160 ± 12.5
154 ± 14.0
156 ±25.9
136 ±86.5
101
3.2
318 ± 21.1
306 ±32.1
275 ± 26.7
155 ±39.3
164
Liverb
0.1
0.4
1.6
3.2
-
-
-
29.2 ± 10.5
107 ± 22.7
388±167
610±142
7.92
30.6
86.5
230
Urine3
0.1
0.05 ±0.02
0.06 ±0.03
0.07 ± 0.04
0.06 ±0.01
-
0.4
0.28 ±0.19
0.31 ±0.20
0.53 ±0.23
0.55 ±0.16
-
1.6
0.96 ±0.39
1.10 ±0.57
0.36 ±0.35
2.71 ±2.07
-
3.2
1.53 ±0.87
1.60 ±0.97
0.52 ±0.28
1.61 ±0.53
-
Feces'3
0.1
0.50 ±0.14
0.49 ±0.11
0.66 ±0.10
0.42 ±0.10
-
0.4
2.42 ±0.49
2.16 ±0.43
2.93 ±0.62
2.39 ± 1.21
-
1.6
10.3 ±3.01
9.20 ±2.68
11.1 ±3.28
9.94 ±4.51
-
3.2
23.9 ±4.16
33.0 ± 10.0
29.5 ±8.92
20.1 ±4.21
—
Notes: GD = gestation day.
B-35
-------
APRIL 2024
aData presented in mean ± standard deviation (|xg/mL).
b Data presented in mean ± standard deviation (|xg/g).
This same study also included a subset of dams allowed to litter naturally and dosed through
lactation day (LD) 4. Liver and serum samples were collected from dams and pups on LD 5. In
this sampling, serum PFOS levels were similar between the dam and offspring, but the liver
values were now higher in the neonates than in the respective dams.
Twenty-five female Sprague-Dawley rats/group were administered 0 mg/kg/day, 0.1 mg/kg/day,
0.3 mg/kg/day, or 1.0 mg/kg/day potassium PFOS by gavage from GD 0 through PND 20. An
additional 10 mated females served as satellite rats to each of the four groups and were used to
collect additional blood and tissue samples. Further details from this study are provided in the
Toxicity Assessment (U.S. EPA, 2024) as reported in Butenhoff et al. (2009). Samples were
taken from the dams, fetuses, and pups for serum and tissue PFOS concentrations and the results
were reported by Chang et al. (2009) (Table B-13).
B-36
-------
APRIL 2024
Table B-13. Serum, Liver, and Brain Tissue PFOS Concentrations of Sprague-Dawley Dams and Offspring as Reported by
Chang et al. (2009)
Time
Dose
Serum PFOSa
Liver PFOSb
Brain PFOSb
(mg/kg)
Dam
Offspring
Dam
Offspring
Dam
Offspring
GD 20°
Control
-------
APRIL 2024
On GD 20, PFOS concentrations in maternal serum, liver, and brain correlated with the daily
doses administered. Maternal liver-to-serum PFOS ratios ranged from 1.8 to 4.9, while the
maternal brain-to-serum ratios were 0.04 to 0.09 (Chang et al., 2009). The concentrations in the
brains of fetuses was about 10 times higher than in their dams for all doses. Given the maternal
and offspring data on GD 20, there is placental transfer of PFOS from rat dams to developing
fetuses. Serum values were approximately 1-2 times greater in the fetuses than in the dams at
GD 20. The concentration of PFOS in fetal liver was less than that of dams, and the brain values
were much higher; this is possibly due to the lack of development of the blood-brain barrier at
this stage of offspring development. PFOS serum concentrations in the offspring were lower than
those for the dams on postnatal day (PND) 4 and continued to drop through PND 72. However,
based on the concentrations still present in the neonate serum, lactational transfer of PFOS was
occurring. At PND 72, the males appeared to be eliminating PFOS more quickly as the serum
values were lower than those in the females; this difference was not observed at earlier
timepoints. In the liver, PFOS was the greatest in the offspring at PND 4 and decreased
significantly by PND 72. Liver values were similar at all timepoints between males and females.
On GD 20, the brain levels for the pups were tenfold higher than those for the dam. The levels in
pup brains gradually declined between PND 4 and PND 21.
Ishida et al. (2017) also examined distribution to livers and brains in Wistar rat dams and pups on
PND 4. Tissue-to-plasma partition coefficients (KPs) for brain/plasma decreased with increasing
dose in dams (0.92 in dams at 1 mg/kg and 0.87 in dams at 2 mg/kg). In pups, the brain/plasma
KP values were 0.447 and 0.408 at 1 mg/kg and 2 mg/kg, respectively. Liver/plasma KP values
were 4.13 and 3.85 in dams and 3.30 and 2.07 in pups at the lower and higher doses,
respectively. Thus, the brain-plasma ratio of PFOS in pups is approximately 5 times higher than
that in dams despite very similar liver/plasma ratios in pups and dams, indicating an age-
dependent accumulation of PFOS in the CNS.
In a study by Zeng et al. (2011), 10 pregnant Sprague-Dawley rats/group were administered
0 mg/kg/day, 0.1 mg/kg/day, 0.6 mg/kg/day, or 2.0 mg/kg/day of PFOS by oral gavage in 0.5%
Tween 80 from GD 2 to GD 21. On GD 21, dams were monitored for parturition, and the day of
delivery was designated PND 0. On PND 0, five pups/litter were sacrificed, and the trunk blood,
cortex, and hippocampus were collected for examination. The other pups were randomly
redistributed to dams within the dosage groups and allowed to nurse until PND 21, when they
were sacrificed with the same tissues collected as previously described. PFOS concentrations in
the hippocampus, cortex, and serum increased in a dose-dependent manner but overall was lower
in all tissues on PND 21 compared with PND 0 (Table B-14).
Table B-14. Serum, Hippocampus, and Cortex PFOS Concentrations of Sprague-Dawley
Rat Pups as Reported by Zeng et al. (2011)
Time
Dose (mg/kg/day)
Serum3
Hippocampusb
Cortexb
PND0
Control
ND
ND
ND
0.1
1.50 ±0.43*
0.63 ±0.19*
0.39 ±0.09*
0.6
24.60 ±3.02"
7.43 ± 1.62*
5.23 ± 1.58"
2.0
45.69 ±4.77"
17.44 ±4.12*
13.43 ±3.89"
PND21
Control
ND
ND
ND
0.1
0.37 ± 1.12*
0.25 ±0.14*
0.06 ± 0.04*
B-38
-------
APRIL 2024
Time
Dose (mg/kg/day)
Serum3
Hippocampusb
Cortexb
0.6
1.86 ±0.35"
1.59 ±0.78"
1.03 ±0.59"
2.0
4.26 ± 1.73*"
6.09 ± 1.30"*
3.69 ±0.95***
Notes: ND = not detected; PND = postnatal day.
*p < 0.05 compared with control in the same day.
** p < 0.05 compared with 0.1 mg/kg group in the same day.
*** p < 0.05 compared with 0.6 mg/kg group in the same day.
aData presented as mean ± standard deviation (|xg/mL).
bData presented as mean ± standard deviation (|xg/g).
Sprague-Dawley rats were administered PFOS in 0.05% Tween (in deionized water) once daily
by gavage from GD 1 to GD 21 at 0 mg/kg/day, 0.1 mg/kg/day, or 2.0 mg/kg/day. There was a
postnatal decline in the serum and brain PFOS levels between PND 0 and PND 21. PFOS
concentrations were higher in the serum when compared with the lung in offspring on both PND
0 and PND 21 (Chen et al., 2012b) (Table B-15).
Table B-15. Serum and Lung PFOS Concentration of Sprague-Dawley Rat Pups (Chen et
al., 2012b)
Age
Dose (mg/kg/day)
Serum3
Lungb
PND 0
0.0
ND
ND
0.1
1.7 ±0.35*
0.92 ±0.04*
2.0
47.52 ±3.72*
22.4 ± 1.03*
PND 21
0.0
ND
ND
0.1
0.41 ± 0.11*
0.21 ±0.04*
2.0
4.46 ± 1.82"
3.16 ± 0.11**
Notes: ND = not detected; PND = postnatal day.
*p < 0.05 compared with control.
** p < 0.01 compared with control.
aData presented as mean ± standard deviation (|xg/mL).
bData presented as mean ± standard deviation (|xg/g).
B.2.3.2.2 Mice
Borg et al. (2010) administered a single dose of 12.5 mg/kg 35S-PFOS by intravenous injection
(n = 1) or gavage (n = 5) on GD 16 to C57B1/6 dams. Using whole-body autoradiography and
liquid scintillation, counting distribution of PFOS was determined for the dams/fetuses (GD 18
and GD 20) and neonates (PND 1). Distribution of PFOS in the dams was similar regardless of
the route of exposure, with the highest levels in the liver and lungs at all timepoints (liver and
lung PFOS levels approximately 4 times and 2 times that of blood, respectively). The distribution
of PFOS in the kidneys was similar to blood and the amount in the brain was lower than that of
the blood. In the fetuses, the highest concentrations of PFOS were found in the kidneys and liver.
In the kidneys, the highest concentration of PFOS was observed in the fetuses on GD 18 (3 times
higher than maternal levels). In the fetuses on GD 18, values in the lungs were similar to the
maternal lungs, and this value Increased by GD 20.
Accumulation in fetal liver was also observed C57BL/6 mice (Lai et al., 2017b). In the offspring
at all timepoints, PFOS was homogeneously distributed in the liver at a level 2.5 times higher
than maternal blood and 1.7 times lower than maternal liver. In pups on PND 1, PFOS was
B-39
-------
APRIL 2024
mostly concentrated in the lungs and liver. Pups on PND 1 had PFOS levels that were 3 times
higher in the lungs compared with maternal blood with a heterogeneous distribution. In the
kidneys, the levels in pups on PND 1 were similar to their respective dams despite being higher
in fetuses on GD 18. Levels in the brain were similar at all timepoints in the offspring and higher
than in the maternal brain, likely due to an immature brain-blood barrier. Select data are
provided in Table B-16.
Table B-16. Concentration Ratios of 35S-PFOS Maternal Serum to Various Organs of
C57BL/6 Mouse Dams, Fetuses, and Pups (Lai et al., 2017b)
[35S-PFOS]orga„/[35S-PFOS]mater„al blood
Group
Liver3
Lungs3
Kidneys3
Brain3
Bloodb
(n = 6-8)
(n = 5-6)
(n = 3-6)
(n = 6-9)
(n = 1-6)
Dams
4.2** ±0.7
2.0* ±0.4
0.9 ±0.1
0.2" ±0.05
1.0
Fetuses on GD 18
2.6**± 0.8
2.1* ± 0.6
2.8**± 0.3
1.2 ±0.3
2.3
Fetuses on GD 20
2.4**± 0.5
2.5**± 0.4
1.4 ±0.2
0.9 ±0.1
1.1 ±0.04
Pups on PND 1
2.4*± 0.4
3.0**± 0.5
1.0 ±0.5
0.9 ±0.2
1.7" ±0.3
Notes: S-PFOS = S-radioisotope perfluorooctance sulfonic acid; GD = gestation day; PND = postnatal day.
'Statistically significant (p < 0.01) in comparison to maternal blood.
"Statistically significant (p < 0.001) in comparison to maternal blood.
aData presented as mean ± standard deviation (|xg/g).
bData presented as mean ± standard deviation (|xg/mL).
Male and female KM mice were administered PFOS by subcutaneous injection one time on PND
7, PND 14, PND 21, PND 28, or PND 35 at concentrations of 0 mg/kg or 50 mg/kg (Liu et al.,
2009). Animals were killed 24 hours after treatment and the PFOS concentration levels obtained.
The percent distribution found in the blood, brain, and liver are provided in Table B-17. The
distribution shows that, beyond PND 14, the levels in the liver are approximately 2-4 times
greater than those found on PND 7.
Table B-17. Percent Distribution of PFOS in Male and Female KM Mice After 50 mg/kg
Subcutaneous Injection (Liu et al., 2009)
Males
Females
PND
Blood3
Brainb
Liverb
Blood3
Brainb
Liverb
7
14
21
28
35
11.78 ±2.88
13.78 ± 1.52
9.85 ±2.74
9.89 ±2.94
13.33 ±0.89
5.04 ± 1.49
1.61 ±0.80"
2.40 ± 0.60"
0.85 ±0.19"
1.02 ±0.28"
14.84 ±4.01
26.50 ±7.36
51.35 ± 11.06*
63.39 ± 19.78*
73.68 ±6.86"
10.77 ± 1.16
12.31 ±2.24
12.37 ±3.80
12.16 ±2.32
11.54 ± 1.28
4.17 ± 1.17
3.26 ±0.58
2.14 ±0.38**
2.10 ±0.73**
0.90 ±0.23**
16.23 ±4.84
26.30 ±4.54
51.48 ±3.44"
51.05 ± 10.59*
69.92 ± 18.52*
Notes: PND = postnatal day
"Statistically significant from PND 7 (p < 0.01).
aData presented as mean percentage ± standard deviation (|xg/mL).
bData presented as mean percentage ± standard deviation (|xg/g).
B-40
-------
APRIL 2024
B.2.4 Volume of Distribution
B.2.4.1 Human Studies
None of the available studies provide data for calibration of the volume of distribution (Vd) of
PFOS in humans. However, several researchers have attempted to characterize PFOS exposure
and intake in humans (Egeghy and Lorber, 2011; Thompson et al., 2010) through
pharmacokinetic modeling. In the models discussed below, Vd was defined as the total amount of
PFOS in the body divided by the blood or serum concentration.
Both research groups defined a Vd for humans using a simple, first-order, one-compartment
pharmacokinetic model (Egeghy and Lorber, 2011; Thompson et al., 2010). The models
developed were designed to estimate intakes of PFOS by young children and adults (Egeghy and
Lorber, 2011) and the general population of urban areas on the east coast of Australia
(Thompson et al., 2010). In both models, the Vd was calibrated using human serum concentration
and exposure data from NHANES, and it was assumed that most PFOS intake was from
contaminated drinking water. Thus, the value for Vd was calibrated so that model prediction of
elevated blood levels of PFOS matched those seen in the study population.
Thompson et al. (2010) adjusted the Vd for PFOS (230 mL/kg) based on the calibrated PFOA
data by 35% in accordance with the differences in PFOA and PFOS volumes of distribution
calculated by Andersen et al. (2006). The original Andersen et al. (2006) model was developed
from oral data in monkeys and optimized a Vd of 220 mL/kg for PFOS and 140 mL/kg for
PFOA. Thus, the Vd in monkeys for PFOS was approximately 35% greater than that for PFOA in
the optimized models. Therefore, Thompson et al. (2010) used a Vd of 230 mL/kg for humans in
their model.
Egeghy and Lorber (2011) used high and low bounding estimates of 3,000 mL/kg and
200 mL/kg for Vd since data in humans were not available. The two separate estimates of Vd
were used in a first-order, one-compartment model to estimate a range of intakes of PFOA. They
concluded that the Vd was likely closer to the lower value based on a comparison of predicted
modeled intake with estimates of intakes based on exposure pathway analyses. Use of the lower
value gave a modeled intake prediction similar to that obtained by a forward-modeled median
intake based on an exposure assessment. The authors concluded that the lower value of
200 mL/kg was appropriate for their analysis.
Both of the models described above used a Vd calibrated from actual human data on serum
measurements and intake estimates. A calibration parameter obtained from human studies, where
constant intake was assumed and blood levels were measured, is considered a more robust
estimate for Vd than that optimized within a model developed from animal data.
The application of Vd values used in several modeling studies are shown in Table B-18. A single
value of 239 mL/Kg has been uniformly applied for most PFOS studies. Gomis et al. (2017) used
a Vd of 235 mL/kg by averaging Vd values estimated for both humans and animals. Vd values
may be influenced by differences in distribution between males and females, between pregnant
and nonpregnant females, and across serum, plasma, and whole blood.
B-41
-------
APRIL 2024
Table B-18. Summary of PFOS Volume of Distribution Values Assigned in Human Studies
AUC or
Mean/Median
Notes and
Study
Population
Sex
Compartment
y Concentration
Measured in
Compartment
(ng/mL)
Considerations; Was
Steady State
Achieved?
Zhang et
Adult
Males
Whole blood
230 mL/kg Mean: 12.8; GM: 8.62
Steady state assumed.
al.
and
(2015b)
females
Pregnant,
adult
Females
Whole blood
230 mL/kg Mean: 14.7; GM: 13.4
Steady state not
assumed due to variable
PFAS levels during
pregnancy.
Worley et
al. (2017)
>12 yr
Males
and
Females
Blood (2016)
230 mL/kg Mean: 23.4 (18.5,
bodyweight 28.4)
>12 yr
Males
and
Females
Blood (2010)
230 mL/kg Mean: 39.8 (30.9,
bodyweight 48.9)
Fu et al.
Adult,
Males
Serum
230 mL/kg Mean: 5,624; median:
-
(2016)
occupational
and
females
1,725
Zhang et
Adults
Males
Serum and
230 mL/kg Mean: 31
-
al.
and
whole blood
(2013c)
Females
Gomis et
Humans and
Males
Serum
235 mL/kg Reports an average of
Authors note that due to
al. (2017)
Animals
and
Females
human and animal Vd
values
declining values in U. S.
and Australian
populations, steady state
was not achieved.
Notes: AUC = area under the curve; GM = geometric mean; Vd = volume of distribution; U.S. = United States; yr = years.
B.2.4.2 Animal Studies
The Chang et al. (2012) series of pharmacokinetic studies on rats, mice, and monkeys described
above, included Vd calculations. Values for all species were calculated following a single oral or
IV dose of PFOS. In accordance with these studies, the authors concluded that the Vds for
monkeys, rats, and mice are likely in the range of 200-300 mL/kg.
Two recent studies in rats (Huang et al., 2019a; Kim et al., 2016b) measured toxicokinetic
parameters including Vd (Table B-19). In the Kim et al. (2016b) study, Vd values were calculated
as Dose x AUMC/(AUC0 -co)2, where AUMC is the area under the first moment curve. Rats were
dosed with 2 mg/kg PFOS by both oral and IV routes. Vd values were higher after oral
administration (382.55 ± 17.59 mL/kg in males and 351.50 ± 19.20 mL/kg in females) compared
with the IV administration (279.81 ± 16.71 mL/kg in males and 288.97 ± 15.59 mL/kg in
females), but results between the sexes were similar. While organ-specific Vd values were not
determined, only the liver exhibited a partition coefficient (Pc) greater than 1, and the liver Pc in
males was significantly higher than the Pc in females (2.63 ± 0.04 and 2.04 ± 0.03, respectively).
This observation may contribute to the slightly lower Vds observed after IV administration in
B-42
-------
APRIL 2024
males relative to females. Pcs in other tissues were 1 (kidney, lung) or 2 (heart, spleen), lower
that those observed in the liver for both males and females.
Huang et al. (2019a) calculated the apparent volume of central (Vi) and peripheral (V2)
distribution in rats using standard equations (Gabrielsson and Weiner, 2000). In this study, a
two-compartment model was the best fit for male rats for both IV and gavage routes of
administration and females dosed by the IV route whereas a one-compartment model was the
best fit for female rats dosed by oral gavage. As detailed in Table B-19, males and females were
administered the same dose (2 mg/kg) used by Kim et al. (2016b). In males, Vd values by the IV
route were 417 ± 31 mL/kg and 264 ±71 mL/kg in the central and peripheral compartments,
respectively. Interestingly, it was the Vd in the peripheral compartment that was most similar to
that observed by Kim et al. (2016b). Vd values in females after IV administration were lower
than that observed in males in both the central and peripheral compartments (297 ± 43 mL/kg,
and 124 ± 62 mL/kg, respectively). For the oral route, striking sex differences were noted
between the central and peripheral compartments. While Vd values were quite similar in males
(244-280 mL/kg) for both compartments, they were notably higher in the central compartment
(222 ± 84 mL/kg) compared with the peripheral compartment (93.4 ± 93 mL/kg) in females.
In a third study (Iwabuchi et al., 2017), PFOS was administered to male Wistar rats as a single
bolus dose (BD) and Vd was measured as BD/elimination rate constant (ke) x plasma
concentration (AUC). Vd values were calculated for whole blood, serum, and several tissues. The
Vd of whole blood was much higher than that observed for serum (2.5 kg tissue volume/g bw and
0.96 kg tissue volume/kg bw, respectively). Organ Vd values were highest in the brain (7.9 kg
tissue volume/kg bw), heart (4.5 kg tissue volume/kg bw) and spleen (2.8 kg tissue
volume/kg bw). VdS were lower by 1 (kidney) or 2 (liver) orders of magnitude. Interestingly, for
this analysis of PFOS, the body organs behaved as an assortment of independent one-
compartments, with a longer elimination half-life in liver than serum in the elimination phase.
B-43
-------
APRIL 2024
Table B-19. Summary of PFOS Volume of Distribution in Rats
Study
Method of
Vd
Calculation
Route Dose Species Age Sex
Vd Compartment
AUC or
Mean/Median
Concentration
Measured in
Compartment
Steady State
Cmax Consideratio
ns
Kim et al. Dose x AU IV
(2016b) MC/(AUC0-
co) 2
2 mg/kg Sprague-
Dawley
8-12 wk Males
382.55 ± 17. Blood Plasma
59 mL/kg
AUC:
216.47 ±8.63 (ig
day/mL
5.23 ± 0.24 (i NR
g/mL
Females 351.50 ± 19. Blood Plasma
20 mL/kg
AUC:
203.60 ±8.42 (ig
day/mL
5.69 ±0.33 (iNR
g/mL
Oral 2 mg/kg Sprague-
Dawley
8-12 wk Males
279.81 ± 16. Blood plasma
71 mL/kg
AUC:
272.69 ±20.39 (i
g day/mL
6.71 ± 0.30 (iNR
g/mL
Females 288.97 ± 15. Blood Plasma
59 mL/kg
AUC:
234.61 ± 10.05 (i
g day/mL
6.66 ± 0.29 (i NR
g/mL
Huang et Standard
al. (2019a) equations
(Gabrielsson
and Weiner,
2000)
IV 2 mg/kg Sprague-
Dawley
I wk Males
417 ±31 m Central
L/kg
AUC:
7.32 ±0.42 |iM-
hr
0.01 ±0.01 NR
mM
264 ± 71 m Peripheral
L/kg
AUC:
7.32 ±0.42 |iM-
hr
0.01 ±0.01 NR
mM
Females
297 ± 43 m Central
L/kg
AUC:10.72 ± 0.7
8 |xM-hr
0.01 ±0.01 NR
mM
124 ± 62 m Peripheral
L/kg
AUC:10.72 ± 0.7
8 |xM-hr
0.01 ±0.01 NR
mM
Oral 2 mg/kg Sprague-
Dawley
I wk Males
280 ± 48 m Central
L/kg
244 ± 81 m Peripheral
L/kg
AUC:
9.86 ± 0.74 |iM-
hr
AUC:
9.86 ± 0.74 |iM-
hr
0.01 ±0.01 NR
mM
0.01 ±0.01 NR
mM
Females 222 ± 84 m Central
L/kg
AUC:
17.74 ± 1.02 |iM-
hr
0.02 ±0.01 NR
mM
B-44
-------
APRIL 2024
Study
Method of
Vd
Calculation
Route
Dose Species Age
Sex
Vd
Compartment
AUC or
Mean/Median
Concentration
Measured in
Compartment
Steady State
Cmax Consideratio
ns
93.4 ± 93 m Peripheral
L/kg
AUC:
17.74 ± 1.02 |iM-
hr
0.02 ±0.01 NR
mM
2 mg/kg Sprague-
(x5 d) Dawley
I wk
Males 176 ± 27 m Central
L/kg
123 ± 42 m Peripheral
L/kg
AUC:
58.18 ±3.00 |iM-
hr
AUC:
58.18 ±3.00 pM-
hr
0.11±0.01 NR
mM
0.11±0.01 NR
mM
Females
136 ± 25 m Central
L/kg
AUC:
89.18 ±5.00 |iM-
hr
0.14 ±0.02 NR
mM
86.3 ±37.3 Peripheral
mL/kg
AUC:
89.18 ±5.00 |iM-
hr
0.14 ±0.02 NR
mM
20 mg/kg
Sprague- 8 wk
Dawley
Males 34.6 ± 4.8 m Central
L/kg
43.9 ± 7.7 m Peripheral
L/kg
AUC:
149.76 ± 10.60 (i
M-hr
AUC:
149.76 ± 10.60 (i
M-hr
AUC: NR
0.21± 0.03 (i
M-hr
AUC: NR
0.21± 0.03 (i
M-hr
Females 27.9 ± 4.7 m Central
L/kg
27.5 ± 6.5 m Peripheral
L/kg
AUC:
213.94 ± 16.00 (i
M-hr
AUC:
213.94 ± 16.00 (i
M-hr
AUC: NR
0.27 ±0.03 (i
M-hr
AUC: NR
0.27 ±0.03 (i
M-hr
Iwabuchi
Dose/elimin Oral
100 (ig/kg Wistar 7-9 wk Males
7.9 kg tissue Brain
180 (ig/kg tissue
9.17 (ig/kg NR
et al.
ation rate
at start of
volume/kg
volume - day
tissue
(2017)
constant
exposure
BW
volume
(ke) x plasm
4.5 kg tissue Heart
380 (ig/kg tissue
27.7 (ig/kg NR
a
volume/kg
volume - day
tissue
BW
volume
B-45
-------
APRIL 2024
Study
Method of
Vd
Calculation
Route
Dose Species Age
AUC or
Mean/Median
Sex Vd Compartment Concentration
Measured in
Compartment
Steady State
Consideratio
ns
concentratio
0.043 kg Liver
240,000 (ig/kg
2,730 (ig/kg NR
n (AUC).
tissue
tissue volume -
tissue
volume/kg
day
volume
BW
2.8 kg tissue Spleen
650 (ig/kg tissue
46.9 (ig/kg NR
volume/kg
volume - day
tissue
BW
volume
0.85 kg Kidney
2,300 (ig/kg
197 (ig/kg NR
tissue
tissue volume -
tissue
volume/kg
day
volume
BW
2.5 kg tissue Whole blood
1,800 (ig/kg
52.6 (ig/kg NR
volume/kg
tissue volume -
tissue
BW
day
volume
0.96 kg Serum
2,200 (ig/kg
127 jig/kg NR
tissue
tissue volume -
tissue
volume/kg
day
volume
BW
Notes: AUMC = area under the first moment curve; AUC = area under the curve; BW = body weight; Cmax = Maximum concentration achieved; IV = intravenous; NR = not
reported; Vd = volume of distribution; wk = weeks.
B-46
-------
APRIL 2024
Unlike the sex differences observed in rats, Vd calculations were similar in male and female
monkeys as shown in Table B-20 (Chang et al., 2017). Young adult cynomolgus monkeys
(Macaca fascicularis) (6 per sex) were sham-dosed with vehicle, a single dose of PFOS
(9 mg/kg, low-dose group), or three separate PFOS doses (11-17.2 mg/kg, high-dose group).
Blood samples were drawn from all monkeys prior to, during, and after PFOS administration for
up to 1 year. Toxicokinetic parameters were determined using a noncompartmental analysis. At
the lower dose, a Vd of 127 mL/kg was calculated for both males and females. At the higher
dose, the Vd in males was calculated to be 135 mL/kg. Vd was slightly higher in females
(141 mL/kg).
Table B-20. Pharmacokinetic Parameters After Acute PFOS Exposure in Cynomolgus
Monkeys3 (Chang et al., 2017)
Parameter
9 mg/kg
14 mg/kg
Male
Female
Male
Female
Ti/2 (day)
Kei (1/day)
CI (mL/day/kg)
Vd (mL/kg)
AUC/dose
(ng/day/mL/mL/kg)
124 ±3.89
0.00559 ±0.000175
0.712 ±0.0812
127 ± 10.9
271,333 ±21,733
102 ±29.2
0.00729 ± 0.00223
0.897 ±0.196
127 ± 18.9
265,200 ± 15,057
117 ± 17.2
0.00605 ±0.000951
0.816 ±0.111
135 ±6.69
249,667 ± 14,468
102 ±45.6
0.00757 ± 0.00270
1.06 ±0.510
141 ±38.5
220,333 ±9,019
Notes: AUC/dose = area under the curve per dose; CI = clearance; Kei = elimination rate per day; T1/2 = half-life (time);
Vd=volume of distribution.
a Data presented in mean ± standard deviation.
B.3 Metabolism
A summary of studies that provide information on PFOS metabolism from recent
literature search and review efforts conducted after publication of the 2016 PFOS
shown in Figure B-3.
Evidence Stream Grand Total
Animal
Human
In Vitro
Grand Total 1
Figure B-3. Summary of PFOS Metabolism Studies
Interactive figure and additional study details available on HAWC.
a Figure does not include studies discussed in the 2016 PFOS HESD or those that solely provided background information on
toxicokinetics.
b Select reviews are included in the figure but are not discussed in the text.
systematic
HESD is
The literature contains no studies on the metabolism of PFOS. It appears that PFOS is not further
metabolized once absorbed. Several studies investigating PFOA found no evidence of
metabolism (U.S. EPA, 2016d), and it is likely that PFOS is similarly resistant to metabolism in
humans, primates, and rodents.
B-47
-------
APRIL 2024
B.4 Excretion
A summary of studies that provide information on PFOS excretion from recent systematic
literature search and review efforts conducted after publication of the 2016 PFOS HESD is
shown in Figure B-4.
Evidence Stream Grand Total
Animal
6
Human
29
In Vitro
2
Grand Total
35
Figure B-4. Summary of PFOS Excretion Studies
Interactive figure and additional study details available on HAWC.
a Figure does not include studies discussed in the 2016 PFOS HESD or those that solely provided background information on
toxicokinetics.
b Select reviews are included in the figure but are not discussed in the text.
B.4.1 Urinary and Fecal Excretion
B.4.1.1 Human Studies
A study in uremic patients illustrates the importance of kidney function in urinary PFOS
excretion (Liu et al., 2018c). Uremic patients exhibit higher concentrations of PFOS than the
general population, indicating the important role of urinary excretion in PFOS elimination.
Interestingly, PFOS can be removed by dialysis, and serum PFOS is negatively correlated with
number of hours of dialysis (p = 0.029). Three additional studies investigated urinary excretion
of PFOS in humans in detail. Zhang et al. (2015b) derived estimates for PFOS's urinary
excretion rate using paired urine and blood samples from 54 adults (29 male, 25 female, ages 22-
62) in the general population and 27 pregnant females (ages 21-39) in Tainjin, China. Urinary
excretion was calculated by multiplying PFOS concentration in first-draw morning urine samples
by the predicted urinary volume (1.6 L/day for males and 1.2 L/day for females). PFOS was
detected in the blood samples for all participants but only for 48% of the urine samples from the
general population (mostly males) and 11% of samples from the pregnant females. Total daily
PFOS intake was modeled for the general population with a geometric mean of 89.2 ng/day,
resulting in an estimated daily urinary excretion rate of 16% of the estimated total daily intake
for PFOS. There was no significant difference in excretion rate between males and females, but a
significantly (p = 0.015) higher rate among the younger adults. Nonpregnant females aged 21-50
had a higher urine:blood ratio than those age 51-61 (0.0018 and 0.0006, respectively). A lower
urine:blood ratio was found in pregnant females compared with nonpregnant females (0.0004
and 0.0013, respectively), suggesting the placenta and cord blood as possible elimination
pathways.
Zhang et al. (2013c) measured renal clearance of PFOS in 86 paired blood and morning urine
samples from healthy volunteers in Hebei province, China. The calculated median renal
clearance rates of 0.044 mL/kg/day in young women and 0.024 mL/kg/day in men and older
women for total PFOS. The authors also observed that major branched PFOS isomers were more
efficiently excreted than the corresponding linear isomer.
B-48
-------
APRIL 2024
In a later study, Fu et al. (2016) determined renal clearance of PFOS, PFOA and PFHxS in 302
occupational workers (213 male, 89 female) from one of the largest producers of PFOS-related
compounds in China. Paired serum and urine samples were collected. Mean and median urine
concentrations for PFOS among all workers were 4.4 ng/mL and 1.2 ng/mL, respectively; in
serum, the mean and median concentrations PFOS were 5,624 and 1,725 ng/mL, respectively.
The correlation coefficient of PFOS concentrations in paired serum and urine samples of 0.72
was found to be highly statistically significant (p < 0.01), suggesting that urine concentrations
could serve as effective bioindicators for PFOS exposure in occupational settings. Daily renal
clearance was calculated for each PFAA as follows:
Urine PFAA Concentrations Daily x Daily urine excretion volume
Serum PFAA concentrations x Body weight
Urine excretion volumes were assigned as 1.4 L/day and 1.2 L/day for males and females,
respectively), and body weight as reported in questionnaires. The daily renal clearance was the
highest for PFOA (GM 0.067 mL/kg/day) and lowest for PFOS (GM 0.010 mL/kg/day). Sex did
impact PFOS daily renal clearance values, which were significantly lower in males compared
with females (p < 0.01).
Fu and colleagues noted their half-life estimates are the shortest values ever, suggesting that the
overall elimination potential of PFAAs might have been underestimated. The shorter half-life
values presented could suggest that pathways other than renal clearance play important roles in
elimination of PFAAs in humans. Another possibility is that the apparent half-lives of PFAAs
calculated through annual decline rates could be affected by the high ongoing levels of exposure.
B.4.1.2 Animal Studies
In a study by Chang et al. (2012), three Sprague-Dawley rats/sex/timepoint were administered
[i4C]pfos as the potassium salt, one time by oral gavage at a dose of 4.2 mg/kg. Urine and feces
were collected after 24 and 48 hours. The amounts recovered in urine and feces were
approximately equivalent at each time point: 1.57% and 1.55%, respectively, at 24 hours and
2.52% and 3.24%, respectively, at 48 hours.
Further investigation by Kim and colleagues measured the amounts of unchanged PFOS excreted
into the urine and the feces of male and female Sprague-Dawley rats with a single dose of
2 mg/kg by oral or intravenous administration (Kim et al., 2016b). After dosing, urine and feces
were measured weekly throughout the 70-day study period. The highest concentrations were
found in urine under all conditions. In males, the levels detected in urine (76.13 ± 16.83 jag) and
feces (61.65 ± 7.29 jag) were similar after oral administration. After intravenous dosing, urine
levels in males (103.04 ± 21.56 jag) were more than 2-fold higher than fecal levels
(43.73 ± 5.29 (j,g). Females also excreted higher levels in urine compared with feces by both
dosing routes. After oral administration, urine and fecal levels were 95.42 ± 22.14 jag and
53.29 ± 8.64 jag, respectively. Similar values in urine (88.29 ± 14.91 jag) and feces
(48.37 ± 4.98 |ig) were measured after intravenous dosing. The similar concentrations in urine
and feces translated to similar half-life estimates for PFOS (26.44 and 28.70 days in males and
23.50 and 24.80 days in females by the oral and intravenous routes).
B-49
-------
APRIL 2024
Another study evaluated repeat dosing in 10 male Sprague-Dawley rats (~9 weeks old)/group
which were administered 0 mg/kg/day, 5 mg/kg/day, or 20 mg/kg/day PFOS by gavage once
daily for 4 weeks (Cui et al., 2010). Urine and feces were collected for 24-hour intervals on the
day prior to treatment (day 0), and days 1, 3, 5, 7, 19, 14, 18, 21, 24, and 28. Both dose groups
exhibited increased excretion over time, with greater excretion rates in the urine. No notable
difference in excretion between the dose groups remained after accounting for decreased food
intake and mortality in the high-dose group.
Another study (Gao et al., 2015) compared concentrations in urine and feces of male and female
Wistar rats. A mixture of PFOA/PFNA/PFOS were administered to rats by drinking water for
90 days, with each compound at doses of 0 mg/L, 0.05 mg/L, 0.5 mg/L, and 5 mg/L. While the
focus of this study was measuring concentrations in the hair of animals (discussed below under
Other Routes of Excretion), the authors measured concentrations of each PFAA in urine and
feces samples by collecting excreta in standard metabolism cages overnight for 24 h intervals on
day 84 (week 12). The intake for each compound was calculated as the drinking volume
multiplied by water concentration of 0.05 mg/L, 0.5 mg/L, and 5 mg/L. In contrast to
observations by others, there were far higher levels of PFOS in feces compared with urine for
both males and females. However, this trend was also observed among female Crl:CD(SD)IGS
VAF/Plus rats by Luebker et al. (2005a), in which five groups of 16 dams each were
administered 0 mg, 0.1 mg, 0.4 mg, 1.6 mg, or 3.2 mg PFOS/kg bw/day by oral gavage
beginning 42 days prior to cohabitation and continuing through GD 14 or GD 20. Urine and
feces were collected overnight from dams on the eve of cohabitation day 1 and during GDs 6-7,
GDs 14-15, and GDs 20-21. The concentrations in the feces were consistently about 5 times
greater than in the urine. It is unclear whether the higher levels of PFOS in feces reflects rat
strain or dose differences among the various studies or is driven by differential excretion
pathways in rats exposed to a mixture of PFAAs.
In summary, limited evidence supports excretion through the fecal route in both animals and
humans. Most studies indicate excretion by the fecal route is substantially lower than that
observed by the urinary route. There are sex-specific differences in excretion of PFOS through
feces. Excretion through the fecal route appears to be more efficient in males compared with
females. Also, in male rats, fecal and urinary concentrations were similar after oral but not
intravenous dosing. Finally, exposures to mixtures of PFNAs suggests that PFOS in the context
of a mixture may be preferentially excreted through the fecal route. The extent to which
resorption by hepatic and enteric routes impacts fecal excretion has not been established in either
humans or animals.
B.4.2 Physiological and Mechanistic Factors Impacting Excretion
B.4.2.1 Renal Resorption
Urinary excretion is the major route of elimination for PFOS. Excretion through urine is
impacted by saturable renal resorption of PFOS from the glomerular filtrate via transporters in
the kidney tubules.
Urinary excretion of PFOS in humans is also impacted by the isomeric composition of the
mixture present in blood and the sex/age of the individuals. The half-lives of the branched chain
B-50
-------
APRIL 2024
PFOS isomers are shorter than those for the linear molecule, an indication that renal resorption is
less likely with the branched chains.
Zhang et al. (2013c) determined half-lives for PFOS isomers based on paired serum samples and
early morning urine samples collected from healthy volunteers in two large Chinese cities. Half-
lives were determined using a one-compartment model and an assumption of first-order
clearance. The mean half-life values for the six branched chain isomers of PFOS were lower than
the value for the linear chain with the exception of the 1-methyl heptane sulfonate, suggesting
that resorption transporters may favor uptake of the linear chain and 1-methyl branched chain
over the other isomers.
B.4.2.2 Enterohepatic Resorption
Early evidence of enterohepatic resorption of PFOS was revealed by Johnson and colleagues
(1984), who demonstrated that cholestyramine (CSM) treatment increased mean cumulative
carbon-14 elimination in feces by 9.5-fold for male CD rats administered 3.4 mg/kg [14C]PFOS.
CSM is a bile acid sequestrant, and its facilitation of PFOS GI clearance suggests enterohepatic
circulation.
Evidence of enterohepatic excretion and potential resorption in humans includes Harada et al.
(2007), in which serum and bile samples from patients (two male and two female; aged 63-76)
undergoing gallstone surgery exhibited higher PFOS levels in the bile than in the serum,
suggesting bile as a route of excretion. The biliary resorption rate was 0.97, which could
contribute to the long half-life in humans. Method of exposure to PFOS was unknown.
Biliary excretion in humans and the potential for resorption from bile discharged to the GI tract
is supported by the Genuis et al. (2010) self-study of the potential for CSM to lower the levels of
PFAS in blood. This was a case report and the sole example of excretion analyzed after
inhalation PFOS exposure. A 51-year-old exposed through carpet treated with soil/dirt repellants
presented with elevated serum levels of perfluorinated compounds including PFOS. After
treatment with CSM for 1 week (ingested 4 g/day, 3 times a day), PFOS serum levels decreased
from 23 ng/g serum to 14.4 ng/g serum. Additionally, the stool concentration of PFOS was
increased from undetectable before treatment (LOD = 0.5 ng/g) to 9.06 ng/g and 7.94 ng/g in the
weeks after treatment, suggesting that it may help with removing PFOS that gains access to the
GI tract with bile.
Table B-21 summarizes enterohepatic transporters identified in liver hepatocytes and intestinal
enterocytes in humans and rats by Zhao and colleagues (2017b; 2015) and suggests that PFOS is
a substrate of both sodium-dependent and -independent enterohepatic transporters involved in
recirculation of bile acids. For these in vitro studies, the authors used transformed ovary (CHO)
and kidney (HEK293) cells stably or transiently transfected with cDNA constructs encoding for
the transporters as well as CHO Flp-In cells expressing human OATP2B.Wild-type CHO cells
and HEK293 cells transfected with vector only were used as controls. With the exception of rat
ASBT, PFOS was demonstrated to be a substrate for all transporters as well as OSTalpha/beta.
Binding efficiency to the enterohepatic transporters was chain-length dependent. Sodium-
taurocholate cotransporting polypeptide (NTCP) transported PFSAs with decreasing affinity but
increasing capacity as the chain length increased (Zhao et al., 2015). The opposite trend was seen
for OATP-mediated uptake (Zhao et al., 2017b). For these five OATPs, PFOS was transported
B-51
-------
APRIL 2024
with the highest affinity compared with transport of PFBS and PFHxS. The authors suggest that
transport efficiency generally increased with the increase in chain length, and that this may, at
least in part, account for the shorter half-lives of short chained versus long chained
perfluoroalkyl sulfonates. While these in vitro studies demonstrate that PFOS is a substrate of
enterohepatic transporters found in the livers and intestines of humans and rats, it is unknown
whether and to what extent these transporters function in vivo.
Table B-21. Enterohepatic Transporters of PFOS
Type
Human Transporters
Rat Transporters
Liver
Intestinal
Liver
Intestinal
Hepatocyte
Enterocyte
Hepatocyte
Enterocyte
Sodium-dependent
(Zhaoet al.,2015)
NTCP
ASBT
NTCP
Sodium-independent
(Zhaoetal., 2017b)
0ATP1B13
OATP1B33
OATP2B13
OATP2B13
0ATP1A13
OATP1B2
OATP2B1
OATP1A5
OATP2B1
Notes: ASBT = human apical sodium-dependent bile salt transporter; NTCP = Na+/taurocholate cotransporting polypeptide;
OATP = organic anion transporting polypeptide.
a Transporter examined in transfection studies; PFOS also shown to be a substrate of these transporters in HEK293 cells
transiently transfected with cDNA constructs encoding these transporters (Zhao et al., 2017b; Zhao et al., 2015).
B.4.3 Maternal Elimination Through Lactation and Fetal
Partitioning
PFOS can readily pass from mothers to their fetuses during gestation and through breast milk
during lactation. In conjunction with elimination through menstruation discussed in Section
BAA, females may eliminate PFOS through routes not available to males. The total daily
elimination of PFOS in pregnant females was estimated to be 30.1 ng/day, higher than the
11.4 ng/day for PFOA (Zhang and Qin, 2014). The ratio of branched:total PFOS isomers in cord
blood was 0.27 and was statistically greater in cord blood compared with maternal blood and
placenta. These findings suggest branched PFOS isomers may transfer to the fetus more readily
than linear forms.
The distribution of PFOS from maternal serum to the fetus and infants is discussed in detail
above (Section B.2.3). A study by Zhang et al. (2013b) exemplifies the routes and amounts of
PFOS eliminated by pregnant females. Paired maternal whole blood and cord blood samples
were analyzed from 32 females from Tianjin, China. The maternal blood concentration of PFOS
was 14.6 ng/mL. The mean levels in the cord blood, placenta, and amniotic fluid were 21%,
56%, and 0.1%, respectively, of those in the mother's blood. Although levels in amniotic fluid
correlated to maternal blood for PFOA, the correlation was poor for PFOS. Nevertheless, in
addition to cord blood, placenta and amniotic fluids are additional potentially substantial routes
of elimination in pregnant females. Blood loss during childbirth could be another source of
excretion.
B-52
-------
APRIL 2024
The elimination of PFOS in pregnant women corresponds to an increase in concentrations in the
placenta. Mamsen et al. (2019) observed an increase in PFOS accumulation from gestational age
50 to 300 days, with male placentas showing higher levels of than female placentas. The authors
estimated a placenta PFOS accumulation rate of 0.13% increase per day during gestation.
Mamsen and colleagues (2017) measured placental samples and fetal organs in relation to
maternal plasma levels of 5 PFAS in 39 Danish women who underwent legal termination of
pregnancy before gestational week 12 (Mamsen et al., 2017). All PFAS were transferred from
mother to fetus with different efficiencies and a significant positive correlation was observed for
fetal age (exposure duration) and for fetal:maternal plasma ratios for all PFAS compounds. Fetal
organ levels of PFOS were lower than maternal blood. The average concentration of PFOS was
0.6 ng/g in fetal organs compared with 1.3 ng/g in placenta and 8.2 ng/g in maternal plasma.
Increasing fetal PFOS levels with fetal age suggests that the rate of elimination of PFOS from
mother to fetus may increase through the gestational period.
The same group (Mamsen et al., 2019) measured PFOS accumulation in fetal tissues across
the three trimesters from 78 pregnant women who underwent elective pregnancy
terminations and from cases of intrauterine fetal death. Fetal tissues (placenta, liver, lung,
heart, central nervous system (CNS) and adipose) were collected for 38 first trimester
pregnancies, 18 second trimester pregnancies and 22 third trimester pregnancies. PFOS was
above LOQ in 100% of maternal serum samples, in 93% of placenta samples and 76% of
fetal organs. In general, the concentrations of PFOS in fetal tissue increased from first
trimester to third trimester except for liver and heart which showed highest levels in the
second trimester compared with the third trimester. Analysis of the placenta: serum ratio of
PFOA revealed a higher ratio in male fetuses than in female fetuses, but unlike PFOA, the
difference between the sexes did not reach statistical significance. These studies support the
placenta and fetus as important routes of PFOS elimination in pregnant women.
Underscoring the importance of pregnancy as a lifestage when excretion is altered, Zhang et al.
(2015a) observed that the partitioning ratio of PFOS concentrations between urine and whole
blood in pregnant women (0.0004) was significantly lower (p = 0.025) than the ratios found in
nonpregnant women (0.0013) and may be affected by the increase in blood volume during
pregnancy (Pritchard, 1965).
After birth, women can also eliminate PFOS via lactation. Tao and colleagues measured 45
human breast milk samples collected in 2004 from Massachusetts and PFOS (mean 131 ng/L)
and PFOA (mean 43.8 ng/L) were the predominant PFAS compounds measured (Tao et al.,
2008). Elimination through breast was more recently measured in 293 samples collected from
127 mothers in the Children's Health and Environmental Chemicals in Korea (CHECK) Cohort
(Lee et al., 2017). Results were stratified by age, parity, body mass, delivery method, and infant
sex. The median PFOS concentrations in breast milk across all samples was 47.4 ng/L (range of
36.4-63.8 ng/L) and the median concentration for all PFAS chemicals measured was 151 ng/L
(range of 105-212 ng/L). Pooled breast milk samples were measured to follow the time course of
PFOS in breast milk after birth. Concentrations in breast milk measured 30 days after birth were
significantly higher than those measured prior to 7 days after birth. These findings are contrast
with results of Thomsen et al. (2010) that reported that breast milk levels of PFOA and PFOS
decreased by 7% and 3.1%, respectively, during the first month after birth. Demographic factors,
B-53
-------
APRIL 2024
maternal diets, sample sizes, the lactational periods measured may account for these
discrepancies.
Lee and colleagues also observed that parity impacts PFOA levels in breastmilk. While
primiparous mothers showed higher levels of PFOA in breast milk to mothers giving birth to
more than one child (p < 0.05), levels of PFOS were not significantly different between these
two groups. In contrast, another study of a Slovakian cohort, multivariable models estimated that
parous women had 40% lower PFOS (95% CI: —56%, —17%) concentrations in colostrum
compared with nulliparous women (Jusko et al., 2016). The geometric mean concentration in
was 35.3 ng/L for PFOS and 32.8 ng/L for PFOA. These findings are also consistent with higher
PFOS levels (p < 0.001) in second trimester maternal serum (18.1 ± 10.9 ng/mL) than maternal
serum levels at delivery (16.2 ± 10.4 ng/mL), which were higher than the levels found in cord
serum (7.3 ± 5.8 ng/mL; p < 0.001) (Monroy et al., 2008). In this study, samples were measured
in 101 pregnant women at 24-28 weeks of pregnancy, at delivery, and in umbilical cord blood.
PFOS was also measured in maternal serum, cord serum and breast milk from 102 female
volunteers hospitalized between June 2010 and January 2013 for planned cesarean delivery in
Tolouse, France (Cariou et al., 2015). Mean PFOS concentrations were 3.67, 1.38 and 0.040 in
maternal serum, cord serum and breast milk respectively (compared with 1.22, 0.9191 and
0.041 ng/mL for PFOA). The observed ratios of cord and maternal serum for PFOS was 0.38 in
this study, much lower than the ratio of 0.78 for PFOA. However, the ratio between breast milk
and maternal serum was 0.038 ± 0.016 (essentially the same as measured for PFOA). Thus,
PFOS exhibits a low transfer from maternal blood to cord blood and a 10-fold lower transfer
from maternal blood to breast milk.
In summary, partitioning to the fetus and breast milk represent important routes of elimination in
humans, and may account for some of the differences observed for blood and urinary levels of
PFOS by sex and age.
BAA Other Routes of Elimination
Wong et al. (2014) looked at the role of menstrual blood as an excretory pathway to explain the
shorter half-life of PFOS in females than males. They fit a population-based pharmacokinetic
model to six cross-sectional National Health and Nutrition Examination Survey (NHANES)
datasets (1999-2012) for males and females. They concluded that menstruation could account
for about 30% of the elimination half-life difference between females and males. Wong et al.
(2014) did not account for other possible loss pathways of PFOS that are unique to women of
reproductive age such as the amount of blood loss in child delivery, amniotic fluid, breast
feeding. Verner and Longnecker (2015) suggested a need to consider the non-blood portion of
the menstrual fluid and its albumin content in the Wong et al. (2014) estimate for the menstrual
fluid volume. A yearly estimate for serum loss of 868 mL/year by Verner and Longnecker (2015)
compared with the 432 mL/year estimate of Wong et al. (2014) suggests that the menstrual fluid
loss can account for >30% of the difference in the elimination half-life between females and
males.
Two earlier studies supported an association between increased serum concentrations of PFOA
and PFOS and early menopause (Taylor et al., 2014; Knox et al., 2011). However, a re-analysis
of this data (Ruark et al., 2017) suggested that this association could be explained by reversed
B-54
-------
APRIL 2024
causality and more specifically, that pharmacokinetic bias could account for the observed
association with epidemiological data. Also challenging the assumption that this is due to
menstruation, Singer et al. (2018) failed to find evidence of associations between menstrual cycle
length and PFAS concentrations.
Furthermore, Lorber et al. (2015) compared individuals who had undergone blood removal
treatments for medical reasons to menstruating females. Measurements showed lower PFOA and
PFOS concentrations in the groups experiencing regular blood loss. Estimated concentrations
based on a one-compartment model were consistent with measured serum concentrations.
Overall, this study provides data and modeling that support the initial hypothesis that ongoing
blood loss explains lower PFAA concentrations in humans. These authors suggested that factors
other than blood loss, such as exposure to or disposition of PFOA/PFOS, may also help explain
the differences in elimination rates between males and females. Curiously, studies providing
direct measurements of PFOS in menstrual blood were not identified. However, for PFOS to be
selectively retained from the blood lost through menstruation would require a specific
mechanism for that process and no such mechanism has been demonstrated or proposed.
Hair has been demonstrated as a route of elimination in animals (Gao et al., 2015). Adult male
and female Wistar rats were exposed via drinking water to 0 mg/L, 0.05 mg/L, 0.5 mg/L, and
5 mg/L of PFOS, PFNA, and PFOA for 90 days. At the end of the exposure period, dorsal hair
samples were collected, washed twice in Triton buffer to remove external contaminants and
alkaline digested to extract PFAS. PFOS was detected in hair samples of all the treatment
groups, suggesting a potential route of elimination. Hair from male and female rats contained
PFOS concentration ranged from 20.3 ng/g to 2,086 ng/g in 0.05 mg/L and 5 mg/L treatment
groups, respectively. Notably, the PFOS concentration in hair was significantly higher than the
levels of PFOA (3.31-444 ng/g) and PFNA (14.2-1,604 ng/g) at 0.05 mg/L to 5 mg/L doses.
Unlike PFOA and PFNA which showed a sexually dimorphic pattern, where male rats have
significantly higher hair concentrations than female rats, hair PFOS levels were lower in males
of the 0.05 mg/L group than females of the same dose group and there were no significant
differences in hair PFOS concentrations between males and females of the 0.5 mg/L and 5 mg/L
dose.
Gao et al. (2015) also measured the composition of the mixture excreted in in urine, feces, and
hair after administration of 0.5 mg/L or 0.05 mg/L of a mixture of PFAS (Table B-22). At the
lower dose of 0.05 mg/L, PFOS was the dominant constituent in urine of males and made up a
smaller proportion of total mixture excreted in hair but not feces. In females however, PFOA was
the predominant constituent excreted in urine, but made up the minority constituent excreted in
feces and especially in hair. These findings underscore the impact of mixtures and sex on PFOA
excretion.
Table B-22. Estimated Percentage of the Sum of PFOS, PFNA, and PFOA in Excreta and
Serum of Male and Female Wistar Ratsa as Reported by Gao et al. (2015)
Sex
PFAA
Serum
Urine
Feces
Hair
Males
PFOS
24.6
89.0
20.8
30.0
PFNA
59.9
11.0
53.0
45.4
PFOA
15.6
ND
26.1
24.6
B-55
-------
APRIL 2024
Sex
PFAA
Serum
Urine
Feces
Hair
Females
PFOS
89.0
ND
62.4
78.0
PFNA
11.0
38.9
21.7
18.0
PFOA
ND
61.1
16.1
4.2
Notes: ND = not detected; PFAA = perfluoroalkyl acid; PFNA = perfluorononanoic acid.
aData are presented in % total perfluoroalkyl acids administered. Animals exposed to 0.05 mg/L in Gao (2015)
A single case report study (Genuis et al., 2010) examined PFOS excretion through sweat. PFOS
was measured in sweat as well as urine and stool from a single male subject exposed to
perfluorinated chemicals via inhalation exposure and subjected to treatment with bile
sequestrants. With the exception of PFHxS, no other PFAS chemicals, including PFOS, were
detected in sweat.
Thus far, no single study has conducted a comparative analysis of elimination of PFOS through
all possible routes of excretion. A comprehensive analysis stratified by age and sex would be
necessary to advance the understanding PFOS excretion by all possible routes, and to establish
factors that influence the proportion of PFOS excreted through urine versus other excreta
matrices.
B.4.5 Half life Do to
B.4.5.1 Overview
In general a half-life represents elimination by all routes, which includes metabolism for other
chemicals, but because PFOA/PFOS are not metabolized, it can be interpreted for excretion
(after correction for body weight (BW) changes). The calculated values of PFOS half-lives
reported in the literature vary considerably, which poses challenges in predicting both the routes
and rates of excretion. Several interrelated physiological and mechanistic factors impacting
excretion are summarized here:
1. The capacity of PFOS to be reabsorbed via renal and enterohepatic routes of excretion
and binding affinities to relevant transporters including organic anion transporters
(OATs), organic anion transporting polypeptides (OATPs), MRPs, and sodium-
dependent transporters involved in bile acid transport including NTCP and the apical
sodium-dependent bile acid transporter (ASBT). Exposure to high levels of PFOS under
acute conditions (e.g., contaminated drinking water) or in occupational settings may
result in saturation of resorption transporters and increased excretion.
2. Binding affinity to serum proteins limiting the concentration of the unbound fraction
available for resorption through renal or enterohepatic transporters. Moreover, binding to
serum proteins may limit passive diffusion of perfluorinated chemicals across the
placental barrier.
3. Phospholipid lipid binding affinity (phospholipidphilicity), which can further reduce the
unbound fraction of PFOS as well as uptake into cells. As reported by Sanchez Garcia
and colleagues, phospholipophilicity shows the highest correlation to cellular
accumulation data compared with other measures of lipophilicity. Also, phospholipid
binding affinity could distinguish between high and low accumulating compounds as well
B-56
-------
APRIL 2024
as half-life measures (Sanchez Garcia et al., 2018).
4. Chain length and branching. The half-lives of the branched chain PFOS isomers are
shorter than those for the linear molecule, an indication that renal resorption is less likely
with the branched chains. Interactions with transporters also vary by chain length.
5. Exposure to mixtures of perfluorinated compounds with differential binding affinities to
transporters, serum binding proteins and phospholipids could impact both the rate and
route of PFOS excretion.
6. Sex and species can influence both the rate and route excretion. First, several elimination
pathways are specific to females including menstruation, pregnancy, and lactation.
Second, sex-specific hormones can impact expression of transporters involved in
resorption. Furthermore, elimination half-lives vary dramatically by species, with much
longer half-lives calculated in humans compared with animals.
Differences between species were observed in studies determining the elimination half-life (T1/2)
of PFOS in rats, mice, monkeys, and humans. Sex differences in rats do not appear to be as
dramatic for PFOS as they are for PFOA (Loccisano et al., 2012b, a).
B.4.5.2 Human Studies
Blood sampling was performed on retirees from the 3M plant in Decatur, Alabama where PFOS
was produced. These samples were taken approximately every 6 months over a 5-year period to
predict the half-life of PFOS. Results ranged from approximately 4 years to 8.67 years (3M,
2002, 2000). Both of these studies exhibited some deficiencies in sample collection and methods.
More recently, Olsen et al. (2007) obtained samples from 26 retired fluorochemical production
workers (24 males and 2 females) from the 3M plant in Decatur, Alabama to determine the half-
life of PFOS. Periodic serum samples (total of 7-8 samples per person) were collected over a
period of 5 years, stored at -80 °C, and at the end of the study, High-performance liquid
chromatography/mass spectrometry was used to analyze the samples. The study took place from
1998 to 2004. The mean number of years worked at the plant was 31 years (range: 20-36 years),
the mean age of the participants at the initial blood sampling was 61 years (range: 55-75 years),
and the average number of years retired was 2.6 years (range: 0.4-11.5 years). The initial
arithmetic mean serum concentration of PFOS was 0.799 [j,g/mL (range: 0.145-3.490 (j,g/mL),
and when samples were taken at the end of the study the mean serum concentration was
0.403 [j,g/mL (range: 0.037-1.740 (j,g/mL). Semi-log graphs of concentration versus time for
each of the 26 individuals were created, and individual serum elimination half-lives were
determined using first-order elimination. The arithmetic and geometric mean serum elimination
half-lives of PFOS were 5.4 years (95% confidence interval (CI): 3.9, 6.9 years) and 4.8 years
(95% CI: 4.1, 5.4 years), respectively.
The rate of serum PFOS decline was measured in residents of two communities exposed to
contaminated municipal drinking water contaminated in Bleking County, Sweden in 2013 (Li et
al., 2018). A biomonitoring program ensued between 2014 and 2016 for residents exposed to
contaminated water and an unexposed community. A subset of residents (age range of 15-
50 years) were included in a panel study to estimate PFOS half-lives. Drinking water PFOS
levels were 8,000 ng/L prior to closure of the waterworks facility and 27 ng/L in the unexposed
B-57
-------
APRIL 2024
community. The mean serum levels among the 106 participants 6 months after the end of
exposure was 387 ± 259 ng/mL. The average decrease in PFOA was 20% of its previous value
each year. The excretion rate constant after the end of exposure was 0.20 (95% CI: 0.19, 0.22)
and was significantly higher in females (0.22) than males (0.15). The mean half-life was
3.4 years and was also significantly shorter in females (3.1 years) than in males (4.6 years).
There was a high level of inter-individual variation in half-lives.
Fu et al. (2016) determined the half-life of PFOS in 302 occupational workers (213 male and 89
female) from one of the largest producers of PFOS-related compounds in China. The half-lives
of PFAAs in workers were estimated by daily clearance rates and annual decline rates of PFAAs
in serum by a first-order model based on fasting blood and urine samples collected over a period
of five years. Mean and median serum concentrations for PFOS among all workers were
5,624 ng/mL and 1,725 ng/mL, respectively, whereas in urine, mean and median PFOS were 4.4
and 1.2 ng/mL. Fu et al. calculated that the renal clearance rate for PFOS ranged from
5.0 x 10 5 mL/kg/day to 0.54 mL/kg/day (Geometric mean of 0.010 mL/kg/day).
Half-lives were calculated by Ln2/k using two approaches. In the first approach, k was defined
as Cltotai/Vd, where Vd stands for the volume of distribution of PFAAs in the human body and
Cltotai represents the total daily PFAAs clearance in the human body. Cltotai was defined as renal
clearance for men and women older than 50, and as the sum of menstrual and renal clearance in
young women. Vd was set to 230 mL/kg for PFOS. In the second approach, k was defined as the
average annual decline rates of PFAAs in workers who participated in this study.
The half-life of PFOS estimated using daily clearance rate of all workers had a geometric mean
and median value of 32.6 and 21.6 years, respectively. However, when measured by annual
decline rate, the half-life of PFOS was estimated to be 1.9 years. The GM values of the half-life
of PFOS for men here was 60.9 years and 8.0 years for women. The authors suggest that half-
lives estimated by the limited clearance route information could be considered as the upper limits
for PFAAs and that the unrealistically long half-lives determined using urine clearance values
may indicate that other clearance play important roles in elimination of PFAAs in humans
including fecal elimination. Another possibility is that the apparent half-lives of PFAAs
calculated through annual decline rates could be affected by the high ongoing levels of exposure.
Calculated half-lives of PFOS were much longer than for PFOA. The authors postulate
differential accumulation kinetics of the pollutants and suggest that PFOS reaches a steady state
much faster than PFHxS and PFOA in humans. The longer half-life estimates for PFOS
compared with PFOA may also reflect its stronger affinity for serum albumin as reported
previously (Salvalaglio et al., 2010). Other factors impacting half-lives could include higher
enterohepatic and renal reabsorption rates of PFOS relative to PFOA. The authors conclude that
the shorter half-lives of PFHxS and PFOS estimated by annual decline compared with those
estimated by daily clearance rates suggest that other important elimination pathways operate to
remove PFOS and might have been underestimated.
Worley and colleagues (2017) calculated PFOS half-lives in community members (age 12-years
old or older) living near a PFAS manufacturer in Alabama that had discharged waste into a local
wastewater treatment plant. Sewage sludge from this plant was applied to local agricultural
fields. In 2010, ATSDR collected blood samples from subjects and followed up with blood and
urine measurements in 2016. Biological half-lives were estimated for PFOA and PFOS using a
B-58
-------
APRIL 2024
one-compartment pharmacokinetic model. Geometric mean serum PFOS concentrations were
significantly higher in subjects in both 2010 (39.8 (J,g/L) and 2016 (23.4 (J,g/L) relative to national
averages reported by NHANES (9.32 [j,g/L in 2009-2010 and 4.99 [j,g/L in 2013-2014).
Half-lives for PFOA and PFOS were estimated to be 3.9 and 3.3 years, respectively. When Vd
and intake values were altered by ± 20%, half-life values varied by several months (half-life
estimates for PFOA and PFOS ranged from 3.5 to 4.1, and 3.0 to 3.6 years). The authors suggest
these parameters have a significant impact on half-life estimates.
Xu et al. (2020c) estimated the half-life of PFAS by sampling urine (4 times) and blood (5 times)
from 26 airport employees between 2 weeks and 5 months after the end of a 2-month exposure to
PF AS-contaminated drinking water. The levels of PFOS in the airport's contaminated water
were 62 ng/L (0.062 ng/mL) for linear PFOS and 64 ng/L (0.064 ng/mL) for branched PFOS.
Specific gravity adjusted urine concentrations for PFOS were generally below detectable limits
for linear and branched forms of PFOS with respective ranges of
-------
APRIL 2024
Table B-23. Summary of PFOS Concentration in Blood and Urine in Relation to Half-life Values in Humans
Study
Number of
Subjects
Age Range
Primary
Exposure
Route
Intake
Plasma/Serum
Concentrations
Urinary
Concentrations
Estimated Half-
Life (y)
Considerations
Xu et al.
26
22-62 yr
Oral
Drinking
Linear PFOS:
Linear PFOS:
Linear PFOS: 2.91,
• 1 woman was
(2020c)
19 Males
water at
Median: 10 ng/mL
mean
-------
APRIL 2024
Study
Number of
Subjects
Age Range
Primary
Exposure Intake
Route
Plasma/Serum
Concentrations
Urinary
Concentrations
Estimated Half-
Life (y)
Considerations
• Clearance rate
was not reported
Fu et al.
302
Males: 19-
Occupational NR
Mean 5,624 ng/mL
Mean: 4.4 ng/mL,
Male (n = 136): GM
• Urinary samples
(2016)
213 Males
65, median
(assuming oral
Median
Median 1.2 ng/mL
60.9
were only taken
89 Females
41
and inhalation
1,725 ng/mL
(not creatinine
Females (n= 71):
from 274
Females:
but not directly
(50.3-
adjusted)
GM 8.0
participants while
19-50,
addressed in
118,000 ng/mL)
Overall (n = 207):
there were serum
Median 37
study)
GM 32.6
samples for every
participant
• For half -life
calculation for
females,
menstrual
clearance was
added to renal
clearance
• PFOS clearance
rate 0.017 mL/kg-
day
Zhang et al.
86
22-68
Unspecified NR
Mean 21 ng/mL
Mean 47 ng/g
Young females: 6.2
• All participants
(2013c)
47 Males
(Oral likely,
Median 19 ng/mL
creatinine
Males and older
had paired (whole
37 Females
Shijazhuang is a
capital city and
Handan is an
industrial city)
(1.4-180 ng/mL)
Branched
Median 28 ng/g
creatinine
(range 2.8-232 ng/g
creatinine)
females: 27
blood/serum and
urine)
• For young
females,
menstrual
clearance was
estimated and
added to renal
clearance.
• Renal clearance
rate for total
PFOS: mean
0.050 mg/kg/day
(young females),
0.037 mg/kg/day
B-61
-------
APRIL 2024
Study
Number of . „
Subjects e Range
Primary
Exposure
Route
Intake
Plasma/Serum
Concentrations
Urinary
Concentrations
Estimated Half-
Life (y)
Considerations
(males and older
females)
Notes: CI = confidence interval; GM = geometric mean; LOD = limit of detection; NR = not reported; yr = years.
B-62
-------
APRIL 2024
All human PFOS half-life values identified in the literature review are provided in Table B-24. A
prominent feature of this data includes a very wide range of values ranging from less than 1 year
in a single male child of 16 years of age (Genuis et al., 2014) to up to 60.9 years for males
occupationally exposed in a plant in China (Fu et al., 2016). Second, with one exception (Genuis
et al., 2014), half-lives estimated for males are longer than those estimated for females. Third,
studies that stratified by ages show an age-related increase in half-life values (Genuis et al.,
2014; Zhang et al., 2013d). Fourth, linear isomers exhibit longer half-lives than branched
isomers (Zhang et al., 2013c).
Gomis et al. (2017) estimated half-lives in the general populations in the U.S. and in Australia
and reported a range of 3.3 to 5.4 years. Olsen et al. (2012b) estimated a similar value in blood
samples from Red Cross volunteer donors of 4.3 years. Interestingly, these values were also in
line with the half-life (2.3 y) estimated for subjects likely exposed to contaminated drinking
water in West Virginia and Ohio (Bartell et al., 2010). Other studies of subjects exposed to
contaminated drinking water in Sweden (Li et al., 2018) estimated half-lives of 3.1 (for females)
to 4.6 years (for males). Among the highest values are those for occupationally exposed workers
that ranged from 8.67 years (retired workers from a PFOS production plant in Decatur, Alabama)
to 60.9 years for workers in Hubei province, China.
While most studies were conducted in adults and/or adolescents, at least one study examined
PFOS half-lives in newborns (Spliethoff et al., 2008). Whole blood was collected as dried spots
on filter paper from almost all infants born in the United States. One hundred and ten of the
Newborn Screening Programs (NSPs) collected in the state of New York from infants born
between 1997 and 2007 were analyzed for PFOS. The analytical methods were validated by
using freshly drawn blood from healthy adult volunteers. The mean whole blood concentration
for PFOS ranged from 0.00081 [j,g/mL to 0.00241 [j,g/mL. The study grouped the blood spots by
two different time points; those collected in 1999-2000 and in 2003-2004, which corresponded
to the intervals reported by NHANES. The PFOS concentrations decreased with a mean value of
0.00243 [j,g/mL reported in 1999-2000 and 0.00174 \iglmL in 2003-2004. The study authors
determined the half-life of PFOS using the regression slopes for natural log blood concentrations
versus the year 2000 and after. The calculated half-life for PFOS was 4.1 years.
B-63
-------
APRIL 2024
Table B-24. Summary of Human PFOS Half-Life Values
Study
Number of Subjects
Age Range
Estimated Half-Life (y)
Subjects
3M (2002)
9
7 males
2 females
61 (55-64)
8.67 ±6.12
(range: 2.29-21.3)
Retirees from the 3M plant in Decatur,
Alabama where PFOS was produced. Derived
from 4 measurements over 18-month time
period from November of 1998 to May of 2000.
Subjects were a subcohort of the C8
Health Project, conducted in 2005-2006, who
had lived in at least one of six affected water
districts near the DuPont Washington Works
plant.
Fu et al. (2016)
302
Males: 19-65, median 41
Based on daily clearance rate
Occupationally exposed subjects working in
213 males
Females: 19-50, median 37
Male (n = 136): GM60.9
one of the largest fluorochemical plants
89 females
Females (n = 71): GM 8.0
(Henxin Chemical Plant) in Yingcheng, Hubei
Overall (n = 207): GM 32.6
province, China
Based on annual decline rate
Overall (n = 207): GM 1.9
Genuis et al. (2014)
53 father
16-53
Father: 1.14
A family (6 patients) identified to have elevated
47 mother
Mother: 1.93
serum concentrations of PFAAs, likely through
22 first male child
First male child: 0.65
repeated commercial spraying of their home
19 second female child
Second female child: 1.03
carpets with stain-repellants. Patients were
17 third male child
Third male child: 0.78
treated by intermittent phlebotomy over a 4- to
16 fourth male child
Fourth male child: 0.61
5-yr period.
Glynn et al. (2012)
413 women
19-41
8.2
Primiparous women 3 wk after delivery in
Uppsala County, Sweden 1996-2010 (the
POPUP study; Persistent Organic Pollutants in
Uppsala Primiparas)
Population-based model using Australian
biomonitoring studies from Toms et al. (2014,
2009) and NHANES from the U.S. A total of
24-84 pools per survey were obtained, with
each pool containing between 30 (2007) and up
to 100 individual samples (2003, 2009 and
2011)
Study reports intrinsic elimination half-lives.
Bartelletal. (2010) 200 54.5 ± 15 2.3
100 males
100 females
Gomis et al. (2017) Australia: A total of 12+ (USA) Australian men: 4.9
24-84 pools per survey <16->60 (Australia) American men: 3.8
containing between 30 Australian women: 5
and 100 individual American women: 3.3
samples.
USA: 2,000 individuals
were sampled
throughout the USA
B-64
-------
APRIL 2024
Study
Number of Subjects
Age Range
Estimated Half-Life (y)
Subjects
Li etal. (2018)
50
Males: 20
Females 30
15-50
Males: 4.6
Females: 3.1
Subjects in Ronneby, Sweden, exposed to
contaminated water through a municipal water
source.
Olsen et al. (2007)
26
24 males
2 females
55-75
5.4
Retirees from the 3M plant in Decatur,
Alabama where PFOS was produced.
Olsen et al. (2012b)
600
Males: 300
Females: 300
5 age groups (20-29, 30-
39, 40-49, 50-59, 60-69)
4.3
Six American Red Cross adult blood donor
centers each provided 100 plasma samples for
analysis of 11 PFAA concentrations in 2010: 10
samples per every 10-yr age interval (20-29,
30-39, 40-49, 50-59, and 60-69) for each sex.
The six American Red Cross blood donor
centers represented the following areas: Boston,
Massachusetts; Charlotte, North Carolina;
Hagerstown, Maryland; Los Angeles,
California; Minneapolis-St. Paul, Minnesota;
and Portland, Oregon
Splitehoffetal. (2008) 240
Newborn infant (1-2 d)
4.1
New York State newborn screening program
blood spot specimens from newborn infants
Wong et al. (2014)
Approx. 2,000 per
dataset (6 datasets)
Males and females
analyzed separately
Eight age groups (age 12-
19, 20-29, 30-39, 40-49,
50-59, 60-69, 70-79, 80+)
Males: 4.7
Females: 3.7
Females (accounting for rate of
menstrual blood loss): 4.0
Population-based pharmacokinetic model
(Ritter) to six cross-sectional datasets from
1999 to 2012 from U.S. NHANES. Data from
age-stratified biomonitoring data for PFOS
extracted from U.S. NHANES from the years
1999-2000, 2003-2004, 2005-2006, 2007-
2008, 2009-2010, and 2011-2012
Worley et al. (2017)
153 (2010)
63 males
90 females
45 (2016)
22 males
23 females
2010: mean 52.0
2016: mean 62.6
3.9 (2010)
3.3 (2016)
Residentially exposed population from
Lawrence, Morgan and Limestone Counties,
Alabama recruited by ATSDR
Xu et al. (2020d)
26
19 males
7 females
22-62 yr
Linear PFOS: 2.91
lm-PFOS: 1.27
3/4/5m-PFOS: 1.09
2/6m-PFOS: 1.04
Subjects in Arvidsjaur, Sweden exposed to
contaminated drinking water occupationally
(working at the airport) and through residential
drinking water
B-65
-------
APRIL 2024
Study
Number of Subjects Age Range
Estimated Half-Life (y)
Subjects
Yeung et al. (2013)
420 20-29
Munster: 4.3
Residents of Munster and Halle, Germany;
Munster: 270
Halle: 4.8
samples collected between 1982 and 2009 in
Halle: 150
Munster and between 1995 and 2009 in Halle.
Zhang et al. (2013c)
86 22-68
47 males
37 females
XPFOS
Young females: 6.2
males and older females: 27
n-PFOS
Young females: 6.7
males and older females: 34
Healthy volunteers in Shijiazhuang and
Handan, Hebei province, China, in April-May
2010
iso-PFOS
Young females: 5.9
males and older females: 24
lm-PFOS
Young females: 10
males and older females: 90
4m-PFOS
Young females: 5.8 y
males and older females: 27
3 + 5m-PFOS
Young females: 5 y
males and older females: 21
£m2-PFOS
Young females: 5.1
males and older females: 14
Notes: ATSDR = Agency for Toxic Substances and Disease Registry; C8HP = C8 Health Project; GM = geometric mean; NHANES = National Health and Nutrition Examination
Survey; PFAA = perfluoroalkyl acid; POPUP = Persistent Organic Pollutants in Uppsala Primiparas; U.S. = United States; USA = United States of America; yr = year.
B-66
-------
APRIL 2024
B.4.5.3 Animal Studies
B. 4.5.3.1 S'onhuman Primates
In the study by Chang et al. (2012), three male and three female monkeys were administered a
single IV dose of PFOS of 2 mg/kg and followed for 161 days. All monkeys were observed twice
daily for clinical signs, and body weights were obtained weekly. Urine and serum samples were
taken throughout the study. There was no indication that elimination was different from males
versus females. Serum elimination half-lives ranged 122-146 days in male monkeys and 88-
138 days in females. Mean values are shown in Table B-25. The Vd values suggest that
distribution was predominately extracellular.
In a second primate study, Seacat et al. (2002) administered 0, 0.03, 0.15, or 0.75 mg/kg/day
potassium PFOS orally in a capsule by intragastric intubation to six young-adult to adult
cynomolgus monkeys/sex/group, except for the 0.03 mg/kg/day group, which had 4/sex, daily for
26 weeks (182 days) in a GLP study. Two monkey s/sex/group in the control, 0.15, and
0.75 mg/kg/day groups were monitored for 1 year after the end of the treatment period for
reversible or delayed toxicity effects. The elimination half-life for potassium PFOS in monkeys
was estimated from the elimination curves as approximately 200 days. This value is consistent
with that reported by Chang et al. (2012) above.
B.4.5.3.2 Rots and Mice
Half-lives rodents are very short relative to those observed in humans and primates (Table B-25).
In mice, Chang et al. (2012) measured slightly higher half-lives in males (36-43 days) compared
with females (30-38 days). Ranges in mice were similar to those observed in rats.
Two recent studies evaluated toxicokinetic parameters informing half-lives in rats (Huang et al.,
2019a; Kim et al., 2016b). In the Kim study, Sprague-Dawley rats were administered 2 mg/kg
PFOS by either the IV or oral route. Urine and feces were collected weekly, and blood was
collected at 10 time points over the first day and then up to 70 days after exposure. Half-lives in
females and males were similar. In females, half-lives of 23.50 ± 1.75 and 24.80 ± 1.52 days
were estimated after oral and IV dosing, respectively. In males, values were slightly longer
(26.44 ± 2.77 and 28.70 ± 1.85 after oral and IV dosing, respectively).
In a similar study (Huang et al., 2019a), male and female Sprague-Dawley rats were
administered a single dose of 2 mg/kg by IV injection or a single dose of 2 mg/kg or 20 mg/kg
by oral gavage and observed from 5 minutes to 20 weeks after dosing. After IV administration of
2 mg/kg, the overall half-life was 22 and 23 days in males and females, respectively days.
Similar values were obtained after a single gavage administration of the same dose (19.9 days in
males and 28.4 days in females) and after repeated dosing by oral gavage (19.0 in males and 21.1
in females). Half-lives in females administered the higher dose of 20 mg/kg were slightly longer
(18 days) than in males (14.5 days) and were slightly longer after repeated administration (19.0
and 21.1. days in males and females, respectively). Half-life values in the terminal elimination
phase were much longer than those measured in the initial elimination phase.
B-67
-------
APRIL 2024
Table B-25. Summary of Animal PFOS Half-Life Values Identified in the Literature Review
Study
„ . , . Exposure
Species and Strain „ ,
r Route
Age or Lifestage
Sex/Half-Life
Approach
Dose
Estimated Half
Life3
Chang et al. (2012)
Cynomolgus Monkey IV
NR
Male
2 mg/kg and followed for
161 d
132 ±7
Female
2 mg/kg and followed for
161 d
110± 15
Oral
4-6 yr
Male
9 mg/kg
14 mg/kg
124 ±3.89
117 ± 17.2
Female
9 mg/kg
14 mg/kg
102 ±29.2
102 ±45.6
Seacat et al. (2002)
Cynomolgus Monkey Oral
Young-adult to
Male
0.15 mg/kg
-200
adult
Female
0.75 mg/kg
-200
Chang et al. (2012)
Mice, CD-I Oral
8-10 wk
Male
1 mg/kg, followed for 20 wk
20 mg/kg, followed for 20 wk
42.81
36.42
Female 1 mg/kg, followed for 20 wk 37.80
20 mg/kg, followed for 20 wk 30.45
Benskinetal. (2012) Rat, Sprague-Dawley Oral Adult (429 g) male 0.4 mg/kg PFOS (0.27 mg/kg n-PFOS: 33.7
n-PFOS) iso-PFOS: 23.4
5m-PFOS: 24.4
4m-PFOS: 23.1
3m-PFOS: 33.8
lm-PFOS: 102
tb-PFOS: 19.6
B7-PFOS: 15.4
B8-PFOS: 11.3
B9-PFOS11.1
Chang et al. (2012) Rat, Sprague-Dawley IV 8-10 wk Male 2 mg/kg, followed for 24 hr 7.99 ± 4.94
Female (1 rat) 2 mg/kg, followed for 24 hr 5.62
Oral 8-10 wk Male 4.2 mg/kg, followed for 144 hr 8.23 ± 1.53
2 mg/kg, followed for 10 wk 38.31 ± 2.32
15 mg/kg, followed for 10 wk 41.19 ± 2.01
Male (1 rat) 2 mg/kg, followed for 24 hr 3.1
Female 2 mg/kg, followed for 24 hr 1.94 ± 0.13
2 mg/kg, followed for 10 wk 62.30 ± 2.09
B-68
-------
APRIL 2024
Study
Species and Strain
Exposure
Route
Age or Lifestage
Sex/Half-Life
Approach
Dose
Estimated Half
Life3
15 mg/kg, followed for 10 wk
71.13 ± 11.25
Huang et al. (2019a)
Rat, Sprague-Dawley
IV
8 wk
Male - overall
elimination half-life
2 mg/kg
22.0 ±2.1
Male - initial phase
2 mg/kg
4.6 ±2.7
Male - terminal
2 mg/kg
39.7 ±4.4
phase
Female - overall
2 mg/kg
23.0 ±3.7
elimination half-life
Female - initial
2 mg/kg
0.3 ±0.3
phase
Female - terminal
2 mg/kg
32.8 ±3.7
phase
Oral
8 wk
Male - overall
elimination half-life
2 mg/kg
2 (x5) mg/kg
20 mg/kg
19.9 ±3.8
19.0 ±3.2
14.5 ±2.1
Male - initial phase
2 mg/kg
2 (x5) mg/kg
20 mg/kg
3.1 ±2.4
0.3 ±0.1
4.0 ±2.9
Male - terminal
2 mg/kg
40.5 ±5.5
phase
2 (x5) mg/kg
20 mg/kg
33.4 ±4.2
35.8 ±4.2
Female - overall
2 mg/kg
28.4 ± 11.0
elimination half-life
2 (x5) mg/kg
20 mg/kg
21.1 ±4.3
18.0 ±3.1
Female - initial
2 mg/kg
0.8 ±2.1
phase
2 (x5) mg/kg
20 mg/kg
0.3 ±0.2
2.2 ±3.0
Female - terminal
2 mg/kg
40.7 ±3.5
phase
2 (x5) mg/kg
20 mg/kg
40.0 ±2.5
36.0 ±4.0
Kim et al. (2016b)
Rat, Sprague-Dawley
IV
8-12 wk
Male
2 mg/kg
28.70 ± 1.85
Female
2 mg/kg
24.80 ± 1.52
B-69
-------
APRIL 2024
Study
„ . , . Exposure
Species and Strain „ a
Route
Age or Lifestage
Sex/Half-Life
Approach
Dose
Estimated Half
Life3
Oral
8-12 wk
Male
2 mg/kg
26.44 ± 2.77
Female
2 mg/kg
23.50 ± 1.75
Notes: d = days; IV = intravenous; NR = not reported; wk = weeks; yr = years.
aData reported in mean days ± standard deviation.
B-70
-------
APRIL 2024
Appendix C. Nonpriority Health Systems Evidence
Synthesis and Integration
C.l Reproductive
The U.S. Environmental Protection Agency (EPA) identified 60 epidemiological and 22 animal
studies that investigated the association between perfluorooctane sulfonic acid (PFOS) and
reproductive effects. Of the epidemiological studies addressing male reproductive endpoints, 2
were classified as high confidence, 15 as medium confidence, 6 as low confidence, and 1 was
considered uninformative (Section C.l.l). Of the epidemiological studies addressing female
reproductive endpoints, 5 were classified as high confidence, 24 as medium confidence, 17 as
low confidence, and 2 were considered uninformative (Section C.l.l). Of the animal studies, 2
were classified as high confidence, 15 as medium confidence, 4 as low confidence, and 1 was
considered mixed {medium/low) (Section C.l.2). Studies may have mixed confidence ratings
depending on the endpoint evaluated. Though low confidence studies are considered qualitatively
in this section, they were not considered quantitatively for the dose-response assessment (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.l.l Human Evidence Study Quality Evaluation and Synthesis
C.l.l.l Male
C.l. 1.1.1 Introduction
The 2016 Health Advisory (U.S. EPA, 2016b) and Health Effects Support Document (HESD)
(U.S. EPA, 2016c) reports identified limited evidence of effects of PFOS on reproductive effects
in men and boys. Analyses of male children in the C8 Health Project (Lopez-Espinosa et al.,
2011) suggested an association between increasing PFOS exposure and delayed onset of puberty,
defined by measured testosterone levels (>50 ng/dL testosterone and >5 pg/mL free
testosterone). The effects of PFOS on semen quality parameters were mixed. In healthy, young
Danish males Joensen et al. (2014) observed significantly inverse associations with testosterone,
calculated free testosterone, free androgen index (FAI), and ratios of testosterone/luteinizing
hormone (LH), free testosterone/LH, and FAI/LH. Significant associations for semen quality
parameters were not observed among these young men. Regarding other studies examining
semen quality parameters, three studies (Buck Louis et al., 2015; Toft et al., 2012; Joensen et al.,
2009) out of nine observed associations with morphologically abnormal sperm. In a cross-
sectional sample of military recruits (n = 105), Joensen et al. (2009) observed significantly lower
sperm counts in men with higher combined perfluorooctane sulfonic acid/perfluorooctanoic acid
(PFOS/PFOA) exposure. A Texas- and Michigan-based cohort (n = 462), the Longitudinal
Investigation of Fertility and the Environment (LIFE) study (Buck Louis et al., 2015), observed
limited evidence of the effects of PFOS. Only one significant association was observed for a
morphological parameter, namely decreased percentage of sperm with coiled tails.
For this updated review, 23 studies5 (24 publications) report on the association between PFOS
and endocrine effects since the 2016 document. Eleven of the studies were in children and
adolescents (Jensen et al., 2020b; Liu et al., 2020b; Di Nisio et al., 2019; Ernst et al., 2019;
5 Zhou, 2016, 3856472 and Zhou, 2017, 3858488 analyze participants from the same population using the same outcome.
C-l
-------
APRIL 2024
Wang et al., 2019a; Goudarzi et al., 2017a; Lind et al., 2017a; Zhou et al., 2017c; Itoh et al.,
2016; Lopez-Espinosa et al., 2016; Zhou et al., 2016), one study was in pregnant women
(Anand-Ivell et al., 2018) and the remainder of the publications were in the general population.
Different study designs were utilized, including four cohort studies (Jensen et al., 2020b; Ernst et
al., 2019; Goudarzi et al., 2017a; Itoh et al., 2016), one case-control study (Anand-Ivell et al.,
2018) with the remainder of the studies following a cross-sectional design. All observational
studies measured PFOS in blood components (i.e., blood, plasma, or serum), however, PFOS
was additionally measured in semen for four studies (Cui et al., 2020; Di Nisio et al., 2019; Pan
et al., 2019; Song et al., 2018) and amniotic fluid in one study (Anand-Ivell et al., 2018). The
studies were conducted in different study populations including populations from Australia,
China, Denmark, the Faroe Islands, Greenland, Italy, Japan, the Netherlands, Poland, Taiwan,
Ukraine, and the United States. There were several pairs of studies investigating the same
population, including the Biopersistent Organochlorines in Diet and Human Fertility
(INUENDO) cohort (Leter et al., 2014; Kvist et al., 2012), the Odense Child Cohort (OCC)
(Jensen et al., 2020b; Lind et al., 2017a), the Genetic and Biomarkers study for Childhood
Asthma (GBCA) (Zhou et al., 2017c; Zhou et al., 2016), and a cross-sectional sample of men
from an infertility clinic in Nanjing, China (Cui et al., 2020; Pan et al., 2019). Two studies
assessed populations from related cohorts belonging to the Hokkaido Study on the Environment
and Children's Health (Goudarzi et al., 2017a; Itoh et al., 2016).
C. 1.1.1.2 Study Quality
There are 24 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and male reproductive effects. Study quality evaluations for these 24 publications
are shown in Figure C-l.
Of the 24 studies identified since the 2016 assessment, two studies were classified as high
confidence, 15 studies as medium confidence, six studies as low confidence, and one study (Song
et al., 2018) was determined to be uninformative. Anand-Ivell, 2018, 4728675 was considered
low confidence for cryptorchidism and uninformative for amniotic fluid hormones. Publications
from the GBCA (Zhou et al., 2017c; Zhou et al., 2016) were rated low confidence because of
concerns of selection bias and confounding. Cases and controls in Zhou, 2017, 3858488 were
drawn from separate sources resulting in some concern for selection bias by recruiting
individuals from different catchment areas. One low confidence study (Di Nisio et al., 2019)
adjusted results only for age, resulting in concerns about potential for residual confounding by
socioeconomic status (SES). One National Health and Nutrition Examination Survey (NHANES)
study (Lewis et al., 2015) did not adjust for the participant sampling design in the analysis which
contributed to a low confidence rating. Song, 2018, 4220306 only reported bivariate correlations
between exposure levels and semen parameters with no accounting for potential confounders
which contributed to the study being classified as uninformative.
C-2
-------
APRIL 2024
Anand-lvell et al., 2018, 4728675
Arbuckle et al„ 2020, 6356900 -
Cui et al, 2020, 6833614-
Di Nisio et al., 2019, 5080655
Ernst et al., 2019, 5080529
Goudarzi et al., 2017, 3981462
Itoh et al., 2016, 3981465
Jensen et al.. 2020, 6311643
Kim et al., 2020, 6833596
Kvist et al., 2012, 2919170
Leter et al., 2014, 2967406
Lewis et al., 2015, 3749030
Lind et al., 2017, 3858512
Liu et al., 2020, 6569227-
Lopez-Espinosa etal., 2016, 3859832
Pan et al., 2019, 6315783-
Petersen et al., 2018, 5080277 -
Song et al., 2018, 4220306 -
Tian etal., 2019, 5390052
Tsai etal., 2015, 2850160-
Wang et al., 2019, 5080598 -
Zhou etal., 2016, 3856472
Zhou et al., 2017, 3858488 -
van den Dungen et al., 2017, 5080340 -
P Legend
Good (metric) or High confidence (overall)
+ Adequate (metric) or Medium confidence (overall)
- Deficient (metric) or Low confidence (overall)
Qj Critically deficient (metric) or Uninformative (overall)
* Multiple judgments exist
Figure C-l. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Male Reproductive Effects
Interactive figure and additional study details available on HAWC.
C-3
-------
APRIL 2024
C.1.1.1.3 Findings From Children and Adolescents
Sex hormone levels and related steroid hormone levels were examined in nine studies (Jensen et
al., 2020b; Liu et al., 2020b; Di Nisio et al., 2019; Wang et al., 2019a; Goudarzi et al., 2017a;
Zhou et al., 2017c; Itoh et al., 2016; Lopez-Espinosa et al., 2016; Zhou et al., 2016) and three
observed significant effects (Appendix D). A high confidence prospective study on the Odense
cohort (Jensen et al., 2020b; Lind et al., 2017a) did not find evidence of effects on steroid
hormones in the sex hormone metabolic pathway (e.g., dehydroepiandrosterone (DHEA), 17-
hydroxyprogesterone (17-OHP)) in four-month-old male infants. Similarly, a prospective cohort
study (Goudarzi et al., 2017a) in boys from the Hokkaido Study on the Environment and
Children's Health reported no significant results with steroid hormones in cord blood. However,
a medium confidence study (Itoh et al., 2016) from a related cohort within the Hokkaido Study
observed a significant positive association (p = 0.033) for estradiol (E2). Increases in E2
potentially contributed to a significant decrease (p = 0.002) in the testosterone-E2 ratio in male
infants. Inverse associations were also observed for progesterone (p = 0.043) and inhibin B
(p < 0.001), and quartile analyses supported significant trends for E2 (p-trend = 0.027), T/E2
(p-trend = 0.015), and inhibin B (p-trend < 0.001) but did not support a significant trend for
progesterone (p-trend = 0.231). A medium confidence cross-sectional study (Lopez-Espinosa et
al., 2016) observed inverse associations for E2 and total testosterone in children 6-9 years of
age. Analyses by quartile of exposure supported this trend for decreasing testosterone. A cross-
sectional analysis in a medium confidence study (Wang et al., 2019a) from China observed a
positive association (p < 0.001) for estriol (E3) in cord blood but did not find an association for
E2.
Decreases in testosterone were seen in low confidence cross-sectional analyses (Zhou et al.,
2017c; Zhou et al., 2016) in children and adolescents (10-15 years of age) from the GBCA in
Taiwan. In boys, testosterone was observed to have a significant inverse association, and a
decreasing trend. No effects on E2 in boys were observed. A follow-up study (Zhou et al.,
2017c) observed significant decreases in testosterone among children with asthma but not in
children without asthma. Sex-stratified analyses for reproductive hormones were not conducted
in this study.
A cross-sectional study (Di Nisio et al., 2019) in Italian high school students examined
associations between PFOS levels and possible risk factors for diseases of the male reproductive
system and observed significantly higher serum PFOS levels and testosterone (p < 0.001) in
exposed individuals compared with unexposed controls.
Pubertal development and semen parameters were examined in two studies (Di Nisio et al., 2019;
Ernst et al., 2019) and effects were seen in one (Appendix D). One medium confidence study
(Ernst et al., 2019) observed no associations between prenatal PFOS exposure from first-
trimester maternal serum samples and pubertal stages (i.e., Tanner stages) and pubertal
landmarks (e.g., acne, voice break, or first ejaculation). Comparisons of semen analysis in Italian
high school students (Di Nisio et al., 2019) observed a reduced number of sperm with normal
morphology (p < 0.001) and a slight increase in semen pH (p = 0.005).
Anthropometric measurements of male reproductive organs were examined in four studies
(Arbuckle et al., 2020; Di Nisio et al., 2019; Tian et al., 2019b; Lind et al., 2017a) and three
observed effects (Appendix D). A high confidence Danish study (Lind et al., 2017a) in children
C-4
-------
APRIL 2024
from the Odense cohort observed a significant positive association with anoscrotal distance
(AGDas) in the highest prenatal PFOS exposure group. Positive non-significant associations
were observed for anopenile distance (ADGap). Children from the Shanghai-Minhang Birth
Cohort Study (Tian et al., 2019b) were evaluated at birth, six months, 12 months of age for
changes in anogenital distance (AGD). At birth, significant decreases in AGDas (p = 0.043) were
observed in continuous analyses, and in the highest quartile of exposure. Results were similar at
six months of age. In contrast, associations were positive and largely not significant at 12 months
of age. However, a significant increase in ADGap was observed among boys in the third quartile
of exposure at 12 months. Results from a medium confidence study (Arbuckle et al., 2020) in
children from the Maternal-Infant Research on Environmental Chemicals (MIREC) cohort were
inconsistent regarding the relationship between prenatal PFOS exposure and AGD. Di Nisio et
al. (2019) reported smaller AGD in exposed compared with unexposed adolescents (p = 0.019).
Significant differences (p < 0.001) were also observed for penile and testicular measurements
among adolescents, including smaller testicular volume, shorter penis length, and smaller penis
circumference. A smaller borderline significant pubis-to-floor distance was also observed
(p = 0.064).
C.1.1.1.4 Findings From the General Adult Population
Serum sex hormones were examined in four studies (Cui et al., 2020; Petersen et al., 2018; Lewis
et al., 2015; Tsai et al., 2015) and two observed effects (Appendix D). A medium confidence
study (Cui et al., 2020) evaluated serum hormone concentrations in men with fecundity issues
and men from couples with female factor infertility. Serum and semen PFOS were significantly
correlated (Spearman's r = 0.793, p < 0.01). Total testosterone and sex hormone binding globulin
(SHBG) were inversely associated (p < 0.05) with serum and semen PFOS. The total
testosterone-LH ratio was negatively associated (p < 0.05) with semen PFOS, and borderline
significant with serum PFOS (p = 0.058). Results for total testosterone remained among those
30 years old or younger after stratifying by age but were no longer observed in men over
30 years of age. The pattern was similar for SHBG, but the association with serum PFOS did not
reach significance (p = 0.069). Analyses by quartile showed agreement with the continuous
regression analyses, indicating significant trends for total testosterone and SHBG with serum and
semen levels of PFOS. A medium confidence cross-sectional study (Petersen et al., 2018) on
Faroese men observed a significant increase (p = 0.04) in luteinizing hormone with increasing
serum PFOS levels.
Semen characteristics and genomic effects in sperm were examined in five studies (Pan et al.,
2019; Petersen et al., 2018; Song et al., 2018; Leter et al., 2014; Kvist et al., 2012) and three
observed effects (Appendix D). One medium confidence study (Kvist et al., 2012) evaluating
men from the INUENDO cohort from Greenland, Poland, or Ukraine observed a significant
positive association (p = 0.026) with the Y:X-chromosome ratio in sperm when pooling data
across countries. This association was also observed in trend analyses for the Greenland subset
of the cohort but not in other country-specific analyses. Chromosomal changes were further
characterized in another INUENDO study (Leter et al., 2014) using a sperm DNA global
methylation assay. Methylation of the Sata repeats, a non-transposonic repetitive satellite DNA
sequence generally found in or adjacent to every centromere, was significantly increased
(p < 0.05) in men from Ukraine, but no effect was observed in other INUENDO communities or
in the pooled analysis. Another method of analysis of sperm DNA methylation utilized flow-
cytometry to measure cell-by-cell methylated cytosines (% 5-mCs) by immunodetection. A
C-5
-------
APRIL 2024
significant inverse relationship was observed among Polish men but was not seen in other
populations or the entire cohort. These results indicate hyper- and hypomethylated states,
respectively. Differences in results may be related to differences in each method's approach.
A medium confidence cross-sectional study (Pan et al., 2019) on a sample of men from Nanjing,
China, described above (Cui et al., 2020), investigated the effects of PFOS on semen
characteristics. Two separate analyses were conducted, each using either serum or semen as the
biomonitoring matrix for PFOS exposure determination. In linear regression analyses using
semen PFOS exposure levels, significant positive associations (p < 0.05) were observed for the
sperm DNA fragmentation index (DFI)—a measure of the percentage of sperm with damaged
DNA. Significant inverse associations were observed for progressive motility, and sperm
straight-line velocity, suggesting an overall deleterious effect on sperm motility. No significant
associations were observed in analyses using serum PFOS levels.
C.l.1.2 Female
C. 1.1.2.1 Introduction
Reproductive health outcomes of interest in females vary with biological maturity over the life
course and by pregnancy status. Of interest across the life stages, reproductive hormone levels,
such as prolactin, follicle stimulating hormone (FSH), LH, testosterone, and E2, are commonly
examined as indicators of reproductive health. Additional reproductive health outcomes of
interest include timing of pubertal milestones among children and adolescents; fertility
indicators, impacts to menstruation, and occurrence of menopause among non-pregnant adult
females; and preeclampsia, gestational hypertension, pregnancy loss, and breastfeeding duration
among pregnant females.
The 2016 Health Assessment and Health Effects Support Document for PFOS (U. S. EPA, 2016c)
concluded that there was suggestive evidence of an association with risk of gestational
hypertension or preeclampsia (Zhang et al., 2015a; Darrow et al., 2013; Stein et al., 2009). There
was generally consistent evidence of associations between serum PFOS and reduced female
fertility and fecundity (Bach et al., 2015; Velez et al., 2015; torgensen et al., 2014; Fei et al.,
2009). There were concerns over the possibility of reverse causality explaining observed
associations between PFOS exposure and various female reproductive outcomes due to
menstruation being a route of PFOS excretion (Whitworth et al., 2012b).
There are 48 studies (50 publications) that have investigated relationships between PFOS
exposure and female reproductive outcomes since the 2016 document (U.S. EPA, 2016c).
Among the 50 publications available for review, there were 20 cohort studies, 17 cross-sectional
studies, and 13 case-control studies. 19 studies were conducted in adults, 6 were conducted in
children and adolescents, 13 were conducted in both adults and children, and 12 were conducted
in pregnant women. Most studies used blood PFOS measures to assess exposure while others
used amniotic fluid and follicular fluid.
C.l.1.2.2 Study Quality
There are 48 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and female reproductive effects. Study quality evaluations for these 48 studies are
shown in Figure C-2 and Figure C-3.
C-6
-------
APRIL 2024
Among the 48 publications available for review, 5 were classified as high confidence, 24 as
medium confidence, 16 as low confidence, and three were considered uninformative. Because
menstruation is a primary route of PFOS excretion for people who menstruate, reverse causality
was a specific concern for cross-sectional studies that measured blood PFOS and certain
reproductive hormones with known menstrual fluctuations without reporting sample collection
timing (Heffernan et al., 2018; Zhang et al., 2018b). Several low confidence studies lacked an
appropriate strategy for identifying potential confounders (Mccoy et al., 2017; Zhou et al.,
2017a) or failed to adjust for key confounders, such as age and SES (Heffernan et al., 2018;
Zhou et al., 2016). Low confidence studies had deficiencies in participant selection (Bach et al.,
2018; Heffernan et al., 2018; Zhang et al., 2018b), exposure measurement methods (Campbell et
al., 2016), reliance on self-reporting for exposure, outcome, or covariate information (Campbell
et al., 2016), and small sample size (Heffernan et al., 2018; Mccoy et al., 2017). Maekawa, 2017,
4238291 was considered uninformative due to lack of information on participant selection, lack
of adjustment in analyses for key confounders. Lee, 2013, 3859850 was also considered
uninformative due to lack of consideration of key confounders in analyses. Arbuckle, 2013,
2152344 was considered uninformative because PFOS was evaluated as the outcome and
reproductive measures were considered as predictors.
C-7
-------
APRIL 2024
Arbuckle etal., 2013, 2152344-
Bach etal., 2015, 3981559-
Bach etal., 2018, 5080557
Bangma et al., 2020, 6833725 -
Borghese et al., 2020, 6833656 -
Campbell et al.,2016, 3860110-
Caserta et al., 2013, 2000966 -
Caserta etal , 2013, 2001177-
Crawford et al., 2017, 3859813 -
Ding etal. 2020, 6833612
Donley et al., 2019, 5381537 -
Ernst etal., 2019, 5080529-
Goudarzi et al., 2017, 3981462 -
Heffeman etal., 2018, 5079713 -
Huang et al., 2019, 5083564 -
Huo etal., 2020, 6505752
Itoh etal., 2016, 3981465
Jensen etal , 2020, 6311643-
Kim et al., 2020, 6833596
Lee etal., 2013, 3859850
Lewis etal., 2015, 3749030
Liewet al., 2020, 6387285-
Liu etal., 2020, 6569227-
Lopez-Espinosa et al., 2016, 3859832 -
Legend
I Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
* Multiple judgments exist
Figure C-2. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Female Reproductive Effects
Interactive figure and additional study details available on HAWC.
C-8
-------
APRIL 2024
~y>v>v>v
Louis etal., 2012, 1597490-
Lyngso et al.,
Legend
p
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
D
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-3. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Female Reproductive Effects (Continued)
Interactive figure and additional study details available on IiAWC.
C-9
-------
APRIL 2024
C.l.1.23 Findings From Children and Adolescents
Two high confidence, eight medium confidence, and three low confidence studies assessed
relationships between PFOS exposure and female reproductive outcomes in children and
adolescents (Appendix D). Studies in infants primarily focused on reproductive hormone levels,
while studies in adolescents focused on reproductive hormone levels as well as pubertal
milestones.
Two high confidence (Jensen et al., 2020b; Yao et al., 2019) and four medium confidence studies
(Liu et al., 2020b; Wang et al., 2019a; Goudarzi et al., 2017a; Itoh et al., 2016) examined the
effects of PFOS exposure on reproductive hormone levels in female infants, reporting mixed
results. Itoh, 2016, 3981465, a study of the Hokkaido birth cohort, observed a significant
negative association between maternal serum PFOS and progesterone in cord blood (regression
coefficient per unit change in PFOS (loglO-ng/mL) = -0.6; 95% CI: -0.9, -0.2) as well as
prolactin in cord blood (regression coefficient per unit change in PFOS (loglO-ng/mL) = -0.5;
95% CI: -0.8, -0.2). A significant positive association was observed between cord blood PFOS
and E3 (regression coefficient per unit increase in cord blood PFOS (loglO-ng/mL) = 0.5; 95%
CI: 0.3, 0.7) in another medium confidence study (Wang et al., 2019a). The two high confidence
studies and four medium confidence studies found no significant associations between maternal
serum or cord blood PFOS and reproductive hormones such as testosterone, the testosterone-to-
estradiol ratio (Yao et al., 2019); E2, testosterone, SHBG, the testosterone-to-SHBG ratio (Itoh et
al., 2016); 17-OHP, androstenedione, FSH, LH, DHEA, dehydroepiandrosterone sulfate
(DHEAS) (Jensen et al., 2020b); 17-OHP, progesterone (Liu et al., 2020b); androstenedione,
DHEA (Goudarzi et al., 2017a); P-E2, and estrone (Wang et al., 2019a).
Three medium confidence (Lopez-Espinosa et al., 2016; Maisonet et al., 2015a; Tsai et al., 2015)
and three low confidence (Zhou et al., 2017c; Zhou et al., 2016; Lewis et al., 2015) studies
assessed the relationship between PFOS and reproductive hormone levels in adolescent females.
As part of the C8 Health Project, Lopez-Espinosa, 2016, 3859832 observed negative associations
for total testosterone across serum PFOS quartiles and per unit increase in serum PFOS among
females 6-9 years old with high exposure (percent difference for quartile 2 vs. quartile 1 = -1.1;
95% CI: -8.6, 7.1; percent difference for quartile 3 vs. quartile 1: —7.8%; 95% CI: -15, -0.1;
percent difference for quartile 4 vs. quartile 1: -11.1%; 95% CL-18.2, -3.5; percent difference
per unit increase in serum PFOS (ln-ng/mL) = -6.6%; 95% CI: -10.1, -2.8). Maisonet, 2015,
3859841 found significantly increased serum testosterone among 15-year-old females in the
highest tertile of maternal serum PFOS during pregnancy (beta: 0.18, 95% CI: 0.01, 0.35). No
significant associations were observed for E2 (Zhou et al., 2017c; Lopez-Espinosa et al., 2016;
Zhou et al., 2016), testosterone (Zhou et al., 2016; Lewis et al., 2015), SHBG (Maisonet et al.,
2015a; Tsai et al., 2015), or FSH (Tsai et al., 2015).
One medium confidence study drew data from the Danish National Birth Cohort (DNBC) to
examine the effects of prenatal PFOS exposure on pubertal milestones in female adolescents,
such as breast development (age at attainment of Tanner stages 2-5), pubic hair development
(age at attainment of Tanner stages 2-5), axillary hair development, and age at menarche in
adolescent girls (Ernst et al., 2019). Average age at attainment for all pubertal indicators was
significantly reduced across PFOS tertiles), while no other significant associations were observed
for breast development, age at menarche, axillary hair development, or pubic hair development.
C-10
-------
APRIL 2024
C.l.1.2.4 Findings From Pregnant Women
One high confidence, five medium confidence studies, and one low confidence study examined
the relationship between PFOS exposure and preeclampsia (Appendix D). One medium
confidence study (Wikstrom et al., 2019) reported significant positive associations between
serum PFOS and odds of preeclampsia in both continuous and quartile analyses (OR = 1.53;
95% CI: 1.07, 2.2; OR for PFOS highest vs. lowest quartile = 2.68; 95% CI: 1.17, 6.12). The
remaining five studies reported mixed non-significant associations (Borghese et al., 2020; Huo et
al., 2020; Rylander et al., 2020; Huang et al., 2019b; Starling et al., 2014a). Huo, 2020, 6505752,
a high confidence cohort study of 3,220 pregnant women study observed a non-significant
reduction in odds of preeclampsia for women above the 80th percentile for plasma PFOS
compared with women in or below the 80th percentile and observed a non-significant increase in
odds of preeclampsia. In two medium confidence cohort studies, non-significant positive
associations were observed (Borghese et al., 2020; Starling et al., 2014a). Non-significant
negative associations were observed in medium confidence case-control (Rylander et al., 2020)
and cross-sectional (Huang et al., 2019b) studies. A low confidence study found no association
between median PFOS levels and hypertensive disorders of pregnancy (Bangma et al., 2020).
One high confidence and two medium confidence studies examined the relationship between
PFOS exposure and gestational hypertension reporting non-significant mixed associations for
gestational hypertension and significant positive associations for blood pressure. Huo, 2020,
6505752, a high confidence cohort study of 3,220 pregnant women, observed a non-significant
negative association between plasma PFOS and odds of gestational hypertension. Borghese,
2020, 6833656, a medium confidence prospective cohort study, followed 1,708 women from
early pregnancy to delivery for gestational hypertension, preeclampsia, and changes in blood
pressure, measuring plasma PFOS once per trimester and again at delivery. Borghese, 2020,
6833656 observed a non-significant positive association between plasma PFOS and odds of
gestational hypertension. A significant positive association was reported for systolic blood
pressure (SBP) mmHg) per log2-(^g/L increase PFOS at delivery (beta: 1.19, 95% CI: 0.28, 2.1).
Significant positive associations were also observed in each trimester for diastolic blood pressure
(DBP) (mmHg) (beta for trimester 3: 0.66, 95 % CI 0.18, 1.14) but not at delivery. No
association between plasma PFOS levels and gestational hypertension was observed by Huang,
2019, 5083564.
Two medium confidence studies (Liew et al., 2020; Buck Louis et al., 2016) and one low
confidence study (Jensen et al., 2015) investigated the effect of PFOS exposure on pregnancy
loss and reported non-significant mixed results. In a cohort study of 501 couples, Louis, 2016,
3858527 reported a non-significant, negative association between serum PFOS levels and
pregnancy loss during the first seven weeks of pregnancy. A case-control study nested within the
DNBC comparing 222 pregnancies ending in miscarriage to 218 pregnancies resulting in live
births observed non-significant positive associations across maternal plasma PFOS levels for
odds of miscarriage in both continuous and quartile analyses. Jensen, 2015, 2850253 also
reported non-significant positive associations for odds of miscarriage in both continuous and
tertile analysis.
Two medium confidence studies assessed the relationship between serum PFOS levels in
pregnancy and breastfeeding duration, with both reporting significant, inverse associations
between the two (Timmermann et al., 2017b; Romano et al., 2016). Using data from two Faroese
C-ll
-------
APRIL 2024
birth cohorts (n = 1,130), Timmermann, 2017, 3981439 observed a significant reduction in total
breastfeeding duration per doubling of maternal serum PFOS (regression coefficient per
doubling of serum PFOS (ng/mL) = -1.4; 95% CI: -2.1, -0.6) and a non-significant reduction in
exclusive breastfeeding duration per doubling of maternal serum PFOS (regression coefficient
per doubling of serum PFOS (ng/mL) = -0.3; 95% CI: -0.6, 0.1). These observations were
supported by a prospective birth cohort study of 336 women investigating the relationship
between serum PFOS levels during pregnancy and relative risk of breastfeeding termination at
three and six months postpartum (Romano et al., 2016). This study observed a positive trend for
relative risk of breastfeeding termination across maternal serum PFOS quartiles for both time
points. Relative risk for stopping breastfeeding by 3 months increased in maternal serum PFOS
quartiles 2, 3, and 4 compared with quartile 1, with a significant increase observed for quartile 3
(relative risk for PFOS quartile 2 vs. 1 = 1.32; 95% CI: 0.97, 1.79; relative risk for PFOS quartile
3 vs. quartile 1 = 1.39; 95% CI: 1.04, 1.88; relative risk for PFOS quartile 4 vs. quartile 1 = 1.08;
95% CI: 0.79, 1.46). Relative risk for stopping breastfeeding by 6 months was non-significantly
increased in maternal serum PFOS quartiles 2, 3, and 4 compared with quartile 1 as well.
One high confidence study and one medium confidence study examined relationships between
PFOS exposure and female reproductive hormone levels in pregnant women. In a medium
confidence case-control study of 545 mother-infant pairs, Toft, 2016, 3102984 observed a
significant, positive association between PFOS in amniotic fluid and 17-OHP, with a significant
percent difference in the continuous analysis and a significant increase for tertile 3 compared
with tertile 1 (percent difference in median 17-OHP level per unit increase in amniotic fluid
PFOS (ln-ng/mL) = 0.15; 95% CI: 0.11, 0.2; percent difference in median 17-OHP for women in
amniotic fluid PFOS tertile 3 vs. tertile 1 = 18%; 95% CI: 11, 26). A significant, positive
association was also observed between amniotic fluid PFOS and androstenedione in the
continuous analysis and for tertile 3 compared with tertile 1 (percent difference in median
androstenedione level per unit increase in amniotic fluid PFOS (ln-ng/mL) = 0.15; 95% CI: 0.1,
0.21; percent difference in median androstenedione for women in amniotic fluid PFOS tertile 3
vs. tertile 1 = 17; 95% CI: 8, 25). Significant, positive associations across tertiles of PFOS were
observed for progesterone (percent difference per 1% increase in PFOS (ln-ng/mL) = 0.21; 95%
CI: 0.14, 0.29; percent difference for PFOS tertile 2 vs. 1 = 11%; 95% CI: 0, 23; percent
difference for PFOS tertile 3 vs. 1 = 22; 95% CI: 11, 34) and testosterone (percent difference per
1% increase in PFOS (ln-ng/mL) = 0.16; 95% CI: 0.09, 0.23; percent difference for PFOS tertile
2 vs. tertile 1 = 9%; 95% CI: -2, 20; percent difference for PFOS tertile 3 vs. tertile 1 = 18%;
95%) CI: 7, 29), but no association was observed for DHEAS. In a high confidence study, Mitro,
2020, 6833625, no significant association was observed between plasma PFOS during pregnancy
and SHBG levels 3 years postpartum.
One medium confidence study (Lyngs0 et al., 2014) examined the effects of serum PFOS levels
on pre-pregnancy menstruation. While evidence of increased odds of menstrual cycle irregularity
was reported, the association was not significant.
C.1.1.2.5 Findings From the General Adult Population
Five medium confidence (Kim et al., 2020b; Donley et al., 2019; Crawford et al., 2017; Lum et
al., 2017; Wang et al., 2017), three low confidence studies (Bach et al., 2018; Zhang et al.,
2018b; Mccoy et al., 2017) and one uninformative study (Arbuckle et al., 2013) examined
implications of PFOS exposure on female fertility, reporting mixed results (Appendix D).
C-12
-------
APRIL 2024
Significant positive associations were reported in low confidence studies, including for odds of
premature ovarian insufficiency (POI) across plasma PFOS quartiles (Zhang et al., 2018b) and
for the fecundity ratio for parous women in plasma PFOS quartiles (Bach et al., 2018). Non-
significant positive associations were observed for day-specific probability of pregnancy (Lum et
al., 2017) and cycle and day-specific time to pregnancy (Crawford et al., 2017). Associations
with indicators of ovarian function were largely non-significant, including no association
observed between serum PFOS and anti-Mullerian hormone (AMH) (Crawford et al., 2017).
Associations between maternal serum PFOS during pregnancy and female adolescent AMH
levels were also not observed (Donley et al., 2019). No significant associations were reported for
infertility measures including endometriosis-related infertility (Wang et al., 2017), and
fertilization rate (Kim et al., 2020b). Additionally, McCoy, 2017, 3858475 reported non-
significant negative correlations between PFOS in follicular fluid and blast conversion rate,
fertilization rate, and follicle count. No associations were observed for other outcomes related to
menstrual cycles and gynecologic pathologies, including menstrual cycle length (Lum et al.,
2017), endometriosis, polycystic ovary syndrome (PCOS), genital tract infections, and idiopathic
infertility (Kim et al., 2020b).
One high confidence study examined the relationship between PFOS exposure and age at natural
menopause: the Study of Women's Health Across the Nation (SWAN), a prospective cohort of
1,120 premenopausal women aged 45-56 (Ding et al., 2020). Significant, positive associations
were reported between serum Sm-PFOS and risk of natural menopause for women in Sm-PFOS
tertile 3 versus tertile 1 (HR = 1.27; 95% CI: 1.01, 1.59) and between serum n-PFOS and risk of
natural menopause for women in n-PFOS tertile 3 versus tertile 1 (HR = 1.26; 95% CI: 1.02,
1.57). Non-significant positive associations were observed for both Sm-PFOS and n-PFOS when
analyzed as a continuous variable and for women in tertile 2 versus tertile 1.
One medium confidence (Tsai et al., 2015) and five low confidence studies (Heffernan et al.,
2018; Zhang et al., 2018b; Mccoy et al., 2017; Lewis et al., 2015; Petro et al., 2014) reported
associations between PFOS and female reproductive hormone levels in non-pregnant adult
women. Three low confidence studies reported significant mixed effects. In women with and
without PCOS, Heffernan, 2018, 5079713 observed significant negative associations with FAI
only in controls. McCoy, 2017, 3858475 observed a negative correlation with plasma E2. In
women with and without POI, Zhang, 2018, 5079665 observed significant negative associations
for E2 in both cases and controls and positive associations for FSH and prolactin in cases only.
No significant associations were observed for testosterone (Lewis et al., 2015); mean FSH and
SHBG in young women (ages 12-30 years) (Tsai et al., 2015); testosterone, E2, and SHBG
(Heffernan et al., 2018); E2 (Petro et al., 2014); or for LH and testosterone (Zhang et al., 2018b).
C.1.2 Animal Evidence Study Quality Evaluation and Synthesis
There are 6 studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and 16 studies from recent
systematic literature search and review efforts conducted after publication of the 2016 PFOS
HESD that investigated the association between PFOS and reproductive effects. Study quality
evaluations for these 22 studies are shown in Figure C-4.
C-13
-------
APRIL 2024
Alam et al., 2021, 9959508-
Argus, 2000, 5080012-
Butenhoff et al., 2009, 757873 -
Butenhoffetal., 2012, 1276144-
Chen et al., 2018, 5080460 -
Lai etal., 2017, 3981773-
Lau et al., 2003, 757854-|
Lee et al., 2015, 2851075-
Lopez-Doval et al., 2015, 2848266-
Luebker et al., 2005, 1276160 -
Luebker et al., 2005, 757857 -
NTP, 2019, 5400978-
Qiu etal., 2013, 2850956
Qiu etal., 2016, 3981408
Qiu et al., 2020, 7276729
Qu et al., 2016, 3981454
Salgado et al., 2015, 3981583
Seacat et al., 2002, 757853
Thomford, 2002, 5432419
Zhang et al., 2019, 5918673
Zhang et al., 2020, 6315674
Zhong et al., 2016, 3748828
++ ++
++ ++
NR
++ ++
¦
+
NR
++ ++ ++
+
+
NR
-
+
+
-
NR
NR
++
++
+
+
+
+
NR
4
+
+
+
NR
+
+
-
"
+
NR
+
+
+
+
-
NR
++ ++
-
-
+
NR
+
4
-
-
+
NR
+
-
-
-
NR
NR
+
+
+
Legend
| Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
NR| Not reported
* Multiple judgments exist
Figure C-4. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Reproductive Effects
Interactive figure and additional study details available on IiAWC.
C-14
-------
APRIL 2024
Short-term, subchronic, chronic, and reproductive/developmental animal studies suggest that oral
exposure to PFOS can adversely affect the male and female reproductive systems. However, it is
not often clear whether the observed alterations reflect specific toxicity to the reproductive
system or if they result from concurrent systemic toxicity (i.e., reductions in body weight).
Effects observed in male rodents included alterations to hormone levels (prolactin, luteinizing
hormone, FSH, E2, and testosterone), as well as decreased testis weights, and decreased sperm
count. In female mice exposed to PFOS, effects on prolactin-family hormones were observed.
Although effects were predominately seen in rodent species there were inconsistencies among
rats and mice. In cynomolgus monkeys no effects were noted in reproductive organ weights and
histopathology, although a decrease in male E2 levels was observed (Seacat et al., 2002).
C. 1.2.1 Male and Female Fertility Parameters and Pregnancy Outcomes
Male and female fertility parameters and pregnancy outcomes were evaluated in rodent and
rabbit species. Mating and fertility parameters, such as number of pregnancies per number of rats
that mated, number of days to inseminate, and number of matings during the first week of
cohabitation were unaffected by PFOS doses as high as 3.2 mg/kg/day in a two-generation
reproduction study in rats(Butenhoff et al., 2009; Luebker et al., 2005a). Gestation and fertility
indices were unaffected in one- and two-generation rat reproduction studies (Luebker et al.,
2005b; Luebker et al., 2005a); however, gestation length was significantly decreased in a dose-
dependent manner in dams exposed to > 0.8 mg/kg/day in the one-generation study (Luebker et
al., 2005b) and in Po dams exposed to 3.2 mg/kg/day in the two-generation study(Luebker et al.,
2005a) (Figure C-5). Decreases in maternal bodyweight change were noted in both studies
(Luebker et al., 2005b; Luebker et al., 2005a)(see Toxicity Assessment, (U.S. EPA, 2024)). In
contrast, Butenhoff et al. (2009) reported no significant differences in gestation length for rats
treated with up to 1 mg/kg/day PFOS from GD 0 to PND 20. That study also found no
significant differences in the number of litters delivered or live litter size at birth (Butenhoff et
al., 2009).
Lix)point Sludv Name Sludv Desis-n
Figure C-5. Gestation Length in Rats Following Exposure to PFOS
LD = lactation day; GD = gestation day; Po = parental generation.
Interactive figure and additional study details available on HAWC.
In mice, reproductive outcomes were examined in pregnant CD1 mice treated at 1.5, 3, and
6 mg/kg/day from GD 6-GD 18. Body weight and body weight change were significantly
reduced in dams given PFOS at 6 mg/kg/day in comparison to the controls (Fuentes et al., 2006).
The number of live and dead fetuses per litter and number of implantation sites were not
statistically significant even though high fetal mortality was observed in dams exposed to PFOS
at 6 mg/kg. Lastly, there was no observed effect on gravid uterine weight in pregnant CD1 mice
on GD 18.
Observation r>mt" Animal Ucxcnpnon
(n>i.»i t I LJ(V. l.'-Wt Hi 1
t»i: "iiu ,i.p Ji^ h, \19 n :'
f'MM rn s: i >h iKsii ?\: ;5>
PH )S Kr|w
-------
APRIL 2024
In a single study in New Zealand white rabbits, dams were administered 0 mg/kg/day,
0.1 mg/kg/day, 1.0 mg/kg/day, 2.5 mg/kg/day, or 3.75 mg/kg/day PFOS via intubation from GD
7 to GD 20 (Argus Research Laboratories, 2000). The number of rabbits pregnant at the time of
sacrifice (GD 29) decreased with increasing dose due to an increased incidence of abortion with
higher PFOS doses (see Toxicity Assessment, (U.S. EPA, 2024)). Only 12/21 (57%) of dams
that became pregnant in the study from the 3.75 mg/kg/day dose group were pregnant on GD 29
compared with 100% pregnancy maintained in the 0 mg/kg/day, 0.1 mg/kg/day, and
1.0 mg/kg/day groups and 94% pregnancy maintained in the 2.5 mg/kg/day group. Each
individual doe that aborted exhibited weight loss and severely reduced feed consumption.
Overall, maternal body weight gains were significantly reduced in the 1.0 mg/kg/day,
2.5 mg/kg/day, and 3.75 mg/kg/day groups (Argus Research Laboratories, 2000).
C. 1.2.2 Male Sperm Parameters
Sperm parameters were evaluated in studies of male rats and mice, with conflicting results
(Figure C-6). In a 28-day study conducted by NTP in which Sprague-Dawley rats, exposed to
PFOS for 28 days had no effect on spermatid headcount in the testis, sperm count in the
epididymis and cauda epididymis, or epididymal sperm motility in animals treated with
1.25 mg/kg/day to 5.0 mg/kg/day (NTP, 2019). In contrast, a general reduction in epidydimal
sperm count was observed in mice among studies of varying durations including two 4-week
studies in ICR mice exposed to 2.5 mg/kg/day or 5 mg/kg/day, a 4-week study in ICR mice
exposed to 5 mg/kg/day and 10 mg/kg, a 5-week study in C57 mice exposed to 10 mg/kg/day,
and CD-I pups on PND 63 exposed to 3 mg/kg/day during gestation (Qiu et al., 2020; Lai et al.,
2017a; Qiu et al., 2016; Qu et al., 2016; Qiu et al., 2013). Qiu et al. (2016) did not observe
alterations in epididymis weight that may have influenced epididymal sperm counts.
PFOS Reproductive Effects - Sperm Parameters
Endpoint
Epididymis Sperm Count
Study Name Study Design
Lai etal., 2017, 3981773 developmental (GD1-17) PND63
Observation Time
Animal Description
F1 Mouse, CD-1 (
-------
APRIL 2024
increased serum testosterone and progesterone levels in comparison to the controls at 0.015 and
0.15 mg/kg via oral gavage for 60 days in Wistar rats. However, in a 28-day study conducted by
NTP (2019), no effects on testosterone levels were noted in male rats treated with up to
5 mg/kg/day. A 46% decrease relative to controls was also noted in mice treated with
10 mg/kg/day for five weeks (Qu et al., 2016). Developmental studies in mice showed a 31%
decrease in testosterone at PND 63 in CD-I mice exposed to 3 mg/kg/day throughout gestation
(Lai et al., 2017a). C57BL/6 mouse pups treated with 1 and 5 mg/kg/day showed 35% and 52%
decreases, respectively, at postnatal week 4 (PNW 4) after maternal oral exposure from GD 1 to
GD 17 (significantly different in the 5 mg/kg/day group) (Zhong et al., 2016). In the same study,
38% and 34% decreases were observed in the 1 and 5 mg/kg/day groups, respectively, at PNW 8,
though only the response in the 1 mg/kg/day group was statistically different from controls.
Similarly, Qiu et al. (2020) observed a significant decrease in serum testosterone levels at 5, and
10 mg/kg/day in comparison to the controls for four weeks in ICR mice. Cynomolgus monkeys
treated up to 0.75 mg/kg/day for 182 days showed no statistically significant effects on
testosterone levels (Seacat et al., 2002).
PFOS Reproductive Effects - Testosterone Levels
Endpoint Study Name Study Design Observation Time Animal Description Dose (mg/kg/day) | O Statistically significant £ Not statistically significant!—| 95% CI |
Testosterone Seacat et al., 2002,757853 chronic (26wk) 182d Monkey, Cynomolgus (,f , IM=4-6> 0
0.03
0.15
0.75
I
>
—1 1
I
II •'
1
1
—1 1
Zhong etal., 2016, 3748828 developmental (GD 1-17) PNW4 F1 Mouse. C57BL/6 (
-------
APRIL 2024
GD = gestation day; PND = postnatal day; PNW = postnatal week; Fi = first generation
Changes in E2 levels in males were noted in rats, mice, and cynomolgus monkeys across studies
of varying durations (Figure C-8); however, the direction of the change was not consistent across
the studies. In two studies from the same laboratory, following a 28-day exposure, Salgado et al.
(2015) and Lopez-Doval et al. (2015) noted decreases in E2 ranging from 13% to 19% in rats
treated with 3.0 and 6.0 mg/kg/day and >1.0 mg/kg/day, respectively. Decreases were similar
across dose groups and were not dose dependent. In mice, subchronic exposure to PFOS
(35 days) at doses of 0.5 mg/kg/day and 10 mg/kg/day showed no statistically significant effect
on E2 levels, but there was a general increasing trend with increasing dose (5% and 10%
increase, respectively) (Qu et al., 2016). Male mouse pups exposed to 5.0 mg/kg/day from GD 1
to GD 17 exhibited a 42% increase in serum E2 levels at PNW 4 (Zhong et al., 2016). By PNW 8
the increase was no longer statistically significant but remained 28% higher than the control
group (Zhong et al., 2016). There was an apparent dose-dependent increase in serum E2 at both
PNW 4 and PNW 8. Conversely, no significant change or trend in serum E2 levels was observed
in adult ICR male mice exposed to 0 mg/kg/day, 0.5 mg/kg/day, 5 mg/kg/day, and 10 mg/kg/day
for four weeks (Qiu et al., 2020). Seacat et al. (2002) observed a 97% decrease in serum E2 in
male cynomolgus monkeys treated at 0.75 mg/kg/day for 182 days (Seacat et al., 2002).
PFOS Reproductive Effects - Male Estradiol Levels
Study Design Observation Time Animal Description Dose (mg/kg/day)
Estradiol Seacat et al., 2002. 757853 chronic (26wk)
Zhong et al„ 2016.3748828 developmental (GD1-17) PNW4
Qiuetal.. 2020, 7276729 short-term (4wk)
Quetal.,2016. 3981454
subchronic (35d) 35d
etal., 2015,3981583 short-term (28d) 28d
Lopez-Doval et al.. 2015, 2848266 short-term (28d)
Monkey. Cynomolgus (:?, N=4-6) 0
0.03
0.15
0.75
F1 Mouse. C57BL/6 (<•;', N=12) 0
0.1
F1 Mouse. C57BL/6 (<*, IM=12) 0
Mouse. ICR N=10)
Mouse. C57 N=12)
Rat, Sprague-Dawley (o. N=7) 0
Rat, Sprague-Dawley {3, N=15) 0
O Statistically significant £ Not statistically significant |—| 95% CI
-120 -100 -8(
-40 -20 0 20 40 60 80 100 120
Percent control response (%)
Figure C-8. Percent Change in Estradiol Levels Relative to Controls in Male Rodent and
Non-Human Primates Following Exposure to PFOS
Interactive figure and additional study details available on HAWC.
GD = gestation day; PND = postnatal day; PNW = postnatal week; Fi = first generation
Short-term exposure studies examining the effect of PFOS exposure on LH, FSH, and prolactin
levels in male rats were available (Figure C-9). Groups treated for 28 days with
C-18
-------
APRIL 2024
doses > 0.5 mg/kg/day as well as 3.0 mg/kg/day and 6.0 mg/kg/day exhibited decreases in LH
(15%—30%) and prolactin (54%-78%), respectively (Lopez-Doval et al., 2015; Salgado et al.,
2015; Lopez-Doval et al., 2014). Additionally, increases ranging from 88% to 133% in serum
FSH levels were observed in all treated groups (0.5 mg/kg/day-6 mg/kg/day) when compared
with controls (Lopez-Doval et al., 2014). However, in a study by Qiu et al. (2020), PFOS
exposure did not significantly alter serum FSH and LH levels.
PFOS Reproductive Effects - Male LH and Prolactin Levels
Study Design Observation Time Animal Description Dose (mg/kg/day)
Follicle Stimulating Hormone (FSH) Qiu et al., 2020. 7276729 short-term (4wk) 4wk
Mouse, ICR (._¦"', N=10)
Leuteinizing Hormone (LH)
Qiu et al.. 2020, 7276729
short-term (4wk) 4wk
Mouse, ICR (J'. N=10) 0
0.5
5
10
Lopez-Doval et al., 2015.2848266
short-term (28d) 29d
Rat. Sprague-Dawley (. \ N=15) 0
0.5
3
6
Prolactin (PRL)
Salgado et al., 2015, 3981583
short-term (28d) 28d
Rat, Spragua-Dawley (;';', N=7) 0
3
6
9 Statistically significant 0 Not statistically significant |—• 95% CI I
i
-120 -100 -60 -60 -40 -20 0 20 40 60 80 100
Parcsnt control response (%)
Figure C-9. Percent Change in LH and Prolactin Levels Relative to Controls in Male Rats
Following Exposure to PFOS
Interactive figure and additional study details available on HAWC.
The red dashed line indicates a 100% decrease from the control response.
C. 1.2.3.2 Females
Evidence that oral exposure to PFOS results in changes to levels of prolactin-family hormones in
female mice was noted in an investigation by Lee et al. (2015) (Figure C-10). In this study, the
authors reported dose-dependent reductions in prolactin-family hormones, including mouse
placental lactogen (mPL-II) (46%-71%), mouse prolactin-like protein (mPLP)-Ca (20%-53%),
and mPLP-K (30%-57%), in pregnant CD-I mice exposed to 0.5 mg/kg/day, 2.0 mg/kg/day, and
8 mg/kg/day PFOS from GD 11 to GD 16. Concurrent dose-dependent decreases in bodyweight
of 2%>, 6%, and 21%, respectively, were also observed in these mice (Lee et al., 2015).
PFOS Reproductive Effects - Prolactin Family Hormone Levels in Female Mice
Endpolnt Study Name Study Design Observation Time Animal Description Dose (mg/kg/day)
Mouse placental lactogen (mPL)-ll Lee et al.. 2015. 2851075 developmental (GD11-16) GD17 P0 Mouse, CD-1 (_-,N=3) 0
Mouse prolactin-like protein (mPLP)-Ca Lee et al., 2015. 2851075 developmental (GD11-16) GD17
Mouse prolactin-like protein (mPLP>-K Lee et al., 2015, 2851075 developmental (GD11-16) GD17
PO Mouse, CD-1 , N=3) 0
P0 Mouse. CD-1 (,;, N=3) 0
# Statistically significant % Not statistically significant!—I 95% CI |
-70 -60 -50 -40 -30 -20 -10 0 10 20
Percent control response (%)
Figure C-10. Percent Change in Prolactin-Family Hormone Levels Relative to Controls in
Female Mice Following Exposure to PFOS
C-19
-------
APRIL 2024
Interactive figure and additional study details available on HAWC.
GD = gestation day; Po = parental generation.
In female cynomolgus monkeys treated with PFOS for 182 days, E2 levels decreased in a dose-
dependent manner relative to controls (decreases of 16%, 52%, and 73% in the 0.03 mg/kg/day,
0.15 mg/kg/day, and 0.75 mg/kg/day dose groups, respectively) (Seacat et al., 2002) (Figure
C-l 1). In the same study, testosterone levels were not affected in females in a dose-dependent or
statistically significant manner, though a decrease of 72% was observed in the 0.15 mg/kg/day
dose group (Seacat et al., 2002). In contrast to female monkeys, evaluation of Fi female mouse
pups treated with 0.1 mg/kg/day, 1.0 mg/kg/day, or 5.0 mg/kg/day from GD 1 to GD 17 showed
increases in E2 levels relative to the control at PNW4 (increases of 10%, 17%, and 8%,
respectively) and PNW8 (increases of 11%, 19%, and 12%, respectively), although statistical
significance was not achieved (Zhong et al., 2016). A dose-dependent decrease in testosterone
levels when compared with controls was noted at PNW4 in females (decreases of 18%, 26%, and
30% in the 0.1 mg/kg/day, 1 mg/kg/day, and 5 mg/kg/day groups, respectively), but was not
statistically significant (Zhong et al., 2016). In female rats exposed to PFOS for 28 days,
testosterone levels were significantly increased with 1.25 mg/kg/day and 2.5 mg/kg/day PFOS
(increases of approximately 37% in both groups) but not in the 5 mg/kg/day dose group (NTP,
2019).
Endpoint
Study Name
Study Design
Observation Time
Animal Description
Dose (mg/kg/day)
Estradiol
Seacat et al.. 2002, 757853
chronic (26wk)
182d
Monkey, Cynomolgus (2, N=4-6)
0
0.03
0.15
0.75
Zhong etal., 2016, 3748828
developmental (GD1-17)
PNW4
F1 Mouse. C57BL/6 (2, N=12)
0
0,1
1
5
PNW8
F1 Mouse, C57BL/6 (2. N=12)
0
0.1
1
5
Testosterone
Seacat et al., 2002, 757853
chronic (26wk)
182d
Monkey, Cynomolgus N=4-6)
0
0.03
0.15
0.75
Zhong etal., 2016, 3748828
developmental (GD1-17)
PNW4
F1 Mouse, C57BL/6 N=12)
0
0.1
1
5
PNW8
F1 Mouse, C57BL/6{ ' . N=12)
0
0.1
1
5
NTP, 2019, 5400978
short-term {28d)
29d
Rat, Sprague-Dawley (9, N=9-10)
0
0.312
0.625
1.25
2.5
5
PFOS Reproductive Effects - Female Estradiol and Testosterone Levels
-140-120-100 -80 -60 -40 -20 0 20 40 60
Percent control response (%)
00 120 140
Figure C-ll. Percent Change in Estradiol and Testosterone Levels Relative to Controls in
Female Rodents and Non-Human Primates Following Exposure to PFOS
Interactive figure and additional study details available on HAWC.
GD = gestation day; Po = parental generation; Fi = first generation; PNW = postnatal week; d = day; wk = week.
C-20
-------
APRIL 2024
The red dashed lines indicate a 100% increase or 100% decrease from the control response.
C. 1.2.4 Estrous Cyclicity and Ovarian Function (Female) and
Reproductive System Development, Including Markers of Sexual
Maturation (Female and Male)
In females, a dose-dependent increase in estrous cycle length was observed in rats treated with
0.625 mg/kg/day to 2.5 mg/kg/day over the course of 28-days (increased length of 0.4 days in the
2.5 mg/kg/day group compared with controls); however, this finding was not statistically
significant (NTP, 2019). Summary statistics indicated that the proportion of time spent in each
phase was unaffected, although Markov analysis indicated that females in all assessed groups
had an increased likelihood of transitioning to prolonged diestrus when compared with controls.
In the same study, the number of cycles was considered unaffected by treatment (NTP, 2019). In
a two-generation reproduction study in rats, no significant effects were observed on the number
of estrous cycles of Po females treated with up to 3.2 mg/kg/day for 6 weeks prior to mating
(Luebker et al., 2005a).
No significant changes in the number or distribution of corpora lutea were noted in Po rats
exposed prior to mating and during gestation in the one- and two-generation reproductive
toxicity studies (Luebker et al., 2005b; Luebker et al., 2005a). Likewise, no changes in the
number of corpora lutea were seen in Po female rabbits exposed during gestation (Argus
Research Laboratories, 2000). Reproductive and developmental studies additionally reported no
impact of gestational PFOS exposure on the timing of preputial separation or vaginal opening in
rats (Butenhoff et al., 2009; Luebker et al., 2005a; Lau et al., 2003).
C. 1.2.5 Reproductive Organ Weights and Histopathology
C.1.2.5.1 Mole
Several studies investigated the effect of PFOS exposure on male reproductive organ weights.
No effects were noted in the absolute or relative epididymal and testes weights in rats treated up
to 5.0 mg/kg/day for 28 days (NTP, 2019) or in absolute or relative testis weight in rats exposed
to 20 ppm in the diet for 53 weeks (equivalent to 0.984 mg/kg/day) (Butenhoff et al., 2012). In a
subchronic study, no significant changes were observed in relative or absolute testis weight upon
exposure to PFOS at doses of 0.015 mg/kg/day and 0.5 mg/kg/day for a duration of 60 days in
Wistar rats (Alam et al., 2021). Effects in mice exposed to PFOS were observed in a subchronic
study in which significant decreases in absolute and relative testis weights were noted in mice
exposed to 10 mg/kg/day for 35 days (Qu et al., 2016). No effects were seen in relative
epididymis or testis weights of mice treated up to 10 mg/kg/day for four weeks (Qiu et al., 2016),
nor were any effects noted in the relative testes weight of mouse pups treated from GD 1 to GD
17 (Lai et al., 2017a). Similarly, no significant changes in relative epididymis or testis weight
were observed for ICR mice treated up to 10 mg/kg/day for four weeks (Qiu et al., 2020). Male
cynomolgus monkeys treated with up to 0.75 mg/kg/day for 182 days showed no changes in
absolute or relative epididymis or testis weights (Seacat et al., 2002).
Histopathological examination of rats following 28 days or 2 years of exposure revealed no
treatment-related changes in the testes, epididymis, seminal vesicle, or prostate (NTP, 2019;
Butenhoff et al., 2012). However, Lopez-Doval et al. (2014) noted edema around seminiferous
C-21
-------
APRIL 2024
tubules and malformed spermatids in male rats treated at > 1 mg/kg/day with marked edema and
loss and degeneration of the spermatozoids observed at 6 mg/kg/day following PFOS exposure
up to 6 mg/kg/day for 28 days. The specific incidences of histopathological findings were not
reported in this study, and statistical analysis was not conducted. In another study, subchronic
exposure in rats revealed lesions including vacuolations in spermatogonia, spermatocytes, and
Leydig cells, as well as exaggerated intracellular space and disturbed germ cells in rats treated at
10 mg/kg/day; however, incidences of specific findings were not reported, and statistical
analyses were not conducted (Qu et al., 2016).
Relevant histopathological findings in a 28-day study in mice included Sertoli cell vacuolization
and derangement of the cell layers at 2.5 mg/kg/day, 25 mg/kg/day, and 50 mg/kg/day and
dislocated immature germ cells in seminiferous tubules at 50 mg/kg/day (Qiu et al., 2013);
however, incidences of specific findings were not reported, and methods used for statistical
analysis are unclear. These findings were confirmed by observing the ultrastructure of
seminiferous epithelia by electron microscopy. In addition, PFOS was observed to disrupt the
blood-testis barrier in vitro and in vivo in two studies, suggesting that Sertoli cells in the testes
are a target for PFOS toxicity (Qiu et al., 2016; Qiu et al., 2013). Along with observations of
reduced epididymal sperm count in these studies, these results collectively suggest the potential
for PFOS exposure to induce deterioration of the testis and impair spermatogenesis in mice.
In a single study in cynomolgus monkeys, histopathology of the testes, prostate, and seminal
vesicles and cell proliferation in the testes were examined following exposure to PFOS for
182 days, however no differences were noted when compared with controls (Seacat et al., 2002).
C.1.2.5.2 Females
Female organ weight and histopathological data in rats were only available from the 28-day NTP
study (NTP, 2019). In females, relative and absolute uterus with cervix and vagina weights in
Sprague-Dawley rats were not affected following a 4-week exposure to PFOS at doses up to
5 mg/kg/day. In addition, no treatment-related histopathological changes were observed in the
uterus or ovary (NTP, 2019). A chronic study in rats (Butenhoff et al., 2012) measured the
weight of the uterus with cervix at the 53-week interim evaluation and evaluated histopathology
of the ovaries, uterus, vagina, and cervix after two years of exposure to concentrations up to
20 ppm in the diet (equivalent to 1.251 mg/kg/day) and reported no significant findings for those
organs. Similarly, Seacat et al. (2002) did not report alterations in ovary weight or uterine or
vaginal histopathology in female cynomolgus monkeys dosed with up to 0.75 mg/kg/day PFOS
for 182 days. Effects on placental characteristics such as weight and capacity, as well as
histopathological effects were noted in rats and mice exposed to PFOS during gestation (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.1.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse reproductive outcomes is discussed in
Sections 3.2.5, 3.3.4, and 3.4.1.2 of the 2016 PFOS HESD (U.S. EPA, 2016c). There are 57
studies from recent systematic literature search and review efforts conducted after publication of
the 2016 PFOS HESD that investigated the mechanisms of action of PFOS that lead to
reproductive effects. A summary of these studies is shown in Figure C-12. Additional
C-22
-------
APRIL 2024
mechanistic synthesis will not be conducted since evidence suggests but is not sufficient to infer
that PFOS may cause respiratory effects.
Mechanistic Pathway Animal Human In Vitro Grand Total
Angiogenic, Antiangiogenic, Vascular Tissue Remodeling
1
0
1
2
AtherogenesisAnd Clot Formation
Big Data, Non-Targeted Analysis
4
1
5
9
Cell Growth, Differentiation, Proliferation, Or Viability
8
0
23
29
Cell Signaling Or Signal Transduction
9
1
18
27
Extracellular Matrix Or Molecules
2
0
2
4
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
5
1
1
6
Hormone Function
10
1
13
23
Inflammation And Immune Response
Oxidative Stress
2
0
5
7
Renal Dysfunction
Xenobiotic Metabolism
2
0
4
6
Other
2
0
1
3
Not Applicable/Not Specified/Review Article
1
0
0
1
Grand Total
21
3
38
57
Figure C-12. Summary of Mechanistic Studies of PFOS and Reproductive Effects
Interactive figure and additional study details available on HAWC.
C.1.4 Evidence Integration
C. 1.4.1 Reproductive Effects in Males
There is slight evidence for an association between PFOS exposure and male reproductive
effects based on inverse associations with testosterone in male children. Inverse associations
with testosterone were observed in two medium confidence studies in children, and one study
reported an inverse association for E2. Among low confidence studies, there was mixed evidence
for an association between PFOS exposure and testosterone in cross-sectional studies (Di Nisio
et al., 2019; Zhou et al., 2017c; Zhou et al., 2016) in children and adolescents. However, these
mixed associations were observed in populations at different stages of pubertal development.
Results showed decreasing testosterone with increasing serum PFOS in children, but increased
C-23
-------
APRIL 2024
testosterone with higher PFOS exposure levels in adolescents. In adolescents, there were no
effects on pubertal development, but associations were observed for penile measurements,
testicular measurements, and sperm parameters (Di Nisio et al., 2019). Evidence was also
inconsistent for AGD in infants. In adults, there was evidence in one study (Cui et al., 2020) of
an inverse association between serum PFOS and testosterone, and these associations were also
observed using semen PFOS. Inverse associations were also seen for E2, SHBG, and the total
T/LH ratio. Regarding semen and sperm characteristics in adults, associations were observed for
several parameters in analyses of semen PFOS, including increased sperm DNA fragmentation
and decreased measures of sperm motility. Other results for markers of genotoxic effects
(e.g., sperm Y:X-chromosome ratio, sperm DNA methylation) in sperm were inconsistent.
The animal evidence for an association between PFOS exposure and reproductive toxicity in
males is slight based on several high or medium confidence studies of varying durations showing
that oral exposure to PFOS can affect the male reproductive system. However, many of the
observed reproductive effects (e.g., decreased E2 levels in male monkeys) occurred at doses that
also resulted in reduced body weight which can be confounding effects for reproductive
endpoints. Additionally, several of the observed effects were not consistent across species
(e.g., sperm parameters, testis weight, E2 levels in males) which increases uncertainty about the
relevance of these effects to humans or potential differences in the MOA between species.
Several studies reported effects of PFOS exposure on male mouse and rat reproductive organ
histopathology (Qiu et al., 2016; Qu et al., 2016; Lopez-Doval et al., 2015; Lopez-Doval et al.,
2014; Qiu et al., 2013). However, these studies did not report incidence data which hinders
further quantification or conclusions about these results. In male mice, these histopathological
alterations were accompanied by a reduction in epididymal sperm count, though this effect was
not observed in male rats. Although reductions in epididymal sperm counts across mouse studies
ranged from 25% to 70% at the highest doses tested (Lai et al., 2017a; Qiu et al., 2016; Qu et al.,
2016; Qiu et al., 2013) and are consistent with effects seen in humans, fertility may be normal in
male rodents even with sperm reductions as great as 90% (Gray et al., 1988). Without further
evidence of reduced fertility or quantitative evidence of histopathological changes in the testes or
epididymis, it is unclear whether reductions in sperm counts can be considered adverse.
Similar uncertainties arise when linking the observed hormonal alterations with functional
reproductive consequences. Changes in LH, FSH, and prolactin were observed in male rats,
however, lack of histopathological and sperm parameter effects (specific to rats), as well as
inconsistent effects on testosterone levels, make it difficult to assess the relevance of these
changes. It is difficult to ascertain the magnitude of change in hormone levels that can be
considered adverse without concurrent supporting evidence of functional or histopathological
reproductive consequences.
C.l.4.2 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause reproductive effects in
males under relevant exposure circumstances (Table C-l). This conclusion is based primarily on
effects on inverse associations with testosterone in male children and adults, and decreased AGD
in children observed in studies in humans exposed to median PFOS ranging from 1.4 to
34.8 ng/mL. Although there is some evidence of negative effects of PFOS exposure on semen
and sperm characteristics in adults, there is considerable uncertainty in the results due to
C-24
-------
APRIL 2024
inconsistency across studies and limited number of studies. For male reproductive toxicity, the
conclusion is based primarily on observed changes in hormonal parameters in adult rodents
following exposure to doses as low as 0.5 mg/kg/day PFOS. However, findings from animal
studies are similarly inconsistent as in epidemiological studies. In animal studies, there are
uncertainties in the adversity of the observed effects, a lack of quantifiable histopathological
evidence in reproductive organs, and inconsistencies in responses observed across studies and
species.
C-25
-------
APRIL 2024
Table C-l. Evidence Profile Table for PFOS Reproductive Effects in Males
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.l.l)
Male reproductive
hormones
1 High confidence study
8 Medium confidence
studies
6 Low confidence
studies
In children and
adolescents, inverse
associations for total
testosterone were
observed in two studies
(2/8), including a medium
confidence study
reporting a significant
inverse trend. One study
reported higher total
testosterone levels among
highly exposed
adolescents but was of
low confidence. Findings
for estradiol in male
children were generally
less precise, however,
one medium confidence
study (1/6) observed a
significant, dose-
dependent increase in
estradiol, which was
accompanied by a
significant decrease in the
testosterone/estradiol
ratio. Findings for LH
and FSH were mixed, but
significantly increased
LH was observed in one
low confidence study,
and significantly
decreased FSH was
• High and medium
confidence studies
• Consistent direction
of effects for
testosterone levels
• Low confidence studies
• Lmprecision of most
findings
• Potential for residual
confounding by SES and
smoking status
©OO
Slight
Evidence for male
reproductive effects is
based on several studies
reporting consistent and
coherent changes to sex
hormones. Effects on sex
hormones were supported
by adverse effects
observed for other
outcomes such as sperm
quality (i.e., sperm DFI
and HDS) and
anthropometric measures.
Uncertainties remain
regarding mixed results
in adults and imprecise
results in some medium
confidence studies. There
were also a limited
number of studies
evaluating certain
endpoints such as semen
parameters and pubertal
development.
©OO
Evidence Suggests
Primary basis:
Human evidence indicted
effects on inverse
associations with
testosterone in male
children and adults, and
decreased AGD in
children observed in
studies in humans exposed
to median PFOS.
Although there is some
evidence of negative
effects of PFOS exposure
on semen and sperm
characteristics in adults,
there is considerable
uncertainty in the results
due to inconsistency
across studies and limited
number of studies. Animal
evidence indicated
changes in hormonal
parameters in adult rodents
following exposure to
PFOS. However, findings
from animal studies are
similarly inconsistent as in
epidemiological studies. In
animal studies, there are
uncertainties in the
C-26
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
observed in a medium
confidence study. In
adults, one study (1/3)
observed significant
decreases in total
testosterone and the
testosterone/estradiol
ratio. Another medium
confidence study reported
a non-significant increase
in total testosterone, but
other results for
testosterone were
imprecise. One study
reported a non-significant
decrease in estradiol, and
one study reported a
significant increase in
LH. Findings for SHBG
were mixed.
Semen parameters
4 Medium confidence
studies
1 Low confidence
studies
One study examined
semen parameters in high
school students and
observed significant
increases in semen pH
and increased deficits in
sperm morphology.
Semen parameter
findings in adults were
generally consistent
between endpoints but
did not always indicate
adverse effects. Sperm
count was non-
significantly increased in
• Medium confidence
studies
• Consistent direction
of effects for most
findings
• Low confidence study
• Lmprecision of most
findings
• Lncoherence of direction
of effect for adult semen
parameters
• Potential for residual
confounding by SES and
smoking status
adversity of the observed
effects, a lack of
quantifiable
histopathological evidence
in reproductive organs,
and inconsistencies in
responses observed across
studies and species.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
C-27
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
two studies (2/3), non-
significant positive
associations were
observed for sperm
concentration in three
studies (3/4), and semen
volume was reported to
be non-significantly
increased in two studies
(2/4). Adverse effects
were also observed,
including decreased
normal morphology (1/2),
increased sperm HDS,
and significantly
increased sperm DFI.
Sperm HDS and DFI are
measures of sperm
chromatin integrity and
sperm DNA damage,
respectively.
Anthropometric
measurements of male
reproductive organs
1 High confidence study
2 Medium confidence
studies
1 Low confidence study
Three studies examined
measurements in male
infants. Non-significant
increases in AGD were
observed in two studies
(2/3), but findings were
not consistent across
timepoints. One study
examined anthropometric
measurements in male
• High and medium
confidence studies
• Consistent direction
of effects
• Coherence of
findings
• Low confidence study
• Lmprecision of some
findings
• Potential for residual
confounding by SES and
smoking status
C-28
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Male pubertal
development
1 Medium confidence
study
Male mating and
fertility
1 Medium confidence
study
Male reproductive
hormones
1 High confidence study
8 Medium confidence
studies
Summary and Key
Findings
high school students.
Adverse effects were
observed in adolescents
with higher exposure
levels, including smaller
testicular volume, shorter
penis length, and smaller
penis circumference.
Tanner stages (G2 and
G5) and earlier onset of
voice break, but none
were significant.
Factors That Increase
Certainty
Factors That Decrease
Certainty
• Inconsistent direction of
effects across species
• Changes in body weight
may limit ability to
interpret these responses
Evidence Stream
Judgment
©oo
Slight
Evidence was based on
15 high and medium
confidence studies. There
were no observed effects
on mating or fertility in
the only available two-
generation reproduction
study; however, other
studies observed effects
on hormone levels, sperm
count, and testis weight
and histopathology. Some
of the reproductive
Evidence Integration
Summary Judgment
Findings for changes in • Medium confidence • Limited number of
timing of pubertal study studies examining
development were largely outcome
non-significant. Study
authors reported earlier
onset of individual
Evidence From In Vivo Animal Studies (Section C.1.2)
No effects on male • Medium confidence • Limited number of
mating or fertility study studies examining
parameters were observed outcome
in a two-generation
reproduction study in rats
with exposure beginning
six weeks prior to mating
(1/1).
Alterations in • High and medium
testosterone levels in confidence studies
male rats (3/8), mice
(4/8), and monkeys (1/8)
were inconsistent.
Reports of decreases
(5/8), increases (1/8), and
C-29
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
no change (2/8) in serum
testosterone were
reported following
developmental, short-
term, and subchronic
exposure. Mixed effects
on serum estradiol
included decreased levels
in rats and monkeys
(3/6), increases (1/6) in
mice, or no effects (2/6).
Short-term studies in
male rodents reported no
effect onFSH (1/1),
decreases in LH (1/2),
and decreases in prolactin
(1/1).
Sperm parameters
1 High confidence study
5 Medium confidence
studies
In mice, five short-term,
subchronic, or
developmental studies
observed dose-dependent
reductions in epididymal
sperm count (5/5).
However, in rats, no
effects on epididymal or
testicular sperm counts or
epididymal sperm
motility were reported
(1/1).
• High and medium
confidence studies
• Consistent direction
of effects within
species
• Inconsistent direction of
effects across species
Male pubertal
development
3 Medium confidence
studies
No effects on age at
preputial separation were
observed in reproductive
and developmental
studies in male rats (3/3).
• Medium confidence
studies
• Consistent direction
of effects
• Limited number of
studies examining
outcome
effects observed (e.g.,
decreased testosterone
and estradiol levels) may
be secondary effects
because they occurred at
doses that also resulted in
reduced body weight.
Additionally, several of
the observed effects were
not consistent across
species (e.g., sperm
parameters, testosterone
and estradiol levels)
which increases
uncertainty about the
relevance of these effects
to humans or potential
differences in the mode
of action between
species. Studies reporting
alterations in testis
histopathology did not
report incidence data
which hinders
conclusions about these
results. In male mice,
these histopathological
alterations were
accompanied by a
" reduction in epididymal
sperm count. Without
further evidence of
reduced fertility or
quantitative evidence of
C-30
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Organ weights
2 High confidence
studies
7 Medium confidence
studies
Most studies in rats,
mice, or monkeys found
no effects on absolute or
relative testis weight
(8/9). One subchronic
study in mice observed
decreases in absolute and
relative testis weight
(1/9) only at the highest
dose tested. No effects on
absolute or relative
epididymis weight were
observed (4/4).
' High and medium
confidence studies
No factors noted
Histopathology
2 High confidence
studies
4 Medium confidence
studies
Two high confidence
studies in rats and one
medium confidence study
in monkeys found no
histopathological changes
in the testes, prostate,
epididymides, or seminal
vesicles following short-
term or chronic exposure
(3/6). Three studies in
mice observed
histopathological changes
in the testes following 4-
5 wk of exposure (3/6).
These changes included
vacuo lations in
spermatogonia,
spermatocytes, Leydig
cells, and Sertoli cells,
and disturbed germ cell
layers; however, results
• High and medium
confidence studies
• Inconsistent direction of
effects across species
histopathological changes
in the testis or
epididymis, it is unclear
whether reductions in
sperm counts can be
considered adverse.
Changes in LH, FSH, and
prolactin were observed
in male rats; however, the
lack of histopathological
and sperm parameter
effects (specific to rats),
as well as inconsistent
effects on testosterone
levels, make it difficult to
assess the relevance of
these changes.
C-31
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Evidence Integration
Studies and Summary and Key Factors That Increase Factors That Decrease Evidence Stream Summary Judgment
Interpretation Findings Certainty Certainty Judgment
were all reported
qualitatively only.
Notes: LH = luteinizing hormone; FSH = follicle stimulating hormones; SHBG = sex hormone binding globulin; SES = socioeconomic status; DFI = DNA fragmentation index;
HDS = high DNA stainability; DNA = deoxynucleic acid; AGD = anogenital distance; G2 = genital stage 2; G5 = genital stage 5; wk = weeks.
C-32
-------
APRIL 2024
C. 1.4.3 Reproductive Effects in Females
There is slight evidence for an association between PFOS exposure and female reproductive
effects in humans based on observed increases in preeclampsia and gestational hypertension,
with most studies observing positive non-significant associations, in populations with high
exposure levels and at levels typical in the general population.
Epidemiological evidence of a relationship between PFOS exposure and female fertility is
mixed. Since the 2016 Health Assessment, nine studies have investigated associations between
PFOS exposure and fertility. While some studies reported more frequent or intense female
fertility problems with increasing PFOS exposure (Zhang et al., 2018b; Crawford et al., 2017;
Mccoy et al., 2017), others found PFOS to be positively associated with female fertility
indicators (Kim et al., 2020b; Bach et al., 2018; Lum et al., 2017), and some did not observe any
clear trends (Wang et al., 2017). Kim et al. (2020b) also observed some non-significant, positive
associations between follicular fluid PFOS and fertility etiology factors for other gynecologic
pathologies, including endometriosis, PCOS, genital tract infections, and idiopathic infertility.
There is limited, consistent epidemiological evidence of an inverse association between serum
PFOS levels in pregnancy and breastfeeding duration. Timmermann et al. (2017b) observed
negative associations between PFOS exposure and exclusive and total breastfeeding duration,
while Romano et al. (2016) observed increased relative risk of breastfeeding termination with
increasing PFOS exposure.
Human epidemiological evidence of a relationship between PFOS exposure and the female
reproductive milestones of age at menarche and menopause is mixed. In the 2016 Health
Assessment, Christensen et al. (2011) observed a non-significant decreased adjusted OR for
earlier age at menarche for continuous prenatal PFOS exposure. Since the 2016 Health
Assessment, Ernst et al. (2019) observed a significant inverse association between age at
attainment for overall puberty indicators and a non-significant inverse association for continuous
prenatal PFOS exposure and age at menarche. In the 2016 Health Assessment, Knox et al. (2011)
observed significant increased odds of natural menopause across PFOS quintiles for women ages
51-65 years in the C8 Health Project. Since the 2016 Health Assessment, Ding et al. (2020)
observed significant, positive associations for serum Sm-PFOS and n-PFOS and risk of natural
menopause in women aged 45-56. However, findings from studies concurrently assessing
menstruation events and PFOS levels in blood must be interpreted with caution due to potential
reverse causality, as menstruation is a primary route of PFOS excretion for people who
menstruate.
Since the 2016 Health Assessment, 20 studies have assessed relationships between PFOS
exposure and various female reproductive hormones. 12 of these studies were conducted in
female infants and adolescents. Commonly assessed female reproductive hormones were 17-
OHP, DHEA, E2, FSH, SHBG, and testosterone. While most studies did not report significant
associations or consistent trends between PFOS exposure and these outcomes, Itoh et al. (2016)
observed significant negative associations for maternal serum PFOS and cord blood prolactin
and progesterone levels and Wang et al. (2019a) observed significant positive associations for
cord blood PFOS and cord blood estrone and E3. In pregnant women, Yao et al. (2019) observed
significant, positive associations for cord blood PFOS and testosterone and testosterone to E2
C-33
-------
APRIL 2024
ratio and Toft et al. (2016) observed significant, positive trends in 17-OHP, androstenedione,
progesterone, and testosterone across amniotic fluid PFOS tertiles.
The recent epidemiological evidence is also suggestive of an association between PFOS and
preeclampsia and gestational hypertension, though there is conflicting evidence on altered
puberty onset and limited data suggesting reduced fertility and fecundity. The association are
inconsistent across reproductive hormone parameters, and it is difficult to assess the adversity of
these alterations.
The animal evidence for an association between PFOS exposure and female reproductive toxicity
is slight based on several high or medium confidence studies of varying durations showing that
oral exposure to PFOS can affect the female reproductive system. However, many of the
observed reproductive effects (e.g., decreased gestation length in female rats, decreased prolactin
levels in female mice) occurred at doses that also resulted in decreased gestational body weight
which can be confounding effects for reproductive endpoints.
Uncertainties arise when linking the observed hormonal alterations with functional reproductive
consequences. NTP (2019) reported modest increases in testosterone concentrations (37%
increase) in female rats with PFOS doses of 1.25 mg/kg/day and 2.5 mg/kg/day, but not the
highest dose of 5 mg/kg/day. The response in the highest dose was confounded by decreased
body weight. The alterations in testosterone were accompanied by dose-dependent increases in
estrous cycle length, though this increase was not statistically significant and alterations in the
estrous cycle were not observed in a second study in female rats (Luebker et al., 2005a). It is
difficult to ascertain the magnitude of change in hormone levels that can be considered adverse
without concurrent supporting evidence of functional or histopathological reproductive
consequences.
C.l.4.4 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause reproductive effects in
females under relevant exposure circumstances (Table C-2). This conclusion is based primarily
on effects on preeclampsia and gestational hypertension, female reproductive milestones, and
female reproductive hormonal outcomes observed in studies in humans exposed to median PFOS
ranging from 1.4 ng/mL to 34.8 ng/mL. There is considerable uncertainty in the results due to
inconsistency across studies and the limited number of studies. For female reproductive toxicity,
the conclusion is based primarily on observed changes in hormonal parameters in adult rodents
following exposure to doses as low as 1.25 mg/kg/day PFOS. However, findings from animal
studies are similarly inconsistent as in epidemiological studies. In animal studies, there are
uncertainties in the adversity of the observed effects, a lack of quantifiable histopathological
evidence in reproductive organs, and inconsistencies in responses observed across studies and
species.
C-34
-------
APRIL 2024
Table C-2. Evidence Profile Table for PFOS Reproductive Effects in Females
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.l.l)
Female reproductive
hormones
3 High confidence
studies
10 Medium confidence
studies
7 Low confidence
studies
Results from assessment
of female reproductive
»High and medium
confidence studies
hormones were mixed. In • Coherence of
13 studies of female
children and adolescents,
7 studies reported
significant associations.
One medium confidence
study reported increased
El and E3 and an inverse
association with E2 (1/7).
Two other studies
reported increased E2
(2/7), and one also
reported increased FSH
(1/2). Three studies, one
high, one medium, and
one low confidence,
reported increases in
testosterone (3/7). One
medium confidence study
observed inverse
associations with
progesterone and
prolactin (1/7). Eight
studies examined adult
women, though many
were low confidence
(5/8). Four studies
reported significant
effects (4/8). Two low
confidence studies
observed inverse
findings for
testosterone
• Low confidence studies
• Lnconsistent direction of
effects
• Lmprecision of most
findings
• Potential for selection
bias and residual
confounding by age and
SES
©oo
Slight
Evidence for female
reproductive effects is
based on several studies
reporting effects on sex
hormones and increased
odds of preeclampsia.
There was also evidence
for changes in age at
natural menopause.
Uncertainties remain
regarding mixed findings
in studies of sex
hormones, and a limited
number of studies
examining outcomes such
as female reproductive
milestones and
anthropometric
measurements.
©OO
Evidence Suggests
Primary basis:
Human evidence indicted
effects on preeclampsia
and gestational
hypertension, female
reproductive milestones,
and female reproductive
hormonal outcomes
observed in studies in
humans exposed to median
PFOS. There is
considerable uncertainty in
the results due to
inconsistency across
studies and the limited
number of studies. Animal
evidence indicated
changes in hormonal
parameters in adult rodents
following exposure to
PFOS. However, findings
from animal studies are
similarly inconsistent as in
epidemiological studies. In
animal studies, there are
uncertainties in the
adversity of the observed
effects, a lack of
quantifiable
histopathological evidence
in reproductive organs,
C-35
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
associations with E2
(2/4), one medium study
observed increased
progesterone,
testosterone, and 17-OHP
(1/4) and one observed an
inverse association with
free androgen index (1/4).
and inconsistencies in
responses observed across
studies and species.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
Preeclampsia and
gestational
hypertension
1 High confidence study
5 Medium confidence
studies
Six studies examined
preeclampsia in pregnant
women. One study
reported significant
positive results, while
four studies of medium
and high confidence
reported non-significant
positive associations with
preeclampsia. Three
studies reported inverse
associations (3/6).
Of the three studies
examining gestational
hypertension (3/6), two
reported inverse
associations but neither
reached significance
(2/3). After observing
non-significant increased
odds of gestational
hypertension, one
medium confidence study
reported significantly
increased DBP.
»High and medium
confidence studies
• Imprecision of all
findings
• Inconsistent direction of
effects
• Potential for reverse
causality
C-36
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Female reproductive
milestones
1 High confidence study
2 Medium confidence
studies
1 Low confidence study
Three studies examined
reproductive milestones
related to menstruation,
two in adolescent
populations (2/3) and one
in an adult population
(1/3). Two studies, one
low confidence study in
adolescents (1/2) and one
medium confidence study
in adults (1/1), reported
non-significant increases
in long menstrual cycles.
A significant inverse
association was observed
among adolescents for
average age at attainment
for all pubertal indicators
(1/2). One high
confidence study reported
significant positive
associations with age at
natural menopause.
• High and medium
confidence studies
• Consistent direction
of effects
• Low confidence study
• Potential for residual
confounding by not
identifying confounders
• Limited number of
studies examining
specific outcomes
Fertility indicators
6 Medium confidence
studies
6 Low confidence
studies
Examinations of fertility
indicators include
fecundability,
fertilization rate, and
measures of ovarian
health, such as AMH
levels or endometriosis.
Twelve studies evaluated
fertility indicators in non-
pregnant women with
mixed results. One
medium confidence study
• Medium confidence
studies
• Low confidence studies
• Lmprecision of most
findings
• Potential for residual
confounding by not
identifying confounders
• Limited number of
studies examining
specific outcomes
C-37
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
reported significant
inverse associations with
endometriosis-driven
infertility. In contrast,
low confidence studies
observed significantly
increased odds of
endometriosis (1/3) and
ovarian syndromes (2/3).
Other studies reported
non-significant positive
associations with
endometriosis (2/12).
Results from remaining
studies were inconsistent
and did not reach
significance.
Breastfeeding
Two medium confidence
• Medium confidence
• Limited number of
2 Medium confidence
cohort studies reported
studies
studies examining
studies
significant inverse
• Consistent direction
outcome
associations with
of effects
breastfeeding duration
• Precision of findings
(2/2).
Anogenital distance
1 High confidence study
1 Medium confidence
study
Two studies examined
measures of AGD,
including anoclitoris and
anofourchette distances,
in female infants. A high
confidence study reported
significant inverse
associations with
anoclitoris distance for
the highest exposure
group and in continuous
analysis. Results for
»High and medium
confidence studies
• Limited number of
studies examining
outcome
• Inconsistent direction of
effects
C-38
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
anofourchette distances
were inverse but not
significant. A medium
confidence study
observed non-significant
mixed results for both
measures.
Evidence From In Vivo Animal Studies (Section C.1.2)
Female mating and
No effects on female • Medium confidence
• Limited number of
fertility
mating or fertility studies
studies examining
2 Medium confidence
parameters were observed • Consistent direction
outcome
studies
in one- and two- 0f effects
generation reproduction
studies in rats with PFOS
exposure beginning six
weeks prior to mating
(2/2).
Female gestation
Duration of gestation was • Medium confidence
• Limited number of
length
slightly decreased in a studies
studies examining
3 Medium confidence
one-generation rat • Consistent direction
outcome
studies
reproduction study and in 0f effects
• Small magnitude of
a two-generation rat
effect
study, both with exposure
beginning six weeks prior
to mating (2/3). No effect
on gestation length was
observed in another rat
study with exposure
beginning on the first day
of gestation.
Female reproductive
hormones
1 High confidence study
3 Medium confidence
studies
Significant alterations in
female testosterone levels
were found (1/3). No
significant changes in
serum E2 were found in
• High and medium
confidence studies
• Dose-response
relationship
• Limited number of
studies examining
specific outcomes
©OO
Slight
Evidence is based on 10
high and medium
confidence studies. There
were no observed effects
on mating or fertility in
the only available two-
generation reproduction
study; however, other
studies observed effects
on length of gestation,
hormone levels, and
estrous cyclicity. Some of
the observed reproductive
effects (e.g., decreased
gestation length in female
rats, decreased prolactin-
family hormones in
female mice) may be
secondary effects because
they occurred at doses
that also resulted in
decreased gestational
body weight. One study
reported modest increases
C-39
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
female monkeys exposed
for 26 wk or in female
mice exposed in utero
from GD 1-17. One
mouse study measured
maternal serum
concentrations of
prolactin-family
hormones (i.e., mPL-II,
mPLP-Ca, mPLP-K)
during pregnancy and
found dose-dependent
decreases (1/1).
Changes in body weight
may limit ability to
interpret these responses
Estrous cyclicity and
ovarian function
1 High confidence study
3 Medium confidence
studies
No significant effect on
estrous cyclicity were
found in two rat studies
(2/2). However, a high
confidence study in rats
observed a dose-
dependent, but not
significant, increase in
estrous cycle length and
prolonged diestrus (1/1)
compared with controls.
No effects on the number
and distribution of
corpora lutea in the
ovaries were observed in
pregnant rats and rabbits
(3/3).
»High and medium
confidence studies
• Limited number of
studies examining
specific outcomes
• Small magnitude of
effect
Female pubertal
development
3 Medium confidence
studies
No effects on age at
vaginal opening were
observed in reproduction
and developmental
studies in rats (3/3).
• Medium confidence
studies
• Consistent direction
of effects
• Limited number of
studies examining
outcome
in testosterone
concentrations in
females, but the response
in the highest dose was
affected by decreased
body weight. The
increases in testosterone
were accompanied by
dose-dependent increases
in estrous cycle length,
though this increase was
not statistically
significant and alterations
in the estrous cycle were
not observed in a second
study in female rats.
C-40
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Organ weights
2 High confidence
studies
1 Medium confidence
study
No effects were observed
on absolute or relative
weights of the uterus
(2/2) or ovaries (1/1).
• High and medium
confidence studies
• Consistent direction
of effects
• Limited number of
studies examining
outcome
Histopathology
2 High confidence
studies
1 Medium confidence
study
No exposure-related • High and medium
histopathological findings confidence studies
were reported for the • Consistent direction
ovaries (2/2), uterus of effects
(3/3), vagina (2/2), or
cervix (1/1).
• Limited number of
studies examining
outcome
Notes: El = estrone; E3 = estriol; E2 = estradiol; FSH = follicle stimulating hormones; 17-OHP = 17-hydroxyprogesterone; SES = socioeconomic status; DBP = diastolic blood
pressure; AMH = anti-Mullerian hormone; AGD = anogenital distance; wk = weeks; GD = gestation day; mPL-II = mouse placental lactogen II; mPLP-Ca = mouse prolactin-like
protein-Ca; mPLP-K = mouse prolactin-like protein-K.
C-41
-------
APRIL 2024
C.2 Endocrine
EPA identified 35 epidemiological and 14 animal studies that investigated the association
between PFOS and endocrine effects. Of the epidemiological studies, 4 were classified as high
confidence, 15 as medium confidence, 9 as low confidence, 4 as mixed (1 high/medium, 1
medium/low, 1 medium/uninformative, and 1 low/uninformative) confidence, and 3 were
considered uninformative (Section C.2.1). Of the animal studies, 1 was classified as high
confidence, 10 as medium confidence, 2 as low confidence, and 1 was mixed {medium/low)
(Section C.2.2). Studies may have mixed confidence ratings depending on the endpoint
evaluated. Though low confidence studies are considered qualitatively in this section, they were
not considered quantitatively for the dose-response assessment (see Toxicity Assessment, (U.S.
EPA, 2024)).
C.2.1 Human Evidence Study Quality Evaluation and Synthesis
C.2.1.1 Introduction
Thyroid disease is more common in females than in males and encompasses conditions such as
hypothyroidism and hyperthyroidism. Hypothyroidism is characterized by elevated thyroid
stimulating hormone (TSH) and concurrently low thyroxine (T4) concentrations, while
subclinical hypothyroidism is characterized by elevated TSH in conjunction with normal T4 and
triiodothyronine (T3) levels. Hyperthyroidism is characterized by elevated T4 and low TSH, and
subclinical hyperthyroidism is characterized by low levels of TSH with normal T4 and T3 levels.
The 2016 Health Advisory (U.S. EPA, 2016a) and HESD (U.S. EPA, 2016c) reports identified
evidence of endocrine effects of PFOS for thyroid disease, hypothyroidism, and
hypothyroxinemia. Occupational studies examining the relationship between PFOS exposure and
endocrine outcomes did not find any significant associations. Studies on NHANES populations
(Wen et al., 2013; Melzer et al., 2010) reported associations between PFOS exposure (serum
PFOS concentrations) and thyroid disease. One study (Melzer et al., 2010) reported associations
with thyroid disease in men, and another study (Wen et al., 2013) saw associations with
subclinical hypothyroidism in men and women. In people without diagnosed thyroid disease or
without biomarkers of thyroid disease, thyroid hormones (i.e., TSH, T3 or T4) show mixed
effects across cohorts. In cross-sectional studies where thyroid hormones were measured in
association with serum PFOS, increased TSH was associated with PFOS exposure in most cases
(Berg et al., 2015; Webster et al., 2014; Wang et al., 2013). Increasing PFOS was associated with
increased T4 in children aged 1 to 17 years from the C8 cohort (Lopez-Espinosa et al., 2012);
however, PFOS was not associated with hypothyroidism. A small South Korean study examining
correlations between maternal PFAS during pregnancy and fetal thyroid hormones in cord blood
(Kim et al., 2011) found an association for PFOS and increased fetal TSH, as well as with
decreased fetal T3. TSH was the outcome most frequently associated with PFOS in studies of
pregnant women. In studies of pregnant women, PFOS was associated with increased TSH levels
(Berg et al., 2015; Wang et al., 2013). Pregnant women testing positive for the anti-thyroid
peroxidase (TPO) biomarker for autoimmune thyroid disease showed a positive association with
PFOS and TSH (Webster et al., 2014). A case-control study of hypothyroxinemia (normal TSH
and low free T4) in pregnant women (Chan et al., 2011), did not show associations of
hypothyroxinemia with PFOS exposure.
C-42
-------
APRIL 2024
For this updated review, 34 studies (35 publications)6 report on the association between PFOS
exposure and endocrine effects. Seven of the publications were studies in pregnant women
(Aimuzi et al., 2020; Dreyer et al., 2020; Inoue et al., 2019; Itoh et al., 2019; Reardon et al.,
2019; Berg et al., 2017; Shah-Kulkarni et al., 2016), and the remainder of the publications were
on the general population. Different study designs were utilized, including seven cohort studies
(Kim et al., 2020a; Lebeaux et al., 2020; Reardon et al., 2019; Liu et al., 2018a; Berg et al., 2017
Blake, 2018, 5080657; Crawford et al., 2017), seven cohort and cross-sectional studies (Dreyer
et al., 2020; Itoh et al., 2019; Xiao et al., 2019; Preston et al., 2018; Kato et al., 2016; Wang et
al., 2014), one case-control study (Predieri et al., 2015), one case-control and cross-sectional
study (Zhang et al., 2018b), and 19 cross-sectional studies (Abraham et al., 2020; Aimuzi et al.,
2020; Aimuzi et al., 2019; Caron-Beaudoin et al., 2019; Inoue et al., 2019; Jain and Ducatman,
2019b; Byrne et al., 2018; Dufour et al., 2018; Heffernan et al., 2018; Kang et al., 2018; Khalil et
al., 2018; Seo et al., 2018; Li et al., 2017; Tsai et al., 2017; van den Dungen et al., 2017; Shah-
Kulkarni et al., 2016; Yang et al., 2016a; Lewis et al., 2015; Audet-Delage et al., 2013; Jain,
2013). All observational studies measured PFOS in blood components (i.e., blood, plasma, or
serum). Six studies measured PFOS in cord blood (Liu et al., 2020b; Aimuzi et al., 2019; Dufour
et al., 2018; Tsai et al., 2017; Shah-Kulkarni et al., 2016; Yang et al., 2016a) and eight studies
measured PFOS in maternal blood or serum during pregnancy (Dreyer et al., 2020; Lebeaux et
al., 2020; Reardon et al., 2019; Xiao et al., 2019; Preston et al., 2018; Kato et al., 2016; Yang et
al., 2016a; Wang et al., 2014). The studies were conducted in different study populations
including populations from Belgium (Dufour et al., 2018), Canada (Caron-Beaudoin et al., 2019;
Reardon et al., 2019), China (Aimuzi et al., 2020; Aimuzi et al., 2019; Zhang et al., 2018b; Li et
al., 2017 Liu, 2020, 6569227; Yang et al., 2016a), Denmark (Dreyer et al., 2020; Inoue et al.,
2019; Xiao et al., 2019), Germany (Abraham et al., 2020), Italy (Predieri et al., 2015), Japan
(Itoh et al., 2019; Kato et al., 2016), Republic of Korea (Kim et al., 2020a; Kang et al., 2018;
Shah-Kulkarni et al., 2016), Taiwan (Tsai et al., 2017; Wang et al., 2014), the United Kingdom
(Heffernan et al., 2018), and the United States (Lebeaux et al., 2020; Jain and Ducatman, 2019b;
Blake et al., 2018; Byrne et al., 2018; Khalil et al., 2018; Liu et al., 2018a; Preston et al., 2018;
Crawford et al., 2017; Lewis et al., 2015; Jain, 2013). Two studies (Itoh et al., 2019; Kato et al.,
2016) belonged to the same cohort, the Hokkaido Study on the Environment and Children's
Health. While most studies evaluated the relationship between exposure to PFOS and thyroid
hormone concentrations, other endocrine outcomes were investigated as well, including: thyroid
disease, thyroid antibodies (thyroglobulin antibodies (TgAb) and thyroid peroxidase antibody
(TPOAb)), and thyroid hormone-associated proteins (e.g., thyroglobulin, thyroxine-binding
globulin).
C.2.1.2 Study Quality
Several considerations were specific to evaluating the quality of studies. First, timing of
exposure and hormone concentration measurements was important. Several studies on mother-
child dyads examined relationships between maternal serum PFOS measurements and thyroid
hormones in both mothers (i.e., a cross-sectional analyses) and in cord blood or children's serum
(i.e., a longitudinal analyses). Longitudinal comparisons between maternal PFOS concentrations
measured during pregnancy and thyroid hormone levels in cord blood or the child's blood
attenuate any concerns for potential reverse causality. Measuring PFOS and thyroid hormone
Itoh et al. (2019) reports thyroid-related hormone levels in a population overlapping with Kato et al. (2016).
C-43
-------
APRIL 2024
concentrations concurrently in maternal serum was considered adequate in terms of exposure
assessment timing. Given the long half-life of PFOS (median half-life = 3.4 years) (Li et al.,
2018), current blood concentrations are expected to correlate well with past exposures. Second,
timing of thyroid hormone assessment was a recurring concern due to the diurnal variation in
thyroid hormones. Thyroid hormone outcome misclassification due to timing of blood collection
is non-differential, however, study sensitivity may be impacted in cases where timing of
collection was uncontrolled.
There are 35 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and endocrine effects. Study quality evaluations for these 35 studies are shown in
Figure C-13 and Figure C-14.
Of the 35 studies identified since the 2016 assessment, 4 studies were classified as high
confidence, 15 as medium confidence, 9 as low confidence, 4 as mixed (1 high/medium, 1
medium/low, 1 medium/uninformative, and 1 low/uninformative) confidence, and 3 studies
(Abraham et al., 2020; Predieri et al., 2015) as uninformative.
C-44
-------
APRIL 2024
pe
-------
APRIL 2024
Khalil et al., 2018, 4238547-
—"—
+
l
+
— —
i
+
i
+
i
n
Kim et al., 2020, 6833758-
++
++
++
++
+
++
l++
Lebeaux et al., 2020, 6356361 -
++
++
++
++
++
+
+
Lewis et al., 2015, 3749030 -
+
+
+
+
-
+
+
-
Li et al., 2017, 3856460-
-
+
-
-
+
+
-
Liu et al., 2018, 4238396-
-
+
++
+
+
+
+
+
Predieri et al., 2015, 3889874 -
+
+
-
-
+
+
-
~
Preston et al., 2018, 4241056 -
+
+
+
+
+
+
+
+
Reardon et al., 2019, 5412435 -
+
+
-
+
+
+
+
+
Seo et al., 2018, 4238334-
-
+
D
-
-
-
~
Shah-Kulkarni et al., 2016, 3859821 -
+
+
+
+
+
+
+
+
Tsai et al., 2017, 3860107-
-
+
-
+
+
+
+
-
Wang et al., 2014, 2850394 -
+
+
+
+
++
+
+
+
Xiao et al., 2020, 5918609-
++
++
+
+
++
+
++
++
Yang et al., 2016, 3858535 -
+
+
+
+
+
+
+
+
Zhang et al., 2018, 5079665-
-
-
-
+
+
+
+
-
van den Dungen et al., 2017, 5080340 -
-
+
+
-
+
+
-
-
Legend
Q
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
B
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-14. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Endocrine Effects (Continued)
Interactive figure and additional study details available on IiAWC.
C-46
-------
APRIL 2024
The main concerns with low confidence and uninformative studies included a lack of
consideration for outcome sampling time, small sample sizes, or use of statistical methods that
did not account for confounding. Other studies rated as low or uninformative had issues
regarding the analysis, including a lack of accounting for population sampling methods (Lewis et
al., 2015), or use of statistical methods that did not account for confounding (Abraham et al.,
2020). Case-control studies (Kim et al., 2016a; Predieri et al., 2015) were rated uninformative
and presented issues with insufficient detail regarding participant recruitment and case
definitions. However, the largest issues identified in these studies included use of statistical
methods that did not account for potential confounding factors, and the sensitivity of both case-
control studies was impacted by small sample sizes.
C.2.1.3 Findings From Children
One high confidence study (Kim et al., 2020a) observed an inverse association between PFOS
concentrations and subclinical hypothyroidism (defined by reference thyroid hormone levels) at
age six which was consistent after additional adjustment for dietary iodine intake. The
association was observed in boys, but not in girls. A positive association was also observed for
PFOS and T3 at six years of age which was significant among boys but not girls, before and after
adjustment for dietary iodine intake.
Thyroid hormone levels were examined in 19 studies (Abraham et al., 2020; Kim et al., 2020a;
Lebeaux et al., 2020; Aimuzi et al., 2019; Caron-Beaudoin et al., 2019; Itoh et al., 2019; Xiao et
al., 2019; Dufour et al., 2018; Kang et al., 2018; Khalil et al., 2018; Preston et al., 2018; Tsai et
al., 2017; Kato et al., 2016; Kim et al., 2016a; Shah-Kulkarni et al., 2016; Yang et al., 2016a;
Predieri et al., 2015; Wang et al., 2014) and five observed significant effects (Appendix D). One
high confidence study (Xiao et al., 2019) on children from the Faroe Islands showed a large,
significant positive association between maternal third trimester PFOS concentrations and cord
serum TSH. The effect size for TSH was similar in both sexes but was no longer significant in
female infants. Additionally, sex-stratified analyses showed positive associations between
maternal PFOS and the free thyroxine index (FTI) in cord serum for girls. A medium confidence
study (Kato et al., 2016) on infants in Sapporo, Japan from the Hokkaido Study observed
positive associations with infant TSH which were consistent after stratifying by the infant's sex.
Analyses by quartile revealed a significant increasing trend (p for trend = 0.024) for infant TSH
and maternal blood. A related medium confidence study (Itoh et al., 2019) of a separate Japanese
cohort from the same region also found a significant positive association between maternal
serum PFOS and TSH among boys. When stratifying by the mother's thyroid antibody (TA)
status, the effect remained among boys born to TA-negative mothers. No effect was seen in TA-
positive mothers, but the sample size was small (n = 48).
Other medium confidence cross-sectional studies in newborns (Aimuzi et al., 2019) showed
significant inverse associations with TSH in single pollutant models. These associations
remained for girls after stratifying by sex. A significant positive association was observed for
free T3 (FT3) among this study sample, but a sensitivity analysis including only those infants
with detectable free FT3 concentrations was conducted due to the low detection rate.
Associations between PFOS and FT3 were no longer significant after removing participants with
non-detectable levels. A medium confidence study (Preston et al., 2018) in infants did not show
significant associations in continuous analyses; however, a significant inverse association was
C-47
-------
APRIL 2024
found for T4 among all infants in the highest PFOS exposure quartile and among boys in in
exposure quartile.
A study in Taiwan (Tsai et al., 2017) found significant positive associations for TSH and inverse
associations for T4 in cord blood among the entire sample and among boys in continuous
analyses. Analyses by exposure quantiles (<30th, 30th-59th, 60-89th, and >90th percentile)
were consistent in the direction of effect, but only reached significance for each effect comparing
the highest PFOS exposure quantile to the reference in the overall population. A significant
effect was also seen among boys in the second quantile (30th-59th) for TSH. However, only
27% of the initially recruited population had available PFOS and thyroid measurements, and
reasons for missing data were not provided. This limited the sample size (n = 118) and raised
concern for potential selection bias, contributing to a low confidence rating.
C.2.1.4 Findings From Pregnant Women
Thyroid hormone levels were examined in six studies (Aimuzi et al., 2020; Inoue et al., 2019;
Itoh et al., 2019; Reardon et al., 2019; Berg et al., 2017; Shah-Kulkarni et al., 2016) and five
observed significant effects (Appendix D). One high confidence study (Xiao et al., 2019)
observed a positive association between third trimester PFOS concentrations and maternal TSH
in mothers giving birth to girls. This association was not seen in the analysis of the entire cohort
or in mothers of boys only. A medium confidence study (Reardon et al., 2019) on a Canadian
cohort of pregnant women investigated associations between multiple PFOS isomers and thyroid
hormones at several timepoints during and after pregnancy. Accounting for all timepoints, a
significant positive association was observed for increasing branched PFOS concentrations and
TSH. The same association was not observed for linear PFOS, except at 3 months postpartum. In
this study, the authors note linear PFOS contributed to 69.0% of exposure concentrations while
branched PFOS constituted only 31.0%. Total PFOS exposure was not assessed. A medium
confidence cross-sectional study (Preston et al., 2018) observed a significant inverse association
for maternal TSH among TPOAb-positive mothers. One low confidence analysis (Kato et al.,
2016) of mothers in Sapporo, Japan from the Hokkaido Study observed significant decreases for
maternal TSH concentrations with increasing serum PFOS, which were also observed after
stratifying by the infant's sex. Analyses by quartile confirmed this decreasing trend (p < 0.001).
No significant effects were observed in mothers from the other Hokkaido cohort (Itoh et al.,
2019). Another low confidence study (Berg et al., 2017) from Norway showed positive
associations between maternal PFOS concentrations and TSH levels during the second trimester.
Analysis by quartile showed significant associations for the two highest exposure groups,
suggesting a consistent trend.
One cross-sectional study (Dufour et al., 2018) on mother-child dyads showed evidence of
increased risk of hypothyroidism in mothers. Analysis by quartile showed a consistent effect, but
only reached significance for mothers in the third PFOS exposure quartile. This study contained
a great deal of uncertainty regarding timing of outcome ascertainment and the method of disease
classification which diminish confidence in the findings for maternal hypothyroidism.
One high confidence study examined adrenal hormones among pregnant women in the OCC and
showed a significant decrease in diurnal urinary (dU) -cortisone and increase in dU-
cortisol/cortisone with twofold increases in serum PFOS concentrations (Dreyer et al., 2020).
However, dU- and serum Cortisol showed non-significant decreases.
C-48
-------
APRIL 2024
C.2.1.5 Findings From the General Adult Population
Thyroid function was examined in 13 studies (Lebeaux et al., 2020; Jain and Ducatman, 2019b;
Blake et al., 2018; Byrne et al., 2018; Liu et al., 2018a; Seo et al., 2018; Zhang et al., 2018b;
Crawford et al., 2017; Li et al., 2017; van den Dungen et al., 2017; Christensen et al., 2016b;
Lewis et al., 2015; Audet-Delage et al., 2013; Jain, 2013) and six observed significant effects
(Appendix D). A medium confidence study (Blake et al., 2018) in individuals residing near a
uranium processing facility in an area with PFAS-contaminated drinking water (Fernald
Community Cohort (FCC)) reported a positive association for TSH in whole study sample.
Stratifying by sex showed a difference in direction of effect between men and women, however,
the interaction term did not reach significance (sex interaction p-value = 0.12). In men, the
association for TSH was consistent and was accompanied by a significant inverse association
with total T4; no significant associations were observed for women.
Results were mixed in three overlapping NHANES studies (Jain and Ducatman, 2019b; Lewis et
al., 2015; Jain, 2013). One low confidence study (Lewis et al., 2015) showed several significant
and borderline significant results among NHANES (2011-2012) participants. Significant
positive associations were found between TSH in males (12-20 years old) and females (20-
40 years old), but other results were not consistent among the same stratified groups (by sex and
age). There is no evidence that the NHANES complex sampling design was accounted for in the
analysis which contributed to a low confidence rating. Jain (2013), another low confidence study,
did not find any significant effects among NHANES (2007-2008) participants. A medium
confidence follow-up study (Jain and Ducatman, 2019b) examined effects on thyroid hormones
stratified by glomerular filtration (GF) stage in a pooled NHANES dataset (2007-2012). A
significant effect was found for total T4 in those individuals with stage 3 A GF, the second most
severe stage. Associations for total T4 among other stages were non-significant and inconsistent
in direction of effect.
One additional cross-sectional study (Byrne et al., 2018) of Alaska Natives found a significant
sex interaction for free T3. Women showed a positive association between serum PFOS and free
T3 while an inverse association was found in men. Borderline significant inverse associations for
TSH and total T3 were also observed among men (p = 0.085 and p = 0.08, respectively). The
sensitivity of the study, however, was limited by the population size (total n = 85; male n = 38)
and resulted in a low confidence rating. Another low confidence study (Li et al., 2017) conducted
in China found significant associations for TSH, free T3, and free T4 among a population
oversampled for thyroid conditions (70%). Inverse associations were observed for free T3 and
free T4, while a positive association was found for TSH amongst the whole population.
Associations were not significant when stratified by thyroid disease state (i.e., normal,
hypothyroidism, Hashimoto's disease). The study was found to be low confidence due to missing
information on recruitment and participation, especially considering this was a convenience
sample. Additionally, there were concerns for selective reporting and residual confounding
because individuals (n = 202) varied greatly by age (1 month to 90 years) and lifestyle factors
were not addressed.
A case-control study (Zhang et al., 2018b) examined women with and without POI and observed
positive associations for TSH among both cases and controls. Additionally, inverse associations
were found among cases for free T3 and free T4. The thyroid hormone concentrations were
within normal ranges in both cases and controls. The study was rated as low confidence due to
C-49
-------
APRIL 2024
insufficient information on control recruitment and potential for reverse causation from irregular
menstruation (a PFOS elimination route) for those women with PCOS.
C.2.2 Animal Evidence Study Quality Evaluation and Synthesis
There are 4 studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and 10 studies from recent
systematic literature search and review efforts conducted after publication of the 2016 PFOS
HESD that investigated the association between PFOS and endocrine effects. Study quality
evaluations for these 14 studies are shown in Figure C-15.
C-50
-------
APRIL 2024
Conleyetal., 2022, 10176381 -
Curran et al., 2008, 757871
Dong et al., 2011, 1424949
Fuentes et al., 2006, 757859 - +
Fuentes et al., 2007, 757865 -
Lau et al., 2003, 757854
Li et al., 2016, 3981495
Luebker et al., 2005, 757857
NTP, 2019, 5400978-
Pereiro et al., 2014, 2230732 -
+
+
NR
+
+
+
+*
Salgado-Freiria et al., 2018, 5079767 -
+
+
NR
+
+
+
++
a
++
+
Seacat et al., 2002, 757853 -
++
+
NR
+
+
+
++
++
++
+
Thomford, 2002, 5432419-
+
-
NR
++ ++
-
++
++
++
-
Zhang et al., 2020, 6315674 -
-
+
NR
+
-
-
+
+
-
-
Legend
a
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
0
Critically deficient (metric) or Uninformative (overall)
F
Not reported
*
Multiple judgments exist
Figure C-15. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Endocrine Effects
Interactive figure and additional study details available on HAWC.
Animal studies suggest that exposure to PFOS can result in adverse effects to the endocrine
system. Overall, studies of varying durations in rodent models and a single study in cynomolgus
monkeys (Seacat et al., 2002) have reported reductions in endocrine hormone levels and changes
C-51
-------
APRIL 2024
in endocrine organ weights. There are insufficient data to support non-neoplastic lesions
(histopathology), and potential neoplastic lesions (see Toxicity Assessment, (U.S. EPA, 2024)).
Moreover, reductions were observed in thyroid hormone levels, including total and free
thyroxine (TT4 and FT4) and total and free triiodothyronine (TT3 and FT3) (NTP, 2019;
Luebker et al., 2005b; Lau et al., 2003), as well as reductions in adrenocorticotropic hormone
(ACTH), corticosterone, and/or corticotropin releasing hormone (CRH) (Salgado-Freiria et al.,
2018; Pereiro et al., 2014). Absolute and relative adrenal gland weights were reduced in rats
(NTP, 2019), however adrenal glands subject to histopathologic examination appeared normal
(Pereiro et al., 2014; Chang et al., 2009; Luebker et al., 2005b) (see Toxicity Assessment, (U.S.
EPA, 2024)).
C.2.2.1 Thyroid and Thyroid-Related Hormone Levels
Several 28-day studies provide evidence that exposure to PFOS can result in adverse effects on
rat thyroid hormone levels (Table C-3). Male and female rats were fed PFOS at doses of 0, 2, 20,
50, or 100 ppm (equivalent to 0, 0.14, 1.33, 3.21, or 6.34 mg/kg/day in males and 0, 0.15, 1.43,
3.73, or 7.58 mg/kg/day in females) for 28-days (Curran et al., 2008). In both males and females,
serum TT4 levels were significantly reduced at doses of >20 ppm. Serum TT3 was decreased at
the 100 ppm and >50 ppm dose groups in males and females, respectively (Curran et al., 2008).
In another study in rats, male and female Sprague-Dawley rats were exposed to PFOS at doses of
0 mg/kg/day, 0.312 mg/kg/day, 0.625 mg/kg/day, 1.25 mg/kg/day, 2.5 mg/kg/day, or
5 mg/kg/day via oral gavage (NTP, 2019). At study termination, TT4 and FT4 levels were
decreased in all male and female dose groups. In addition, TT3 was significantly decreased in
males and females treated with >0.625 mg/kg/day. No treatment-related effects were seen on
TSH levels (NTP, 2019). Yu et al. (2009a) exposed male Sprague-Dawley rats to 0 mg/L,
1.7 mg/L, 5.0 mg/L, or 15.0 mg/L-PFOS in drinking water for 91 days (drinking water
consumption was not reported). Significant dose-dependent reductions in TT4 were noted in
animals treated at > 1.7 mg/L; however, FT4 was only decreased in the 5.0 mg/L group. A
statistically significant increase in TT3 was observed in the 1.7 mg/L dose group, though TT3 in
the 5 mg/L and 15 mg/L groups returned to control levels. No treatment-related effects were seen
in TSH (Yu et al., 2009a).
A number of reproductive/developmental studies investigated the effect of PFOS on thyroid
hormone production in parental and Fi rodents (Table C-3).
Lau et al. (2003) analyzed thyroid hormones in offspring of pregnant rats exposed by gavage to
PFOS at 0 mg/kg/day, 1 mg/kg/day, 2 mg/kg/day, or 3 mg/kg/day from GD 2-GD 21. The
authors reported statistically significant reductions in TT4 and FT4 on PND 5 in rat pups treated
with 2 mg/kg/day and 3 mg/kg/day during gestation. Signs of recovery in TT4 were noted at
weaning, while reduced FT4 persisted through PND 35. No effects were noted in serum TT3 nor
TSH of pups when compared with controls (Lau et al., 2003). In a cross-fostering study
conducted by Yu et al. (2009b), pregnant Wistar rats were fed a diet containing 0 mg/kg/day or
3.2 mg/kg/day PFOS throughout gestation and/or lactation. PFOS-exposed groups consisted of
pups treated with PFOS during gestation only, pups treated with PFOS during lactation only, and
pups treated with PFOS during gestation and lactation. Pups in all exposure groups had
significant decreases in TT4 on PND 21 and PND 35. In contrast, TT3 and reverse T3 (rT3) were
not affected with PFOS exposure in rat pups (Yu et al., 2009b). Another study measured serum
TSH in pups and dams (GD 20, PND 4, or PND 21) following oral gavage exposure of pregnant
C-52
-------
APRIL 2024
Sprague-Dawley rats to PFOS (0 mg/kg/day, 0.1 mg/kg/day, 0.3 mg/kg/day, or 1.0 mg/kg/day)
from GD 0-PND 20. No statistically significant effects were observed in dams or offspring at
any timepoint assayed (Chang et al., 2009).
Luebker et al. (2005b) exposed pregnant Female Crl:CD®(SD)IGS VAF/Plus rats to
0.4 mg/kg/day, 0.8 mg/kg/day, 1.0 mg/kg/day, 1.2 mg/kg/day, 1.6 mg/kg/day, or 2.0 mg/kg/day
for 42 days prior to mating through LD 4. Exposed dams showed decreased TT4 and TT3 at
doses > 0.4 mg/kg/day and >1.2 mg/kg/day, respectively, although no perturbations were seen in
TSH or FT4 levels. In the pups, no perturbations were noted in TT3, FT4, or TSH, however, TT4
was reduced at doses ranging from 0.4 mg/kg/day to 1.6 mg/kg/day (2.0 mg/kg/day group not
assessed due to high pup mortality). The authors noted that the contributions of prenatal versus
postnatal effects of PFOS on thyroid hormones were not clear (Luebker et al., 2005b). The
authors also conducted follow-up analyses due to potential for negative bias from immeasurable
levels of FT3 and FT4 using equilibrium dialysis-radioimmunoassay (ED-RIA) methods and
measurements of TT3 and TT4 with chemiluminometric methods to ensure the validity of their
initial radioimmunoassay (RIA)-based results. While the ED-RIA reference method indicated
potential bias in the results for FT4 in pups, a true comparison could not be made due to
insufficient sample sizes (Luebker et al., 2005b). Conley et al. (2022b) also determined levels of
thyroid hormones in maternal serum following gestational exposure to PFOS. The authors
reported TT3 and TT4 on GD 18 in Sprague-Dawley rats exposed to PFOS at 0 mg/kg/day,
0.1 mg/kg/day, 0.3 mg/kg/day, 1 mg/kg/day, 3, 10, or 30 mg/kg/day from GD 14-GD 18. PFOS
significantly reduced TT3 and TT4 at 10 and 30 mg/kg/day. Non-significant decreases ranging
from 7% to 34% in TT3 and 3%-24% in TT4 were observed in dams exposed to doses below
10 mg/kg/day. Fuentes et al. (2006) examined the effects of PFOS on thyroid hormones in CD1
mice. The dams were exposed during gestation from GD 6-GD 18 to 0 mg/kg/day,
1.5 mg/kg/day, 3 mg/kg/day, or 6 mg/kg/day. At GD 18, the dams had an overall percent
reduction ranging from 11% to 57% in TT3, 36%-57% in FT3, and 42%-67% in FT4.
Conversely, increases in TT4 levels ranged from 158%> to 188%>. Nonetheless, the differences
between the exposed and control dams were not statistically significant due to high variability.
Only one study was included that investigated the effects of PFOS exposure on hormone levels
during development in mice. Lau et al. (2003) exposed pregnant CD-I mice to 0 mg/kg/day,
1 mg/kg/day, 5 mg/kg/day, 10 mg/kg/day, 15 mg/kg/day, or 20 mg/kg/day PFOS from GD 1-GD
17 and evaluated TT4 in sera of pooled mouse pups of each sex at several timepoints across
postnatal development. Because of mortality in the 15 mg/kg/day and 20 mg/kg/day groups, TT4
was only assessed in the 1 mg/kg/day, 5 mg/kg/day, and 10 mg/kg/day groups. TT4 levels varied
across the different time points with different trends based on treatment group. On PND 7, PND
14, and PND 28 there was a general trend for decreased TT4 in the 5 mg/kg/day and
10 mg/kg/day exposure groups when compared with control animals (Lau et al., 2003).
However, this was not observed on PND 3 or PND 21. Results were not significant at any time
point but may be limited by small sample size (3-7 determinations per group).
Male and female cynomolgus monkeys (4-6/sex/group) were orally exposed to PFOS at doses of
0 mg/kg/day, 0.03 mg/kg/day, 0.15 mg/kg/day, or 0.75 mg/kg/day for 182 days (Seacat et al.,
2002). Recovery animals from the 0 mg/kg/day, 0.15 mg/kg/day, and 0.75 mg/kg/day dose
groups were then monitored for an additional year. On the last day of dosing (day 182), thyroid
hormone levels, including TSH, TT3, and TT4 were evaluated. In males, TT3 was significantly
C-53
-------
APRIL 2024
reduced across all dose groups while TSH and TT4 remained unaffected. In females, significant
reductions in TT3 were noted in animals treated with 0.15 mg/kg/day and 0.75 mg/kg/day.
Significant reductions in TT4 were noted in the mid-dose group only (0.15 mg/kg/day). TSH
remained unaffected in females. Sixty-one days after cessation of treatment there was still a trend
for decreased TT3 in 0.15 mg/kg/day males and 0.75 mg/kg/day males and females. Because
there were only 2 animals per group at this time, statistical analyses were not appropriate. TT4
and TSH results were not reported in the recovery period (Seacat et al., 2002).
C-54
-------
APRIL 2024
Table C-3. Summary of Results for Thyroid and Thyroid-Related Hormones in Toxicological Studies Following Exposure to
PFOS
Study Name
Species
Study Design
Life Stage Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
Total Thyroxine (TT4)
Seacat et al.
Cynomolgus
Chronic (26 wk)
Adult M
0
4.38 ±0.61
NA
(2002)b
Monkey
0.03
4.72 ±0.68
7.8
0.15
3.99 ±0.62
-8.9
0.75
5.34 ± 1.57
21.9
F
0
5.66 ±0.89
NA
0.03
4.33 ± 1.46
-23.5
0.15
3.91 ±0.62*
-30.9
0.75
5.61 ± 1.00
-0.9
Fuentes et al.
CD-I Mice
Developmental
Po Adult (GD 18) F
0
0.50 ±0.13
NA
(2006)°
(GD 6-18)
1.5
1.29 ±0.59
158
3
1.41 ±0.39
182
6
1.44 ±0.57
188
Conley et al.
Sprague-Dawley
Developmental
Po Adult (GD 18) F
0
3.27 ±0.83
NA
(2022b)°
(GD 14-18)
0.1
2.49 ±0.43
-24
0.3
2.42 ±0.35
-26
1
3.18 ±0.95
-3
3
2.49 ±0.42
-24
10
1.67 ±0.47
-49
30
1.04 ±0.50
-68
Lau et al. (2003)c d
CD-I Mice
Developmental
Fi Pups (PND 28) M/F
0
4.2 ±0.9
NA
(GD 1-17)
1
3.8 ±0.5
-9.5
5
3.6 ±0.5
-14.3
10
3.5 ±0.3
-16.7
Curran et al.
Sprague-Dawley
Short-term (28 d)
Adult M
0
6.27 ±0.92
NA
(2008)b
Rats
0.14
5.19 ± 1.14
-17.3
1.33
1.11 ± 0.32*
-82.3
3.21
1.00 ±0.21*
-84.1
6.34
1.03 ±0.20*
-83.6
F
0
2.92 ± 1.19
NA
0.15
2.51 ±0.81
-14.1
1.43
1.52 ±0.19*
-48.0
C-55
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
3.73
1.17 ± 0.15*
-60.1
7.58
1.27 ±0.36*
-56.5
NTP (2019)°
Sprague-Dawley
Short-term (28 d)
Adult M
0
3.51 ±0.3
NA
Rats
0.312
1.33 ±0.19*
-62.1
0.625
0.53 ±0.09*
-84.9
1.25
0.26 ±0.07*
-92.6
2.5
0.22 ±0.04*
-93.7
5
0.48 ±0.07*
-86.3
F
0
2.21 ±0.24
NA
0.312
1.11 ± 0.12*
-49.8
0.625
0.55 ±0.07*
-75.1
1.25
0.33 ±0.07*
-85.1
2.5
0.35 ±0.09*
-84.2
5
0.38 ±0.05*
-82.8
Yu et al. (2009a)°
Sprague-Dawley
Subchronic (91 d)
Adult M
0
4.09 ±0.18
NA
Rats
0.0017
2.39 ± 0.13*
-41.6
0.005
1.64 ±0.54*
-59.9
0.015
0.85 ±0.16*
-79.2
Lau et al. (2003)c d
Sprague-Dawley
Developmental
F, Adult (PND 35) M/F
0
4.3 ±0.5
NA
Rats
(GD 2-21)
1
3 ±0.2
-30.2
2
2.5 ±0.2*
-41.9
3
2 ± 0.1*
-53.5
Luebker et al.
Crl:CD®(SD)IGS
Reproductive
Po Adult (LD 5) F
0.0
1.5 ±0.63
NA
(2005b)b
VAF/Plus® Rats
(80 d (42 d pre-
0.4
0.81 ±0.41*
-46.0
mating, GD 0-21,
0.8
0.6 ±0.44*
-60.0
LD 1-4))
1.0
0.73 ±0.24*
-51.3
1.2
0.28 ±0.32*
-81.3
1.6
0.27 ±0.17*
-82.0
2.0
0.24 ±0.15*
-84.0
Fi Pups M/F
0.0
0.54 ±0.22
NA
(PND 5)e
0.4
0±0
-100.0
0.8
0±0
-100.0
1.0
0.02 ±0.05
-96.3
1.2
0.01 ±0.02
-98.1
C-56
-------
APRIL 2024
Study Name Species
Study Design
Life Stage Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
1.6
0.01 ±0.0
-98.1
2.0
-
Fi Pups M/F
0.0
2.1 ±0.6
NA
(PND5)g
0.4
1.6 ±0.4
-23.8
0.8
1.5 ±0.7
-28.6
1.0
1.5 ±0.5
-28.6
1.2
-
-
1.6
-
-
2.0
-
-
Yu et al. (2009b)ch Wistar Rats
Reproductive
Fi Pups (PND 14) M/F
0
6.78 ±0.35
NA
(GD 0-PND 35)
3.2
6.36 ±0.25
-6.2
(Gestation Only)
3.2
5.97 ±0.39
-11.9
(Lactation Only)
3.2
4.29 ±0.17*
-36.7
(Gestation and Lactation)
Fi Pups (PND 21) M/F
0
5.81 ±0.31
NA
3.2
4.63 ±0.27*
7.9
(Gestation Only)
3.2
4.15 ±0.26*
-3.3
(Lactation Only)
3.2
4.38 ±0.24*
2.1
(Gestation and Lactation)
Fi Pups (PND 35) M/F
0
6.75 ±0.35
NA
3.2
5.44 ±0.33*
-19.4
(Gestation Only)
3.2
4.33 ±0.30*
-35.9
(Lactation Only)
3.2
4.23 ±0.22*
-37.3
(Gestation and Lactation)
Free Thyroxine (FT4)
NTP(2019)° Sprague-Dawley
Short-term (28 d)
Adult M
0
0.00253 ± 0.00022
NA
Rats
0.312
0.00095 ±0.0001*
-62.5
0.625
0.00047 ± 0.00005*
-81.4
1.25
0.0004 ± 0.00002*
-84.2
C-57
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage
Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
2.5
5
0.00036 ± 0.00005*
0.00033 ±0.00001*
-85.8
-87.0
F
0
0.312
0.625
1.25
2.5
5
0.00174 ± 0.00023
0.00107 ±0.00009*
0.0007 ±0.00003*
0.00064 ± 0.00005*
0.00056 ± 0.00005*
0.00048 ± 0.00003*
NA
-38.5
-59.8
-63.2
-67.8
-72.4
Yu et al. (2009a)°
Sprague-Dawley
Rats
Subchronic (91 d)
Adult
M
0
0.0017
0.005
0.015
1.9 ± 0.13
1.67 ±0.14
1.26 ±0.15*
1.73 ±0.11
NA
-12.1
-33.7
-8.9
Fuentes et al.
(2006)°
CD1 Mice
Developmental
(GD 6-18)
Po Adult (GD 18)
F
0
1.5
3
6
0.078 ±0.038
0.045 ±0.007
0.060 ±0.011
0.026 ±0.014
NA
-42%
-23%
-67%
Lau et al. (2003)c d
Sprague-Dawley
Rats
Developmental
(GD 2-21)
F, Adult (PND 35) M/F
0
1
2
3
0.02 ± 0.002
0.014 ±0.000
0.009 ±0.001
0.011 ±0.001
NA
-30.0
-55.0
-45.0
Luebker et al.
(2005b)b
Crl:CD®(SD)IGS
VAF/Plus® Rats
Reproductive
(80 d (42 d pre-
mating, GD 0-21,
LD 1-4))
P0 Adult
(LD 5)
F
0.0
0.4
0.8
1.0
1.2
1.6
2.0
0.00236 ±0.00061
0.00212 ±0.00058
0.00261 ±0.00056
0.00248 ± 0.00022
0.00259 ± 0.00082
NA
-10.2
10.6
5.1
9.7
Fi Pups
(PND 5)
M/F
0.0
0.4
0.8
1.0
1.2
1.6
2.0
0.0019 ±0.0009
0.0013 ±0.0004
NA
-31.6
C-58
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage Sex Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
Free Triiodothyronine (FT3)
Fuentes et al.
(2006)°
CD1 Mice
Developmental
(GD 6-18)
Po Adult (GD 18) F
0
1.5
3
6
0.014 ±0.003
0.009 ±0.001
0.006 ±0.001
0.008 ± 0.003
NA
-36
-57
-43
Luebker et al.
(2005b)b
Crl:CD®(SD)IGS
VAF/Plus® Rats
Reproductive
(80 d (42 d pre-
mating, GD 0-21,
LD 1-4))
Fi Pups M/F
(LD 5)
0.0
0.4
0.8
1.0
1.2
1.6
2.0
0.00019 ±0.00002
0.0002 ± 0.00003
0.000151
0.00018 ±0.00006
NA
5.3
-21.1
-5.3
Total Triiodothyronine (TT3)
Fuentes et al.
(2006)°
CD1 Mice
Developmental
(GD 6-18)
Po Adult (GD 18) F
0
1.5
3
6
0.105 ±0.034
0.045 ± 0.002
0.051 ±0.008
0.093 ±0.017
NA
-57
-51
-11
Conley et al.
(2022b)°
Sprague-Dawley
Developmental
(GD 14-18)
Po Adult (GD 18) F
0
0.1
0.3
1
3
10
30
0.106 ±0.013
0.082 ±0.016
0.070 ±0.001
0.099 ± 0.022
0.079 ±0.014
0.069 ±0.015*
0.040 ± 0.006*
NA
-23
-34
-7
-25
-35
-62
Seacat et al.
(2002)b
Cynomolgus
Monkey
Chronic (26 wk)
Adult M
0
0.03
0.15
0.75
0.16 ±0.007
0.119 ± 0.031*
0.125 ±0.015*
0.066 ± 0.027*
NA
-25.6
-21.9
-58.8
F
0
0.03
0.15
0.75
0.135 ±0.031
0.12 ±0.024
0.097 ±0.008*
0.085 ±0.012*
NA
-11.1
-28.1
-37.0
Curran et al.,
2008, 757871b
Sprague-Dawley
Rats
Short-term (28 d)
Adult M
0
0.14
1.33
10.39 ±2.14
11.75 ± 1.23
8.83 ± 1.69
NA
13.1
-15.0
C-59
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
3.21
8.38 ±8.38
-19.4
6.34
7.86 ± 1.49*
-24.4
F
0
11.88 ± 1.10
NA
0.15
11.17 ± 0.91
-6.0
1.43
11.36 ± 1.75
-4.4
3.73
9.15 ± 1.43*
-23.0
7.58
8.25 ± 1.30*
-30.6
NTP (2019)°
Sprague-Dawley
Short-term (28 d)
Adult M
0
0.08737 ±0.00532
NA
Rats
0.312
0.07781 ±0.00544
-10.9
0.625
0.06063 ± 0.00464*
-30.6
1.25
0.0575 ± 0.00267*
-34.2
2.5
0.05535 ±0.00275*
-36.6
5
0.051*
-42.8
F
0
0.09305 ± 0.00504
NA
0.312
0.0814 ±0.00302
-12.5
0.625
0.07252 ± 0.00427*
-22.1
1.25
0.0692 ±0.00363*
-25.6
2.5
0.06203 ±0.00178*
-33.3
5
0.05157 ±0.00143*
-44.6
Yu et al. (2009a)°
Sprague-Dawley
Subchronic (91 d)
Adult M
0
0.029 ± 0.004
NA
Rats
0.0017
0.048 ±0.008*
65.5
0.005
0.023 ± 0.005
-20.7
0.015
0.023 ± 0.003
-20.7
Lau et al. (2003)c d
Sprague-Dawley
Developmental
Fi Adult (PND 35) M/F
0
0.08 ±0.00
NA
Rats
(GD 2-21)
1
0.09 ±0.00
12.5
2
0.09 ±0.01
12.5
3
0.11 ±0.01
37.5
Luebker et al.
Crl:CD®(SD)IGS
Reproductive
Po Adult F
0.0
0.0760 ±0.0185
NA
(2005b)b
VAF/Plus® Rats
(80 d (42 d pre-
(LD 5)
0.4
0.0729 ±0.0135
-4.1
mating, GD 0-21,
0.8
0.0638 ± 0.00668
-16.1
LD 1-4))
1.0
0.0624 ±0.0132
-17.9
1.2
0.0529 ±0.015*
-30.4
1.6
0.0470 ± 0.020*
-38.2
2.0
0.0533 ±0.0173*
-29.9
C-60
-------
APRIL 2024
Study Name Species
Study Design
Life Stage
Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
Fi Pups
M/F
0.0
0.054 ±0.018
NA
(PND 5)e
0.4
0.056 ±0.019
3.7
0.8
0.049 ±0.018
-9.3
1.0
0.048 ± 0.009
-11.1
1.2
0.045 ± 0.022
-16.7
1.6
0.033 ±0.008
-38.9
2.0
0.033 ±0.012
-38.9
Fi Pups
M/F
0.0
0.0424 ± 0.0057
NA
(PND 5)s
0.4
0.0362 ± 0.0062
-14.6
0.8
0.031
-29.2
1.0
0.03 ±0*
-29.2
1.2
-
-
1.6
—
—
2.0
-
-
Yu et al. (2009b)ch Wistar Rats
Reproductive
Fi Pups
M/F
0
0.057 ± 0.004
NA
(GD 0-PND 35)
(PND 14)
3.2
0.052 ± 0.004
00
00
1
(Gestation Only)
3.2
0.051 ±0.003
-10.5
(Lactation Only)
3.2
0.043 ± 0.003
-24.6
(Gestation and Lactation)
Fi Pups
M/F
0
0.058 ±0.003
NA
(PND 21)
3.2
0.065 ± 0.007
12.1
(Gestation Only)
3.2
0.058 ±0.004
0.0
(Lactation Only)
3.2
0.059 ±0.003
1.7
(Gestation and Lactation)
Fi Pups
M/F
0
0.059 ±0.003
NA
(PND 35)
3.2
0.052 ±0.003
-11.9
(Gestation Only)
3.2
0.049 ± 0.004
-16.9
(Lactation Only)
3.2
0.055 ±0.002
-6.8
(Gestation and Lactation)
C-61
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage
Sex
Dose (mg/kg/day)
Value (jLtg/d L)" Percent Change
Reverse Triiodothyronine (rT3)
Yu et al. (2009b)ch Wistar Rats
Reproductive
(GD 0-PND 35)
Fi Pups M/F 0
(PND 14) 3.2
(Gestation Only)
3.2
(Lactation Only)
3.2
(Gestation and Lactation)
Fi Pups
M/F
0
0.0251
NA
(PND 21)
3.2
(Gestation Only)
0.025 ± 0.003
0.0
3.2
0.029 ±0.001
16.0
(Lactation Only)
3.2
0.025 ± 0.002
0.0
(Gestation and Lactation)
Fi Pups
M/F
0
0.02 ± 0.002
NA
(PND 35)
3.2
(Gestation Only)
0.02 ± 0.002
0.0
3.2
0.015 ±0.000
-25.0
(Lactation Only)
3.2
0.02 ±0.001
0.0
(Gestation and Lactation)
Thyroid Stimulating Hormone (TSH)
Seacat et al.
(2002)b
Cynomolgus
Monkey
Chronic (26 wk) Adult
M
0
0.03
0.15
0.75
0.43 ± 0.52J
0.34 ± 0.3J
0.74 ± 0.75J
0.93 ± 0.57J
0
0.03
0.15
0.75
0.73 ± 1.12J
0.68 ± 0.82J
1.27 ± 1.52J
0.84 ± 0.79J
NA
-20.9
72.1
116.3
NA
-6.8
74.0
15.1
NTP (2019)°
Sprague-Dawley
Rats
Short-term (28 d) Adult
M
0
0.312
0.625
1.25
2.039 ±0.14
1.494 ±0.174
1.479 ±0.12
2.333 ±0.294
NA
-26.7
-27.5
14.4
C-62
-------
APRIL 2024
Study Name Species Study Design Life Stage Sex Dose (mg/kg/day) Value (jLtg/dL)" Percent Change
25 2.419 ±0.338 18^6
5 1.890 ±0.239 -7.3
F 0 1.286 ±0.073 NA
0.312 1.476 ±0.088 14.8
0.625 1.276 ±0.085 -0.8
1.25 1.325 ±0.115 3.0
2.5 1.4914 ±0.195 16.0
5 1.536 ±0.073 19.4
Yu et al. (2009a)° Sprague-Dawley Subchronic (91 d) Adult M 0 0.072 ±0.030 NA
Rats 0.0017 0.067 ±0.027 -6.9
0.005 0.112 ±0.034 55.6
0.015 0.162 ±0.067 125.0
Chang etal. Sprague-Dawley Developmental Po Adult F 0 1.304 ±0.102 NA
(2009)c d Rats (GD 0-PND 20) (GD 20) 0.1 1.202 ±0.096 -7.8
0.3 1.061 ±0.058 -18.6
1 1.1 ±0.077 -15.6
Po Adult (PND 4) F 0 1.036 ±0.115 NA
0.1 1.119 ± 0.121 8.0
0.3 0.863 ±0.032 -19.3
1 1.023 ±0.083 -1.3
Po Adult F 0 1.714 ±0.205 NA
(PND 21) 0.1 1.758 ±0.166 2.6
0.3 1.483 ±0.128 -13.5
1 1.95 ±0.198 13.8
FiPups M 0 0.765 ±0.060 NA
(PND 21) 0.1 0.994 ±0.089 29.93
0.3 0.949 ±0.080 24.05
1 0.880 ±0.045 15.03
FiPups F 0 0.880 ±0.06 NA
(PND 21) 0.1 0.889 ±0.074 1.0
0.3 0.865 ±0.07 -1.7
1 0.840 ± 0.065 -4.5
FiPups M/F 0 1.212 ±0.134 NA
(GD 20) 0.1 1.053 ±0.08 -13.1
C-63
-------
APRIL 2024
Study Name
Species
Study Design
Life Stage
Sex
Dose (mg/kg/day)
Value (ng/dL)a
Percent Change
0.3
1
0.934 ±0.075
0.969 ±0.075
-22.9
-20.0
Fi Pups
(PND 4)
M/F
0
0.1
0.3
1
0.557 ±0.065
0.552 ±0.02
0.477 ± 0.07
0.542 ±0.06
NA
-0.9
-14.4
-2.7
Lau et al. (2003)c d
Sprague-Dawley
Rats
Developmental
(GD 2-21)
Fi Adult
(PND 35)
M/F
0
1
2
3
0.62 ±0.08
0.73 ±0.16
0.65 ± 0.06
0.29 ±0.02
NA
17.7
4.8
-53.2
Luebker et al.
(2005b)b
Crl:CD®(SD)IGS
VAF/Plus® Rats
Reproductive
(80 d (42 d pre-
mating, GD 0-21,
LD 1-4))
P0 Adult
(LD 5)
F
0.0
0.4
0.8
1.0
1.2
1.6
2.0
0.163 ±0.096
0.114 ±0.023
0.144 ±0.092
0.111 ±0.052
0.145 ±0.103
0.167 ±0.077
0.153 ±0.068
NA
-30.1
-11.7
-31.9
-11.0
2.5
-6.1
Fi Pups
(PND 5)
M/F
0.0
0.4
0.8
1.0
1.2
1.6
2.0
0.102 ±0.017
NA
0.2361
0.101 ±0.025
0.145 ±0.034*
0.151
131.4
-1.0
42.2
47.1
Notes: wk = weeks; F = female; Fi = first generation; GD = gestation day; LD = lactation day; M = male; NA = not applicable; Po = parental generation; PND = postnatal day;
TT4 = total thyroxine; FT4 = free thyroxine; FT3 = free triiodothyronine; TT3 = total triiodothyronine; rT3 = reverse Triiodothyronine; d = days; TSH = thyroid stimulating
hormone.
* Statistically significant atp < 0.05.
a Values were converted to (xg/dL for Seacat et al. (2002) (ng/dL TT3, FT3, FT4; uU/mL TSH); Curran et al. (2008) (nmol/L T4; nmol/L TT3); NTP (2019) (ng/dL FT4, ng/dL
TT3; ng/mL TSH); Yu et al. (2009a) (ng/L TT4; ng/L FT4; ng/L TT3; ng/L TSH); Lau et al. (2003) (ng/mL TT4; pg/mL FT4; ng/mL TT3; ng/mL TSH); Luebker et al. (2005b)
(ng/dL FT4; pg/mL FT3; ng/dL TT3; ng/mL TSH); Yu et al. (2009b) (ng/mL TT4; ng/mL TT3; ng/mL rT3); Chang et al. (2009) (ng/mL TSH); Conley et al. (2022b) (ng/mL
TT3, TT4); Fuentes et al. (2006) (ng/dL TT3, FT3, FT4).
b Data are presented as mean ± standard deviation.
c Data are presented as mean ± standard error.
d Values were estimated from a figure using a digital ruler.
e Analyzed by analog radioimmunoassay (RIA).
C-64
-------
fInsufficient sample for analysis.
g Analyzed by analog chemiluminometric assay (CL).
h Cross-foster study.
1 n = 1.
J Units in |xU/mL.
APRIL 2024
C-65
-------
APRIL 2024
C.2.2.2 Hypothalamic, Pituitary, and/or Adrenal Hormone Levels
Effects of PFOS exposure on hormones of the hypothalamus, pituitary gland, and adrenals were
available in two rat studies conducted by the same laboratory (Figure C-16). Salgado-Freiria et
al. (2018) and Pereiro et al. (2014) investigated the effect of PFOS exposure on hypothalamic
CRH, ACTH, and corticosterone of male Sprague-Dawley rats treated at 0 mg/kg/day,
0.5 mg/kg/day, 3.0 mg/kg/day, and 6.0 mg/kg/day for 28 days. Following exposure, decreases in
serum CRH and corticosterone concentrations in all dose groups were observed, but there was no
dose-related trend. However, a dose-dependent decrease in ACTH was observed. In a
reproductive/developmental study, pregnant Sprague-Dawley rats were administered
0 mg/kg/day, 5 mg/kg/day, and 20 mg/kg/day from GD 12-GD 18 via gavage (Li et al., 2016).
Fetal serum corticosterone levels were significantly increased in animals treated with
5 mg/kg/day and 20 mg/kg/day.
Three studies in mice have examined the effects of PFOS exposure on serum corticosterone
(Dong et al., 2011; Fuentes et al., 2007b; Fuentes et al., 2006). Fuentes et al. (2006) observed 1%
and 5% decreases at 1.5 mg/kg and 6 mg/kg respectively; and an 8% increase at 3 mg/kg
indicating there was no dose-related trend in pregnant CD1 mice. Dose-dependent increases of
approximately 20% and 50% were recorded in male CD1 mice following a 4-week exposure to 3
or 6 mg/kg/day PFOS (Fuentes et al., 2007b). In male C57BL/6 mice exposed to 0 mg/kg/day,
0.008 mg/kg/day, 0.017 mg/kg/day, 0.083 mg/kg/day, 0.417 mg/kg/day, or 0.833 mg/kg/day over
the course of 60 days, serum corticosterone decreased by 2%, 13%, and 17% at 0.008, 0.017, and
0.083 mg/kg/day (low doses) and increased by 2% and 19% at 0.417 mg/kg/day and 0.833 mg/kg
(high doses), indicating a biphasic dose-response trend (Dong et al., 2011). Although the changes
in serum corticosterone seem to be related to exposure, they were not statistically significant,
likely due to variability.
C-66
-------
APRIL 2024
PFOS Endocrine Effects-Ad renal H
Study Name Study Design Observation Time Animal Description Dose (mgfkgJday) | Statistically significant 0 Not statistically significant! 195% CI |
Adrenocorticotropic Hormone (ACTH) Salgado-Freiria et al., 2018. 5079767 !
Corticosterons
l (28d) 29d
/elopmental (GD6-18) GDia
Fuentesetal., 2007, 757865 short-term (4wk) 39d
DongctaL, 2011, 1424949 subchronic (60d) 60d
Jo-Freiria et al., 2018, 5079767 short-term (28d) 29d
Corticotropin Releasing Hormone (CRH) Salgado-Freiria et al., 2018, 5079767 s
^ (23d) 29d
Rat. Sprague-Dawley (;?. N=8) 0
Mouse, CD-1 (,, N=10-11) 0
Mouse, CD-1 (¦•>, N=10) 0
Mouse, C57BU6 (J. N=6) 0
0.008
0.017
0.Q83
0.417
0.833
Rat, Sprague-Dawley (-. N=8) 0
0.5
Rat, Sprague-Dawley N=8) 0
-80 -60 -40 -20 0 20 40 60
Percent control response i'-'ii)
00 120 140
Figure C-16. Percent Change in Adrenal Hormones Relative to Controls in Rodents
Following Exposure to PFOSa'b
Interactive figure and additional study details available on HAWC.
ACTH = adenocorticotropic hormone; CRH = corticotropin releasing hormone; CI = confidence interval.
aPereiro et al. (2014) reported on the same data as Salgado-Freiria et al. (2018) and is not shown in the figure.
bThe red dashed lines indicate a 100% increase from the control response.
C.2.2.3 Organ Weights
No adverse effects on male and female thyroid weights (Table C-4) were noted in the previously
mentioned NTP study (NTP, 2019). In a longer-term study conducted by Yu et al. (2009a), no
treatment-related effects were observed on absolute and relative thyroid weights in Sprague-
Dawley rats exposed to PFOS in drinking water at doses of 0 mg/L, 1.7 mg/L, 5.0 mg/L, or
15 mg/L for 91 days (Yu et al., 2009a). However, in Sprague-Dawley rats exposed to
2 mg/kg/day, 20 mg/kg/day, 50 mg/kg/day, or 100 mg/kg/day in the diet for 28 days, relative
thyroid weight was significantly increased in females and males in the highest dose group. No
treatment-related effects were observed on absolute thyroid weight or thyroid weight relative to
brain weight (Curran et al., 2008).
PFOS exposure was associated with changes in adrenal gland weighs in rats and non-human
primates (Table C-4). In Sprague-Dawley rats, absolute right adrenal gland weights in male rats
were reduced at doses > 1.25 mg/kg/day. No effects were observed in females (NTP, 2019). No
effects were observed in relative adrenal weights at any dose for either sex after 28 days of
exposure to 0 mg/kg/day-5 mg/kg/day via gavage (NTP, 2019). Additionally, relative adrenal
gland weight was decreased in male rats treated at doses of > 0.5 mg/kg/day for 28 days (Pereiro
et al., 2014). Curran et al. (2008) observed significant trends toward increased adrenal gland
weight relative to body weights and increased adrenal gland weight relative to brain weights in
male and female Sprague-Dawley rats exposed to 0 mg/kg/day, 2 mg/kg/day, 20 mg/kg/day,
50 mg/kg/day, or 100 mg/kg/day PFOS for 28 days. Seacat et al. (2002) measured absolute and
relative adrenal weights in male cynomolgus monkeys exposed to PFOS at doses of
C-67
-------
APRIL 2024
0 mg/kg/day, 0.03 mg/kg/day, 0.15 mg/kg/day, or 0.75 mg/kg/day for 182 days. The only
significant treatment-related effect was an increase in left adrenal-to-body weight percentages in
males of the high dose group (Seacat et al., 2002). No studies were available evaluating the
effect of PFOS exposure on mouse organ weights.
Effects on the relative weight of the hypothalamus were observed by Salgado et al. (2015) (see
Toxicity Assessment, (U.S. EPA, 2024)).
Table C-4. Associations Between PFOS Exposure and Endocrine Organ Weights in
Rodents and Non-human Primates
Endpoint
Study Name
Species
Exposure
Length
Dose (mg/kg/day) Sex
Change
Adrenal
NTP (2019)
Sprague-
28 d
0,0.312,0.625, 1.25, M
4 1.25-
Weight, Right,
Dawley rat
2.5, 5 mg/kg/day
5.0 mg/kg/da
Absolute
y
F
n.s.
Adrenal
Weight, Right,
Relative
NTP (2019)
Sprague-
Dawley rat
28 d
0,0.312,0.625, 1.25, M
2.5, 5 mg/kg/day
F
n.s.
n.s.
0,0.14,1.33,3.21, M n.s.
6.34 mg/kg/day
0,0.15,1.43,3.73, F |
7.58 mg/kg/day 1.43 mg/kg/
day
Adrenal Curran et al. (2008) Sprague- 28 d
Weight, Dawley rat
Relative to
Body Weight
Adrenal Curran et al. (2008) Sprague- 28 d
Weight, Dawley rat
Relative to
Brain Weight
Adrenal Pereiro et al. (2014) Sprague- 28 d 0,0.5,1,3, M J, 0.5-6
Weight, Dawley rat 6 mg/kg/day mg/kg/day
Relative
Adrenal Seacat et al. (2002) Cynomolgus 182 d 0,0.03,0.15, M f
Weight, Left, monkeys 0.75 mg/kg/day 0.75 mg/kg/
Relative to day
Body Weight F n.s.
Adrenal Seacat et al. (2002) Cynomolgus 182 d 0,0.03,0.15, M n.s.
Weight, Left, monkeys 0.75 mg/kg/day
Relative to ^ ~
Brain Weight
Adrenal Curran et al. (2008) Sprague- 28 d
Weight Dawley rat
Absolute
0,0.14,1.33,3.21, M n.s.
6.34 mg/kg/day
0,0.15,1.43,3.73, F |
7.58 mg/kg/day 3.73 mg/kg/
day
0,0.14,1.33,3.21, M n.s.
6.34 mg/kg/day
0,0.15,1.43,3.73, F |
7.58 mg/kg/day 1.43 mg/kg/
day
C-68
-------
APRIL 2024
Endpoint
Study Name
Species
Exposure
Length
Dose (mg/kg/day)
Sex
Change
Thyroid
Weight,
NTP (2019)
Sprague-
Dawley rat
28 d
0,0.312,0.625, 1.25, M
2.5. 5 me/ke/dav
n.s.
Absolute
F
n.s.
Curran et al. (2008)
Sprague-
Dawley rat
28 d
0, 2, 20, 50,
100 mg/kg/day
M
n.s
Yu et al. (2009a)
Sprague-
Dawley rat
91 d
0, 1.7, 5.0, or
15 mg/L
M
n.s.
Thyroid
Weight,
NTP (2019)
Sprague-
Dawley rat
28 d
0,0.312,0.625, 1.25, M
2.5. 5 me/ke/dav
n.s.
Relative to
Body Weight
F
n.s.
Curran et al. (2008)
Sprague-
Dawley rat
28 d
0, 2, 20, 50,
100 mg/kg/day
M
n.s.
Yu et al. (2009a)
Sprague-
Dawley rat
91 d
0, 1.7, 5.0, or
15 mg/L
M
n.s.
Seacat et al. (2002)
Cynomolgus
monkeys
182 d
0,0.03,0.15,
0.75 mg/kg/day
M
n.s.
F
n.s.
Thyroid
weight,
Relative to
Brain Weight
Curran et al. (2008)
Sprague-
Dawley rat
28 d
0, 2, 20, 50,
100 mg/kg/day
M
n.s.
Notes: F = female; M = male; n.s. = non-significant.
C.2.2.4 Histopathology
Few histological and morphometric abnormalities were observed in fetal and neonatal thyroid
glands in Sprague-Dawley rats that were orally administered PFOS at doses of 0 or 1 mg/kg/day
from GD 0-PND 20 (Chang et al., 2009). On GD 20, female fetuses had a significantly higher
number of thyroid follicular epithelial cells compared with controls (2.1-fold increase); the
number of follicular epithelial cells were not statistically different from controls in male fetuses.
No other treatment-related histologic changes in number of follicles present and the distribution
of follicle sizes were observed in fetuses at GD 20 or in neonates at PND 4 or PND 21 (Chang et
al., 2009). Luebker et al. (2005b) examined the thyroid gland of one male and female
Crl:CD®(SD)IGS VAF/Plus pup exposed to 2 mg/kg/day (highest dose group) PFOS through
LD4. No microscopic changes were noted (Luebker et al., 2005b).
Pereiro et al. (2014) examined the effect of oral PFOS exposure on the adrenal cortex of male
Sprague-Dawley rats treated with 0 mg/kg/day, 0.5 mg/kg/day, 1.0, 3.0 and 6.0 mg/kg/day for
28 days. Fasculata zona cells appeared more activated (presenting spongy cytoplasm due to the
presence of liposomes) in animals treated with PFOS when compared with control animals.
However, incidence data of non-neoplastic lesions and statistical analysis were not
reported/conducted (Pereiro et al., 2014). In contrast, NTP (2019) did not observe
histopathological changes in the thyroid, adrenal, or pituitary glands of male or female rats dosed
with up to 5 mg/kg/day PFOS for 28 days.
C-69
-------
APRIL 2024
In male and female cynomolgus monkeys orally exposed to PFOS at doses of 0 mg/kg/day,
0.03 mg/kg/day, 0.15 mg/kg/day, or 0.75 mg/kg/day for 182 days, no treatment-related effect on
cell proliferation of the pancreas was observed (Seacat et al., 2002).
C.2.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse endocrine outcomes is discussed in
Sections 3.2.5, 3.3.2, 3.3.6, and 3.4.1.5 of the 2016 PFOS HESD (U.S. EPA, 2016c). There are
29 studies from recent systematic literature search and review efforts conducted after publication
of the 2016 PFOS HESD that investigated the mechanisms of action of PFOS that lead to
endocrine effects. A summary of these studies is shown in Figure C-17. Additional mechanistic
synthesis will not be conducted since evidence suggests but is not sufficient to infer that PFOS
leads to endocrine effects.
Mechanistic Pathway
Animal
In Vitro Grand Total
Big Data, Non-Targeted Analysis
2
0
2
Cell Growth, Differentiation, Proliferation, Or Viability
3
12
15
Cell Signaling Or Signal Transduction
2
6
8
Extracellular Matrix Or Molecules
0
1
1
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
1
2
3
Hormone Function
8
13
20
Inflammation And Immune Response
0
1
1
Oxidative Stress
3
0
3
Xen obi otic Metabolism
1
1
2
Other
0
2
2
Not Applicabtej'Not Specified/Review Article
1
0
1
Grand Total
12
18
29
Figure C-17. Summary of Mechanistic Studies of PFOS and Endocrine Effects
Interactive figure and additional study details available on HAWC.
C.2.4 Evidence Integration
There is slight evidence for an association between PFOS exposure and endocrine effects in
humans based on studies reporting positive associations for TSH in children and adults. The
C-70
-------
APRIL 2024
2016 PFOS HESD (U.S. EPA, 2016c) included two studies reporting positive associations with
thyroid disease in NHANES participants. In this updated review, further evidence on the
relationship between PFOS and thyroid disease was limited to two studies, one of which reported
an inverse association in children (Kim et al., 2020a) and the other was classified as
uninformative. The most consistent effects were for TSH in children. Three medium confidence
studies (Itoh et al., 2019; Xiao et al., 2019; Kato et al., 2016) reported elevated TSH among
infants with increasing PFOS exposure, but other studies found the opposite effect (Aimuzi et al.,
2019). General population studies in adults also suggested a positive association between PFOS
exposure and TSH, but results were limited to one medium confidence study, while the rest were
low confidence. Interestingly, two general population studies identified seemingly sexually
dimorphic effects for TSH (Blake et al., 2018) and T3 (Byrne et al., 2018). The 2016 Health
Assessment included three studies reporting positive associations between serum PFOS and TSH
in pregnant women. In the recent literature, one high and one medium confidence study reported
positive association, while there was inconsistent evidence in low confidence studies. Additional
uncertainty exists due to the potential for confounding by other PFAS. One study (Aimuzi et al.,
2019) on infants reported correlations across PFAS (i.e., PFOA, PFNA, PFDA,
perfluoroundecanoic acid (PFUnDA), PFHxS, and PFDoA) and found them to be moderately
correlated (r = 0.37-82). Results for PFOS were not significant, however, the direction and
magnitude of effect were similar in single-pollutant and multipollutant models.
The animal evidence for an association between PFOS exposure and effects in the endocrine
system is considered moderate based on effects from 13 high or medium confidence studies.
Decreases in free T4, total T4, and total T3 were observed in rats, mice, and monkeys after PFOS
exposure; however, a compensatory increase in TSH was not reported, nor was there evidence of
thyroid gland histopathology, which is consistent with findings of hypothyroxinemia. Although
evidence of thyroid hormone disruption in humans is inconsistent, EPA concluded that the
sensitive and consistent changes in thyroid hormone levels in multiple animal models indicate
toxicity of relevance to humans.
Reductions in ACTH, corticosterone, and CRH in studies with animal models suggest that
exposure to PFOS may interfere with the hypothalamic-pituitary-adrenal axis. However, changes
in adrenal weights were inconsistent among studies and among species. More data on the
interactions between corticosterone and ACTH are required, as well as potential histological
effects in the adrenal gland, to understand the relevance of an effect of PFOS on adrenocortical
hormone levels.
C.2.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause endocrine effects in
humans under relevant exposure circumstances (Table C-5). This conclusion is based primarily
on evidence from animal models showing alterations in circulating thyroid and adrenocortical
hormone levels following exposure to doses as low as 0.03 mg/kg/day PFOS. Although a few
associations between PFOS exposure and TSH were observed in medium confidence
epidemiological studies, there is considerable uncertainty in the results due to inconsistency
across studies and limited number of studies.
C-71
-------
APRIL 2024
Table C-5. Evidence Profile Table for PFOS Endocrine Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.2.1)
Thyroid and thryoid-
related hormones and
thyroid disease
3 High confidence
studies
17 Medium confidence
studies
10 Low confidence
studies
In adults, findings
indicated significantly
increased levels of the
thyroid-related hormone
TSH (2/11); however,
one of the studies was of
low confidence. Findings
for thyroid hormones
(i.e., T3 and T4) were
generally inconsistent
across studies, and
considerable differences
were observed by sex
within studies. TSH was
significantly increased
among children in three
studies (3/19), including
a high confidence study.
However, other studies
reported inverse
associations for TSH,
including one significant
finding. Findings for free
T4 in children were
mixed, but significant
decreases (2/6) in T4 and
significant increases in
T3 (2/6) were reported.
Two studies in pregnant
women (2/3) reported
non-significant positive
associations for free T4
and free T3.
»High and medium
confidence studies
• Low confidence studies
• Inconsistent direction of
effect in adults which
may be influenced by
timing of outcome
sampling (i.e., diurnal
variations)
• Lmprecision offindings
©oo
Slight
Evidence for endocrine
effects is based on
increased TSH in adults,
decreased T4 in children,
and increased T3 in
children. Findings from
medium confidence
studies were frequently
inconsistent or imprecise.
There was limited
evidence reporting effects
on thyroid disease.
Uncertainties remain
regarding diurnal
variation of thyroid
hormones, differential
effects in males and
females, and consistency
across outcome timing.
©OO
Evidence Suggests
Primary basis:
Animal evidence
demonstrated alterations in
circulating thyroid and
adrenocortical hormone
levels. Although a few
associations between
PFOS exposure and TSH
were observed in medium
confidence
epidemiological studies,
there is considerable
uncertainty in the results
due to inconsistencies
across studies and the
limited number of studies.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
C-72
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Thyroid hormone
antibodies
2 Medium confidence
studies
Findings for thyroid
hormone antibodies were
generally imprecise,
however, hormone
antibody (i.e., TPOAb-
negative) status was
reported to play a role in
the association between
exposure and TSH levels
in male children.
• Medium confidence • Limited number of
studies
studies examining
outcome
• Imprecision of findings
Steroid and adrenal
hormones
1 High confidence study
One study reported
decreases in diurnal
urinary cortisone among
pregnant women, and the
diurnal urinary
cortisol/cortisone ratio
was correspondingly
increased.
»High confidence
study
• Limited number of
studies examining
outcome
Evidence From In Vivo Animal Studies (Section C.2.2)
Thyroid and thyroid-
related hormones
1 High confidence study
7 Medium confidence
studies
Reductions in total T4,
free T4, and/or total T3
was observed following
short-term and
developmental exposure
in male and female
rodents (5/7) and chronic
exposure in male and
female non-human
primates (1/1). No
significant change in
TSH levels was reported
in rats, mice, or non-
human primates (4/4).
»High and medium
confidence studies
» Coherent changes
across thyroid
hormone levels
> Consistent findings
across species, sex,
and study design
»Dose-response
relationship observed
for free T4, total T4,
and total T3
¦ Contributions of prenatal
versus postnatal
exposure to PFOS on
thyroid hormones
unclear
0©O
Moderate
Evidence was based on
high and medium
confidence studies that
demonstrated decreased
thyroid hormone levels
(free T4, total T4, total
T3). A compensatory
increase in TSH was not
reported, nor was there
evidence of thyroid gland
histopathology, which is
C-73
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Adrenocortical
hormones
5 Medium confidence
studies3
Mixed effects on
corticosterone levels were
observed in rodent
studies but most reported
no significant changes
(3/5). A dose-dependent
decrease in ACTH and a
non-monatomic decrease
in CRH were reported in
male rats (1/1).
»High and medium
confidence studies
• No factors noted
consistent with findings
of hypothyroxinemia.
Organ weights
1 High confidence study
3 Medium confidence
studies
In rodents, absolute (1/2)
and relative (1/3) adrenal
gland weights were
decreased in males while
absolute (1/2) and
relative (1/2) adrenal
gland weights were
increased in females
following a 28-day
exposure in rats. One
chronic study in non-
human primates reported
an increase in relative
adrenal weights in males
(1/1). No significant
changes were observed in
absolute or relative
thyroid gland weight
(4/4).
»High and medium
confidence studies
• Inconsistent direction of
effect in organ weights
across studies
• Limited number of
studies examining
outcomes
Histopathology
1 High confidence study
2 Medium confidence
studies
No significant effects
were observed in
incidence of non-
neoplastic lesions in the
thyroid gland, adrenal
gland, and/or pituitary
• High and medium
confidence studies
• Limited number of
studies examining
outcomes
C-74
-------
APRIL 2024
Studies and
Interpretation
Evidence Stream Summary and Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
gland following exposure
to male and female mice,
rats, and non-human
primates (3/3).
Notes: TSH = thyroid stimulating hormone; T3 = triiodothyronine; T4 = thyroxine; TPOAb = thyroid peroxidase antibody; ACTH = adrenocorticotropic hormone;
CRH = corticotropin releasing hormone.
aPereiro et al. (2014) reported on the same data as Salgado-Freiria et al. (2018) for adrenocortical hormone measurements.
C-75
-------
APRIL 2024
C.3 Metabolic/Systemic
EPA identified 69 epidemiological and 29 animal studies that investigated the association
between PFOS and systemic and metabolic effects. Of the epidemiological studies, 10 were
classified as high confidence, 36 as medium confidence, 14 as low confidence, 5 as mixed (4
medium/low and 1 medium/uninformative) confidence, and 4 were considered uninformative
(Section C.3.1). Of the animal studies, 3 were classified as high confidence, 20 as medium
confidence, 5 as low confidence, and 1 was considered mixed ( 6.5%, fasting plasma
glucose > 126 mg/dL, a 2-hour plasma glucose > 127 in an oral glucose tolerance test, or a
random plasma glucose > 200 mg/dL (in patients with classic symptoms of hyperglycemia or a
hyperglycemic crisis).
Metabolic syndrome is a combination of medical disorders and risk factors that increase the risk
of developing cardiovascular disease (CVD) and diabetes, including abnormalities in
triglycerides, waist circumference, blood pressure, cholesterol, and fasting glucose. It is highly
prevalent in the general population of the United States. Risk factors for metabolic syndrome
include insulin resistance and being overweight or obese.
The 2016 EPA Health Assessment for PFOS concluded that there is no evidence of an
association with metabolic syndrome. One study observed an association with gestational
diabetes (Zhang et al., 2015a), but no associations were observed with type 1 or type 2 diabetes.
Among adults, serum PFOS was significantly associated with increased beta-cell function.
Serum PFOS concentration was not associated with metabolic syndrome, glucose concentration,
homeostasis model of insulin resistance (HOMA-IR), or insulin levels in adults or adolescents
(Lin et al., 2009). Another study reported no association with metabolic syndrome or glucose
homeostasis parameters (Fisher et al., 2013). Overall, these studies show a lack of association of
PFOS with diabetes, metabolic syndrome, and related outcomes.
For this updated review, 69 new epidemiologic studies examined the association between PFOS
and metabolic outcomes. Of these, 32 were cohort studies, six were case-control studies, 27 were
cross-sectional studies, two were nested case-control studies, and two were controlled trials.
Most studies measured exposure to PFOS using biomarkers in blood. Di Nisio et al. (2019)
measured exposure to PFOS using biomarkers in blood and in semen) Shapiro et al. (2016)
C-76
-------
APRIL 2024
measured the exposure to PFOS in urine. Biomarkers in maternal blood were used in 16 studies
and cord blood was used in 2 studies. Most studies identified were conducted in the United
States and China. Other study locations included Canada, Croatia, Denmark (including the Faroe
Islands), France, Italy, Japan, Korea, Norway, Spain, Sweden, Taiwan, the Netherlands, and the
United Kingdom.
Twenty-two studies examined diabetes (one in children, nine in pregnant women), and four
examined metabolic syndrome in general adult populations. Other metabolic outcomes examined
included blood glucose levels or glucose tolerance, HbAlc, insulin or insulinogenic index,
insulin resistance, insulin sensitivity, adiponectin, leptin, beta-cell function, proinsulin, insulin-
like factor 1, c-peptide, body mass index (BMI) or ponderal index, body weight, gestational
weight gain, body fat, and anthropometric measurements (Appendix D).
C.3.1.2 Study Quality
Several criteria were specific to evaluating the quality of studies on metabolic outcomes.
Because of concerns for potential reverse causality (where the exposure may be affected by
disease status), studies evaluating diabetes were considered critically deficient if exposure and
prevalent diabetes were measured concurrently, since the cross-sectional design would not allow
for a reliable characterization of exposure before the onset of diabetes. Another concern is for the
evaluation of insulin, homeostasis model assessment of beta-cell function (HOMA-B), or
HOMA-IR without consideration of diabetes status, as the treatment of diabetes, particularly in
those being treated with hypoglycemic medications, influences insulin production and secretion.
There are 69 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and metabolic effects. Study quality evaluations for these 69 studies are shown in
Figure C-18, Figure C-19, and Figure C-20.
On the basis of the considerations mentioned, 10 studies were classified as high confidence, 36
as medium, 14 as low confidence, and 4 as uninformative for all metabolic outcomes. Five
studies have split ratings and were classified as medium confidence for one outcome and low
confidence for other outcomes). One study (Liu et al., 2018a) was considered uninformative for
insulin resistance and medium confidence for other metabolic outcomes. Uninformative studies
had critical deficiencies in at least one domain. These deficiencies included a lack of control for
confounding (Huang et al., 2018; Predieri et al., 2015; Jiang et al., 2014), lack of fasting
measures for glucose measurements (Jiang et al., 2014), and treating PFOS as an outcome
instead of an exposure, which limits the ability to make causal inference for the purpose of
hazard determination (Jain, 2020b; Predieri et al., 2015). Other concerns leading to an
uninformative rating included inadequate reporting of population selection (Jiang et al., 2014),
small sample size, and narrow ranges for exposure (Predieri et al., 2015).
The most common reason for a low confidence rating was potential for residual confounding,
particularly by SES (Fassler et al., 2019; Convertino et al., 2018; Heffernan et al., 2018; Khalil et
al., 2018; Koshy et al., 2017; Christensen et al., 2016a; Lin et al., 2013), by adiposity (Lin et al.,
2013), by age (Koshy et al., 2017), or by diabetes status (Lind et al., 2014). Low confidence
studies presented concerns with the outcome measures including potential for outcome
misclassification (He et al., 2018; Christensen et al., 2016a; Zong et al., 2016), failing to account
C-77
-------
APRIL 2024
for diabetes status (Lind et al., 2014) or use of medications that would impact insulin levels or
beta-cell function (He et al., 2018; Fleisch et al., 2017), analytical methods (Koshy et al., 2017),
and failure to establish temporality between PFOS exposure and diabetes (Lind et al., 2014).
Other concerns included selection bias (Fassler et al., 2019; van den Dungen et al., 2017), which
resulted from self-selection (Christensen et al., 2016a), failure to provide information on control
group selection (Heffernan et al., 2018), or differential recruitment for cases and controls (Lin et
al., 2013). Small sample size was also a concern in some studies (Heffernan et al., 2018; Khalil
et al., 2018; van den Dungen et al., 2017; Christensen et al., 2016a). In the evidence synthesis
below, high, and medium confidence studies were the focus, although low confidence studies
were still considered for consistency in the direction of association.
C-78
-------
APRIL 2024
Alderete et al.
Ashley-Martin et al.
Ashley-Martin et al.
Blake et al,
Legend
S3
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
r
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-18. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects
Interactive figure and additional study details available on IiAWC.
C-79
-------
APRIL 2024
He etal., 2018, 4238388
Huang et al., 2018. 5024212
Jain et al., 2019, 5080621
Jensen et al., 2018, 4354143
Jensen et al., 2020, 6833719¦
Jiang etal., 2014, 2850910
Kang et al., 2018, 4937567
Karlsen et al., 2017, 3858520
Koshy et al., 2017, 4238478
Lauritzen et al., 2018, 4217244
Lopez-Espinosa et al., 2016, 3859832
Mancini et al., 2018, 5079710 ¦
£
-*
+
-
+
+
~
+
-*
+
+
~
++
++
+
-
+
+
+
-
+
•1
+
++
+
+
+
++
++
D
+
+
+
+
-
+
+
+
+
+
J
-
+
D
+
-
+
J
+
++
+
+
El
+
+
+
+
++
1
'A
-
¦
+
-
-
+
+
+
+
+
+
+
+
+
+*
+
+
++
+
+
+
+
-
+
+
-
•
+
++
D
+
+
+
+
+
+
+
+
+
-
+
++
+
+
++
+
+
+
+
+
+
+
+
-
+
+
+*
~
+*
+
+
-
+
+
a
+
+
+
+
£
-
+
1
+
+
-
+
-
+
+
++
+
D
++
++
-
+
+
+
+
+
+
+
+
+
++
+
*
+
+
+
Legend
| Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
^ Critically deficient (metric) or Uninformative (overall)
* Multiple judgments exist
Figure C-19. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects (Continued)
Interactive figure and additional study details available on IiAWC.
C-80
-------
APRIL 2024
9^°V^O^O*#**5^ 0^
Manzano-Salgado et al.,
Marks et al.,
Martinsson et al..
Matilla-Santander et al.,
Minatoya et al.,
Mitro et al.,
2017, 4238509
2019, 5381534-
2020, 6311645-
2017, 4238432-
2017, 3981691
2020, 6833625
Mora et al.,2017, 3859823-
Predieri et al., 2015, 3889874
Preston et al., 2020, 6833657
Rahman et al.. 2019, 5024206
Ren et al., 2020, 6833646-
Shapiroet al., 2016, 3201206
Starling et al., 2017, 3858473
Suet al., 2016, 3860116
Sun et al.
Tian et al.,
Valvi et al.,
Wang et al.,
Wang et al.,
Xu et al.,
Yang et al.,
Zong et al.,
van den Dungen et al.,
2018, 4241053
2019, 5080586-
2017, 3983872
2018, 5079666-
2018, 5080352-
2020, 6833677
2018, 4238462
2016, 3350666
2017, 5080340
Legend
I Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
* Multiple judgments exist
Figure C-20. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Metabolic/Systemic Effects (Continued)
Interactive figure and additional study details available on IiAWC.
C-81
-------
APRIL 2024
C.3.1.3 Findings From Children and Adolescents
Three medium confidence studies and two low confidence studies evaluated glucose levels in
children, with mixed non-significant results. Two medium confidence studies (Kang et al., 2018;
Domazet et al., 2016) observed positive, non-significant associations with fasting blood glucose.
Negative, non-significant associations with fasting blood glucose were observed in three studies,
one of medium confidence (Alderete et al., 2019), and two of low confidence (Fassler et al.,
2019; Khalil et al., 2018). Alderete et al. (2019) also reported a positive, non-significant
association with 2-hour glucose (Alderete et al., 2019). (Appendix D).
Seven studies examined insulin measures, and two reported statistically significant associations.
Insulin resistance, as described by the HOMA-IR, was examined in five studies with mixed
results. Fleisch et al. (2017) observed a significant negative association with HOMA-IR in mid-
childhood in a study of female children. Five studies (two medium and three low confidence)
reported non-significant negative associations with HOMA-IR (Alderete et al., 2019; Fassler et
al., 2019; Khalil et al., 2018; Koshy et al., 2017; Domazet et al., 2016). In a medium confidence
study, a non-significant decrease in HOMA-IR at age 15 and 21 years per increase in PFOS
exposure from 9 years and a non-significant increase in HOMA-IR at 21 per increase in PFOS
measured at age 15 (Domazet et al., 2016).
Three studies examined fasting insulin levels. All three of these studies reported negative, non-
significant associations with fasting insulin (Fassler et al., 2019; Khalil et al., 2018; Domazet et
al., 2016).
A positive non-significant association was observed with insulin sensitivity, measured through
both the insulin sensitivity index and the Children's Health and Environmental Chemicals in
Korea (CHECK) Index/Quantitative Insulin Sensitivity Check Index (QUICKI) (Fassler et al.,
2019).
One medium confidence study of reported significant negative associations with insulin-like
growth factor 1 (IGF-1) in 6-9-year-old children in the C8 Health Project (Lopez-Espinosa et al.,
2016). Significant negative associations for both girls and boys persisted after stratification by
sex, and statistically significant decreasing trends across quartiles were also observed (Lopez-
Espinosa et al., 2016).
One medium confidence study examined HOMA-B. Negative, non-significant associations were
observed between PFOS levels at age 9 and beta-cell function at ages 15 or 21, but a positive
non-significant association was observed between PFOS levels at age 15 and beta-cell function
at age 21 (Domazet et al., 2016).
Two high and two medium confidence studies examined adiponectin and leptin, and one
observed significant association. For adiponectin, all studies observed positive associations. A
high confidence study on the Sapporo Cohort of the Hokkaido Study observed a statistically
significant positive association between maternal PFOS and cord blood adiponectin
(p-value = 0.028) (Minatoya et al., 2017). Three other studies (one high and two medium
confidence studies) reported positive, non-significant associations with adiponectin (Domazet et
al., 2020; Buck et al., 2018; Fleisch et al., 2017). Buck et al. (2018) observed a positive, non-
significant association between maternal PFOS and adiponectin, but a negative-non-significant
association between mid-childhood PFOS and adiponectin.
C-82
-------
APRIL 2024
Two medium and one high confidence study reported negative, non-significant association with
leptin (Domazet et al., 2020; Fleisch et al., 2017; Minatoya et al., 2017). Minatoya et al. (2017)
observed a negative association with leptin among male children and a positive association
among female children; the interaction between child sex and PFOS was statistically significant.
Another study observed a positive, non-significant association with PFOS; after stratification by
sex, a negative non-significant association with leptin was observed among males, but a positive
non-significant association was observed among females (Buck et al., 2018).
Six studies examined body fat measures, and one reported a significant negative association. A
medium confidence study from the Avon Longitudinal Study of Parents and Children (ALSPAC)
reported a statistically significant negative association between maternal PFOS and trunk fat
percentage in female children (Hartman et al., 2017). One study observed non-significant
negative associations with body fat percentage (Braun et al., 2016), and two studies observed a
non-significant negative association with body fat mass (Domazet et al., 2020; Jeddy et al.,
2018).
A high confidence study of 5-year-old children observed positive, non-significant associations
with body fat percentage and fat mass; after stratification by sex, the non-significant positive
associations persisted for boys, but non-significant negative associations with fat mass and body
fat percentage were observed among girls (Chen et al., 2019b). Another study of medium
confidence observed positive, non-significant associations with mid-childhood total fat mass
index, total fat-free mass index, and trunk fat mass index among children from Project Viva
(Mora et al., 2017).
Eleven studies examined BMI and related measures with mixed results. In the European Youth
Heart Study (EYHS) study, Domazet et al. (2016) observed a positive significant association
between PFOS at age 9 and BMI at age 15. Positive, but non-significant associations were
observed between PFOS measured at either age 9 or age 15 and BMI measured at age 21
(Domazet et al., 2016). Additionally, two medium confidence studies observed significant
positive associations with children's BMI (Lauritzen et al., 2018; Mora et al., 2017). Mora et al.
(2017) reported a positive, significant association between maternal PFOS and early childhood
BMI; the association was positive but not significant for the association with mid-childhood BMI
(Mora et al., 2017). After stratification by sex, the association with BMI remained positive
(though non-significant) for boys and girls in early childhood and for girls in mid-childhood but
was negative and non-significant for boys in mid-childhood (Mora et al., 2017).
Significant negative associations were observed between maternal serum PFOS levels and BMI
of girls from the ALSPAC study (Hartman et al., 2017) and between serum PFOS levels and
BMI of girls from the Breast Cancer and Environment Research Program (BCERP) study
(Fassler et al., 2019). Three studies (one of high confidence and two of low confidence) reported
negative, non-significant associations with BMI (Chen et al., 2019b; Khalil et al., 2018; Koshy et
al., 2017). In a sex-stratified analysis, Chen et al. (2019b) observed a negative, non-significant
association among girls, but a positive non-significant association among boys.
Di Nisio et al. (2019) reported no difference between BMI between Italian male high school
students exposed to PFOS pollution compared with those who were not exposed.
C-83
-------
APRIL 2024
A medium confidence study reported a significant negative association between serum PFOS
levels and ponderal index at birth in infants from the Hokkaido Study on Environment and
Children's Health (Kobayashi et al., 2017).
Seven studies evaluated BMI z-score, and two observed an association with PFOS. In a medium
confidence study of children from the Faroe Islands, a significant positive association was
observed between maternal PFOS and BMI z-score among 18-month old children (Karlsen et
al., 2017). In children from the POPUP study, Gyllenhammar et al. (2018b) observed a positive,
significant association with BMI z-score among children 4- and 5-years old; the association with
BMI z-score among 3-year-old children was positive, but not significant. Three other studies
(two medium and one high confidence) reported positive, non-significant associations with BMI
z-score (Jensen et al., 2020a; Manzano-Salgado et al., 2017b; Mora et al., 2017). In an age-
stratified analysis, Jensen et al. (2020a) observed a positive, non-significant association with
BMI z-score at birth, but a negative, non-significant association with BMI z-score at 3-months
and 18-months of age.
Two studies reported negative, non-significant associations with BMI z-score (Koshy et al.,
2017; Braun et al., 2016).
Seven studies evaluated the risk of being overweight or obese, and three reported significant
associations. A medium confidence study reported increased odds of being overweight at 4 years
old, with significantly increased odds of being overweight in the 4th quartile of maternal PFOS
exposure (Martinsson et al., 2020). Another medium confidence study observed significantly
increased odds of being overweight with increasing maternal PFOS among 5-year-old children
(Lauritzen et al., 2018). A medium confidence study of mother-child pairs in the Faroe Islands
reported a significantly increased risk of being overweight at 18 months (Karlsen et al., 2017).
Two medium confidence studies observed an increased, non-significant risk of being overweight
(Manzano-Salgado et al., 2017b; Mora et al., 2017). Manzano-Salgado et al. (2017b) observed an
increased, non-significant risk of being overweight at age 4, but a non-significant, decreased risk
of being overweight at age 7.
Two studies (one medium and one low confidence) reported non-significant, decreased risks of
being overweight or obese (Koshy et al., 2017; Braun et al., 2016). Braun et al. (2016) observed
a non-significant decreased risk of being overweight or obese in the second tertile of PFOS
exposure, but a non-significant increased risk of being overweight or obese in the third tertile of
PFOS exposure.
Six studies examined waist circumference, and two reported an association. A significant,
positive association was observed between PFOS exposure at age 9 and waist circumference at
age 15 and 21 years old; a positive, non-significant association was reported for PFOS exposure
at age 15 and waist circumference at age 21 (Domazet et al., 2016). Two studies, one high
confidence and one low confidence observed negative, non-significant associations with waist
circumference (Chen et al., 2019b; Mora et al., 2017). After stratification by sex, Mora et al.
(2017) observed negative, non-significant associations with waist circumference among boys,
and positive, non-significant associations with waist circumference among girls.
A medium confidence study of mother-daughter dyads reported a statistically significant negative
association with girls' waist circumference at age 9 (Hartman et al., 2017). In a tertiles analysis,
C-84
-------
APRIL 2024
Braun et al. (2016) observed a negative association with waist circumference in the second tertile
of PFOS exposure, but a positive association in the third tertile.
One low confidence study reported no statistical difference in waist circumference among PFOS-
exposed children compared with non-exposed children (Di Nisio et al., 2019).
Two studies assessed waist circumference z-score among children, and none reported an
association. Both studies observed negative, non-statistical associations with waist circumference
z-score (Jensen et al., 2020a; Manzano-Salgado et al., 2017b). Manzano-Salgado et al. (2017b)
observed a negative, non-significant association with waist circumference z-score at age 4 and a
null association at age 7; after stratification by sex, negative, non-significant associations were
observed for both boys and girls at age 7. In an age-stratified analysis, Jensen et al. (2020a)
reported a positive association with waist circumference z-score at birth, but a negative
association at 3-months and at 18-months.
Three studies evaluated waist-to-height ratio among children, and one observed a significant
association. A low confidence study reported a significant negative association was observed
with waist-to-height ratio among 6-8 year-old girls (Fassler et al., 2019).
A high confidence study of children from the Shanghai Prenatal Cohort observed negative, non-
significant associations with waist-to-height ratio (Chen et al., 2019b). In a medium confidence
study, a decreased risk of high waist-to-height ratio was observed at age 4, while an increased
risk of waist-to-height ratio was observed at age 7 (Manzano-Salgado et al., 2017b).
Two studies examined waist-to-hip ratio in children, with no significant associations reported. A
medium confidence study observed a positive, non-significant association with waist-to-hip ratio
(Fassler et al., 2019), while a null association was observed in a medium confidence study (Mora
et al., 2017). After stratification by sex, Mora et al. (2017) observed a positive, non-significant
association among girls, but a negative, non-significant association among boys.
Three studies examined skinfold thickness metrics, with two studies reporting significant
associations. A study from the EYHS reported significant positive associations between PFOS
measured at age 9 and skinfold thickness at age 15 and age 21; the association between PFOS at
age 15 and waist circumference at age 21 was positive, but not significant (Domazet et al., 2016).
Additionally, a significant positive association was observed with tricep skinfold thickness z-
score, while associations with subscapular skinfold thickness z-score were positive, but non-
significant (Lauritzen et al., 2018).
Mora et al. (2017) observed positive, non-significant associations with subscapular and tricep
skin thickness measures in mid- and early childhood. Negative, non-significant associations were
observed with the sum of subscapular and tricep skinfold thickness among all children in mid-
childhood, as well as with the subscapular-to-tricep skinfold thickness ratio among girls in early
childhood (Mora et al., 2017).
C.3.1.4 Findings From Pregnant Women
Ten studies examined diabetes or gestational diabetes and overall results were mixed, with no
significant associations (Appendix D).
C-85
-------
APRIL 2024
Positive, non-significant associations with gestational diabetes were reported in four studies
(Preston et al., 2020; Liu et al., 2019; Wang et al., 2018a; Matilla-Santander et al., 2017). A
medium confidences study observed an increased, non-significant risk of gestational diabetes
among women with a family history of type 2 diabetes and women were had an overweight pre-
pregnancy BMI; a decreased, non-significant risk of gestational diabetes was observed among all
women, women without a family history of type 2 diabetes, and with a normal pre-pregnancy
BMI (Rahman et al., 2019).
Four medium and one low confidence studies reported inverse, non-significant associations with
gestational diabetes (Xu et al., 2020b; Wang et al., 2018c; Valvi et al., 2017; Shapiro et al., 2016;
Zong et al., 2016). With the exception of the low confidence study (Zong et al., 2016),
gestational diabetes was determined through standard clinical methods. The nested case-control
study conducted by Xu et al. (2020b) recruited pregnant women with no history of diabetes and
reported inverse, non-significant odds of gestational diabetes across quartiles of PFOS exposure
and log-transformed PFOS exposure. Similarly, Shapiro et al. (2016) observed inverse, non-
significant odds of gestational diabetes or gestational impaired glucose tolerance, but increased
odds of gestational diabetes in the second quartile of PFOS exposure.
Fasting glucose was examined in six studies, and one reported a positive association. A medium
confidence study observed a significant increase in fasting glucose levels with increasing tertiles
of PFOS, but a negative association between PFOS analyzed continuously and fasting glucose
(Wang et al., 2018a). Two high confidence studies and one medium confidence study reported
negative, non-significant associations with fasting glucose (Liu et al., 2019; Jensen et al., 2018;
Starling et al., 2017). In contrast, two medium confidence studies reported positive, non-
significant associations with fasting glucose among pregnant women (Ren et al., 2020; Wang et
al.,2018c).
Results from oral glucose tolerance tests were assessed in five studies, two of which reported an
association. A high confidence study from Project Viva observed non-significant positive
associations with 1-hour glucose; a significant association with 1-hour glucose was observed in
the fourth quartile of PFOS exposure (Preston et al., 2020). Additionally, a medium confidence
study reported a significant association with 1-hour glucose levels among pregnant women in the
Shanghai-Minhang Birth Cohort (Ren et al., 2020). Three studies observed positive, non-
significant associations with oral glucose tolerance test results (Liu et al., 2019; Jensen et al.,
2018; Wang et al., 2018a).
Three studies examined impaired glucose tolerance among pregnant women. One low confidence
study reported positive, statistically significant effect estimates between plasma PFOS levels and
impaired glucose tolerance among pregnant women from the INMA birth cohort in Spain
(Matilla-Santander et al., 2017). A high confidence study and a medium confidence study both
reported positive, non-significant associations with impaired glucose tolerance in the second and
third quartiles of PFOS exposure, and a negative, non-significant association with impaired
glucose tolerance in the fourth quartile of PFOS exposure (Preston et al., 2020; Shapiro et al.,
2016).
Two high confidence studies evaluated associations between plasma PFOS levels and
hyperglycemia or HbAlc among members of Project Viva. Preston et al. (2020) reported a
positive, non-significant association with hyperglycemia. Conversely, Mitro et al. (2020)
C-86
-------
APRIL 2024
observed a negative, non-significant association with HbAlc; negative non-significant
associations persisted after stratification by maternal age.
Two studies, one of high and one of medium confidence observed positive, non-significant
associations with both fasting insulin and HOMA-IR in pregnant women (Jensen et al., 2018;
Wang et al., 2018c). These studies evaluated members of the OCC in Denmark with high risk of
gestational diabetes (Jensen et al., 2018) and women in China in early pregnancy (Wang et al.,
2018c). Jensen et al. (2018) reported a negative, non-significant association with insulin
sensitivity as reported by the Matsuda index.
One high confidence study of members of the OCC examined HOMA-B and levels of fasting c-
peptide among pregnant women with high risk of gestational diabetes and reported positive, non-
significant associations with both HOMA-B and fasting c-peptide (Jensen et al., 2018).
Two high confidence studies compared levels of PFOS and adiponectin or leptin among pregnant
women. One medium confidence study observed a negative, non-significant association with
adiponectin (Mitro et al., 2020) while another medium confidence study reported a positive, non-
significant association with adiponectin (Ashley-Martin et al., 2017). After stratification by age
during pregnancy, Mitro et al. (2020) reported a negative association with adiponectin among
women aged 35 and older, and a positive, non-significant association among women under 35.
Among the two medium confidence studies examining leptin, one reported a positive, non-
significant association (Mitro et al., 2020), while the other reported a negative, non-significant
association (Ashley-Martin et al., 2017).
Three medium confidence studies examined gestational weight gain, with mixed results.
Jaacks et al. (2016) observed a positive, non-significant association with gestational weight gain
among all mothers, and mothers with a BMI < 25, and a negative non-significant association in
mothers with a BMI > 25. Increased odds of excessive gestational weight gain and decreased
odds of inadequate weight gain were observed and were non-significant (Jaacks et al., 2016).
Ashley-Martin et al. (2016) used data from mother-infant pairs from the MIREC to estimate the
odds of having high cord blood PFOS (>0.39 ng/mL) per increase in gestational weight gain.
ORs were significant for both 1 kg increase in gestational weight gain and IQR increase in
gestational weight gain (Ashley-Martin et al., 2016).
Marks et al. (2019) observed a negative, non-significant association with gestational weight gain.
However, a significant interaction was observed between PFOS and pre-pregnancy BMI (Marks
etal., 2019).
One high confidence study reported a significant positive association with skinfold thickness, as
well as a non-significant positive association with waist circumference among pregnant women
from Project Viva (Mitro et al., 2020).
In a high confidence study, a positive non-significant association was observed between plasma
PFOS levels and BMI in pregnant women from the Project Viva study (Mitro et al., 2020).
C-87
-------
APRIL 2024
C.3.1.5 Findings From the General Adult Population
Eleven studies evaluated diabetes in the general population and four reported significant
associations with diabetes. A medium confidence study of Taiwanese adults aged 20-60 reported
a significant positive association with type 2 diabetes (Su et al., 2016). In a quartile analysis,
odds of type 2 diabetes significantly increased with increasing quartiles of PFOS (Su et al.,
2016). Another medium confidence study reported significantly increased odds of type 2 diabetes
in the second and third tertile of PFOS exposure among female nurses in the Nurses' Health
Study (NHS) II (Sun et al., 2018). A medium confidence study from the E3N cohort reported a
non-significant increased risk of type 2 diabetes in the 2nd-4th, 6th, 8th-9th deciles of PFOS
exposure, and a non-significant decreased risk of type 2 diabetes was observed in the 5th and
10th deciles of PFOS exposure (Mancini et al., 2018) (Appendix D).
Three low confidence studies reported non-significant positive associations with diabetes (He et
al., 2018; Christensen et al., 2016a; Lind et al., 2014) and prediabetes (Christensen et al., 2016a).
Significant decreased odds of type 1 and type 2 diabetes were observed among 6,889 participants
in the C8 Health Project (Conway et al., 2016). The decrease in odds of uncategorized diabetes
was not significant. After stratifying by age, significant decreased odds of type 1 diabetes were
observed among adults and children (Conway et al., 2016). One high confidence cohort study
from the Diabetes Prevention Program followed adults at increased risk of type 2 diabetes and
observed a decreased non-significant risk of diabetes (Cardenas et al., 2017). After stratification
by sex, a significant decreased risk of type 2 diabetes was observed among males, and the
decreased risk among females was not significant (Cardenas et al., 2017). Two other medium
confidence study reported non-significant negative associations with type 2 diabetes (Cardenas et
al., 2019; Donat-Vargas et al., 2019a).
Four studies (three medium confidence and one low confidence) evaluated metabolic syndrome
(MetS) and one study reported an association. In an adult population of the island of Hvar
(Croatia) Chen et al. (2019a) observed a positive non-significant association with risk of MetS as
defined by the Adult Treatment Panel III (ATP III) criteria (OR: 2.19; 95% CI: 0.88, 5.44). Two
medium confidence studies using overlapping data from NHANES reported non-significant
negative associations with MetS. Liu et al., 2018 observed adults aged 20 and older from the
2013-2014 NHANES cycle and Christensen et al. (2019) observed adults aged 18 and older
from 2007-2014 NHANES. In a model simultaneously adjusted for PFDE, PFOA, PFHxS, N-
methyl-PFOSA (MPAH), PFNA and PFUnDA, Christensen et al. (2019) reported non-
significant increased odds of MetS in the third and fourth quartiles of PFOS exposure; the
decreased odds observed in the second quartile of PFOS were not significant.
A low confidence study observed lower non-significant odds of MetS for participants with serum
PFOS >1.90 ng/mL compared with those with serum PFOS < 1.90 ng/mL (Yang et al., 2018).
However, concerns for selection bias, outcome misclassification, and residual confounding by
SES diminish confidence in the study results.
There were nine studies examining glucose. Three studies reported associations with fasting
blood glucose, one reported an association with 2-hour glucose, one reported an association with
glucose area under the curve (AUC).
C-88
-------
APRIL 2024
A medium confidence study of adults aged 19-87 years from China reported a significant
positive association with fasting blood glucose (Duan et al., 2020). Additionally, a study using
NHANES 1999-2014 data observed a significant positive correlation between fasting glucose
and serum PFOS (Huang et al., 2018). Su et al. (2016) reported a non-significant positive
association with fasting glucose; in a quartiles analysis, mean fasting blood glucose significantly
increased with increasing quartiles of PFOS. Liu et al. (2018b) reported a negative statistically
significant association with fasting blood glucose, but non-significant increased odds of fasting
glucose levels >100 mg/dL.
A low confidence study observed a positive, non-significant association with fasting blood
glucose (Heffernan et al., 2018), while another reported lower non-significant odds of blood
glucose >1.6 mmol/L for participants with serum n-PFOS >3 ng/mL compared with those with
serum n-PFOS < 3 ng/mL (Yang et al., 2018).
Two studies (one high confidence and one medium confidence) observed non-significant positive
associations with 2-hour glucose (Cardenas et al., 2017; Su et al., 2016) and 30-minute glucose
(Cardenas et al., 2017). Another medium confidence study reported a negative, non-significant
association with 2-hour glucose (Liu et al., 2018b).
One medium confidence study observed a significant decrease in glucose AUC with increasing
quartiles of PFOS and a non-significant negative association between PFOS (measured
continuously) and glucose AUC (Su et al., 2016). In the POUNDS Lost clinical trial, a positive,
non-significant correlation was observed between PFOS and glucose levels (Liu et al., 2018b).
Blood glucose levels were examined in a medium confidence study from NHANES (2007-
2014), which reported increased odds of high blood glucose in the second and third quartiles of
PFOS, and decreased odds in the fourth quartile of PFOS exposure (Christensen et al., 2019). A
low confidence study reported a negative association with blood glucose levels (van den Dungen
et al., 2017). None of the associations for these two studies reached statistical significance.
Significant associations were reported between resting metabolic rate and PFOS. The association
with resting metabolic rate was assessed in the POUNDS Lost trial, a clinical trial of overweight
and obese adults aged 30-70. A non-significant negative correlation between PFOS and resting
metabolic rate was observed (Liu et al., 2018a). In the first 6 months of the trial, resting
metabolic rate decreased non-significantly with increasing tertiles of PFOS exposure for the
entire study population, men, and women. The interaction between PFOS and sex were
significant (Liu et al., 2018a). In months 6-24 of the trial, a significant positive association was
observed with mean resting metabolic rate in all tertiles of PFOS exposure, and average resting
metabolic rate significantly decreased with increasing tertiles of PFOS (Liu et al., 2018a). In a
sex-stratified analysis, average resting metabolic rate significantly decreased with increasing
tertiles of PFOS among men and women (Liu et al., 2018a).
Twelve studies examined insulin resistance measures and one observed significant association
with fasting insulin, insulin resistance, fasting plasma insulin, 30-minute insulin, fasting
proinsulin, and insulin (corrected response), and one reporting associations with the ratio of
proinsulin to insulin.
C-89
-------
APRIL 2024
Four studies measured fasting insulin. One high confidence study used a subset of data on 954
adults at high risk of type 2 diabetes from the Diabetes Prevention Program and observed a
positive significant association between PFOS and fasting insulin (Cardenas et al., 2017). Two
low confidence reported non-significant positive associations with fasting insulin (Chen et al.,
2019a; Sun et al., 2018), and one reported a non-significant negative association (He et al.,
2018). One medium confidence study reported a positive, non-significant association with insulin
levels (Liu et al., 2018b).
Nine studies examined insulin resistance (measured as HOMA-IR), and one reported a
significant association. A high confidence study of 956 adults at high risk for type 2 diabetes in
the Diabetes Prevention Program reported a significant, positive association with HOMA-IR
(Cardenas et al., 2017). A medium confidence study of 1871 adults in NHANES observed a non-
significant positive association with HOMA-IR (Liu et al., 2018b). However, Donat-Vargas et al.
(2019a) reported a non-significant negative association with HOMA-IR in both continuous and
tertile analyses. In a sensitivity analysis, a non-significant negative association was observed
between HOMA-IR and the third tertile of baseline PFOS, and between HOMA-IR and PFOS
measured at the end of follow-up for both the second and third tertile of PFOS exposure. A non-
significant positive association with HOMA-IR was reported in the second tertile of baseline
PFOS exposure (Donat-Vargas et al., 2019a).
Four low confidence studies investigated the association between PFOS and insulin resistance.
Of these studies, two reported a positive, non-significant association with insulin resistance
(Chen et al., 2019a; Lind et al., 2014; Lin et al., 2013). In a sex-stratified tertile analysis, a non-
significant negative association was observed between PFOS and insulin resistance in both males
and females; among females, a significant negative association with insulin resistance was
observed in the third quartile of PFOS exposure (He et al., 2018). These studies were of low
confidence due to concerns with the statistical analysis (not accounting for design of NHANES)
(He et al., 2018), failure to account for diabetes status (Lind et al., 2014) or medications that
could affect insulin levels (Chen et al., 2019a), and concerns for residual confounding and
selection bias (Lin et al., 2013).
The association between plasma PFOS and insulinogenic index 1 was investigated in a high
confidence study from the Diabetes Prevention Program. A non-significant positive association
was observed with insulinogenic index among 945 adults at high risk for type 2 diabetes
(Cardenas et al., 2017).
In a high confidence study, Cardenas et al. (2017) reported significant positive associations
between PFOS and fasting plasma insulin, 30-minute insulin, and fasting proinsulin. A non-
significant positive association was observed with insulin (corrected response) (Cardenas et al.,
2017).
In a low confidence study, a non-significant positive association was reported for the ratio of
proinsulin to insulin and PFOS (Lind et al., 2014). This study was given a low confidence rating
due to failure to adjust for diabetes status in statistical analyses.
Four studies measured the association between PFOS and beta-cell function and two reported a
significant association. Cardenas et al. (2017) reported a significant positive association with
beta-cell function (measured as HOMA-B) in adults at high risk for type 2 diabetes from the
C-90
-------
APRIL 2024
Diabetes Prevention Program. Positive non-significant associations with HOMA-B were reported
in adults from NHANES (Liu et al., 2018b) and (Chen et al., 2019a). A medium confidence study
reported negative, non-significant associations with HOMA-B (Donat-Vargas et al., 2019a).
Four studies examined adiponectin, and none reported significant associations. Two high
confidence studies reported non-significant positive associations with adiponectin (Buck et al.,
2018; Ashley-Martin et al., 2017). In contrast, a non-significant negative association with
adiponectin was observed among 945 adults in the Diabetes Prevention Program (Cardenas et al.,
2017). A medium confidence study reported a negative non-significant correlation between
PFOS and plasma adiponectin (Sun et al., 2018).
Three studies examined associations with leptin. One study reported a significant association.
Two high quality studies measured associations with leptin; one reported a non-significant
positive association (Buck et al., 2018), and the other reported a non-significant negative
association (Ashley-Martin et al., 2017). A medium confidence study reported a positive, non-
significant correlation between plasma PFOS and leptin concentrations, and a non-significant,
positive correlation with soluble leptin receptors (Liu et al., 2018a).
Nine studies examined HbAlc, and three reported associations. A high confidence study on
participants in the Diabetes Prevention Program reported a significant positive association with
HbAlc (Cardenas et al., 2017). A significant positive association with HbAlc was also reported
among adults under age 55 in a medium confidence study of adults living in China; the
association with HbAlc among adults aged 55 and older was also positive, but not significant
(Duan et al., 2020). Two medium confidence studies observed positive correlations with HbAlc;
one was non-significant (Sun et al., 2018) and the other was significant (Huang et al., 2018).
Another medium confidence cross-sectional study assessed the association between plasma
PFOS and HbAlc in adults aged 20-60 (Su et al., 2016). A positive, non-significant association
between HbAlc and continuous PFOS was reported, and a significant increase in average HbAlc
was observed with increasing quartiles of PFOS (Su et al., 2016).
In the POUNDS Lost trial, a negative, non-significant correlation was observed between PFOS
and HbAlc (Liu et al., 2018a). Additionally, a medium confidence study of 1871 adults from
NHANES reported a non-significant negative association with HbAlc (Liu et al., 2018b).
One low confidence study reported a non-significant negative association with HbAlc
(Heffernan et al., 2018). Another low confidence study observed a non-significant positive
association between PFOS and HbAlc (Chen et al., 2019a). Concerns with measurement of
confounders and inclusion of medications that could affect insulin levels (Chen et al., 2019a), as
well as concerns with case selection and residual confounding (Heffernan et al., 2018) resulted in
low confidence ratings.
There were four studies evaluating body weight measures. Associations were observed in one
study of body weight, and two studies reported associations with being overweight or obese.
One study, from the POUNDS Lost clinical trial, evaluated body weight and observed a
negative, non-significant association with weight loss in the first 6 months of the trial, and a
positive, significant association with weight loss in months 6-24 of the trial (Liu et al., 2018a). A
C-91
-------
APRIL 2024
significant increase in average weight gain during months 6-24 of the trial was observed with
increasing tertiles of PFOS (Liu et al., 2018a).
Two studies evaluated being overweight, one of which reported an association. A medium
confidence study reported significantly greater serum PFOS among obese adults compared with
non-obese adults (Jain and Ducatman, 2019e). One medium confidence study evaluated maternal
PFOS and risk of being overweight or obese in their children; this study reported increased, non-
significant odds of being overweight at age 4 in the second and third quartiles of PFOS exposure,
and significant increased odds of being overweight at age 4 in the fourth quartile (Martinsson et
al., 2020).
One low confidence study observed significant increased odds of being overweight or obese
(Tian et al., 2019c). Another low confidence study reported non-significant negative associations
with being overweight and obese (Yang et al., 2018).
Five studies evaluated body fat measures, and one reported an association Four studies of
medium confidence evaluated body fat. A significant negative association was observed between
maternal plasma PFOS and trunk fat in young girls ALSPAC. After stratification by age at
menarche, the association remained negative but was not significant in either age group
(Hartman et al., 2017). A negative, non-significant association was observed between maternal
plasma PFOS and body fat percentage (Hartman et al., 2017).
Three medium confidence studies reported positive, non-significant associations with body fat
measures (Liu et al., 2019; Mora et al., 2017; Braun et al., 2016).
Two medium confidence studies evaluated fat mass; one reported a non-significant negative
association with fat mass among children (Jeddy et al., 2018) and a non-significant positive
association with fat mass among overweight and obese adults (Liu et al., 2019).
Eleven studies assessed BMI; one significant association was reported for BMI, and one
significant association was reported for BMI z-score.
In the Health Outcome Measures of the Environment (HOME) study, a cohort study of 285
mother-child pairs, PFOS exposure was measured during pregnancy and BMI was recorded at
age 8 (Braun et al., 2016). Negative, non-significant associations with BMI z-score were
observed in the second and third tertile of maternal PFOS exposure (Braun et al., 2016). Liu et
al. (2018a) reported a non-significant negative correlation between PFOS and BMI.
One high confidence study and two medium confidence studies observed positive, non-
significant associations with BMI (Chen et al., 2019a; Blake et al., 2018; Cardenas et al., 2017).
In a medium confidence cohort study from the ALSPAC, a significant negative association with
children's BMI was observed among 312 mother-child pairs (Hartman et al., 2017). Another
medium confidence study reported non-significant positive association with BMI; in a sex-
stratified analysis, a non-significant percent decrease was observed for males, and a non-
significant percent increase was observed among females (Blake et al., 2018). In the single low
confidence study, Tian et al. (2019c) reported a non-significant association with BMI. In a sex-
stratified analysis, a non-significant negative association was observed among men and a
positive, non-significant association was reported for women. (Tian et al., 2019c). This study
C-92
-------
APRIL 2024
was given a low confidence designation due to concerns for PFOS to be potentially related to
BMI.
A high confidence study measured PFOS in maternal serum and BMI z-score in children. Non-
significant negative associations with BMI z-score were observed in children at 3- and 18-
months, and a non-significant positive association with BMI z-score was observed at birth.
(Jensen et al., 2020a) A medium confidence study of 412 mother-child pairs observed a positive,
significant association between maternal serum PFOS and 5-year-old child's BMI z-score
(Lauritzen et al., 2018).
Five studies examined waist circumference. Two single medium confidence studies observed a
negative, non-significant association with waist circumference (Liu et al., 2018b; Liu et al.,
2018a). One low confidence study reported a non-significant positive association with waist
circumference (Tian et al., 2019c). Non-significant decreased odds of increased waist
circumference were observed among men, and non-significant increased odds were observed for
women; the interaction between PFOS and sex was significant but was not significant in
continuous analyses (Tian et al., 2019c). In another low confidence study, non-significant
increased odds of increased waist circumference were observed with increasing quartiles for
PFOS; these estimates were adjusted for multiple PFAS (Christensen et al., 2019).
C.3.1.6 Findings From Occupational Studies
No occupational studies examined metabolic outcomes and PFOS.
C3.2 Animal Evidence Study Quality Evaluation and Synthesis
C3.2.1 Metabolic Homeostasis
There are three studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and four studies from
recent systematic literature search and review efforts conducted after publication of the 2016
PFOS HESD that investigated the association between PFOS and metabolic effects. Study
quality evaluations for these four studies are shown in Figure C-21.
C-93
-------
APRIL 2024
Conley et al., 2022, 10176381 -
+
+
NR
++
+
+
~
++
+
Curran et al., 2008, 757871 -
++
NR
NR
++
+
+
++
++
+
Lai etal., 2018, 5080641 -
+
+
NR
++
-
+
+
+
+
+
Luebker et al., 2005, 757857 -
+
+
NR
+
-
-
++
+
+
+
NTP, 2019, 5400978-
Seacat etal., 2003, 1290852-
Thomford, 2002, 5432419-
Legend
3
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
D
Critically deficient (metric) or Uninformative (overall)
NR
Not reported
Figure C-21. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Metabolic Effects
Interactive figure and additional study details available on HAWC.
PFOS has been observed to cause perturbations in glucose homeostasis in rodents. Several
studies in adult and perinatal rats and mice investigate glucose homeostasis, including serum
glucose levels, glucose tolerance, and gluconeogenesis, among other measures. Alterations in
these metabolic endpoints were observed, but the data is inconclusive as there are inconsistencies
within the literature with too few studies to assess possible difference across life stages, sexes,
and species.
NTP (2019) reported no statistical differences in serum glucose in adult male and female
Sprague-Dawley rats exposed to PFOS doses up to 5 mg/kg/day for 28 days. In contrast, Seacat
et al. (2003) observed a significant decrease in serum glucose in adult male Sprague-Dawley rats
compared with controls following 1.51 mg/kg/day PFOS exposure in the diet for 4 weeks. No
statistically significant change was seen in females at the 4-week interim timepoint. After
14 weeks, serum glucose concentrations were no longer statistically different in males from any
treatment group. In females at 14 weeks, serum glucose was significantly lower in the
0.40 mg/kg/day group, but not in the high dose group (1.56 mg/kg/day).
C-94
-------
APRIL 2024
In a rat reproductive toxicity study, Luebker et al. (2005b) noted significantly higher serum
glucose levels on lactational day (LD) 5 in dams treated with 2 mg/kg/day PFOS for 42 days
prior to mating until LD 4. This change was not seen in dams sacrificed at GD 21. Serum glucose
levels were not significantly altered in fetuses at GD 21 or in pups at LD 5. In a glucose
tolerance test, Lv et al. (2013) observed a dose-related increase in serum glucose 10 weeks
postweaning in rats perinatally exposed to PFOS from GD 0-PND 20 with significance in the
high dose exposure group of 1.5 mg/kg/day. At 15 weeks postweaning, only the low dose
(0.5 mg/kg/day) group had significantly elevated serum glucose during the glucose tolerance test.
Elevated serum glucose in this test indicates decreased glucose clearance or tolerance. In
addition, at 18 weeks postweaning, rats in the high dose group had elevated serum insulin, higher
insulin resistance indices, increased leptin levels, and decreased adiponectin levels, all of which
indicate dysregulation of glucose homeostasis and insulin resistance, potential signs of
prediabetes (Lv et al., 2013).
Wan et al. (2014) exposed CD-I mouse dams to 0 mg/kg/day, 0.3 mg/kg/day, or 3 mg/kg/day
PFOS from GD 3-PND 21. Offspring were then fed either a standard or high-fat diet from PND
21-PND 63. At PND 21, no statistical difference was detected in the fasting serum glucose or
insulin levels in dams. However, the Homeostatic Model Assessment for Insulin Resistance
(HOMA-IR) index was significantly increased in both the 0.3 and 3 mg/kg/day dose groups.
Increases in this metric indicate increased risk of insulin resistance, hypertension, and type 2
diabetes (Wan et al., 2014). There was no significant difference in fasting serum glucose or the
HOMA-IR index in male or female pups at PND 21, though males from both the 0.3 mg/kg/day
and 3 mg/kg/day groups had significantly increased fasting serum insulin levels. No difference
was found in fasting serum insulin levels in female pups at PND 21. In pups fed a standard diet,
at PND 63, fasting serum glucose levels were significantly higher for males and females at both
PFOS doses. Serum insulin and HOMA-IR were significantly increased only at the high dose of
3 mg/kg/day PFOS in both sexes. No significant differences between treatment groups in glucose
tolerance were observed in either sex. In the high-fat diet group, fasting serum insulin was
increased at PND 63 in the 3 mg/kg/day PFOS group of both sexes. Fasting serum glucose was
significantly higher in females dosed with both 0.3 and 3 mg/kg/day, but only for the
3 mg/kg/day males. In the glucose tolerance test, serum glucose was significantly higher only in
the high dose group in both sexes, indicating decreased glucose tolerance in these animals. The
HOMA-IR index in each sex was elevated in the high dose groups compared with the high-fat
diet control group. However, the HOMA-IR indices were significantly higher for the high-fat
diet groups compared with the standard diet groups within a specific PFOS treatment group and
sex. In contrast, Ngo et al. (2014) did not observe significant changes in blood glucose at PNW
6, PNW 11, or PNW 20 in wild-type or tumorigenic transgenic C57BL/6J-/1//// mice offspring
gestationally exposed to 0 mg/kg/day, 0.01 mg/kg/day, 0.1 mg/kg/day, or 3 mg/kg/day PFOS
from GD 1-GD 18, though it should be noted that the animals were not fasted prior to serum
sample collection.
Lai et al. (2018) exposed CD-I female mice to 0, 0.3, or 3 mg/kg/day for 7 weeks with
conflicting results. The authors conducted an oral glucose tolerance test and an intraperitoneal
insulin tolerance test. In both tests, blood glucose levels were significantly lower in the
3 mg/kg/day dose group compared with controls, potentially indicating increased glucose
tolerance and reduced insulin resistance, respectively. Pyruvate tolerance was also significantly
C-95
-------
APRIL 2024
decreased in both the 0.3 mg/kg/day and 3 mg/kg/day dose groups which could indicate reduced
gluconeogenesis.
C3.2.2 Survival, Clinical Observations, Body Weight, and Food
Consumption
There are 6 studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and 21 studies from recent
systematic literature search and review efforts conducted after publication of the 2016 PFOS
HESD that investigated the association between PFOS and systemic effects. Study quality
evaluations for these 27 studies are shown in Figure C-22 and Figure C-23.
C-96
-------
APRIL 2024
Alam et al., 2021, 9959508
Butenhoff etal., 2009, 757873
Curran et al., 2008, 757871
Dong et al., 2011, 1424949
Han etal., 2018, 4238554
Han etal., 2018, 4355066
Kawamoto et al., 2011, 2919266
Lefebvre et al., 2008, 1276155
Li etal., 2021, 7643501
Luebker et al., 2005, 1276160
Luebker et al., 2005, 757857
Lv et al., 2015, 3981558
NTP, 2019, 5400978
++
++
++
++
++
++
Qiu etal., 2016, 3981408-
+
+
NR
+
+
+
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
B Critically deficient (metric) or Uninformative (overall)
Not reported
* Multiple judgments exist
Figure C-22. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Systemic Effects3
Interactive figure and additional study details available on HAWC.
a Lefebvre et al. (2008) reported on the same animals as Curran et al. (2008).
C-97
-------
APRIL 2024
Qiu et al., 2020, 7276729-
i.
NR
i
NR
++
I
+
I
+
++
++
+
Qu et al., 2016, 3981454-
+
+
NR
++
+
+
++
++
+
Salgado et al., 2015, 3981583 -
+
+
NR
+
+
-
+
+*
-
-
Salgado et al., 2016, 3179088-
+
+
NR
++
+
+
++
n
n
+
Seacat et al., 2002, 757853 -
++
NR
+
+
+
++
++
++
+
Seacat et al., 2003, 1290852 -
++
NR
+
+
+
++
D
B
+
Thomford, 2002, 5432419-
+
-
NR
++
++
-
++
++
++
-
Wan et al., 2016, 3981504-
+
+
NR
++
-
+
¦
B
++
+
Xing et al., 2016, 3981506-
++
+
NR
++
++
++
++
+
+
Yan et al., 2014, 2850901 -
++
+
NR
++
¦
a
++
++
Yang etal., 2021, 7643494 -
++
+
NR
++
++
++
++
-
-
Zhang et al., 2019, 5918673 -
-
+
NR
+
+
-
+
-
-
Zhong etal., 2016, 3748828-
-
NR
NR
+
+
+
+
++
+
Legend
0
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
"
Deficient (metric) or Low confidence (overall)
b
Critically deficient (metric) or Uninformative (overall)
F
Not reported
*
Multiple judgments exist
Figure C-23. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Systemic Effects (Continued)3
Interactive figure and additional study details available on HAWC.
a Lefebvre et al. (2008) reported on the same animals as Curran et al. (2008).
A number of subchronic, chronic, and developmental studies suggest that PFOS exposure can
induce whole-body toxicity, which can manifest as decreased body weight, partly due to a
reduction in food consumption. These changes were more prominent following high exposures to
C-98
-------
APRIL 2024
PFOS. Although one study in non-human primates suggests PFOS-related mortality, PFOS-
induced mortality and clinical observations were not supported by rodent studies.
C.3.2.2.1 Mortality and Clinical Observations
PFOS-related mortality was observed in 2 of 6 male cynomolgus monkeys administered
0.75 mg/kg/day PFOS for 26 weeks. Pulmonary inflammation was identified as the probable
cause of death of one monkey that died on day 155 of dosing, and hyperkalemia was suggested
for the other monkey that died on day 179 (Seacat et al., 2002). Mortality was not affected in
female monkeys administered 0.75 mg/kg/day PFOS or male or female monkeys receiving
0.03 mg/kg/day or 0.15 mg/kg/day PFOS (Seacat et al., 2002).
Rodent studies did not observe mortality with doses up to 10 mg/kg/day and durations up to
60 days. No mortality was observed in C57 male mice exposed to 0.5 mg/kg/day or
10 mg/kg/day PFOS for 5 weeks, but the study did not report if there were any overt clinical
observations (Qu et al., 2016). NTP (2019) exposed male and female Sprague-Dawley rats to
0.312-5 mg/kg/day PFOS for 28 days. All rats survived to the end of the study, except for one
female Sprague-Dawley rat administered 5 mg/kg/day (NTP, 2019). There were no treatment-
related clinical observations reported in male or female rats (NTP, 2019). Similarly, Alam et al.
(2021) reported that there was no mortality in male Wistar rats over the course of a 60-day study
exposure to 0, 0.015, or 0.15 mg/kg/day PFOS. Xing et al. (2016) did not observe an effect on
mortality in C57BL/6J male mice exposed to PFOS at 2.5 mg/kg/day, 5 mg/kg/day, or
10 mg/kg/day for 30 days. Clinical observations such as rough hair, slow movement, and
constipation were reported, although neither the exposure group associated with these effects nor
incidence were specified (Xing et al., 2016). Study authors indicated that there were no
treatment-related clinical signs or mortality in Po male Crl:Cd(Sd)lgs rats following 6 weeks of
pre-mating exposure to 1.6 mg/kg/day, 2.0 mg/kg/day, or 3.2 mg/kg/day (Luebker et al., 2005a).
No mortality was observed in the Po females, but timing of the clinical observations
(i.e., localized areas of partial alopecia) were not specified when they occurred (Luebker et al.,
2005b; Luebker et al., 2005a) (see Toxicity Assessment, (U.S. EPA, 2024)).
C3.2.2.2 Body Weight in Adults
Many studies with rodent models report reductions in body weight following short-term to
subchronic PFOS exposure (Figure C-24). A dose-dependent reduction in body weight change
was observed in C57BL/6J male mice exposed to PFOS at 2.5 mg/kg/day, 5 mg/kg/day, or
10 mg/kg/day via gavage for 30 days (Xing et al., 2016). All dose groups had a significant
difference in body weight gain when compared with the control with the 10 mg/kg/day group
having a 31% reduction in body weight over the study period compared with a 27.75% weight
gain in the controls. This reduction may be attributed to reduced food consumption reported
across all doses, but the correlation between body weight and food intake was not significant in
the treatment groups suggesting that this may not be the only explanation (Xing et al., 2016).
Body weight was significantly changed in the highest dose group (50 mg/kg total administered
dose equivalent to 0.833 mg/kg/day) in a 60 day study in C57BL/6 mice; this reduction may be
attributed to reduced food consumption reported in this group (Dong et al., 2011). C57 male
mice exposed to 0 mg/kg/day, 0.5 mg/kg/day, or 10 mg/kg/day by oral gavage for 5 weeks also
showed decreased body weight, but only in the 10 mg/kg/day group, which weighed 83% of
controls (Qu et al., 2016). In a separate study, although reductions in body weight were observed
in male BALB/c mice after 1 week of exposure to 10 mg/kg/day PFOS via gavage, this effect
C-99
-------
APRIL 2024
was attenuated at the end of the exposure period at 3 weeks (Lv et al., 2015). Additionally, a
significant increase in body weight was observed in 2.5 mg/kg/day exposure group at the end of
the 3-week exposure period (Lv et al., 2015). Food consumption was not reported in these
studies (Qu et al., 2016; Lv et al., 2015). No change in body weights were observed across 8
timepoints in male ICR mice exposed to 0.5 mg/kg/day, 5 mg/kg/day, or 10 mg/kg/day by oral
gavage for 28 days (Qiu et al., 2016).
Three studies using Sprague-Dawley rats reported decreased body weights following PFOS
exposure via oral gavage for 28 days, which usually occurred at the highest dose tested. Of these,
Han et al. (2018a) and Wan et al. (2016) exposed males to 1 mg/kg/day or 10 mg/kg/day and
observed an approximate 10% reduction in body weight following 10 mg/kg/day. NTP (2019)
reported decreased body weights in male and female Sprague-Dawley rats exposed to
5 mg/kg/day PFOS. However, body weights of all dose male and female groups were within
10% of control groups. The decrease in body weights was not associated with reduced food
consumption in Han et al. (2018a), and food consumption was not reported in the other studies
(NTP, 2019; Wan et al., 2016). Two studies by Salgado et al. (2016; 2015) using the same
animals reported no change in body weight variation or food consumption in male Sprague-
Dawley rats administered 3 mg/kg/day or 6 mg/kg/day PFOS by oral gavage for 28 days, but
data were not provided.
A reduction in body weight was also observed following 6 weeks of PFOS exposure via gavage
in male and female Crl:Cd(Sd)lgs Br Vaf rats exposed to 3.2 mg/kg/day (weighing 93 and 88%
of control, respectively), which was associated with decreased food consumption (Luebker et al.,
2005a). Although a 6-week exposure to 2 mg/kg/day did not reduce body weights in female
Crl:CD(SD)Igs Vaf/Plus rats, this dose did reduce mean female body weight gain and food
consumption (Luebker et al., 2005b). In a study assessing the dietary PFOS exposure in the same
rat strain, no change was observed in body weights or food consumption in male and female
Crl:CD(SD)IGS BR rats exposed to PFOS in the diet at concentrations of 0 ppm, 0.5 ppm,
2 ppm, 5 ppm, or 20 ppm (equivalent to 0 mg/kg, 0.05 mg/kg, 0.18 mg/kg, 0.37 mg/kg, or
1.51 mg/kg in males and 0 mg/kg, 0.05 mg/kg, 0.22 mg/kg, 0.47 mg/kg, or 1.77 mg/kg in
females) for 4 weeks (Seacat et al., 2003).
Chronic PFOS exposure studies also suggest an effect of PFOS on body weight. Male and female
Cynomolgus monkeys exposed to 0 mg/kg/day, 0.03 mg/kg/day, 0.15 mg/kg/day, or
0.75 mg/kg/day PFOS (equivalent to cumulative doses of 0 mg/kg, 4.6 mg/kg, 22.9 mg/kg, or
114.7 mg/kg) via intragastric intubation for 26 weeks (182 days) showed a reduction in body
weight change in the highest dose group (8% reduction in males and 4% reduction in females),
although no change in absolute body weight was observed (Seacat et al., 2002). This is in
contrast to the 14% and 5% body weight increases in control males and females, respectively.
However, chronic (14 weeks) exposure to PFOS in the diet at 0 ppm, 0.5 ppm, 2 ppm, 5 ppm,
and 20 ppm (equivalent to 0 mg/kg, 0.05 mg/kg, 0.18 mg/kg, 0.37 mg/kg, and 1.51 mg/kg in
males and 0 mg/kg, 0.05 mg/kg, 0.22 mg/kg, 0.47 mg/kg, and 1.77 mg/kg in females) showed
had no effect on Crl:CD(SD)IGS BR male or female rats. For 20 ppm dose-group males,
terminal body weights appeared to be reduced in a dose-dependent manner, however this
difference was not statistically significant (Seacat et al., 2003). In line with reduced body
weights, food consumption was significantly decreased in the 20 ppm exposure group, but these
data were not shown and the sex of the animals affected was not specified (Seacat et al., 2003).
C-100
-------
APRIL 2024
PFOS Whole Body Effects - Body Weights
Endpoint Study Name
Study Design
Observation Time
Animal Description
0 No significant change^ Significant increase ~ Significant decrease
Body Weight, Absolute Seacat et al.. 2002, 757853
chronic (26wk)
184d
Monkey, Cynomolgus (y, N=4-6)
Monkey. Cyriomolgus N=4-6)
Zhong Qt al., 2016. 3748828
developmental (GD1-17)
PNW8
F1 Mouse, C57BL/6 N=12)
« « « ~
F1 Mouse, C57BL/6 (V. N=12)
X • ~
Qiu et al., 2020, 7276729
short-term (4wk)
4wk
Mouse, ICR N=10)
Qiu etal., 2016, 3981408
short-term (4wk)
4wk
Mouse, ICR (:£, N=10)
Yanetal.. 2014, 2850901
short-term (28d)
28d
Mouse. BALB/c ({'¦, N=16)
t ——»
Xing etal.. 2016. 3981506
subchronic (30d)
31d
Mouse. C57BL/6J (
-------
APRIL 2024
2005a), and in female Crl:Cd(Sd)lgs Vaf/Plus rats following a 6 week exposure to 2.0 mg/kg/day
(Luebker et al., 2005b) (see Toxicity Assessment, (U.S. EPA, 2024)).
Food and water consumption was not observed to be affected in Sprague-Dawley rats exposed to
PFOS via gavage at doses of 1 mg/kg/day or 10 mg/kg/day (Han et al., 2018a), 3 or 6 mg/kg/day
(Salgado et al., 2015), nor 0.5 mg/kg/day, 1 mg/kg/day, 3 mg/kg/day, or 6 mg/kg/day (Salgado et
al., 2016) for 28 days. Seacat et al. (2003) fed Crl:CD(SD)IGS Br male or female rats 0, 0.5, 2, 5,
and 20 ppm PFOS for 4 or 14 weeks (equivalent to 0, 0.05, 0.18, 0.37, and 1.51 mg/kg in males
and 0, 0.05, 0.22, 0.47, and 1.77 mg/kg in females). The authors noted that food consumption
was slightly reduced in the 20 ppm female dose group during the first 4 weeks of dosing, but
these data were not provided (Seacat et al., 2003). By 14 weeks, food consumption was noted to
be significantly decreased in the 20 ppm dose group, but these data were not provided and the
sex of the animals affected was not specified.
C.3.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse metabolic outcomes is discussed in
Sections 3.2.2, 3.3.2, and 3.3.4 of the 2016 PFOS HESD (U.S. EPA, 2016c). There are 32 and 36
studies from recent systematic literature search and review efforts conducted after publication of
the 2016 PFOS HESD that investigated the mechanisms of action of PFOS that lead to metabolic
and systemic effects, respectively. A summary of these metabolic and systemic studies is shown
in Figure C-25 and Figure C-26, respectively. Additional mechanistic synthesis will not be
conducted since evidence suggests but is not sufficient to infer that PFOS leads to metabolic and
systemic effects.
Mechanistic Pathway Animal Human In Vitro Grand Total
Big Data, Non-Targeted Analysis
0
2
1
3
Cell Growth, Differentiation, Proliferation, Or Viability
1
0
11
12
Cell Signaling Or Signal Transduction
a
1
e
11
Fatty Add Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
7
1
e
15
Hormone Function
1
4
3
e
Oxidative Stress
2
1
2
5
Xen obi otic Metabolism
0
0
2
2
Other
2
0
0
2
Not Applicable/Not Specified/Review Article
1
0
0
1
Grand Total
11
7
16
32
Figure C-25. Summary of Mechanistic Studies of PFOS and Metabolic Effects
Interactive figure and additional study details available on HAWC.
C-102
-------
APRIL 2024
Mechanistic Pathway Animal Human In Vitro Grand Total
Atherogenesis And Clot Formation
0
0
1
1
Big Data, Non-Targeted Analysis
3
0
2
4
Cell Growth, Differentiation, Proliferation, Or Viability
4
1
11
16
Cell Signaling Or Signal Transduction
2
1
7
10
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
4
1
9
13
Hormone Function
1
0
0
1
Inflammation And Immune Response
0
0
2
2
Oxidative Stress
5
1
7
13
Xenobiotic Metabolism
3
1
1
4
Other
0
0
4
4
Not Applicable/Not Specified/Review Article
2
0
0
2
Grand Total
10
3
25
36
Figure C-26. Summary of Mechanistic Studies of PFOS and Systemic Effects
Interactive figure and additional study details available on HAWC.
C.3.4 Evidence Integration
There is slight evidence of an association between PFOS exposure and metabolic effects in
humans based on observed effects for diabetes, gestational weight gain, HOMA-IR, HOMA-B,
leptin, and adiponectin in high and medium confidence studies. Five studies observed non-
significant positive associations with gestational diabetes. In the general population, six studies
reported positive associations with type 2 diabetes. Three epidemiological studies observed
positive associations with gestational weight gain. Seven studies reported non-significant
positive associations with HOMA-IR in pregnant women and in general populations, or in adults
at high risk for type 2 diabetes. Of the six studies on HOMA-IR in children, only one reported a
positive association with HOMA-IR. Four studies reported positive associations with HOMA-B,
but an inverse association was observed in children (one study). There is limited evidence
suggesting a potential association between PFOS exposure and adiponectin in children, but not
adults. Findings for an association between PFOS exposure and MetS were mixed in four general
population epidemiological studies identified since 2016: two reported negative associations with
MetS, and two reported positive associations.
The animal evidence for an association between PFOS and systemic or metabolic effects is
indeterminate. Although some alterations related to glucose homeostasis were reported in the
available animal toxicity literature, the results from 6 high or medium confidence studies are
inconclusive as there are too few studies to assess possible difference across life stages, sexes,
C-103
-------
APRIL 2024
and species. In addition, the effects on body weight, clinical observations, and mortality from 20
high or medium confidence studies indicate that the systemic effects occur only at the high doses
tested. NTP (2019) and Seacat et al. (2003) reported differing observations on the impact of
PFOS on serum glucose in male rats at 4 weeks, which may be explained by differing methods of
exposure (gavage and dietary, respectively). Additionally, the statistically significant
observations reported by Seacat et al. (2003) and Curran et al. (2008) differ between males and
females, are not consistent across timepoints, and sometimes did not follow a linear dose-
response relationship. Given the differences noted in timing of measurement, duration of
exposure, and differences across sex, the biological significance of the increase or decrease in
metabolic endpoints such as serum glucose in these animal models is unclear, especially
considering the sensitivity of these parameters to increases in animal stress.
There were also inconsistencies in results reported in developmental studies. Lv et al. (2013)
reported dose-dependent increases in serum glucose during a glucose tolerance test at PNW10 in
rat offspring. This trend did not continue through PNW 15 in this study. In addition, Wan et al.
(2014) did not report significantly altered results of the glucose tolerance test at PND 63 in
mouse offspring gestationally exposed to PFOS and fed standard diets. Although multiple studies
indicate potential effects of PFOS on glucose homeostasis, the responses were inconsistent
and/or transient for specific endpoints across studies and the biological significance of the
observed effects is uncertain.
Though the observed metabolic effects were inconsistent, evidence from animal studies suggests
that PFOS exposure may induce whole-body toxicity, but only at the higher doses tested.
Decreased body weight and food consumption were observed in a number of subchronic and
chronic studies using rodents and non-human primates. While signs of decreased body weights
can be indicative of poor health in animals and a relevant endpoint demonstrating whole-body
toxicity, the effects reported in these studies were generally minimal and only surpassed a >10%
change in body weight at the highest doses tested.
C.3.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause systemic and metabolic
effects in humans under relevant exposure circumstances (Table C-6). This conclusion is based
primarily on diabetes, gestational weight gain, HOMA-IR, HOMA-B, leptin, and adiponectin
effects observed in high and medium confidence studies in humans exposed to median PFOS
levels between 5.4 and 35.7 ng/mL. Although there is some evidence of negative effects of PFOS
exposure on MetS, there is considerable uncertainty in the results due to inconsistency across
studies and limited number of studies.
C-104
-------
APRIL 2024
Table C-6. Evidence Profile Table for PFOS Systemic and Metabolic Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.3.1)
Glucose metabolism
4 High confidence
studies
13 Medium confidence
studies
7 Low confidence
Studies
Findings for FBG in
adults were primarily
positive (7/12), but only a
few reached significance.
OGTT results were
examined only in studies
finding significant
increases in FBG and
were congruent with FBG
findings. In children,
decreases in FBG were
observed (3/5), but none
were significant. Findings
for FBG in pregnant
women were similarly
non-significantly inverse
(3/4), however, the three
high and medium
confidence studies
conducting OGTT
observed increases in 1-
hr glucose levels, two of
which were significant.
• High and medium
confidence studies
• Consistent direction
of effect for FBG in
adults
• Low confidence studies
• Imprecision of findings
• Potential for selection
bias and residual
confounding by SES
Diabetes (and
gestational diabetes)
3 High confidence
studies
16 Medium confidence
studies
5 Low confidence
studies
Findings in adults were
mixed. Among the high
and medium confidence
studies (8/11), two
reported significant
positive associations
(2/8), 1 reported a
significant inverse
association (1/8),
»High and medium
confidence studies
• Low confidence studies
• Inconsistent direction of
effect
• Lmprecision of findings
• Potential for outcome
misclassification, self-
selection, residual
©OO
Slight
Evidence for metabolic
effects is based on
increases in FBG,
increased odds of
diabetes, and increases in
measures of adiposity in
adults. Positive
associations were
reported for heightened
glucose levels, effects on
insulin regulation,
diabetes, and adiposity,
but many medium and
high confidence studies
presented non-
statistically significant
results and several studies
presented conflicting
associations.
Uncertainties remain due
- to mixed results,
contrasting findings, and
potential for residual
confounding in the
analysis of outcomes
such as glucose
metabolism, diabetes, and
insulin levels.
©OO
Evidence Suggests
Primary basis:
Human evidence indicted
effects on diabetes,
gestational weight gain,
HOMA-IR, HOMA-B,
leptin, and adiponectin and
there was limited animal
evidence. Although there
is some evidence of
negative effects of PFOS
exposure on MetS, there is
considerable uncertainty in
the results due to
inconsistency across
studies and limited number
of studies.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
C-105
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
and 5 reported imprecise
associations (5/8). The 3
low confidence studies all
reported non-significant
positive associations and
typically relied on self-
reported data. Findings
for HbAlc were less
consistent. In pregnant
women, findings for
gestational diabetes were
mixed. The only study
examining diabetes in
children was considered
uninformative.
confounding by SES,
and failure to establish
temporality
Insulin levels
2 High confidence
studies
7 Medium confidence
studies
10 Low confidence
studies
Findings from a high
confidence study in
adults reported
significant increases in
fasting insulin, HOMA-
IR, HOMA-B, and
insulin responses during
an OGTT, however, this
population was at high
risk for type 2 diabetes.
Findings for adults
among medium and low
confidence studies were
generally mixed, but
there were multiple
contrasting findings for
HOMA-IR, indicating an
inverse association (5/9).
Studies in children
reported mixed and
• High and medium
confidence studies
• Low confidence studies
• Inconsistent direction of
effects
• Imprecision of findings
• Potential for residual
confounding by diabetes
status or use of
medications that would
impact insulin levels in
some studies
C-106
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
generally imprecise
findings for measures of
insulin resistance.
Similarly, findings in
studies among pregnant
women were imprecise.
Adiponectin and leptin
5 High confidence
studies
3 Medium confidence
studies
1 Low confidence study
Inverse associations with
adiponectin were reported
in two studies of adults
(2/2), while one study
(1/1) reported increases
in leptin. None reached
significance. Findings for
adiponectin in children
were positive (5/6), but
only one reached
significance. Findings for
leptin were mixed among
children. Only one study
reported findings from
pregnant women,
observing non-significant
increases in both
adiponectin and leptin.
• High and medium
confidence studies
• Consistent direction
of effect for
adiponectin in
children
• Low confidence study
• Imprecision of findings
Adiposity
4 High confidence
studies
17 Medium confidence
studies
4 Low confidence
studies
In adults, findings for
BMI were primarily
positive (4/6), indicating
increased BMI. Increases
in the odds of being
overweight or obese were
also reported, which was
significant for women in
one study. Results were
mixed for WC, but one
study observed
»High and medium
confidence studies
• Low confidence studies
• Inconsistent direction of
effects
• Imprecision of findings
C-107
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
differences in direction of
effect between men and
women. Findings for
BMI in children were
mixed, with studies of
medium confidence
reporting significant
positive and significant
inverse associations with
measures of BMI. In
pregnant women, positive
associations were
reported for gestational
weight gain, but results
were inconsistent
between studies after
stratification of weight
status (i.e., under-,
normal-, or overweight).
Metabolic syndrome
4 Medium confidence
studies
1 Low confidence study
In adults, findings for
MetS were mixed, and
none reached significance
(0/4). Significant
reduction in the resting
metabolic rate were
observed in a single study
of adults. MetS was not
evaluated in children or
pregnant women.
• Medium confidence
studies
»Low confidence study
»Lnconsistent direction of
effects in medium
confidence studies
» Concern for selection
bias, outcome
misclassification, and
residual confounding by
SES in low confidence
study
Evidence From In Vivo Animal Studies (Section C.3.2)
Glucose homeostasis
1 High confidence study
5 Medium confidence
studies
Mixed results were
reported on glucose
levels in rodent studies
(6). Of these, 2 reported
non-significant effects,
»High and medium
confidence studies
»Lnconsistent direction
and magnitude of effects
across study designs- and
sex
ooo
Lndeterminate
Alterations related to
glucose homeostasis were
C-108
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
and 4 reported significant
effects with inconsistent
directionality. Reduced
glucose levels were
reported in female
rodents (3/4) at the
highest PFOS exposure
group tested. No
significant effects on
glucose levels were
observed in males (3/3)
and dams (1/2). One
study in female mice
reported decreased
insulin resistance (1/1)
and pyruvate tolerance
(1/1).
• Limited number of
studies examining
outcomes
Body weight
Statistically significant
• High and medium • Effects do not follow a
3 High confidence
reductions in body
confidence studies linear dose-responsive
studies
weights (9/20) and body
• Consistent direction relationship
17 Medium confidence
weight changes (2/2)
of effects
studies
were reported in various
• Confounding
studies, including studies
variables such as
in rats (11), mice (9), and
food consumption
monkeys (2).
were considered in
most studies
Survival and mortality
1 High confidence
studies
6 Medium confidence
studies
No effects on survival
and mortality were
reported in rodent studies
(6/6). One study in non-
human primates observed
increased mortality at the
highest dose tested (1/1).
• High and medium
confidence studies
• Consistent direction
of effects across sex,
species, and duration
of exposure
• Limited number of
studies examining
outcomes
reported in 6 high or
medium confidence
studies were inconclusive
as there are too few
studies to assess possible
difference across life
stages, sexes, and species
and results from the
existing studies are
inconsistent or transient.
Systemic effects
(e.g., body weight,
clinical observations,
survival, food
consumption, and water
consumption) from 20
high or medium
confidence studies
indicate that biologically
significant effects
(e.g., body weight change
exceeding 10% of
control) tend to occur
only at the highest doses
tested.
C-109
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Clinical observations
1 High confidence study
3 Medium confidence
studies
Clinical observations
were observed in most
rodent studies (3/4).
Findings found across
these studies included:
hyperkalemia, rough
hair, slow movement,
constipation, and
localized areas of partial
alopecia.
»High and medium
confidence studies
• Limited number of
studies examining
outcomes
• Qualitative and
subjective data reporting
Food and water
consumption
9 Medium confidence
studies
Reduced food
consumption (6/9) was
reported in the higher
dose groups in male and
female rodents. No
significant effects were
reported on water
consumption in male rats
following short-term
exposure (2/2).
• Medium confidence
studies
• Consistent direction
of effects on water
consumption
• Limited number of
studies examining
outcomes
Notes: FBG = fasting blood glucose; hr = hour; OGTT = oral glucose tolerance testing; HbAlc = hemoglobin Ale; SES = social economic status; HOMA-IR = homeostatic model
assessment for insulin resistance; HOMA-B = homeostasis model assessment of P-cell function; BMt = body mass index; WC = waist circumference; MetS = metabolic
syndrome.
C-110
-------
APRIL 2024
C.4 Nervous
EPA identified 36 epidemiological and 16 animal studies that investigated the association
between PFOS and nervous effects. Of the epidemiological studies, 3 were classified as high
confidence, 28 as medium confidence, and 5 were considered low confidence (Section C.4.1). Of
the animal studies, 1 was classified as high confidence, 8 as medium confidence, 4 as low
confidence, 2 as mixed (2 medium/low) confidence, and 1 was considered uninformative (Section
C.4.2). Studies may have mixed confidence ratings depending on the endpoint evaluated. Though
low confidence studies are considered qualitatively in this section, they were not considered
quantitatively for the dose-response assessment (see Toxicity Assessment, (U.S. EPA, 2024)).
C.4.1 Human Evidence Study Quality Evaluation and Synthesis
C.4.1.1 Introduction
The 2016 Health Assessment (U.S. EPA, 2016c) reviewed studies examining associations
between PFOS exposure and neurodevelopmental disorders in children, including attention
deficit hyperactivity disorder (ADHD) and learning disabilities and concluded there was limited
evidence to suggest an effect. A significant increase in risk of development of cerebral palsy in
males was observed in a case-control study of maternal PFOS levels of participants within the
DNBC (Liew et al., 2014). One study observed a significant positive association of child PFOS
levels with parent-reported ADHD in children aged 12-15 in the general population (Hoffman et
al., 2010). No association between maternal plasma PFOS concentrations and Apgar score or
between maternal plasma PFOS concentrations and mother reported assessments of fine motor
skills, gross motor skills or cognitive skills in children at 6 and 18 months of age were observed
in one study of pregnant women and their children (Fei et al., 2008a). No association between
parent-reported behavioral or coordination problems in children 7 years of age and prenatal
PFOS levels was reported in another study (Fei and Olsen, 2011). No associations were observed
between prenatal PFOS and parent-reported motor development scores in children ages 7 to 9;
however, the highest PFOS tertile was associated with a 0.5-point higher hyperactivity score for
participants within one country with higher exposures, but not for participants within other
countries (H0yer et al., 2015). Data interpretations within these studies were limited in some
cases by use of a cross-sectional study design (Hoffman et al., 2010; Fei et al., 2008a), potential
random misclassification error resulting from using current PFOS levels as proxy measures of
etiologically relevant exposures (Hoffman et al., 2010), outcomes defined by parental report
(H0yer et al., 2015; Fei and Olsen, 2011; Hoffman et al., 2010; Fei et al., 2008a), and limited
sample sizes in some countries (H0yer et al., 2015).
For this updated review, 35 studies (35 publications) investigated the association between PFOS
and neurological outcomes that have been identified since the 2016 document. One was
conducted in a high-exposure community (Spratlen et al., 2020a). One publication (Vuong et al.,
2020b) was conducted in pregnant women. The remainder were conducted in the general
population. Study designs included 3 case-control (Shin et al., 2020; Long et al., 2019; Ode et
al., 2014), 2 nested case-control (Lyall et al., 2018; Liew et al., 2015), 26 cohort, and 5 cross-
sectional studies (Appendix D). The studies measured PFOS in different matrices including
blood, serum, plasma, cord blood, breast milk (Lenters et al., 2019; Forns et al., 2015), maternal
serum, maternal plasma, and amniotic fluid (Long et al., 2019). Several studies (Vuong et al.,
2020b; Vuong et al., 2020a; Vuong et al., 2019; Vuong et al., 2018b; Vuong et al., 2018a; Zhang
C-lll
-------
APRIL 2024
et al., 2018a; Vuong et al., 2016; Braun et al., 2014) were conducted on subsets of data from the
HOME study. Two studies (Lenters et al., 2019; Forns et al., 2015) utilized data from the
Norwegian Human Milk Study (HUMIS). Two studies (Liew et al., 2018; Liew et al., 2015)
utilized the DNBC data. The studies were conducted in multiple locations including populations
from China, Denmark, the Faroe Islands, Great Britain, Japan, the Netherlands, Norway,
Sweden, Taiwan, and the United States (Appendix D). Neurological effects were determined for
numerous clinical conditions and by assessing performance on neuropsychological tests
assessing various neurological domains, including developmental, general intelligence
(i.e., intelligence quotient (IQ)), social-emotional, executive function, ADHD and attention,
autism spectrum disorder (ASD) and intellectual disability (ID), and visuospatial performance.
C.4.1.2 Study Quality
There are 34 studies (36 publications)7 from recent systematic literature search and review
efforts conducted after publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated
the association between PFOS and nervous effects. Study quality evaluations for these 36 studies
are shown in Figure C-27 and Figure C-28.
Of the 36 studies identified since the 2016 assessment, three (Niu et al., 2019; Harris et al., 2018;
Oulhote et al., 2016) were classified as having high confidence, 28 studies were classified as
medium confidence, and five were low confidence. Studies rated as low confidence had
deficiencies including potential residual confounding, exposure misclassification, selection bias,
and small sample size. One low confidence NHANES study (Berk et al., 2014) had a high
likelihood of residual confounding due to the use of an insensitive marker of SES, and the
analysis did not account for the population's complex sampling design. Differences in laboratory
extraction methods, collection timing, and missing details on storage raised concerns for
exposure misclassification in a study on children from the HUMIS cohort (Forns et al., 2015).
Additionally, children were only evaluated on some, but not all, test instrument (Ages and Stages
Questionnaire (ASQ)) domains, and rationale for domain selection was not provided. Concerns
for Lien et al. (2016) included a high loss to follow-up, lack of detail on completion rates of
ADHD questionnaires and low detection rate for PFOS. Small sample size, temporality and
reporting issues were cited as limitations in Weng et al. (2020). Finally, limitations in Ode et al.
(2014) included sensitivity concerns due to the limited number of ADHD cases and potential for
residual confounding due to the lack of data on other exposures potentially related to ADHD. In
the evidence synthesis below, high, and medium confidence studies were the focus, although low
confidence studies were still considered for consistency in the direction of association.
Vuong et al. (2018b) reports score trajectories for the same population and test as Vuong et al. (2016). Vuong et al. (2020a)
reports on an overlapping population with the same test as Zhang et al. (2018a).
C-112
-------
Berketal., 2014, 2713574
Braunetal., 2014, 2345999
Chen etal.,2013, 2850933
Ding and Park, 2020, 6711603
Fornsetal.,2015, 3228833
Gallo etal.,2013, 2272847
Ghassabian et al., 2018, 5080189
Goudarzi et al., 2016, 3981536
Harris etal., 2018, 4442261
Jeddy et al., 2017, 3859807
Lenters et al., 2019, 5080366
Li, 2020, 6833686
Lien etal., 2016, 3860112
Liew etal., 2015, 2851010
Liew etal., 2018, 5079744
Long etal., 2019, 5080602
Lyall etal., 2018, 4239287
Niu etal., 2019, 5381527
APRIL 2024
.e\e° ^
s^
s?
D Legend
Good (metric) or High confidence (overall)
+ Adequate (metric) or Medium confidence (overall)
- Deficient (metric) or Low confidence (overall)
Jj Critically deficient (metric) or Uninformative (overall)
Figure C-27. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Neurological Effects
Interactive figure and additional study details available on HAWC.
C-113
-------
Ode et al.
Oulhote et al.
Oulhote et al.
Quaak et al.
Shin et al.
Shrestha et al.
Skogheim et al.
Spratlen et al.
Stram et al,
Vuong et al.
Vuong et al.
Vuong et al.
Vuong et al,
Vuong et al,
Vuong et al,
Wang et al.
Weng et al.
Zhang et al.
APRIL 2024
9#
2014, 2851245-
++
+
i
+
'
|
+
'
+
-
n
2016, 3789517-
+
+
+
++
2019, 6316905-
+
+
-
+
+
+
+
2016, 3981464-
+
+
+
+
+
+
-
+
2020,6507470 -
+
+
++ ++ ++
+
+
+
2017, 3981382-
+
+
+
+
+
+
+
+
2019, 5918847-
2
++
+
+*
+
2020,6364693 -
+
-
+
+
+
+
-
+
2014, 2922190-
++
++
+
+
++
+
+
+
2016, 3352166-
+
+
+
+
+
+
+
+
2018, 5079675-
+
++
+
+
+
+
-
+
2018, 5079693-
+
++
+
++
+
+
+
+
2019, 5080218-
+
+
+
+
++
+
+
+
2020,6356876 -
+
+
+
+
+
+
-
+
2020, 6833684 -
++
+
+
+
+
+
+
+
2015, 3860120-
-
+
+
+
+
+
+
2020, 6718530-
-
+
+
-
-
-
-
2018,4238294-
+
-
++
+
+
+
Legend
p
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
h
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-28. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Neurological Effects (Continued)
Interactive figure and additional study details available on HAWC.
C-114
-------
APRIL 2024
C.4.1.3 Findings From Children and Adolescents
Six cohort studies (Niu et al., 2019; Jeddy et al., 2017; Shrestha et al., 2017; Goudarzi et al.,
2016b; Forns et al., 2015; Chen et al., 2013), and one high-exposure community study (Spratlen
et al., 2020a) examined developmental outcomes in children. In a high confidence study (Niu et
al., 2019) from the Shanghai-Minhang Birth Cohort Study (S-MBCS), maternal PFOS
concentrations (median =10.8 ng/mL) during pregnancy were inversely associated with
neuropsychological development (especially for personal-social skills) assessed by the ASQ in 4-
year-old children. A medium confidence study of data from the Taiwan Birth Panel Study (Chen
et al., 2013) observed associations between in utero PFOS (mean = 7.4 ng/mL) and decreases in
Comprehensive Developmental Inventory (CDI) developmental quotients in the highest exposure
group compared with the lowest exposure group for the whole test as well as for gross motor,
fine motor, and self-help domains. Effect sizes were generally greater with increasing PFOS
levels. A medium confidence study (Jeddy et al., 2017) utilizing data from the ALSPAC
observed significant associations between maternal PFOS (median = 19.8 ng/mL) and verbal
comprehension scores as assessed by the adapted MacArthur Communicative Development
Inventories for Infants (MCDI) in children at 15 months of age, but not for vocabulary
comprehension and production, nonverbal communication, or social development. Significant
inverse associations were also observed between maternal PFOS and language and intelligibility
scores in children at 38 months of age. Results for this study varied by maternal age at delivery.
A statistically significant inverse association was reported for vocabulary comprehension and
production scores in 15-month infants with mothers <25 years of age. A significant inverse
association was observed for intelligibility scores in children 38 months of age with mothers
>30 years of age, and a significant positive association was observed for intelligibility scores in
children 38 months of age with mothers <25 years of age. Results from a medium confidence
study (Goudarzi et al., 2016b) reported no significant associations between prenatal PFOS levels
(median = 5.7 ng/mL at 6 months; median = 5.8 at 18 months) and Mental (MDI) and
Psychomotor (PDI) Development Indices in infants at 6 and 18 months. Similarly, no significant
adverse associations or apparent trends between delivery or cord blood PFOS concentrations
(median = 6.0 ng/mL) and age 1 mental or psychomotor developmental indices were reported in
a high-exposure community study of children prenatally exposed to the World Trade Center
(WTC) Disaster, however a significant interaction by sex with MDI at ages 2 and 3, with
stronger positive associations for females compared with males was observed (Spratlen et al.,
2020a).
Ten studies evaluated cognitive function and IQ measures among children, with most conducted
within the general population (Vuong et al., 2020a; Oulhote et al., 2019; Skogheim et al., 2019;
Vuong et al., 2019; Harris et al., 2018; Liew et al., 2018; Zhang et al., 2018a; Wang et al.,
2015b; Stram et al., 2014), and one within a high-exposure community (Spratlen et al., 2020a).
In a high confidence analysis of participants within Project Viva, children born to women with
top quartile PFOS (34.9-168.0 ng/mL) concentrations had higher nonverbal IQ scores, although
dose-response patterns appeared non-linear (Harris et al., 2018). Positive associations were
observed between prenatal PFOS (median = 12.7 ng/mL) and reading skills at age eight years in
a medium confidence study (Vuong et al., 2020a) which utilized data from the HOME study.
Childhood serum PFOS concentrations at ages three and eight years were positively associated
with higher children's reading scores at ages five and eight years, respectively in an additional
medium confidence study of data within the HOME study (Zhang et al., 2018a). No significant
C-115
-------
APRIL 2024
associations were reported between maternal prenatal PFOS (median = 21.4 ng/mL) and
offspring scholastic achievement in a medium confidence prebirth cohort study of participants
within the Danish Fetal Origins 1988 (DaF088) cohort (Stram et al., 2014). Maternal prenatal
PFOS (median = 27.7 ng/mL) concentrations were associated with lower cognitive function as
assessed by the Boston Naming Test in a medium confidence study of children aged seven years
(Oulhote et al., 2019).
In a medium confidence study in a highly exposed community, sex-specific trends between
PFOS exposures and some cognitive outcomes (verbal and full-scale IQ only) at 4 and 6 years
were observed, suggesting stronger positive associations for females compared with males
(Spratlen et al., 2020a). Another medium confidence study investigated associations between
prenatal exposure to PFOS and IQ at age five in a sample of children from the DNBC with no
consistent associations observed (Liew et al., 2018). Consistent adverse associations with age
eight cognitive development as assessed by IQ were not observed in an additional medium
confidence study (Vuong et al., 2019). Similarly, utilizing data from participants within the
Taiwan Maternal and Infant Cohort Study, a medium confidence prospective cohort study by
Wang (Wang et al., 2015b) reported no significant associations between maternal serum PFOS
(median = 13.3 ng/mL) and IQ measurements in children five or eight years of age. Evidence
was inconsistent, with significant decreases in nonverbal working memory only in the highest
quintile and no significant associations with verbal working memory, for the evaluation of the
association between prenatal exposure to PFOS (median = 11.5 ng/mL) and cognitive
dysfunction in preschool children in a medium confidence study from The Norwegian Mother,
Father, and Child Cohort Study (MoBa) (Skogheim et al., 2019).
Six studies assessed the relationship between PFOS and behavioral development problems and
behavioral regulation problems (Weng et al., 2020; Oulhote et al., 2019; Ghassabian et al., 2018;
Vuong et al., 2018a; Oulhote et al., 2016; Quaak et al., 2016). No significant associations
between prenatal PFOS (1,650 ng/L) and externalizing problems at age 18 months assessed using
the Child Behavior Checklist 1.5-5 (CBCL 1.5-5) were reported in a high confidence study
utilizing data from the Dutch cohort LINC (Linking Maternal Nutrition to Child Health) (Quaak
et al., 2016). No consistent associations in total Strengths and Difficulties Questionnaire (SDQ)
behavior scores with serum PFOS (median = 16.8 ng/mL) at age five was observed, but a
twofold increase in serum PFOS (median = 15.3 |ig/L) in children aged seven years was
associated with higher SDQ total behavioral difficulties scores in girls, and lower scores in boys
(gender interaction p < 0.05) in a high confidence study (Oulhote et al., 2016). Maternal prenatal
PFOS concentrations (median = 27.7 ng/mL) were positively associated with total scores on the
SDQ, indicating more behavioral problems, in a medium confidence study of children
seven years of age (Oulhote et al., 2019). Higher newborn PFOS levels (median =1.7 ng/mL) in
dried blood spots were associated with increased odds of having behavioral difficulties, driven
mostly by problems in conduct and emotional symptoms, as assessed by the maternal completed
SDQ at age 7 in another medium confidence birth cohort study (Ghassabian et al., 2018). Child
sex modified the associations between prenatal PFOS and attention, with males having better
performance than females, but not enough evidence was observed to support an overall
association between prenatal PFOS (median = 12.9 ng/mL) and inattention and impulsivity as
assessed by the Connors' Continuous Performance Test-II in a medium confidence study (Vuong
et al., 2018a). A low confidence study on adolescents reported a significant, inverse correlation
C-116
-------
APRIL 2024
between prenatal PFOS levels (mean = 14.85 ng/mL) and in the right putamen brain region
associated with impulsive behavior as assessed by MRI in teenage offspring (Weng et al., 2020).
One medium confidence study (Strain et al., 2014) from the DaF088 cohort examined the
association between prenatal PFOS exposure and depression among offspring with 20 years of
follow-up. No significant association was observed between clinical depression and maternal
PFOS (median = 21.4 ng/mL) levels.
Three medium confidence studies (Vuong et al., 2018b; Shrestha et al., 2017; Vuong et al., 2016)
examined the relationship between PFOS concentrations and executive function in children with
mixed results. Executive function was assessed with the parent-rated Behavior Rating Inventory
of Executive Function (BRIEF) in two studies (Vuong et al., 2018b; Vuong et al., 2016) among
HOME study participants at five and eight years of age. Higher BRIEF scores indicate executive
function impairments. Maternal serum PFOS concentrations were significantly associated with
poorer behavior regulation, metacognition, and global executive functioning, with approximately
a 3-point increase in all summary measures with a 1 ln-unit increase in PFOS concentrations
(Vuong et al., 2016). Vuong et al. (2018b) again utilized data from the HOME study in a medium
confidence cross-sectional analysis to examine associations of child PFOS levels measured in
children aged eight years with executive function and reported no significant associations
between PFOS and executive function.
Five medium confidence studies assessed relationships between PFOS exposures and ADHD
(Lenters et al., 2019; Skogheim et al., 2019; Quaak et al., 2016; Liew et al., 2015; Stram et al.,
2014). One medium confidence study (Lenters et al., 2019) examined early-life high PFOS
exposures in breast milk in relation to ADHD among children (range: 7.2-14.1 years old) from
the HUMIS and reported significant associations with PFOS concentrations
(median = 117.7 ng/L) and increased odds of ADHD (OR = 1.75, 95% CI: 1.11, 2.76) with
significant sex-specific effects. Stram et al. (2014) investigated the association between maternal
prenatal PFOS and ADHD among offspring (follow-up to age 20) of participants within the
DaF088 cohort. No significant association between maternal PFOS (median = 21.4 ng/mL) and
offspring ADHD was reported in this medium confidence study. A medium confidence nested
case-control study (Liew et al., 2015) within the framework of the DNBC examined prenatal
PFOS exposures and ADHD in children. No consistent evidence was observed to suggest that
prenatal PFOS exposures (ADHD cases median = 26.8 ng/mL; controls median = 27.4 ng/mL)
increase the risk of ADHD. Quaak et al. (2016) explored the relationship between prenatal PFOS
exposures and parent-reported ADHD using the CBCL 1.5-5. This medium confidence study
utilized data from the Dutch cohort, LINC. No significant associations were reported between
cord blood PFOS (median = 1,600 ng/L) exposures and ADHD scores in the whole population or
in the sex-stratified analyses.
Two low confidence studies (Lien et al., 2016; Ode et al., 2014) examined PFOS exposures in
relation to ADHD. Ode et al. (2014) investigated the association in a case-control study between
cord blood PFOS (median = 6.9 ng/mL for cases, 6.8 ng/mL for controls) exposures and ADHD
diagnosis in childhood (age range 5-17 years), but no associations between PFOS and ADHD
were observed. Lien, 2016, 3860112 evaluated the association between cord blood PFOS
(mean = 4.8 ng/mL) exposures and neurobehavioral symptoms related to ADHD among 7-year-
old participants from the Taiwan Birth Panel Study and the Taiwan Early-Life Cohort, but no
effects were observed.
C-117
-------
APRIL 2024
One high (Oulhote et al., 2016)and five medium confidence studies since the 2016 assessment
evaluated PFOS exposures in relation to autism, autistic behaviors, and ID (Shin et al., 2020;
Long et al., 2019; Lyall et al., 2018; Liew et al., 2015; Braun et al., 2014). A twofold increase in
serum PFOS (median = 15.26 (J,g/L) at age seven was associated with significantly higher SDQ
autism screening scores at age seven, with higher autism scores in females than in males, in a
high confidence study (Oulhote et al., 2016). In a medium confidence prospective birth cohort
study from the HOME study, increasing maternal serum PFOS concentrations
(median =13 |ig/L) were associated with increased autistic behaviors in children 4 to 5 years of
age as assessed by maternal completed Social Responsiveness Scale (SRS) scores, although not
significantly so, and PFOS levels were positively associated with SRS scores in boys, but not
girls (Braun et al., 2014). No consistent evidence of an association between maternal plasma
PFOS (median = 25.4 ng/mL for cases; 27.4 ng/mL for controls) and diagnosed childhood
autism identified by linkage to the Danish National Hospital Registry was observed in a medium
confidence nested case-control study of mother-child pairs with an average of ten years of
follow-up within the DNBC (Liew et al., 2015). Autism cases had significantly lower PFOS
levels in a medium confidence case-control study of amniotic fluid PFOS (median = 0.6 ng/mL
for cases; 1.4 ng/mL for controls) and diagnosed ASD, with cases identified as born 1982-1999
within the Danish Psychiatric Central Registry (Long et al., 2019). Prenatal maternal serum
PFOS (median = 17.5 ng/mL for ASD cases; 15.9 ng/mL for ID cases; 17.9 ng/mL for controls)
was inversely associated with ASD and ID in a medium confidence study of children aged 4.5-
9 years with diagnosed ASD and ID (Lyall et al., 2018). An association was reported in a
medium confidence study of modeled prenatal maternal PFOS and clinically confirmed ASD
from mother-child pairs in the Childhood Autism Risk from Genetics and Environment
(CHARGE) study of children ages two to five years, with modeled prenatal maternal PFOS
(median = 3.1 ng/mL for cases; 3.3 ng/mL for controls) associated with increased odds of child
diagnosis of ASD and among boys when stratified by sex (Shin et al., 2020).
The effects on visuospatial performance were evaluated in one high confidence study of
participants of Project Viva (Harris et al., 2018). Visual-motor test scores (Wide Range
Assessment of Visual Motor Abilities) were consistently lower with increasing prenatal or
childhood PFOS exposures. Children in the upper quartile of prenatal PFOS (Q4 = 34.9-
168.0 ng/mL) had lower mid-childhood visual-motor scores, and participants in the third quartile
of childhood PFOS (Q3 = 6.3-9.7 ng/mL) had significantly decreased visual-motor scores.
Participants from the HOME study were assessed using the Virtual Morris Water Maze
(VMWM), but no significant effects were observed (Vuong et al., 2018a).
C.4.1.4 Findings From Pregnant Women
No evidence was observed to support an adverse relationship between serum PFOS during
pregnancy and maternal depressive symptoms assessed by the Beck Depression Inventory (BDI)
from pregnancy to eight years postpartum in a medium confidence study based on women from
the HOME study (Vuong et al., 2020b).
C.4.1.5 Findings From the General Adult Population
The effects of PFOS on general intelligence and IQ test outcomes were examined in a medium
confidence study (Shrestha et al., 2017) of adults (ages 55-74 years) in New York state. Findings
C-118
-------
APRIL 2024
indicated higher PFOS was significantly associated with improved performance in tests of
delayed recall.
Findings of a medium confidence study (Shrestha et al., 2017), described above, indicated no
significant associations between serum PFOS in adults and tests of executive function.
Two medium confidence studies investigated a possible association between PFOS and
depression (Vuong et al., 2020b; Shrestha et al., 2017). No significant associations were
observed in a medium confidence study of depression assessed by the BDI and serum PFOS
(median = 33.7 ng/mL) in a cross-sectional study of adults aged 55 to 74 years (Shrestha et al.,
2017). Additionally, no evidence was observed to support a relationship in adults between serum
PFOS during pregnancy and maternal depressive symptoms assessed by the BDI from pregnancy
to 8 years postpartum in a medium confidence study based on women from the HOME study
(Vuong et al., 2020b). One low confidence study (Berk et al., 2014) of data from adults
participating in NHANES reported no adverse associations between PFOS levels and depression
as assessed by the nine-item depression module of the Patient Health Questionnaire (PHQ-9).
The effects on visuospatial performance were evaluated in one medium confidence cross-
sectional study of older adults (Shrestha et al., 2017). A significant association between serum
PFOS and improved tests of visual and spatial function results was reported.
Two medium confidence studies explored the relationships between PFOS and memory loss.
(Shrestha et al., 2017; Gallo et al., 2013). Statistically significant inverse associations between
PFOS and memory impairment were reported in a medium confidence study of adults in the C8
Health Project (Gallo et al., 2013). No adverse effects of PFOS on memory impairment were
again reported in a separate medium confidence study of older adults (Shrestha et al., 2017).
Two medium confidence cross-sectional studies investigated PFOS and hearing impairment in
adult NHANES participants. Li, 2020, 6833686 reported positive correlations between PFOS and
hearing impairment, while Ding and Park (2020) observed no significant associations.
C.4.2 Animal Evidence Study Quality Evaluation and Synthesis
There are three studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and 13 studies from
recent systematic literature search and review efforts conducted after publication of the 2016
PFOS HESD that investigated the association between PFOS and nervous effects. Study quality
evaluations for these 16 studies are shown in Figure C-29.
C-119
-------
APRIL 2024
^01 p^\Q $lPv 0°° S®xe s^OT o>v° ^ 0-
' L
Butenhoff et al„ 2009, 757873 -
Curran et al., 2008, 757871 -
Era et al„ 2009, 2919358-
Fuentes et al., 2007, 757865 -
Kawamoto et al., 2011, 2919266-
Li et al., 2021, 7643501 -
Lopez-Doval et al., 2015, 2848266 -
Luebker et al., 2005, 1276160 -
Mehrietal., 2016, 8776814-
Mshaty et al., 2020, 6833692 -
NTP, 2019, 5400978
NR
NR
NR
NR
NR
+*
NR
B
++
+*
Pereiro et al., 2014, 2230732 -
+
+
NR
++
+
+
Salgado et al., 2015, 3981583 -
+
+
NR
+
+
-
Salgado etal., 2016, 3179088-
+
+
NR
++
+
Thomford, 2002, 5432419-
+
-
NR
++
-
Zhang et al., 2019, 5080461 -
+
+
NR
+
+
+
*'
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
B Critically deficient (metric) or Uninformative (overall)
Not reported
* Multiple judgments exist
Figure C-29. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Nervous Effects
Interactive figure and additional study details available on HAWC.
C-120
-------
APRIL 2024
There are few studies evaluating neurotoxicity, including neurodevelopmental toxicity,
associated with short-term, subchronic, or gestational exposure to PFOS in experimental models
(Table C-7). No study indicates morphological changes or damage attributed to PFOS. However,
there is some evidence suggesting that PFOS exposure may be associated with neurobehavioral
and physiological effects (e.g., impairments in spatial learning and memory, increases in
locomotor activity, and changes in neuronal electrophysiology and neurotransmitter levels).
Further research may be warranted.
Brain weight was assessed in only one developmental study and two short-term study in rats.
Absolute and relative brain weights were unchanged in the offspring of Crl:CD (SD) rats dosed
with 0.1-1 mg/kg/day PFOS during gestation and lactation (Butenhoff et al., 2009). In male and
female Sprague-Dawley rats exposed to 2 mg/kg-100 mg/kg PFOS in diet for 28 days, relative
brain weights were increased in the 50 mg/kg and 100 mg/kg exposure groups in a
concentration-dependent manner, which may have been secondary to a decrease in body weights
as absolute brain weights were unchanged (Curran et al., 2008). The relative weights of the
amygdala, hippocampus, and prefrontal cortex were unchanged in male Sprague-Dawley rats
dosed with 0.5 mg/kg/day-6 mg/kg/day PFOS for 28 days (data not provided) (Salgado et al.,
2016); the absolute weights of these brain regions were not provided. One developmental and
one short-term study examined the gross pathology or histopathology of the brain, and no effects
were seen in rats exposed to 0.1 mg/kg/day-5 mg/kg/day PFOS (NTP, 2019; Butenhoff et al.,
2009). The authors of a subchronic study in female BALB/c mice dosed with 0.1 mg/kg/day and
1 mg/kg/day PFOS for 2 months noted a small amount of neuron phagocytosis and that neuronal
cells were contracted, deeply stained, and lacked clearly defined cytoplasm and nuclei (Li et al.,
2021b).
One developmental (Mshaty et al., 2020), one short-term (Fuentes et al., 2007b), and one
subchronic study (Long et al., 2013) in mice and several reproductive (Luebker et al., 2005a) and
developmental studies (Wang et al., 2015a; Butenhoff et al., 2009; Johansson et al., 2008;
Fuentes et al., 2007a) in rats assessed the neurobehavioral effects associated with PFOS. Mshaty
et al. (2020) assessed learning and memory in male C57BL/6J mice exposed to 0.1-1 mg/kg/day
PFOS from PND 1-PND 14 using the object location test, object recognition test, and pairwise
visual discrimination task. The discriminatory index for the object location and recognition
memory tests were decreased in mice exposed to 1 mg/kg/day, as was the learning curve for the
1 mg/kg/day group during the visual discrimination task. Spatial learning and memory were also
reduced in adult male C57BL6 mice dosed with 2.15 mg/kg/day and 10.75 mg/kg/day but not
0.43 mg/kg/day PFOS for 3 months, as seen by increases in escape latency and decreases in the
time spent in the target quadrant using the Morris water maze (Long et al., 2013). Time spent in
the target quadrant was also decreased in male CD1 mice dosed with 3 mg/kg/day but not
6 mg/kg/day PFOS for 4 weeks (Fuentes et al., 2007b). In this study, swimming speed was
increased in mice exposed to 3 and 6 mg/kg/day and distance traveled was increased in mice
exposed to 6 mg/kg/day, whereas no effects on motor activity were seen with the open-field test
or rotarod test. Similar effects on spatial learning and memory were seen in the offspring of
Wistar rat dams exposed to 15 mg/mL but not 5 mg/mL PFOS in drinking water throughout
gestation and/or lactation (drinking water consumption not reported); swimming speed was not
affected by exposure (Wang et al., 2015a). However, two studies reported no changes in learning
and memory, as tested with the Morris water maze or the Biel swimming maze, in male and
female rats exposed to 0.1 mg/kg/day-3.2 mg/kg/day PFOS pre- and postnatally (Butenhoff et
C-121
-------
APRIL 2024
al., 2009; Luebker et al., 2005a). In a two-generation study, Luebker et al. (2005a) also reported
no effects on learning, memory, and short-term retention, as measured in a passive avoidance
paradigm, and Butenhoff et al. (2009) reported no effects on the acoustic startle response.
However, increased motor activity (ambulatory and total locomotor activity) and lack of
habituation was seen at PND 17 in males exposed to >0.3 mg/kg/day or 1 mg/kg/day,
respectively, throughout development (Butenhoff et al., 2009). In male NMRI mice given a
single dose of 11.3 mg/kg at PND 10, during a period of development, lack of habituation was
also observed at 2 and 4 months of age (Johansson et al., 2008); this effect was not observed with
a single dose of 0.75 mg/kg at PND 10. In this study, locomotion, rearing, and total activity was
significantly decreased in both the 0.75 mg/kg and 11.3 mg/kg dose groups at 2 months of age.
Another development study exposed CD-I mice to 6 mg/kg/day PFOS from GD 12-GD 18 and
assessed neuromotor maturation with surface righting reflex, open-field test, and rotarod test
(Fuentes et al., 2007a). Surface righting reflex was delayed at PND 4 and PND 8. Significant
effects were also observed during the climb test, with PFOS exposure resulting in diminished
resistance to backwards pull and reduced climb ability at PND 10 and PND 11 but not PND 12.
Climbing ability and forelimb grip strength was reduced with PFOS exposure at PND 11 but not
PND 10 or PND 12. The authors state that these transient effects may support delayed
neuromotor maturation due to gestational PFOS exposure. However, no effects were observed
with the open-field or rotarod tests at 3 months of age.
A short-term study reported that male CD-I mice displayed increased anxiety-like behavior in
the open-field test, as seen by decreased time in the center of the chamber in the 3 mg/kg/day
PFOS group and decreased vertical activity in the 6 mg/kg/day group (Fuentes et al., 2007b).
However, in a developmental study by the same authors, no effects on anxiety-like behavior
were observed in CD-I mice exposed to 6 mg/kg/day PFOS from GD 12-GD 18 (Fuentes et al.,
2007a). Similarly, no effects on this behavior were observed in a single-dose study in male
NMRI mice dosed with 0.75 mg/kg or 11.3 mg/kg PFOS (Johansson et al., 2008).
C-122
-------
APRIL 2024
Table C-7. Associations Between PFOS Exposure and Neurobehavioral Effects in Rodents
Reference
Study Design
Learning and
Memory
Acoustic
Startle
Anxiety-like Behavior
Motor
Activity/ Coordination
Neuromaturation
Mice
Fuentes et al.
(2007a)a
Developmental
exposure (GD12-18) to
0 or 6 mg/kg/day
NT
NT
Open field: No effect
Open field: No effect
Rotarod: No effect
Surface righting reflex: J,
at 6 mg/kg/day
Grip strength: J, at
6 mg/kg/day
Mshaty et al. (2020)b
Developmental
exposure (PND1-14) to
0,0.1, 0.25, or
1 mg/kg/day
Object location and
recognition test, and
pairwise visual
discrimination task:
| at 1 mg/kg/day
NT
NT
NT
NT
Johansson et al.
Single dose (PND10) to
Spontaneous
NT
Elevated plus maze: No
Spontaneous behavior,
NT
(2008)b
0, 0.75, or 11.3 mg/kg
behavior,
effect
total activity: j at
habituation: I at
>0.75 mg/kg in first
11.3 mg/kg
test block; f at
11.3 mg/kg in final test
block
Fuentes, et al.
Short-term exposure to
Morris water maze
NT
Open field, time in
Open field: No effect
NT
(2007b)b
0, 3, or 6 mg/kg/day
(acquisition): no
center: I at
effect
3 mg/kg/day;
Rotarod: No effect0
vertical activity: j at
Morris water maze
6 mg/kg/day
Morris water maze
(probe): | at
(probe), swimming
3 mg/kg/day
speed: t at
> 3 mg/kg/day; distance
traveled: t at
6 mg/kg/day
Long et al. (2013)d
Subchronic exposure
Morris water maze
NT
NT
NT
NT
(3 mo) to 0,0.43,2.15,
(acquisition, probe):
or 10.75 mg/kg/day
i at >2.15 mg/kg/day
C-123
-------
APRIL 2024
Reference
Study Design
Learning and
Memory
Acoustic
Startle
Anxiety-like Behavior
Motor ,T
i m. i ft Neuromaturation
Activity/ Coordination
Rats
Wang et al. (2015a)e
Developmental
exposure (gestational
and/or lactational) to 0,
5, 15 mg/L (0, 0.8, or
2.4 mg/kg/dayf)
Morris water maze
(acquisition, probe):
i at 15 mg/mL
NT
NT
Morris water maze, NT
swimming speed: No
effect
Butenhoff et al.
(2009)a
Developmental
exposure (GD 0-PND
20) to 0,0.1, 0.3, or
1.0 mg/kg/day
Males, habituation: J,
at 1 mg/kg/day
Biel swimming
maze: No effect
No effect
NT
Males, motor activity: \ NT
at 0.3 mg/kg/day
Females: No effect
Luebker et al.
(2005a)a
Reproductive exposure
(GD 0-PND 112) to
0.0,0.1,0.4, 1.6, or
3.2 mg/kg/day
Modified M-maze:
No effect
Passive avoidance:
No effect
NT
NT
NT NT
Notes: GD = gestation day; NT = not tested; PND = postnatal day.
aMales and females analyzed separately.
b Study conducted in males.
cNo quantitative data were presented for this endpoint, which was consequently rated as low confidence.
d Sexes combined.
e Sex was not specified.
fDoses in mg/kg/day were derived and presented in the 2016 PFOS HESD (U.S. EPA, 2016c).
C-124
-------
APRIL 2024
Several short-term studies in mice and rats (Salgado et al., 2016; Lopez-Doval et al., 2015;
Salgado et al., 2015), one developmental study in mice (Mshaty et al., 2020), and one subchronic
study in mice (Long et al., 2013) examined the effects of PFOS on neurotransmitter levels (Table
C-8). Glutamine, glycine, and serotonin were each examined in only one study. Neither
glutamine nor glycine were altered in the dorsal hippocampus of male C57BL/6J mice exposed
to 1 mg/kg/day PFOS from PND 1-PND 14 (Mshaty et al., 2020). Serotonin was increased in the
anterior hypothalamus, mediobasal hypothalamus, and the median eminence of male Sprague-
Dawley rats dosed with 0.5 mg/kg/day-6 mg/kg/day for 28 days (Lopez-Doval et al., 2015). The
effect of PFOS on dopamine and/or gamma-aminobutyric acid (GAB A) in various brain regions
was examined in three studies (Mshaty et al., 2020; Salgado et al., 2015; Long et al., 2013). A
subchronic study found no changes in GABA in the hippocampus of male C57BL6 mice dosed
with 0.43-10.75 mg/kg/day PFOS (Long et al., 2013). However, GABA was increased in the
dorsal hippocampus of male C57BL/6J mice exposed to 1 mg/kg/day PFOS from PND 1-PND
14 (Mshaty et al., 2020). In adult male Sprague-Dawley rats dosed with 3 and 6 mg/kg/day
PFOS for 28 days, GABA was unaltered in the mediobasal hypothalamus and increased in the
anterior hypothalamus in both dose groups (Salgado et al., 2015). In male Sprague-Dawley rats
dosed with 0.5 mg/kg/day-6 mg/kg/day PFOS for 28 days, dopamine was increased in the
hippocampus in the 0.5 mg/kg, 1 mg/kg, and 3 mg/kg groups, but not the 6 mg/kg/day group
(Salgado et al., 2016). Increased dopamine levels were also detected in the prefrontal cortex of
the 1 mg/kg/day group only and in the anterior hypothalamus of the 3 mg/kg/day and
6 mg/kg/day groups (Salgado et al., 2016; Salgado et al., 2015). No changes in dopamine levels
were seen in the mediobasal hypothalamus (Salgado et al., 2015). In male C57BL6 mice dosed
with 0.43 mg/kg/day-10.75 mg/kg/day PFOS, dopamine in the caudate putamen was decreased
only at the highest dose (Long et al., 2013). In this study, glutamate in the hippocampus was also
increased at the highest dose. However, glutamate was increased in the dorsal hippocampus of
male C57BL/6J mice exposed to 1 mg/kg/day PFOS from PND 1-PND 14 (Mshaty et al., 2020).
Greater sensitivity of the developing brain to PFOS exposure might explain why glutamate
increases in the hippocampus were only seen at higher doses in the Long et al. (2013) study
compared with increases seen at a lower dose in the Mshaty et al. (2020) study.
C-125
-------
APRIL 2024
Table C-8. Associations Between PFOS Exposure and Neurotransmitters in Rodents
Reference
Study Design
Glutamine/
Glutamate
Glycine
Serotonin
GABA
Dopamine
Mice
Mshaty et al. (2020)a
Developmental exposure
(PND1-14) to 0 or
1 mg/kg/day
Dorsal hippocampus,
glutamate: t at
1 mg/kg/day
Dorsal hippocampus,
glutamine: No effect
Dorsal
hippocampus:
No effect
NT
Dorsal
hippocampus: t at
1 mg/kg/day
NT
Long et al. (2013)b
Subchronic exposure
(3 mo) to 0,0.43, 2.15,
or 10.75 mg/kg/day
Hippocampus,
glutamate: t at
10.75 mg/kg/day
NT
NT
Hippocampus: No
effect
Caudate putamen: I
at 10.75 mg/kg/day
Rats
NT NT Mediobasal Mediobasal
hypothalamus: hypothalamus:
No effect No effect
Anterior Anterior
hypothalamus: | at hypothalamus: | at
>3 mg/kg/day >3 mg/kg/day
Salgado et al. (2016)a Short-term exposure NT NT NT NT Amygdala: No
(28 d) to 0, 0.5, 1, 3, or effect
6 mg/kg/day Prefrontal cortex: f
at 1 mg/kg/day but
not at 3 and
6 mg/kg/day
Hippocampus: t at
0.5, 1, and
3 mg/kg/day but not
at 6 mg/kg/day
Salgado et al. (2015)a Short-term exposure NT
(28 d) to 0, 3, or
6 mg/kg/day
C-126
-------
APRIL 2024
Reference
Study Design
Glutamine/
Glutamate
Glycine
Serotonin
GABA
Dopamine
Lopez-Doval et al.
(2015)a
Short-term exposure
(28 d) to 0, 0.5, 1, 3, or
6 mg/kg/day
NT
NT
Mediobasal NT
hypothalamus: t at
>0.5 mg/kg/day
Anterior
hypothalamus: t at
>0.5 mg/kg/day
Median eminence: t
at >0.5 mg/kg/day
NT
Notes: PND = postnatal day; d = days; NT = not tested.
a Study conducted in males.
b Sexes combined.
C-127
-------
APRIL 2024
Synaptic transmission and plasticity were assessed in one electrophysiology study in SD rats
exposed to 0.35 mg/kg/day-2.17 mg/kg/day PFOS throughout development until PND 90
(Zhang et al., 2019). Zhang et al. (2019) observed moderate inhibition of paired pulse facilitation
(at highest dose) and the input/output curve (at all doses) in the hippocampus. Long-term
potentiation was also decreased in a dose-dependent manner in the 0.72 mg/kg/day and
2.17 mg/kg/day dose groups.
C.4.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse nervous outcomes is discussed in
Sections 3.2.4, 3.2.5, 3.2.6, 3.3.4, 3.3.6, and 3.4.1.4 of the 2016 PFOS HESD (U.S. EPA, 2016c).
There are 54 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD that investigated the mechanisms of action of PFOS that
lead to nervous effects. A summary of these studies is shown in Figure C-30. Additional
mechanistic synthesis will not be conducted since evidence suggests but is not sufficient to infer
that PFOS leads to nervous effects.
Mechanistic Pathway Animal Human In Vitro Grand Total
Big Data, Non-Targeted Analysis
1
0
0
1
Cell Growth, Differentiation, Proliferation, Or Viability
8
0
25
31
Cell Signaling Or Signal Transduction
12
0
21
29
Extracellular Matrix Or Molecules
0
0
1
1
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
2
0
1
3
Hormone Function
7
0
5
11
Inflammation And Immune Response
1
1
5
6
Oxidative Stress
1
0
10
11
Xen obi otic Metabolism
0
0
1
1
Other
2
0
1
3
Not Applicable/Not Specified/Review Article
a
0
0
3
Grand Total
26
1
32
54
Figure C-30. Summary of Mechanistic Studies of PFOS and Nervous Effects
Interactive figure and additional study details available on HAWC.
C.4.4 Evidence Integration
There is slight evidence of an association between PFOS and nervous system effects in humans
based on the mostly mixed results. There were no new neurological studies identified that
C-128
-------
APRIL 2024
evaluated cerebral palsy. Outcomes investigated include depression, memory impairment,
hearing impairment, ASD, and ID.
Epidemiological studies in this current review provide limited indication of adverse effects of
PFOS on neurodevelopment or neuropsychological outcomes (Niu et al., 2019; Jeddy et al.,
2017; Chen et al., 2013), cognitive development (Oulhote et al., 2019; Harris et al., 2018), and
executive function (Vuong et al., 2016) in human populations. No adverse effects were observed
for PFOS and depression or memory impairment, and only one study indicated effects of PFOS
on hearing impairment (Li, 2020), however the number of studies was limited. Overall, results
from studies of neurodevelopmental, neuropsychological, and cognitive outcomes were mixed.
The recent studies provide limited indication of adverse effects of PFOS on behavioral problems,
ADHD, ASD, and ID. The studies reviewed provide some indication of behavioral problems
associated with PFOS (Oulhote et al., 2019; Ghassabian et al., 2018; Oulhote et al., 2016),
however overall results were mixed. Of the multiple studies examining associations between
PFOS and ADHD, only one (Lenters et al., 2019) reported a significant relationship between
PFOS and ADHD, with results indicating heterogeneity with respect to gender. No adverse
associations of ID with PFOS were reported in the single study reviewed (Lyall et al., 2018).
There was an indication of a potential relationship between PFOS and autistic behaviors or ASD
diagnosis in some studies (Shin et al., 2020; Oulhote et al., 2016; Braun et al., 2014). However,
many studies have methodological concerns, as PFOS exposures in cases and controls within the
ADHD and ASD studies were often either similar to or had mean control exposures greater than
cases in some studies. A single category outcome for ASD may also not adequately encompass
the heterogeneity in terms of developmental history, intelligence, comorbidity, and severity that
might be important in accurately revealing associations. The current evidence examining PFOS
exposure and neurodevelopmental disorders in children, including ADHD and learning
disabilities, is limited.
The animal evidence for an association between PFOS and neurological effects is moderate.
There are several medium confidence studies available where changes in neurobehavioral effects
were observed. Although the studies varied by design, endpoints measured, and methods of
measurement leading to some inconsistencies across studies, there is evidence of effects on
learning and memory. Of the studies available in animal models, no effects were noted for brain
weight with limited changes observed for histopathology. Some neurobehavioral effects were
observed but these results and the methods used to quantify them were relatively inconsistent.
Alterations in neurotransmitter levels and synaptic transmission and plasticity were also
observed, though it is often unclear what magnitude of change in neurotransmitters levels can be
considered adverse. Notably, Mshaty et al. (2020) observed dose-dependent effects of PFOS in
both the object recognition memory test and object location recognition memory test, as well as
dose-dependent effects of PFOS across 9 days of a visual discrimination task. These behavioral
changes in the 1 mg/kg/day dose group were accompanied by significant increases in
hippocampal neurotransmitter concentrations, including glutamate and GAB A. Increased
hippocampal glutamate levels may cause excitotoxicity which could explain the spatial learning
deficits seen by Mshaty et al. (2020). Importantly, the exposure period in this study encompassed
a sensitive period of neurodevelopment (i.e., lactation) and the observed effects occurred at
relatively low doses. In addition, the deficits in spatial learning and increased hippocampal
C-129
-------
APRIL 2024
glutamate concentrations observed by Long et al. (2013) in PFOS-exposed adult mice support
these results.
C.4.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause nervous system effects
in humans under relevant exposure circumstances (Table C-9). This conclusion is based
primarily on alterations in neurodevelopment, neurobehavior, and neurotransmitter levels in
animals following exposure to doses as low as 0.5 mg/kg/day PFOS. Although there is some
evidence of adverse effects of PFOS exposure on neurodevelopment or neuropsychological
outcomes, cognitive development, executive function and behavioral problems in humans, there
is considerable uncertainty in the results due to inconsistency across studies and limited number
of studies.
C-130
-------
APRIL 2024
Table C-9. Evidence Profile Table for PFOS Nervous System Effect
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.4.1)
Neurodevelopment
2 High confidence study
4 Medium confidence
studies
1 Low confidence study
Inverse associations were
reported for
neurodevelopmental
outcomes in three studies
of children (3/7). One
high confidence study
observed a significant
inverse association with
social measures among
girls. Of the medium
confidence studies, one
observed significant
inverse associations with
total neurodevelopment
and motor skill measures.
Another study reported
significant inverse
associations with
communication measures
but a positive association
with cognition. The same
study reported
inconsistent effects when
stratified by maternal age.
Results reported in the
remaining studies were
inconsistent.
1 High and medium
confidence studies
• Low confidence studies
• Lnconsistent direction of
effects across studies
©OO
Slight
Evidence of nervous
system effects is based on
high confidence studies
reporting significant
associations, which varied
in magnitude and were
inconsistent across
neurological outcomes.
Uncertainties remain due
to inconsistent findings
within studies and mixed
findings across studies.
Studies with mixed
findings were primarily of
medium or low
confidence.
Cognitive function
1 High confidence
studies
9 Medium confidence
studies
Reported results were
largely inconsistent
across studies, with both
positive and inverse non-
significant associations
1 High and medium
confidence studies
©OO
Evidence Suggests
Primary basis:
Animal evidence indicated
alterations in
neurodevelopment,
neurobehavior, and
neurotransmitter levels.
Although there is some
evidence of adverse
effects of PFOS exposure
on neurodevelopment or
neuropsychological
outcomes, cognitive
development, executive
function and behavioral
problems in humans, there
is considerable uncertainty
in the results due to
inconsistency across
studies and limited
number of studies.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
• Lnconsistent direction of
effects across studies
• Small magnitude of
effect
C-131
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
reported. One high
confidence study
observed non-
significantly increased
nonverbal IQ scores
among the highest
exposure group. Positive
associations with reading
scores were observed in
some medium confidence
studies (2/9).
Social-emotional and
behavioral regulation
1 High confidence study
4 Medium confidence
studies
1 Low confidence study
One high confidence
study found no
significant associations
with behavioral measures
at age 5 but observed a
positive association
among females and a
negative association
among males at age 7. Of
the medium confidence
studies, two observed
positive associations with
behavioral difficulties
(2/4). Another medium
confidence study
observed that the
association with
impulsivity was modified
by sex, with males
performing better than
females (1/4). One low
confidence study of
adolescents observed a
significant inverse
• High and medium
confidence studies
• Low confidence study
• Lnconsistent direction of
effects across and within
studies
C-132
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
correlation with the
region of the brain
associated with impulsive
behavior.
Depression
One medium confidence • Medium confidence • Low confidence study
3 Medium confidence
study reported positive studies
studies
but non-significant results
1 Low confidence study
for depression in the
general population adults.
Another medium
confidence study
explored depression in
children followed for
20 yr, reporting no
association. An additional
study of medium
confidence reported no
association with
depression among
pregnant women. A low
confidence study reported
no association.
Executive function
3 Medium confidence
studies
Two medium confidence
studies examined
executive function
measures, including
behavior regulation and
metacognition, among
children from the HOME
study (2/3). One of these
studies reported
significantly inversed
associations with
executive function
measures, while the other
• Medium confidence
studies
• Inconsistent direction of
effects across studies
• Limited number of
studies examining
outcome
C-133
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
reported no significant
associations. A medium
confidence study of
adults did not observe
significant associations.
Attention
5 Medium confidence
studies
2 Low confidence
studies
Studies examining
measures of attention
reported mixed findings.
One medium confidence
study reported
significantly increased
odds of ADHD. When
stratified by child sex,
significant effects
remained. The remaining
medium confidence
studies (4/5) did not
report significant
associations.
Additionally, the two low
confidence studies
observed no associations
with measures of
attention.
• Medium confidence
studies
• Low confidence studies
• Lnconsistent direction of
effects across studies
Autism, autistic
behaviors, and
intellectual disability
1 High confidence study
5 Medium confidence
studies
One high confidence
study observed a positive
association with autism
scores when measured at
age 7. When stratified by
sex, higher scores were
observed in females
compared with males.
Findings from the five
medium confidence
studies were mixed. Two
1 High and medium
confidence studies
• Lnconsistent direction of
effects across studies
C-134
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
studies observed positive
associations, with one
study reporting
associations for the
overall study population
and the other study
reporting the association
only among males.
Another medium
confidence study reported
inverse associations.
Other reported results
were not significant.
Visuospatial
performance
1 High confidence study
1 Medium confidence
studies
Two studies examined
visuospatial performance
effects among children.
One high confidence
study among children
observed a significant
inverse association with
visual-motor performance
across quartiles of
exposure. The medium
confidence study reported
no association with
visuospatial
performance.
• High and medium
confidence studies
• Large magnitude of
effect
• Inconsistent direction of
effects across studies
• Limited number of
studies examining
outcome
Memory impairment
2 Medium confidence
studies
Two studies reported
associations with memory
loss among adult
populations. One medium
confidence study
observed a significant
inverse association with
memory impairment. No
• Medium confidence
studies
• Inconsistent direction of
effects across studies
• Small magnitude of
effect
• Limited number of
studies examining
outcome
C-135
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
significant effects were
reported from the
remaining medium
confidence study.
Hearing impairment
2 Medium confidence
studies
Two medium confidence
studies examined hearing
impairment among adults
from NHANES. One
study observed positive
correlations with hearing
impairment, while the
other reported no
associations.
• Medium confidence
study
• Large magnitude of
effect
• Inconsistent direction of
effects across studies
• Limited number of
studies examining
outcome
Evidence From In Vivo Animal Studies (Section C.4.2)
Neurobehavior
4 Medium confidence
studies
Changes in
neurobehavior endpoints
were altered and
decreases in learning and
memory tasks were
largely consistent among
studies (3/4). Motor
activity was found to be
increased (2/2), with
anxiety-like behavior
being decreased (1/1). A
single study measured
acoustic startle and found
no changes (1/1).
• Medium confidence
studies
• Coherence of
findings in
neurotransmitters
• Limited number of
studies examining
specific outcomes
• Inconsistent direction of
effects
Neurotransmitters
3 Medium confidence
studies
Changes in
neurotransmitter levels in
short-term studies in male
mice included a dose-
responsive increase in
serotonin (1/1) and
• Medium confidence
studies
• Coherence of
findings in
neurobehavior
endpoints
• Limited number of
studies examining
outcome
• Biological significance
of the magnitude of
effect is unclear
0©O
Moderate
Several medium
confidence studies are
available where changes
in neurobehavioral effects
were observed. Although
the studies varied by
design, endpoints
measured, and methods of
measurement leading to
some inconsistencies
across studies, there is
evidence of effects on
learning and memory. No
effects were noted for
brain weight and limited
changes were observed
for histopathology.
Alterations in
C-136
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
region-specific decreases
of GABA (1/1) and
dopamine (2/2).
• Dose-response
relationship
Organ weights
3 Medium confidence
studies
No effects were observed
on absolute brain weights
(2/2). One study reported
a significant increase in
relative brain weights;
however, this increase
was confounded by a
reduction in body weight.
• Medium confidence
studies
• Limited number of
studies examining
outcomes
• Confounding variables
such as decreases in
body weights
Histopathology
1 High confidence study,
1 Medium confidence
study
One study found no
effects on brain
histopathology in male
and female rats, whereas
some phagocytosis in the
brains of PFOS-exposed
mice was noted in
another study.
• High and medium
confidence studies
• Limited number of
studies examining
outcomes
Electrophysiology
1 Medium confidence
study
One developmental study
in male and female rats
found inhibition of paired
pulse facilitation and the
input/output curve in the
hippocampus.
Hippocampal long-term
potentiation was also
decreased (1/1).
• Medium confidence
study
• Dose-response
relationship
• Limited number of
studies examining
outcome
neurotransmitter levels
and synaptic transmission
and plasticity were also
" observed, though it is
often unclear what
magnitude of change in
neurotransmitters levels
can be considered
adverse.
Notes: yr = years; ADHD =
IQ = intelligence quotient;
attention deficit/hyperactivity disorder; GABA = gamma-aminobutyric acid; HOME = Health Outcomes and Measures of the Environment;
NHANES = National Health and Nutrition Examination Survey.
C-137
-------
APRIL 2024
C.5 Renal
EPA identified 19 epidemiological and 12 animal studies that investigated the association
between PFOS and renal effects. Of the epidemiological studies, 17 were classified as low
confidence and two were considered uninformative (Section C.5.1). Of the animal studies, two
were classified as high confidence, eight as medium confidence, and two were considered low
confidence (Section C.5.2). Studies may have mixed confidence ratings depending on the
endpoint evaluated. Though low confidence studies are considered qualitatively in this section,
they were not considered quantitatively for the dose-response assessment (see Toxicity
Assessment, (U.S. EPA, 2024)).
C.5.1 Human Evidence Study Quality Evaluation and Synthesis
C.5.1.1 Introduction
PFOS has the potential to affect the kidney's function given the saturable resorption from the
renal tubules (U.S. EPA, 2016c). Biomarkers of renal function include blood urea nitrogen
(BUN), estimated glomerular filtration rate (eGFR), serum creatinine, and uric acid. eGFR is a
marker of non-malignant renal disease.
The 2016 PFOS HESD (U.S. EPA, 2016c) concluded there was evidence of a suggestive
association between PFOS and chronic kidney disease (CKD; defined as glomerular filtration
rate (GFR) < 60 mL/min/1.73 m2) based on two studies on the general population (Shankar et al.,
2011; Steenland et al., 2010) and two on children (Geiger et al., 2014b; Watkins et al., 2013);
however, given the cross-sectional study designs, the potential for reverse causality could not be
ruled out.
For this updated review, 19 studies examined the association between PFOS and renal health
outcomes. Five studies were in children and adolescents (Khalil et al., 2018; Qin et al., 2016;
Kataria et al., 2015; Predieri et al., 2015; Geiger et al., 2013), one study in pregnant women
(Nielsen et al., 2020), one study in occupational workers (Rotander et al., 2015), and the
remainder of the studies were in the general population. Fifteen of the studies utilized a cross-
sectional study design; the remaining study designs included one case-control study (Predieri et
al., 2015), and three cohorts (Nielsen et al., 2020; Blake et al., 2018; Conway et al., 2018)
(Appendix D). All studies measured PFOS in blood components (i.e., plasma or serum). Two
studies conducted in China investigated the same population from the Isomers of C8 Health
Project (Wang et al., 2019b; Zeng et al., 2019c). Among the studies investigating populations in
the United States, five studies utilized data from the NHANES (Scinicariello et al., 2020b; Jain
and Ducatman, 2019a, c; Kataria et al., 2015; Geiger et al., 2013). Outcomes evaluated in these
studies include clinical conditions, such as CKD and gout and biomarkers of renal function,
including uric acid, eGFR, albumin, and creatinine.
C.5.1.2 Study Quality
Several considerations were specific to evaluating the quality of studies examining kidney
function and kidney disease. Since PFOS is removed from the blood by the kidney, cross-
sectional analyses using serum PFOS as the exposure measure are problematic if individuals
with compromised kidney function are included: PFOS concentrations could be increased in
C-138
-------
APRIL 2024
those individuals and an apparent association with GFR would be observed, even if one did not
exist (Dhingra et al., 2017).
There are 19 studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and renal effects. Study quality evaluations for these 19 studies are shown in
Figure C-31.
Of the 19 studies identified since the 2016 assessment, 17 studies were classified as low
confidence and the remaining two as uninformative (Seo et al., 2018; Predieri et al., 2015). No
studies were classified as high or medium confidence. The main concerns with the low
confidence studies included potential for residual confounding, selection bias, and reverse
causality. Another concern included small sample sizes (Nielsen et al., 2020; Khalil et al., 2018).
Additionally, low confidence studies utilizing cross-sectional analyses of kidney function with
serum PFOS were impacted by the potential for reverse causation.
Deficiencies identified in Predieri et al. (2015) included a small sample size and narrow ranges
of exposures which contributed to an uninformative rating. Seo et al. (2018) presented bivariate
correlations between PFOS exposure and renal outcomes, limiting the ability to interpret the
results. Other potential sources of bias were identified, including a lack of information on
participant recruitment and selection, unexplained discrepancies in samples sizes, and missing
details on outcome assessment methods. Neither uninformative study adjusted for key
confounders (e.g., age and SES), resulting in a high potential for residual confounding.
C-139
-------
Arrebola et al., 2019, 5080503
Blake etal., 2018, 5080657
Chen et al., 2019, 5387400
Conway et al., 2018, 5080465
Geigeret al., 2013, 2919148
Jain and Ducatman, 2019, 5080378
Jain and Ducatman, 2019, 5381566
Katariaetal., 2015, 3859835
Khalil et al., 2018,4238547
Linet al., 2013, 2850967
Liuet al., 2018,4238514
Nielsen et al., 2020, 6833687
Predieri etal., 2015, 3889874
Qinet al., 2016, 3981721
Rotanderet al., 2015, 3859842
Scinicariello et al., 2020, 6833670
Seo et al., 2018, 4238334
Wang etal., 2019, 5080583
Zeng et al., 2019, 5918630
APRIL 2024
,c®
1
+
1
+
1
+
++
I
i
+
i
+
+
¦
++
+
++
+
+
-
+
+*
++
+
+
+
+
-
+
-
++
+
+
+
+
-
++
-
+
+
++
+
+
-
+
-
+
+
-
+
+
-
+
+
¦
+
-
+
+
-
+
-
+
+
++
+
+
-
-
+
+
-
+
+
-
-
"
+
-
-
+
+
+
-
+
+
+
+
+
+
+
-
+
+
++
-
-
+
+
¦
+
+
-
-
+
+
~
+
-
+
+
+
+
+
-
-
+
¦
+
+
+
-
-
+
-
+*
+
+
+
+
-
-
+
-
B
-
-
n
+
-
+
+
+
+
+
~
+
-
++
+
++
+
+
~
Legend
D
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
b
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-31. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Renal Effects
Interactive figure and additional study details available on HAWC.
C-140
-------
APRIL 2024
C.S.I.3 Findings From Children and Adolescents
Three low confidence studies reported on uric acid among children and adolescents (Qin et al.,
2016; Kataria et al., 2015; Geiger et al., 2013) with two also reporting on hyperuricemia (Qin et
al., 2016; Geiger et al., 2013), defined as serum uric acid levels > 6 mg/dL. The three studies
reported mixed results. Among adolescents aged 12 to 18 years from NHANES (1999-2008),
Geiger et al. (2013) observed statistically significant positive associations between increasing
quartiles of PFOS and hyperuricemia (p-trend = 0.0221), and serum uric acid (p-trend = 0.0575).
An overlapping NHANES (2003-2010) study (Kataria et al., 2015) also reported a positive
association with uric acid levels among adolescents, where the highest PFOS quartile
(> 19.4 ng/mL) was associated with a 0.19 mg/dL (95% CI: 0.032, 0.34 mg/dL, p < 0.05)
increase in uric acid levels compared with the lowest PFOS quartile (<7.9 ng/mL). Qin et al.
(2016) did not observe significant associations for hyperuricemia or uric acid in children aged 12
to 15 years from the GBCA in Taiwan.
One low confidence study (Kataria et al., 2015) reported on GF in children aged 12 to 19 years
from NHANES (2003-2010). Significant negative associations were observed for eGFR in the
second, third, and fourth quartiles of PFOS exposure compared with the lowest quartile.
Two low confidence studies and one uninformative study investigated serum creatinine among
children and adolescents (Khalil et al., 2018; Kataria et al., 2015; Predieri et al., 2015). One low
confidence study (Kataria et al., 2015) on NHANES (2003-2010) adolescents (12-19 years old)
reported a significant positive association with serum creatinine in the third and fourth quartiles
of PFOS exposure. One low confidence study (Khalil et al., 2018) examined serum creatinine
levels among obese children aged 8 to 12 years, but no significant effect was observed.
C.5.1.4 Findings From the General Adult Population
Two low confidence studies examined CKD in the general population (Wang et al., 2019b;
Conway et al., 2018) and both observed positive associations. CKD was defined as an eGFR of
< 60 mL/min/1.73 m2. In C8 Health Project participants, Conway, 2019, 5080465 observed
significantly elevated odds of CKD among non-diabetic participants; a negative association was
observed among participants with diabetes. The prevalence of CKD in the diabetic population
was higher (22%) than the non-diabetic population (7%). Wang et al. (2019b) observed non-
significantly elevated odds of CKD in participants from the Isomers of C8 Health Project in
China. However, a concern for reverse causality makes interpretation of the results difficult in
both studies.
Gout was examined in one low confidence study (Scinicariello et al., 2020b) in adults from
NHANES (2009-2014). Positive associations were observed between serum PFOS and self-
reported gout, however, none were significant.
Six low confidence general population studies (Scinicariello et al., 2020b; Arrebola et al., 2019;
Chen et al., 2019a; Jain and Ducatman, 2019a; Zeng et al., 2019c; Lin et al., 2013) and one low
confidence occupational study (Rotander et al., 2015) examined PFOS and uric acid levels, and
three of those studies evaluated uric acids as they pertained to hyperuricemia (Scinicariello et al.,
2020b; Arrebola et al., 2019; Zeng et al., 2019c).
C-141
-------
APRIL 2024
A low confidence NHANES (2009-2014) study (Scinicariello et al., 2020b) observed
significantly elevated serum uric acid across increasing PFOS exposure quartiles, and the trend
was significant (p-trend = 0.003). Higher odds of hyperuricemia among participants in the
highest exposure quartile (>11.90 ng/mL) compared with the lowest (< 4.43 ng/mL) was also
observed, but the trend was not significant (p-trend = 0.15). Results were similar when restricted
to participants without CKD. Another low confidence study (Zeng et al., 2019c) on participants
from the Isomers of C8 Health Project reported significantly elevated uric acid levels with
increasing PFOS exposure, and a marginally significant association (OR: 1.17, 95% CI: 0.99,
1.39, p = 0.074) for hyperuricemia., Jain and Ducatman (2019a) examined uric acid by
glomerulation stage among NHANES (2007-2014) participants. For males, positive associations
with uric acid were observed for stages GF-1 (p < 0.01) and GF-2 (p = 0.05), but the effect was
negative for stages GF-3A (p = 0.66) and GF-3B/4 (p < 0.01). For females, all associations were
positive across stages of GF with significant associations (p < 0.05) for GF-1 and GF-3A. Two
low confidence studies did not observe associations with plasma uric acid in Croatian adults aged
44-56 years (Chen et al., 2019a), or in adolescents and young adults aged 12-30 years in the
Young Taiwanese Cohort Study (Lin et al., 2013). Another low confidence study (Arrebola et al.,
2019) using pooled cohort data (the BIOAMBIENT.ES study) observed a non-significant
increase in serum uric acid with increasing PFOS.
One low confidence occupational study examined serum uric acid levels among firefighters with
past exposure to aqueous film-forming foam (AFFF) (Rotander et al., 2015). No significant
association was observed for serum uric acid and increasing PFOS exposure.
Two general population studies evaluated PFOS and eGFR (Wang et al., 2019b; Blake et al.,
2018). A low confidence study (Blake et al., 2018) assessed participants of the FCC with high
exposure to PFAS from their household water supplies. A significant inverse association with
eGFR was observed in the latent effects mixed effect model (LME), but not in the repeated
measures LME. These results were consistent with the low confidence study (Wang et al.,
2019b) which assessed participants of the Isomers of C8 Health Project and observed negative
association between total PFOS serum concentrations and eGFR.
The evidence of association between PFOS and renal effects among pregnant women was
limited. Only one low confidence study reported on pregnant women (Nielsen et al., 2020) using
a small sample of women (n = 73) from the Pregnancy Obesity Nutrition and Child Health study
(PONCH) study. No significant Spearman rank correlations were reported between PFOS and
kidney function parameters.
Two studies examined albumin and creatinine as biomarkers for renal function (Chen et al.,
2019a; Jain and Ducatman, 2019c). The two low confidence studies provided differing
conclusions. Jain, 2019, 5381566 utilized NHANES (2005-2014) data and reported statistically
significant positive associations with serum and urine creatinine, and serum albumin.
Statistically significant negative associations were also reported with urine albumin and urine
albumin-creatinine ratios. Stratification by stages of GF was noted as better representing more
severe stages of renal failure. For PFOS, stratification by stages of GF had inconsistent effects.
One low confidence study (Chen et al., 2019a) did not observe significant associations with
plasma creatinine in Croatian adults ages 44-56 years.
C-142
-------
APRIL 2024
One low confidence study, Liu et al. (2018b) examined serum proteins among NHANES (2013-
2014) participants, and positive associations (p < 0.01) were observed for serum protein with
increasing PFOS exposure. The effect was consistent when stratified by linear and branched
PFOS.
C.5.2 Animal Evidence Study Quality Evaluation and Synthesis
There are four studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and eight studies from
recent systematic literature search and review efforts conducted after publication of the 2016
PFOS HESD that investigated the association between PFOS and renal effects. Study quality
evaluations for these 12 studies are shown in Figure C-32.
C-143
-------
APRIL 2024
Butenhoffetal., 2012, 1276144
Curran et al., 2008, 757871
Dong et a]., 2011, 1424949-
Fuentes et a!., 2006, 757859-
Kawamoto et al., 2011, 2919266-
Li etal., 2021, 7643501 -
NTP, 2019, 5400978-
++
++
NR
++
~
NR
++
+
NR
+
+
~
NR
-
+
NR
+
+
NR
%
++ ++ ++ ++
Seacat et al., 2002, 757853 -
++
+
+
H
+
Seacat etal., 2003, 1290852-
++ ++
NR
+
+
+
Thomford, 2002, 5432419-
+
-
NR
++
++
-
Xing etal., 2016, 3981506-
++
+
NR
++
++
++
Zhong et al., 2016, 3748828-
-
NR
NR
+
+
+
s
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
NR Not reported
* Multiple judgments exist
A Bias away from null
Figure C-32. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Renal Effects
Interactive figure and additional study details available on HAWC.
Few renal effects were observed across multiple studies assessing PFOS toxicity in animal
models. Most studies did not observe significant effects of PFOS exposure on kidney weight or
histopathology (Li et al., 2021b; Zhong et al., 2016; Dong et al., 2011; Peden-Adams et al., 2008;
Yahia et al., 2008; Fuentes et al., 2006; Seacat et al., 2003; Seacat et al., 2002). Flowever, two
subchronic studies in male mice reported significant decreases in relative kidney weight with
PFOS treatment for 30 days at the highest dose tested of 10 mg/kg/day (approximately 10%
C-144
-------
APRIL 2024
decrease) (Xing et al., 2016) and treatment for 60 days at doses of 0.83 mg/kg/day or
2.083 mg/kg/day (approximately 18% and 16% decreases, respectively) (Dong et al., 2009).
Neither of these studies reported absolute kidney weight and, in both studies, PFOS treatment
resulted in decreased body weight at these doses which precludes evaluation of the significance
of relative weights. One developmental study in mice reported no significant changes in maternal
relative or absolute kidney weight (Fuentes et al., 2006).
In contrast to the mouse studies, four short-term/sub chronic studies in male rats reported
significant increases in relative kidney weight at doses as low as 1.25 mg/kg/day (NTP, 2019),
5 mg/kg/day (Cui et al., 2009), 6 mg/kg/day (Goldenthal et al., 1978), and 6.34 mg/kg/day
(Curran et al., 2008). NTP (2019) observed an approximately 14% increase in relative kidney
weight at the highest dose tested (5 mg/kg/day) that occurred along with significantly decreased
body weight. Small but significant increases (approximately 8%) in relative kidney weight were
also observed at 1.25 and 2.5 mg/kg/day; however, no significant changes were observed in
absolute kidney weight at any dose level. While Cui et al. (2009) did not provide absolute kidney
weight data, no significant difference was observed in body weight in the 5 mg/kg/day dose
group; the study authors indicate that the increased relative kidney weight may be due to renal
hypertrophy. Body weight was affected in all other dose groups showing changes in relative
kidney weight in Goldenthal et al. (1978), Cui et al. (2009), and Curran et al. (2008). Curran et
al. (2008) also reported that absolute kidney weight and kidney weight relative to brain weight
were both significantly decreased in male rats exposed to 6.34 mg/kg/day, which also indicates
that the increase in relative kidney weight in that dose group was driven by decreased body
weight.
NTP (2019) also observed small but significant increases (approximately 9%) in relative kidney
weight in female rats at doses as low as 0.625 mg/kg/day, but the increase was not significant at
the highest dose tested (5 mg/kg/day). Curran et al. (2008) observed a significant increase in
relative kidney weight for female rats at doses as low as 3.73 mg/kg/day, but body weights were
significantly decreased in the same dose groups and there were no significant changes in absolute
kidney weight or kidney weight relative to brain weight. Similarly, a chronic study in female rats
reported significant increases in kidney weight relative to body weight with the highest dose
tested (1.25 mg/kg/day) but reported no change in kidney weight relative to brain weight at the
same dose, indicating these effects were also driven by the significant decreases in body weight
seen at this dose (Butenhoff et al., 2012).
Cui et al. (2009) observed altered kidney histopathology in male rats, including turbidness and
tumefaction in the epithelia of the proximal convoluted tubule, congestion in the renal cortex and
medulla, and enhanced cytoplasmic acidophilia, though only in the highest dose group
(20 mg/kg/day). Besides Cui et al. (2009), all other studies reported no treatment-related changes
in kidney histopathology (Li et al., 2021b; NTP, 2019; Xing et al., 2016; Butenhoff et al., 2012;
Curran et al., 2008; Yahia et al., 2008; Seacat et al., 2003).
Several studies also analyzed clinical chemistry endpoints relevant to renal toxicity. At the
highest dose tested in each study (1.3 mg/kg/day-5 mg/kg/day), Seacat et al. (2003) and NTP
(2019) (males only) both reported significant increases in BUN in rats after 14-week and 28-day
exposures, respectively. Similarly, Curran et al. (2008) observed a significant trend toward
increased serum urea in male rats exposed to doses up to 6.34 mg/kg/day for 28 days, although
no significant differences were detected between exposure groups. In an extension of the Seacat
C-145
-------
APRIL 2024
et al. (2003) study, Butenhoff et al. (2012) reported increased BUN in both males and females of
the high dose group (approximately 0.98 mg/kg/day and 1.25 mg/kg/day, respectively) at
27 weeks and significantly increased BUN in doses >0.1 mg/kg/day in males and
>0.3 mg/kg/day in females at 53 weeks. However, the studies that reported increased BUN did
not see concurrent increases in serum creatinine concentrations at the same dose levels and time
points (NTP, 2019; Butenhoff et al., 2012; Curran et al., 2008; Seacat et al., 2003); NTP (2019)
and Butenhoff et al. (2012) consider mild increases in BUN without increases in creatinine to be
more consistent with decreased water intake and mild dehydration rather than a direct
toxicological effect of chemical exposure, though these studies did not quantify water intake in
exposed animals. Additionally, increases in BUN were not seen in male mice treated with up to
10 mg/kg/day PFOS for 30 days (Xing et al., 2016) or in male or female monkeys treated with up
to 0.75 mg/kg/day PFOS for 26 weeks (Seacat et al., 2002). Other clinical chemistry endpoints,
including creatine kinase (NTP, 2019; Curran et al., 2008; Seacat et al., 2002), uric acid (Curran
et al., 2008), urinary N-acetyl-b-glucosaminidase (NAG) (Xing et al., 2016), and urinalysis
parameters including urine pH (Butenhoff et al., 2012; Curran et al., 2008; Seacat et al., 2003;
Seacat et al., 2002), were not widely assessed across multiple studies and either showed no
significant changes or inconsistent responses between studies.
C.5.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse renal outcomes in the
2016 PFOS HESD (U.S. EPA, 2016c). There are three studies from recent systematic literature
search and review efforts conducted after publication of the 2016 PFOS HESD that investigated the
mechanisms of action of PFOS that lead to renal effects. A summary of these studies is shown in Figure
C-33. Additional mechanistic synthesis will not be conducted since evidence suggests but is not
sufficient to infer that PFOS leads to renal effects.
C-146
-------
APRIL 2024
Mechanistic Pathway
Animal
In Vitro
Grand Total
Big Data, Non-Targeted Analysis
0
1
1
Cell Growth, Differentiation, Proliferation, Or Viability
2
2
2
Cell Signaling Or Signal Transduction
1
2
2
Extracellular Matrix Or Molecules
1
1
1
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
0
1
1
Inflammation And Immune Response
1
1
2
Oxidative Stress
0
1
1
Renal Dysfunction
1
1
2
Xen obi otic Metabolism
0
1
1
Grand Total
3
2
3
Figure C-33. Summary of Mechanistic Studies of PFOS and Renal Effects
Interactive figure and additional study details available on HAWC.
C.5.4 Evidence Integration
There is slight evidence for an association between PFOS exposure and renal effects in humans
based on observed effects on measures of renal function and kidney disease in 17 low confidence
studies. The 2016 PFOS HESD (U.S. EPA, 2016c) concluded there was evidence of a suggestive
association between PFOS and CKD. The epidemiological evidence in this review observed
positive associations between serum PFOS concentrations and CKD only in low confidence
studies (Wang et al., 2019b; Conway et al., 2018). There is suggestive evidence of associations
with decreased kidney function, although reverse causality (i.e., increases in serum
perfluoroalkyl levels could be due to a decrease in GF and shared renal transporters for
perfluoroalkyls and uric acid) cannot be ruled out. There were mixed results across the measures
of renal function. Results were more consistent for eGFR, in which inverse associations were
reported by two low confidence studies (Wang et al., 2019b; Blake et al., 2018). Regarding
hyperuricemia and uric acid levels, results varied across glomerular function and sex. Among
children, there were mixed results for associations with creatinine and uric acid. One low
confidence study reported a statistically significant decrease in eGFR in adolescents across PFOS
quartiles (Kataria et al., 2015). Additionally, given the limited evidence, conclusions cannot be
drawn between PFOS and renal effects among pregnant women and occupational workers.
The animal evidence for an association between PFOS exposure and effects on renal toxicity is
considered indeterminate based on 10 high or medium confidence animal studies. The renal
system does not appear to be sensitive to PFOS toxicity. Effects on kidney weight were
C-147
-------
APRIL 2024
inconsistent between species and mainly consisted of changes in relative kidney weights
occurring at relatively high doses where body weights were also decreased. These changes in
relative kidney weight are considered a reflection of changes in body weight rather than adverse
effect on the kidney. Additionally, changes in clinical chemistry parameters such as increased
BUN without further evidence of kidney dysfunction (e.g., increased serum creatinine) are not
generally considered adverse and may be more reflective of changes in water consumption than
effects on the kidney.
C.5.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause renal effects in humans
under relevant exposure circumstances (Table C-10). This conclusion is based primarily on
effects on measures of kidney function observed in studies in humans exposed to median PFOS
ranging from 3.5 ng/mL to 11.9 ng/mL. Although there is some evidence of negative effects of
PFOS exposure on CKD, there is considerable uncertainty in the results due to inconsistency
across studies, mixed findings, limited number of studies, and potential for reverse causation.
C-148
-------
APRIL 2024
Table C-10. Evidence Profile Table for PFOS Renal Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.5.1)
Uric acid
10 Low confidence
studies
Increases in uric acid were«
observed in both children
(3/3) and adults (4/7).
Significant increases in
uric acid were observed in
adults (2/7). Results were
consistently stratified by
CKD status, but the
direction of effect was less
consistent when stratified
by eGFR. Increases in uric
acid led to increased odds
of hyperuricemia in all
studies that assessed
hyperuricemia (5/5).
Consistent direction of* Low confidence studies
effects among children
and adults
Serum and urinary Significant increases in • No factors noted
biomarkers serum albumin were
5 Low confidence studies observed in adults (2/2),
while albumin was not
analyzed in children.
Creatinine was
significantly increased in
children (2/3), but two
studies in adults reported
inconsistent directions of
effect. A study in adults
from NHANES observed
significant positive
associations of serum
proteins with PFOS and
when linear and branched
PFOS were analyzed
separately.
• Low confidence studies
©OO
Evidence Suggests
Primary basis:
No evidence in animals and
human evidence indicted
effects on kidney function.
Although there is some
evidence of negative
effects of PFOS exposure
hyperuricemia, and on CKD'lhcrc is
decreased eGFR. In adults, considerable uncertainty in
studies found evidence of 1'1C rcsu'ls ^ue t0
increased albumin and totalinconsistency across
studies, mixed findings,
limited number of studies
©OO
Slight
All studies were of low
confidence, which found
evidence of decreased
kidney function in adults
and children, including
increased uric acid,
serum proteins, and
children studies reported
evidence of decreased
creatinine. Overall,
inconsistent findings in
direction of effect and
imprecision were observed
for most outcomes. The
limitation of only low
confidence studies, mixed
results, and risk of high
bias leaves uncertainty
regarding renal outcomes
and PFOS exposures.
and potential for reverse
causation.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
C-149
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Chronic kidney disease Two studies examined
2 Low confidence studies CKD in adults. Odds of
CKD was increased
among general population
adults (2/2), with one
reporting a significant
increase. The direction of
effect was not consistent
after stratification by
diabetes status.
• No factors noted
• Low confidence studies
• Limited number of studies
examining outcome
Glomerular filtration One study in children
rate reported significantly
4 Low confidence studies decreased eGFR in all
exposure groups (1/1). In
adults, significant
decreases in eGFR were
observed (2/2), but results
were less consistent after
stratification by sex.
Results in pregnant
women (1/1) were not
significant.
• Consistent direction of* Low confidence studies
effects
Gout
1 Low confidence study
No significant associations* No factors noted
were observed in the
overall study population,
or in analyses stratified by
CKD status.
• Low confidence study
• Limited number of studies
examining outcome
• Potential outcome
misclassification due to
self-reported outcome
Evidence From In Vivo Animal Studies (Section C.5.2)
Kidney weight Relative kidney weight • High and medium
2 High confidence studieswas increased in rats (3/4), confidence studies
7 Medium confidence mainly occurring at
studies relatively high dose levels
that also resulted in
• Lnconsistency of findings
across species
OOO
Lndeterminate
Evidence was based on 10
C-150
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
decreased body weight.
Relative kidney weight
was decreased in mice
(1/4) and absolute kidney
weight was decreased in
rats (1/4), both at dose
levels that also resulted in
decreased body weight.
One study in monkeys
reported no effects on
kidney weight.
Changes in body weight
may limit ability to
interpret these responses
Histopathology None of the studies that •
2 High confidence studiesexamined kidney
4 Medium confidence histopathology (0/6) found .
studies evidence of morphological
damage or exposure-
related lesions following
short-term, subchronic, or
chronic exposure to PFOS.
High and medium
confidence studies
Consistent effects
across study design,
sex, and species
• No factors noted
Serum biomarkers Serum BUN was increased«
2 High confidence studies(3/6) mainly at the highest
4 Medium confidence dose tested and only in rats
studies (1 study each in monkeys,
rats, or mice found no
effects on BUN). One high
confidence study with
chronic exposure observed
increased BUN in male
and female rats at several
timepoints throughout the
study with a dose response
evident in female rats after
53 wk of exposure. No
significant changes in
serum creatinine were
High and medium
confidence studies
high and medium
confidence studies. The
renal system does not
appear to be sensitive to
PFOS toxicity. Effects on
kidney weight were
inconsistent between
species and mainly
consisted of changes in
relative kidney weights
occurring at relatively high
-doses with body weights
also decreased. There were
no apparent exposure-
related changes observed
in kidney histopathology or
urinalysis endpoints.
Changes in clinical
chemistry parameters such
-as increased BUN without
»Incoherence of findings in further evidence of kidney
serum biomarkers of renal dysfunction (e.g., increased
function serum creatinine) are not
generally considered
adverse and may be more
reflective of changes in
water consumption than
effects on the kidney.
C-151
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
observed (5/5), including
all studies that observed
increased BUN. No
exposure-related changes
were observed for serum
uric acid (1/1) or creatine
kinase (2/2).
Urinalysis
1 High confidence study
4 Medium confidence
studies
No exposure-related «
changes were observed for
urinalysis endpoints (5/5).
Urine pH was increased or
decreased (2/5), but the
changes were not
exposure-related. One
subchronic study in mice
found no changes in
urinary N-acetyl-b-
glucosaminidase.
High and medium
confidence study
• No factors noted
Notes: BUN = blood urea nitrogen; CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; NHANES = National Health and Nutrition Examination Survey;
wk = weeks.
C-152
-------
APRIL 2024
C.6 Hematological
EPA identified eight epidemiological and five animal studies that investigated the association
between PFOS and hematological effects. Of the epidemiological studies, three were classified
as medium confidence, two as low confidence, and three were considered uninformative (Section
C.6.1). Of the animal studies, one was classified as high confidence, three as medium confidence,
one was considered low confidence (Section C.6.2). Studies may have mixed confidence ratings
depending on the endpoint evaluated. Though low confidence studies are considered qualitatively
in this section, they were not considered quantitatively for the dose-response assessment (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.6.1 Human Evidence Study Quality Evaluation and Synthesis
C.6.1.1 Introduction
The mechanisms for PFOS effects on hematological parameters might include immune
suppression, shifts in nutrients absorbed from the diet, or the influences related to other health
outcomes such as cardiometabolic or kidney dysfunction (Abraham et al., 2020; Jain, 2020a;
Chen et al., 2019a). PFOS has been implicated in endocrine disruption, which may affect vitamin
D homeostasis (Etzel et al., 2019). It could also alter epigenetics via DNA methylation (van den
Dungen et al., 2017). The effects of PFOS on hematological outcomes may differ by
characteristics such as age, gender, race, and genetics.
Hematological health outcomes in humans were previously reviewed in the 2016 PFOS HESD
(U.S. EPA, 2016d). Six occupational studies and one general population study, published prior to
2010, provided hematology data. No statistically significant associations between PFOS
exposure and hematology parameters were identified. The 2016 PFOS HESD did not specifically
discuss or draw conclusions about these parameters independent of other health outcomes.
For this updated review, eight studies examined the association between PFOS hematological
health outcomes (Appendix D). The specific hematological parameters investigated included
hematology tests (calcium, erythrocytes, ferritin, fibrinogen, hematocrit, hemoglobin, iron),
blood coagulation tests, Vitamin D levels and deficiency and anemia.
All studies assessed exposure to PFOS using biomarkers in blood. Samples were taken from
participating pregnant women, children, adolescents, or adults. All included studies were cross-
sectional designs. Four were from the United States, three from Europe, and one from Asia.
Three studies used overlapping data from a large, ongoing survey in the United States, NHANES
(Jain, 2020a, b; Etzel et al., 2019). Etzel et al. (2019) used 2003-2010 NHANES data for
adolescents and adults 12 years and older, and Jain (2020a) and Jain (2020b), used 2003-2016
NHANES data for adults 20 years and older. Also in the United States, Khalil et al.(2018)
included 48 obese children 8-12 years old from a hospital lipid clinic in Dayton, Ohio. Abraham
et al.(2020) included 101 healthy one-year-old German children in the Berlin area, including 27
children living near a former copper smelting site. Jiang et al.(2014) recruited 141 pregnant
women in Tianjin, China. Chen et al.(2019a) conducted a pilot study with 1,430 male and female
adults from the island of Hvar, off the coast of Croatia. A study conducted by van den Dungen et
al.(2017) included 80 men aged 40-70 years in the Netherlands who regularly consumed eel.
C-153
-------
APRIL 2024
C.6.1.2 Study Quality
Several considerations were specific to evaluating the quality of studies on hematological
parameters. Important considerations included the influence of diet, supplement or medication
use, adiposity (due to lipid binding), disease status, and SES on both PFOS exposure and
hematology. In particular, the duration of breastfeeding is expected to be associated with both
PFOS exposure and nutrition intake (Abraham et al., 2020). The blood matrix (whole blood
versus plasma or serum) could also affect the interpretation of results. Measuring PFOS and
serum lipids concurrently was considered adequate in terms of exposure assessment timing.
Given the long half-life of PFOS (median half-life = 3.5 years) (Li et al., 2018), current blood
concentrations are expected to correlate well with past exposures.
There are eight studies from recent systematic literature search and review efforts conducted
after publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and hematological effects. Study quality evaluations for these eight studies are
shown in Figure C-34.
On the basis of the considerations mentioned, three studies were classified as medium
confidence, two as low confidence and three as uninformative. Two low confidence studies had
deficiencies in participant selection, confounding, or sample size. Khalil et al. (2018) was
affected by a small sample size, the cross-sectional design, and potential residual confounding
attributable to differences in participants' SES. Van den Dungen et al. (2017) was affected by a
small sample size, concerns about selection bias, and a lack of information on key confounders
such as SES.
Three studies were rated as uninformative for hematological outcomes. For Jain (2020b), the use
of PFOS as the dependent variable and health outcomes as the independent (predictive) variable
rendered the study uninformative for hazard assessment (Jain, 2020b). Abraham et al. (2020) and
Jiang et al. (2014) only performed unadjusted correlation analyses and therefore did not consider
the influence of potential confounding factors.
C-154
-------
APRIL 2024
?/V5V
¦e«*
Abraham et al., 2020, 6506041 -
i
+
i
+
++
l
1
+
i
+
~
Chen et al., 2019, 5387400-
+
+
++
+
+
+
+
Etzel et al., 2019, 5043582-
++
+
+
+
++
+
+
+
Jain, 2020, 6333438-
+
+
++
+
-
+
+
+
Jain, 2020, 6833623-
+
+
+
+
+
+
+
--
Jiang et al., 2014, 2850910-
-
++
+
¦
-
+
-
--
Khalil et al., 2018, 4238547-
-
+
+
+
+
-
-
den Dungen et al., 2017, 5080340 -
-
+
+
-
+
+
-
-
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
Figure C-34. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Hematological Effects
Interactive figure and additional study details available on HAWC.
C.6.1.3 Findings
Two studies examined levels of 25-hydroxy vitamin D and vitamin D deficiency and a
significant association was observed in one study (Etzel et al., 2019). In adolescents and adults
from NHANES (2003-2010), Etzel et al.(2019) observed a statistically significant decrease in
total serum 25-hydroxy vitamin D per a 2-fold increase in PFOS and comparing the top quintile
of PFOS exposure (25.9 ng/mL-435.0 ng/mL) to the lowest quintile. Statistically significant
decrease in total serum 25-hydroxy vitamin D were also observed in participants 60 and older. A
positive non-significant association with prevalence ORs for vitamin D deficiency was also
observed. In 8-12-year-old U.S. children, Khalil et al. (2018) also observed a decrease in 25-
hydroxy vitamin D levels, but it did not reach significance.
In adults from NHANES (2003-2016), Jain (2020a) observed small statistically significant
increases in whole blood hemoglobin levels (WBHGB) with increased PFOS exposure among
C-155
-------
APRIL 2024
adult males or females > 20 years (Appendix D). This was true for subgroups with or without
anemia, although the magnitude of the effect was larger among those defined as anemic. Anemia
was defined as WBHGB concentrations < 12 g/dL for females or < 13 g/dL for males. Jain
(2020a) also evaluated the impact of deteriorating kidney function, by stratifying results by
stages of GF. For anemic males, association between WBHGB and PFOS concentrations were
uniformly positive across worsening stages of renal failure. For anemic females, association
between WBHGB and PFOS concentrations were positive except at GF-1
(eGFR > 60 mL/min/1.73 m2). Overall, the association between WBHGB and PFOS followed U-
shaped distributions. Hemoglobin levels were also examined in pregnant women (Jiang et al.,
2014). Small significant positive correlations were observed between total PFOS and
hemoglobin levels (r = 0.280, p < 0.01) as well as total PFOS and red blood cell count (RBC)
(r = 0.206, p < 0.01), although these results did not consider the influence of confounding factors
and should be interpreted with caution. In high-exposed population (van den Dungen et al.,
2017), observed non-significant decreases in hemoglobin and hematocrit levels, and non-
significant increases in retinol.
Chen et al.(2019a) found that serum calcium levels among Croatian adults were statistically
significantly decreased in association with an increase in the natural log of PFOS exposure.
C.6.2 Animal Evidence Study Quality Evaluation and Synthesis
There are three studies from the 2016 PFOS HESD (U.S. EPA, 2016c) and two studies from
recent systematic literature search and review efforts conducted after publication of the 2016
PFOS HESD that investigated the association between PFOS and hematological effects. Study
quality evaluations for these five studies are shown in Figure C-35
C-156
-------
APRIL 2024
nO<*
Curran etal., 2008, 757871 -
++
NR
++
B
B
++
++
++
B
NTP, 2019, 5400978-
++ ++
NR
++
++
++
++
Seacat et al„ 2002, 757853 -
B
NR
++
++
++
B
Seacat et al„ 2003, 1290852 -
Thomford, 2002, 5432419-
++ ++
NR
B
B
++
B
B
+
NR
++
++
++
++
++
Legend
B
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
b
Critically deficient (metric) or Uninformative (overall)
NR
Not reported
Figure C-35. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Hematological Effects
Interactive figure and additional study details available on HAWC.
Hematological measures, along with other biomarkers or histopathological findings, may be
informative for assessment of the health and function of blood-forming tissues such as the spleen
and bone marrow. The focus of this section is clinical hematological endpoints including
alterations in hemoglobin and hematocrit levels and changes in red blood cell production and
structure. Four oral studies in rodents or monkeys with short-term to chronic exposure durations
evaluated the effects of PFOS on the hematological system (see Toxicity Assessment, (U.S.
EPA, 2024)).
Significantly decreased reticulocyte counts were observed in male and female Sprague-Dawley
rats following 28-day oral gavage exposure to 2.5 mg/kg/day or 5 mg/kg/day (NTP, 2019). The
percent decrease from control was 42% and 49% in the 5 mg/kg/day dose group for males and
females, respectively, indicating potential deficiencies in red blood cell maturation. Increased
incidences of decreased splenic hematopoiesis, as well as increased bone marrow hypocellularity
characterized by minimal increases in the number of adipocytes and reductions in hematopoietic
cells, were observed in both males and females at these doses (see Toxicity Assessment, (U.S.
EPA, 2024)). NTP (2019) suggests that a combination of these findings may indicate a
suppression in erythropoiesis.
No other effects on hematocrit, hemoglobin, mean cell volume, platelet count, and red blood
cells were reported in male or female Sprague-Dawley rats in the NTP (2019) report or in male
C-157
-------
APRIL 2024
or female Sprague-Dawley rats administered up to 20 ppm PFOS (equivalent to 1.51 mg/kg/day
or 1.77 mg/kg/day in females and males, respectively) in feed for 28 days (Seacat et al., 2003). In
a third 28-day study, female Sprague-Dawley rats exposed to 100 mg/kg of PFOS in diet (highest
dose tested, equivalent to 7.58 mg/kg/day), displayed significantly reduced red blood cell
numbers, hemoglobin levels, hematocrit, and mean cell hemoglobin concentrations, though these
effects were generally within 10% of control levels (Curran et al., 2008). In male rats, there was
a trend toward reduced red blood cell distribution widths (i.e., decreased range in the volume and
size of erythrocytes) with increasing PFOS dose. Circulating blood platelet numbers were
unaffected, but mean platelet volume was significantly reduced in male rats at 6.34 mg/kg/day
(100 mg/kg of PFOS in the diet) and in female rats at 3.73 mg/kg/day (50 mg/kg of PFOS in the
diet). In both males and females exposed to 100 mg/kg PFOS in the diet, equivalent to
6.34 mg/kg/day and 7.34 mg/kg/day, respectively, the red blood cell deformability index was
significantly reduced over a range of shear stress levels. Effects on blood electrolyte levels were
also noted in these rats. Notably, the sodium/potassium ratio was increased in males and females
at 100 mg/kg PFOS in the diet (7.34 mg/kg/day) while inorganic phosphate was decreased in
females only at this same dose (Curran et al., 2008).
Other reported hematologic effects following subchronic or chronic exposure to PFOS appear to
be minimal in the low dose range. For example, male and female Sprague-Dawley rats exposed
to 0.5-20 ppm PFOS in feed (equivalent to 0.03 mg/kg/day-1.33 mg/kg/day and
0.04 mg/kg/day-1.56 mg/kg/day in males and females, respectively) for 14 weeks showed no
effects on hematocrit, hemoglobin, mean cell volume, platelet count, and red blood cells (Seacat
et al., 2003). Hemoglobin levels were decreased in male Cynomolgus monkeys following a
chronic 182-day exposure to 0.75 mg/kg/day, although no changes were observed in female
monkeys. While the hemoglobin levels in males reported by Seacat et al. (2002) are statistically
significant, they are within 10% of control and no other hematologic changes were reported in
the study.
C.6.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse hematological outcomes is discussed in
Section 3.1.1.1 of the 2016 PFOS HESD (U.S. EPA, 2016c). There are two studies from recent
systematic literature search and review efforts conducted after publication of the 2016 PFOS
HESD that investigated the mechanisms of action of PFOS that lead to hematological effects. A
summary of these studies is shown in Figure C-36. Additional mechanistic synthesis will not be
conducted since evidence is inadequate to infer that PFOS leads to hematological effects.
Mechanistic Pathway Animal Human Grand Total
Atherogenesis And Clol Formation
0
1
1
Other
0
1
Grand Total
1
1
2
Figure C-36. Summary of Mechanistic Studies of PFOS and Hematological Effects
C-158
-------
APRIL 2024
Interactive figure and additional study details available on HAWC.
C.6.4 Evidence integration
The evidence evaluating an association between PFOS exposure and hematological effects in
humans is indeterminate. The limited number of studies reporting on hematological effects of
PFOS in humans is limited and relevant outcomes were not studied in more than in one study,
hence coherence is impossible to evaluate. There is evidence for an association between
increased PFOS and slightly increased WBHGB levels (Jain, 2020a), particularly among anemic
adults in a large NHANES study. Increases in hemoglobin and RBC may also affect pregnant
women (Jiang et al., 2014). However, it is unclear whether the observed changes are clinically
adverse. The two studies that examined 25-hydroxy vitamin D levels reported mixed non-
significant effects; three studies examined hemoglobin and also reported mixed effects.
There is indeterminate animal evidence of an association between PFOS exposure and
hematological effects. Although the available 28-day studies in rats observed some
hematological effects, the alterations were generally within 10% of control, except for reduced
reticulocyte counts observed by NTP (2019). These reductions in reticulocyte counts support
histopathological changes in the spleen (splenic extramedullary hematopoiesis) that have been
identified as notable immune endpoints (see Toxicity Assessment, (U.S. EPA, 2024)).
Reticulocyte counts do not appear to be as sensitive as the corresponding histopathological
findings in the spleen; decreases in reticulocytes were observed at doses > 2.5 mg/kg/day
whereas histopathological alterations were observed at a slightly lower dose of 1.25 mg/kg/day
and higher. Further, the available subchronic and chronic studies measured hematology at
various timepoints did not observe any consistent effect of treatment on red blood cells.
C.6.4.1 Evidence Integration Judgment
Overall, there is inadequate evidence to assess whether PFOS exposure can cause hematological
effects in humans under relevant exposure circumstances (Table C-l 1).
C-159
-------
APRIL 2024
Table C-ll. Evidence Profile Table for PFOS Hematological Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.6.1)
25-hydroxy vitamin D
1 Medium confidence
study
1 Low confidence study
Two studies observed
decreases in 25-hydroxy
vitamin D. One of the
studies observed a
significant decrease
among the whole study
population. Results were
similar in all
stratifications and study
authors reported
increased vitamin D
deficiency.
• Medium confidence
study
• Consistent direction
of effects
• Low confidence study
• Limited number of
studies examining
outcome
Anemia and whole
blood hemoglobin
(WBHGB)
1 Medium confidence
study
1 Low confidence study
One study (1/2) observed
significantly increased
WBGHB, and one study
(1/2) observed non-
significant decreases in
hemoglobin.
• Medium confidence
study
• Low confidence study
• Inconsistent direction of
effects
OOO
Lndeterminate
Evidence for
hematological effects is
based on two studies
reporting decreased 25-
hydroxy vitamin D and
one study reporting
increased WBGHB.
Considerable uncertainty
due to limited number of
studies and unexplained
inconsistency across
studies and endpoints.
Serum electrolytes
1 Medium confidence
study
One study observed
significantly decreased
serum calcium among
adults.
• Medium confidence
study
• Limited number of
studies examining
outcome
Evidence From In Vivo Animal Studies (Section C.6.2)
Complete blood count
1 High confidence study
3 Medium confidence
study
One short-term study in
rats found evidence of
decreased reticulocyte
counts in male and
female following PFOS
exposure (1/1).
Hematocrit levels were
decreased in female rats
»High and medium
confidence studies
• Limited number of
studies examining
outcome
• Lnconsistent direction of
effect for reticulocyte,
hematocrit, hemoglobin,
and RBC levels
OOO
Lndeterminate
Evidence was limited,
inconsistent with
direction of effect, and
largely non-significant
for hematological
OOO
Inadequate Evidence
Primary basis:
Evidence in humans and
animals were limited and
largely non-significant.
Human relevance, cross-
stream coherence, and
other inferences'.
No specific factors are
noted.
C-160
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Evidence Integration
Studies and Summary and Key Factors That Increase Factors That Decrease Evidence Stream Summary Judgment
Interpretation Findings Certainty Certainty Judgment
at the highest dose tested endpoints in animal
following short-term models.
exposure (1/4).
Decreased hemoglobin
(2/4) was observed in
male monkeys following
chronic exposure (1/4)
and in female rats
following short-term
exposure (1/4). No
effects on hemoglobin
were found after short-
term and chronic
exposure in rats (2/4).
RBC was decreased (1/4)
in females at the highest
dose tested and only in
rats (2 additional studies
in rats and 1 study in
monkeys found no effects
on RBC). No significant
changes In mean cell
volume (2/2) and red cell
distribution width (1/1)
were observed.
An increase in the RBC
deformity index
associated with increased
PFOS dose and log shear
stress in both male and
female rats in a short-
term study (1/1).
Decreased mean platelet
volume (1/1) was
observed in male and
female rats following
C-161
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Evidence Integration
Studies and Summary and Key Factors That Increase Factors That Decrease Evidence Stream Summary Judgment
Interpretation Findings Certainty Certainty Judgment
short-term exposure to
PFOS. No significant
exposure-related changes
were observed in platelet
count (3/3).
Serum electrolytes
Inorganic phosphate • Medium confidence
• Limited number of
inorganic phosphate,
levels were decreased studies
studies examining
chloride, and Na/K
(1/2) in female rats
outcome
ratio
chronically exposed to
2 Medium confidence
the highest dose tested
studies
(1/1) but no significant
findings were observed in
a short-term study for
male or female monkeys
(1/1). In a chronic rat
study, increased Na/K
ratio (1/1) was observed
in males and females and
no exposure-related
changes were observed in
chloride levels (1/1).
Notes: WBHGB = whole blood hemoglobin; RBC = red blood count; Na/K = sodium/potassium ratio.
C-162
-------
APRIL 2024
C.7 Respiratory
EPA identified five epidemiological and five animal studies that investigated the association
between PFOS and respiratory effects. All five of the epidemiological studies were classified as
medium confidence (Section C.7.1). Of the animal studies, one was classified as high confidence,
three as medium confidence, and one was considered low confidence (Section C.7.2). Studies
may have mixed confidence ratings depending on the endpoint evaluated. Though low
confidence studies are considered qualitatively in this section, they were not considered
quantitatively for the dose-response assessment (see Toxicity Assessment, (U.S. EPA, 2024)).
C.7.1 Human Evidence Study Quality Evaluation and Synthesis
C.7.1.1 Introduction
Respiratory health can be ascertained by several measurements. The most informative are
measurements of pulmonary function (e.g., lung volume and air flow measures determined by
spirometry, as well as respiratory sounds, sputum analysis, and blood gas tension) or pulmonary
structure (e.g., lung weight, histopathology, and chest radiography), while respiratory symptoms
(shortness of breath, cough/presence of sputum, chest tightness), history of respiratory illnesses,
and respiratory mortality have low specificity and sensitivity.
The 2016 Health Assessment for PFOS (U.S. EPA, 2016c) did not examine any epidemiological
evidence of association between exposure to this chemical and respiratory health effects.
For this updated review, five epidemiological studies investigated the association between PFOS
and respiratory outcomes. All studies measured PFOS using biomarkers in blood. Three studies
were mother-child cohort studies conducted in Europe (Agier et al., 2019; Manzano-Salgado et
al., 2019; Impinen et al., 2018), one was a cross-sectional case-control study (cross-sectional
analyses were performed in asthmatic cases and non-asthmatic controls) conducted in Taiwan
(Qin et al., 2017); and one was a cross-sectional study of adolescents and young adults residing
near the WTC (Gaylord et al., 2019). The five available studies examined lung function
measures in children and young adults, including forced expiratory volume in one second
(FEVi), forced vital capacity (FVC), FEVi/FVC ratio, forced expiratory flow at 25%-75%
(FEF25%-75%), peak expiratory flow rate (PEF), lung volume, resistance at oscillation frequencies
of 5 Hz or 20 Hz, lung function at birth, and severity of obstructive airways disease (Appendix
D).
Studies that examined respiratory illnesses or symptoms reflecting immune system responses
(e.g., asthma and allergies) and respiratory tract infections (e.g., cough) are analyzed in the
immune system section.
C.7.1.2 Study Quality
There are five studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and respiratory effects. Study quality evaluations for these five studies are shown
in Figure C-37. The five general population studies identified since the last assessment were all
classified medium confidence. These studies had minor deficiencies, including concerns that co-
exposures in the WTC disaster could confound the results (Gaylord et al., 2019), reduced
C-163
-------
APRIL 2024
sensitivity because of low exposure levels and narrow ranges (Impinen et al., 2018), or concerns
with potential bias in selection of non-asthmatic controls (Qin et al., 2017).
Figure C-37. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Respiratory Effects
Interactive figure and additional study details available on HAWC.
C.7.1.3 Findings From Children and Adolescents
Four studies examined respiratory health effects in children up to 15 years old (Agier et al.,
2019; Manzano-Salgado et al., 2019; Impinen et al., 2018; Qin et al., 2017) and one examined
adolescents and young adults ages 13-22 years (Gaylord et al., 2019) (Appendix D).
Of the four studies examining FEV1, three reported negative associations (i.e., decrease in FEV1
with higher PFOS levels), while one reported a positive association. Qin et al. (2017) observed
significant inverse associations for children ages 10-15 years old with asthma (beta = -0.061,
95% CI: -0.101, -0.021), and in boys with asthma, but not in girls with asthma. There was also a
significantly decreasing trend by quartiles of PFOS in children with asthma (p-trend = 0.003).
No effects were observed in children without asthma. Results from other studies examining
FEV1 were inconsistent and non-significant, with two studies (Gaylord et al., 2019; Manzano-
Salgado et al., 2019) observing inverse associations and one study (Agier et al., 2019) reporting a
positive association.
For other lung function measures examined there was also limited evidence of associations. Qin
et al. (2017) reported a statistically significant association with FVC (beta = -0.055, 95%
C-164
-------
APRIL 2024
CI: -0.1, -0.01) but a non-significant decreasing trend by quartiles of PFOS (p-trend = 0.186).
Non-significant associations were observed for FEF25%-75% or PEF or for any lung function
measures in children without asthma. Impinen et al. (2018) reported a statistically significant
association with severe obstructive airways disease at age 2 measured by the Oslo Severity Score
(OSS), but only for the lowest severity category (OSS 1-5) (OR per log2 increase PFOS = 1.71,
95% CI: 1.16, 2.53). The study also reported a non-significant decrease in odds of reduced lung
function at birth, as measured by tidal flow volume. Clear patterns were not observed for other
lung function measures (i.e., FVC, FVC/FEVi, lung resistance, total lung capacity, functional
residual capacity, and residual volume) in the remaining studies (Gaylord et al., 2019; Manzano-
Salgado et al., 2019).
C. 7.2 Animal Evidence Study Quality Evaluation and Synthesis
There are five studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and respiratory effects. Study quality evaluations for these five studies are shown
in Figure C-38.
Argus, 2000, 5080012-
++
++
++ ++
++
++
B
Li et al., 2021, 7643501 -
B
B
B
++
B
B
NTP, 2019, 5400978-
++
++
++ ++
++ ++
++
++
Thomford, 2002, 5432419-
B
B
++ ++
B
++ ++
++
B
Yang et al„ 2021, 7643494-
++
B
++ ++ ++
B
Legend
| Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
B Critically deficient (metric) or Uninformative (overall)
Not reported
* Multiple judgments exist
Figure C-38. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Respiratory Effects
Interactive figure and additional study details available on HAWC.
Several studies have reported adverse pulmonary effects resulting from oral PFOS exposure. The
available literature primarily focuses on fetal and neonatal outcomes as several groups
C-165
-------
APRIL 2024
hypothesized that the interactions of PFOS with pulmonary surfactants and subsequent
reductions in lung function or maturity may play a role in the increased perinatal mortality
resulting from gestational PFOS exposure (U.S. EPA, 2016c; Chen et al., 2012b; Ye et al., 2012;
Yahia et al., 2008; Grasty et al., 2005; Grasty et al., 2003; Argus Research Laboratories, 2000).
There are also several available studies that reported pulmonary effects in adult mammalian
models (Li et al., 2021b; Yang et al., 2021; NTP, 2019; Cui et al., 2009; Goldenthal et al., 1979).
Yahia et al. (2008) exposed mouse dams to 0, 1, 10, or 20 mg/kg/day PFOS from GD 0-GD 17
and assessed neonatal and maternal lung histopathology. Initially, a single surviving pup from
each dam (n = 5/treatment group) was analyzed at PND 0; all five pups in the 20 mg/kg/day
group showed lung atelectasis (i.e., complete or partial lung collapse) which was characterized
by alterations in the alveolar epithelium, congestion of alveolar capillary vessels, and reduced
alveolar space. Focal or severe atelectasis was also present in some of the pups from the
10 mg/kg/day group (incidence not provided) but not in pups from the control or 1 mg/kg/day
groups. No observed histological effects of PFOS exposure were observed on the maternal lung.
Yahia et al. (2008) dosed additional dams with 20 mg/kg/day PFOS from GD 0-GD 17 or
10 mg/kg/day PFOS from GD 0-GD 18 to further examine pulmonary effects in fetuses and
pups, respectively. Immediately at birth, 27% (4/15) of pups (n = 3 pups/dam) from 3/5 dams
dosed with 10 mg/kg/day PFOS showed at least mild lung atelectasis. In contrast, all fetuses in
the 20 mg/kg/day group showed normal lung histopathology at GD 18. The authors suggested an
increase in the incidence of moderate to severe intracranial blood vessel dilation in fetuses at GD
18 as a cause of the pulmonary effects that were not seen until birth (Yahia et al., 2008).
Chen et al. (2012b) similarly assessed rat pup lung histopathology at PND 0 and PND 21 after
gestational exposure to 0 mg/kg/day, 0.1 mg/kg/day, or 2 mg/kg/day PFOS from GD 1-GD 21.
With PFOS exposure of 2 mg/kg/day, pups showed marked alveolar hemorrhaging, thickened
interalveolar septum, and focal lung consolidation at PND 0 (incidence data not provided). These
effects lasted through PND 21, when pups from the 2 mg/kg/day treatment group also showed
alveolar hemorrhaging, thickened interalveolar septum, and inflammatory cell infiltration. The
2 mg/kg/day group PND 0 and PND 21 pups also had higher percentages of pulmonary apoptotic
cells. There were no pulmonary abnormalities observed in pups from the control or
0.1 mg/kg/day groups.
Zhang et al. (2021) reported that Sprague-Dawley rat pups exposed to 1 or 5 mg/kg/day from GD
12 to GD 18 had higher lung injury scores and that pups in the 5 mg/kg/day group had lower
radial alveolar counts on PND 1, 3, 7, and 14 compared with controls.
In an attempt to identify the prenatal window of susceptibility to PFOS in neonatal rats, Grasty et
al. (2003) dosed dams with 0 mg/kg/day, 25 mg/kg/day, or 50 mg/kg/day PFOS during several 4-
day gestational timepoints, including GD 17-GD 20, a period of development they identified in
this study as a particularly sensitive window for neonatal mortality. As the last few days of fetal
development involve central nervous system and pulmonary maturation, the authors conducted a
second exposure of 0 mg/kg/day, 25 mg/kg/day, or 50 mg/kg/day PFOS from GD 19-GD 21 and
sacrificed fetuses at GD 21 or pups at PND 0 to examine lung histology (Grasty et al., 2003). No
histological differences between lung samples of control and treated fetuses sacrificed at GD 21
were observed, though it appeared that PFOS reduced lung expansion and slowed or
compromised lung maturation of pups by PND 0; epithelial thickness of lungs of PFOS-treated
pups at PND 0 was similar to that of lungs from fetal control animals at GD 21 (incidence data
C-166
-------
APRIL 2024
not provided). Grasty et al. (2005) conducted a follow-up study with the same GD 19-GD 21
exposure paradigm to further explore mechanisms of developmental pulmonary dysfunction and
potential methods of therapeutic rescue of delayed lung maturation and effects on pulmonary
surfactants seen after gestational PFOS exposure. Grasty et al. (2005) found several
morphometric changes in pup lung tissue after 25 mg/kg/day or 50 mg/kg/day PFOS exposure,
including increases in the proportion of lung occupied by solid tissue, decreases in the proportion
of lung occupied by small airways, and increases in the ratio of solid tissue to small airway
space. The authors also note that some lung samples from the 50 mg/kg/day group did not appear
to fill fully upon perfusion, potentially indicating a failure of inflation upon birth or atelectasis.
Similar to the results of Grasty et al. (2003), the lungs of some PFOS-exposed pups at PND 0
resembled the lungs of control fetuses at GD 21 (incidence of 17% and 50% of pups from the
25 mg/kg/day and 50 mg/kg/day groups, respectively). Co-treatment with the therapeutic agents
dexamethasone or retinyl palmitate did not increase neonatal survival, indicating the pulmonary
effects of PFOS do not drive neonatal mortality, though the authors did not report histological
analyses showing improved pulmonary outcomes in co-treated animals. Ye et al. (2012) did not
observe effects on rat fetal lung histopathology following gestational exposure to 5 or
20 mg/kg/day, though the exposure period lasted from GD 12-GD 18 and may have missed the
sensitive period of lung development in rats (Grasty et al., 2005; Grasty et al., 2003).
In a rabbit teratology study, Argus (2000) reported a significant increase in the number of fetuses
with absent intermediate lung lobes after exposure to 0.1 mg/kg/day PFOS from GD 7-GD 20
(7/172 fetuses compared with 2/175 in controls). However, this increase was not statistically
significant when analyzed by litter (4/19 litters compared with 2/20 in controls) and no increase
was observed in the higher dose groups of 1 mg/kg/day, 2.5 mg/kg/day, or 3.75 mg/kg/day.
Argus (2000) noted that this fetal malformation was likely not related to the test article as varied
lung development is frequently observed in New Zealand white rabbits.
Pulmonary effects were observed in adult animals after short-term and subchronic exposures to
PFOS. Cui et al. (2009) reported dose-related increases in pulmonary congestion and focal or
diffuse thickening of epithelial walls in the lungs of male rats gavaged with 5 mg/kg/day or
20 mg/kg/day PFOS for 28 days (incidence data not provided). Focal or diffuse neutrophil,
acidophilia, and lymphocyte cellular infiltration and vasodilatation due to leakage of erythrocytes
was also especially apparent in the 20 mg/kg/day dose group (incidence data not provided). In a
study with limited sample size (n = 2/sex/treatment), Goldenthal et al. (1979) reported increased
moderate diffuse atrophy of the serous alveolar cells in 3/4 rhesus monkeys from the highest
dose group (4.5 mg/kg/day) treated with PFOS for 90 days. NTP (2019) did not report nasal,
olfactory, or pulmonary histopathological effects in adult male or female rats dosed with up to
5 mg/kg/day PFOS for 28 days. However, female rats dosed with 1.25 mg/kg/day,
2.5 mg/kg/day, or 5 mg/kg/day had significantly increased relative lung weight. The biological
significance of this increase is unclear as absolute lung weight was only significantly increased
in the 1.25 mg/kg/day group and there were no accompanying histopathological alterations in the
lung. Yang et al. (2021) examined the impacts of PFOS exposure on male C57BL/6 mice
pulmonary system in a 28-day oral gavage study. Relative lung weights displayed an 1% and 6%
increase in 0.25 mg/kg/day and 2.5 mg/kg/day groups, respectively, compared with the control
group. The toxicological significance of the increase is unclear due to both low sample size with
six animals per group and lack of report on body weight or absolute lung weight. Li et al.
(2021b) examined the histopathological effects of PFOS exposure on female BALC/c mice
C-167
-------
APRIL 2024
pulmonary system in a 60-day oral gavage study. Authors reported zero incidence of lesions
following respiratory histopathological examination among the female mice gavaged with
0.1 mg/kg/day and 1 mg/kg/day.
Immunological responses in lungs were investigated in Yang et al. (2021). No significant
differences in bronchoalveolar lavage fluid (BALF) macrophages, eosinophils, neutrophils, and
total cell counts were observed among control or dosed groups. Cytokine IL-4 in BALF
displayed significant increases in both the 0.25 mg/kg/day and 2.5 mg/kg/day dose groups. IL-13
in BALF showed a significant increase in the 2.5 mg/kg/day dose group whereas IFN-y in BALF
did not display a significant difference. In the same study, PFOS was found to likely exacerbate
asthmatic responses. In the BALF, total cell count and eosinophil numbers were higher in
ovalbumin (OVA)-induced mice exposed to 0.25 mg/kg/day or 2.5 mg/kg/day PFOS than to
OVA-induced alone. 2.5 mg/kg/day PFOS-treated OVA-induced mice showed a 33% increase in
the eosinophil infiltration and 67% increase in mucus production compared with OVA-induced
alone mice.
C.7.3 Mechanistic Evidence
Mechanistic evidence linking PFOS exposure to adverse respiratory outcomes is discussed in
Sections 3.2.5 and 3.4.1.2 of the 2016 PFOS HESD (U.S. EPA, 2016c). There are three studies
from recent systematic literature search and review efforts conducted after publication of the
2016 PFOS HESD that investigated the mechanisms of action of PFOS that lead to respiratory
effects. A summary of these studies is shown in Figure C-39. Additional mechanistic synthesis
will not be conducted since evidence suggests but is not sufficient to infer that PFOS leads to
respiratory effects.
Mechanistic Pathway In Vitro Grand Total
Cell Growth, Differentiation, Proliferation, Or Viability
3
3
Inflammation And Immune Response
1
1
Oxidative Stress
1
1
Grand Total
3
3
Figure C-39. Summary of Mechanistic Studies of PFOS and Respiratory Effects
Interactive figure and additional study details available on HAWC.
C.7.4 Evidence Integration
The evidence evaluating associations between PFOS exposure and respiratory effects in humans
is slight, with an indication of decreased lung function in infants, children, and adolescents.
However, the results across studies are inconsistent, and there are a lack of studies examining
respiratory effects in both children and adults. Specifically, no studies were available that
assessed respiratory health effects in older adults. While there is some evidence of detrimental
C-168
-------
APRIL 2024
respiratory health effects, particularly in children with asthma, the available epidemiological
evidence examining PFOS exposure and respiratory health is limited.
The animal evidence for an association between PFOS exposure and respiratory effects is slight,
with an indication that the developing lung may be affected in animal models, but at high doses.
Evidence in adults is less consistent with no lesions observed in medium or high confidence
studies (Li et al., 2021b; NTP, 2019), but an exacerbated immune response appears to occur in
the lung based on a medium confidence study (Yang et al., 2021). Several studies in animal
models indicate that PFOS may influence fetal and neonatal lung development which may be
consistent with epidemiological assessments of reduced lung function in children, though none
of the animal studies provide quantifiable incidence data. Additionally, effects on the pulmonary
systems of fetuses and neonates generally occurred at doses above those that result in other
adverse developmental effects (see Toxicity Assessment, (U.S. EPA, 2024)), indicating that
respiratory toxicity is not likely a highly sensitive health outcome for PFOS exposure.
C.7.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause respiratory effects in
humans under relevant exposure circumstances (Table C-12). The conclusion is based on limited
evidence of an association between PFOS and detrimental respiratory health effects, particularly
in children with asthma, in a small number of epidemiologic studies with median exposure levels
from 5.2 ng/mL to 31.5 ng/mL, and on evidence from animal models showing changes in pup
lung tissue following exposure to doses as low as 2 mg/kg/day PFOS. However, the limited
number of studies and issues with inconsistency across studies raise considerable uncertainty.
C-169
-------
APRIL 2024
Table C-12. Evidence Profile Table for PFOS Respiratory Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.7.1)
Lung function
measures
4 Medium confidence
studies
Two studies (2/4)
observed decreases in
forced expiratory volume
in 1 s (FEV1) and forced
vital capacity (FVC) in
children, with one study
reporting significant
decreases among
asthmatic children. Other
studies observed small
increases in FEV1/FVC
and FEF25%-75% at age 4,
but the associations were
imprecise at age 7.
• Medium confidence
studies
»Imprecision of study findings ©OO
in children Slight
©OO
Evidence Suggests
Obstructive disease
1 Medium confidence
study
One study in infants
under 2 years old
observed significantly
increased odds of low
severity obstructive
airway disease.
• Medium confidence
study
• Limited number of studies
examining outcome
Primary basis:
Several studies of Human evidence indicted
medium confidence founddetrimental ^spiratoty
evidence for decreases in health effects' Particularly
lung function measures in children with asthma
among infants, children, wlule ammal evldence
and adolescents, though indlcalcd changes in pup
other medium confidence 'un^ tissue following
studies did not observe exposure. However, the
significant effects. Few llimted number of studies
and issues with
imprecision across studies
raise considerable
studies examined
_obstructive disease
effects. Uncertainty
remains about respiratory uncertainty
outcomes among adults
in occupational settings
and in the general
population.
Evidence From In Vivo Animal Studies (Section C.7.2)
Human relevance, cross-
stream coherence, and
other inferences:
¦No specific factors are
noted.
Histopathology
1 High confidence study
3 Medium confidence
studies
One teratology study in <
rabbits (1/1) reported a
significant increase in the
number of fetuses with
absent intermediate lung
lobes after gestational
exposure to the lowest
dose of PFOS. This
increase was not
significant when analyzed
by litter and no increase
High and medium
confidence studies
• Inconsistency of findings
across species and life stage
©OO
Slight
Evidence indicates that
the developing lung may
be affected. Evidence in
adults is less convincing
as limited findings were
observed in adult mice
and rats.
C-170
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
was observed following
exposure to higher doses.
Three short-term and
subchronic studies in
adult male and female
mice and rats reported no
histopathological effects
in the respiratory system
after exposure (3/3).
Organ weight One short-term study
1 High confidence study reported female rats had
significantly increased
relative lung weight while
the absolute weight only
increased in one dose
group. No change in lung
weight was reported in
male rats.
> High confidence study
• Limited number of studies
examining outcome
Notes: FEF25%-75% = forced expiratory flow at 25%-75%; FEV1 = forced expiratory volume; FVC = forced vital capacity.
C-171
-------
APRIL 2024
C.8 Musculoskeletal
EPA identified six epidemiological and one animal studies that investigated the association
between PFOS and musculoskeletal effects. Of the epidemiological studies, six were classified as
medium confidence and two as low confidence (Section C.8.1). The animal study was classified
as low confidence (Section C.8.2). Studies may have mixed confidence ratings depending on the
endpoint evaluated. Though low confidence studies are considered qualitatively in this section,
they were not considered quantitatively for the dose-response assessment (see Toxicity
Assessment, (U.S. EPA, 2024)).
C.8.1 Human Evidence Study Quality Evaluation and Synthesis
C.8.1.1 Introduction
Musculoskeletal health outcomes include bone mineral density, risk of bone fractures, and risk of
osteoarthritis. Osteoporosis (characterized by weak, brittle bone) and osteoarthritis
disproportionately affect women, older individuals, and certain racial/ethnic groups (Khalil et al.,
2016; Uhl et al., 2013).
The 2016 PFOS HESD (U.S. EPA, 2016c) did not previously evaluate musculoskeletal health
outcomes in humans.
For this updated review, eight studies (eight publications) examined the association between
PFOS exposure and musculoskeletal health outcomes. All studies were in the general population.
Different study designs were used, including cross-sectional, prospective cohort, and one clinical
trial (Hu et al., 2019). All studies measured PFOS in blood components (i.e., blood, plasma, or
serum), and one study (Di Nisio et al., 2019) measured PFOS in semen. Three studies (Khalil et
al., 2016; Lin et al., 2014; Uhl et al., 2013) used data from participants in the NHANES, but the
study years and outcomes examined in these studies did not overlap. Other studies used data
from various cohorts for cross-sectional analyses, including Project Viva (Cluett et al., 2019), the
POUNDS Lost clinical trial (Hu et al., 2019), and the ALSPAC (Jeddy et al., 2018). The studies
were conducted in different populations, including participants from England, Italy, and the
United States. The specific outcomes investigated were osteoporosis; osteoarthritis; bone area,
mineral content, mineral density, thickness (e.g., endosteal and periosteal thickness), or
circumference; bone stiffness; ultrasound attenuation and speed of sound; lean body mass;
height; arm span; bone fracture; and plasma concentrations of P-C-telopeptides of type I collagen
(CTX), a marker for bone turnover.
C.8.1.2 Study Quality
Considerations specific to evaluating the quality of studies on the musculoskeletal system relate
to the causal pathways for PFOS to alter musculoskeletal development. Expectations for
musculoskeletal condition should be interpreted relative to participants' age, pubertal and/or
menopause status, thyroid hormone levels, and adiposity (BMI), which could likewise be
influenced by PFOS exposure (Cluett et al., 2019; Jeddy et al., 2018; Khalil et al., 2018; Khalil et
al., 2016). Ideally, studies would characterize these factors, adjust models for confounding where
appropriate, and capture a range of human life stages with prospective measurement of PFOS
exposure relative to health outcomes. The outcomes should be well-defined and validated by
C-172
-------
APRIL 2024
biometric testing, a physician diagnosis, or medical records where possible. An exception may be
acute traumatic injuries such as fractures, which are less likely to be subject to recall bias.
There are eight studies from recent systematic literature search and review efforts conducted
after publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and musculoskeletal effects. Study quality evaluations for these eight studies are
shown in Figure C-40.
On the basis of the considerations mentioned, six studies were classified as medium confidence
and two as low confidence. The two cross-sectional studies (Di Nisio et al., 2019; Khalil et al.,
2018) classified as low confidence had deficiencies in participant selection, confounding, and
study sensitivity. Participant selection was considered a deficiency mainly due to underreporting
about participation rates and participant characteristics. Other deficiencies included potential for
residual confounding by SES, small sample sizes and limited ranges of participant exposure to
PFOS.
C-173
-------
APRIL 2024
1 1 1
Cluett etal., 2019, 5412438-
++
+
++
+
++
+
+
+
Di Nisio et al., 2019, 5080655 -
-
+
+
-
+
+
-
-
Hu et al., 2019, 6315798-
+
++
++
-
++
+
+
+
Jeddy et al., 2018, 5079850 -
+
B
++
-
+
+
+
+
Khalil etal., 2016, 3229485-
++
+
+
+
+
+
Khalil et al., 2018, 4238547-
-
++
-
+
+
-
-
Lin et al., 2014, 5079772-
+
+
+
+
+
+
Uhl etal., 2013, 1937226-
+
+
+
+
++
+
+
+
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
* Multiple judgments exist
Figure C-40. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Musculoskeletal Effects
Interactive figure and additional study details available on HAWC.
C.8.1.3 Findings From Children and Adolescents
Three studies (Cluett et al., 2019; Jeddy et al., 2018; Khalil et al., 2018) examined
musculoskeletal outcomes in children and adolescents, and two observed effects. While the
medium confidence studies observed few statistically significant associations between PFOS and
musculoskeletal health outcomes, the associations consistently supported a harmful, rather than
beneficial, direction of effect (Appendix D). Cluett et al. (2019) observed a statistically
significant inverse association with areal bone mineral density (aBMD) z-score (a standardized
measure of bone mineral amount relative to bone area) in children aged 6-10 years. The sex-
stratified results were not statistically significant. Inverse non-significant associations were also
observed with a bone mineral density (BMD) in boys and in girls with bone mineral content
C-174
-------
APRIL 2024
(BMC) z-score. Jeddy et al. (2018) identified a statistically significant inverse association
between prenatal PFOS exposure and total lean body mass and height in 17-year-old girls. The
same study initially showed inverse associations between PFOS exposure and BMC or bone
area, but these were not statistically significant after adjusting for participant height.
A low confidence study in 8-12-year-old children from a hospital lipids clinic in Dayton, Ohio,
(Khalil et al., 2018) observed non-significant inverse associations with bone stiffness index,
broadband ultrasound attenuation, or speed of sound.
None of the studies identified in this updated review examined musculoskeletal outcomes in
pregnant women and infants.
C.8.1.4 Findings From the General Adult Population
Five studies (Di Nisio et al., 2019; Hu et al., 2019; Khalil et al., 2016; Lin et al., 2014; Uhl et al.,
2013) examined musculoskeletal outcomes in adults in the general population and three observed
effects (Appendix D).
The four medium confidence studies observed a small number of statistically significant
associations but a consistently harmful direction of effect. The same outcomes were not
examined by multiple studies. Uhl et al. (2013) observed higher odds of osteoarthritis with
increased PFOS exposure only in women aged 20-84 from NHANES (2003-2008), who may
have differing susceptibility to endocrine disruption. Significant associations were observed only
by younger women aged 20-49. In an overlapping NHANES study (Lin et al., 2014), observed
decreased total lumbar spine BMD only among younger women not in menopause; no
statistically significant association with a history of bone fractures were observed in women aged
20 or older. Khalil et al.(2016) observed a statistically significant inverse association with BMD
of the total femur or femoral neck in women aged 12-80 years from NHANES (2009-2010). The
same was true for the femoral neck only in males aged 12-80 years. In adults aged 30-70 years
from the POUNDS Lost study, Hu et al.(2019) observed small but statistically significant inverse
associations with BMD (or two-year change in BMD) in three of the six sites examined: the
spine, total hip, and hip intertrochanteric area.
A low confidence study in young men (18-24 years) from the Padova area of northeastern Italy
(Di Nisio et al., 2019) did not find evidence of associations between PFOS exposure and arm
span.
C.8.2 Animal Evidence Study Quality Evaluation and Synthesis
There is one study from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and musculoskeletal effects. Study quality evaluation for this one study is shown
in Figure C-41.
C-175
-------
APRIL 2024
Thomford, 2002, 5432419-
s
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
NR Not reported
Figure C-41. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Musculoskeletal Effects
Interactive figure and additional study details available on HAWC.
Limited data are available on the effect of PFOS on the musculoskeletal system other than
developmental skeletal defects resulting from gestational exposure (see Toxicity Assessment,
(U.S. EPA, 2024)). EPA did not identify any publications that reported musculoskeletal effects
outside of those associated with developmental toxicity from the 2016 PFOS HESD (U.S. EPA,
2016c) or the recent literature searches that were PECO relevant and determined to be medium or
high confidence rating during study quality evaluation.
C.8.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse musculoskeletal outcomes
in the 2016 PFOS HESD (U.S. EPA, 2016c). There are six studies from recent systematic
literature search and review efforts conducted after publication of the 2016 PFOS HESD that
investigated the mechanisms of action of PFOS that lead to musculoskeletal effects. A summary
of these studies is shown in Figure C-42. Additional mechanistic synthesis will not be conducted
since evidence suggests but is not sufficient to infer that PFOS leads to musculoskeletal effects.
C-176
-------
APRIL 2024
Mechanistic Pathway
Animal
In Vitro
Grand Total
Big Data, Non-Targeted Analysis
0
2
2
Cell Growlh, Differentiation, Proliferation, Or Viability
0
4
Cell Signaling Or Signal Transduction
1
4
Extracellular Matrix Or Molecules
0
1
1
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
0
1
1
Hormone Function
1
0
1
Other
1
0
1
Grand Total
2
4
e
Figure C-42. Summary of Mechanistic Studies of PFOS and Musculoskeletal Effects
Interactive figure and additional study details available on HAWC.
C.8.4 Evidence Integration
There is slight evidence of an association between PFOS exposure and musculoskeletal effects in
humans based on observed effects on BMD and bone health in a limited number of medium
confidence studies. Limited evidence from individual studies supported possible negative effects
of PFOS on skeletal size (height), lean body mass, and connective tissue disorders
(osteoarthritis). No musculoskeletal health outcome epidemiologic studies were previously
reviewed in the 2016 PFOS HESD (U.S. EPA, 2016c).
Although relatively few studies have investigated musculoskeletal health outcomes related to
PFOS exposure, some shared conclusions can be drawn. This review observed evidence of
statistically significant associations in about 13% of all tests conducted. The observed
associations were primarily between increased PFOS exposure and decreased BMD
(inconsistently among various skeletal sites), height and lean body mass in adolescence, and
osteoarthritis. These issues with bone density may correspond with the reports of reduced
ossification and skeletal deformities in developmental animal models with gestational PFOS
exposure (see Toxicity Assessment, (U.S. EPA, 2024)). Arm span measures in adolescents were
not associated with PFOS exposure. More severe clinical outcomes, such as fracture, were not
observed to be associated with PFOS exposure. No evidence supported beneficial
musculoskeletal effects of PFOS exposure. In general, links to musculoskeletal disease were
more commonly observed among older women. Some outcomes, such as osteoporosis and
osteoarthritis, may be more relevant to examine in females, due to greater prevalence and
potentially greater susceptibility to endocrine-disrupting chemicals. Study limitations have
somewhat reduced the confidence of most studies; common issues included cross-sectional
design or potential for residual confounding.
C-177
-------
APRIL 2024
The animal evidence for an association between PFOS and effects in the musculoskeletal system
is considered indeterminate based on lack of information in animal models.
C.8.4.1 Evidence Integration Judgment
Overall, evidence suggests that PFOS exposure has the potential to cause musculoskeletal effects
in humans under relevant exposure circumstances (Table C-13). This conclusion is based
primarily on effects on BMD and bone health observed in studies in humans exposed to median
PFOS ranging from 6.4 ng/mL to 32.2 ng/mL. Although there is some evidence of negative
effects of PFOS exposure on skeletal size (height and arm span) and connective tissue disorders
(osteoarthritis, especially in older women), there is considerable uncertainty in the results due to
inconsistency across studies and limited number of studies.
C-178
-------
APRIL 2024
Table C-13. Evidence Profile Table for PFOS Musculoskeletal Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Bone parameters
5 Medium confidence
studies
1 Low confidence study
Fractures
1 Medium confidence
study
Size measures
1 Medium confidence
study
1 Low confidence study
Summary and Key
Findings
Decreases in BMC were
observed in two studies
(2/6), with significant
decreases observed
among female children.
Reductions in BMD were
also observed in children
and adults (4/6),
including site specific
BMD measures.
Significant decreases in
BMD were also observed
in analyses stratified by
sex. Decreases in other
measures of bone health,
such as the stiffness
index, bone area, and
broadband ultrasound
attenuation, were
observed in children.
Findings regarding
incidence of fractures in
adults ages 20 yr or older
were largely imprecise.
One study reported
significantly decreased
height in girls at age 17
(1/2). Findings for arm
Factors That Increase
Certainty
• Medium confidence
studies
• Consistency of BMD
reduction findings
across three medium
studies
• Medium confidence
study
• Medium confidence
study
Factors That Decrease
Certainty
• Lmprecision of findings
across exposure groups
and studies
• Low confidence study
• Lmprecision of findings
• Limited number of
studies examining
outcome
• Lmprecision of findings
• Limited number of
studies examining
outcome
Evidence Stream
Judgment
©oo
Slight
Evidence for
musculoskeletal effects is
based on studies
reporting reductions in
bone health, bone
density, lean body mass,
and increased odds of
osteoarthritis.
Uncertainties remain due
to inconsistent or
imprecise results, and
limited evidence for
fractures, size measures,
and odds of osteoarthritis
or osteoporosis.
Evidence Integration
Summary Judgment
©OO
Evidence Suggests
Primary basis:
No animal evidence and
human evidence indicated
effects on BMD and bone
health. Although there is
some evidence of negative
effects of PFOS exposure
on skeletal size (height
and arm span) and
connective tissue disorders
(osteoarthritis, especially
in older women), there is
considerable uncertainty in
the results due to
inconsistency across
studies and limited number
of studies.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
Evidence From Studies of Exposed Humans (Section C.8.1)
C-179
-------
APRIL 2024
Evidence Stream Summary and Interpretation
Evidence Integration
Studies and Summary and Key Factors That Increase Factors That Decrease Evidence Stream Summary Judgment
Interpretation Findings Certainty Certainty Judgment
span were largely • Low confidence study
imprecise in a study on
male high school
students.
Lean body mass
1 Medium confidence
study
Osteoarthritis
1 Medium confidence
study
Osteoporosis
1 Medium confidence
study
One study found a
significant reduction of
total lean body mass in
girls at age 17.
Odds of osteoarthritis
among adults aged 20-84
and among females aged
20-49 were significantly
increased.
Findings for osteoporosis
in women aged 12-80
were largely imprecise.
• Medium confidence
study
• Medium confidence
study
• Medium confidence
study
• Limited number of
studies examining
outcome
• Limited number of
studies examining
outcome
• Lmprecision of findings
• Limited number of
studies examining
outcome
Notes: BMC = bone mineral content; BMD = bone mineral density; yr = years.
C-180
-------
APRIL 2024
C.9 Gastrointestinal
EPA identified four epidemiological and two animal studies that investigated the association
between PFOS and gastrointestinal effects. Of the epidemiological studies, three were classified
as medium confidence and one as low confidence (Section C.9.1). Of the animal studies, one was
classified as high confidence, and one was considered low confidence (Section C.9.2). Studies
may have mixed confidence ratings depending on the endpoint evaluated. Though low
confidence studies are considered qualitatively in this section, they were not considered
quantitatively for the dose-response assessment (see Toxicity Assessment, (U.S. EPA, 2024)).
C.9.1 Human Evidence Study Quality Evaluation and Synthesis
C.9.1.1 Introduction
GI health outcomes were not previously evaluated in the 2016 PFOS HESD, although
gastroenteritis frequency was considered as a marker of immune system function. Causation of
gastroenteritis cases may be difficult to disentangle, as underlying susceptibility varies, and the
infectious agent or irritant is rarely confirmed. Granum et al. (2013) did not observe a
statistically significant association between prenatal PFOS exposure and the frequency of
gastroenteritis episodes in a child's first three years of life, as they did for PFOA (Granum et al.,
2013).
PFOS exposure may affect GI health by altering molecular processes (such as those involved in
inflammation), gut mucosa integrity (by acting as surfactants) and intestinal permeability, gut
microbiota, and/or systemic susceptibility to infection (Xu et al., 2020d; Steenland et al., 2018).
GI outcomes only assessed in the context of immune system health, including ulcerative colitis
and Crohn's disease, are discussed (see Toxicity Assessment, (U.S. EPA, 2024)). However, some
research suggests an overall immunosuppressive effect of PFOS could reduce the efficiency of
routine childhood immunizations (Dalsager et al., 2016) which might include that for rotavirus, a
common childhood cause of diarrhea and vomiting. In addition, inflammatory bowel disease
(IBD), or the chronic inflammation of the GI tract in response to environmental triggers, can be
considered an immune dysregulation response occurring in genetically susceptible individuals
(Hammer et al., 2019).
For this updated review, four studies examined the association between PFOS and GI health
outcomes. The specific outcomes investigated were diarrhea, vomiting, IBD, and IBD
biomarkers (zonulin and calprotectin). PFOS was measured in serum or blood
Dalsager et al. (2016) used data from the ongoing, prospective OCC, a group of pregnant women
recruited 2010-2012 and their children living in the Odense area of Denmark. Hammer et al.
(2019) examined participants in the Children's Health and the Environment in the Faroes (CHEF)
cohort, which enrolled mother-child pairs, the children's fathers and grandparents, and young
men from the Faroe Islands hospital system between 1986 and 2009. Xu et al. (2020d) examined
child and adult participants from the Ronneby, Sweden exposed to PFAS in drinking water), and
unexposed individuals from a nearby town. Timmermann et al. (2020) examined a subset of 4-
18-month-old children from a randomized controlled trial of early measles vaccination,
conducted in Guinea-Bissau in West Africa from 2012 to 2015.
C-181
-------
APRIL 2024
C.9.1.2 Study Quality
Several considerations were specific to evaluating the quality of the studies of GI symptoms. For
example, fever or a stool test might help to confirm that diarrhea and vomiting are attributable to
infection, as opposed to a chronic underlying condition or other chemical or dietary irritant.
Medical diagnoses are preferred to self-reported symptoms, although knowledge of GI disorders
has developed substantially over recent decades and diagnostic indicators continue to rapidly
evolve. Causal factors in developing GI conditions have likewise shifted over time, such as
changes in emerging contaminants, hygiene, the gut microbiome, activity and stress levels, and
dietary trends. These underlying trends may affect cohort studies with extended recruitment or
follow-up periods. Reverse causation is possible if GI conditions lead to increased intake of
PFOS from food packaging or preparation methods, increased PFOS absorption through the GI
tract, or reduced fecal excretion (Xu et al., 2020d). Measuring PFOS and GI outcomes
concurrently was considered adequate in terms of exposure assessment timing. Given the long
half-life of PFOS (median half-life = 3.5 years) (Li et al., 2018), current blood concentrations are
expected to correlate well with past exposures.
There are four studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and gastrointestinal effects. Study quality evaluations for these four studies are
shown in Figure C-43.
On the basis of the considerations mentioned, one study was considered medium confidence
(Timmermann et al., 2020) and three as low confidence (Xu et al., 2020d; Hammer et al., 2019;
Dalsager et al., 2016). The medium confidence study (Timmermann et al., 2020) relied on
retrospective reporting of GI outcomes, which is subject to recall bias, and did not detail the
interview question used. Study sensitivity was also limited by small case numbers and relatively
low PFOS exposure levels. However, the concerns were considered relatively minor and likely to
minimally impact interpretation of the results.
Concerns in the low confidence studies included potential for selection bias, including using
unclear recruitment methods and, a convenience sample (Xu et al., 2020d). Another concern was
potential for outcome misclassification or underreporting due to inconsistent participation and
adherence to the parent reporting mechanism (Dalsager et al., 2016). Another common reason for
low confidence was a serious risk for residual confounding by SES (Hammer et al., 2019).
Exposure misclassification was also a concern in Xu et al. (2020d), due to use of residential
history as a proxy. Deficiencies in multiple domains contributed to an overall low confidence
rating.
C-182
-------
APRIL 2024
,g6
Dalsager et al., 2016, 3858505 -
-
+ +
+
+
+
-
Hammer et al., 2019, 8776815-
+
-
+
-
-
Timmermann et al., 2020, 6833710 -
+
+
-
+
++
+
"
+
Xu et al., 2020, 6315709-
*
+
"
+
+
-*
Legend
D
Good (metric) or High confidence (overall)
+
Adequate (metric) or Medium confidence (overall)
-
Deficient (metric) or Low confidence (overall)
b
Critically deficient (metric) or Uninformative (overall)
*
Multiple judgments exist
Figure C-43. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Gastrointestinal Effects
Interactive figure and additional study details available on HAWC.
C.9.1.3 Findings
Both studies examining diarrhea observed non-significant increased association with PFOS
(Appendix D). Timmermann et al. (2020) observed increased odds of diarrhea in very young
children (up to 9 months old) in Guinea-Bissau. Dalsager et al. (2016) observed non-significant
increased incidence and inconsistent odds of diarrhea; similar inconsistent associations were
observed for vomiting when comparing exposure tertiles to the referent one in 1-4-year-old
children in Denmark.
Both studies examining IBD observed no associations with PFOS. Hammer et al. (2019)
observed a non-significant decrease in incidence of IBD in Faroese children and adults. Xu et al.
(2020d) observed non-significant decreases in levels of IBD biomarkers calprotectin or zonulin
in children and adults from Sweden.
C.9.2 Animal Evidence Study Quality Evaluation and Synthesis
There are two studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
C-183
-------
APRIL 2024
between PFOS and gastrointestinal effects. Study quality evaluations for these two studies are
shown in Figure C-44.
l
NTP, 2019, 5400978-
++ ++
++ ++
D
++
Thomford, 2002, 5432419-
++ ++
¦
++ ++
¦
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
Not reported
Figure C-44. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Gastrointestinal Effects
Interactive figure and additional study details available on HAWC.
Studies on the GI effects of PFOS exposure are limited. In a study conducted by NTP (2019),
male and female Sprague-Dawley rats were orally administered 0 mg/kg/day, 0.312 mg/kg/day,
0.625 mg/kg/day, 1.25 mg/kg/day, 2.5 mg/kg/day, or 5 mg/kg/day PFOS for 28 days. Animals
treated at 0 or 5 mg/kg/day showed no effects in the forestomach, glandular stomach, intestines,
pancreas, or salivary gland during histopathological examination (NTP, 2019).
The 2016 PFOS HESD identified an acute study in which male and female CD rats were
gavaged with a single dose of 0 mg/kg, 100 mg/kg, 215 mg/kg, 464 mg/kg, or 1,000 mg/kg of
PFOS suspended in a 20% acetone/80% corn oil mixture. Rats were observed for abnormal signs
for 4 hours after exposure and then daily for up to 14 days. All rats died in the 464 mg/kg and
1,000 mg/kg groups, and 3/10 rats died in the 215 mg/kg group. Necropsy results indicated
stomach distension and irritation of the glandular mucosa. According to the findings, the acute
oral LDso was 233 mg/kg in males, 271 mg/kg in females, and 251 mg/kg combined (Dean et al.,
1978).
The 2016 PFOS HESD also identified a sub-acute study in rhesus monkeys in which Goldenthal
et al. (1979) exposed two rhesus monkeys/sex/dose to 0 mg/kg/day, 0.5 mg/kg/day,
1.5 mg/kg/day, or 4.5 mg/kg/day of PFOS in distilled water by gavage for 90 days. All monkeys
in the 4.5 mg/kg/day group died or were euthanized in extremis by week 7 and exhibited signs of
GI tract toxicity (anorexia, emesis, black stool) (Goldenthal et al., 1979).
C-184
-------
APRIL 2024
C.9.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse GI outcomes in the 2016
PFOS HESD (U.S. EPA, 2016c). There are 10 studies from recent systematic literature search
and review efforts conducted after publication of the 2016 PFOS HESD that investigated the
mechanisms of action of PFOS that lead to GI effects. A summary of these studies is shown in
Figure C-45. Additional mechanistic synthesis will not be conducted since evidence is
inadequate to infer that PFOS leads to GI effects.
Mechanistic Pathway Animal Human In Vitro Grand Total
Big Data, Non-Targeted Analysis
1
0
0
1
Cell Growth, Differentiation, Proliferation, Or Viability
1
0
2
2
Extracellular Matrix Or Molecules
1
0
0
1
Fatty Acid Synthesis, Metabolism, Storage, Transport, Binding, B-Oxidation
4
0
0
4
Inflammation And Immune Response
1
0
2
2
Other
5
1
1
7
Grand Total
7
1
3
10
Figure C-45. Summary of Mechanistic Studies of PFOS and Gastrointestinal Effects
Interactive figure and additional study details available on HAWC.
C.9.4 Evidence Integration
The evidence evaluating an association between PFOS exposure and GI effects in humans is
indeterminate due to the limited number of studies available for evaluation and the
methodological shortcomings of those studies. In the 2016 PFOS HESD, GI outcomes in humans
were only assessed in the context of immune system health. Evidence is limited due to a paucity
of research and the quality of the available studies. The available research has not demonstrated
conclusive effects of PFOS on GI effects including vomiting, or diarrhea.
The animal evidence for an association between PFOS exposure and GI effects is indeterminate
based on the limited data available. The few studies that demonstrated GI effects in animal
models appeared to only observe effects in moribund or deceased individuals.
C.9.4.1 Evidence Integration Judgment
Overall, there is inadequate evidence to assess whether PFOS exposure can cause GI effects in
humans under relevant exposure circumstances (Table C-14).
C-185
-------
APRIL 2024
Table C-14. Evidence Profile Table for PFOS Gastrointestinal Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.9.1)
Diarrhea and vomiting
1 Medium confidence
study
1 Low confidence study
Two studies observed
modest, non-significant
positive associations for
diarrhea in children under
4 yr of age. One study
observed inconsistent
non-significant
associations with
vomiting across exposure
tertiles in children ages
1-4 yr. No studies were
conducted in adults.
• Medium confidence
study
• Low confidence study
• Lnconsistent direction of
effects across exposure
levels and endpoints
• Limited number of
studies examining
outcome
• Lmprecision of findings
• Potential outcome
misclassification or
underreporting due to
inconsistent parental
participation
Inflammatory bowel
disease
2 Low confidence
studies
One study in children and • No factors noted
adults observed a modest,
non-significant negative
association for IBD
incidence. One
community-based study
observed no clear
associations for IBD
biomarkers calprotectin
and zonulin.
• Low confidence studies
• Limited number of
studies examining
outcome
• Lmprecision of findings
• Potential for residual
confounding by
socioeconomic status and
decreased study
sensitivity
OOO
Lndeterminate
Evidence for
gastrointestinal effects is
based on one study
observing a modest, non-
significant association for
diarrhea and vomiting in
children under 4 yr of
age. Considerable
uncertainty due to limited
number of studies and
unexplained
inconsistency across
exposure levels and
endpoints.
OOO
Inadequate Evidence
Primary basis:
Evidence in humans and
animals are largely non-
significant.
Human relevance, cross-
stream coherence, and
other inferences:
No specific factors are
noted.
Evidence From In Vivo Animal Studies (Section C.9.2)
Histopathology
1 High confidence study
No changes in
forestomach, glandular
stomach, intestines,
pancreas, or salivary
gland histopathology in
one 28-day study in male
and female rats.
• High confidence
study
• Limited number of
studies examining
outcome
OOO
Lndeterminate
Evidence was limited to
one study reporting no
findings of
gastrointestinal toxicity.
C-186
-------
APRIL 2024
Notes: IBD = inflammatory bowel disease; yr = years.
C-187
-------
APRIL 2024
C.10 Dental
EPA identified two epidemiological studies that investigated the association between PFOS and
dental effects. No animal studies were identified. The two epidemiological studies were both
classified as medium confidence (Section C.10.1). Studies may have mixed confidence ratings
depending on the endpoint evaluated. Though low confidence studies are considered qualitatively
in this section, they were not considered quantitatively for the dose-response assessment (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.10.1 Human Evidence Study Quality Evaluation and Synthesis
C.10.1.1 Introduction
PFOS exposure could potentially adversely affect both dentin and bone mineralization, skeletal
formation, thyroid hormones that stimulate tooth maturation and enamel sufficiency, and
immune responses to cariogenic bacteria (Puttige Ramesh et al., 2019). At a molecular level,
PFAS such as PFOS may influence tooth growth and development via activation of peroxisome
proliferator-activated receptor alpha, initiation of oxidative stress, altering gene expression in the
vascular endothelial growth factor signaling pathway for gastric cells, hemoprotein binding,
estrogen disruption, or disruption of carbonic anhydrase (needed for enamel development)
(Wiener and Waters, 2019).
For this updated review, two studies examined the association between PFOS exposure and
dental caries (Puttige Ramesh et al., 2019; Wiener and Waters, 2019). The dental caries effect
was defined as presence of decay or a restoration on any tooth surface or the loss of a tooth
following tooth decay, excluding third molars (Puttige Ramesh et al., 2019). Trained dentists
performed visual and tactile exams using appropriate tools, but X-rays were not taken. No other
dental health outcomes were evaluated.
The two cross-sectional studies used data from the NHANES: Puttige Ramesh et al. (Puttige
Ramesh et al., 2019) assessed data from 2,869 12-19-year-old adolescents included in 1999-
2012 NHANES and Wiener and Waters (2019) examined data from 639 children ages 3-
11 years in the 2013-2014 NHANES cycle. Therefore, no participant overlap is expected
between these studies. Exposure to PFOS was assessed via biomarkers in blood.
C.10.1.2 Study Quality
Important considerations specific to evaluating the quality of studies on dental outcomes relate to
the difficulty of characterizing risk factors for dental caries, such as diet and oral hygiene
practices. Self-reported frequency of brushing, fluoridated product use, and dental visits may be
useful indicators. Fluoride levels in local public drinking water supplies are also thought to
influence development of dental caries and tap water consumption habits differ among
households and individuals (Wiener and Waters, 2019). Measuring PFOS and dental outcomes
concurrently was considered adequate in terms of exposure assessment timing. Given the long
half-life of PFOS (median half-life = 3.5 years) (Li et al., 2018), current blood concentrations are
expected to correlate well with past exposures.
There are two studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
C-188
-------
APRIL 2024
between PFOS and dental effects. Study quality evaluations for these two studies are shown in
Figure C-46.
On the basis of the considerations mentioned, the two included studies were considered medium
confidence, wherein limitations were not expected to severely affect results interpretation.
Limitations included cross-sectional study design, which introduces some concern about whether
the exposure preceded the outcome or vice versa (Puttige Ramesh et al., 2019; Wiener and
Waters, 2019). Puttige Ramesh et al. (2019) was primarily limited by participant selection, since
NHANES data necessarily excluded participants who were unable or unwilling to submit to a
dental examination. This could have resulted in selection bias against individuals with the most
severe tooth decay. Dental examinations were performed on all NHANES participants aged 2+
who did not have orofacial pain, specific medical conditions, physical limitations, inability to
comply, or were uncooperative.
Puttige Ramesh etal., 2019, 5080517-
Wiener et al., 2019, 5386081 -
O"W'^
,0®
Legend
| Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
Figure C-46. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Dental Effects
Interactive figure and additional study details available on HAWC.
C. 10.1.3 Findings
The two studies observed mixed effects (Puttige Ramesh et al., 2019; Wiener and Waters, 2019).
Wiener and Waters (2019) observed borderline significant increased odds of dental caries with
increased PFOS exposure in children (OR: 1.41; 95% CI: 0.97, 2.05; p-value = 0.069). The
analysis adjusted for age, sex, race/ethnicity, ratio of family-income-to-poverty guidelines, tooth
brushing frequency, fluoride in water, percentage of sugar in the diet, and dental visits. Puttige
Ramesh et al. (2019) observed increased odds of dental caries only in the third quartile of
exposure, but decreased odds in the second and highest quartiles compared with the lowest, and
per doubling of PFOS. Analyses did not account for age, but considered gender, race, education
level of parent/guardian, family-poverty-to-income ratio, blood lead level, and serum cotinine
level (an indicator of exposure to smoking). No studies of dental health outcomes were available
for pregnant women, adults, or occupational workers (Appendix D).
C-189
-------
APRIL 2024
C.10.2 Animal Evidence Study Quality Evaluation and Synthesis
In the available literature, there is no reported biological consequence of PFOS exposure on
dental outcomes in animals.
C.10.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse dental outcomes in the
2016 PFOS HESD (U.S. EPA, 2016c). There are no studies from recent systematic literature
search and review efforts conducted after publication of the 2016 PFOS HESD that investigated
the mechanisms of action of PFOS that lead to dental effects. Additional mechanistic synthesis
will not be conducted since evidence is inadequate to infer that PFOS may cause dental effects.
C.10.4 Evidence Integration
The evidence evaluating an association between PFOS exposure and dental effects in humans is
indeterminate based on the limited number of available studies. Dental health outcomes were not
previously reviewed in the 2016 PFOS HESD (U.S. EPA, 2016c). The present review was
limited by the availability of only two studies. Only one outcome was examined (prevalence of
dental caries), and while both studies observed increased odds of dental carries, the associations
were non-significant (Puttige Ramesh et al., 2019; Wiener and Waters, 2019). These studies have
exposure levels typical in the general population, large sample sizes and low risk of bias.
The animal evidence for an association between PFOS exposure and dental effects is
indeterminate because there are no available studies in animal models examining the effects of
PFOS exposure on dental outcomes.
C. 10.4.1 Evidence Integration Judgment
Overall, there is inadequate evidence to assess whether PFOS exposure can cause dental effects
in humans under relevant exposure circumstances (Table C-15).
C-190
-------
APRIL 2024
Table C-15. Evidence Profile Table for PFOS Dental Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Dental caries
2 Medium confidence
studies
Summary and Key
Findings
Two studies observed
non-significant increases
and decreases in odds of
dental caries. No
significant associations
observed in studies of
children and adolescents
fromNHANES.
Factors That Increase
Certainty
• Medium confidence
studies
Factors That Decrease
Certainty
• Inconsistent direction of
effects across studies and
across exposure levels
• Limited number of
studies examining
outcome
• Imprecision of findings
Evidence Stream
Judgment
ooo
Indeterminate
Evidence was limited to
two studies that reported
non-significant positive
associations to dental
caries in children and
adolescents, but results
are imprecise.
Uncertainty remains
regarding adults and
other age groups from the
general population.
Evidence Integration
Summary Judgment
OOO
Inadequate Evidence
Primary basis:
No evidence in animals
and evidence in humans is
largely non-significant.
Human relevance, cross-
stream coherence, and
other inferences'.
No specific factors are
noted.
Evidence From Studies of Exposed Humans (Section C.10.1)
Notes: NHANES = National Health and Nutation Examination Survey; N/A = not applicable.
C-191
-------
APRIL 2024
C.ll Ocular
EPA identified one epidemiological and two animal studies that investigated the association
between PFOS and ocular effects. The epidemiological study was classified as medium
confidence (Section C.l 1.1). Of the animal studies, one was classified as high confidence, and
one was considered low confidence (Section C.ll .2). Studies may have mixed confidence ratings
depending on the endpoint evaluated. Though low confidence studies are considered qualitatively
in this section, they were not considered quantitatively for the dose-response assessment (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.l 1.1 Human Evidence Study Quality Evaluation and Synthesis
C.ll. 1.1 Introduction
For this updated review, there is one epidemiological study that investigated the association
between PFOS and ocular effects (Zeeshan et al., 2020).
This cross-sectional study conducted in Shenyang, China as part of the Isomers of C8 Health
Project in China focused on a high-exposed population, including adults aged 20 years and older,
who were randomly selected using multistage, stratified cluster sampling. Median total PFOS
serum concentrations among the 1,202 study participants were 24.07 ng/mL. Participants were
subject to a complete ophthalmic examination which included ocular history, visual acuity, and
anterior and posterior segment examinations. Several ocular conditions, reflecting effects on
different segments of the eyes, were assessed, including visual impairment (VI), vitreous
disorder, synechia, macular disorder, corneal pannus, anterior chamber depth (ACD)-shallow,
retinal disorder, lens opacity, and conjunctival disorder.
C.ll. 1.2 Study Quality
There is one study from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and ocular effects. Study quality evaluation for this one study is shown in Figure
C-47.
Zeeshan et al. (2020) was classified as medium confidence. The main limitation of this study is
the cross-sectional design, which does not allow for establishing temporality. Participants' serum
samples were collected at study enrollment only and the utilization of a single exposure
measurement may not adequately represent exposure variability; additionally, it is unclear
whether exposure occurred at an etiologically relevant time period to reflect changes in ocular
function.
C-192
-------
APRIL 2024
Zeeshan et al., 2020, 6315698-
,6\6'
.„r^®
?8^' 0^0°^°
,G©
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
Figure C-47. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Ocular Effects
Interactive figure and additional study details available on HAWC.
C. 11.1.3 Findings
Zeeshan et al. (2020) examined the effects of exposure to PFOS in adults aged 22-96 years, who
had lived for at least 5 years in in Shenyang, China (Appendix D). Ocular outcomes examined
include VI, vitreous disorder, synechia, macular disorder, corneal pannus, and ACD, and
combined eye disease (aggregating all ocular conditions examined). A positive statistically
significant association between VI and total serum PFOS was observed (OR: 3.11; 95% CI: 2.30,
4.20). When stratified by age, the association between combined eye disease and total serum
PFOS was statistically significant for participants aged < 65 years (OR: 1.52; 95%, 1.21, 1.91),
but not for the older participants (OR: 0.91; 95% CI: 0.55, 1.51). No other associations were
observed.
C.11.2 Animal Evidence Study Quality Evaluation and Synthesis
There are two studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and ocular effects. Study quality evaluations for these two studies are shown in
Figure C-48.
C-193
-------
APRIL 2024
NTP, 2019, 5400978
Thomford, 2002, 5432419
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
NR Not reported
Figure C-48. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Ocular Effects
Interactive figure and additional study details available on HAWC.
An eye irritation study in rabbits suggests that PFOS acts as an ocular irritant (Biesemeier and
Harris, 1974); however, in a 28-day oral toxicity study conducted by NTP, no histological
abnormalities were noted in male or female Sprague-Dawley rats exposed to 5 mg/kg/day PFOS
(NTP, 2019).
C.11.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse ocular outcomes in the
2016 PFOS HESD (U.S. EPA, 2016c). There is one study from recent systematic literature
search and review efforts conducted after publication of the 2016 PFOS HESD that investigated
the mechanisms of action of PFOS that lead to ocular effects. A summary of these studies is
shown in Figure C-49. Additional mechanistic synthesis will not be conducted since evidence is
inadequate to infer that PFOS leads to ocular effects.
C-194
-------
APRIL 2024
Mechanistic Pathway
In Vitro
Grand Total
Atherogenesis And Clot Formation
Cell Growth, Differentiation, Proliferation, Or Viability
Cell Signaling Or Signal Transduction
Inflammation And Immune Response
Grand Total
1
Figure C-49. Summary of Mechanistic Studies of PFOS and Ocular Effects
Interactive figure and additional study details available on HAWC.
The evidence evaluating an association between PFOS exposure and ocular effects in humans is
indeterminate due to limited evidence available from epidemiological studies. In the 2016 Health
Assessment for PFOS, no epidemiological evidence of an association between PFOS exposure
and ocular health effects was examined. One epidemiological study reported an association
between PFOS and VI and combined eye disease in humans. However, since only one study was
available for review and given its cross-sectional design, existing epidemiological evidence does
not allow for a definitive conclusion regarding potential detrimental ocular health effects due to
exposure to PFOS.
The association between PFOS and ocular effects is indeterminate due to the limited evidence
available in animal models. One available study in an animal model did not report
histopathological ocular abnormalities.
C. 11.4.1 Evidence Integration Judgment
Overall, there is inadequate evidence to assess whether PFOS exposure can cause ocular effects
in humans under relevant exposure circumstances (Table C-16).
C.11.4 Evidence Integration
C-195
-------
APRIL 2024
Table C-16. Evidence profile table for PFOS Ocular effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase
Certainty
Factors That Decrease
Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.ll.l)
Eye disease
1 Medium confidence
study
Histopathology
1 High confidence
study
The only study
examining eye disease
was a cross-sectional
study that observed
significant positive
associations between
visual impairment and
serum PFOS. The
association was also
significant for combined
eye disease, but only in
participants aged <65 yr.
• Medium confidence
study
• Limited number of
studies examining
outcome
Indeterminate
ooo
Evidence was limited to
one study reporting
increases in visual
impairment in all ages and
increases in combined eye
disease in participants
aged <65 yr.
Evidence From In Vivo Animal Studies (Section C.11.2)
No changes in ocular
histopathology were
reported in one 28-day
study in male and female
rats.
• High confidence
study
• Limited number of
studies examining
outcome
Indeterminate
OOO
Evidence was limited to
one study reporting no
findings of ocular
toxicity.
OOO
Inadequate Evidence
Primary basis:
Evidence in humans is
limited and evidence in
animals is largely non-
significant.
Human relevance, cross-
stream coherence, and
other inferences'.
No specific factors are
noted.
Notes: yr = years.
C-196
-------
APRIL 2024
C.12 Dermal
EPA identified one epidemiological and two animal studies that investigated the association
between PFOS and dermal effects. The epidemiological study was classified as medium
confidence (Section C.12.1). Of the animal studies, one was classified as high confidence, and
one was considered low confidence (Section C.12.2). Studies may have mixed confidence ratings
depending on the endpoint evaluated. Though low confidence studies are considered qualitatively
in this section, they were not considered quantitatively for the dose-response assessment (see
Toxicity Assessment, (U.S. EPA, 2024)).
C.12.1 Human Evidence Study Quality Evaluation and Synthesis
C.12.1.1 Introduction
For this updated review, one study examined the association between age at the occurrence of
acne and PFOS exposure. In the Puberty Cohort, a large sub-cohort of the DNBC in Denmark,
Ernst et al. (2019) examined the association between prenatal PFOS exposure and pubertal
development. Mother-child pairs were recruited for the DNBC from 1996 to 2002, and eligibility
for the Puberty Cohort was determined in 2012. PFAS levels in maternal blood, largely collected
during the first trimester of pregnancy, were used to assess prenatal exposure, and age at the
occurrence of acne was self-reported by children via bi-annual questionnaire starting in 2012 or
at 11 years of age.
C.12.1.2 Study Quality
There is one study from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and dermal effects. Study quality evaluation for this one study is shown in Figure
C-50.
Ernst et al. (2019) was considered a medium confidence study, with no major concerns with the
overall quality of the study and any identified concerns were not likely to impact the results.
Self-reporting was used to assess the occurrence of acne, a study limitation that could introduce
minor bias to the outcome assessment. Additionally, some children were sampled for the Puberty
Cohort after the onset of puberty, thus their self-reported outcome information has increased risk
of inaccurate recall. However, this was not expected to substantially impact the accuracy of the
outcome measures.
C-197
-------
APRIL 2024
0^0°^
AO'
wt^Ve°
#>
O"4
,o®
Ernst etal., 2019, 5080529-
II 1 1 1 1 1 1 II
++
+
+
+
++
+
+
~
Legend
Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
Figure C-50. Summary of Study Quality Evaluation Results for Epidemiology Studies of
PFOS Exposure and Dermal Effects
Interactive figure and additional study details available on HAWC.
C. 12.1.3 Findings
Ernst et al. (2019) observed negative non-significant associations between prenatal PFOS
exposure and age at the occurrence of acne in both boys and girls. Associations remained
negative and non-significant in analyses stratified by tertiles, except for girls in the second tertile
of PFOS exposure compared with the lowest (P: 0.09; 95% CI: -4.69, 4.87) (Ernst et al., 2019).
Associations in boys were negative and non-significant (Appendix D).
C.12.2 Animal Evidence Study Quality Evaluation and Synthesis
There are two studies from recent systematic literature search and review efforts conducted after
publication of the 2016 PFOS HESD (U.S. EPA, 2016c) that investigated the association
between PFOS and dermal effects. Study quality evaluations for these two studies are shown in
Figure C-51.
C-198
-------
APRIL 2024
NTP, 2019, 5400978-
Thomford, 2002, 5432419-
Legend
| Good (metric) or High confidence (overall)
Adequate (metric) or Medium confidence (overall)
Deficient (metric) or Low confidence (overall)
Critically deficient (metric) or Uninformative (overall)
NR Not reported
Figure C-51. Summary of Study Quality Evaluation Results for Animal Toxicological
Studies of PFOS Exposure and Dermal Effects
Interactive figure and additional study details available on HAWC.
There is no evidence in the literature that oral PFOS exposure results in dermal toxicity. In a 28-
day oral gavage study in male and female Sprague-Dawley rats with PFOS concentrations up to
5 mg/kg/day, no dermal lesions were observed during histopathological observation (NTP,
2019).
C.12.3 Mechanistic Evidence
There was no mechanistic evidence linking PFOS exposure to adverse dermal outcomes in the
2016 PFOS HESD (U.S. EPA, 2016c). There are no studies from recent systematic literature
search and review efforts conducted after publication of the 2016 PFOS HESD that investigated
the mechanisms of action of PFOS that lead to dermal effects. Additional mechanistic synthesis
will not be conducted since evidence is inadequate to infer that PFOS may cause dermal effects.
C.12.4 Evidence Integration
The evidence evaluating an association between PFOS exposure and dermal effects in humans is
indeterminate based on the limited number of studies available. In the 2016 PFOS HESD (U.S.
EPA, 2016c), the association between oral PFOS exposure and dermal effects was not examined.
In this updated review of the epidemiologic literature, one study examined the association
between prenatal PFOS exposure and dermal effects during puberty (Ernst et al., 2019) and
observed negative non-significant associations in both boys and girls in the study cohort.
However, conclusions regarding PFOS exposure and resulting dermal effects are limited by the
lack of studies examining the association. Dermal effects beyond acne are not currently
represented in the epidemiological literature.
C-199
-------
APRIL 2024
The evidence for potential dermal effects in experimental animals in indeterminate and limited to
a single high confidence study with no dermal lesions observed. In the available literature from
animal models, there is no reported biological consequence of oral PFOS exposure on dermal
tissue.
C. 12.4.1 Evidence Integration Judgment
Overall, there is inadequate evidence to assess whether PFOS exposure can cause dermal effects
in humans under relevant exposure circumstances (Table C-17).
C-200
-------
APRIL 2024
Table C-17. Evidence Profile Table for PFOS Dermal Effects
Evidence Stream Summary and Interpretation
Studies and
Interpretation
Summary and Key
Findings
Factors That Increase Factors That Decrease
Certainty Certainty
Evidence Stream
Judgment
Evidence Integration
Summary Judgment
Evidence From Studies of Exposed Humans (Section C.12.1)
Acne
1 Medium confidence
study
One study found negative
non-significant
associations with age of
acne onset among
adolescent girls and boys.
• Medium confidence
study
• Limited number of
studies examining
outcome
• Imprecision of findings
ooo
Indeterminate
Evidence was limited to
one study reporting non-
significant associations.
Evidence From In Vivo Animal Studies (Section C.12.2)
Histopathology No changes in skin
1 High confidence study histopathology were
reported in one 28-day
study in male and female
rats.
• High confidence
study
• Limited number of
studies examining
outcome
OOO
Indeterminate
Evidence was limited to
one study reporting no
findings of dermal
toxicity.
OOO
Inadequate Evidence
Primary basis:
Evidence in humans and
animals are largely non-
significant.
Human relevance, cross-
' stream coherence, and
other inferences'.
No specific factors are
noted.
C-201
-------
APRIL 2024
Appendix D. Detailed Information from Epidemiology Studies
D.l Developmental
Table D-l. Associations Between PFOS Exposure and Developmental Effects in Recent Epidemiological Studies
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels"
Resultsb
Ashley-Martin
etal. (2017)
High
Canada, 2008-
2011
Cohort
Pregnant women Maternal blood BW (z-score): Regression
(enrolled if
<14 wk gestation,
>18 yr of age) and
their infants at
recruitment and
from MIREC
N = 1,509
Early pregnancy
4.6 (3.2-6.8)
adequate,
inadequate,
and excess
weight gain
coefficient per
loglO-unit
increase PFOS
BW: 0.05 (-0.18,0.29)
Females: 94.31 (-76.3, 264.92)
Males: -11.15 (-174.26, 151.95)
Adequate weight gain: -0.03
(-0.49, 0.41)
Excess weight gain: 0.25 (-0.11,
0.62)
Inadequate weight gain: -0.24
(-0.95, 0.45)
MIREC = Maternal-Infant Research on Environmental Chemicals
Outcome: Weight gain adequacy based on Institute of Medicine (IOM) guidelines
Confounding: Maternal age, pre-pregnancy BMI, parity, household income, smoking, each PFAS.0
Bach et al.
Denmark,
Cohort
Pregnant women
Maternal serum
BL (cm), BW
Regression
BL: 0 (-0.1, 0.2)
(2016)
2008-2013
and their infants
Early pregnancy
(g, z-score),
coefficient or
Q2
-0.3 (-0.7, 0)
High
from the Aarhus
8.3 (6.0-10.8)
gestational
OR (PTB) per
Q3
-0.1 (-0.4,0.3)
Birth Cohort
length
IQR increase in
Q4
-0.1 (-0.5,0.2)
N = 1,507
(weeks), HC
PFOS and by
(cm), PTB
quartiles
BW (g): -8 (-30, 14)
Q2
-66 (-122,-11)
Q3
-30 (-86, 26)
Q4
-58 (-105, 8)
Females: -32 (-71, 7)
Q2
-44 (-140, 52)
Q3
-55 (-148, 38)
Q4
-71 (-174, 31)
Males: 26 (-13, 65)
Q2
-129 (-239, -19)
Q3
9 (-93, 110)
Q4
-37 (-141, 67)
D-l
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
BW (z-score): -0.02 (-0.07, 0.04)
Q2
Q3
Q4
-0.15 (-0.29,-0.02)
-0.06 (-0.19,0.07)
-0.11 (-0.25,0.02)
Gestational length: 0 (-0.1, 0.1)
Q2:-0.1 (-0.4,0.1)
Q3:0(-0.2, 0.3)
Q4: 0 (-0.3, 0.2)
HC: 0 (-0.1, 0.1)
Q2
Q3
Q4
-0.2 (-0.5, 0)
-0.1 (-0.4,0.1)
-0.1 (-0.3,0.2)
PTB: 0.85 (0.6, 1.21)
Q2
Q3
Q4
0.96 (0.53, 1.74)
0.65 (0.34, 1.26)
0.82 (0.44, 1.53)
Results: Lowest quartile used as reference.
Confounding: Maternal age, pre-pregnancy BMI and educational level, GA.
Bell et al.
United States, Cross-sectional Singleton and twin Blood BL (cm), BW
Regression
BL
(2018)
2008-2010 infants born in
Later pregnancy (g), GA
coefficient per
S: -0.04 (-0.10,0.1)
High
from Upstate
Singletons: 1.72 (weeks), HC
log(PFOS+l)
T: 0.23 (-0.07, 0.53)
KIDS
(1.14-2.44) (cm),
unit increase
N = 2,071
Twins: 1.64 ponderal
BW
singletons; 1,040
(1.09-2.33) index
S: -18.32 (-42.41, 5.78)
twins
T: 3.91 (-31.07, 38.89)
GA
S: 0.05 (-0.03,0.13)
T: -0.02 (-0.15,0.11)
HC
S: 0.03 (-0.19,0.24)
D-2
-------
APRIL 2024
Reference, Location, Population,Ages, MaS°S«mple Outcome Comparison Rente-
„ j.. . Design Matrix, Sample
Confidence Years N Timing, Levels'
T: 0.23 (-0.04, 0.49)
Ponderal index
S: -0.01 (-0.03, 0.01)
T: -0.01 (-0.04, 0.01)
Comparison: Logarithm base not specified.
Results: S = Singletons; T = Twins
Confounding: Maternal age, maternal BMI, maternal education, infertility treatment, parity.
Bjerregaard
Olesen et al
(2019)
High
Denmark, Cohort Pregnant women Maternal serum BL (cm), BW Regression BL:-0.1 (-0.3, 0.2)
2011-2013 and their children Early pregnancy (g), HC (cm) coefficient per Females:-0.4 (-0.8, 0)
from FETOTOX IQR = 4.12 IQR increase in Males: 0.2 (-0.1, 0.5), Interaction
N = 671 serum PFOS p-value = 0.022
BW: -15 (-62, 32)
Females: -81 (-147, -14)
Males: 38 (-28, 105), Interaction
p-value = 0.013
HC:0 (-0.2,0.1)
Females: -0.1 (-0.4, 0.1)
Males: 0 (-0.2, 0.2), Interaction
p-value = 0.404
Confounding: Age at delivery, pre-pregnancy BMI, educational level, smoking, alcohol intake, GA at birth.
Buck Louis et
United States, Cohort Pregnant women
Maternal blood
Umbilical Regression
Umbilical circumference: 0.04
al. (2018)
2009-2013 (age range 18-
Early pregnancy
circumference coefficient per
(-0.09, 0.16)
High
40 yr) with
5.13 (3.39-7.89) (cm), upper SD increase in
Upper arm length: -0.04 (-0.1,
singleton
arm length logPFOS
0.1)
pregnancies from
(cm), upper
Upper thigh length: -0.03 (-0.1,
the NICHD Fetal
thigh length
0.04)
Growth Studies
(cm)
N = 2,106
NICHD = National Institute of Child Health and Human Development
Comparison: Logarithm base not specified.
Confounding: Maternal age, education, pre-pregnancy BMI, serum cotinine, infant sex, chemical-maternal race/ethnic interaction.
D-3
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Chu et al.
(2020)
High
China,
2013
Cohort Pregnant women
(aged 18-45 yr)
and infants from
Guangzhou Birth
Cohort Study
N = 372
Maternal serum
Later pregnancy
1.538 (0.957-
2.635)
Females: 1.497
(0.920-2.642)
Males: 1.558
(0.988-2.628)
BW (g), GA
(weeks),
LBW, PTB
Regression
coefficient (BW,
GA) or OR
(LBW, PTB) per
ln-unit increase
in PFOS or by
quartiles
BW: -83.28 (-133.2, -33.36)
Females: -71.91 (-143.86, 0.05)
Males: -71.52 (-142.44, -0.61)
p-value for interaction by
sex = 0.678
GA:-0.32 (-0.53,-0.11)
Females: -0.61 (-0.9, -0.32)
Males: 0.004 (-0.31,0.32)
p-value for interaction by
sex = 0.003
LBW: 2.43 (1.08, 5.47)
Q2: 0.83 (0.11,6.47)
Q3: 1.41 (0.23,8.82)
Q4: 3.7 (0.61, 22.58)
p-trend < 0.001
PTB: 2.03 (1.24, 3.32)
Q2
Q3
Q4
2.22 (0.55, 9.05)
4.52 (1.21, 16.88)
4.99 (.134, 18.56)
p-trend = 0.003
Outcome: LBW defined as BW < 2500 g
Results: Lowest quartile used as reference.
Confounding: Maternal age, maternal occupation, maternal education, family income, parity for all outcomes; GA for BW and LBW; child
sex for BW and GA.
Costa et al. Spain, 2003- Cohort Pregnant women
(2019) 2008 and their children
High from INMA study
N = 1,230
(Girls = 597,
Boys = 633)
Maternal plasma AC, FL, BPD, Percent change
estimated fetal per twofold
6.05 (4.52-7.82) weight at increase in
12 wk, 20 wk, PFOS
and 34 wk
AC
12 wk: 1.4 (-2.1,4.9)
Girls: 2.3 (-2.8,7.1)
Boys: 0.8 (-3.8, 5.4)
20 wk: 2.2 (-1.3, 5.6)
Girls: 4.0 (-0.9, 8.8)
Boys: 0.5 (-4.1, 5.0)
34 wk: 2.1 (-1.3, 5.5)
Girls: 1.2 (-3.6, 5.8)
D-4
-------
APRIL 2024
Reference, Location, Population,Ages, MaS°S«mple Outcome Comparison Rente-
„ j.. . Design Matrix, Sample
Confidence Years N Timing, Levels'
Boys: 2.8 (-1.8, 7.2)
FL
12 wk: 1.2 (-2.3,4.8)
Girls: 0.3 (-4.7, 4.9)
Boys: 2.0 (-2.6, 6.6)
20 wk: -0.6 (-4.1,2.9)
Girls: -1.7 (-6.5, 3.1)
Boys: 0.0 (-4.6, 4.7)
34 wk: 1.2 (-4.1,6.5)
Girls: 1.3 (-3.6,6.1)
Boys: 1.7 (-2.9, 6.2)
BPD
12 wk: 0.5 (-3.0, 3.9)
Girls: 1.6 (-3.3, 6.4)
Boys: -0.9 (-8.2, 6.3)
20 wk: 1.3 (-2.3,4.8)
Girls: 1.2 (-3.7, 6.0)
Boys: 1.2 (-3.5, 5.9)
34 wk: 0.9 (-2.7, 4.4)
Girls: 0.0 (-4.9, 4.7)
Boys: 1.2 (-3.5, 5.9)
Estimated Fetal Weight
12 wk: 1.9 (-1.7, 5.4)
Girls: 1.3 (-3.5,6.2)
Boys: 2.5 (-2.3,7.1)
20 wk: 2.6 (-0.9,6.1)
Girls: 2.4 (-2.4, 7.2)
Boys: 1.0 (-3.7, 5.3)
34 wk: 2.6 (-0.9,6.1)
Girls: 1.8 (-3.2, 6.5)
Boys: 3.0 (-1.7, 7.5)
INMA = INfancia y Medio Ambiente (Environment and Childhood) Project
D-5
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Confounding: Cohort, parity, maternal age, country of birth, smoking at week 12, maternal pre-pregnancy BMI, studies, season of last
menstrual period.
Darrow et al. United States Cohort Pregnant women
(2013) 2005-2011 from the C8HP
High exposed through
drinking water,
Ages >19
LBW, all births
N = 1,629
LBW, first
prospective birth
N = 783
B W, all births
N = 1,470
BW, first
prospective birth
N = 710
PTB, all births
N = 1,628
PTB, first
prospective birth
N = 783
Maternal serum LBW, BW
at enrollment (g), PTB
13.9(9.5-19.7)
OR (LBW,
PTB)and
regression
coefficient (BW)
per ln-unit
increase in
PFOS, per IQR
increase in
PFOS, or by
quintiles
LBW
All births
Per ln-unit increase: 1.12 (0.75,
1.67)
Per IQR increase: 1.12(0.87, 1.44)
Q2:1.48 (0.71,3.08)
Q3: 1.23 (0.57,2.65)
Q4: 1.31 (0.59,2.94)
Q5:1.33 (0.60,2.96)
p-value for trend = 0.651
First prospective birth
Per ln-unit increase: 0.97 (0.61,
1.54)
Per IQR increase: 0.93 (0.63, 1.37)
Q2: 1.65 (0.52, 5.20)
Q3: 0.95 (0.30, 3.01)
Q4: 1.17 (0.36,3.78)
Q5: 0.82 (0.25, 2.70)
p-value for trend = 0.484
BW
All births
Per ln-unit increase: -29 (-66, 7)
Per IQR increase: -23 (-48, 3)
Q2: -25 (-96, 48)
Q3: -37 (-109,35)
Q4: -83 (-152, -13)
Q5: -54 (-124, 17)
p-value for trend = 0.045
First prospective birth
Per ln-unit increase: -49 (-90, -8)
Per IQR increase: -29 (-58, 0)
Q2: -33 (-140, 74)
Q3:-115 (-216,-14)
D-6
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Q4: -149 (-244, -54)
Q5: -105 (-196, -13)
p-value for trend = 0.006
PTB
All births
Per ln-unit increase: 1.02 (0.78,
1.35)
Per IQR increase: 1.03 (0.83, 1.27)
Q2: 1.11 (0.63, 1.94)
Q3: 0.76 (0.42, 1.36)
Q4: 1.00 (0.56, 1.78)
Q5: 1.07 (0.58, 1.95)
p-value for trend = 0.976
First prospective births
Per ln-unit increase: 1.02 (0.72,
1.45)
Per IQR increase: 0.95 (0.73, 1.25)
Q2: 1.07 (0.44,2.59)
Q3: 0.63 (0.25, 1.59)
Q4: 1.08 (0.47,2.46)
Q5: 0.86 (0.36,2.04)
p-value for trend = 0.818
C8HP = C8 Health Project
Outcome: PTB defined as births occurring before 37 wk gestation. LBW defined as those weighing less than 2,500 g.
Results: Lowest quintile used as reference.
Confounding: Maternal age, educational level, smoking status, parity, BMI, self-reported diabetes, time between conception and serum
management (year strata). Additional confounding for B W: indicator variables for gestational week.
Eick et al.
(2020)
High
United States
2014-2018
Cohort
Second trimester Maternal serum BW (g, z- Regression
pregnant women
from the CIOB
cohort
BW (g)
N = 461
from the second score), GA coefficient by
trimester (weeks), PTB tertile
1.93 (1.18-3.13) PTB:
OR by tertile
BW (g)
T2: 1.62 (-105.53, 108.77)
T3: 14.26 (-101.51, 130.03)
BW (z-score)
T2: -0.01 (-0.24,0.22)
T3: 0.02 (-0.23,0.27)
D-7
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
GA, BW (z-score),
PTB
N = 506
GA
T2:-0.19 (-0.64, 0.26)
T3: -0.08 (-0.59,0.43)
PTB
T2: 1.21 (0.50,2.91)
T3: 1.87 (0.72,4.88)
CIOB = Chemicals in our Bodies
Outcome: PTB defined as birth at <37 wk gestation.
Results: Lowest tertile used as reference.
Confounding: Maternal age, maternal race/ethnicity, pre-pregnancy BMI, maternal education, smoking status, parity, food insecurity.
Gardener et al. United States Cohort Pregnant women Maternal serum GA at birth GA at birth and
(2021) Recruitment: in third trimester from primarily (weeks), BW BW: Mean by
High 2009 (ages 18-49) and third trimester (z-score), GA quartile
children at birth 3.9(2.6-5.9) <37 wk
from the Vanguard GA <37 weeks
Pilot Study of the and B W: OR by
GA at birth
Mean
Ql: 38.92 (38.58,39.26)
Q2: 38.53 (38.19, 38.87)
Q3: 38.77 (38.43,39.09)
Q4: 38.77 (38.42,39.10)
NCS
quartile p-trend = 0.77
GA at birth
BW
N = 433
Mean
BW
Ql:-1.15 (-4.63,2.32)
N = 403
Q2: 0.56 (-2.72, 3.84)
Q3: 1.16 (-2.06,4.38)
Q4: 1.10 (-2.29,4.46)
p-trend = 0.35
OR
Q2: 0.93 (0.43, 2.04)
Q3: 1.41 (0.66,3.03)
Q4: 0.81 (0.36, 1.82)
p-trend = 0.40
GA <37 wk
OR
Q2: 1.94 (0.66, 5.68)
Q3: 1.13 (0.34,3.73)
D-8
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Q4: 1.41 (0.46,4.33)
p-trend = 0.82
NCS = National Children's Study
Results: Lowest quartile used as reference.
Confounding: Maternal age, education, race/ethnicity, pre-pregnancy BMI, prenatal smoking, parity, GA at serum collection.
Govarts et al.
(2016)
High
Belgium, 2008- Cohort
2009
Mother-newborn
pairs from FLEHS
II
N = 213
Cord blood
2.63 (iL (1.70-
3.90 |xL)
BW (g)
Regression
coefficient per
IQR increase in
PFOS
10.82 (-72.4, 94.05),
p-value = 0.798
FLEHS II = Flemish Environmental and Health Study II
Confounding: GA, child's sex, smoking of the mother during pregnancy, parity, maternal pre-pregnancy BMI.
Huo et al., 2020, China, 2013- Cohort Mothers Maternal blood
6835452 2016 (aged > 20 yr) and Later pregnancy
High their children from 9.33 (6.54-
the Shanghai Birth 13.65)
Cohort
N = 2,849
GA (weeks),
PTB
(indicated,
non-
spontaneous,
spontaneous,
and overall)
Regression
coefficient (GA)
per ln-unit
increase in
PFOS and per
tertile
OR (PTB) per
ln-unit increase
in PFOS and per
tertile
GA: 0.02 (-0.08,0.12)
Tl: -0.27 (-0.62,0.08)
T2: 0.26 (-0.43, 0.96)
T3: 0.03 (-0.24,0.29)
OR T2: 0.08 (-0.06,0.21)
OR T3: 0.06 (-0.08,0.19)
PTB, overall: 0.86 (0.63, 1.17)
T2: 0.61 (0.4, 0.94)
T3: 0.73 (0.48, 1.1)
Tl (per ln-unit increase): 2.67
(0.85, 8.29)
T2 (per ln-unit increase): 0.63
(0.05, 8.04)
T3 (per ln-unit increase): 0.83
(0.33, 2.08)
Females: 0.74 (0.45, 1.16)
Males: 0.94 (0.62, 1.41)
PTB, indicated: 1.13 (0.64, 2.01)
T2: 0.79 (0.35, 1.78)
T3: 0.99 (0.46,2.12)
PTB, non-spontaneous
Females: 1.35 (0.56, 3.26)
D-9
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Males: 0.98 (0.46, 2.09)
PTB, spontaneous: 0.77 (0.53,
1.11)
T2: 0.56 (0.34, 0.94)
T3: 0.65 (0.4, 1.05)
Females: 0.59 (0.33, 1.06)
Males: 0.93 (0.57, 1.5)
Results: Lowest tertile used as reference.
Confounding: Maternal age, pre-pregnancy BMI, parity, parental education levels, pregnancy complicated with chronic disease, infant sex,
GA at blood drawing.
Lauritzen et al.
Norway and Cohort
Mother-infant Maternal serum
BL (cm), BW
Regression
BL: -0.3 (-0.7,0.1),
(2017)
Sweden, 1986-
pairs from NICHD Later pregnancy
(g), GA
coefficient or
p-value = 0.139
High
1988
SGA Norway: 9.74
(weeks), HC
OR (SGA) per
NO: 0 (-0.4, 0.4), p-value = 0.987
N = 424 (265 from (Range = 0.95-
(cm), SGA
ln-unit increase
SE:-1.2 (-2.1,-0.3),
Norway, 159 from 59.6)
in PFOS
p-value = 0.007
Sweden (78 girls,
81 boys)) Sweden: 16.4
BW: -15.1 (-111, 80.7),
(Range = 2.28-
p-value = 0.757
55.2)
NO: 74 (-31, 178), p-value = 0.167
SE: -292 (-500, -84),
p-value = 0.006
GA: -0.07 (-0.34, 0.2),
p-value = 0.601
NO: -0.01 (-0.3, 0.3),
p-value = 0.952
SE: -0.4 (-0.9, 0.2),
p-value = 0.201
HC: 0.04 (-0.19,0.27),
p-value = 0.748
NO: 0.2 (-0.1,0.4),
p-value = 0.189
SE: -0.4 (-0.9, 0.04),
p-value = 0.073
D-10
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
SGA: 0.95 (0.62, 1.48),
p-value = 0.833
NO: 0.71 (0.42, 1.2),
p-value = 0.201
SE: 2.51 (0.93,6.77),
p-value = 0.068
NICHD SGA = The U.S. National Institute of Child Health and Human Development (NICHD) Scandinavian Successive Small-for-
Gestational-Age Births Study
Outcome: SGA defined as BW below the 10th percentile for GA, sex, and parity.
Results: NO = Norway; SE = Sweden
Confounding: Maternal age, height, pre-pregnancy BMI, education, parity, smoking status at conception, interpregnancy interval, offspring
sex.
Lind et al. Denmark Cohort Infants prenatally Maternal serum BW (g), HC Regression
(2017a) 2010-2012 exposed to PFAS Early pregnancy (cm), coefficient per
High from the Odense 8.1(6.0-11.1) gestational ln-unit increase
Child Cohort length in PFOS or by
N = 212 girls, 299 (days) quartiles
boys
BW
Males
Continuous: -17 (-130, 97)
p-trend by quartiles = 0.73
Females
Continuous: 92 (-15, 199)
p-trend by quartiles = 0.15
HC
Males
Continuous: -0.2 (-0.6, 0.2)
p-trend by quartiles = 0.38
Females
Continuous: 0.3 (-0.1, 0.7)
p-trend by quartiles = 0.12
Gestational length
Males
Continuous: -0.5 (-3.4, 2.3)
p-trend by quartiles = 0.74
Females
Continuous: -1.0 (-4.2, 2.1)
p-trend by quartiles = 0.83
D-ll
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Quartile analysis did not show any
statistically significant associations
Results: Lowest quartile used as reference.
Confounding: Age at examination, weight-for-age z-score, pre-pregnancy BMI, parity, smoking.
Luo et al.
China Cohort Mother-newborn Maternal blood BW (g), BL Regression
BW
(2021)
2017-2019 pairs and cord blood (cm), coefficient per
-93.34 (-157.92, -28.75), p-value
High for BW
N = 224 within three ponderal ln-unit increase
<0.05
Medium for
days of delivery index (kg/m3) in PFOS
birth length and
BL
ponderal index
5.01 (3.32, 7.62)
-0.05 (-0.38, 0.28)
Ponderal index
-0.67 (-1.08, -0.26), p-value <
0.05
Confounding: Maternal age, pre-pregnancy BMI, education, parity, environmental tobacco smoke exposure, alcohol drinking, GA, newborn
sex.
Manzano- Spain, 2003- Cohort Mother (aged
Salgadoetal. 2008 >16 yr)-child pairs
(2017a) from INMA
High N = 1,202
Maternal plasma BL (cm), BW
Early pregnancy (g), GA
Mean= 6.05
(SD = 2.74)
(weeks), HC
(cm), LBW,
LBW at term,
PTB, SGA
Regression BL: 0.03 (-0.12, 0.17)
coefficient per p-value for sex interaction = 0.98
doubling of
PFOSorby BW: 0.44 (-32.48, 33.36)
quartiles p-value for sex interaction = 0.75
LBW, LBW at GA: -0.06 (-0.19, 0.06)
term, PTB, Q2:-0.09 (-0.33, 0.16)
SGA: OR per Q3: -0.02 (-0.26, 0.23)
log2-unit Q4: -0.31 (-0.55, -0.06); p-value
increase in <0.05
PFOS p-value for sex interaction = 0.38
HC: 0 (-0.1, 0.1)
p-value for sex interaction = 0.53
LBW: 1.06 (-0.71, 1.58)
Females: 0.73 (0.46, 1.19)
Males: 1.90 (0.98, 3.68)
D-12
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
p-value for sex interaction = 0.01
LBW at term: 0.91 (0.55, 1.50)
p-value for sex interaction = 0.15
PTB: 1.10(0.70, 1.74)
p-value for sex interaction = 0.35
SGA: 0.92 (0.70, 1.22)
p-value for sex interaction = 0.57
BL, BW, HC: No statistically
significant associations by quartiles
All outcomes: No statistically
significant associations by sex
INMA = INfancia y Medio Ambiente [Environment and Childhood Project]
Outcome: SGA defined as newborns weighing below the 10th percentile for GA and sex according to national references.
Results: Lowest quartile used as reference.
Confounding: Maternal age, parity, pre-pregnancy BMI, fish intake during pregnancy, type of delivery.
Minatoya et al. Japan Cohort Pregnant women Maternal serum
(2017) 2002-2005 and their children
High from the Sapporo 5.1(3.7-6.7)
Cohort (Hokkaido Female mean:
Study on 5.04 (SD = 2.33)
Environment and Male mean: 5.85
Children's Health) (SD = 2.63)
N = 168 (90 girls,
78 boys)
BW (g), Regression BW
ponderal coefficient per -29 (-289, 232); p-value = 0.828
index (kg/m2) loglO-unit Females: -251 (-645, 143)
increase in Males: 190 (-162, 543)
PFOS and LSM p-value for sex interaction = 0.201
by tertiles LSM T1: 3196 (3095, 3298)
LSMT2: 3076 (2976, 3176)
LSM T3: 3158 (3057, 3258)
p-trend = 0.424
Ponderal index
-2.25 (-4.01,-0.50);
p-value = 0.012
Females: -2.11 (-4.86, 0.64)
Males: -2.46 (-4.74, -0.18)
p-value for sex interaction = 0.658
D-13
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
LSMT1: 28.39 (27.71,29.06)
LSM T2: 26.68 (26.02, 27.34)
LSM T3: 27.23 (26.57, 27.90)
p-trend = 0.003
Confounding: Maternal BMI, parity, smoking during pregnancy, blood sampling period, GA.
Rokoff et al.
(2018)
High
United States
1999-2002
Case-control
Pregnant women
and their children
from Project Viva
N = 1,597
Maternal plasma BW-for-GA
z-score
Mean= 29.1
(SD = 16.5)
Regression
coefficient per
IQR increase in
PFOS
-0.03 (-0.07, 0.02)
Confounding: Maternal age, race/ethnicity, education, pre-pregnancy BMI, and parity, black carbon, prenatal smoking.
Sagiv et al. United States, Cohort Pregnant women
(2018) 1999-2002 and infants from
High Project Viva
N = 1,644
Maternal blood
Early pregnancy
25.7
(IQR = 16.0)
BW-for-GA
(z-score),
gestational
length
(weeks), PTB
Regression
coefficient per
IQR increase in
PFOS and by
quartiles
PTB:
OR per IQR
increase in
PFOS and by
quartiles
BW-for-GA
-0.04 (-0.08,0.01)
Q2: -0.09 (-0.22, 0.04)
Q3: -0.09 (-0.22,0.04)
Q4:-0.13 (-0.26,0.00)
No statistically significant
associations or interactions by sex
Gestational length
-0.08 (-0.17,0.02)
Q2: -0.20 (-0.47, 0.06)
Q3: -0.08 (-0.35,0.19)
Q4: -0.36 (-0.64, -0.09)
Females: 0.01 (-0.11,0.14)
Males: -0.19 (-0.33,-0.05)
p-value for sex interaction = 0.09
PTB
1.1 (1.0,
1.3)
Q2
Q3
Q4
2.0(1.1,3.7)
2.0(1.1,3.7)
2.4(1.3,4.4)
Outcome: PTB was defined as <37 wk
Results: Lowest quartile used as reference.
D-14
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Confounding: Maternal age at enrollment, race/ethnicity, education, prenatal smoking, parity, history of breastfeeding, pre-pregnancy BMI,
paternal education, household income, child's sex, GA at blood draw.
Shoaff et al. United States, Cohort Pregnant women
(2018) 2003-2006; (aged >18 yr) and
High follow-up 4 wk their children at
to 2 yr from birth, 4 wk and
recruitment 2 yr from the
HOME study
N = 345
Maternal blood
Later pregnancy
14 (9.6-18)
BW (z-score), Regression
length-for-age coefficient by
(z-score),
rapid weight
gain, weight-
for-age (z-
score),
weight-for-
length (z-
score)
tertile (per
doubling in
PFOS)
Rapid weight
gain: RR by
tertile
BW z-score
T2: -0.05 (-0.29,0.19)
T3: -0.12 (-0.36, 0.13)
p-value for trend = 0.36
Length-for-age z-score
T2: 0.05 (-0.33, 0.44)
T3: -0.24 (-0.64,0.15)
p-value for trend = 0.08
Weight-for-age z-score
T2:0.01 (-0.31,0.32)
T3:-0.33 (-0.65,-0.01)
p-value for trend = 0.07
Weight-for-length z-score
T2:-0.16 (-0.41, 0.09)
T3:-0.31 (-0.56,-0.05)
p-value for trend = 0.66
Rapid weight gain
T2: 0.79 (0.55, 1.14)
T3: 1.11 (0.81, 1.53)
HOME = Health Outcomes and Measures of the Environment
Outcome: Rapid weight gain defined as increase in weight z-score > 0.67 SDs any time between age 4 wk and 2 yr.
Results: Lowest tertile used as reference
Confounding: Maternal age at delivery, race, marital status, insurance, income, education, parity, serum cotinine, depressive symptoms, mid-
pregnancy BMI, food security, fruit/vegetable and fish consumption during pregnancy, prenatal vitamin use.
Starling et al.
United States, Cohort Pregnant women
Maternal serum
Adiposity (% Regression
Adiposity: 0.08 (-0.33, 0.49)
(2017)
2009-2014 (aged >16 yr) and
fat mass), BW coefficient per
T2: 0.26 (-0.46, 0.98)
High
infants from
2.4 (1.5-3.7)
(g) ln-unit increase
T3: -0.41 (-1.15,0.33)
Healthy Start at
in PFOS and by
birth
tertiles
BW:-13.8 (-102.8, 35.2)
N = 628
T2: -33.8 (-102.8, 35.2)
D-15
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Starling et al.
(2019)
High
T3: -71.1 (-142.6,0.5)
Results: Lowest tertile used as reference.
Confounding: Maternal age, pre-pregnancy BMI, race/ethnicity, education, gestational weight gain, smoking during pregnancy, gravidity, GA
at blood draw, infant sex, and GA at birth.
United States,
2009-2014
Cohort
Pregnant women
Maternal serum
Adiposity
Regression
(aged >16 yr) and
(%), weight-
coefficient per
infants from
2.2 (1.4-3.4)
for-age z-
ln-unit increase
Healthy Start
score (WAZ),
in PFOS and by
assessed up to
weight-for-
tertiles
5 mo
length z-score
N = 415 (202 girls,
(WLZ), WAZ
Rapid growth:
213 boys)
and WLZ
OR per ln-unit
growth from
increase in
birth to 5 mo,
PFOS
rapid growth
in WAZ or
WLZ
Adiposity at 5 mo
-0.13 (-0.83,0.57)
Females: -0.91 (-1.84,0.02)
Female T3: -2.08 (-3.81, -0.35)
Males: 0.73 (-0.36, 1.81)
Male T2: 1.85 (0.14, 3.47)
p-value for sex interaction = 0.05
WAZ at 5 mo: -0.10 (-0.23, 0.02)
T3: -0.28 (-0.51,-0.05)
Females: -0.26 (-0.43, -0.10)
Female T3: -0.56 (-0.87, -0.26)
Males: 0.07 (-0.13,0.27)
p-value for sex interaction = 0.10
WLZ at 5 mo: -0.08 (-0.23, 0.06)
Females: -0.08 (-0.23, 0.06)
Female T3: -0.52 (-0.88, -0.17)
Males: 0.06 (-0.17,0.28)
p-value for sex interaction = 0.17
WAZ or WLZ growth from birth to
5 mo, rapid growth: No statistically
significant associations
Outcome: Rapid growth defined as change in WAZ or WLZ >0.67 between birth and 5 mo
Results: Lowest tertile used as reference
Confounding: Maternal age, race/ethnicity, pre-pregnancy BMI, any previous pregnancies, any smoking during pregnancy, education,
gestational weight gain z-score, infant sex, exclusive breastfeeding to follow-up visit, infant age (days) at follow-up.
Valvi et al.
Faroe Islands Cross-sectional Pregnant women Maternal serum
HC (cm),
Regression
HC
(2017)
1997-2000 and their children
body length
coefficient per
0 (-0.28, 0.27)
High
N = 604 (288 girls, 27.2 (23.1-33.1)
(cm), BW
doubling of
Girls: 0.48 (0.05, 0.90)
316 boys)
(g)
PFOS
Boys: -0.28 (-0.65, 0.09)
D-16
-------
APRIL 2024
Reference, Location, Population,Ages, MaS°S«mple Outcome Comparison Rente-
„ j.. . Design Matrix, Sample
Confidence Years N Timing, Levels'
p-value for sex interaction = 0.01
Body length
0.05 (-0.33, 0.43)
Girls: 0.32 (-0.24, 0.89)
Boys: -0.18 (-0.60, 0.23)
p-value for sex interaction = 0.17
BW
-81 (-173, 11)
Girls: 5 (-124, 135)
Boys: -150 (-282, -17)
p-value for sex interaction = 0.08
Confounding: Maternal age at delivery, education, parity, pre-pregnancy BMI, smoking during pregnancy, child sex.
Whitworth et al.
Norway
Cohort
Pregnant women
Maternal plasma
PTB, BW (z-
OR by quartile
PTB
(2012a)
2003-2004
and their children
around 17 wk of
score),
Q2
0.9 (0.3, 2.8)
High
from MoBa
gestation
SGA, LGA
BW:
Q3
0.9 (0.3, 2.7)
13.0(10.3-16.6)
Regression
Q4
0.3 (0.1, 1.0)
PTB, LGA, SGA
coefficient per
p-trend = 0.03
N = 901
unit increase in
BW
PFOS, or by
LGA
N = 838
quartile
Q2
0.8 (0.5, 1.6)
Q3
1.0(0.5, 1.7)
Q4
0.7 (0.3, 1.4)
p-trend = 0.33
SGA
Q2: 1.2(0.5,3.0)
Q3: 2.2 (1.0, 5.1)
Q4: 1.3 (0.5, 3.4)
p-trend = 0.51
BW
Per increase: -0.01 (-0.02, 0.01)
Q2: -0.08 (-0.29,0.13)
Q3:-0.17 (-0.39, 0.05)
D-17
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Q4:-0.18 (0.41,
p-trend = 0.12
0.05)
MoBa = Norwegian Mother and Child Cohort Study
Outcome: PTB defined as GA <37 wk. SGA defined as gender- and gestation age-specific BW less than the 10th percentile. LGA defined as
gender- and GA-specific BW greater than the 90th percentile.
Confounding: Maternal age, pre-pregnancy BMI, parity. Additional confounding for BW: albumin concentration, maternal education,
interpregnancy interval, quadratic interpregnancy interval, consumption of lean fish.
Wikstrom et al. Sweden Cohort Infants exposed Maternal serum
(2020) 2007-2010 prenatally to PFAS Early pregnancy
High from the SELMA 5.38 (3.97-7.60)
study
N = 1533 (732
girls, 801 boys)
BW (g), BW-
SDS, SGA
Regression
coefficient (BW,
BW-SDS) and
OR (SGA) per
ln-unit increase
in PFOS or by
quartiles
BW
Per increase: -46 (-88, -3)
Q2: -27 (-89, 35)
Q3: -22 (-84,41)
Q4: -80 (-144, -16)
Girls
Per increase: -85 (-145, -25)
Q2: -32 (-115, 52)
Q3: -51 (-137, 34)
Q4: -142 (-231,-54)
Boys
Per increase: -13 (-73, 47)
Q2
Q3
Q4
-28 (-118, 63)
5 (-86, 96)
-28 (-119, 63)
BW-SDS
Per increase: -0.100 (-0.197,
-0.004)
Q2:-0.045-0.185, 0.096)
Q3:-0.024 (-0.166,0.118)
Q4:-0.172 (-0.317,-0.027)
Girls
Per increase: -0.167 (-0.301,
-0.034)
Q2:-0.044 (-0.232,0.143)
Q3:-0.092 (-0.283,0.100)
Q4: -0.296 (-0.494, -0.098)
Boys
D-18
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Per increase: -0.027 (-0.166,
0.112)
Q2
Q3
Q4
-0.055 (-0.263,0.153)
0.038 (-0.171, 0.246)
-0.066 (-0.276,0.144)
SELMA = Swedish Environmental Longitudinal Mother and Child, Asthma and Allergy
Outcomes: SGA defined as BW below the 10th percentile for GA and sex.
Results: Lowest quartile used as reference.
Confounding: Sex, GA, maternal weight, parity, cotinine levels.
SGA
Per increase: 1.19(0.87, 1.64)
Q2: 0.69 (0.43, 1.08)
Q3: 0.79 (0.53, 1.18)
Q4: 1.56 (1.09,2.22)
Girls
Per increase: 1.40 (0.83, 2.35)
Q2: 0.89 (0.39,2.03)
Q3: 0.82 (0.36,2.03)
Q4: 2.05 (1.00,4.21)
Boys
Per increase: 1.08 (0.72, 1.63)
Q2: 1.26 (0.67,2.37)
Q3: 0.86 (0.45, 1.67)
Q4: 1.30 (0.7,2.4)
Wikstrom et al. Sweden, 2007- Nested case- Pregnant women Serum during Miscarriage OR per doubling Per doubling: 1.13 (0.82, 1.52)
(2021)
High
2010
control from the SELMA
study
N = 1,527
first trimester
Case: 6.09
(3.99-8.77)
Control: 5.45
(4.00-7.68)
in PFOS
SELMA = Swedish Environmental Longitudinal Mother and Child, Asthma and Allergy
Confounding: Parity, age, cotinine (tobacco smoke) exposure.
Xiao et al. Denmark Cohort Pregnant women Maternal blood Z-scores for Regression
(2019) 1994-1995 and their children Later pregnancy BL, BW, and coefficient per
High N=171 log2-unit
BL z-score
-0.33 (-0.69, 0.03)
Girls: -0.23 (-0.75, 0.30)
D-19
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
GM = 20.8 (ig/g
(range: 6.9-
47.6 jig/g)
cranial
circumference
increase in
PFOS
Boys: -0.41 (-0.87, 0.05)
BW z-score
-0.47 (-0.85, -0.09)
Girls:-0.56 (-1.12,0.00)
Boys: -0.40 (-0.89, 0.08)
Cranial circumference z-score
-0.26 (-0.68,0.16)
Girls: -0.42 (-1.05, 0.21)
Boys: -0.15 (-0.68, 0.39)
Confounding: Child sex, parity, maternal BMI, maternal height, maternal education, maternal age, smoking and drinking alcohol during
pregnancy, total PCB, mercury.
Yao et al.
(2021)
High
China
2010-2013
Cross-sectional
Parents and their
children from
LWBC
N = 369
Maternal and
paternal serum
within three
days of birth
Maternal: 4.55
(Range = 0.55-
29.85)
Paternal: 10.15
(
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
length
(WFL) z-
score
S: -0.08 (-0.12, -0.04); p-value <
0.05
Q2: 0.03 (-0.09,0.15)
Q3: -0.06 (-0.18,0.06)
Q4: -0.20 (-0.32, -0.09); p-value
<0.05
S-girls: -0.11 (-0.17, -0.05);
p-value < 0.05
Q2: 0.07 (-0.10,0.24)
Q3: 0.03 (-0.16,0.17)
Q4: -0.26 (-0.26, -0.10); p-value
<0.05
S-boys: -0.05 (-0.11,0.01)
Q2:-0.01 (-0.16, 0.15)
Q3:-0.11 (-0.27,0.06)
Q4:-0.15 (-0.32,0.02)
T: -0.03 (-0.10,0.05)
Q2: 0.11 (-0.09,0.32)
Q3: 0.07 (-0.14,0.28)
Q4: 0.0005 (-0.2, 0.2)
Length
S: 0.07 (-0.06,0.19)
S-girls: 0.03 (-0.14, 0.20)
S-boys: 0.10 (-0.07, 0.27)
T: 0.18 (-0.07, 0.42)
Length z-score
S: 0.03 (-0.03, 0.08)
S-girls: 0.008 (-0.07, 0.08)
S-boys: 0.05 (-0.03,0.12)
T: 0.07 (-0.04,0.18)
Weight
S: -21.99 (-59.52, 15.55)
D-21
-------
APRIL 2024
Reference, ^cation, Population,Ages, MaS°Sam|,le Outcome Comparison Rl,„l,s"
Confidence Years N Timing, Levils-
S-girls: -51.57 (-102.32, -0.82);
p-value < 0.05
S-boys: 6.15 (-48.31, 60.61)
T: 62.47 (-13.97, 138.92)
Weight z-score
S: -0.03 (-0.08, 0.01)
S-girls: -0.07 (-0.13,-0.01);
p-value < 0.05
S-boys: -0.001 (-0.06, 0.06)
T: 0.04 (-0.04,0.12)
WFL z-score
S: -0.08 (-0.12,-0.04)
S-girls: -0.10 (-0.16, -0.05);
p-value < 0.05
S-boys:-0.05 (-0.11,0.01)
T: -0.03 (-0.11,0.05)
Rapid weight gain, obesity: not
statistically significant for all
children
Outcome: Rapid weight gain defined as the child's weight gain SD above 0.5 for 4 or 9 mo or about 0.67 for 12 mo.
Comparison: Logarithm base not specified.
Results: Lowest quartile used as reference.
Confounding: Child's age at measurement, age squared, age cubed, sex-age interactions, maternal age, pre-pregnancy BMI category, maternal
education, maternal race, private insurance, infertility treatment.
Andersen et al.
Denmark,
Cohort
Pregnant women
Maternal plasma
BW (g, z- Regression
BW
(2010)
1996-2002
and their children
from first and
score), BMI at coefficient per
z-score: -0.002 (-0.006, 0.002)
Medium
followed up at
second trimester
5 and 12 mo, unit increase in
g: "1 ("3.1, 1.0)
birth, 5 mo, and
33.4 (6.4, 106.7)
height at 5 PFOS
Boys
12 mo from
and 12 mo
z-score: 0.003 (-0.003, 0.008)
DNBC
(cm), weight
g: 1.3 (-1.6, 4.2)
at 5 and
Girls
12 mo (g)
z-score: -0.006 (-0.011, -0.001),
p-value <0.05
D-22
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
N at birth = 1,114
(552 boys, 562
girls)
g: -3.2 (-6.0, -0.3), p-value <0.05
BMI at 5 mo
z-score: -0.001 (-0.006, 0.003)
g: -0.002 (-0.10,0.005)
Boys
z-score: -0.004 (-0.011, 0.002)
g: -0.007 (-0.018, 0.003)
Girls
z-score: 0.001 (-0.005, 0.007)
g: 0.002 (-0.008, 0.012)
BMI at 12 mo
z-score: -0.007 (-0.011, 0.002),
p-value <0.05
g: -0.011 (-0.019,-0.003)
Boys
z-score: -0.01 (-0.017, -0.003),
p-value <0.01
g: -0.017 (-0.028, -0.005),
p-value <0.01
Girls
z-score: -0.005 (-0.011, 0.002)
g: -0.007 (-0.018, 0.003)
Height at 5 mo
z-score: 0.002 (-0.002, 0.006)
g: 0.006 (-0.004, 0.017)
Boys
z-score: 0.0004 (-0.006, 0.006)
g: 0.001 (0.014, 0.016)
Girls
z-score: 0.004 (-0.001, 0.010)
g: 0.011 (-0.004, 0.026)
Height at 12 mo
D-23
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
z-score: 0.003 (-0.001, 0.008)
g: 0.010 (-0.003, 0.023)
Boys
z-score: 0.003 (-0.004, 0.009)
g: 0.008 (-0.011,0.027)
Girls
z-score: 0.004 (-0.002, 0.010)
g: 0.011 (-0.007, 0.030)
Weight at 5 mo
z-score: -0.001 (-0.005, 0.003)
g: -0.8 (-4.2, 2.6)
Boys
z-score: -0.004 (-0.009, 0.001)
g: -3.7 (-8.7, 1.3)
Girls
z-score: 0.002 (-0.004, 0.007)
g: 1.3 (-3.3, 5.9)
Weight at 12 mo
z-score:-0.005 (-0.009, 0.001),
p-value <0.05
g: -5.8 (-10.4, -1.2), p-value
<0.05
Boys
z-score: -0.008 (-0.013, -0.002),
p-value <0.05
g: -9 (-15.9, -2.2), p-value <0.05
Girls
z-score: -0.003 (-0.009, 0.003)
g: -3.3 (-9.3,2.7)
DNBC = Danish National Birth Cohort
Results: "Models for weight at 5 or 12 mo included BW, models for length at 5 or 12 mo included birth length, and models for BMI at 5 or
12 mo included birth BMI."; adjusted models were used for all results.
D-24
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Confounding: Maternal age, parity, pre-pregnancy BMI, smoking, socioeconomic status, GA at blood drawing, breastfeeding. Additional
confounding for BMI and 5 and 12 mo: birth BMI. Additional confounding height at 5 and 12 mo: birth height. Additional confounding for
weight at 5 and 12 mo: BW.
Apelberg et al. United States
(2007b) 2004-2005
Medium
Cross-sectional Pregnant women Cord blood at
and their newborns birth
from Baltimore 5(3.4-7.9)
THREE Study,
N = 293
BW (g), HC Regression BW
(cm), BL coefficient per Per ln-unit increase: -69 (-149,
(cm), ln-unit increase 10)
ponderal in PFOS, Per IQR increase: -58 (-125, 9)
index regression
(g/cm3 x 100), coefficient per HC
GA (days) IQR increase in Per ln-unit increase: -0.32 (-0.56,
PFOS -0.07), p-value <0.05
Per IQR increase: -0.27 (-0.48,
-0.06), p-value <0.05
BL
Per ln-unit increase: 0.13 (-0.26,
0.52)
Per IQR increase: 0.11 (-0.22,
0.44)
Ponderal index
Per ln-unit increase: -0.074
(-0.123, -0.025), p-value <0.05
Per IQR increase: -0.062 (-0.104,
-0.021), p-value <0.05
GA
Per ln-unit increase: 1.9 (-1.3, 5)
Per IQR increase: 1.0 (-0.7, 2.8)
Confounding: GA, maternal age, BMI, race, parity, smoking, baby sex, height, net weight gain, diabetes, hypertension. Additional
confounding for HC: delivery mode.
Arbuckle et al.
Canada, 2008- Cohort
Pregnant women
Maternal blood
Anoclitoris
Regression
ACD: 0.07 (-1.03, 1.18)
(2020)
2011
(age range =17-
distance
coefficient per
Q2: -0.06 (-1.7, 1.58)
Medium
42 yr) and their
4.50 (ig/L (3.30-
(ACD, mm),
ln-unit increase
Q3: 0.17 (-1.5, 1.85)
infants from
6.10 (ig/L)
anofourchette
in PFOS and by
Q4: -0.05 (-1.68, 1.57)
MIREC
distance
quartiles
D-25
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
N = 205
(AFD, mm),
anopenile
distance
(APD, mm),
anoscrotal
distance
(ASD, mm)
AFD: -0.29 (-1.62, 1.04)
Q2
Q3
Q4
-0.12 (-2.09, 1.85)
0.89 (-1.12,2.9)
-0.33 (-2.31, 1.65)
APD: 0.13 (-1.13, 1.38)
Q2
Q3
Q4
-0.97 (-2.81, 0.87)
-1.28 (-3.22, 0.66)
0.22 (-1.68,2.13)
ASD: 1.05 (-0.24, 2.35)
Q2
Q3
Q4
-0.87 (-2.78, 1.04)
0.33 (-1.67,2.33)
0.49 (-1.47, 2.46)
No statistically significant trends
MIREC = Maternal-Infant Research on Environmental Chemicals (MIREC)
Results: Lowest quartile used as reference.
Confounding: Household income, education, active smoking status, GA, WLZ, and recruitment site.
Chang et al.
(2022)
Medium
United States
2014-2018
Cohort
Mother-infant Maternal serum, BW (g), SGA BW: Regression BW
pairs from the Early
Emory University pregnancy,
African American 2.19 (1.45-3.24)
Vaginal, Oral, and
Gut Microbiome
in Pregnancy
Study
N = 370
coefficient per
doubling in Q2
PFOS and by Q3
quartiles Q4
SGA: Odds ratio
per doubling in
PFOS and by
quartiles
Per doubling: -7 (-48, 34)
78 (-98, 196)
20 (-98, 138)
-16 (-136, 105)
p-trend = 0.48
SGA
Per doubling: 1.12 (0>
Q2: 0.92 (0.47, 1.78)
Q3: 1.32 (0.69,2.53)
Q4: 1.09 (0.56,2.13)
p-trend = 0.65
Chen et al. Taiwan, 2004- Cross-sectional Mother-infant Cord blood at BW (g), BL
(2012a) 2005 pairs from TBPS birth (cm), GA
BW: Regression BW
coefficient per
1.42)
Outcome: SGA defined as a BW below the 10th percentile for GA.
Confounding: maternal age, education, BMI, parity, tobacco use, marijuana use, and infant's sex (BW only).
D-26
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Medium
N = 429
GM (SD) = 5.94
(1.95)
(weeks), HC
(cm), LBW,
ponderal
index (g/cm3),
PTB, SGA
unit increase in
PFOS
BW, BL, GA,
HC, ponderal
index:
Regression
coefficient per
ln-unit increase
in PFOS, or by
quartile
PTB, LBW,
SGA: OR per ln-
unit increase in
PFOS, or by
quartile
Per ln-unit increase: -110.2 (-176,
-44.5), p-value <0.01
Per unit increase: -11.3 (-17.4,
-5.2)
Q2: 54 (-44, 152)
Q3: 2 (-95, 102)
Q4: -92 (-190, 6)
p-trend = 0.045
BL
Per ln-unit increase: -0.17 (-0.42,
0.09)
Q2
Q3
Q4
0.08 (-0.39, 0.55)
0.14 (-0.33,0.62)
-0.32 (-0.80, 0.15)
p-trend = 0.234
GA
Per ln-unit increase: -0.37 (-0.6,
-0.13), p-value <0.01
Q2: 0.13 (-0.30,0.57)
Q3: -0.65 (-1.07, -0.20)
Q4: -0.44 (-0.88, 0.00)
p-trend = 0.004
HC
Per ln-unit increase: -0.25 (-0.46,
-0.05), p-value <0.05
Q2:-0.16 (-0.53, 0.21)
Q3: -0.26 (-0.63,0.12)
Q4: -0.42 (-0.80, -0.05)
p-trend = 0.025
Ponderal index
Per ln-unit increase: -0.01 (-0.05,
0.02)
D-27
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Q2
Q3
Q4
0.03 (-0.03, 0.09)
-0.02 (-0.08, 0.04)
-0.03 (-0.09, 0.04)
p-trend = 0.232
PTB
Per ln-unit increase: 2.45 (1.47,
4.08), p-value <0.001
Q2: 1.0 (0.2, 5.0)
Q3: 6.5 (2.0, 24)
Q4: 5.5 (1.5,20)
p-trend = 0.0006
LBW
Per ln-unit increase: 2.61 (0.85,
8.03)
Q2: 0.5 (0.02, 13)
Q3: 1.0(0.06, 18)
Q4: 4.5 (0.50, 57)
p-trend = 0.062
SGA
Per ln-unit increase: 2.27 (1.25,
4.15), p-value <0.01
Q2: 0.8 (0.2, 2.5)
Q3: 0.5 (0.1, 2.0)
Q4: 1.5 (0.6,4.5)
p-trend = 0.422
TBPS = Taiwan Birth Panel Study
Outcome: PTB defined as GA <37 wk. Low BW defined as a BW <2,500 g. SGA defined as a BW below the 10th percentile for GA.
Confounding: Maternal age, pre-pregnancy BMI, education level, ln-transformed cord blood cotinine levels, type of delivery, parity, infant
sex. Additional confounding for BW, BL, HC, ponderal index, low BW, PTB: GA.
Chen et al.
Taiwan, 2004- Cohort
Mother-infant
Cord blood
BMI (z-score, Regression
BMI
(2017c)
2005
pairs from the
kg/m2), height coefficient per
Birth: -0.11 (-0.25,0.02)
Medium
Taiwan Birth
5.7 (IQR =5.0)
(z-score, cm), ln-unit increase
0-6 mo: 0.002 (-0.17, 0.18)
in PFOS
6-12 mo:-0.12 (-0.31, 0.08)
D-28
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
Panel Study
(TBPS)
N = 429
weight (z-
score, kg)
Girls 6-12 mo: -0.33 (-0.59,
-0.08); p-value < 0.05
12-24 mo: -0.09 (-0.29, 0.11)
Girls 12-24 mo: -0.25 (-0.45,
-0.05); p-value < 0.05
24-60 mo: -0.17 (-0.41, 0.06)
60-108 mo: -0.02 (-0.33, 0.28)
Girls 60-108 mo: 0.34 (0.007,
0.68); p-value < 0.05
Height
Birth: -0.16 (-0.31,-0.02),
p-value < 0.05
0-6 mo:-0.04 (-0.23,0.16)
6-12 mo:-0.02 (-0.23,0.18)
12-24 mo: 0.04 (-0.17, 0.26)
24-60 mo: 0.09 (-0.12, 0.3)
Boys 24-60 mo: 0.18 (0.03, 0.33);
/rvalue < 0.05
60-108 mo: 0.06 (-0.19, 0.31)
Boys 60-80 mo: 0.19 (0.01, 0.38);
p-value < 0.05
Weight
Birth: -0.14 (-0.26,-0.01),
p-value < 0.05
0-6 mo:-0.008 (-0.17,0.16)
6-12 mo: -0.13 (-0.32,0.07)
Girls 6-12 mo: -0.25 (-0.47,
-0.04); p-value < 0.05
12-24 mo: -0.005 (-0.25, 0.16)
Girls 12-24 mo: -0.24 (-0.41,
-0.06); p-value < 0.01
24-60 mo: -0.07 (-0.3, 0.16)
60-108 mo: 0.02 (-0.27, 0.31)
D-29
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Chen et al.
(2021)
Medium
BMI, height, and weight: no
statistically significant interactions
by sex at any age
Population: Infants were followed up at 4, 6, 13, 24, 60, 84, and 108 mo
Confounding: Maternal age, pre-pregnancy BMI, education level, ln-cord blood cotinine, infant sex, PTB, postnatal ETS exposure,
breastfeeding.
China
Recruitment:
2013-2015
Cohort
Mother-child pairs Maternal plasma BW (g), BL Regression
BW
from the SBC,
Ages > 20,
N = 214
95 male children,
119 female
children
from the first (cm), HC coefficient per 2.7 (-84.3, 89.7)
trimester (cm) ln-unit increase
9.70(6.75- inPFOS BL
15.35) -0.27 (-0.51, -0.02), p-value
<0.05
Males
-0.14 (-0.55, 0.26)
Females
-0.4 (-0.74, -0.06), p-value <0.05
HC
-10.6 (-60.7, 39.6)
SBC = Shanghai Birth Cohort
Confounding: Maternal age, BMI, educational level, occupation, income, fetal sex, parity, GA, smoking and alcohol.
Darrow et al. United States, Cohort Pregnant women Serum collected Primary OR per ln-unit
(2014) Recruitment: with known PFAS before analysis increase in
Medium 2005-2006, exposure (ages pregnancy miscarriage, PFOS, OR by
Follow-up: >20 yr) from 15.1(10.4-21.2) first quintile
2008-2011 C8HP pregnancy
N = 1,438 miscarriage
First pregnancy
N = 1,129
Primary analysis: 1.21 (0.94, 1.55)
Q2: 1.34 (0.84,2.16)
Q3: 1.40 (0.88,2.25)
Q4: 1.59 (0.99,2.54)
Q5: 1.41 (0.88,2.26)
First pregnancy: 1.34 (1.02, 1.76)
Q2: 1.68 (0.99,2.84)
Q3: 1.93 (1.13,3.31)
Q4: 1.94(1.14,3.31)
Q5: 1.80 (1.06,3.06)
C8HP = C8 Health Project
Outcome: Primary analysis includes more than one pregnancy for some women (304 miscarriages). First pregnancy is restricted to the first
pregnancy conceived per woman after serum measurement (213 miscarriages)
D-30
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Confounding: Maternal age, educational level, smoking status, BMI, self-reported diabetes, time between conception, serum measurement.
De Cock et al.
(2014)
Medium
The Netherlands Cohort
Recruitment:
2011-2013
Follow-up at 1,
2, 4, 6, 9, and
11 mo after
birth
Mother-child pairs
N = 89
Cord blood
1,600.0 ng/L
(Range = 570-
3,200 ng/L)
BMI (kg/m2),
HC (cm),
height (cm),
weight (kg)
Regression
coefficient for
quartiles of
PFOS
BMI, HC, height, and weight: no
statistically significant associations
Confounding: BW, GA, maternal height.
de Cock et al.
(2016)
Medium
The Cross-sectional Mother-infant
Netherlands, pairs
2011-2013 N = 64
Cord blood
1,600 ng/L
(Range = 570-
3,200 ng/L)
BW (g)
Regression
coefficient by
tertiles
T2: 254.8 (-99.47, 609.09),
p-value = 0.153
T3: 438.4 (55.09, 821.68),
p-value = 0.026
Females
T2: 143.3 (-361.63, 648.32),
p-value = 0.566
T3: 301.1 (-124.87, 727.05),
p-value = 0.159
Males
T2: 486.9 (-1.21, 975.03),
p-value = 0.051
T3: 724.4 (193.83, 1,254.97),
p-value = 0.009
Results: Lowest tertile used as reference.
Confounding: GA, maternal BMI, maternal height, maternal age at birth, and parity, paternal BMI, paternal height, education, fish intake.
Fei et al.
Denmark Cohort
Pregnant women
Maternal plasma
Gross motor
Gross motor Gross motor milestone
(2008a)
Recruitment:
and their children
during the first
milestone,
milestone: Q2: 0.93 (0.79, 1.08)
Medium
1996-2002,
at 6 and 18 mo
trimester
language
Hazard ratio by Q3: 0.85 (0.72, 0.99)
Assessment 6-
from the DNBC
33.3 (26.0-43.2)
milestone,
quartile Q4: 0.86 (0.73, 1.01)
18 mo later
Apgar score
Language p-trend = 0.041
Total
<10
milestone: OR
N = 1,400
by quartile Language milestone
18-mo olds
Apgar score: OR Q2: 1.39 (0.46, 4.25)
N = 1,380
for Q4 vs. Q1 Q3: 1.58 (0.51, 4.91)
Q4: 2.93 (1.00, 8.56)
D-31
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
p-trend = 0.039
Apgar score
Q4: 1.2(0.67,2.14)
DNBC = Danish National Birth Cohort
Outcome: Gross motor milestone defined as sitting without support. Language milestone defined as children not using word-like sounds to
indicate what they want.
Results: Lowest quartile used as reference group.
Confounding: Maternal age, maternal occupational and educational status, parity, pre-pregnancy BMI, smoking and alcohol consumption
during pregnancy, gestational weeks at blood drawing, child's sex. Additional confounding for gross motor milestone and language milestone:
parity, out-of-home childcare, home density (rooms/person). Additional confounding for language milestone: child's age at interview.
Fei et al. Denmark Cohort Pregnant women Maternal plasma
(2008b) 1996-2002 and their newborns between 4 and
Medium from the DNBC 14 wk gestation
33.4 (26.1-43.3)
Placental weight
N = 1,337
Birth length
N = 1,376
HC
N = 1,347
Abdominal
circumference
N = 1,325
Placental
weight (g),
HC (cm), BL
(cm),
abdominal
circumference Mean difference
Regression
coefficient per
unit increase in
PFOS
(cm)
by quartile
Placental weight
Per unit increase: -0.24 (-0.85,
0.37)
-6.6 (-28.8, 15.5)
-13.7 (-36.4, 8.9)
-10.8 (-33.4, 11.8)
Q2
Q3
Q4
HC
Per unit increase: 0.0 (-0.006,
0.007)
Q2
Q3
Q4
0.14 (-0.09,0.36)
0.09 (-0.14,0.32)
0.03 (-0.20, 0.27)
BL
Per unit increase: -0.002 (-0.011,
0.006)
Q2: 0.21 (-0.08,0.51)
Q3: 0.06 (-0.24,0.36)
Q4: 0.05 (-0.25, 0.35)
Abdominal circumference
Per unit increase: -0.003 (-0.012,
0.005)
Q2: 0.24 (-0.07, 0.55)
D-32
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Q3: 0.10 (-0.21,0.42)
Q4: 0.00 (-0.32, 0.32)
DNBC = Danish National Birth Cohort
Results: Lowest quartile used as reference group
Confounding: GA, quadratic GA, infant sex, maternal age, socio-occupational status, parity, cigarette smoking, pre-pregnancy BMI,
gestational week at blood drawing.
Govarts et al. Belgium, the Cohort Mother-child pairs Cord blood
(2018) Netherlands, from FLEHSI and
Medium Norway, and II, HUMIS, LINC, 1,984 ng/L
Slovakia and PCB Cohort (1,200-
2002-2012 N = 657 3,008 ng/L)
SGA
OR per IQR
increase in
PFOS
0.823 (0.742, 0.913)
FLEHS = Flemish Environmental and Health Study; HUMIS = Human Milk Study; LINC = Linking EDCs in Maternal Nutrition to Child
Health
Outcome: SGA defined as newborns weighing below the 10th percentile for the norms defined by GA, country, and infant's sex.
Confounding: Maternal education, maternal age at delivery, maternal height, maternal pre-pregnancy BMI, smoking during pregnancy,
parity, child's sex.
Gyllenhammar
et al. (2018b)
Medium
Sweden, 1996-
2011 and
follow-up at
5 yr of age
Cohort and
cross-sectional
Mother-infant
pairs of singleton
births from
POPUP study
N = 377
Maternal serum
Later pregnancy
13 (7.4-19)
BL (SD
scores), BW
(SD scores),
gestational
length (days),
HC (SD
scores), length
(SD scores),
weight (SD
scores)
Regression BL: 0.1377 (-0.0971, 0.3725)
coefficient per BW: 0.0167 (-0.1878, 0.2225)
IQR increase in Gestational length: -2.0342
maternal PFOS (-4.1139, 0.0455)
HC: 0.0703 (-0.1602, 0.2974)
HC, length, and weight: no
statistically significant associations
by sex
POPUP = Persistent Organic Pollutants in Uppsala Primiparas
Confounding: Sampling year, maternal age, pre-pregnancy BMI, maternal weight gain during pregnancy, maternal weight loss after delivery,
years of education, smoking during pregnancy, total fish consumption.
Hamm et al.
Canada Cohort
Pregnant women
Maternal serum
BW (g, z-
BW: Regression
BW (g per ln-unit): -31.3 (-43.3,
(2010)
Recruitment:
(>18 yr of age)
collected at 15-
score), SGA,
coefficient per
105.9), p-value = 0.03
Medium
2005-2006
and their singleton
16 wk gestation
PTB, length
ln-unit or per
T2: -13.51 (-136.57, 109.55)
Follow-up at
children delivered
of gestation
unit increase in
T3: 71.25 (-54.97, 197.48)
delivery: 2006-
at or after 22 wk
GM (SD) = 7.4
(weeks)
PFOS and by
BW (g per unit): 1.5 (-7.6, 10.6)
2007
gestation
(2.0)
tertiles
D-33
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
N = 252
SGA, PTB:
Relative risk by
tertile
Length of
gestation:
Regression
coefficient per
ln-unit increase
in PFOS and by
tertile
BW (z-score per ln-unit): 0.06
(-0.11,0.23)
T2: -0.006 (-0.29, 0.27)
T3: 0.16 (-0.13, 0.44)
SGA:
T2: 0.99 (0.27, 3.61)
T3: 0.26 (0.10,0.70)
PTB:
T2: 1.06 (0.33, 3.45)
T3: 1.11 (0.36, 3.38)
Length of gestation:
Per ln-unit: 0.21 (-0.12, 0.53)
T2: 0.13 (-0.42,0.67)
T3: 0.046 (-0.51,0.60)
Outcome: SGA defined as BW <10th percentile for GA and infant gender; PTB defined as delivery at 22-36 wk
Results: Lowest tertile used as reference
Confounding: Maternal age, maternal race, gravida, maternal weight and height, smoking. Additional confounding for PTB and BW (g):
infant gender. Additional confounding for BW (g): GA at birth.
Hjermitslev et Greenland, Cohort Pregnant women
al. (2019) Recruitment: (>18 yr of age)
Medium 2010-2011, and their children
2013-2015 from ACCEPT
N = 256
Maternal serum
Early
pregnancy, later
pregnancy
8.99
(Range = 1.50-
61.3)
BW (g), GA
at birth
(weeks), HC
(cm), preterm
birth
Regression
coefficient per
ln-unit increase
in PFOS
Preterm birth:
OR per ln-unit
increase in
PFOS
BW: -5.47 (-12.6, 1.67)
Females: -5.65 (-14.9, 3.55)
Males: -1.9 (-14, 10.2)
GA: 0.001 (-0.02, 0.03)
Females: 0.002 (-0.03, 0.03)
Males: -0.006 (-0.05, 0.04)
HC: -0.01 (-0.04, 0.01)
Females: -0.02 (-0.05, 0.01)
Males: 0.005 (-0.04, 0.05)
Preterm birth: 0.95 (0.87, 1.05),
p-value = 0.321
D-34
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Jensen et al.
(2020a)
Medium
No statistically significant
associations
ACCEPT = Adapting to Climate Change, Environmental Pollution and Dietary Transition
Confounding: Maternal age, plasma cotinine, alcohol consumption during pregnancy, pre-pregnancy BMI, GA at birth.
Denmark,
2010-2012 and
follow-up at
18 mo of age
Cohort
Pregnant women Maternal serum Ponderal
Regression
and infants at 3
and 18 mo of age 8.04 (3.82-
from Odense Child 15.45)
Cohort
N = 593
index standard coefficient per
deviation unit increase in
score (SDS) PFOS
-0.004 (-0.03, 0.02)
Birth: 0.03 (0.01, 0.05),
p-value = 0.02
3 mo: -0.005 (-0.03, 0.016)
18 mo: -0.003 (-0.03, 0.02)
3 and 18 mo: no statistically
significant associations
Outcome: Ponderal index (kg/m3) was calculated as weight (kg) divided by the length cubed (m3)
Confounding: Maternal age, parity, pre-pregnancy BMI, pre-pregnancy BMI2, education, smoking, sex, visit, adiposity marker at birth.
Kashino et al. Japan, 2003- Cohort Mother-infant
(2020) 2009 pairs from the
Medium Hokkaido Study
on Environment
and Children's
Health
N = 1,949
Plasma
Later pregnancy
3.4 (2.6-4.7)
Birth HC
(cm), BL
(cm), BW (g)
Regression
coefficient per
loglO-unit
increase in
PFOS
HC: -0.067 (-0.418, 0.283)
Females: 0.001 (-0.531, 0.532)
Males: -0.142 (-0.605, 0.321)
Length: 0.092 (-0.311, 0.494)
Females: 0.25 (-0.321, 0.821)
Males: -0.019 (-0.589, 0.551)
BW: -35 (-109, 39)
Females: -19.9 (-128, 88.2)
Males: -46.3 (-148.4, 55.8)
HC, BL, and BW: no statistically
significant associations overall or
stratified by sex
Confounding: GA, maternal age, pre-pregnancy BMI, parity, infant sex, maternal educational level, plasma cotinine concentration during
pregnancy.
Kishi et al.
(2015)
Medium
Japan, 2002-
2005
Cross-section
Pregnant women
(aged 28-34 yr)
and infants from
the Hokkaido
Maternal blood BW (g)
Mean= 5.89
(SD = 0.20)
Regression
coefficient by
quartiles
Females
Q2: -70.1 (-242.5, 102.2)
Q3: -39.1 (-216.1, 137.8)
D-35
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Study on
Environment and
Children's Health
Females, N = 165
Males, N= 141
Q4: -186.6 (-363.4, -9.8), p-value
<0.05
p-trend = 0.031
Males
Q2: -56.7 (-255.9, 142.4)
Q3: 95.9 (-116.5, 308.4)
Q4: 30.5 (-169.7, 230.8)
p-trend = 0.187
Results: Lowest quartile used as reference.
Confounding: GA, maternal age, pre-pregnancy BMI, smoking and drinking during pregnancy, parity, annual household income, blood
sampling period.
Kobayashi et al. Japan, 2002-
(2017) 2005
Medium
Cross-sectional Pregnant women Maternal serum BL (cm), BW Regression
at 22-35 wk
gestation and
infants from
Hokkaido Study
on Environment
and Children's
Health
N = 177
(g)
5.3 (3.9-7.2)
coefficient per
ln-unit increase
in PFOS
Length: 0.32 (-0.19, 0.82)
BW: -56 (-162.8, 50.8)
Length and B W: no statistically
significant associations
Confounding: Maternal age, pre-pregnancy BMI, parity, maternal education, maternal smoking during pregnancy, GA, infant sex, maternal
blood sampling period.
Kobayashi et al.
(2022)
Medium
Japan
Recruitment:
2002-2005
Cohort
Mother-child pairs Maternal blood BW (g), BL Regression
BW
from the Sapporo in the third
(cm)
Cohort of the
Hokkaido Birth
Cohort
N = 372 (198
female children,
trimester
5.2 (3.7-7.2)
Females
174 male children) 5.2(3.4-7.3)
Males
5.3 (3.9-7.0)
IVVglVOiJlVJil J-/ V*
coefficient per -182.3 (-3 36.5,-28.2),
loglO-unit p-value = 0.021
increase in Females: -292.1 (-504.3, -79.8),
PFOS p-value = 0.007
Males: 17.7 (-207, 242.5),
p-value = 0.876
BL
-0.552 (-1.433,0.328),
p-value = 0.218
Females: -1.384 (-2.472, -0.297),
p-value = 0.013
D-36
-------
APRIL 2024
Reference,
Confidence
Location, Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison Resultsb
^cars ^ Timing, Levels3
Males: 0.635 (-0.832, 2.102),
p-value = 0.394
Confounding: Maternal age (continuous), pre-pregnancy BMI (continuous), maternal smoking in the third trimester (yes/no), maternal
alcohol consumption during pregnancy (yes/no), parity (primiparous, multiparous), educational level, annual household income, cesarean
section (yes/no), maternal blood sampling period, GA (continuous), infant sex.
Kwon et al.
(2016)
Medium
Korea, 2006- Cohort Pregnant women Cord blood BW (g) Regression -49.41 (-95.57,-3.25),
2010 and infants from coefficient per p-value = 0.04
EBGRC 0.64(0.29-1.09) log-unit increase
N = 268 in PFOS
EBGRC = Ewha Birth and Growth Retrospective Cohort
Comparison: Logarithm base not specified.
Confounding: Mother's age, pre-pregnancy BMI, past history of alcohol consumption and child's GA, gender, parity.
Lenters et al.
(2016)
Medium
Greenland, Cohort Pregnant women Maternal serum BW at term Regression -114.36 (-206.81,-21.91),
Poland, and and singleton Later pregnancy (g) coefficient per p-value = 0.015
Ukraine infants from GM= 9.357 (2- 2-SD increase in
2002-2004 INUENDO SD In- ln-PFOS
N= 1,250 PFOS = 1.600)
INUENDO = Biopersistent Organochlorines in Diet and Human Fertility
Confounding: Study population, maternal age, pre-pregnancy BMI, parity.
Liew et al.
(2020)
Medium
Denmark, Nested case- Females from the Plasma Miscarriage OR per doubling 1.2(0.9,1.8)
1996-2002 control Danish National Control: 23.35 ofPFOSorby Q2: 1.1 (0.6, 1.9)
Birth Cohort, (18.1,30.30) quartiles Q3: 1.3 (0.8,2.4)
N = 438 Cases: 24.55 Q4: 1.4 (0.8,2.4)
(19.5, 32.25)
Results: Lowest quartile used as the reference group.
Confounding: Maternal age, parental socio-occupational status, maternal smoking in the first trimester, maternal alcohol intake in the first
trimester, gestational week of blood sampling, parity.
Louis et al.
(2016)
Medium
United States, Cohort Females from the Plasma Pregnancy HR per log-unit 0.81 (0.65,1.00)
2005-2009 LIFE Study, Pregnant: 12.2 loss increase in T2:0.81 (0.50, 1.33)
Ages 18-40, (8.3,17.8) PFOSorby T3:0.60 (0.35, 1.03)
N = 344 Infertile: 12.1 tertiles
(7.1, 17.1)
Comparison: Logarithm base not specified.
Confounding: Age, BMI, prior pregnancy loss conditional on previous pregnancy, any alcohol consumption during pregnancy, any cigarette
smoking during pregnancy.
D-37
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Maisonet et al. Great Britain Cohort
(2012)
Medium
Recruitment:
1991-1992,
followed up
until 20 mo of
age
Pregnant women Maternal serum
and their singleton during
girls assessed at pregnancy
birth, 9, and 20 mo (median 15 wk)
from ALSPAC
19.6
BW (Range = 3.8-
N = 422 112.0)
BL
N = 356
GA
N = 444
Ponderal index
N = 360
Weight at 20 mo
N = 320
(106 upper tertile
of BW,
107 middle tertile
of BW,
107 lower tertile
of BW)
BW (g), BL
(cm), GA
(weeks),
ponderal
index (g/cm3),
weight at
20 mo (g)
Regression BW
coefficient by T2: -111.71 (-208.24, -15.17)
tertile T3:-140.01 (-238.14,-41.89)
p-trend = 0.0053
BL
T2: -0.72 (-1.19,-0.25)
T3: -0.63 (-1.11,-0.15)
p-trend = 0.0103
GA
T2: -0.02 (-0.39, 0.35)
T3:-0.15 (-0.53, 0.23)
p-trend = 0.4352
Ponderal index
T2: 0.00 (-0.07, 0.06)
T3: 0.05 (-0.01,0.12)
p-trend = 0.1120
Weight at 20 mo
T2: 310.64 (27.19, 594.08)
T3: 579.82 (301.4, 858.25)
p-trend < 0.0001
Upper tertile of B W
T2: 333.57 (-301.28, 968.42)
T3: 596.22 (-52.98, 1245.42)
p-trend = 0.0714
Middle tertile of BW
T2: -262.83 (-884.25, 358.60)
T3: 165.43 (-439.52, 770.37)
p-trend = 0.5886
Lower tertile of BW
T2: 602.64 (-150.79, 1356.07)
T3: 932.71 (186.90, 1678.52)
p-trend = 0.0148
D-38
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Meng et al.
(2018)
Medium
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
ALSPAC = Avon Longitudinal Study of Parents and Children
Results: Lowest tertile used as reference
Confounding: BW: maternal smoking during pregnancy, maternal pre-pregnancy BMI, previous live births, and GA; BL additionally adjusted
for maternal education. GA: GA when maternal serum sample was obtained. Ponderal index: maternal pre-pregnancy BMI, previous live
births, and GA when maternal serum sample was obtained. Weight at 20 mo (all tertiles): height at 20 mo, BW, maternal education, maternal
age at delivery, and previous live birth; intratertile analyses adjusted for maternal education, maternal age at delivery, previous live birth, and
BW.
Manzano- Spain, 2003- Cohort Mother (aged Maternal blood
Salgadoetal. 2008 >16 yr)-child pairs
(2017b) from INMA GM = 5.80
Medium assessed at birth (4.52-7.84)
and 6 mo
N = 1,154 (568
girls, 586 boys)
Weight gain Regression Weight gain z-score
z-score, rapid coefficient or -0.02 (-0.11, 0.07)
growth RR per log2-unit Girls:-0.09 (-0.21, 0.04)
increase in Boys: -0.05 (-0.08, 0.19)
PFOS p-value for sex interaction = 0.54
Rapid growth
0.92 (0.80, 1.06)
INMA = INfancia y Medio Ambiente [Environment and Childhood Project]
Outcome: Rapid growth defined as a z-score >0.67 standard deviation for weight gain from birth until 6 mo.
Confounding: Maternal characteristics (i.e., region of residence, country of birth, previous breastfeeding, age, pre-pregnancy BMI), age and
sex of child
Denmark,
1996-2002
Cohort
Pregnant women Maternal serum BW (g), GA Regression
BW
and their infants
from DNBC
N = 3,522 (1,533
girls, 1,969 boys)
Early (days), low
pregnancy, Later BW, PTB
pregnancy
30.1 (22.9-39.0)
coefficient (BW, -45.2 (-76.8., -13.6)
GA) or OR Q2: 24.7 (-24.8, 74.1)
(LBW, PTB) per Q3: -50.1 (-101.1, 0.9)
doubling of Q4: -48.2 (-99, 2.5)
PFOS and by Females: -65.3 (-111.7, -18.9)
quartiles Males: -24.3 (-67.1, 18.6)
p-value for sex interaction = 0.31
GA
-1.1 (-1.7,-0.4)
Q2: -1.1 (-2.1, -0.1)
Q3:-2 (-3.1,-1)
Q4: -1.5 (-2.6, -0.5)
Females: -1 (-2, 0)
Males:-1.1 (-2.0,-0.3)
p-value for sex interaction = 0.72
D-39
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
LBW
1.3 (0.9, 2.0)
Q2: 1.4 (0.7, 2.8)
Q3: 1.8(0.9,3.6)
Q4: 1.2 (0.6, 2.4)
PTB
1.5(1.1,2.2)
Q2: 2.0 (1.1, 3.6)
Q3: 3.3 (1.8, 5.8)
Q4: 1.9(1.0,3.5)
DNBC = Danish National Birth Cohort
Results: Lowest quartile used as reference.
Confounding: Infant sex, infant birth year, gestational week of blood draw, maternal age, parity, socio-occupational status, pre-pregnancy
BMI, smoking during pregnancy, alcohol intake during pregnancy, study sample
Ou et al. (2021) China, 2014- Nested case-
Medium 2018 control
Pregnant women
and their children
with (cases) and
without (controls)
CHD
N = 316
Maternal blood
and cord blood
at delivery
Maternal blood
Cases: 5.752
(3.655-8.683)
Controls: 5.742
(4.156-6.850)
Cord blood:
Cases: 1.928
(0.823-3.295)
Controls: 2.237
(1.505-3.072)
Septal defects,
conotruncal
defects, total
CHD
OR for >75th
percentile vs.
<75th percentile
PFOS
Maternal PFOS
Septal defects: 1.92 (0.80, 4.60)
Conotruncal defects: 1.65 (0.59,
4.63)
Total CHD: 1.61 (0.91,2.84),
p-value <0.10
Cord PFOS
Septal defects: 1.15 (0.38, 3.54)
Conotruncal defects: 0.63 (0.16,
2.57)
Total CHD: 1.03 (0.46,2.3)
CHD = Congenital heart defects
Outcome: Total congenital heart defects included septal defects and conotruncal defects, as well as individual congenital heart defect subtypes
with a large number of cases.
Confounding: Maternal age, parity, infant sex.
D-40
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Robledo et al. United States, Cohort Couples and their
(2015) 2005-2009 children from the
Medium LIFE study
N = 234
Serum
Early pregnancy
Girls:
GM = 12.44
(95%
CI = 11.50,
13.44)
Boys:
GM = 21.6
(95%
CI = 19.97,
23.39)
BW (g), HC
(cm), BL
(cm),
ponderal
index (g/cm3)
Regression Maternal PFOS
coefficient for Girls:
mean change per BW: 14.16 (-81.83, 110.15)
1-SD increase in HC: -0.04 (-0.46, 0.38)
ln(maternal BL: 0.30 (-0.26, 0.86)
PFOS) and in Ponderal Index: -0.03 (-0.10,
ln(paternal 0.03)
PFOS) Boys:
BW: 37.51 (-73.45, 148.46)
HC: 0.07 (-0.45, 0.60)
BL: 0.22 (-0.43, 0.86)
Ponderal Index: 0.00 (-0.07, 0.08)
Paternal PFOS
Girls:
BW: 38.58 (-59.29, 136.45)
HC: 0.29 (-0.14, 0.71)
BL: -0.05 (-0.62, 0.52)
Ponderal Index: 0.05 (-0.02, 0.11)
Boys:
BW: 36.85 (-73.14, 146.84)
HC: 0.16 (-0.37,0.68)
BL: -0.20 (-0.84, 0.43)
Ponderal Index: 0.06 (-0.02, 0.13)
LIFE = Longitudinal Investigation of Fertility and the Environment
Confounding: Maternal and paternal serum lipids, serum cotinine, BMI, maternal age, difference in paternal age, infant gender, individual
and partner sum of remaining chemical concentrations in each chemical's respective class
Stein et al.
United States Cohort
Pregnant women
Maternal serum
Birth defects,
OR per IQR
Birth defects
(2009)
2005-2006
and their infants
within 5 yr after
PTB, LBW
increase in
Per IQR increase: 1.1 (0.9, 1.3)
Medium
from the C8HP
pregnancy
PFOS
13.6 (9.0-17.7)
PTB
Birth defects
PTB, LBW:
Per IQR increase: 1.1 (1.0, 1.3)
N = 3,996
OR by percentile 50th-75thpercentile: 1.1 (0.9, 1.3)
PTB
75th-90th percentile: 1.1 (0.9, 1.3)
N = 4,512
>90th percentile: 1.4(1.1, 1.7)
Low BW
D-41
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
N = 4,561
LBW
Per IQR increase: 1.3 (1.1, 1.6)
50th-75thpercentile: 1.3 (0.9, 1.8)
75th-90th percentile: 1.6 (1.1, 2.3)
>90th percentile: 1.8 (1.2, 2.8)
C8HP = C8 Health Project
Population: Includes "women who lived in the same contaminated water district from the approximate start of the pregnancy through the time
of enrollment... to ensure that the PFOA level measured at C8 Health Project enrollment would reflect the level at the time of pregnancy."
Outcome: PTB defined as birth at <37 wk; low BW defined as <5.5 pounds at birth.
Results: <50th percentile used as reference group.
Confounding: Maternal age, parity, educational level at interview, smoking status at interview, PFOA in the analysis of PFOS.
Tian et al. China Cohort Pregnant women
(2019b) 2012-2014 and their sons at
Medium birth, 6 mo, and
12 mo from the S-
MBCS
Birth N = 439
6-mo N = 322
12-mo N = 301
Maternal serum Weight gain Regression
z-score (0- coefficient per
6 mo or 6-
12 mo),
AGDap,
AGDas
10.70 (7.61-
15.71)
ln-unit increase
in PFOS or by
quartiles AGDap
Weight gain z-score
0-6 mo: -0.06
6-12 mo: 0.12; p-value < 0.05
Weight gain z-
score: Pearson
correlation
coefficient
Quartile analysis showed no other
statistically significant associations
S-MBCS = Shanghai-Minhang Birth Cohort Study; AGDap = anopenile distance; AGDas = anoscrotal distance
Results: Lowest quartile used as reference.
Confounding: Maternal age at delivery, GA, maternal education, parity, pre-pregnancy BMI, infant age at physical examination, infant body
size
Toft et al.
Denmark Case-control Pregnant women
Amniotic fluid
Cryptorchidis
OR per ln-unit
Cryptorchidism
(2016)
1980-1996 and their sons
m,
increase in
0.99 (0.75, 1.30)
Medium
from the DMBR
Second
hypospadias
PFOS or by
T2: 1.08 (0.71, 1.63)
N = 270
exposure tertile:
tertiles
T3: 1.01 (0.66, 1.53)
cryptorchidism
0.8-1.4
cases, 75
Hypospadias
hypospadias cases,
0.87 (0.57, 1.34)
and 300 controls
T2: 0.97 (0.51, 1.87)
T3: 0.69 (0.35, 1.38)
DMBR = Danish Medical Birth Registry
D-42
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Outcome: Cryptorchidism defined as both a diagnosis of undescended testis and a corrective surgical procedure recorded in the Danish
National Patient Registry (DNPR). Hypospadias defined as diagnosis in the DNPR.
Results: Lowest tertile used as reference
Confounding: GA of amniocentesis, maternal age, smoking (cotinine groups), and case or control status
Vesterholm et
al. (2014)
Medium
Wang et al.
(2019a)
Medium
Denmark and
Finland
Recruitment
1997-2002,
follow-up 3 mo
after birth
Nested case-
control
Boys with (cases)
or without
(controls)
cryptorchidism
N = 215
Cord blood
9.1 (5th-95th
percentile: 4.8-
16.4)
Cryptorchidis
m
OR per ln-unit
increase in
PFOS or by
tertiles
Continuous: 0.83 (0.44, 1.58)
T2: 0.70 (0.34, 1.46)
T3: 0.83 (0.39, 1.78)
p-trend = 0.64
Outcome: Cryptorchidism defined as by Scorer (1964).
Exposure Level: Denmark cases: 2.4 (5th-95th percentile: 1.4-4.4); controls: 2.70 (5th-95th percentile: 1.4, 4.0); Finland cases: 1.9 (5th-
95thpercentile: 1.0-3.9); controls: 2.3 (5th-95th percentile: 1.2-4.8)
Results: Lowest tertile used as reference.
Confounding: BW, GA, parity
China
2013
Cross-sectional Pregnant women Cord blood
and their children Later pregnancy
at birth 0.65 (0.40-1.19)
N = 340 (171 girls,
169 boys)
BL (cm), BW
(g), BWz-
score, HC
(cm),
ponderal
index (g/cm3)
Regression BL, BW, HC, ponderal index: no
coefficient per statistically significant associations
loglO-unit or interactions by sex
increase in BL
PFOS -0.01 (-0.40,0.39);
p-value = 0.982
Girls: -0.01 (-0.60, 0.58);
p-value = 0.968
Boys:-0.17 (-0.71, 0.37);
p-value = 0.535
p-value for interaction by
sex = 0.557
BW
54.5 (-149.07, 40.06);
p-value = 0.259
Girls: -57.3 (-201.38, 86.78);
p-value = 0.436
Boys: -61.6 (-184.61, 61.42);
p-value = 0.326
D-43
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
p-value for interaction by
sex = 0.844
BW z-score
-0.15 (-0.41, 0.11);
p-value = 0.258
HC
0.02 (-0.26, 0.29); p-value = 0.915
Girls: -0.01 (-0.42, 0.39);
p-value = 0.947
Boys: -0.04 (-0.41, 0.32);
p-value = 0.821
p-value for interaction by
sex = 0.709
Ponderal index
-0.04 (-0.09, 0.001);
p-value = 0.054
Girls: -0.04 (-0.11,0.02);
p-value = 0.198
Boys: -0.02 (-0.08, 0.03);
p-value = 0.427
p-value for interaction by
sex = 0.637
Confounding: Pregnant age, family income, maternal education level, maternal career, husband's smoking, energy daily intake, daily physical
activity, GA, parity, pre-pregnant maternal BMI, gestational diabetes mellitus, infant sex, delivery mode, gestational weight gain
Woods et al
(2017)
Medium
HOME = Health Outcomes and Measures of Environment
Confounding: Maternal race, age at delivery, infant sex, maternal education, tobacco exposure, household annual income, employment,
maternal insurance status, marital status, prenatal vitamin use, maternal BMI, GA
United States, Cohort Pregnant women Maternal serum BW (g) Regression -8.7 (-52.8, 34.9)
Recruitment: and their children Later pregnancy coefficient per
2003-2006; at birth from the 14.4(10-17.0) loglO-unit
outcome HOME study increase
assessed at birth N = 272 maternal PFOS
D-44
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
Yang et al.
(2022In Press)
Medium
China
2018-2019
Nested case-
control
Infants from the
KBCS,
N = 768
(403 males, 365
females)
PTBs
N = 384
(205 males,
females)
Term births
N = 384
179
Cord blood at PTB, GA PTBs: OR per PTB
birth (weeks) IQR increase in 1.44 (1.18, 1.79), p-value <0.01
Term births PFOS PFAS-residuals model: 1.71 (1.26,
0.266(0.144- 2.4), p-value <0.001
0.444) GA: Males
PTBs Regression 1.45 (1.10,2.03)
0.213(0.112- coefficient per Females
0.483) IQR increase in 1.40 (1.10,1.93)
PFOS p-value for interaction by infant's
sex = 0.99
GA
PTBs, total
-1.26 (-2.46, -0.05),
p-value = 0.04
PFAS-residuals model: -2.01
(-3.42,-0.61), p-value = 0.01
PTBs, males
-0.41 (-2.2, 1.37)
PTBs, females
-1.06 (-2.87,0.74)
p-value for interaction by infant's
sex = 0.14
Term births
-0.16 (-1.81, 1.48), p-value = 0.85
KBCS = Kashgar Birth Cohort Study
Confounding: Maternal age, maternal ethnicity, maternal BMI, household income, maternal education level, maternal tobacco smoking
during pregnancy, maternal alcohol drinking during pregnancy, parity, living near a factory, periconceptional folic acid intake, gestational
diabetes, gestational hypertension. Additional confounding for analyses with both sexes: infant's sex. Additional confounding for PFAS-
residuals model: residuals regressed from PFDoA with PFOA, PFDA, PFUdA, PFNA, and PFTrDA.
Callan et al.
(2016)
Low
Australia 2008-
2011
Cross-sectional
Mother-infant
pairs enrolled in
AMETS,
Ages 19-44,
N = 98
Maternal blood
1.99 (0.45-8.1)
BW (g), BL
(cm),
Proportion of
optimal BW
(POBW), HC
(cm),
Regression BW
coefficient per -69 (-231, 94)
ln-unit increase
in PFOS BL
-0.22 (-1, 0.57)
D-45
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
ponderal
index
(g/cm3 x 100)
, proportion of
optimal birth
length
(POBL),
proportion of
optimal HC
(POHC)
POBW
0.48 (-4.2, 5.2)
HC
-0.39 (-0.98, 0.2)
Ponderal Index
-0.03 (-0.14,0.08)
POBL
-0.12 (-1.4, 1.7)
POHC
-0.6 (-2.3, 1.1)
AMETS = Australian Maternal Exposure to Toxic Substances
Confounding: For BW, BL, HC, and ponderal index: GA, maternal height, pre-pregnancy BMI, weight gain during pregnancy, sex of infant.
For POHC: Weight gain during pregnancy, annual household income. For POBL: Weight gain during pregnancy, maternal age, annual
household income.
Cao et al.
China,
Cohort
Infants from
Cord blood
BW (g), BL
BW, BL, HC
BW
(2018)
2013-2015
Zhoukou City,
(cm),
and ponderal
T2: 103.5 (-17.8, 224.8)
Low
China,
1.01 (0.60-1.76) ponderal
index:
T3: -17.6 (-141.2, 106)
N = 337 (183
index
Regression
Males
males, 154
(g/cm3),
coefficient by
T2: 76.2 (-91.1, 243.6)
females)
postnatal
weight (g),
tertiles
T3: 9.6 (-165.6, 184.8)
Females
Postnatal weight,
postnatal
T2: 146.8 (-36.2, 329.9)
postnatal length,
length (cm),
T3: -6.7 (-184.8, 171.4)
postnatal HC
postnatal HC,
N = 282 (157
birth defects
BL
males, 125
T2: 0.33 (-0.01, 0.68)
females)
T3: 0.07 (-0.27,0.42)
Males
T2: 0.4 (-0.05, 0.84)
T3: 0.27 (-0.19,0.74)
Females
T2: 0.3 (-0.25, 0.86)
D-46
-------
APRIL 2024
Reference, Location, Population,Ages, MaS°S«mple Outcome Comparison Rente-
„ j.. . Design Matrix, Sample
Confidence Years N Timing, Levels'
T3: -0.04 (-0.58,0.5)
Ponderal index
T2: 0.02 (-0.07,0.1)
T3: -0.04 (-0.13,0.06)
Males
T2: -0.03 (-0.17,0.12)
T3: -0.06 (-0.21,0.09)
Females
T2: -0.03 (-0.17 0.12)
T3: -0.06 (-0.21,0.09)
Postnatal weight
T2:-138.1 (-573.7, 297.6)
T3: -78.3 (-531.6,374.9)
Males
T2: -427.6 (-959.2, 104)
T3: -321.2 (-894.3,252)
Females
T2: 239.6 (-519.6, 998.8)
T3: 128 (-620.3,876.3)
Postnatal length
T2: 0.08 (-1.78, 1.95)
T3: -0.1 (-2.04, 1.84)
Males
T2: -1.05 (-3.4, 1.29)
T3: 0.17 (-2.36, 2.7)
Females
T2: 1.07 (-2,4.13)
T3: -0.72 (-3.74,2.31)
Postnatal HC
T2: 0.17 (-0.76, 1.09)
T3: -0.23 (-1.19,0.73)
Males
D-47
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome
^ Timing, Levels3
Comparison
Resultsb
T2: 0.27 (-0.92, 1.45)
T3: 0.28 (-1, 1.56)
Females
T2: -0.19 (-1.69, 1.31)
T3: -1.22 (-2.7,0.25)
Birth defects
T2 OR: 0.84 (0.37, 1.91)
T3 OR: 1.27 (0.59, 2.73)
Comparison: Tertiles were defined as follows: T2 = 0.74-1.52 vs. <0.74. T3 = >1.52 vs. <0.74.
Results: Lowest tertile used as reference.
Confounding: Maternal age, household income, parity, infant's gender. Additional confounding for BW, birth defects, ponderal index:
smoking of father, drinking of father. Additional confounding for B W, birth defects, ponderal index, postnatal weight, postnatal length,
POHC: maternal education. Additional confounding for postnatal weight, postnatal length, and POHC: infant's age.
Espindola
Brazil Cross-sectional Mother-child pairs Cord blood from BW (z-score),
Regression
BW
Santos et al.
Recruitment: of women enrolled newborns BL (z-score),
coefficient per
0.06 (-0.42, 0.54)
(2021)
2017 in the PIP A weight-for-
loglO-unit
Low
project 2.06(1.06-5.21) length (z-
increase in
BL
score), HC (z-
PFOS
-0.02 (-0.54, 0.50)
BW: N = 72 score)
BL: N = 65
W eight-for-length
W eight-for-length:
0.38 (-0.28, 1.04)
N = 64
HC: N = 62
HC
0.18 (-0.46, 0.82)
PIPA = Rio Birth Cohort Study
Population: Mothers were recruited between 29th and 32nd weeks of gestation and were over 16 yr of age.
Exposure: Year of assessment not reported.
Confounding: Education, income, race, pre-gestational BMI, smoking active and passive, alcohol consumption, GA, primiparity, age
(continuous), and fish consumption.
Gennings et al.
Sweden, Cohort Mothers and their Prenatal serum BW (g)
Regression
BW
(2020)
Recruitment: children (age 7) Mean
coefficient per
-70.39 (SE= 16.31), p-value
Low
2007-2010, from the SELMA (SE) = 0.82
Follow-up at study (0.19)logl0-
7 yr N= 1,312 ng/mL
loglO-unit
increase in
PFOS
<0.001
D-48
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Sample Outcome Comparison
^ Timing, Levels3
Resultsb
SELMA = Swedish Environmental Longitudinal, Mother and child, Asthma and Allergy
Confounding: My Nutrition Index (MNI, z-score), sex, maternal smoking status, maternal weight (z-score), premature birth status, maternal
education, total energy intake (z-score)
Gross et al.
(2020)
Low
United States
2012-2014
Nested Case-
control
Healthy and
overweight 18-
mo-old Hispanic
children from
StEP,
N = 98
Newborn blood BW (z-score), Regression
BW
Mean overweight coefficient (BW) -0.62 (-0.96, -0.29), p-value
(SD) = 0.440 and OR <0.00714
(0.364) (overweight) for
PFOS >mean Overweight
level vs. PFOS < 0.43 (0.17, 1.09)
mean level
StEP = Starting Early Program
Outcome: Overweight defined as 18-mo weight-for-length z-score > 85th percentile
Confounding: Maternal age, maternal education, maternal depressive symptoms, pre-pregnancy BMI, GA, parity, intervention status.
Notes: AC = abdominal circumference; ACCEPT = Adapting to Climate Change, Environmental Pollution and Dietary Transition; ACD = Anoclitoris distance;
AFD = anofourchette distance; AGD = anogenital distance; AGDap = anopenile distance; AGDas = anoscrotal distance; ALSPAC = Avon Longitudinal Study of Parents and
Children; AMETS = Australian Maternal Exposure to Toxic Substances; ASD = anoscrotal distance; BL = birth length; BMI = body mass index; BPD = biparietal diameter;
BW = birth weight; C8HP = C8 Health Project; CHD = congenital heart defects; CIOB = Chemicals in our Bodies; DMBR = Danish Medical Birth Registry; DNBC = Danish
National Birth Cohort; DNPR = Danish National Patient Registry; EBGRC = Ewha Birth and Growth Retrospective Cohort FL = femur length; FLEHS = Flemish Environmental
and Health Study; FLEHS II = Flemish Environmental and Health Study II; GA = gestational age; GM = geometric mean; HC = head circumference; HOME = Health Outcomes
and Measures of Environment; HR = hazard ratio; HUMIS = Human Milk Study; INMA = INfancia y Medio Ambiente (Environment and Childhood) Project; IOM = Institute of
Medicine; IQR = interquartile range; KBCS = Kashgar Birth Cohort Study; LBW = low birth weight; LGA = large for gestational age; LIFE = Longitudinal Investigation of
Fertility and the Environment; LINC = Linking EDCs in Maternal Nutrition to Child Health; LSM = least squares mean; LWBC = Laizhou Wan Birth Cohort;
MIREC = Maternal-Infant Research on Environmental Chemicals; MNI = My Nutrition Index mo = months; MoBa = Norwegian Mother and Child Cohort Study;
NICHD = National Institute of Child Health and Human Development; NICHD SGA = National Institute of Child Health and Human Development Scandinavian Successive
Small-for-Gestational-Age Births Study; NCS = National Children's Study; NO = Norway; OR = odds ratio; PFNA = perfluorononanoic acid; PIPA = Rio Birth Cohort Study;
POBL = proportion of optimal birth length; POBW = Proportion of optimal birth weight; POHC = proportion of optimal head circumference; POPUP = Persistent Organic
Pollutants in Uppsala Primiparas; PTB = preterm birth; RR = relative risk ratio; S = singletons; SBC = Shanghai Birth Cohort; SD = standard deviation; SDS = standard deviation
score; SE = standard error; SE = Sweden; S-MBCS = Shanghai-Minhang Birth Cohort Study; SELMA = Swedish Environmental Longitudinal, Mother and child, Asthma and
Allergy; SGA = small for gestational age; StEP = Starting Early Program; T = twins; T1 = tertile 1; T2 = tertile 2; T3 = tertile 3; TBPS = Taiwan Birth Panel Study;
WAZ = weight-for-age z-score; WFL = weight-for-length; WLZ = weight-for-length z-score; wk = weeks; yr = years.
a Exposure reported as median (25th-75th percentile) in ng/mL unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D-49
-------
APRIL 2024
D.2 Reproductive
D.2.1 Male
Table D-2. Associations Between PFOS Exposure and Male Reproductive Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure „ ,
Matrix, Levels3 Outcome
Comparison
Resultsb
Children and Adolescents
Jensen et al.
(2020b)
High
Denmark
2010-2012
Cohort
Infants from
Odense Child
Cohort
N = 208 boys
Maternal serum Levels of FSH
3.33
(IU/L), testosterone
(nmol/L), LH (IU/L),
testosterone /LH
ratio, DHEAS
(nmol/L), DHEA
(nmol/L),
Androstenedione
(nmol/L), 17-OHP
(nmol/L)
Regression
coefficient
(testosterone), or
percent change per
doubling of PFOS
No statistically significant
associations
Confounding: Age of the child at examination time, maternal parity0
Lind et al.
(2017a)
High
Denmark Cohort
2010-2012
Infants from
Odense Child
Cohort
N = 649
(296 boys)
Maternal serum Penile width (mm),
Total cohort: 8.1 anogenital distance
(AGD; scrotal, as;
penile, ap) (mm)
Regression
coefficient per ln-
unit increase in
PFOS, or by
quartiles
AGDas
Continuous: 1.2 (-0.4, 2.7)
Q2: 0.9 (-0.9, 2.8)
Q3: 0.9 (-0.8,2.7)
Q4: 1.9 (0.04, 3.7)
p-trend by quartiles = 0.06
AGDap, penile width: no
statistically significant
associations
AGDap: p-trend by
quartiles = 0.55
Penile width: p-trend by
quartiles = 0.67
D-50
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
Results: Lowest quartile used as reference.
Confounding: Age at examination, WAZ, pre-pregnancy BMI, parity, smoking
Itoh et al. (2016) Japan Cohort Infants from
Medium 2002-2005 Sapporo Cohort
of the Hokkaido
Study on
Environment
and Children's
Health
N = 83 boys
loglO-unit increase
inPFOS, LSMby
quartiles
Maternal serum In cord blood, loglO- Regression
5.40 transformed levels of coefficient per
E2 (ng/mL), FSH
(mlU/mL), Inhibin B
(pg/mL), insulin-like
3 (ng/mL), LH
(mlU/mL),
progesterone
(ng/mL), prolactin
(ng/mL), SHBG (not
log 10-transformed,
nmol/L), testosterone
(pg/mL)
Testosterone/E2
ratio,
testosterone/SHBG
ratio
E2
0.372 (0.057, 0.687)
p-value = 0.021
Ql: 4.34 (3.07, 6.15)
Q2: 5.84 (4.34, 8.01)
Q3: 8.74 (6.33, 12.05)
Q4: 6.39 (4.52, 8.98)
p-trend = 0.027
Inhibin B
-0.439 (-0.620, 0.257)
p-value <0.001
Ql: 53.4 (42.4, 65.6)
Q2: 50.1 (41.2, 60.5)
Q3: 39.1 (31.8, 47.6)
Q4: 33.3 (26.6, 40.0)
p-trend < 0.001
Progesterone
-0.344 (-0.678, 0.01)
p-value = 0.043
Ql: 238.5 (161.5, 354.9)
Q2: 267.6 (192, 375.3)
Q3: 241.5 (168.7, 346.2)
Q4: 184.7 (126.5, 267.6)
p-trend = 0.231
Testosterone/E2
-0.399 (-0.643,-0.156)
p-value = 0.002
Ql: 20.3 (15.2, 26.8)
Q2: 19.5 (15.2, 24.8)
Q3: 14.5 (10.7, 18.6)
Q4: 14.5 (10.8, 18.8)
D-51
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
p-trend = 0.015
FSH, insulin-like 3, LH, prolactin,
SHBG, testosterone,
testosterone/SHBG: No
statistically significant
associations or trends
Confounding: Age, parity, BMI before pregnancy, annual income, smoking during pregnancy, caffeine consumption during pregnancy,
gestational weeks of blood sampling for PFOS/PFOA measurement, gestational age at birth
Lopez-Espinosa
etal. (2016)
Medium
United Cross- Male children Serum
States sectional ages 6-9 yr 22.4
2005-2006 N= 1,169
Total testosterone
(ln-ng/dL)
Percent difference Total testosterone:
between 75th and 25th -5.8 (-9.4, -2.0)
percentile of ln-unit Q2
PFOS or by quartiles Q3
Q4
-4.2 (-11.4, 3.6)
-9.2 (-16.1,-1.6)
-11.8 (-18.6, -4.3)
p-trend = 0.002
Results: Results by quartile used lowest quartile as reference.
Confounding: Age, month and time of sampling
Goudarzi et al.
(2017a)
Medium
Japan
2002-2005
Cohort
Children from
the Hokkaido
Study
N = 185 (81
males)
Serum
Total cohort:
5.20
Levels (loglO ng- Regression
mL) of DHEA,
androstenedione
coefficient per
loglO-unit increase
in PFOS or by
quartiles
Among males
DHEA: 0.308 (0.099, 0.755);
p-value = 0.011
Androstenedione: -0.011 (-0.312,
0.284); p-value = 0.926
Confounding: Gestational age, maternal age, parity, smoking and caffeine intake during pregnancy, maternal educational level, and blood
sampling period
Ernst et al.
(2019)
Medium
Denmark Cohort
1999-2017
Children from
the Puberty
Cohort of the
Danish National
Birth Cohort
N = 565 boys
Maternal blood
Sample 1: 31.9
Sample 2: 27.2
Age (months) at
axillary hair
attainment, voice
break, first
ejaculation, Tanner
stages 2-5 for
genital development
or pubic hair growth;
combined sex-
Regression
coefficient per log2-
unit increase in first
trimester maternal
serum PFOS
Puberty indicator:
mean difference in
No statistically significant
associations
D-52
-------
APRIL 2024
Reference,
Confidence
Population,
L*vca,,on< Design Ages, , Outcome
Years Matrix, Levels
N
Comparison Resultsb
specific puberty
indicator
age at puberty by
tertiles
Confounding: Highest social class of parents, maternal age at menarche, maternal age at delivery, parity, pre-pregnancy BMI, and daily
number of cigarettes smoked in first trimester
Tian et al.
(2019b)
Medium
China Cohort Male infants at Maternal plasma Anopenile distance
2012-2013 birth, 6 mo, and 10.70 (AGDap) (mm),
12 mo anoscrotal distance
N = 500 (AGDas) (mm)
Regression AGDap
coefficient per In- GEE (Birth, 6 mo, and 12 mo):
unit increase in -0.34 (-1.38, 0.69);
maternal PFOS or by p-value = 0.516
quartiles Birth: -0.04 (-0.78, 0.69);
p-value = 0.925
6 mo.:-1.20 (-3.29, 0.88);
p-value = 0.262
12 mo.: 0.69 (-1.83,3.22);
p-value = 0.589
Q2: 1.57 (-1.95, 5.09)
Q3: 5.17 (1.53, 8.81);
p-value < 0.05
Q4: -0.49 (-4.04, 3.07)
AGDas
GEE (Birth, 6 mo, and 12 mo):
-0.83 (-1.71, 0.06);
p-value = 0.067
Birth: -0.65 (-1.27, -0.02);
p-value = 0.0429
Q2: 0.17 (-0.79, 1.13)
Q3:-0.10 (-1.10, 0.90)
Q4: -1.46 (-2.44, -0.49);
p-value < 0.05
p-value for trend < 0.05
6 mo.:-2.21 (-4.28, -0.14);
p-value = 0.0372
12 mo.: 0.47 (-1.63,2.58);
p-value = 0.6587
Results: Lowest quartile used as reference.
D-53
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
Confounding: Maternal age at delivery, gestational age, maternal education, parity, pre-pregnancy BMI, infant age at physical examination,
and infant body size (birth weight at birth; WLZ at 6 and 12 mo of age)
Wang et al. China Cross- Pregnant women Cord blood Levels (loglO- Regression
(2019a) 2013 sectional and their Total cohort: ng/mL) of estrone coefficient per
Medium children 0.65 (0.40-1.19) (El), E2, estriol (E3) loglO-unit increase
N = 340 inPFOS
(169 boys)
Zhou et al.
(2016)
Low
El: 0.071 (-0.05,0.18);
p-value = 0.247
E2: 0.02 (-0.10,0.14);
p-value = 0.761
E3: 0.36 (0.16, 0.55);
p-value <0.001
Confounding: Pregnant age, family income, maternal education level, maternal career, husband's smoking, energy daily intake, daily
physical activity, gestational age, parity, pre-pregnant maternal BMI, gestational diabetes mellitus, infant sex, delivery mode, gestational
weight gain
Arbuckle et al.
(2020)
Medium
Canada Cohort
2008-2011
Newborns from Maternal plasma Anopenile distance Regression
AGDap
the MIREC
cohort
N = 205 boys
4.4
(AGDap) (mm),
anoscrotal distance
(AGDas) (mm)
coefficient per ln-
unit increase in Q2
maternal PFOS or by Q3
quartiles Q4
Per In increase: 0.13 (-1.13, 1.38)
-0.97 (-2.81, 0.87)
-1.28 (-3.22, 0.66)
0.22 (-1.68,2.13)
p-value for trend = 0.908
AGDas
Per In increase: 1.05 (-0.24, 2.35)
Q2: -0.87 (-2.78, 1.04)
Q3: 0.33 (-1.67, 2.33)
Q4: 0.49 (-1.47, 2.46)
p-value for trend = 0.3936
Results: Lowest quartile used as reference.
Confounding: AGDap: recruitment site, education, active smoking status, gestational age; AGDas: household income, active smoking status,
gestational age
Taiwan Cross-
2009-2010 sectional
Adolescents
ages 13-15
N = 225 (102
boys)
Serum
Total: 28.9
Boys: 29.9
Levels (In-
transformed) of E2
(pmol/L),
testosterone
(nmol/L)
Regression
coefficient per unit
increase in PFOS
Testosterone, boys: -0.0029
(-0.0055, -0.0003)
p-value for interaction by
sex = 0.060
E2: No statistically significant
associations or interactions
Confounding: Age, sex, BMI, environmental tobacco smoke exposure, parental education, regular exercise, month of survey
D-54
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
Zhou et al.
(2017c)
Low
Taiwan Case-
2009-2010 control
Children ages
10-15 with
(cases) or
without
(control) asthma
N = 231 cases,
225 controls
Serum
Cases: 33.94
Controls: 28.91
Levels of
testosterone (ln-
nmol/L)
Regression
coefficient per unit
increase in PFOS
Testosterone
Cases: -0.004 (-0.005, -0.003)
Controls: -0.002 (-0.008, 0.003)
Confounding: Age, sex, BMI, parental education, environmental tobacco smoke exposure, physical activity, month of survey
Di Nisio et al.
(2019)
Low
Italy
2017-2018
Cross-
sectional
Male high
school students
N = 100 (50
unexposed
controls, 50
exposed)
Serum
Unexposed
controls: 0.82
Exposed: 1.11
Semen
Unexposed
controls: 0.11
Exposed: 0.11
AGD (cm), crown-
to-pubis distance
(cm), pubis-to-floor
distance (cm),
crown-to-
pubis/pubis-to-floor
ratio, penis
circumference (cm),
penis length (cm),
testicular volume
(mL), normal
morphology (%),
semen pH, immotile
sperm (%),
nonprogressive
motility (%),
progressive motility
(%), total sperm
count (106), semen
volume (mL), sperm
concentration
(106/mL), viability
(%), FSH (U/L),
testosterone
(nmol/L)
Mann-Whitney test
(Exposed vs.
Controls)
AGD
Controls: 4.50 (4.0, 5.2)
Exposed: 4.00 (3.5, 5.0)
Adjusted p-value for comparison
of medians = 0.114
Pubis-to-floor distance
Controls: 97.0 (93.0, 101.1)
Exposed: 95.0 (90.3, 99.0)
Adjusted p-value for comparison
of medians = 0.320
Penis circumference
Controls: 10.10 (9.9, 11.0)
Exposed: 9.50 (9.0, 10.0)
Adjusted p-value for comparison
of medians < 0.001
Penis length
Controls: 10.0(9.0, 11.0)
Exposed: 9.00 (8.0, 10.0)
Adjusted p-value for comparison
of medians < 0.001
Testicular volume
Controls: 16.13 (14.8, 19.0)
Exposed: 14.00 (12.6, 16.0)
D-55
-------
APRIL 2024
Population,
Reference, Location, . A Exposure „ . „ „ b
„ ..., Design Ages, ,, ^ „ Outcome Comparison Results
Confidence Years Matrix, Levels
N
Adjusted p-value for comparison
of medians < 0.001
Normal morphology
Controls: 7.0 (4.0, 12.0)
Exposed: 4.0 (2.0, 6.0)
Adjusted p-value for comparison
of medians < 0.001
Semen pH
Controls: 7.60 (7.5, 7.7)
Exposed: 7.70 (7.6, 7.7)
Adjusted p-value for comparison
of medians = 0.042
Testosterone
Controls: 18.98 (12.9, 17.9)
Exposed: 18.98 (16.3, 21.8)
Adjusted p-value for comparison
of medians < 0.001
Crown-to-pubis, Crown-to-
pubis/pubis-to-floor, sperm
motility, sperm count, semen
volume, sperm concentration,
viability, FSH: No statistically
significant associations after
adjusting for comparison of
medians
Results: Values for each outcome are reported as median (25th-75th percentile).
Confounding: Age
General Population
Cui et al. (2020)
Medium
China
2015-2016
Cross- Adult men
sectional N = 651
Serum Serum levels (ln-
9.94 transformed) of E2
(pmol/L), FSH
Percent change per
ln-unit increase in
serum or semen
SHBG
Serum PFOS: -4.94 (-8.71,
-1.02); p-value = 0.014
D-56
-------
APRIL 2024
Population,
Reference, Location, . A Exposure „ . „ „ b
„ ..., Design Ages, ,, ^ „ Outcome Comparison Results
Confidence Years Matrix, Levels
N
Semen (IU/L), LH (IU/L), PFOS, or by p-trend by quartiles = 0.004
0.15 SHBG (nmol/L), quartiles Ages < 30:-3.11 (-6.58, 0.48);
free testosterone, p-value = 0.069
total testosterone Semen PFOS: -5.29 (-8.94,
(nmol/L); free -1.49); p-value = 0.007
androgen index, total p-trend by quartiles = 0.026
testosterone/LH ratio Ages < 30: -3.13 (-6.25, -0.10);
p-value = 0.009
Total testosterone
Serum PFOS: -3.36 (-6.40,
-0.22); p-value = 0.036
p-trend by quartiles = 0.022
Ages < 30: -4.25 (-7.77, -0.59);
p-value = 0.023
Semen PFOS: -4.20 (-7.13,
-1.18); p-value = 0.007
p-trend by quartiles = 0.019
Ages < 30: -4.82 (-7.96, -1.58);
p-value = 0.004
Total testosterone/LH,
Serum PFOS: -4.53 (-8.99, 0.15);
p-value = 0.058
p-trend by quartiles = 0.044
Semen PFOS: -5.00 (-9.32,
-0.48); p-value = 0.031
p-trend by quartiles = 0.042
No statistically significant
associations by age groups
E2, FSH, free androgen, LH, free
testosterone: No statistically
significant associations or trends
Confounding: Age, BMI, smoking status, blood sampling time, fasting status
D-57
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
Petersen et al.
(2018)
Medium
Denmark Cross-
2007-2009 sectional
Faroese men
born between
1981 and 1984
N = 263
Serum
19.5
Levels (log-
transformed) of E2
(nmol/L), FSH
(IU/L), free
testosterone
(pmol/L), inhibin B
(pg/mL), LH, (IU/L),
SHBG (nmol/L),
testosterone
(nmol/L)
Regression
coefficient per log-
unit increase in
PFOS
LH: 0.35 (0.02,
p-value = 0.04
0.68);
SHBG: 0.31 (0.02,
p-value = 0.04
0.60);
No other statistically significant
associations
Ratios of free
testosterone/E2, free
testosterone/LH,
Inhibin B/FSH,
testosterone/E2,
testosterone/LH
Normal morphology
(%), motile sperm
(logit-%), total
sperm count
((106)i/3) Semen
volume (mL1/3),
sperm concentration
((106/mL)1/3)
Comparison: Logarithm base not specified.
Confounding: Age, BMI groups, current smoking, time of sampling
Kvist et al.
Greenland, Cross-
Male partners of Serum Y:X-chromosome
Linear regression 0.016; p-value = 0.026
(2012)
Poland, and sectional
pregnant women Mean ratio of sperm
adjusted r2
Medium
Ukraine
fromlNUENDO Greenland:
2002-2004
N = 359 51.65
Poland: 12.12
Ukraine: 8.20
Confounding: Age, abstinence time, alcohol intake and CB-153
D-58
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels"
Outcome
Comparison
Resultsb
Leter et al. Greenland, Cross- Male partners of Serum
(2014) Poland, and sectional pregnant women Mean = 27.2
Medium Ukraine from INUENDO
2002-2004 N = 262
Sperm DNA Regression
methylation level (% coefficient per ln-
5-mC) at LINE-1, unit increase in
Alu, or Sat-alpha;
global DNA
methylation level
(FCM DGML
channel no.)
PFOS
Sat-alpha
Total: 1.1 (-3.1, 5.3)
Greenland: -1.8 (-8.6, 5.1)
Poland:-7.2 (-16, 1.6)
Ukraine: 8.2 (0.6, 15.8)
Global
Total: -21 (-63.2, 21.3)
Greenland: -32.1 (-105.6, 41.3)
Poland: -108.4 (-191.5, -25.2)
Ukraine: 27.2 (-43.1, 97.6)
LINE-1, Alu: No statistically
significant associations
Confounding: Site, age (ln-transformed), smoking status
Pan etal. (2019) China Cross-
Medium 2015-2016 sectional
Adult men in
Nanjing
N = 664
Serum
Sperm normal
Regression
No statistically significant
8.378
morphology (%),
coefficient per ln-
associations by serum PFOS
count ((106)1/3),
unit increase in
levels; following results are by
Semen
concentration
serum or serum
semen PFOS
0.097
((106/mL)1/3),
PFOS, or by
progressive motility
quartiles
Progressive motility: -1.700
(%), curvilinear
(-2.867, -0.532); p-value = 0.03
velocity (VCL)
Q2
-2.30 (-5.27, 0.68)
(|im/s): straight-line
Q3
-1.53 (-4.61, 1.56)
velocity (VSL)
Q4
-5.54 (-8.72, -2.36)
(|im/s). DNA
p-trend = 0.01
fragmentation index
(DFI) (ln-%), high
VCL: -0.767 (-1.447, -0.087);
DNA stainability
p-value = 0.1
(HDS) (ln-%);
Q2
-1.60 (-1.50,2.01)
semen volume (ln-
Q3
-2.78 (-2.40, 1.10)d
mL)
Q4
-4.8 (-2.97, -0.72)
p-trend = 0.1
D-59
-------
APRIL 2024
Population,
Reference, Location, . A Exposure „ . „ „ b
„ ..., Design Ages, ,, ^ „ Outcome Comparison Results
Confidence Years Matrix, Levels
N
VSL: -0.773 (-1.337, -0.209);
p-value = 0.04
Q2: -1.00 (-2.44,0.45)
Q3: -1.40 (-2.89,0.09)
Q4: -2.06 (-3.60, -0.52)
p-trend = 0.1
DFI: 0.087 (0.033, 0.142);
p-value = 0.02
Q2: 0.03 (-0.11,0.17)
Q3: 0.08 (-0.07,0.22)
Q4: 0.25 (0.10,0.40)
p-trend = 0.01
Normal morphology, sperm count,
sperm concentration, sperm HDS,
semen volume: No statistically
significant associations
Results: Lowest quartile used as reference.
Confounding: Age, BMI, BMI2, smoking, alcohol intake, abstinence time
iVofes:17-OHP = 17-hydroxyprogesterone; AGD = anogenital distance; AGDap = anopenile distance; AGDas = anoscrotal distance; BMI = body mass index;
DHEA = dehydroepiandrosterone; DFI = DNA fragmentation index; DNA = deoxyribonucleic acid; El = estrone; E2 = estradiol; E3 = estriol; FCM DGML = flow cytometric
sperm DNA global methylation assay; FSH = follicle stimulating hormone; GEE = generalized estimating equation; HDS = high DNA stainability; INUENDO = Biopersistent
Organochlorines in Diet and Human Fertility; LH = luteinizing hormone; LSM = least squares mean; MIREC = Maternal-Infant Research on Environmental Chemicals;
mo = months; PFOA = perfluorooctanoic acid; PFOS = perfluorooctane sulfonic acid; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; SHBG = sex hormone
binding globulin; VCL = curvilinear velocity; VSL = straight-line velocity; WAZ = weight-for-age z-score; yr = years.
a Exposure levels reported as median in ng/mL unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
d Values are reproduced as reported in publication.
D-60
-------
APRIL 2024
D.2.2 Female
Table D-3. Associations Between PFOS Exposure and Female Reproductive Effects in Female Children and Adolescents
Reference,
Confidence
Location,
Year(s)
Study
Design
Population, Exposure
Ages, Matrix,
N Levels3 (ng/mL)
Outcome
Comparison
Resultsb
Jensen et al.
(2020b)
High
Denmark, Cohort
2010-2012
Female
infants from
the Odense
Child Cohort,
Age 4 mo,
N= 165
Maternal serum
8.07
percentile = 4.21,
15.50)
Levels of 17-OHP
(nM), androstenedione
(nM), DHEA (nM),
DHEAS (nM), FSH
(IU/L), LH (IU/L)
Percent change per
doubling in PFOS
17-OHP
2.1 (-11.9, 18.2)
Androstenedione
0.6 (-14.3, 18.2)
DHEA
-9.4 (-22.5, 5.9)
DHEAS
-10.4 (-28.4, 12.2)
FSH
0.2 (-12.5, 14.7)
LH
9.5 (-12.8, 37.6)
Confounding: Age of the child at examination time, maternal parity0
Lind et al.
(2017a)
High
Denmark
2010-2012
Cohort
Infants from
Odense Child
Cohort
N = 649 (353
girls)
Maternal serum
Total cohort: 8.1
Anogenital distance
(AGD) (mm); clitoral
(AGDac), fourchette
(AGDaf)
Regression AGDac
coefficient per ln-unit Continuous: -2.3 (-3.8,
-0.7)
increase in PFOS, or
by quartiles
Q2
Q3
Q4
-1.0 (-2.6, 0.6)
-1.7 (-3.5, 0)
-2.8 (-4.5,-1.1)
p-trend by quartiles < 0.01
AGDaf
Continuous: -0.4 (-1.6, 0.8)
No statistically significant
associations by quartiles, p-trend
by quartiles = 0.31
Results: Lowest quartile used as reference.
Confounding: Age at examination, WAZ, pre-pregnancy BMI, parity, smoking.
Yao et al.
(2019)
High
China,
2010-2013
Cross-
sectional
Pregnant
women
(aged > 18 yr)
Cord blood
1.39 (0.92,2.01)
Testosterone (loglO-
ng/mL),
Estradiol (loglO-
pg/mL).
Regression Testosterone
coefficient per log 10- 0.15 (0.01, 0.29), p-value < 0.05
unit increase in PFOS Estradiol
0.24 (-0.05, 0.07)
D-61
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population, Exposure
Ages, Matrix,
N Levels" (ng/mL)
Outcome
Comparison
Resultsb
and female
infants
N= 171
Testosterone-to-
estradiol ratio (loglO-
transformed)
Testosterone-to-estradiol ratio
0.14 (0.01, 0.27), p-value < 0.05
Confounding: Maternal age, pre-pregnancy BMI, parity, mode of delivery, passive smoking during pregnancy, gestational age, household
income level among male and female infants separately
Donley et al.
(2019)
Medium
United Nested
Kingdom, case-
Recruitment control
1991-1992,
outcome
assessed at
adolescence
Mothers and Maternal serum AMH (loglO-ng/mL) Regression
their
daughters
from the
ALSPAC,
N = 446
19.8(15.1,24.9)
coefficient per unit
increase in PFOS
Complete AMH data:
0.24 (0.00, 0.02)
Multiple imputation model:
0.01 (0.00, 0.015)
Confounding: Maternal age at delivery, pre-pregnancy BMI, maternal education
Ernst et al.
(2019)
Medium
Denmark, Cohort
Recruitment
1996-2002,
outcome
assessed
2012-2017
Female
adolescents
from the
Danish
Maternal blood
Sample 1
(N = 366):
32.3 (lO'MO4
Breast development,
pubic hair
development, age at
attainment of axillary
National Birth percentiles = 19.3 hair (months), age at
Cohort, , 50.8) menarche, age at
N = 555 attainment of
combined puberty
indicator
Combined puberty
indicator:
Mean difference by
tertiles of PFOS
All other outcomes:
Regression
coefficient per log2-
unit increase in PFOS
Combined puberty indicator
T2: -3.73 (-6.59, -0.87)
T3:-0.17 (-2.83, 2.49)
Breast development
-3.01 (-7.96, 1.95), p-value = 0.03
Pubic hair development
1.81 (-2.42, 6.04)
Axillary hair
0.50 (-2.79, 3.79), p-value = 0.02
Menarche
-0.68 (-3.13, 1.77)
Exposure Levels: [Sample 2] Median = 27.9 ng/mL (10th-90th percentiles = 16.5, 42.2 ng/mL). Samples 1 and 2 combined for analysis.
Outcome: Age in months at Tanner stage 5 used to measure breast development and pubic hair development. For combined puberty indicator,
lowest tertile was used as the reference group.
Confounding: Highest social class of parents, maternal age at menarche, maternal age at delivery, parity, pre-pregnancy BMI, daily number of
cigarettes smoked in first trimester
Goudarzi et al.
(2017a)
Medium
Japan,
2002-2005
Cohort
Pregnant
women and
their infants
from the
Hokkaido
Study on the
Maternal serum Levels of Regression
5.20(1.50,16.20) androstenedione coefficient per log 10-
(log 10-ng/mL), DHEA unit increase in PFOS
(loglO-ng/mL)
Androstenedione
0.004 (-0.29, 0.30),
p-value = 0.059
DHEA
0.24 (-0.02, 0.80)
D-62
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population, Exposure
Ages, Matrix,
N Levels" (ng/mL)
Outcome
Comparison
Resultsb
Itoh et al.
(2016)
Medium
Liu et al.
(2020b)
Medium
Environment
and
Children's
Health,
N= 104
Confounding: Gestational age, maternal age, parity, smoking and caffeine intake during pregnancy, maternal educational level, blood
sampling period
Japan,
2002-2005
Cohort
Female
infants from
the Sapporo
Cohort of the
Hokkaido
Study on
Environment
and
Children's
Health,
N= 106
Maternal serum
5.15 (3.45,7.00)
Cord blood levels of
estradiol (loglO-
ng/mL), testosterone
(loglO-pg/mL),
prolactin (loglO-
ng/mL), progesterone
(loglO-ng/mL), SHBG
(nmol/L); testosterone-
to-SHBG ratio,
testosterone-to-
estradiol ratio
Regression
coefficient per log 10-
unit increase in PFOS
Estradiol
0.08 (-0.15,0.31)
Testosterone
0.07 (-0.26, 0.40)
Prolactin
-0.49 (-0.76, -0.22),
p-value = 0.001
Progesterone
-0.55 (-0.89, -0.21),
p-value = 0.002
SHBG
-0.18 (-0.42, 0.06)
Testosterone/SHBG ratio
0.25 (-0.16,0.66)
Testosterone/estradiol ratio
-0.01 (-0.03 0.26)
Confounding: Maternal age, parity, BMI before pregnancy, annual income, smoking during pregnancy, caffeine consumption during
pregnancy, gestational weeks of blood sampling for PFOS/PFOA measurement, gestational age at birth
China,
2013-2014
Cross-
sectional
Female
neonates,
N= 191
Cord blood
4.15(2.81,6.18)
Levels of 17-OHP
(ng/mL), progesterone
(ng/mL)
Percent change per 17-OHP
IQR increase in PFOS -1.27 (-7.52, 5.39)
Progesterone
-1.68 (-6.93, 3.88)
Confounding: Maternal age at delivery, pre-pregnancy BMI, maternal education status, passive smoking during smoking, parity, gestational
weeks, sample-collection time
Lopez-
United Cohort
Females from
Serum
Levels of estradiol (ln-
Percent difference for
Estradiol
Espinosa et al.
States,
the C8 Health
20.9 (15.3,29.4)
pg/mL), total
75th vs. 25th
75th vs. 25th percentiles
(2016)
2005-2006
Project,
testosterone (ln-ng/dL)
percentiles, or by
-0.3 (-4.6, 4.2), p-value = 0.048
Medium
Ages 6-9,
quartiles
Q2: 5.2 (-3.7, 14.9)
D-63
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population, Exposure
Ages, Matrix,
N Levels" (ng/mL)
Outcome
Comparison
Resultsb
N= 1,123
Q3: 3.7 (-5.2, 13.4)
Q4: -1.3 (-9.9, 8.2)
Testosterone
75th vs. 25th percentiles
-6.6 (-10.1,-2.8)
Q2
Q3
Q4
-1.1 (-8.6,7.1)
-7.8 (-15.0,-0.1)
-11.1 (-18.2, -3.5)
Results: Lowest quartile used as the reference group.
Confounding: Age, month, time of sampling
Maisonet et al.
(2015a)
Medium
United
Kingdom,
1991-1992
Cohort
Female
adolescents
from
ALSPAC,
Age 15,
N = 72
Maternal serum
19.2(15.1,25.0)
Levels of serum total
testosterone (nmol/L),
SHBG (nmol/L)
Regression
coefficient by tertiles
ofPFOS
Testosterone
T2: 0.1 (-0.07,0.28)
T3: 0.18 (0.01, 0.35)
SHBG
T2: -2.86 (-18.8, 13.09)
T3: 3.46 (-12.06, 18.98)
Results: Lowest tertile used as the reference group.
Confounding: Maternal education, maternal age at delivery, maternal pre-pregnancy BMI, maternal smoking during pregnancy, time of day
daughter's blood sample was obtained, daughter's age at menarche, daughter's BMI at 15 yr. SHBG concentration included in testosterone
model.
Tsai et al.
(2015)
Medium
Taiwan, Cross- Female
2006-2008 sectional adolescents,
Ages 12-17,
N = 95
Serum, Levels of serum FSH
8.65 (5.37, 13.29) (ln-mlU/mL), serum
SHBG (ln-nmol/L)
Means by quartiles of FSH
PFOS Ql: 1.56 (SE = 0.23)
Q2: 1.67 (SE = 0.23)
Q3: 1.36 (SE = 0.19)
Q4: 1.23 (SE = 0.35)
SHBG
Ql: 3.58 (SE = 0.29)
Q2: 3.36 (SE = 0.29)
Q3: 3.49 (SE = 0.24)
Q4: 3.41 (SE = 0.44)
Confounding: Age, BMI, high-fat diet
Wang et al.
(2019a)
Medium
China, Cross- Pregnant
2013 sectional women and
their children,
N= 171
Cord blood Levels of estrone
0.65 (0.40, 1.19) (log 10-ng/mL), (3-
estradiol (loglO-
Regression Estrone
coefficient per ln-unit 0.15 (0.04, 0.26), p-value = 0.007
increase in PFOS -estradiol
D-64
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population, Exposure
Ages, Matrix,
N Levels" (ng/mL)
Outcome
Comparison
Resultsb
ng/mL), estriol (loglO-
ng/mL)
-0.17 (-0.31,-0.02),
p-value = 0.023
Estriol
0.48 (0.27, 0.70), p-value < 0.001
Confounding: Pregnant age, family income, maternal education level, maternal career, husband's smoking, energy daily intake, daily physical
activity, gestational age, parity, pre-pregnant maternal BMI, gestational diabetes mellitus, infant sex, delivery mode, gestational weight gain
Notes: 17-OHP = 17-hydroxyprogesterone; AGD = anogenital distance; ALSPAC = Avon Longitudinal Study of Parents and Children; AMH = anti-Mullerian hormone;
BMI = body mass index; DHEA = dehydroepiandrosterone; DHEAS = dehydroepiandrosterone sulfate; FSH = follicle stimulating hormone; LH = luteinizing hormone;
mo = months; SHBG = sex hormone binding globulin; T1 = tertile 1; T2 = tertile 2; T3 = tertile 3; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4;
WAZ = weight-for-age z-score; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
Table D-4. Associations Between PFOS Exposure and Female Reproductive Health Effects in Pregnant Women
Reference,
Confidence
Location,
Year(s)
Study
Design
Population,
Ages,
N
Exposure
Matrix,
Levels" (ng/mL)
Outcome
Comparison
Resultsb
Huo et al.
(2020)
High
Mitro et al.
(2020)
High
China, Cohort Females from Plasma Gestational
2013-2016 the Shanghai 9.36 (6.57, 13.69) hypertension,
Birth Cohort Preeclampsia/Eclampsia
Study,
Ages > 20,
N= 3,220
OR per ln-unit
increase in PFOS
Gestational hypertension
0.91 (0.57, 1.43)
Preeclampsia/Eclampsia:
1.24 (0.82, 1.90)
Confounding: Maternal age, pre-pregnancy BMI, parity, parental educational levels, gestational age of blood drawn, fetal sex0
United States, Cohort Females from Plasma
Sex hormone binding Percent difference Sex hormone binding globulin:
Recruitment
1999-2002,
outcome
assessed 3-yr
postpartum
Project Viva,
N = 812
24.7 (18.1, 33.9) globulin (nmol/L)
per log2-unit
increase in PFOS
-0.6 (-7.6, 6.9)
Ages <35:-0.8 (-11.9, 11.7)
Ages >35:-1.5 (-10.0, 7.8)
Confounding: Age, pre-pregnancy BMI, marital status, race/ethnicity, education, income, smoking, parity
D-65
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population,
Ages,
N
Exposure
Matrix,
Levels" (ng/mL)
Outcome
Comparison
Resultsb
Borghese et al.
(2020)
Medium
Canada,
2008-2011
Cohort Females from
the MIREC
study,
Ages > 18,
N = 1,739
Plasma
GM = 4.56 (95%
CI: 4.44, 4.69)
DBP (ininHg). SBP
(mmHg), preeclampsia,
gestational hypertension
Regression
coefficient (DBP,
SBP), OR
(preeclampsia,
gestational
hypertension)
per log2-unit
DBP
Trimester 1 to delivery:
0.47 (0.10,0.85)
Trimester 1: 0.46 (0.01, 0.90)
Trimester 2: 0.33 (-0.10, 0.76)
Trimester 3: 0.66 (0.18, 1.14)
SBP
increase inPFOS or Delivery: 1.19 (0.28, 2.1)
by tertiles Preeclampsia
1.25 (0.84, 1.82)
T2: 1.72 (0.77, 3.82)
T3: 1.55 (0.68,3.49)
Gestational hypertension
1.15 (0.91, 1.45)
T2: 1.43 (0.90,2.29)
T3: 1.38 (0.84,2.23)
Results: Lowest tertile used as the reference group.
Confounding: Maternal age, education, smoking status, pre-pregnancy BMI, parity
Huang et al.
(2019b)
Medium
China,
2011-2012
Cross- Females from
sectional mother-infant
pairs,
N = 687
Plasma
2.38 (1.81,3.23)
Gestational
hypertension,
preeclampsia
OR per increase in
standardized PFOS
Gestational hypertension
0.87 (0.57, 1.34)
Preeclampsia
0.83 (0.52, 1.32)
Comparison: Standardized PFOS calculated by subtracting PFOS concentration from mean PFOS concentration and dividing by the SD.
Confounding: Age, pre-pregnancy BMI, parity, education level
Lyugso et al.
(2014)
Medium
Greenland,
2002-2004
Cross- Pregnant
sectional women from
the
INUENDO
cohort,
N = 1,623
Serum,
8.0 (10th-90th
percentile = 3.6,
25.6)
Menstrual cycle length
(long), irregularity
OR per log-unit
increase in PFOS or
by tertiles
Length
1.1 (0.8, 1.6)
T2: 1.3 (0.8, 2.2)
T3: 1.2 (0.6,2.5)
Irregularity
1.2(0.9, 1.8)
T2: 1.1 (0.6,2.1)
T3: 1.7 (0.8,3.5)
Results: Lowest tertile used as the reference group.
Comparison: Logarithm base not specified.
Confounding: Age at menarche, age at pregnancy, parity, pre-pregnancy BMI, smoking, country
D-66
-------
APRIL 2024
Reference,
Confidence
Population, Exposure
Location, stud, Matriji>
Year(s) DeHgn n ^ (ng/mL)
Outcome
Comparison
Resultsb
Romano et al.
United States, Cohort Females from Serum
Breastfeeding
RR by quartiles of
Termination at 3 mo
(2016)
2003-2006 the HOME 13.9(9.6,18.2)
termination (by 3 mo
PFOS
Q2: 1.08 (0.79, 1.46)
Medium
study,
postpartum),
Q3: 1.39 (1.04, 1.88)
Ages > 18,
Breastfeeding
Q4: 1.32 (0.97, 1.79)
N = 336
termination (by 6 mo
Termination at 6 mo
postpartum)
Q2: 1.17 (0.93, 1.48)
Q3: 1.16 (0.91, 1.48)
Q4: 1.25 (0.98, 1.58)
Results: Lowest quartile used as the reference group.
Confounding: Maternal age at delivery, household income, total weeks of prior breastfeeding, gestational week at blood draw, marital status,
race, parity, maternal serum cotinine during pregnancy, alcohol use during pregnancy
Rylander et al. Sweden, 1989 Case- Females with Serum Preeclampsia
(2020) control or without Primiparous
Medium pre- cases:
eclampsia, 12.9 (Minimum,
Ages 15-44, maximum = 2.15,
N = 876 50.0)
Exposure Levels: [Multiparous cases] Median = 10.9 ng/mL (Minimum, maximum = 1.49, 66.6 ng/mL); [Primiparous controls]
Median = 12.4 ng/mL (Minimum, maximum = 0.52, 54.5 ng/mL); [Multiparous controls] Median = 9.36 ng/mL (Minimum, maximum = 1.13,
47.0 ng/mL)
Results: Lowest quartile used as the reference group.
Confounding: Maternal age, BMI in early pregnancy, maternal smoking in early pregnancy, parity
Timmermann
Denmark,
Cohort Pregnant and
Serum
Total breastfeeding
Regression
Total breastfeeding duration
et al. (2017b)
1997-2000,
postpartum
19.47 (8.67,
duration (months),
coefficient per
-1.4 (-2.1,-0.6)
Medium
2007-2009
females,
28.22)
Exclusive breastfeeding
doubling of PFOS
Exclusive breastfeeding duration
N = 987
duration (months)
-0.3 (-0.6,-0.1)
Confounding: Cohort, maternal age, pre-pregnancy BMI, pregnancy alcohol intake, pregnancy smoking, education, employment, parity
Toft et al.
Denmark
Case- Pregnant
Amniotic fluid
Amniotic fluid levels of
Percent difference
17-OHP
(2016)
1980-1996
control females and
Tertile 2: (Range:
17-OHP (ln-nmol/L),
in median level per
0.15 (0.11,0.20)
Medium
their male
0.8, 1.4)
androstenedione (ln-
1% increase in
T2: 7 (-1, 13)
infants,
nmol/L), DHEAS (ln-
PFOS or by tertiles
T3: 18(11,26)
N = 545
nmol/L), progesterone
p-value for trend < 0.001
(ln-nmol/L), testosterone
Androstenedione
(ln-nmol/L)
0.15 (0.10,0.21)
T2: 8 (0, 17)
OR by quartiles of
PFOS
Q2: 0.81 (0.5, 1.32)
Q3: 1.23 (0.78, 1.93)
Q4: 0.96 (0.60, 1.53)
D-67
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population,
Ages,
N
Exposure
Matrix,
Levels" (ng/mL)
Outcome
Comparison
Resultsb
T3: 17(8, 25)
p-value for trend = 0.001
DHEAS
0.07 (-0.03,0.16)
T2: 5 (-10, 20)
T3: 2 (-14, 17)
p-value for trend = 0.93
Progesterone
0.21 (0.14,0.29)
T2: 11 (0, 23)
T3: 22 (11, 34)
p-value for trend = 0.001
Testosterone
0.16(0.09,0.23)
T2: 9 (-2, 20)
T3: 18(7, 29)
p-value for trend = 0.002
Results: Lowest tertile used as the reference group.
Confounding: Gestational age of amniocentesis, maternal age, smoking (cotinine groups), case or control status.
Wikstrom et al. Sweden,
(2019) 2007-2010
Medium
Cohort Females from Serum
Preeclampsia
the SELMA
study,
Ages 28-35,
N = 1,773
5.39 (3.95,7.61)
OR per log2 1.53 (1.07,2.20)
increase in PFOS or Q4: 2.68 (1.17, 6.12)
by quartiles
Results: Lowest quartile used as the reference group
Confounding: Parity, women's age, body weight, smoke exposure
iVofes:17-OHP = 17-hydroxyprogesterone; BMI = body mass index; DBP = diastolic blood pressure; DHEAS = dehydroepiandrosterone sulfate; GM = geometric mean;
HOME = Health Outcomes and Measures of the Environment; HR = hazard ratio; INUENDO = Biopersistent Organochlorines in Diet and Human Fertility; LIFE = Longitudinal
Investigation of Fertility and the Environment Study; MIREC = Maternal-Infant Research on Environmental Chemicals; OR = odds ratio; Q2 = quartile 2; Q3 = quartile 3;
Q4 = quartile 4; RR = relative risk ratio; SBP = systolic blood pressure; SELMA = Swedish Environmental Longitudinal, Mother and child, Asthma and allergy study; T1 = tertile
1; T2 = tertile 2; T3 = tertile 3; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D-68
-------
APRIL 2024
Table D-5. Associations Between PFOS Exposure and Female Reproductive Health Effects in Non-Pregnant Adult Women
Reference,
Confidence
Location,
Year(s)
Study
Design
Population,
Ages,
N
Exposure
Matrix,
Levels" (ng/mL)
Outcome
Comparison
Resultsb
Ding et al.
(2020)
High
Crawford et al.
(2017)
Medium
United States,
1999-2017
Cohort
Pre-
menopausal
women from
the Study of
Women's
Health Across
the Nation,
Ages 42-52,
N = 1,120
Serum
Sm-PFOS: 7.2
(4.6, 10.8)
n-PFOS: 17.1
(12.2, 24.5)
Natural menopause
HR per doubling
increase in PFOS or
by tertiles
Sm-PFOS:
1.08 (0.99, 1.19)
T2: 1.11 (0.90, 1.37)
T3: 1.27 (1.01, 1.59)
p-value for trend = 0.03
n-PFOS:
1.11 (0.99, 1.23)
T2: 1.06 (0.86, 1.31)
T3: 1.26 (1.02, 1.57)
p-value for trend = 0.03
Results: Lowest tertile used as the reference group.
Confounding: Age at baseline, race/ethnicity, study site, education, parity, BMI at baseline, physical activity, smoking status, prior hormone
use at baseline0
United States, Cohort Females from Serum
Cycle-specific time to Time to pregnancy Cycle-specific time to pregnancy
2008-2009
the Time to
Conceive
Study,
Ages 30-44,
N = 99
9.29 (8.31, 10.38) pregnancy, day-
specific time to
pregnancy, AMH (ln-
ng/mL)
outcomes:
Fecundability ratio
per ln-unit increase in
PFOS
AMH:
Regression
coefficient per ln-unit
increase in PFOS
0.89 (0.49, 1.60)
Day-specific time to pregnancy
0.99 (0.28, 2.32)
AMH
0.07
Confounding: Age, mean cycle length (added for cycle-specific time to pregnancy model)
Kim et al.
Australia, Cross- Females
Follicular fluid Fertilization rate
Regression 2.28 (-0.56, 5.11)
(2020b)
2006-2011 sectional undergoing
Mean= 4.8
coefficient per unit
Medium
fertility
(Minimum,
increase in PFOS
treatment,
Maximum = 0.7,
Ages 23-42,
22.4)
N = 97
Confounding: Age
D-69
-------
APRIL 2024
Reference,
Confidence
Location,
Year(s)
Study
Design
Population,
Ages,
N
Exposure
Matrix,
Levels" (ng/mL)
Outcome
Comparison
Resultsb
Lum et al.
United States,
Cohort
Females from
Serum
Day-specific
Regression
All women:
(2017)
2005-2009
the LIFE
Women
probability of
coefficient by tertiles
T2: 1.0 (0.7, 1.5)
Medium
study, Ages
with < 24-day
pregnancy
ofPFOS
T3: 0.9 (0.6, 1.3)
18-40,
cycle:
N = 483
12.3 (9.7, 17.0)
Women with 25
to 31-day cycle:
12.6 (8.2, 17.6)
Women
with> 32-day
cycle: 11.5 (7.3,
16.9)
Results: Lowest tertile used as the reference group
Confounding: Couple intercourse pattern, female menstrual cycle length, age, BMI, active smoking at enrollment
Tsai et al.
(2015)
Medium
Taiwan, Cross- Females,
2006-2008 sectional Ages 18-30,
N = 265
Serum, Levels of FSH in
8.65 (5.37, 13.29) serum (ln-mlU/mL),
SHBG in serum (ln-
nmol/L)
Means by quartiles of FSH
PFOS Ql: 1.71 (SE = 0.25)
Q2: 1.66 (SE = 0.23)
Q3: 1.71 (SE = 0.25)
Q4: 1.69 (SE = 0.25)
SHBG
Ql: 3.90 (SE = 0.21)
Q2: 3.82 (SE = 0.20)
Q3: 3.89 (SE = 0.22)
Q4: 3.80 (SE = 0.21)
Confounding: Age, BMI, high-fat diet
Wang et al.
(2017)
Medium
China, Case- Females of
2014-2015 control reproductive
age,
N = 335
Plasma,
Cases:
6.40 (4.02, 11.42)
Controls:
6.60 (3.92, 13.54)
Endometriosis-related
infertility
OR by tertiles of
PFOS
T2: 1.11 (0.61, 1.99)
T3: 0.66 (0.36, 1.21)
Confounding: Age, BMI, household income, and education
Notes: AMH = anti-Mullerian hormone; BMI = body mass index; FSH = follicle stimulating hormone; LIFE = Longitudinal Investigation of Fertility and the Environment;
Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; SHBG = sex hormone binding globulin; LI = tertile 1; L2 = tertile 2; L3 = tertile 3.
D-70
-------
APRIL 2024
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D.3 Hepatic
Table D-6. Associations Between PFOS Exposure and Hepatic Effects in Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
Adults
Omoike et al.
(2020)
Medium
United States
2005-2012
Cross-
sectional
Adults from
NHANES,
Age > 20,
N = 6,652
Serum
11.40 (20th-80th
percentile = 5.80-
23.18)
Levels of iron in Percent change per
serum, bilirubin, one percent
and albumin increase in PFOS
Iron concentration in serum
0.05 (0.03, 0.07), p-value < 0.05
Bilirubin
0.03 (0.02, 0.05), p-value < 0.05
Albumin
0.02 (0.02, 0.03), p-value < 0.05
Confounding: Age, sex, race, education, poverty-income ratio, serum cotinine, BMI
Jain (2019)
Medium
United States
2003-2014
Cross-
sectional
Adults from
NHANES,
Ages > 20,
N = 108-3,562
Serum
Levels of ALT
(loglO-IU/L),
AST (loglO-
IU/L)
Regression
coefficient per
loglO-unit increase
in PFOS
ALT,
Non-obese,
GF-1: -0.008
GF-2: 0.011
GF-3A: -0.013
GF-3B/4: -0.088, p-value < 0.01
Obese,
GF-1: 0.048, p-value<0.01
GF-2: 0.005
GF-3A: 0.038
GF-3B/4: 0.0696, p-value < 0.01
AST
Non-obese,
GF-1: -0.013
GF-2: 0.007
GF-3A : -0.015
GF-3B/4: -0.004
Obese,
D-71
-------
APRIL 2024
Reference, Location, _ . Population, Exposure Matrix, „ ^ ^ .
„ .... Design . . . , _ ' Outcome Comparison Select Results
Confidence Years Ages, N Levels (ng/mL)a
GF-1: 0.011
GF-2: -0.013
GF-3A : 0.041, p-value = 0.01
GF-3B/4: 0.023
Liu et al.
(2018b)
Medium
Confounding: Gender, race/ethnicity, smoking status, age, loglO(BMI), diabetes status, hypertension status, fasting time, poverty-income
ratio, survey year, alcohol consumption0
Liu et al.
United States, Controlled Overweight and Plasma
Hepatic fat mass Partial Spearman Hepatic fat mass: 0.11
(2018a)
2004-2007 trial Obese patients Males
correlation
Medium
from the 27.2(19.9-45.2)
coefficient among
POUNDS Lost, Females
baseline PFOS
Age 30-70 22.3 (14.3-34.9)
(ng/ml) and hepatic
study,
fat mass
N= 150
Confounding: age, sex, race, education, smoking status, alcohol consumption, physical activity, menopausal status (women only), hormone
replacement therapy (women only), and dietary intervention groups
United States,
2013-2014
Cross-
sectional
Adults from
NHANES,
Age >18,
N= 1871
Serum
GM = 5.28
(SE= 1.02)
Levels of
albumin (g/dL)
Regression
coefficient per ln-
unit increase in
PFOS
Albumin
0.04, SE = 0.01, p-value < 0.005
Confounding: age, gender, ethnicity, smoking status, alcohol intake, household income, waist circumference, and medications
(antihypertensive, anti-hyperglycemic, and anti-hyperlipidemic agents)
Salihovic et al. Sweden Cohort
Elderly adults in
Plasma
Levels of ALT
Regression
(2018) 2001-2014
Sweden,
Age 70
(|ikat/L)
coefficient per ln-
Medium
Ages 70
13.2 (9.95, 17.8)
unit increase in
N = 1002
Age 75
PFOS
Ages 75
12.6 (7.97, 19.2)
N = 817
Age 80
Age 80
0.57 (5.36, 11.5)
N = 603
0.03 (0.02, 0.04), p-value < 0.0016
Confounding: Sex, LDL and HDL cholesterol, serum triglycerides, BMI, fasting glucose levels, statin use, smoking
D-72
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
Nian et al.
China Cross-
Adults in high Serum Levels of ALT Percent change per
ALT
(2019)
2015-2016 sectional
exposure area in 24.22(14.62-37.19) (ln-U/L), AST 2.71-fold increase
4.1 (0.6,7.7), p-value <0.05
Medium
China, (ln-U/L) in PFOS
Ages 22-96,
AST
N = 1,605
2.0 (-0.3, 4.3)
Confounding: Age, sex,
career, income, education, drink, smoke, giblet, seafood consumption, exercise, BMI
Yamaguchi et
Japan Cross-
Participants Blood Levels of GGT Spearman rank
GGT
al. (2013)
2008-2010 sectional
from the 5.8(3.7-8.8) (IU/L), AST correlation
0.06, p-value = 0.120
Medium
"Survey on the (IU/L), ALT
Accumulation of (IU/L)
AST
Dioxins and
0.11, p-value = 0.010
Other Chemical
Compounds"
ALT
project from
0.12, p-value = 0.004
urban,
agricultural and
fishing areas,
Ages 15-76,
N = 590
Confounding: Age, sex, BMI, regional block, smoking habits, frequency of alcohol intake
Gallo et al. United States Cross- Adults from the Serum
(2012) 2005-2006 sectional C8 Health 20.3 (13.7-29.4)
Medium Project,
Ages > 18 yr,
N = 46, 452
Levels of ALT ALT, GGT, direct
(ln-IU/L), GGT bilirubin:
(ln-IU/L), Direct Regression
bilirubin (In- coefficient per ln-
mg/dL), ALT unit increase in
(IU/L, elevated) PFOS
Elevated ALT:
OR per ln-unit
ALT
0.02 (0.014, 0.026),
p-value <0.001
Direct bilirubin
0.029 (0.024, 0.034),
p-value <0.001
ALT, elevated (OR):
increase in PFOS, Decile 2
1.01 (0.87, 1.16)
or by deciles Decile 3
1.06 (0.91, 1.22)
Decile 4
1.11 (0.96, 1.28)
Decile 5
1.19 (1.04, 1.37)
Decile 6
1.19 (1.04, 1.37)
Decile 7
1.20 (1.04, 1.38)
Decile 8
1.24 (1.08, 1.43)
D-73
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
Lin et al.
(2010)
Medium
van den
Dungen et al.
(2017)
Low
Decile 9: 1.18 (1.02, 1.36)
Decile 10: 1.25 (1.08, 1.44)
p-trend < 0.001
Per ln-unit increase:
1.13 (1.07, 1.18), p-value < 0.001
GGT: No statistically significant
associations
Results: Lowest decile used as the reference group
Confounding: Age, sex, alcohol consumption, socioeconomic status, fasting status, month of blood sample collection, smoking status, BMI,
physical activity, insulin resistance. Additional confounding for ALT, GGT, and direct bilirubin analyses: Race. Additional confounding for
OR analyses: increased serum iron.
United States
1999-2000,
2003-2004
Cross-
sectional
Adults from
NHANES,
Ages > 18 yr,
N = 2,216
Serum
23.50 (15.50-33.80)
Levels of
bilirubin (|iIVI).
GGT (log-U/I),
ALT (U/I)
Regression
coefficient per log-
unit increase in
PFOS
Bilirubin
Separate analysis: -0.30
(SE = 0.24), p-value = 0.223
Composite analysis: -1.06
(SE = 0.27), p-value = 0.001
GGT
Separate analysis: 0.01
(SE = 0.03), p-value = 0.808
Composite analysis: -0.06
(SE = 0.03), p-value = 0.025
ALT
Separate analysis: 1.01
(SE = 0.53), p-value = 0.066
Composite analysis: -0.19
(SE = 0.63), p-value = 0.769
Comparison: Logarithm base not specified.
Confounding: Age, gender, race/ethnicity, smoking status, drinking status, education level, BMI, HOMA-IR, metabolic syndrome, iron
saturation status. Additional confounding for composite analyses: PFHxS exposure, PFNA exposure, PFOA exposure.
The
Netherlands
2015
Cross-
sectional
Men with
habitual eel
consumption,
Ages 40-70,
Serum
40 ng/g wet weight
(15-93)
Levels of ALT,
AST
Standardized
regression
coefficient per unit
increase in PFOS
ALT
0.01 (-0.32,0.34)
AST
D-74
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
N = 37
0.19 (-0.17,0.55)
Confounding: Age, waist-to-hip ratio
Olsen et al.
(2003a)
Medium
United States,
Belgium
1994-2000
Cross-
sectional
Current and
former workers
at two
fluorochemical
production
plants
Male
N = 421,
Female
N = 97,
Regression
analysis
N= 174
Serum
Antwerp Mean
(SD) = 0.96 ppm
(0.97);
Decatur = 1.40 ppm
(1.15)
Levels of ALT
(IU/L),
(IU/L),
(IU/L),
(IU/L)
ALP
AST
GGT
Comparison of
mean outcome by
PFOS quartile
Males
Elevated (p < 0.05) ALT for
employees in Q4 compared with
Qi
Elevated (p < 0.05) ALP for
employees in Q3 and Q4 compared
with Ql
No significant differences in mean
AST or GGT by PFOS exposure
quartile
Females
Elevated (p < 0.05) ALP for
employees in Q4 compared with
Ql and Q2, and in Q3 compared
with Q2
Elevated (p < 0.05) GGT for
employees in Q4 compared with
Ql
No significant differences in mean
ALT or AST by PFOS exposure
quartile
Confounding: Sex
Olsen et al.
United States, Cohort
Male 3M
Antwerp (2000)
Levels of ALT
Regression
ALT
(2001)
Belgium
fluorochemical
Mean (SD):
(ln-IU/L), ALP
coefficient per unit
0.010 (SE = 0.016), p-value = 0.54
Medium
1994-2000
plant workers in
1.16 ppm (1.07);
(ln-IU/L), AST
increase in PFOS
PFOS x Years of observation
Antwerp,
Decatur (2000):
(ln-IU/L), GGT
interaction p-value < 0.001
Belgium and
1.67 ppm (1.39)
(ln-IU/L)
Decatur,
AST
Alabama
0.010 (SE = 0.011), p-value = 0.39
D-75
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
N= 175
PFOS x Years of observation
interaction p-value = 0.79
ALP
0.002 (SE = 0.009), p-value = 0.87
PFOS x Years of observation
interaction p-value = 0.47
GGT
-0.004 (SE = 0.020),
p-value < 0.001
PFOS x Years of observation
interaction p-value = 0.42
Confounding: Years of observation, PFOS x Years of observation, age, BMI, drinks/day, cigarettes/day, location, entry period, baseline years
worked, triglycerides
Olsen et al.
(2012a)
Low
United States
2008-2010
Cohort
3M
fluorochemical
plant employees
and contractors,
N= 179
Serum
Mean change from
baseline,
Employees:
-101.3 ng/mL;
Contractors: 1
Levels of ALT Regression ALT
(IU/L), AST coefficient per unit -0.045 (SD = 0.015),
(IU/L) increase in PFOS p-value = 0.005
AST
-0.007 (SD = 0.009)
Confounding: Sex, age at baseline, BMI at baseline, alcohol consumption at baseline
Rantakokko et Finland Cross-
al. (2015) 2005-2011 sectional
Medium
Morbidly obese
adults
undergoing
bariatric
surgery,
N= 160
Serum
3.2 (5th-95th
percentile: 0.89,
10.3)
Lobular
inflammation
OR per log-unit
increase in PFOS
by level of lobular
inflammation
<2 foci: 0.52 (0.13,2.09)
2-4 foci: 0.14(0.01, 1.66)
Comparison: Logarithm base not specified.
Results: No foci used as the reference group. Foci measured per 200x field.
Confounding: Age, sex, BMI, serum lipids, fasting insulin
Children and Adolescents
Gleason et al.
United States
Cross-
Adolescents
Serum Levels of ALT
Regression
ALT
(2015)
2007-2010
sectional
from NHANES,
11.3(7.0-18.0) (ln-U/L), GGT
coefficient per ln-
(0.013) (-0.009, 0.034)
Medium
Ages > 12,
(ln-U/L), AST
D-76
-------
APRIL 2024
Reference, Location, _ . Population, Exposure Matrix, „ ^ ^ .
„ .... Design . . . , _ ' Outcome Comparison Select Results
Confidence Years Ages, N Levels (ng/mL)a
N = 4,333 (ln-U/L), ALP unit increase in GGT
(ln-U/L) PFOS 0.036 (0.001,0.071)
AST
0.004 (-0.010, 0.018)
ALP
-0.010 (-0.027, 0.007)
Confounding: Age, gender, race/ethnicity, BMI group, smoking, alcohol consumption "if statistically significant associated with both the
exposure and outcome in univariate analysis."
Mora et al.
United States Cohort Children from
Plasma
Levels of ALT
Regression
Prenatal exposure: -0.4 (-1.1, 0.2)
(2018)
1999-2010 Project Viva,
Prenatal exposure:
(U/L)
coefficient per IQR
Mid-childhood exposure: -0.3
Medium
N, prenatal
24.6 (17.9-34.0)
increase in PFOS
(-0.9, 0.2)
exposure = 508,
Mid-childhood
N, mid-
exposure:
childhood
6.2 (4.2-9.7)
exposure = 630
Confounding: Maternal education, prenatal smoking, gestational age at blood draw, and child's sex, race/ethnicity, age at lipids/ALT
measurements
Attanasio
United States Cross- Adolescents
Serum
Levels of ALT
Regression
ALT
(2019)
2013-2016 sectional fromNHANES,
Boys:
(ln-IU/L), AST
coefficient per ln-
Boys,
Medium
Ages 12-19,
GM = 3.68
(ln-IU/L)
unit increase in
(-0.09,0.10)
N, boys = 354
(SE = 0.12)
PFOS or by
Q2: -0.05 (-0.21,0.11)
N, girls = 305
Girls:
quartiles
Q3: 0.07 (-0.05,0.18)
GM = 2.76
Q4:-0.01 (-0.14, 0.13)
(SE = 0.14)
Girls,
0.09 (-0.01,0.18)
Q2: -0.02 (-0.17,0.14)
Q3: 0.01 (-0.11, 0.13)
Q4: 0.11 (-0.02,0.24)
AST
Boys,
-0.02 (-0.11,0.06)
Q2: -0.02 (-0.11,0.08)
Q3: 0.01 (-0.07, 0.10)
D-77
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
Q4: -0.01 (-0.12,0.10)
Girls,
0.07 (0.00, 0.013)
Q2: 0.03 (-0.08,0.14)
Q3: 0.05 (-0.04,0.13)
Q4: 0.12 (0.03, 0.21),
p-value = 0.01
Results: Lowest quartile used as the reference group.
Confounding: Age, race/ethnicity, body weight status, education, poverty-income ratio, exposure to smoking
Khalil et al.
(2018)
Low
United States
2016
Cross-
sectional
Obese children,
Ages 8-12,
N = 48
Serum
2.79 (IQR = 2.10)
Levels of ALT Regression ALT
(U/L), AST coefficient per unit 0.16 (-1.84, 2.15)
(U/L) increase in PFOS
AST
-0.28 (-1.22, 0.65)
Confounding: Age, sex, race
Children and Adolescents - Other Hepatic Outcomes
Jin et al. (2020) United States Cross-
Medium 2007-2015 sectional
Children and
adolescents
diagnosed with
nonalcoholic
fatty liver
disease,
Ages 7-19,
N = 74
Plasma
3.59 (2.35-6.81)
Ballooning,
Grade of
steatosis,
Liver fibrosis,
Lobular
inflammation,
Nonalcoholic
steatohepatitis,
Portal
inflammation
OR per IQR
increase in PFOS
Ballooning
Few balloon cells: 1.11 (0.52,
2.37)
Many cells/prominent ballooning:
1.12 (0.26,4.95)
Grade of steatosis
34%-66% steatosis: 1.37 (0.54,
3.51)
> 66% steatosis: 0.88 (0.39, 1.97)
Liver fibrosis
Mild (stage 1): 1.71 (0.73, 4.03)
Significant (stages 2-4): 1.51
(0.53,4.35)
Lobular inflammation
<2 foci: 0.50 (0.21, 1.22)
2-4 foci: 2.92 (0.92, 9.23)
D-78
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Select Resultsb
Nonalcoholic steatohepatitis
3.32 (1.40, 7.87), p-value < 0.05
Portal inflammation
Mild: 1.85 (0.82, 4.21)
Moderate-to-severe: 2.26 (0.75,
6.79)
Results: For ballooning, none was used as the reference group. For grade of steatosis < 5%—33% was used as the reference group. For liver
fibrosis, none was used as the reference group. For lobular inflammation, no foci used as the reference group. Foci measured per 200x field.
For portal inflammation, none was used as the reference group.
Confounding: Age, sex, ethnicity, and BMI z-score
Notes: ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; GF = glomerular filtration; GGT = y-glutamyltransferase; GM = geometric
mean; HDL = high-density lipoprotein; HOMA-IR = homeostasis model assessment of insulin resistance; HR = hazard ratio; IQR = interquartile range; LDL = low-density
lipoprotein; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; PFHxS = perfluorohexane sulfonic acid; PFNA = perfluorononanoic acid;
POUNDS = Preventing Overweight Using Novel Dietary Strategies; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; SD = standard deviation; SE = standard
error; SMR = standardized mortality ratio; T1 = tertile 1; T2 = tertile 2; T3 = tertile 3.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.4 Immune
Table D-7. Associations Between PFOS Exposure and Vaccine Response in Recent Epidemiological Studies
Reference,
Location, Years
Confidence
„ . „ , ,. . .. Exposure Matrix,
Design Population, Ages, N Jvc|s (ng/mLr
Outcome
Comparison Resultsb
Children
Grandjean et al. Faroe Islands, Cohort
Children followed
Maternal serum
Antibody
Percent change
Child serum
(2012) Denmark
from birth to age 7
(prenatal)
concentrations
per doubling in
Anti-diphtheria,
Medium Recruitment 1997-
Birth and infancy:
Geometric
(log-IU/mL) for
age 5 and
prebooster, age 5
2000,
N = 587
mean = 27.3 (23.2-
tetanus and
maternal PFOS
-16 (-34.9, 8.3)
Follow-up through
Prebooster (mean
33.1)
diphtheria
Anti-diphtheria,
2008
age 5.0)
postbooster, age 5
examination:
Child serum (5 yr)
-15.5 (-31.5, 4.3)
N = 532
Anti-diphtheria, age 7
D-79
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome Comparison
Resultsb
Postbooster (mean
age 5.2)
examination:
N = 456
Age 7 (mean age
7.5) examination:
N = 464
Geometric
mean = 16.7 (13.5—
21.1)
Maternal serum
Anti-diphtheria,
-27.6 (-45.8, -3.3)
Anti-diphtheria, age 7
adjusted for age 5 Ab
-20.6 (-38.2, 2.1)
prebooster, age 5
-38.6 (-54.7, -16.9)
Anti-diphtheria,
postbooster, age 5
-20.6 (-37.5, 0.9)
Anti-diphtheria, age 7
-19.7 (-41.8, 10.7)
Anti-diphtheria, age 7
adjusted for age 5 Ab
-10 (-32.6, 20)
Child serum
Anti-tetanus, prebooster,
age 5
-11.9 (-30, 10.9)
Anti-tetanus, postbooster,
age 5
-28.5 (-45.5, 6.1)
Anti-tetanus, age 7
-23.8 (-44.3, 4.2)
Anti-tetanus, age 7
adjusted for age 5 Ab
-11.4 (-30.5, 12.8)
Maternal serum
Anti-tetanus, prebooster,
age 5
-10.1 (-31.9, 18.7)
Anti-tetanus, postbooster,
age 5
D-80
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
-2.3 (-28.6, 33.6)
Anti-tetanus, age 7
35.3 (-3.9, 90.6)
Anti-tetanus, age 7
adjusted for age 5 Ab
33.1 (1.5,74.6)
Confounding: Age, sex. Additional confounding for postbooster analyses: time since vaccination, booster type. Additional confounding
for year 7 analyses: booster type. Additional confounding for year 7 analyses adjusted for age 5 Ab: booster type, child's specific antibody
concentration at age 5 yr
Granum et al.
Norway
Cohort Mother-infant pairs
Maternal serum with Levels (OD) of
Regression
Rubella antibody
(2013)
1999-2008
from MoBa at 3-yr
three days of
rubella anti-
coefficient per
-0.08 (-0.14, -0.02)
Medium
follow-up
delivery
vaccine antibodies
unit increase
p-value = 0.007
N = 56
5.5 (3.8-7.1)
PFOS
Confounding: maternal allergy, paternal allergy, maternal education, child's gender, and/or age at 3-yr follow-up.
Mogensen et al. Faroe Islands,
Cohort Children aged 5-7 yr Serum
Antibody
Percent change
Anti-diphtheria, age 7
(2015a)
Denmark
N = 443 at age 7
15.5 (12.8-19.2)
concentrations
per doubling of
-30.3 (-47.3, -7.8)
Medium
2002-2007
(log2-IU/mL) for
PFOS
diphtheria or
Anti-tetanus, age 7
tetanus
-9.1 (-32.8, 23)
Confounding: Age, sex, booster type0
Stein et al.
United States,
Cross- Children aged 12-19
Serum
Antibody
Percent change
Measles antibodies
(2016b)
1999-2000, 2003-
sectional years, NHANES
GM = 20.8 (95% CI:
concentrations for
per doubling
All
Medium
2004, 2005-2006
19.1,22.7)
measles,
serum PFOS
-3.5 (-18.3, 14.0)
N = 1,190 (All)
mumps, and
Seropositive
N = 1,152
rubella
-2.9 (-17.3, 13.9)
(Seropositive)
Mumps antibodies
All
-7.4 (-12.8,-1.7)
Seropositive
-5.9 (-9.9,-1.6)
Rubella antibodies
All
-8.4 (-17.9, 2.1)
Seropositive
D-81
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
-13.3 (-19.9,-6.2)
Confounding: Age, sex, race/ethnicity, survey year.
OVAj 1UVV/VUUUV11J ; JU1VVJ J vcu.
Cohort and Children followed up Serum
at7yrandl3yr 13 yr: 6.7 (5.2-8.5)
7 yr: 15.3 (12.4-
N = 505 (13 yr) 19.0)
N = 427 (7 yr)
Grandjean et al. Faroe Islands,
(2017a) Denmark
Medium Enrollment:
1997-2000
cross
sectional
Levels of
diphtheria
antibody (log2-
IU/mL), tetanus
antibody (log2-
IU/mL)
Percent change Diphtheria antibody
per doubling of Age 7: -23.8 (-43.2,
PFOS
p-value = 0.07
Age 13:-8.6 (-27.7,
p-value = 0.454
2.3)
15.6)
Tetanus antibody
Age 7: 30 (-16.1, 101.4)
p-value = 0.24
Age 13: 22.2 (-12.4, 70.3)
p-value = 0.237
Confounding: Sex, age at antibody assessment, booster type at age 5
Grandjean et al. Faroe Islands,
(2017b) Denmark
Medium 1997-2000 and
2007-2009 (year of
birth)
Cross- Infants 2 wk after
sectional expected term date,
followed up at 18 mo
and 5 yr
All: N = 490, 18 mo:
N = 275, 5 yr:
N = 349
Serum
18 mo: 7.1 (4.5-
10.0)
5 yr: 4.7 (3.5-6.3)
Levels of
tetanus antibody
(IU/mL),
diphtheria
antibody (IU/mL)
Percent change 2007-2009 cohort
per doubling of Tetanus antibody
PFOS
Birth: -10.84 (-28.34,
10.94)
p-value = 0.3
18 mo:-7.027 (-21.63,
10.3)
p-value = 0.4
5 yr:-9.076 (-28.1,
14.98)
p-value = 0.43
Diphtheria antibody:
Birth: -14 (-31.59, 8.11)
p-value = 0.20
18 mo: 17.55 (-0.84,
39.34)
p-value = 0.062
5 yr: 17.17 (-8.66, 50.31)
p-value = 0.21
Combined cohort
D-82
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Tetanus antibody
Birth: -10.55 (-24.63,
6.16
p-value = 0.2
18 mo: -7.08 (-21.29,
9.70)
p-value = 0.39
5 yr: -10.52 (-24, 5.35)
p-value = 0.18
Diphtheria antibody
Birth: -24.47 (-36.90,
-9.60)
p-value = 0.002
18 mo: 15.07 (-2.49,
35.79)
p-value = 0.096
5 yr:-1.34 (-17.05,
17.34)
p-value = 0.88
Confounding: Age, sex
Abraham et al.
(2020)
Medium
Berlin, Germany
Enrollment: 1997-
1999
Cross-
sectional
Children, 1 yr old Plasma
All: N = 101, Formula fed:
formula fed: N = 21, mean = 6.8
breastfed: N = 80 (range = 2.8-19.3)
Breastfed:
mean = 15.2
(range = 1.9-34.8)
Levels of Hib
antibody,
tetanus antibody
IgG,
tetanus antibody
IgGl,
diphtheria
antibody
Spearman
correlation
coefficient
Hib antibody: -0.05
Tetanus antibody IgG:
-0.02
Tetanus antibody IgGl:
-0.07
Diphtheria antibody:
-0.02
Confounding: Time since last vaccination
Timmermann et Guinea-Bissau
al. (2020) 2012-2015
Medium
Cohort Infants enrolled at 4- Maternal blood
7 mo old (inclusion), 0.77 (0.53-1.02)
followed up at 9 mo
and 2 yr
Measles antibody
concentration
(mlU/mL)
Percent
difference per
doubling of
PFOS
Inclusion (no measles
vaccination): -13 (-26, 4)
9-mo visit
D-83
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Inclusion: N = 236
9-mo
Unvaccinated
controls: N = 100
Intervention:
N = 133
2-yr
Unvaccinated
controls: N = 100
Intervention: N = 91
Control (no measles
vaccination): -27
(-44, -4)
Intervention (1 measles
vaccination): -21 (-37,
"2)
2-yr visit
Control (1 measles
vaccination): -6 (-25, 18)
Intervention (2 measles
vaccinations): -3 (-20,
J7)
Confounding: Weight and age at inclusion, maternal education, breastfeeding without solids, maternal measles antibody concentration, sex,
and time from vaccination to blood sampling
Timmerman et
al. (2021)
Medium
Greenland
Recruitment:
1999-2005,
Examination:
2012-2015
Cohort and Vaccinated children Maternal serum from Levels (IU/mL) of Percent
cross-
sectional
ages 7-12 yr and
their mothers at
pregnancy
Maternal serum
N = 57
Child serum
N = 169
pregnancy
19.16(15.20-24.06)
Child serum
8.68 (6.52-12.23)
diphtheria and
tetanus antibody
difference per
unit increase in
PFOS
OR per loglO-
unit increase in
PFOS
Diphtheria antibody
Child serum
Percent difference: 9 (-16,
"2)
OR: 1.14(1.04, 1.26)
Maternal serum
Percent difference: 1 (-4,
6)
Tetanus antibody
Child serum
Percent difference: -3 (-8,
3)
Maternal serum
Percent difference: 2 (-3,
6)
Confounding: Area of residence (Nuuk, Maniitsoq, Sisimiut, Ilulissat, Aasiaat, Qeqertarsuaq, Tasiilaq). Additional confounding for percent
difference analyses: duration of being breastfed (<6 mo, 12 mo, >1 yr). Additional confounding for child serum analyses: time since vaccine
booster (only children with known vaccination date were included).
Zeng et al.
(2019b)
China
2013
Cohort Infants from
Guangzhou Birth
Cord blood
3.17 (1.88-4.94)
HFMD antibody
titers (CA16 or
Percent change CA16
or OR (below
D-84
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Low
Cohort Study at birth
and 3 mo
Birth N= 194 (91
girls, 103 boys)
3-mo N = 180 (89
girls, 91 boys)
EV71) in serum of clinical
cord blood or at protection) per
3 mo
doubling of
PFOS
Cord blood: -20.6 (-30.0,
-9.9)
Girls: -14.0 (-27.5, 1.9)
Boys: -24.7 (-37.6, -9.1)
3 mo: -6.9 (-13.9, 0.7)
Girls: -2.8 (-10.9, 6.2)
Boys:-12.2 (-23.7, 1.1)
CA16 below clinical
protection
Cord blood: 1.75 (1.16,
2.63); p-value = 0.007
Girls: 1/43 (0.80, 2.56)
Boys: 1.98 (1.03,3.81)
p-value for interaction by
sex = 0.311
3 mo: 1.71 (1.12,2.60);
p-value = 0.013
Girls: 0.97 (0.88, 1.08)
Boys: 2.29 (1.20, 4.36)
p-value for interaction by
sex = 0.318
EV71
Cord blood: -23.6 (-33.9,
-11.8)
Girls: -23.5 (-37.9, -5.8)
Boys: -23.4 (-37.2, -6.6)
3 mo: -10.6 (-16.9, -3.9)
Girls: -8.6 (-17.1, 0.9)
Boys: -12.2 (-21.3, -1.9)
EV71 below clinical
protection
Cord blood: 1.66 (1.12,
2.45); p-value = 0.011
Girls: 1.48 (0.92,2.37)
D-85
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Boys: 2.01 (1.03, 3.90)
p-value for interaction by
sex = 0.265
3 mo: 2.25 (1.44, 3.51);
p-value < 0.05
Girls: 2.05 (1.11, 3.79)
Boys: 2.35 (1.19,4.65)
p-value for interaction by
sex = 0.579
Outcome: Clinical protection threshold defined as titers > 1:8 in modified cytopathogenic effect assay.
Confounding: Sex, age, parental education, parental occupation, family income, parity, and birth weight
Adults and Adolescents
Looker et al. United States Cohort Adults near water
(2014) Baseline: districts of Ohio and
Medium 2005-2006, West Virginia with
Follow-up: contaminated
2010 drinking water
N = 403
Serum
GM (95% CI) = 8.32
(7.65-9.05)
Influenza
antibodies (titer
ratio and titer rise,
log 10-
transformed):
A/H1N1, A/H3N2,
type B
Regression
coefficient per
loglO-unit
increase, or by
quartiles
Influenza type B titer rise
Per loglO-unit: 0.5 (-0.11,
0.21), p-value = 0.56
Q2: 0.02 (-0.13,0.18),
p-value = 0.76
Q3: -0.03 (-0.19,0.14),
p-value = 0.73
Q4: 0.04 (-0.14,0.21),
p-value = 0.68
Influenza type B titer ratio
Per loglO-unit: 0.05
(-0.09, 0.18),
p-value = 0.52
Q2: 0.004 (-0.14,0.14),
p-value = 0.96
Q3: -0.02 (-0.16,0.12),
p-value = 0.78
Q4: 0.03 (-0.12,0.18),
p-value = 0.71
Influenza A/H3N2 titer
rise
Per loglO-unit: 0.09
(-0.13,0.32),
p-value = 0.42
D-86
-------
APRIL 2024
Reference, Exnosure Matrix. .
Location, Years Design Population, Ages, N T , /• , ,,,, Outcome Comparison Results
Confidence Levels (ng/mL)
Q2: 0.03 (-0.19,0.26),
p-value = 0.78
Q3: 0.18 (-0.06,0.41),
p-value = 0.14
Q4: -0.04 (-0.28, 0.21),
p-value = 0.77
Influenza A/H3N2 titer
ratio
Per loglO-unit: -0.005
(-0.20, 0.19),
p-value = 0.96
Q2: -0.06 (-0.26,0.14),
p-value = 0.56
Q3: 0.02 (-0.18,0.23),
p-value = 0.84
Q4: -0.03 (-0.24,0.19),
p-value = 0.82
Influenza A/H1N1 titer
rise
Per loglO-unit: 0.15
(-0.02, 0.32),
p-value = 0.08
Q2: -0.04 (-0.21,0.14),
p-value = 0.68
Q3: 0.13 (-0.04,0.31),
p-value = 0.14
Q4: 0.10 (-0.09, 0.29),
p-value = 0.30
Influenza A/ H1N1 titer
ratio
Per loglO-unit: 0.10
(-0.11,0.3),
p-value = 0.36
Q2: -0.07 (-0.28,0.13),
p-value = 0.47
D-87
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Q3: 0.03 (-0.18,0.24),
p-value = 0.78
Q4: 0.03 (-0.19,0.26,
p-value = 0.77
Results: Lowest quartile used as reference group
Confounding: Age (cubic spline), gender, mobility, and history of previous influenza vaccination
Pilkerton et al.
(2018)
Medium for
youth
Low for adult
United States Cross- Adults and
1999-2000 sectional adolescents 12 yr
and older
Youths: N= 1,012
Adults:
N = 542 women, 613
men
Serum
Women:
mean = 22.1,
SE = 0.9
Men: mean= 28.1
SE= 1.3
Rubella IgA titers
(log-IU)
Regression
coefficient by
quartiles or per
quartile
increase
Adolescents:
Per quartile increase:
F-value = 1.44,
p-value = 0.251
Adults:
Per quartile increase: F-
value = 3.44,
p-value = 0.030
Women
Q2: 0.05 (-0.34, 0.43)
p-value = 0.81
Q3: 0.04 (-0.51,0.6)
p-value = 0.87
Q4:-0.17 (-1.13, 0.8)
p-value = 0.73
Men
Q2: -0.20 (-0.62, 0.23)
p-value = 0.35
Q3:-0.32 (-0.69, 0.05)
p-value = 0.08
Q4: 0.01 (-0.54, 0.56)
p-value = 0.97
Outcome: Logarithm base not reported
Results: Lowest quartile used as reference group
Confounding: Women: age, ethnicity, BMI, educational level, number of live births; men: age, ethnicity, BMI, educational level
Bulka et al.
(2021)
Medium
Unites States
1999-2000, 2003-
2016
Cross- NHANES Serum
sectional adolescents and 12-19 yr: GM
adults aged 12-49 yr (SE) = 7.54 (0.26)
Persistent
infections of
cytomegalovirus,
Persistent
infections:
Prevalence
Cytomegalovirus
12-19 yr: 0.92 (0.77,
1.09), p-value = 0.36
D-88
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
12-19 yr:N = 3,189
20-49 yr: N = 5,589
20—19 yr: GM
(SE) = 8.67 (0.24)
Epstein-Barr vims,
hepatitis C,
hepatitis E, herpes
simplex virus 1,
herpes simplex
virus 2,
Toxoplasma
gondii, and
Toxocara species;
pathogen burden
ratio per
doubling in
PFOS
Pathogen
burden:
Relative
difference per
log2-unit
increase in
PFOS
20-49 yr: 0.99 (0.92,
1.05), p-value = 0.70
Epstein-Barr virus
12-19 yr: 1.01 (0.96,
1.05), p-value = 0.74
Hepatitis C virus
20-49 yr: 0.96 (0.71,
1.29), p-value = 0.77
Hepatitis E virus
20-49 yr: 1.00 (0.83,
1.20), p-value = 0.99
Herpes simplex virus 1
12-19 yr: 1.05 (0.99,
1.11), p-value = 0.13
20-49 yr: 1.04(1.01,
1.06), p-value < 0.01
Herpes simplex virus 2
20-49 yr: 1.04 (0.99,
1.09), p-value = 0.1
Toxoplasma gondii
12-19 yr: 1.15 (0.90,
1.48), p-value = 0.27
20-19 yr: 1.1 (0.97, 1.26),
p-value = 0.15
Toxocara species
12-19 yr: 1.12(0.66,
1.91), p-value = 0.68
20-19 yr: 1.57(1.26,
1.96), p-value < 0.01
D-89
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Pathogen burden
12-19 yr: 1.30 (1.25, 1.36)
20-49 yr:l. 10 (1.07, 1.12)
Outcome: Pathogen burden defined as the sum of pathogens for which an individual was seropositive (including any pathogens with a
seroprevalence < 1.0%)
Confounding: Age, race/ethnicity, sex, ratio of family income to the federal poverty threshold, educational attainment, serum cotinine
concentrations, and BMI
Lopez-Espinosa United States Cohort and
etal. (2021) 2005-2006,2010 cross-
Medium sectional
Adults from C8HP
2005-2006:
N = 42,782
2010: N = 526
Serum
Levels (ln-cclls/|iL
Counts:
2005-2006: 19.7
or percentage of
Precent
(13.3-28.4)
white blood
difference per
2010: 9.60 (6.10-
cells/lymphocytes)
IQR increase in
14.9)
of white blood
PFOS
cells, neutrophils,
monocytes,
Percentages:
eosinophils,
Difference per
lymphocytes,
IQR increase in
CD3+ T cells,
PFOS
CD3+CD4+ T-
helper cells,
CD3+CD4+CD8+
double positive T
cells, CD3+CD8+
T-cytotoxic cells,
CDS-
CD 16+CD56+
natural killer cells,
CD3-CD19+ B
cells; CD4+/CD8+
ratio
White blood cells, total
2005-2006: -0.55 (-0.84,
-0.26)
2010:0.55 (-1.35,2.49)
Likelihood ratio test
p-value < 0.001 for the
comparison between the
two time periods
Outcome: All cell types reported as cell counts; eosinophils, lymphocytes, monocytes, and neutrophils additionally reported as percentage of
white blood cells; CD3+ T cells, CD3+CD4+ T-helper cells, CD3+CD4+CD8+ double positive T cells, CD3+CD8+ T-cytotoxic cells, CDS-
CD 16+CD56+ natural killer cells, and CD3-CD19+ B cells additionally reported as percentage of lymphocytes
Confounding: Gender, age, smoking, month of sampling, alcohol intake, and educational level
Shih et al.
(2021)
Medium
Faroe Islands,
Denmark
Cohort and
cross-
sectional
Faroe Island
residents at birth, 7,
14, 22, and 28 yr
Cord blood at birth
5.96 (IQR = 3.09)
Levels (IU/mL) of Percent change Hepatitis Type B
hepatitis A per doubling of Cord blood:-23.24
antibody, hepatitis PFOS (-46.77, 10.69)
D-90
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Recruitment:
1986-1987,
Follow-up through
2015
N = 399
Serum
7 yr: 31.89
(IQR = 13.37)
14 yr: 31.29
(IQR = 9.62)
22 yr: 12.55
(IQR = 7.24)
28 yr: 6.85
(IQR =5.29)
B antibody,
diphtheria
antibody, tetanus
antibody; Hepatitis
A antibody signal-
to-cutoff ratio
7-yr serum: -4.65
(-45.87, 67.87)
14-yr serum: 22.17
(-34.09, 126.46)
22-yr serum: 15.26
(-22.88, 72.26)
28-yr serum: 6.12 (-23.36,
46.93)
Hepatitis Type A
Cord blood: 0.11 (-0.36,
0.59)
7-yr serum: 0.21 (-0.54,
0.96)
14-yr serum: -0.14
(-1.01,0.74)
22-yr serum: -0.1 (-0.63,
0.44)
28-yr serum: -0.23
(-0.66, 0.21)
Diphtheria
Cord blood: 28.26 (-5.7,
74.44)
7-yr serum: 5.04 (-36.45,
73.59)
14-yr serum: -3.5
(-42.87, 63.01)
22-yr serum: 5.29 (-21.69,
41.56)
28-yr serum: 6.91 (-14.26,
33.31)
Tetanus
Cord blood: 2 (-20.24,
30.44)
D-91
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
7-yr serum: 8.91 (-25.85,
59.95)
14-yr serum: -19.44
(-48.36, 54.7)
22-yr serum: -9.1
(-28.42, 15.44)
28-yr serum: -2.1
(-17.77, 16.56)
Confounding: Sex
Stein et al. United States Cohort Adults enrolled at
(2016a) 2010 18-49 yr, followed
Low up at day 30
Total population:
N = 75, low baseline
Ab: N = 29
Serum
GM = 5.22 (95% CI:
4.52-6.02)
Anti-A-HINI
antibody response
measured by HAI
or by IHC
RR by tertiles
HAI anti-A-HlNl
antibody
Total population
T2: 2.6 (0.4, 15.1)
T3: 1.3 (0.2,7.3)
p-value for trend = 0.81
Low baseline Ab
T2: 6.7 (1.2, 37.9)
T3: 1.6 (0.3,9.7)
p-value for trend = 0.81
IHC anti-A-HlNl
antibody
Total population
T2: 2.6 (0.9, 7.4)
T3: 2.4 (0.9,6.6)
p-value for trend = 0.12
Low baseline Ab
T2: 4.5 (1,20.3)
T3: 3.1 (1, 10.2)
p-value for trend = 0.13
Results: Lowest tertile used as the reference group.
Confounding: Age, sex, and race/ethnicity
Zeng et al.
China
Cross-
Adults from the
Serum
Hepatitis B surface Regression
HBsAb concentration
(2020)
2015-2016
sectional
Isomers of C8
10.7 (6.82-16.2)
antibody (HBsAb) coefficient or
Linear: -0.51
Low
Health Project
(log-mlU/mL) or OR (HBsAb
(-0.84, -0.18);
N = 605
surface antigen seronegative)
p-value = 0.002
D-92
-------
APRIL 2024
Reference,
Confidence
Location, Years Design Population, Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Branched: -0.31 (-0.7,
0.07); p-value = 0.114
(HBsAg) (mlU- per loglO-unit
mL); HBsAb increase in
seronegative linear or
(<10 mlU/mL) branched PFOS HBsAb seronegative
Linear: 1.96 (1.37, 2.81);
p-value < 0.001
Branched: 1.64 (1.05,
2.56); p-value = 0.03
HBsAg concentration
Linear: 0.74 (-0.02, 1.49);
p-value = 0.056
Branched: 1.08 (0.06,
2.09); p-value = 0.037
Confounding: Age, gender, BMI, career, income, alcohol drinking, smoking, regular exercise; education for HBsAb concentration alone
Zhang et al.
(2023c)
Medium
United States Cross-
2003-2004, 2009- sectional
2010
Children and Serum
adolescents aged 12- Mean 12.44 (7.35-
19 from NHANES 21.90)
N = 819
Levels of rubella Percent change Rubella levels
antibody, mumps per 2.7-fold
antibody, measles increase in
antibody serum PFOA
-8.16 (-13.67,-2.31)
p-value < 0.05
Mumps levels
-2.12 (-8.11,4.25)
Measles levels
-2.38 (-11.94, 8.21)
Confounding: Age, sex, race, income-poverty ratio, BMI, serum cotinine concentrations, survey cycle, and dietary intake of milk.
Notes: Ab = antibody; BMI = body mass index; C8HP = C8 Health Project; CI = confidence interval; GM = geometric mean; HAI = hemagglutinin inhibition; HBsAb = hepatitis
B surface antibody; HBsAg = hepatitis B surface antigen; HFMD = hand, foot, and mouth disease; ICH = immunohistochemistry; IQR = interquartile range; mo = months;
MoBa = Norwegian Mother and Child Cohort Study; NHANES = National Health and Nutrition Examination Survey; OD = optical density; Q2 = quartile 2; Q3 = quartile 3;
Q4 = quartile 4; RR = risk ratio; SE = standard error; T2 = tertile 2; T3 = tertile 3; wk = weeks; yr = year(s).
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D-93
-------
APRIL 2024
Table D-8. Associations Between PFOS Exposure and Infectious Disease in Recent Epidemiological Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Children
Feietal. (2010)
Medium
Denmark,
Recruitment:
1996-2003;
Follow-up: 2008
Cross-sectional
and cohort
Mother-infant
pairs with
follow-up to
llyr(DNBC)
N = 1,400
Maternal plasma Infectious
Mean disease
(range) = 35.3 hospitalizations
(6.4-106.7)
IRR by quartiles Girls
or per quartile
increase in
PFOS
Q2: 1.14 (0.73, 1.79)
Q3: 1.61 (1.05,2.47)
Q4: 1.59 (1.02, 2.49)
Per quartile increase: 1.18 (1.03,
1.36)
Boys
Q2: 0.8 (0.57, 1.13)
Q3: 0.61 (0.42, 0.89)
Q4: 0.77 (0.54, 1.12)
Per quartile increase: 0.90 (0.80,
1.02)
All children
Q2: 0.93 (0.71, 1.21)
Q3: 0.90 (0.68, 1.18)
Q4: 1.0 (0.76, 1.32)
Per quartile increase: 1.0 (0.91,
1.09)
Results stratified by age not
statistically significant
Results: Lowest quartile used as reference group
Confounding: Parity, maternal age, pre-pregnancy BMI, breastfeeding, smoking during pregnancy, socio-occupational status, home density,
child's age, sibling age difference, gestational age at blood drawing, birth year, and birth season
Gourdazi et al. Hokkaido, Cohort Children, early Maternal blood Infectious
(2017b) Japan pregnancy 4.93 (3.67-6.65) diseases, total
Medium 2003-2009 followed up at (including Otitis
4 yr media,
Pneumonia, RS
N = ,1558 (793 virus, Varicella)
boys, 765 girls)
OR by quartiles
Girls
Q2: 1.42 (0.91, 2.23)
Q3: 1.32 (0.86,2.06)
Q4: 1.71 (1.08,2.72)
p-value for trend = 0.036
Boys
D-94
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Q2: 1.45 (0.95, 2.22)
Q3: 1.25 (0.83, 1.91)
Q4: 1.59 (1.03,2.46)
p-value for trend = 0.071
All
Q2: 1.44 (1.06, 1.96)
Q3: 1.28 (0.95, 1.73)
Q4: 1.61(1.18,2.21)
p-value for trend = 0.008
Results: Lowest quartile used as reference group.
Confounding: Maternal age, maternal educational level, number of elder siblings, child sex, breastfeeding period, and smoking during
pregnancy0
Manzano-
Salgado et
(2019)
Medium
al.
Spain,
2003-2008
Cohort
Children ages
1.5, 4, or 7 yr
Age 1.5:
N = 1,188
Age 4:
N = 1,184
Age 7:
N = 1,071
Maternal blood LRTI
6.06 (4.25-7.82)
OR or RR per
log2-unit
increase in
PFOS
OR
1.5 yr: 0.99 (0.83, 1.18)
4 yr: 0.95 (0.79, 1.16)
7 yr: 0.83 (0.57, 1.2)
RR, 1.5-7 yr
All: 0.96 (0.85, 1.09)
Boys: 0.97 (0.81, 1.15)
Girls: 0.94 (0.77, 1.14)
Confounding: OR assessment: Age-at-follow-up of the child; RR assessment: Maternal age at delivery, parity, previous breastfeeding, pre-
pregnancy BMI, region of residence, and country of birth
Ait Bamai
(2020)
Medium
et al. Hokkaido,
Japan
Enrollment:
2003-2012
Cohort Children, early Maternal blood Chicken pox,
pregnancy 5.12 (3.75-7.02) RSV, otitis
followed up at media,
7 yr pneumonia,
wheeze, eczema
N = 2,689
OR or RR per Pneumonia: OR: 1.14 (0.93, 1.38);
ln-unit increase p-value = 0.21
in PFOS
Otitis media: OR: 1 (0.83, 1.2);
p-value = 0.989
Chicken pox: OR: 1.1 (0.91, 1.32);
p-value = 0.348
RSV: OR: 0.72 (0.56, 0.91);
p-value = 0.007
D-95
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Huang et al.
(2020)
Medium
Wheeze: RR: 0.93 (0.82, 1.06);
p-value = 0.255
Eczema: RR: 0.86 (0.76, 0.98);
p-value = 0.02
Confounding: Sex, maternal age, parity, maternal smoking during pregnancy, BMI pre-pregnancy, annual household income during
pregnancy, duration nursing, and presence of siblings
China
Recruitment:
2011-2013,
Follow-up at
5 yr
Cohort
Children ages Cord blood Respiratory tract Recurrent
1-5 yr
N= 344 (182
boys, 162 girls)
2.44 (1.74-3.22) infections (total
and recurrent)
respiratory tract
infections: OR
for > 75th
percentile
vs. < 75th
percentile PFOS
Total respiratory tract infections
-0.64 (-4.38, 3.1), p-value = 0.738
Recurrent respiratory tract
infections
0.91 (0.51, 1.65), p-value = 0.762
Results stratified by age and sex not
statistically significant
Confounding: Infant sex, maternal age, maternal education level, birth weight
Grandjean et al. Denmark Cross-sectional Adults, ages 30-Plasma COVID-19 OR per unit
(2020) 2020 70 yr, with 4.86 (2.85-8.29) severity increase in
Medium known SARS- PFOS
CoV-2 infection
N = 323
Covid-19 severity
0.97 (0.92, 1.02)
Covid-19 severity (hospitalization
vs. no hospitalization)
0.96 (0.84, 1.10)
Covid-19 severity (intensive care
unit and/or deceased vs.
hospitalization)
1.08 (0.94, 1.24)
Confounding: Age, sex, kidney disease, other chronic disease, national origin, place of testing, and days between blood sampling and
diagnosis
Dalsager et al.
Denmark Cohort
Pregnant
Maternal serum Hospitalization
Hazard ratio per Any infection
(2021)
Recruitment:
women and
7.52 (0.49-27.5) from infection
log2-unit 1.23 (1.05, 1.44)
Medium
2010-2012,
their children
(any infection,
increase in Boys: 1.36 (1.10, 1.67)
from the OCC,
upper
PFOS Girls: 1.04 (0.85, 1.28)
D-96
-------
APRIL 2024
Reference,
Confidence
Location,
Years
p . . Exposure
Design A^s n" Matrix, Levels Outcome Comparison
g ' (ng/mL)a
Resultsb
Follow-up until
2015
followed up to
4 yr
N = 1,472
respiratory tract,
lower
respiratory tract,
gastrointestinal,
other)
Upper respiratory infection
1.25 (0.97, 1.61)
Lower respiratory infection
1.54 (1.11,2.15)
Gastrointestinal infection
0.77 (0.46, 1.29)
Other infection
1.17 (0.98, 1.40)
Confounding: Maternal age, parity, maternal educational level, child sex, child age
Ji et al. (2021) China Case-control Adults Urine COVID-19 ORperlog2-SD COVID-19
Medium 2020 N=160 infection change in PFOS 1.94 (1.39,2.96)
Controls: 42.4
(25.5-61.3) ng/g
creatinine
Cases: 67.6
(41.0-96.5) ng/g
creatinine
Confounding: Age, gender, BMI, diabetes, cardiovascular diseases, and urine albumin-to-creatinine ratio
Wang et al.
China Cohort
Pregnant Maternal serum Common cold,
OR per loglO-
Common cold
(2022b)
Recruitment:
women and at delivery bronchitis/pneu
unit increase in
OR: 1.86 (0.53, 6.50),
Medium
2010-2013,
their children at 4.58(3.31-6.14) monia, diarrhea
PFOS
p-value = 0.334
Follow-up after
1 yr from
IRR: 1.24 (0.76, 2.02),
1 yr
LWBC
IRR per log 10-
p-value = 0.382
N = 235
unit increase in
PFOS
Bronchitis/pneumonia
OR: 1.54 (0.30, 7.78),
p-value = 0.602
IRR: 0.76 (0.23, 2.46),
p-value = 0.644
Diarrhea
D-97
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Dalsager et al.
(2016)
Low
OR: 2.6 (0.67, 10.09),
p-value = 0.167
IRR: 1.89 (1.08, 3.32),
p-value = 0.027
Confounding: Maternal age, pre-pregnancy BMI, smoking during pregnancy, maternal education level, and parity
Odense,
Denmark
2010-2012
Cohort
Children,
pregnancy
followed up at
l^t yr
N = 346
Maternal serum
8.07
(range = 2.36-
25.10)
Fever, cough,
nasal discharge,
diarrhea,
vomiting
OR (of
proportion of
days with
symptoms) by
tertiles
Fever
T2: 1.41 (0.81,2.44)
T3: 2.35 (1.34, 4.11);
p-value < 0.05
Cough
T2: 1.16(0.67,2.01)
T3: 1.03 (0.59, 1.79)
Nasal discharge
T2: 1.11 (0.65, 1.93)
T3: 1.07 (0.62, 1.85)
Diarrhea
T2: 0.89 (0.51, 1.56)
T3: 1.04 (0.59, 1.82)
Vomiting
T2: 1.47 (0.86,2.54)
T3: 0.78 (0.45, 1.35)
Results: Lowest tertile used as reference group.
Confounding: Maternal age, maternal educational level, parity, and child age.
Impinen et al.
Oslo, Norway Cohort, Nested
Infants followed
Cord blood
Common cold
Regression
Common cold 0-2 yr
(2018)
Recruited 1992- case-control
up at 2 and
5.2 (4.0-6.6)
episodes from 0
coefficient per
-0.03 (-0.08, 0.01)
Low
1993, followed
10 yr of age
to 2 yr, LRTI
log2-unit
p-value = 0.173
up for 10 yr
N = 641
episodes from 0
increase in
to 10 yr
PFOS
LRTI 0-10 yr
0.5 (0.42, 0.57)
p-value < 0.0001
Confounding: Child sex
D-98
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Impinen et al.
(2019)
Low
Oslo, Norway Cohort
Enrollment:
1999-2008
Pregnant
Maternal blood Common cold, OR per IQR
women and 12.87 (9.92-
their infants 16.63)
followed up at 3
and 7 yr
0-3 yr:
N = 1,207
6-7 yr: N = 921
bronchitis/pneu increase in
monia, throat PFOS
infection with
strep,
pseudocroup,
ear infection,
diarrhea/gastric
flu, urinary tract
infection
Common cold
0-3 yr: 0.94 (0.92, 0.97);
p-value < 0.05
Bronchitis/pneumonia
0-3 yr: 1.20(1.07, 1.34);
p-value < 0.05
6-7 yr: 0.77 (0.50, 1.19)
Throat infection with strep
0-3 yr: 0.90 (0.78, 1.04)
Other throat infections
0-3 yr: 0.90 (0.81, 1.01)
Pseudocroup
0-3 yr: 1.07 (0.96, 1.20)
Ear infection
0-3 yr: 0.88 (0.82, 0.94);
p-value < 0.05
6-7 yr: 1.13 (0.92, 1.40)
Diarrhea/gastric flu
0-3 yr: 0.98 (0.93, 1.03)
6-7 yr: 1.12(1.01, 1.24)
Urinary tract infection
0-3 yr: 0.78 (0.70, 0.87);
p-value < 0.05
6-7 yr: 0.91 (0.63, 1.31)
Confounding: Maternal age, maternal BMI, maternal education, parity, smoking during pregnancy
Kvalem et al.
Norway
Cohort and
Children, 10 yr,
Serum
Common cold,
Colds: OR
Colds, 10-16 yr
(2020)
Enrollment:
cross-sectional
all: 378, boys:
LRTI
(reference: 1-2
3-5 colds
Low
1992-1993
193, girls: 185
All: 19.4 (IQR:
colds)
All: 1.26 (0.34,4.55)
9.23)
p-value = 0.73
D-99
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Follow-up:
2002-2009
Children, 10- Boys: 21.7
16 yr, all: 375, (IQR: 8.86)
boys: 191, girls: Girls: 17.52
184 (IQR: 8.02)
Children, 16 yr,
all: 330, boys:
170, girls: 160
LRTI: RR per
IQR increase in
PFOS
Boys: 2.54 (0.38, 17.3)
p-value = 0.34
Girls: 0.86 (0.16,4.75)
p-value = 0.86
> 5 colds
All: 1.16 (0.33,4.07)12.54
p-value = 0.82
Boys: 1.99 (0.3, 13.2)
p-value = 0.48
Girls: 1.07 (0.21, 5.45)
p-value = 0.93
LTRI
10-16 yr
All: 1.34 (1.17, 1.55)
p-value < 0.001
Boys: 1.33 (1.26, 1.39)
p-value < 0.001
Girls: 1.23 (0.91, 1.66)
p-value = 0.17
16 yr
All: 0.82 (0.4, 1.69)
p-value = 0.6
Boys: 0.62 (0.22, 1.78)
p-value = 0.38
Girls: 1.11 (0.41, 3)
p-value = 0.84
Confounding: Puberty status at 16 yr, mother's education, physical activity level at 16 yr
Notes: BMI = body mass index; CI = confidence interval; DNBC = Danish National Birth Cohort; IQR = interquartile range; IRR = incidence rate ratio; LRTI = lower respiratory
tract infection; LWBC = Laizhou Wan Birth Cohort; OCC = Odense Child Cohort; OR = odds ratio; RR = risk ratio; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4;
RSV = respiratory syncytial virus; SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2; SD = standard deviation; SE = standard error; T2 = tertile 2; T3 = tertile 3;
yr = year(s).
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D-100
-------
APRIL 2024
Table D-9. Associations Between PFOS Exposure and Asthma in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Dong et al. Taiwan, 2009- Case-control
(2013) 2010 andcross-
Medium sectional
Children from
GBCA with
(cases) or
without
(controls)
asthma, ages
10-15 yr,
N = 231 (cases),
N = 225
(controls)
Serum
Cases: 33.9
(19.6-61.1)
Controls: 28.9
(14.1-43.0)
Asthma,
Asthma Control
Test score,
asthma severity
score, IgE in
serum (IU/mL),
AEC (106/L),
ECP in serum
(lig/L)
Asthma: OR by
quartiles of
PFOS
Asthma Control
Test score,
asthma severity
score, IgE,
AEC, ECP:
mean values by
quartiles
Asthma
Q2: 1.96 (1.11,3.47)
Q3: 1.32 (0.75,2.32)
Q4: 2.63 (1.48, 4.69)
p-trend = 0.003
IgE
Ql:517.9 (336.7, 699.2)
Q2: 686.2 (501.3, 871.1)
Q3: 658.1 (475.2, 841.1)
Q4: 877.3 (695.2, 1,059.5),
p-value < 0.05
p-trend = 0.008
AEC
Ql: 329.4 (255.8, 403.0)
Q2: 368.6 (293.9, 443.3)
Q3: 431.3 (358.1, 504.6)
Q4: 453.4 (379.4, 527.3)
p-trend = 0.009
ECP
Ql: 25.9 (10.4, 41.3)
Q2: 37.4 (21.9, 52.8)
Q3: 43.5 (27.5, 59.4)
Q4: 62.4 (46.3, 78.4),
p-value < 0.05
p-trend = 0.001
Asthma severity score
Ql: 3.33 (2.36,4.31)
Q2: 4.18 (3.19, 5.17)
Q3: 4.49 (3.52,5.45)
Q4: 4.57 (3.61,5.54)
p-trend = 0.045
D-101
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Asthma Control Test score: trends
across quartiles not statistically
significant
Results: Lowest quartile used as reference group
Confounding: age, sex, BMI, parental education, ETS exposure, and month of survey
Humblet et al. Unites States, Cross-sectional Adolescents,
(2014) 1999-2008 ages 12-19 yr
Medium old from
NHANES
N = 1,877
Serum
Never asthma
16.8 (10.8-26.2)
Ever asthma
17.0 (10.8-25.8)
No current
asthma
16.8 (10.8-26.2)
Current asthma
16.7 (10.3-25.3)
No wheezing
16.8 (10.8-26.2)
Wheezing
17.2 (10.9-25.4)
Asthma, wheeze
OR per Ever asthma
doubling in Per doubling: 0.88 (0.74, 1.04),
PFOS or per p-value = 0.13
unit increase in Per unit increase: 0.99 (0.98, 1.0),
PFOS p-value = 0.07
Current asthma
Per doubling: 0.88 (0.72, 1.09),
p-value = 0.24
Per unit increase: 0.99 (0.98, 1.01),
p-value = 0.34
Wheeze
Per doubling: 0.83 (0.67, 1.02),
p-value = 0.08
Per unit increase: 0.99 (0.98, 1.01),
p-value = 0.37
Exposure: No wheezing defined as no wheezing in the past 12 mo. Wheezing defined as history of wheezing in the past 12 mo.
Confounding: Sex, smoking, age, race/ethnicity, survey cycle, poverty-income ratio, health insurance
Smit et al.
(2015)
Medium
Ukraine and
Greenland,
Exposure:
2002-2004,
Outcome:
2010-2012
Cohort
Mother-child Maternal blood Asthma
pairs with
follow-up when Ukraine:
the children GM = 4.88 (PS-
were 5-9 yr of P95: 2.34-9.94)
age, N = 1,024 Greenland:
GM = 20.6 (P5-
P95: 10.2-49.6)
OR per SD
increase in
PFOS
Asthma ever (combined): 0.86
(0.67, 1.10)
Ukraine: 0.75 (0.39, 1.42)
Greenland: 0.88 (0.67, 1.15)
Confounding: Maternal allergy, smoking during pregnancy, education level, maternal age, child sex, child age at follow-up, gestational age at
blood sample, parity, breastfeeding, and birthweight0
D-102
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Stein et al.
(2016b)
Medium
United States,
1999-2000,
2003-2004,
2005-2006
Cross-sectional
Children aged
12-19 years,
NHANES
N = 638
Serum
GM = 20.8
(95% CI: 19.1,
22.7)
Asthma and
wheeze
OR [per
IQR(lnPFOS)
increase (0.76
ln-ng/mL)]
Asthma
1.20 (0.88, 1.63)
Wheeze
0.76 (0.45, 1.29)
Impinen et al.
(2018)
Medium
Confounding: Age, sex, race/ethnicity, survey year; for Wheeze: age, gender, race, weight status, serum cotinine.
Oslo, Norway,
1992-2002
Cohort, Nested
case-control
Infants followed Cord blood
up at 2 and 5.2(4.0-6.6)
10 yr of age,
N = 641
Asthma OR per log2- Current asthma (10 yr):
unit increase in 1.14 (0.84, 1.54); p-value = 0.392
PFOS Asthma ever (10 yr):
1.32 (0.89,1.97); p-value = 0.167
Confounding: Sex
Beck et al.
(2019)
Medium
Denmark,
Enrollment:
2010-2012
Cohort
Children, early Maternal blood Wheeze, self- OR per
pregnancy to
5 yr
N = 970 (507
boys, 363 girls)
7.73 (5.68-
10.44)
reported asthma,
doctor-
diagnosed
asthma
doubling in
maternal serum
PFOS
Wheeze
All: 1.01 (0.79, 1.30)
Boys: 1.02 (0.74, 1.39)
Girls: 1.01 (0.67, 1.52)
Self-reported asthma
All: 1.22 (0.65, 2.28)
Boys: 2.39 (0.92, 6.21)
Girls: 0.67 (0.29, 1.53)
Doctor-diagnosed asthma
All: 0.83 (0.52, 1.31)
Boys: 0.74 (0.46, 1.20)
Girls: 1.60 (0.46, 5.59)
Confounding: Parity, maternal education level, maternal pre-pregnancy BMI, asthma predisposition, child sex
Gaylord et al. New York City, Case-control
(2019) NY
Medium 2014-2016
Children with Serum
(cases) or
without
(controls)
asthma aged
13-22,
N = 118 (cases),
Cases: 3.72
(Range: 1.01—
14.2)
Controls: 2.75
(Range: 0.60-
27.8)
Asthma OR per log-unit 0.89(0.45,1.76)
increase in
PFOS
D-103
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
N = 169
(controls)
Comparison: Logarithm base not specified.
Confounding: Sex, race/ ethnicity, age, BMI, tobacco smoke exposure
Impinen et al.
(2019)
Medium
Manzano-
Salgado et al.
(2019)
Medium
Oslo, Norway, Cohort
Enrollment:
1999-2008
Pregnant
women and
their infants
(followed to age
7),
N = 921
Maternal blood Asthma
12.87 (9.92-
16.63)
OR per IQR
increase in
PFOS
Current asthma:
Total: 1.11 (0.72, 1.69);
p-value = 0.643
Boys: 1.17 (0.64,2.15);
p-value = 0.616
Girls: 1.03 (0.56,1.91);
p-value = 0.927
Ever asthma:
Total: 0.93 (0.68, 1.26);
p-value = 0.631
Boys: 0.94 (0.63, 1.40);
p-value = 0.744
Girls: 0.92 (0.57, 1.49);
p-value = 0.745
Confounding: Maternal age, maternal BMI, maternal education, parity, smoking during pregnancy
Spain,
2003-2008
Cohort
Children, 4 yr,
N = 1,184
7 yr, N = 1,068
Maternal blood Asthma
6.06 (4.52-7.82)
OR or RR per
log2-unit
increase in
maternal PFOS
4-yr follow-up: OR = 0.72 (0.45,
1.13)
7-yr follow-up: OR = 0.84 (0.57,
1.25)
4 and 7 yr
Girls: RR = 0.68 (0.38. 1.22)
Boys: RR = 0.91 (0.58, 1.41)
Confounding: OR assessment: Age at follow-up of the child; RR assessment: Maternal age at delivery, parity, previous breastfeeding, pre-
pregnancy BMI, region of residence, and country of birth
Zeng et al.
(2019a)
Medium
Shanghai,
China,
2012-2015
Cohort
Enrolled in
pregnancy,
follow-up at
5 yr
Cord blood
Boys: 2.49
(1.81-3.51)
Girls: 2.38
(1.73-3.13)
Asthma
OR per loglO-
unit increase in
PFOS
All: 1.49 (0.29, 7.54),
p-value = 0.63
Boys: 4.69 (0.51,42.77),
p-value = 0.17
D-104
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
N = 358 (187
boys, 171 girls)
Girls: 0.17(0.01,4.15),
p-value = 0.27
Confounding: Child weight at age 5, gestational age, breastfeeding during the first 6 mo, maternal education, maternal pre-pregnancy BMI,
and annual household income
Jackson-Browne NHANES,
et al. (2020) United States,
Medium 2013-2014
Cross-sectional
Children, ages
3-11 yr,
N = 607
Serum
GM = 3.7 (2.6-
5.5)
Asthma
OR per ln-SD 1.2(0.8,1.7)
increase in
PFOS
By age:
3-5 yr: 1.7 (1.0,3.0)
6-1 lyr: 1.1 (0.7, 1.6)
p-value for interaction by
age = 0.03
By sex:
Females: 1.1 (0.7, 1.7)
Males: 1.2 (0.8, 2.0)
p-value for interaction by
sex = 0.82
By race/ethnicity:
White, non-Hispanic: 1.4 (0.8, 2.6)
Black, non-Hispanic: 1.3 (0.8, 2.2)
Hispanic: 1.3 (0.8, 2.0)
Other: 1.1 (0.7, 1.7)
p-value for interaction by
race = 0.35
Confounding: Sex, age, race/ethnicity, serum cotinine, poverty to income ratio
Kvalem et al.
(2020)
Medium
Norway
Enrollment:
1992-1993;
Follow-up:
2002-2009
Cohort and
Children, 10 yr Serum
Asthma
cross-sectional N = 378
(193 boys,
185 girls)
Children, 10-
16 yr
N = 375
(191 boys,
184 girls)
All: 19.4 (IQR:
9.23)
Girls: 17.52
(IQR: 8.02)
Boys: 21.7
(IQR: 8.86)
RR per IQR
increase in
PFOS
10 yr
All: 1.01 (0.86, 1.19)
Boys: 1.06 (0.89, 1.26)
Girls: 0.76 (0.52, 1.12)
10-16 yr
All: 0.94 (0.74, 1.20)
Boys: 0.96 (0.71, 1.31)
Girls: 0.85 (0.54, 1.31)
D-105
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Children, 16 yr
N = 375 (191
boys, 184 girls)
16 yr
All: 1.00 (0.79, 1.27)
Boys: 1.01 (0.76, 1.36)
Girls: 0.91 (0.60, 1.38)
Confounding: 10 yr: Age at follow-up, physical activity, mothers' education; 16 yr: BMI at 16 yr, puberty status at 16 yr, mothers' education,
physical activity level at 16 yr
Huang et al.
(2020)
Medium
China
Recruitment:
2011-2013,
Follow-up at
5 yr
Cohort
Children ages
1-5 yr
N = 344 (182
boys, 162 girls)
Cord blood IgG (ng/mL),
2.44(1.74-3.22) IgE (ng/mL)
Regression
coefficient per
loglO-unit
increase in
PFOS
IgG
-0.01 (-0.06, 0.04),
p-value = 0.643
IgE
-0.04 (-0.35, 0.27),
p-value = 0.805
Results stratified by age and sex not
statistically significant
Confounding: Infant sex, maternal age, maternal education level, birth weight
Xu et al.
United States Cross-sectional Adults from
Serum Fractional
Percent change
Fractional exhaled nitric oxide
(2020a)
2007-2012 NHANES, ages
Mean exhaled nitric
per doubling of
2.03 (0.11, 4.00), p-value < 0.05
Medium
20-79 yr
(SD) = 13.33 oxide (ppb)
PFOS, or by
T2: 1.80 (-1.53,5.25)
N = 3,630
(12.92) (ig/L
tertile
T3: 5.02 (1.40, 8.77),
p-value < 0.01
p-trend < 0.006
Results: Lowest tertile used as reference group
Confounding: Age, sex, race/ethnicity, BMI, annual family income, education level, serum cotinine, recent respiratory symptom, and
smoking status
Zhou et al.
Taiwan Case-control Children with
Serum Asthma
Asthma:
Asthma: Increased PFOS among
(2017b)
2009-2010 (cases) or
Case boys: 36.9
Comparison of
asthmatics, p-value = 0.002
Low
without
(22.6-67.8)
PFOS
(controls)
Case girls: 28.2
distributions
asthma ages 10-
(13.9-46.0)
(Wilcoxon rank-
15 from the
Control boys:
sum test)
GBCA
29.9 (13.0-43.8)
N = 456
Control girls:
Case boys: 158
28.8 (14.8-42.6)
D-106
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Case girls: 73
Control boys:
102
Control girls:
123
Confounding: Cases and controls were matched on age and sex
Zhu et al.
Taiwan Case-control Children with Serum Asthma
OR for highest
Boys: 4.24 (1.81, 9.42); p-value for
(2016)
2009-2010 (cases) or Case boys:
vs. lowest
trend = 0.001
Low
without 36.94
quartiles of
Girls: No statistically significant
(controls) Case girls:
PFOS
associations or trends
asthma ages 10- 28.16
15 from the Control boys:
GBCA 26.24
N = 456 Control girls:
Case boys: 158 30.12
Case girls: 73
Control boys:
102
Control girls:
123
Confounding: Age, BMI, parental education, environmental tobacco smoke, parental asthma, month of survey
Zhou et al.
Taiwan Case-control Children with Serum Asthma
OR per unit
Females with high testosterone:
(2017c)
2009-2010 (cases) or Cases: 33.94
increase in
0.58 (0.36, 0.93)
Low
without (19.59-61.10)
PFOS
Females with low testosterone: 1.32
(controls) Controls: 28.91
(0.88, 1.99)
asthma ages 10- (14.06-42.02)
p-value for interaction by low/high
15 from the
testosterone = 0.010
GBCA
N = 456
Males with high testosterone: 1.04
Case boys: 158
(0.87, 1.25)
Case girls: 73
Males with low testosterone: 2.54
Control boys:
(1.40, 4.60)
102
p-value for interaction by low/high
Control girls:
testosterone = 0.005
123
D-107
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Sexes evenly
divided into
high/low
hormone
classifications
Females with high estradiol: 1.25
(0.84, 1.86)
Females with low estradiol: 0.65
(0.42, 0.99)
p-value for interaction by low/high
estradiol = 0.026
Males with high estradiol: 1.25
(0.90, 1.72)
Males with low estradiol: 1.06
(0.87, 1.30)
p-value for interaction by low/high
estradiol = 0.407
Confounding: Age, sex, BMI, parental education, environmental tobacco smoke exposure, physical activity, month of survey
Timmermann et
al. (2017a)
Low
Faroe Islands,
recruitment:
1997-2000
Cohort
Pregnant
women and
infants, follow-
up at ages 5, 7,
and 13 yr,
N = 559
Maternal serum Asthma
Prenatal/At
birth: 27.4
(23.3-33.3)
Age 5/7: 16.8
(13.5-21.1)
Age 13: 6.7
(5.2-8.5)
OR per Asthma (age 5): Total: 1.21 (0.64,
doubling of 2.29)
maternal PFOS No MMR vaccine before age 5:
3.96 (0.55, 28.39)
Yes MMR vaccine before age 5:
0.98 (0.55, 1.76)
Asthma (age 13):
Total: 0.69 (0.43, 1.09)
No MMR vaccine before age 5:
5.41 (0.62, 47.16)
Yes MMR vaccine before age 5:
0.94 (0.51, 1.74)
Confounding: Family history of eczema in children, allergic eczema, and hay fever, maternal pre-pregnancy BMI, maternal smoking during
pregnancy, sex, duration of breastfeeding, fish intake at age 5, number of siblings, daycare attendance at age 5, birth weight, and family
history of chronic bronchitis/asthma
Averina et al.
Norway
Cohort
Adolescents in
Serum
Asthma self-
OR by quartiles
TFF1
(2019)
2010-2011
their first year
Girls: GM=5.8
reported,
of PFOS
Q2: 1.51 (0.72, 3.18)
Low
of high school
(IQR = 2.7)
doctor-
Q3: 2.75 (1.36, 5.57);
from TFF1 and
Boys: GM = 6.8
l diagnosed
p-value = 0.005
TFF2
(IQR = 3.0)
D-108
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
N = 675
Q4: 2.11 (1.02, 4.37);
p-value = 0.044
p-value for trend = 0.02
TFF2
Q2: 2.00 (0.96,4.15);
p-value = 0.064
Q3: 2.56 (1.24, 5.30);
p-value = 0.011
Q4: 1.43 (0.65, 3.12)
Trend not statistically significant
Results: Lowest quartile used as reference group.
Confounding: Sex, age, BMI, physical activity, unemployment/disability of parents, living with adoptive parents, fish intake
Workman et al. Canada
(2019) 2010-2012
Low
Cohort
Mothers and
their infants
N = 85
Maternal
plasma
2.2 (Range:
0.18-21)
Recurrent
wheezing
episodes
Difference in
prenatal PFOS
levels for
wheezing vs. no
wheezing
(Mann-Whitney
test)
No significant differences
Confounding: None reported
Notes: AEC = absolute eosinophil counts; BMI = body mass index; CI = confidence interval; ECP = eosinophilic cationic protein; ETS = environmental tobacco smoke;
GBCA = Genetic and Biomarker study for Childhood Asthma; GM = geometric mean; IgE = immunoglobulin E; IQR = interquartile range; NHANES = National Health and
Nutrition Examination Survey; MMR = measles, mumps, rubella; mo = months; NY = New York; OR = odds ratio; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; RR = risk
ratio; SD = standard deviation; T2 = tertile 2; T3 = tertile 3; TFF1 = Tromse Fit Futures; TFF2 = Tromse Fit Futures 2; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D-109
-------
APRIL 2024
Table D-10. Associations Between PFOS Exposure and Allergies in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Study Design PoPulat>«n'
J b Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Children
Wang et al.
(2011)
Medium
Taiwan
2004
Cohort and
cross-sectional
Pregnant
women and
their children at
age 2
N = 244 (133
boys, 111 girls)
Cord blood
5.50 (0.11-
48.36)
Atopic
dermatitis, IgE
levels (log-
KU/L)
Atopic
dermatitis:
OR by quartiles
of PFOS
IgE:
Regression
coefficient per
ln-unit increase
in PFOS
Atopic dermatitis
Q2: 0.68 (0.20, 2.3)
Q3: 2.34 (0.86, 6.41)
Q4: 2.19 (0.78, 6.17)
IgE in cord blood at birth
All: 0.161 (SE = 0.147),
p-value = 0.017
Boys: 0.175 (SE = 0.179),
p-value = 0.053
Girls: 0.151 (SE = 0.165),
p-value = 0.616
IgE in serum at age 2
All: 0.251 (SE = 0.179),
p-value = 0.147
Boys: 0.359 (SE = 0.255),
p-value = 0.238
Girls: 0.095 (SE = 0.325),
p-value = 0.723
Results: Lowest quartile used as reference group.
Confounding: Gender, gestational age, maternal age. Additional confounding for atopic dermatitis: maternal history of atopy, duration of
breast feeding, pre-natal ETS exposure. Additional confounding for IgE: parity.
Okada et al.
(2012)
Medium
Japan
2002-2005
Cohort
Pregnant
women and
children from
the Hokkaido
Study on
Environment
and Children's
Health; follow-
up at 18 mo
N = 343
Maternal serum
5.2 (3.4-7.2)
Food allergy,
eczema, otitis
media, and
wheezing
IgE levels
(loglO-IU/mL)
OR and
regression
coefficients per
loglO-unit
increase in
PFOS
Food allergy
3.72 (0.81, 17.10)
Eczema
0.87 (0.15,5.08)
Otitis media
1.40 (0.33, 6.00)
Wheezing
D-110
-------
APRIL 2024
Reference, Location, study Design Population, Matrix, Levels Outcome Comparison Resultsb
Confidence Years • Ages, N (ng/mL)a
2.68 (0.39, 18.30)
IgE:
Linear regression
-0.342 (-1.230,0.546)
Quadratic regression
-0.681 (-2.50, 1.137)
Cubic regression
1.464 (-5.354, 8.282)
Results stratified by gender not
statistically significant for boys and
combined
Confounding: maternal age, maternal educational level, pre-pregnancy BMI, allergy of parents, parity, infant gender, breastfeeding period,
environmental tobacco exposure, daycare attendance and blood sampling period; for IgE: maternal age, maternal allergic history, distance
from home to highway, parity, birth season, and blood sampling period
Okada et al. Japan Cohort Japanese Maternal blood Total allergic
(2014) 2003-2009 women who had 5.02 (3.71-6.83) diseases
Medium singleton births (eczema,
and their infants wheezing, and
N = 2,062 allergic
rhinoconjunctivi
tis symptoms)
OR by quartiles
ofPFOS
exposure
Total allergic diseases
Q2: 0.97 (0.75, 1.26)
Q3: 0.80 (0.61, 1.04)
Q4: 0.86 (0.66, 1.13)
p-value for trend = 0.139
Buser et al.
(2016)
Medium
Results: Lowest quartile used as reference group.
Confounding: Maternal age, maternal educational level, parental allergic history, infant gender, breast-feeding period, number of siblings,
day care attendance, and ETS exposure in infancy at 24 months.
United States
2005-2016
Cross-sectional
Adolescents
aged 12-19 yr
from NHANES
Nby cycle:
2005-2006: 637
2007-2010: 701
Serum
2005-2006:
GM = 14.98
(10.65-22.69)
2007-2010:
GM = 8.74
(5.96-13.75)
Food allergy or
sensitization
OR by quartiles
ofPFOS
Food allergy, 2007-2010 cycle
Q2: 2.22 (0.85, 5.77)
Q3: 2.43 (1.05, 5.59)
Q4: 2.95 (1.21, 7.24)
p-value for trend = 0.27
Food sensitization, 2005-2006
cycle: No statistically significant
associations or trends
D-lll
-------
APRIL 2024
Reference,
Confidence
Location, study Design Population, Matrix, Levels Outcome
Years Ages,N (ng/mL)a
Comparison
Resultsb
Outcome: Food sensitization defined as at least 1 food specific IgE level >0.35 kU/L.
Results: Lowest quartile used as reference.
Confounding: Age, sex, race/ethnicity, BMI, serum cotinine0
Goudarzi et al.
(2016a)
Medium
Japan Cohort Children at age Maternal blood Allergic OR by quartiles Q2: 0.66 (0.48, 0.90)
2003-2013 4 from the 4.93 (3.67-6. diseases, total ofPFOS Q3:0.79 (0.58, 1.07)
Hokkaido Study 65) Q4: 0.82 (0.60, 1.11)
on Environment p-value for trend = 0.391
and Children's
Health No statistically significant
N = 1,558 (765 associations, trends, or interactions
girls, 793 boys) by sex
Results: Lowest quartile used as reference.
Confounding: Maternal age, maternal educational level, sex, parental allergic history,
attendance, environmental tobacco smoke exposure
number of older siblings, breast feeding, daycare
Stein et al. United States, Cross-sectional Children aged Serum Allergy and OR [per Allergy
(2016b) 1999-2000, 12-19 years, GM = 20.8 rhinitis IQR(lnPFOS) 1.05 (0.80,1.37)
Medium 2003-2004, NHANES (95% CI: 19.1, increase (0.76
2005-2006 22.7) ln-ng/mL)] Rhinitis
N = 638 1.16 (0.90,1.50)
Confounding: Age, sex, race/ethnicity, survey year; for Wheeze: age, gender, race, weight status, serum cotinine.
Timmermann et
al. (2017a)
Medium
Faroe Islands,
Recruitment:
1997-2000
Cohort
Pregnant
women and
infants, follow-
up at ages 5, 7,
and 13 yr,
N = 559
Maternal serum
Prenatal/At
birth: 27.4
(23.3-33.3)
Age 5/7: 16.8
(13.5-21.1)
Age 13: 6.7
(5.2-8.5)
Allergy, allergic
rhino-
conjunctivitis in
past 12 mo,
positive skin
prick test, IgE
OR per
doubling of
PFOS
IgE: Percent
change per
doubling of
PFOS
Allergy (age 5)
OR = 0.73 (0.38,1.41)
Allergic rhino-conjunctivitis in past
12 mo, age 13
1.01 (0.54, 1.89)
Positive skin prick test, age 13
1.15 (0.75, 1.77)
IgE, age 7: -9.38 (-37.17, 30.71)
Confounding: Maternal parity, family history of eczema in children, allergic eczema and hay fever, maternal pre-pregnancy BMI, maternal
smoking during pregnancy, maternal fish intake during pregnancy, and duration of breastfeeding; for IgE: family history of eczema in
D-112
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
children, allergic eczema, and hay fever, maternal pre-pregnancy BMI, maternal smoking during pregnancy, sex, duration of breastfeeding,
fish intake at age 5, number of siblings, and daycare attendance at age 5
Impinen et al.
(2018)
Medium
Oslo, Norway,
1992-2002
Cohort, Nested
case-control
Infants followed Cord blood
up at 2 yr and 5.2 (4.0-6.6)
10 yr of age,
N = 641
Rhinitis, rhino-
conjunctivitis,
SPT
OR per log2-
unit increase in
PFOS
Rhinitis, current, 10 yr
1.00 (0.72, 1.40); p-value = 0.983
Rhinitis, ever, 10 yr
1.05 (0.74, 1.48); p-value = 0.775
Rhino-conjunctivitis, ever, 10 yr
1.02 (0.72, 1.45); p-value = 0.905
Rhino-conjunctivitis, ever, spes
IgE > 0.35, 10 yr
1.02 (0.71, 1.47); p-value = 0.905
SPT, any pos, 10 yr
0.87 (0.65, 1.17); p-value = 0.359
SPT + and/pr slgE > 0.35, 10 yr
0.91 (0.69, 1.19); p-value = 0.476
Confounding: Sex
Impinen et al.
(2019)
Medium
Oslo, Norway, Cohort
Enrollment:
1999-2008
Pregnant
women and
their infants
(followed to age
7),
N = 921
Maternal blood
12.87
(9.92-16.63)
Allergy, food or OR per IQR
inhaled increase in
PFOS
Allergy, food, current
All: 1.02 (0.73, 1.41);
p-value = 0.928
Boys: 1.09 (0.68, 1.74);
p-value = 0.72
Girls: 0.95 (0.59,1.51);
p-value = 0.815
Allergy, food, ever
All: 0.99 (0.72, 1.37);
p-value = 0.969
Boys: 1.11 (0.69, 1.77);
p-value = 0.671
Girls: 0.91 (0.58, 1.42);
p-value = 0.676
D-113
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Allergy, inhaled, current
All: 1.11 (0.72, 1.69);
p-value = 0.643
Boys: 0.86 (0.44, 1.71);
p-value = 0.669
Girls: 1.17 (0.55,2.48);
p-value = 0.679
Allergy, inhaled, ever
All: 1.27 (0.93,1.74);
p-value = 0.135
Boys: 1.2 (0.79, 1.84);
p-value = 0.39
Girls: 1.33 (0.84,2.12);
p-value = 0.224
Confounding: Maternal age, maternal BMI, maternal education, parity, smoking during pregnancy, nursery attendance
Ait Bamai et al. Hokkaido, Cohort Early pregnancy Maternal blood Rhino-
(2020) Japan, to7yr, 5.12 (3.75-7.02) conjunctivitis
Medium 2003-2012 N = 2,689
RR per ln-unit
increase in
PFOS, from
birth to 7 yr old
0.96 (0.79, 1.15); p-value = 0.626
Confounding: Sex, parity, maternal age at delivery, maternal smoking during pregnancy, pre-pregnancy BMI, and annual household income
during pregnancy
Kvalem et al.
Norway,
Cohort and
Children, age
Serum
Rhinitis, skin
Change in RR
Rhinitis
(2020)
Enrollment:
cross-sectional
10 yr: N = 377
All: 19.4 (IQR:
prick test (SPT)
per IQR
10 yr
Medium
1992-1993;
Age 16 yr:
9.23)
increase in
All: 0.98 (0.74,1.30);
Follow-up:
N = 375
Girls: 17.52
PFOS
p-value = 0.92
2002-2009
(IQR: 8.02)
Boys: 0.90 (0.66, 1.23);
Boys: 21.7
p-value = 0.52
(IQR: 8.86)
Girls: 0.97 (0.58, 1.62);
p-value = 0.92
16 yr
All: 1.03 (0.90,1.19);
p-value = 0.69
D-114
-------
APRIL 2024
Reference, Location, study Design Population, Matrix, Levels Outcome Comparison Resultsb
f nnti/lanrio Von~ Anne \ ' *
Confidence Years J 6 Ages, N
(ng/mL)a
Boys: 0.92 (0.72, 1.19);
p-value = 0.55
Girls: 1.15 (0.91, 1.45);
p-value = 0.24
SPT
10 yr
All: 1.10(0.95, 1.26);
p-value = 0.21
Boys: 0.98 (0.96, 1.01);
p-value = 0.17
Girls: 0.97 (0.65, 1.44);
p-value = 0.86
16 yr
All: 1.09 (1.03, 1.15);
p-value = 0.001
Boys: 1.07 (0.97, 1.17);
p-value = 0.18
Girls: 0.99 (0.80, 1.23);
p-value = 0.93
Confounding: 10 yr: Physical activity at 10 yr, mothers' education, BMI at 10 yr; 16 yr: BMI at 16 yr, puberty status at 16 yr, mothers'
education, physical activity level at 16 yr
Notes: BMI = body mass index; CI = confidence interval; ETS = environmental tobacco smoke; GM = geometric mean; IgE = immunoglobulin E; IQR = interquartile range;
MMR = measles, mumps, rubella; mo = months; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile
4; RR = risk ratio; SD = standard deviation; SE = standard error; SPT = skin prick test; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
Table D-ll. Associations Between PFOS Exposure and Eczema in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels Outcome
(ng/mL)a
Comparison
Resultsb
General Population
D-115
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Okada et al. Japan Cohort Japanese Maternal blood Eczema
(2014) 2003-2009 women who had 5.02 (3.71-6.83)
Medium singleton births
and their infants
N = 2,062
OR by quartiles
ofPFOS
exposure
Eczema
Q2: 1.06 (0.77, 1.46)
Q3: 0.93 (0.67, 1.29)
Q4: 0.89 (0.64, 1.24)
p-value for trend = 0.372
Results: Lowest quartile used as reference group.
Confounding: Maternal age, maternal educational level, parental allergic history, infant gender, breast-feeding period, and ETS exposure in
infancy at 24 months.
Goudarzi et al. Japan Cohort Children at age Maternal blood Eczema
(2016a) 2003-2013 4 from the 4.93 (3.67-
Medium Hokkaido Study 65654)
on Environment
and Children's
Health
N = 1,558 (765
girls, 793 boys)
OR by quartiles Q2: 0.64 (0.44, 0.93)
ofPFOS Q3: 0.65 (0.45, 0.95)
Q4: 0.85 (0.591, 1.22)
p-value for trend = 0.427
No statistically significant
associations, trends, or interactions
by sex
Results: Lowest quartile used as reference.
Confounding: Maternal age, maternal educational level, sex, parental allergic history, number of older siblings, breast feeding, daycare
attendance, environmental tobacco smoke exposure0
Timmermann et
al. (2017a)
Medium
Denmark
1997-2000
Cohort
Pregnant
women and
infants from the
CHEF study at
ages 5, 7, and
13 yr
N = 559
Serum Atopic eczema OR per
Prenatal at birth: at age 13 doubling of
16.8 (13.5-21.1) PFOS at age 13
Age 5/7: 27.4
(23.3-33.3)
Age 5: 0.75 (0.42, 1.34)
Age 13:0.8 (0.46, 1.39)
MMR vaccination before age 5
Yes: 8.94 (0.27, 299.11)
No: 0.82 (0.53, 1.28)
Confounding: Confounding: Family history of eczema in children., allergic eczema and hay fever, maternal pre-pregnancy BMI, maternal
smoking during pregnancy, sex, duration of breastfeeding, and fish intake at age 13, birth weight, and family history of chronic
bronchitis/asthma, maternal parity
Chen et al. China Cohort Infants followed
(2018) 2012-2015 up at 6, 12, and
Medium 24 mo
N = 687
children (328
Cord blood Atopic
All: 2.48 dermatitis
(Range = 0.39-
65.61)
Female: 2.47
OR per log-unit All: 1.23 (0.85, 1.76)
increase in
PFOS, or by
quartiles
Q2: 0.93 (0.56, 1.58)
Q3:l (0.59, 1.7)
Q4:1.31 (0.78,2.2)
Female: 1.1 (0.64, 1.87)
Q2: 0.73 (0.33, 1.61)
D-116
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Impinen et al.
(2018)
Medium
female and 359 (Range = 0.39-
male) 18.68)
Male: 2.49
(Range = 0.62-
65.61)
Q3:0.71 (0.32, 1.6)
Q4: 1.08 (0.5, 2.35)
Male: 1.42 (0.84, 2.42)
Q2: 1.34 (0.64, 2.8)
Q3: 1.3 (0.61,2.75)
Q4: 1.65 (0.79, 3.41)
Comparison: Logarithm base not specified.
Results: Lowest quartile used as reference group.
Confounding: Maternal age, maternal pre-pregnancy BMI, gestational week at delivery, birth weight, maternal education, paternal education,
parity, mode of delivery, family history of allergic disorders, infant sex, family income, maternal ethnicity, paternal smoking, breastfeeding
Norway
1992-2002
Cohort, Nested
case-control
Children from
the ECA study
at 0, 2, and
10 yr
N = 641
Cord blood
5.2 (4.0-6.6)
Atopic
dermatitis
diagnosed
anytime
between 0 and
2 yr old, or
between 0 and
10 yr old
OR per log2-
unit increase
PFOS
Ages 0-2: 1.15 (0.88, 1.52)
Ages 0-10: 0.68 (0.38, 1.2)
Confounding: Sex
Manzano-
Salgado et al.
(2019)
Medium
Wen et al.
(2019a)
Spain
2003-2015
Cohort
Pregnant
women and
children
followed up at
ages 1.5, 4, and
7 from the
INMA study
N = 1,188 at
1.5, N= 1,184
at 4 yr,
N = 1,066 at
7 yr
Maternal plasma Eczema
6.06 (4.52-7.82)
OR or RR per Age 1.5: 1.02 (0.83, 1.27)
log2-unit Age 4: 0.8 (0.65, 0.99)
increase in Age 7: 0.82 (0.68, 0.99)
PFOS Boys at ages 1.5, 4, and 7: 0.91
(0.75, 1.11)
Girls at ages 1.5, 4, and 7: 0.77
(0.64, 0.94)
From ages 1.5 to 7 yr: 0.86 (0.75,
0.98)
Confounding: Age at follow-up of the child, maternal age at delivery, parity, previous breastfeeding, pre-pregnancy BMI, region of residence,
and country of birth
Taiwan
2001-2005
Cohort
Children at age
2 yr
Cord blood Atopic
3.49 (2.18-5.05) dermatitis
OR by tertiles of T2: 1.33 (0.57, 3.20)
PFOS T3: 1.86 (0.84, 4.36)
D-117
-------
APRIL 2024
Reference,
Confidence
Population,
Location, .
Design
N
Exposure
Matrix, Levels Outcome Comparison
(ng/mL)a
Resultsb
Medium
N = 839
Results: Lowest tertile used as reference.
Confounding: Sex, family income, maternal atopy, breast feeding, and maternal age at childbirth
Wen et al.
Taiwan Cohort General
Cord blood
Atopic
Hazard ratio for
1.43 (0.82, 2.43)
(2019b)
2001-2005 population,
3.49 (2.18-5.05) dermatitis
PFOS > 5.05 ng
No statistically significant
Medium
children, and
/mL vs.
associations
adolescents <18
< 5.05 ng/mL
yr.; Infants
followed from
birth up to 5 yr
of age
N = 863
Confounding: Sex, parental education, parental atopy, breast feeding, and maternal age at childbirth
Notes: BMI = body mass index; CD = Crohn's disease; CHEF = Children's Health and the Environment in the Faroes; CIS = clinically isolated serum syndrome;
ECA = Environment and Childhood Asthma; INMA = INfancia y Medio Ambiente (Environment and Childhood) Project; MMR = measles, mumps, rubella; mo = months;
OR = odds ratio; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; RR = risk ratio; RRMS = relapsing remitting multiple sclerosis; UC = ulcerative colitis; yr = years.
a Exposure levels are reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
Table D-12. Associations Between PFOS Exposure and Autoimmune Health Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Gaylord et al. United States
(2020)
Medium
Case-control
Children and
adolescents
younger than
21 yr with
(cases) and
without
(controls) celiac
disease
N = 88 (42 girls,
46 boys)
Serum
Cases: 2.02
(IQR = 1.85)
Controls: 1.59
(IQR = 1.64)
Celiac disease
OR per ln-unit
change in PFOS
2.20 (0.78,6.18)
Girls: 12.8(1.17, 141);
p-value < 0.05
Boys: 1.02 (0.24, 4.21)
D-118
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: Genetic susceptibility score, albumin, BMI, age, race (non-Hispanic white vs. other race/ethnicity) and sex0
Steenland et al.
(2018)
Low
United States
1999-2012
Case-control
Patients with
UC, CD, or
healthy controls
N = 114 UC, 60
CD, 75 controls
Serum
UC: 3.95
CD: 3.32
Neither: 4.21
UC
Change in UC vs. CD:
log(PFOS) 0.05 (0.16), p-value = 0.77
comparing cases UC vs. control:
and controls -0.40 (0.21), p-value = 0.06
Comparison: Logarithm base not specified.
Results: Lowest quintile used as reference.
Confounding: Age, sex, ethnic group (white or non-white), year of sample
Sinisalu et al.
Finland Cohort Pregnant
Cord blood
Celiac disease
Comparison of
No significant differences in
(2020)
1999-2005 women and
Case: 2.21
mean PFOS
exposure between cases and control
Low
infants at birth
(min-max:
exposure levels
at birth or 3 mo
and 3 mo from
0.27-8.17)
the Type 1
Control: 2.25
Diabetes
(min-max:
Prediction and
0.27-5.32)
Prevention
Study in Finland
3-mo serum
(DIPP)
Case: 2.93
N = 33 (17
(min-max:
celiac disease,
0.27-7.66)
16 controls)
Control: 3.40
(min-max:
0.71-6.70)
Ammitzboll et
Denmark Case-control Adults with
Serum
Relapsing
Percent change
-17 (-27, -6); p-value = 0.004
al. (2019)
2019 (cases) or
Cases: 7.14
remitting
in PFOS
Females: -14 (-28, 3);
Low
without
(5.76-9.93)
multiple
comparing MS
p-value = 0.093
(controls)
Controls: 9.41
sclerosis
cases vs. healthy
Males: -19 (-32, -3);
RRMS or CIS
(6.41-13.0)
(RRMS)
controls
p-value = 0.023
N = 162 (92
women, 70
men)
Confounding: Age, sex, breastfeeding
D-119
-------
APRIL 2024
Notes: CD = Crohn's disease; CIS = clinically isolated serum syndrome; DIPP = Diabetes Prediction and Prevention Study in Finland; IQR = interquartile range; mo = months;
OR = odds ratio; RRMS = relapsing remitting multiple sclerosis; UC = ulcerative colitis.
a Exposure levels are reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.5 Cardiovascular
D.5.1 Cardiovascular Endpoints
Table D-13. Associations Between PFOS Exposure and Cardiovascular Effects in Recent Epidemiological Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Children and Adolescents
Li et al. (2021a)
High for
gestation, birth,
and childhood
exposures (3-yr
and 8-yr)
Medium for
exposure at 12-
yr follow-up
United States Cohort Pregnant women
2003-2006 and their children
followed up at
birth and ages 3,
8, and 12 from
HOME study
Gestation:
N = 203
At birth: N= 124
Age 3: N = 137
Age 8: N = 165
Age 12: N = 190
Maternal serum
Gestation: 12.9
(8.9-18.0)
Cord serum
At birth: 4.2
(3.0-6.5)
Serum
At age 3: 6.2
(4.5-9.9)
At age 8: 3.6
(2.8-1.7)
At age 12: 2.4
(1.8-3.2)
SBP (z-score),
mean of SBP
and DBP (z-
score)
Regression coefficient
per log2-unit IQR
increase in PFOS
SBP (z-score)
Gestation: 0.1 (-0.1, 0.2)
At birth: 0.2 (0.0,0.4)
Age 3: 0.1 (-0.1,0.4)
Age 8: 0.1 (-0.3,0.4)
Age 12: 0.2 (-0.1,0.5)
Mean of SBP and DBP (z-
score)
Gestation: 0.1 (-0.1, 0.2)
At birth: 0.1 (0.0, 0.3)
Age 3: 0.1 (-0.1,0.3)
Age 8: 0.1 (-0.2,0.4)
Age 12: 0.2 (0.0, 0.4)
Confounding0: visit, visit
concentrations, and parity;
Breastfeeding duration.
x PFAS, maternal age, maternal education, maternal pre-pregnancy BMI, gestational serum cotinine
and child age, sex, race, and pubertal stage. Additional confounding for analyses at age 3, age 8, and age 12:
Ma et al. (2019)
Medium
United States
2003-2012
Cross-
sectional
Adolescents aged
12-20 from
NHANES
Serum DBP, SBP Regression coefficient
median =11.1 per loglO-unit increase
(6.2-18.0) in PFOS
DBP
Total cohort: 0.014 (-0.001,
0.030)
Females: 0 (-0.02, 0.02)
D-120
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population,
Design Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
N = 2,251 (1,048
female, 1,203
male)
Males: 0.025 (0.001, 0.049);
p-value < 0.05
SBP
Total cohort: 0.002 (-0.004,
0.009)
Females: -0.001 (-0.009,
0.008)
Males: 0.003 (-0.006, 0.012)
Warembourg et France, Spain, Cohort
al. (2019)
Medium
Lithuania,
Norway,
Greece,
United
Kingdom
1999-2015
Pregnant women
and their children
at ages 6 and 11
from the HELIX
Project
N = 1,277
Prenatal exposure
Postnatal
exposure
Maternal blood:
6.4 (4.1-9.6)
Plasma: 2.0 (1.3-
3.2)
DBP, SBP
Regression coefficient
per log2-unit IQR
increase in PFOS
DBP
Maternal PFOS: 0.46 (-0.34,
1.27)
Childhood PFOS: 0.48 (-1.06,
0.62)
SBP
Maternal PFOS: -0.22 (-1.06,
0.62)
Childhood PFOS: 0.23 (-0.56,
1.03)
Confounding: Cohort of inclusion, maternal age, maternal education level, maternal pre-pregnancy BMI, parity, parental country of birth,
child age, child sex, child height0
Canova et al.
Italy
Cross-
Adolescents aged
Serum DBP, SBP
Regression coefficient
DBP
(2021)
2017-2019
sectional
14 to 19 yrand
Adolescents: 3.3
per ln-unit increase in
Adolescents
Medium
children aged 8 to
(2.2-1.9)
PFOS, or by quartiles
Per ln-unit increase: -0.44
11 yr from health
(-0.82, 0.05)
surveillance
Children: 2.2
Q2: -0.54 (-1.15,0.08)
program in
(1.6-3.0)
Q3: -0.66 (-1.30,-0.02)
Veneto Region
Q4: -0.78 (-1.45,-0.10)
Adolescents:
Children
N = 6,669
Per ln-unit increase: 0.03
Children:
(-0.54, 0.61)
N = 2,693
Q2: 0.67 (-0.15, 1.54)
Q3: 0.91 (0.05, 1.77)
Q4:-0.10 (-0.95, 0.75)
D-121
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
SBP
Adolescents
Per ln-unit increase: -0.47
(-1.02, 0.08)
Q2: -0.67 (-1.54, 0.20)
Q3: -0.96 (-1.87,-0.06)
Q4: -1.34 (-2.30, -0.38)
Children
Per ln-unit increase: -0.42
(-1.18,0.33)
Q2:-0.13 (-1.22, 0.95)
Q3: 0.18 (-0.95, 1.31)
Q4: -0.80 (-1.92, 0.33)
Results: Lowest quartile used as the reference group.
Confounding: Age, gender, country of birth, data on food consumption, degree of physical activity, salt intake, smoking status (for
adolescents only), time lag between the beginning of the study and the date of enrollment.
Papadopoulou et United Cohort
Mother-child
Maternal plasma
DBP (z-score),
Regression coefficient
DBP
al. (2021)
Kingdom,
pairs from the
(prenatal)
SBP (z-score)
per doubling in PFOS,
Maternal PFOS: 0.04 (-0.06,
Medium
France, Spain,
HELIX Project,
6.15 (3.99-9.16)
or by quartiles
0.14)
Lithuania,
children followed
Q2
-0.06 (-0.23,0.11)
Norway,
up around age 8
Plasma
Q3
0.03 (-0.16,0.23)
Greece
(range 6-12)
(childhood)
Q4
-0.04 (-0.29, 0.21)
Recruitment
N = 1,101
1.93 (1.22-3.11)
p-trend = 0.922
1999-2010,
Childhood PFOS: 0.01 (-0.06,
Follow-up:
0.08)
2013-2015
Q2
-0.02 (-0.18,0.13)
Q3
-0.01 (-0.19,0.17)
Q4
0.01 (-0.20, 0.23)
p-trend = 0.827
SBP
Maternal PFOS: 0.03 (-0.08,
0.14)
Q2
-0.06 (-0.25,0.13)
Q3
0.10 (-0.12,0.13)
Q4
-0.05 (-0.32, 0.23)
D-122
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
p-trend = 0.980
Childhood PFOS:-0.01
(-0.08, 0.07)
Q2
Q3
Q4
-0.04 (-0.21,0.13)
-0.03 (-0.23,0.16)
-0.03 (-0.27, 0.21)
p-trend = 0.763
Comparison: Maternal PFOS quartiles are defined as follows: Ql: 0.28-3.98; Q2: 3.99-6.15; Q3: 6.15-9.15; Q4: 9.16-47.98; childhood
PFOS quartiles are defined as follows: Ql: 0.00-1.22; Q2: 1.22-1.92; Q3: 1.93-3.10; Q4: 3.11-33.83.
Results: Lowest quartile used as the reference group.
Confounding: Maternal age and education, pre-pregnancy BMI, parity, cohort, child ethnicity, age, child gender, PFHxS, PFNA, PFOA
Manzano-
Spain Cohort
Pregnant women
Maternal blood
Blood Pressure
Regression coefficient
BP
Salgado et al.
2003-2008
and their children
GM = 5.80
(BP) (z-
per log2-unit increase
All age 4:-0.05 (-0.15,0.06)
(2017b)
at ages 4 and 7
(4.52-7.84)
score)
in PFOS
Girls: -0.06 (-0.22, 0.09)
Medium
from INMA study
Cardiometaboli
Boys: -0.02 (-0.18,0.14)
Age 4 N= 839
c Risk Score
All age 7: 0.06 (-0.04,0.15)
(412 girls, 427
(CMR)
Girls: 0.06 (-0.09, 0.20)
boys)
Boys: 0.04 (-0.08,0.17)
Age 4 N= 386
(197 girls, 189
CMR
boys) for CMR
All age 4: 0.28 (-0.33, 0.89)
score
Girls: 0.10 (-0.73, 0.93)
measurements
Boys: 0.47 (-0.44, 1.37)
Age 7 N = 1,086
(535 girls, 551
boys)
Confounding: Maternal region of residence, country of birth, previous breastfeeding, age, pre-pregnancy BMI; age/sex of child
Linetal. (2013)
Medium for
CIMT
Low for Systolic
BP
Taiwan
2006-2008
Cross-
sectional
Adolescents and
young adults ages
12-30
N = 637
Serum
8.65 (5.4-13.52)
SBP, CIMT Mean by quartiles
SBP: No associations across
quartiles; p-trend = 0.177
CIMT:
Significant associations across
exposure groups;
p-trend < 0.002
D-123
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Geiger et al.
(2014b)
Medium
Averina et al.
(2021)
Medium
Females: significant
associations across exposure
groups; p-trend < 0.001
Males: no associations across
exposure groups;
p-trend = 0.401
Ages 12-19: significant
associations across exposure
groups; p-trend < 0.001
Ages 20-30: no associations
across exposure groups;
p-trend = 0.084
Confounding: Age, gender, smoking status, alcohol drinking, BMI; for CIMT, also includes systolic blood pressure, low-density lipoprotein
cholesterol, triglyceride, high sensitivity C-reactive protein, homeostasis model assessment of insulin resistance
United States
1999-2000,
2003-2008
Cross-
sectional
Children
ages < 18 yr from
NHANES
N = 1,655
Serum
Mean
(SE) = 18.4 (0.5)
Hypertension OR per ln-unit increase Hypertension
in PFOS, or by quartile Per ln-unit increase: 0.83
(0.58, 1.19)
Q2: 0.99 (0.55, 1.78)
Q3: 0.73 (0.36, 1.48)
Q4: 0.77 (0.37, 1.61)
p-trend = 0.3625
Results: Lowest quartile used as the reference group.
Confounding: Age, sex, race-ethnicity, BMI categories, annual household income categories, moderate activity, total cholesterol, and serum
cotinine
Norway
2010-2011
Cross-
sectional
First level high
school students
ages 15-19 yr
from TFF1
N = 940
Serum
Girls: GM
(IQR) = 5.71
(2.64)
Boys: GM
(IQR) = 6.52
(3.09)
Hypertension OR by quartiles
Hypertension
Q2: 1.40 (0.78,2.51),
p-value = 0.261
Q3: 1.01 (0.56, 1.80),
p-value = 0.980
Q4: 1.86 (1.08,3.19),
p-value = 0.025
Outcome: Hypertension defined as systolic blood pressure > 130 mmHg and/or diastolic blood pressure > 80 mmHg.
Comparison: PFOS quartiles are defined as follows: Ql: 1.28-4.86: Q2: 4.87-6.21; Q3: 6.22-7.80; Q4: 1.28-4.86.
Results: Lowest quartile used as the reference group.
Confounding: Sex, age, BMI and physical activity outside school
D-124
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Lin etal. (2016) Taiwan Cross-
Medium 1992-2000 sectional
Adolescents and
young adults ages
12-30
N = 848
Serum
GM = 6.44 (95%
CI: 6.05-6.89)
8-OHDG (log-
l-ig/g
creatinine)
CIMT
CD31+ /
CD42a-
(log count/|iL)
CD31+ /
CD42a+
(log count/(iL)
CD62E
(log count/|iL)
CD62P
(log count/(iL)
Mean by quartiles
8-OHDG: No associations
across exposure groups;
p-trend = 0.102
CIMT
Ql: 0.433 (0.423,0.442)
Q2: 0.437 (0.428, 0.446)
Q3: 0.456 (0.447, 0.465)
Q4: 0.453 (0.444, 0.463)
p-trend <0.001
CD31+ / CD42a~: Statistically
significant increase across
exposure groups, 4.65-5.30
(Q3); p-trend = 0.010
CD31+ / CD42a+: Statistically
significant increase across
exposure groups, 8.02-8.54
(Q3); p-trend = 0.010
CD62E, CD62P: No
statistically significant
associations across exposure
groups
Confounding: Age, gender, smoking status, BMI, systolic blood pressure, low-density lipoprotein, triglyceride, homeostasis model
assessment of insulin resistance, and high sensitivity C-reactive protein
Khalil et al.
United States Cross-
Obese children
Serum
DBP, SBP
Regression coefficient
DBP: 1.17 (-0.40, 2.74)
(2018)
2016 sectional
ages 8-12
2.79
per unit increase in
SBP: 1.53 (-0.46, 3.51)
Low
OO
-t
II
£
(IQR = 2.10)
PFOS
Confounding: Age, race,
sex
Koshy et al.
United States Cross-
Children and
Serum
Augmentation
Regression coefficient
Al: -0.24 (-2.02,2.41)
(2017)
2011-2012 sectional
adolescents from
3.72
Index (Al)
per ln-unit increase in
BAD: 0.30 (-0.01, 0.62)
Low
the World Trade
(IQR = 2.82)
PFOS
PWV: -0.06 (-0.23,0.11)
Center Health
D-125
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Registry
(WTCHR)
N = 308
Comparison:
2.78
(IQR = 2.18)
Brachial Artery
Distensibility
(BAD)
Pulse Wave
Velocity
(PWV)
Confounding: BMI category, caloric intake, cotinine concentration, physical activity, race, sex
Pregnant Women
Matilla-
Spain Cohort Pregnant women Plasma CRP Percent median change
CRP
Santander et al.
2003-2008 from INMA study 6.05 (4.51-7.81) (loglO mg/dL by quartiles and per
-8.41 (-18.4,3.35)
(2017)
N= 1,240 ) loglO-unit increase in
By quartile:
Medium
PFOS
Q2: 6.18 (-11.3,28.4)
Q3: -6.76 (-22.9, 11.6)
Q4: -5.82 (-22.9, 12.7)
Results: Lowest quartile used as the reference group.
Confounding: Sub-cohort, country of birth, pre-pregnancy BMI, previous breastfeeding, parity, gestational week at blood extraction, physical
activity, relative Mediterranean Diet Score
General Population
Liao et al.
(2020)
High
United States
2003-2012
Cross-
sectional
Adults ages
20+ from
NHANES
N = 6,967 (3,439
females, 3,528
males)
Serum
12.8 (7.2-22.0)
DBP, SBP,
hypertension
DBP and SBP:
Regression coefficient
per loglO-unit increase
in PFOS or around
inflection point
(8.20 ng/mL)
Hypertension: OR by
tertiles
DBP
Levels < 8.20 ng/mL: -2.62
(-4.73,-0.51)
Levels > 8.20 ng/mL: 1.23
(-0.42, 2.88)
SBP
Per loglO-unit change: 1.35
(0.18,2.53)
Hypertension: No statistically
significant associations or
trends by tertiles or age groups
Males
T2: 1.17 (0.93, 1.47)
T3: 1.07 (0.85, 1.34)
Females
D-126
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
T2: 1.08 (0.87, 1.34)
T3: 1.18(0.92, 1.51)
p-value for interaction by
sex = 0.016
Outcome: Hypertension defined as average SBP >140 mmHg and average DBP > 90 mmHg, or self-reported use of prescribed
antihypertensive medication.
Comparison: Tertiles are defined as follows (in ng/mL PFOS): T1 < 8.9; 8.9 < T2 < 18.1; 18.1 < T3.
Results: Lowest tertile used as the reference group.
Confounding: Age, sex, education level, race, diabetes mellitus, consumption of at least 12 alcohol drinks/year, current smoking status, BMI,
waist circumference, hemoglobin, total cholesterol, estimated glomerular filtration rate (eGFR), dietary intake of sodium, dietary intake of
potassium, and dietary intake of calcium
Mattsson et al.
(2015)
High
Sweden
1990-1991,
2002-2003
Case-
control
Rural men
N = 462
Serum
Cases: 22.8
(IQR = 10.0)
Controls: 22.0
(IQR = 10.1)
CHD
OR by quartiles
CHD
Q2: 0.82 (0.46, 1.45)
Q3: 1.30 (0.74,2.26)
Q4: 1.07 (0.6, 1.92)
Results: Lowest quartile used as reference.
Confounding: BMI, systolic blood pressure, total cholesterol, HDL, tobacco use
Mobacke et al.
(2018)
High
Sweden
Years not
reported
Cross-
sectional
Adults aged 70
Serum
Left
from the
Mean
Ventricular
Prospective
(SD) = 14.9
End-Diastolic
Investigation of
(8.88)
Diameter
the Vasculature in
(LVEDD)
Uppsala Seniors
(mm)
(PIVUS) study
Left
N = 801
Ventricular
Mass Index
(LVMI)
(g/m27)
Relative Wall
Thickness
(RWT)
Regression coefficient
per ln-unit increase in
PFOS
LVEDD: 0.47 (0.08,0.87)
LVMI: 0.12 (-0.73, 0.97)
RWT: -0.01 (-0.01, -0.001)
Confounding: Sex, systolic blood pressure, antihypertensive medication, HDL and LDL, cholesterol, blood glucose, waist circumference,
triglycerides, BMI, education levels, exercise habits, smoking, energy, alcohol intake
D-127
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Bao et al. (2017) China Cross-
Medium 2015-2016 sectional
Adults aged 22-
96
N = 1,612(408
females, 1,204
males)
Serum
24.2 (14.6-37.2)
DBP, SBP, Regression coefficient DBP
hypertension per ln-unit change in Total: 2.70 (1.98, 3.42)
PFOS Females: 2.86 (1.51, 4.20)
Males: 0.45 (-0.47, 1.36)
Hypertension: OR per In- p-value for interaction by
unit increase in PFOS sex = 0.001
SBP
Total: 4.84 (3.55,6.12)
Females: 6.65 (4.32, 8.99)
Males: 1.50 (-0.17, 3.18)
p-value for interaction by
sex < 0.001
Hypertension
Total: 1.24 (1.08, 1.44)
Females: 1.63 (1.24,2.13)
Males: 1.08 (0.90, 1.29)
p-value for interaction by
sex = 0.016
Outcome: Hypertension defined as mean SBP > 140 mmHg and/or DBP > 90 mmHg, and/or use of antihypertensive medications.
Confounding: Age, sex, BMI, education, income, exercise, smoking, drinking, family history of hypertension
Liu et al.
United States Controlled Overweight and
Plasma DBP, SBP Partial Spearman DBP: 0.15; p-value < 0.05
(2018a)
2004-2007 trial obese adults ages
Females: 22.3 correlation coefficient SBP: 0.07
Medium
30-70 in the
(14.3-34.9)
POUNDS Lost
Males: 27.2
study
(19.9-45.2)
N = 621 (384
females, 237
males)
Confounding: Age, sex, race, education, smoking status, alcohol consumption, physical activity, menopausal status (women only), hormone
replacement therapy (women only), dietary intervention groups
Linetal. United States Cohort Adults from the Serum DBP, SBP, DBP, SBP: Regression SBP: lifestyle arm, baseline to
(2020b) 1996-2014 Diabetes Baseline: 26.7 pulse coefficient per log2- year 2:-2.13 mmHg/year
Medium Prevention (17.4-40.3) pressure (-3.54, -0.71)
D-128
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Program (DPP)
and Outcomes
Study (DPPOS)
N = 957 at
baseline, 956 at
year 2, and 346 at
year 14
Year 2: 27.6
(19.6-38.9)
Year 14: 9.8
(5.9-14.8)
(mmHg), and
hypertension
unit increase in PFOS,
or by quartiles
Hypertension: HR or RR
per log2-unit increase
DBP, pulse pressure,
hypertension: No statistically
significant associations by
timepoint, by quartiles, or by
in PFOS or by quartiles sex
Outcome: Hypertension defined as SBP > 140 mmHg and DBP > 90 mmHg in those without diabetes, SBP > 130 mmHg, and
DBP > 80 mmHg in those with diabetes, self-reported hypertension diagnosis, or use of antihypertensive medication.
Confounding: Sex, age, race/ethnicity, treatment assignment, education, income, marital status, alcohol intake, smoking, and DASH diet
score
Mitro et al.
United States Cohort Pregnant women Plasma DBP, SBP,
Regression coefficient
SBP: (3= 1.2 (0.3,2.2);
(2020)
1999-2005 and their children 24.7 (18.1-33.9) CRP (mg/L)
per log2-unit increase
p-value < 0.01
Medium
at age 3 from
in PFOS
Ages <35: 0.6% (-0.7, 1.8)
Project Viva
Percent difference (%)
Ages >35: 2.3% (0.9, 3.6);
N = 761 mothers
per log2-unit increase
p-value < 0.01
(496 ages <35,
PFOS
265 ages >35)
DBP, CRP: No statistically
significant associations
Population: For measurements of C-reactive protein, N = 454 mothers (247 ages < 35, 207 ages > 35).
Confounding: age, pre-pregnancy BMI, marital status, race/ethnicity, education, income, smoking, parity; breastfeeding in a prior pregnancy
for BP measurements only
Pitter et al.
Italy
Cross- Adults aged 20-
Serum
DBP, SBP,
DBP, SBP: Regression
DBP
(2020)
2017-2019
sectional 3 9 yr from Veneto
3.7 (2.5-5.6)
hypertension
coefficient per ln-unit
0.44 (0.20, 0.68)
Medium
Region with
Male: 4.8 (3.3-
risk
increase in PFOS, or by
Q2
0.32 (-0.08, 0.72)
PFAS-
6.9)
quartiles
Q3
0.30 (-0.12,0.71)
contaminated
Female: 3 (2-4.4)
Q4
0.57 (0.13, 1.02)
drinking water
Hypertension risk: OR
Males: 0.29 (-0.07, 0.64)
DBP and SBP:
per ln-unit increase in
Females: 0.51(0.17,0.84)
N = 15,380 (7,428
PFOS, or by quartiles
males, 7,952
SBP
females)
0.57 (0.24, 0.90)
Hypertension risk:
Q2
-0.01 (-0.56, 0.53)
N = 15,786 (7,667
Q3
Q4
0.27 (-0.29, 0.84)
0.60 (0.00, 1.21)
D-129
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Liu et al.
(2018b)
Medium
CMstensen et
al. (2019)
Medium
males, 8,119
females)
Males: 0.98 (0.47, 1.48)
Females: 0.32 (-0.13, 0.77)
Hypertension risk
1.12(1.02, 1.22)
Q2: 0.99 (0.85, 1.16)
Q3: 1.06 (0.91, 1.24)
Q4: 1.12 (0.95, 1.32)
Males: 1.17(1.05, 1.31)
Females: 1.06 (0.91, 1.24)
Outcome: Hypertension defined as any self-reported diagnosis, use of antihypertensive drugs, or elevated systolic blood pressure
(SBP >140 mmHg)/DBP > 90 mmHg).
Results: Lowest quartile used as the reference group.
Confounding: Age, BMI, time lag between the enrolment and the beginning of the study, gender, physical activity, smoking habits, food
consumption, salt habit, country of birth, alcohol consumption, education level and center in charge of the BP measurement
United States
2013-2014
Cross-
sectional
Adults ages 18+
from NHANES
N = 1,871
Serum
GM (SE) = 5.28
(1-02)
Hypertension
OR per ln-unit increase
inPFOS
Hypertension: 1.08 (0.88,
1.33)
Outcome: Hypertension defined as average SBP > 130 mmHg and average DBP > 85 mmHg, or self-reported use of prescribed
antihypertensive medication.
Confounding: Age, gender, ethnicity, lifestyle variables (smoking status, alcohol intake and household income), medications
(antihypertensive, anti-hyperglycemic, and anti-hyperlipidemic agents), other components of the metabolic syndrome
United States
2007-2014
Cross-
sectional
Adults ages 20+ Serum Hypertension OR by quartiles Hypertension
from NHANES 8.4 (4.8-14.0) No statistically significant
N = 2,975 associations
Outcome: Hypertension defined as SBP >130 mmHg and/or DBP > 85 mmHg, or use of antihypertensive drug in a patient with a history of
hypertension.
Results: Lowest quartile used as the reference group.
Confounding: Age, alcohol intake, family income, MP AH, PFDE, PFHxS, PFOA, PFUnDA, race/ethnicity, smoking status, survey cycle
D-130
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Donat-Vargas et
al. (2019b)
Medium
Sweden Cohort Adults aged 30-
1990-2013 60 at baseline
N = 187
Plasma
Baseline: 20 (15-
26)
Follow-up: 15
(9.7-21)
Hypertension OR by tertiles or per SD- Hypertension
unit increase in PFOS Baseline OR per increase: 0.71
(0.56, 0.89)
No other statistically
significant associations
Prospective: No statistically
significant associations
Outcome: Hypertension defined as SBP > 140 mmHg or DBP > 90 mmHg, self-reported diagnosis, or use of antihypertensive drugs
Results: Lowest tertile as the reference group.
Confounding: Gender, age, education, sample year, BMI, smoking habit, alcohol consumption, physical activity, healthy diet score
Jeddi et al.
Italy Cross- Residents aged Serum Elevated blood OR per ln-unit increase
Elevated BP: 1.10 (1.03, 1.17),
(2021a)
2017-2019 sectional 20-39 from the GM (range): 4.54 pressure in PFOS
p-value < 0.05
Medium
PFAS- (130 mmHg or DBP >85 mmHg.
Confounding: Age, gender, time lag between the beginning of the study and blood sampling center where BP has been measured, education,
number of deliveries, physical activity, country of birth, diet, alcohol intake, and smoking status, and other components of metabolic
syndrome
Fry and Power
United States Cohort Adults ages Serum Mortality by HR per SD-unit increase
Mortality
(2017)
2003-2006 60+from 4.3 ng/g cerebrovascul in PFOS
0.85 (0.65, 1.12);
Medium
NHANES (SE = 0.2 ng/g) ar or heart
p-value = 0.24
N= 1,036 diseases
Confounding: Age, education, gender, race/ethnicity, smoking status
Lind et al.
Sweden Cross- Adults ages Plasma CIMT, carotid CIMT, CIM-GSM:
CIMT, CIM-GSM,
(2017b)
2001-2004 sectional 70+in Uppsala, 13.23 (9.95- artery intima- Regression coefficient
atherosclerotic plaque: no
Medium
Sweden 17.77) media per ln-unit increase in
statistically significant
N= 1,016 (509 complex grey PFOS
associations
females and 507 scale median
males) (CIM-GSM), Plaque: OR per ln-unit
carotid artery increase in PFOS
atheroscleroti
c plaque
D-131
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Confounding: Sex, HDL and LDL cholesterol and serum triglycerides, BMI, BP, smoking exercise habits, energy and alcohol intake,
diabetes, educational level
Huang et al.
(2018)
Medium
United States
1999-2014
Cross-
sectional
Adults from
NHANES ages
18+
N = 10,859
Serum
12.40 (6.40-
22.60)
CVD, angina
pectoris,
congestive
heart disease,
CHD, heart
attack, stroke,
CRP (mg/L)
OR by quartiles
CRP: Spearman
correlation coefficient
CVD
Q2: 1.04 (0.78, 1.40)
Q3: 1.36 (1.07, 1.74)
Q4: 1.25 (0.92, 1.69)
p-trend = 0.0681
Females: No statistically
significant associations or
trends
Males
Q2: 1.76 (1.11,2.80)
Q3: 2.19 (1.37, 3.51)
Q4: 1.92 (1.20, 3.07)
p-trend = 0.0290; p-trend for
sex interaction = 0.0326
Ages < 50: No statistically
significant associations or
trends
Ages > 50
Q2: 1.01 (0.74, 1.38)
Q3: 1.39 (1.08, 1.78)
Q4: 1.27 (0.92, 1.75)
p-trend = 0.0491; p-trend for
age interaction = 0.1228
Angina pectoris: No
association by quartiles, no
significant trend;
p-trend = 0.4211
Congestive heart disease: No
association by quartiles, no
significant trend;
p-trend = 0.9462
D-132
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
CHD: No association by
quartiles, no significant trend;
p-trend = 0.0910
Heart attack
Q2: 1.30 (0.90, 1.87)
Q3: 1.56 (1.01,2.43)
Q4: 1.53 (0.96,2.45)
p-trend = 0.1026
Stroke: No association by
quartiles, no significant trend;
p-trend = 0.3084
CRP: -0.006;
p-value = 0.6062
Comparison: Age groups were defined as < 50 yr and > 50 yr.
Results: Lowest quartile used as the reference group.
Confounding: Age, sex, race/ethnicity, family poverty-income ratio, education levels, physical activity levels, BMI, alcohol-drinking status,
smoking status, diabetes, hypertension, family history of CVD, total energy intake, log-transformed levels of serum cotinine, log-transformed
levels of serum total cholesterol
Cardenas et al
(2019)
Medium
Umted States Controlled Prediabetic adults Plasma MVD,
1996-2014 trial ages 25+from GM nephropathy,
DPP and DPPOS (IQR) = 26.38 neuropathy,
N = 877 (22.8) retinopathy
OR per log2-unit
increase baseline PFOS
MVD: lifestyle arm: 1.37
(1.04, 1.84)
Nephropathy, neuropathy,
retinopathy: No statistically
significant associations
Confounding: Sex, race/ethnicity, baseline age, marital status, education, income, smoking history, BMI, maternal diabetes, paternal
diabetes, treatment assignment; baseline fasting glucose and HbAlc levels for microvascular disease only
Hutcheson et al. United States Cross- Adults from C8 Serum Stroke OR per ln-unit increase 0.90(0.82,0.98);
(2020) 2005-2006 sectional Health Project With diabetes: PFOS p-value = 0.02
Medium N = 48,206 21.4 (13.8-31.9)
Without diabetes:
20.1 (13.5-29.0)
D-133
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Confounding: Age, BMI, C-reactive proteins, diabetes duration, eGFR, HDL, LDL, history of smoking, race, sex
Osorio-Yanez et
al. (2021)
Medium
United States Cohort Prediabetic adults Plasma CAC
1999 ages 25+ enrolled 27.55 (Agastston
in the DPP trial (IQR = 19.30) score), AsAC
N = 666
OR per doubling in
PFOS
CAC (11-400): 1.20 (0.94,
1.53)
CAC (>400): 1.49(1.01,
2.21), p-value < 0.05
AsAC: 1.67(1.10,2.54),
p-value < 0.05
Results: CAC <11 used as reference group.
Confounding: Sex, age, BMI, race/ethnicity, cigarette smoking, education, treatment assignment, statin use.
He et al. (2018) United States Cross-
Low 2003-2012 sectional
Adults ages 20+
from NHANES
N= 3,948
(females) and
3,956 (males)
Serum
Female Mean
(SE) = 14.51
(0.26)
Male Mean
(SE) = 20.80
(0.32)
DBP, SBP Percent difference in log-
transformed outcome per
interquartile ratio
increase PFOS by
quartiles
DBP
Females:
Q2: -1.12 (-2.55, 0.34)
Q3: 0.00 (-1.45, 1.59)
Q4: 1.47 (-0.11, 3.08)
p-trend = 0.022
Males: No statistically
significant associations;
p-trend = 0.119
SBP:
Females:
Q2: 0.11 (-0.90, 1.02)
Q3: 0.34 (-0.56, 1.36)
Q4: 1.13 (0.23,2.16)
Males: No statistically
significant associations;
p-trend = 0.171
Comparison: Logarithm base not specified.
Results: Lowest quartile used as the reference group. Interquartile ratio = 75th/25th percentiles of serum PFOS: 3.08 ng/mL.
Confounding: None listed
Yang et al.
(2018)
Low
China
Years not
reported
Cross-
sectional
Adult men
N = 148
Serum
3.00 (Range:
0.3-14.6)
DBP, SBP,
hypertension
Regression coefficient
per log-unit increase in
n-PFOS
DBP, SBP, hypertension: no
statistically significant
associations
D-134
-------
APRIL 2024
Reference,
Confidence
Location,
Design
Years
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Hypertension: OR
comparing above or
below median
Outcome: Hypertension evaluated by individual BP components
Comparison: Logarithm base not specified.
Confounding: Age
Chen et al.
(2019a)
Low
Croatia Cross-
2007-2008 sectional
Adults aged 44-
56
N = 122
Plasma
GM = 8.91
(Range = 2.36-
33.67)
DBP, SBP
Regression coefficient
per ln-unit increase
PFOS
DBP: 1.42 (-0.95, 3.79)
SBP: 1.40 (-3.46, 6.25)
Confounding: Age, sex, education, socioeconomic status, smoking, dietary pattern, physical activity
Graber et al.
(2019)
Low
United States Cross-
2016-2017 sectional
Members of
community with
exposed water
supply
(Paulsboro, NJ)
ages 12+
N = 105
Serum
5.66 (3.09-9.28)
Cardiovascular
conditions,
self-reported
OR per unit increase in
PFOS
Any condition
1.08 (0.98, 1.21)
Confounding: Age, BMI
Occupational Populations
CMstensen et
al. (2016a)
Low
United States Cross-
2012-2013 sectional
Male anglers ages
50+
N = 154
Serum
19.00 (9.80-
28.00)
Cardiovascular
condition
(any), CHD,
hypertension
OR per unit increase in
PFOS
Any condition: 1.00 (0.98,
1.02)
CHD: 1.01 (0.98, 1.03)
Hypertension: 0.99 (0.96,
1.01)
Outcome: Hypertension was self-reported
Confounding: Age, BMI, work status, and alcohol consumption
Notes'. 8-OHdG = 8-hydroxy-2-deoxyguanosine; AI = augmentation index; BAD = brachial artery distensibility; BMI = body mass index; BP = blood pressure; CAC = coronary
artery calcium; CHD = coronary heart disease; CI = confidence interval; CIM-GSM = carotid artery intima-media complex grey scale median; CIMT = carotid artery intima-
media thickness (mm); CMR = cardiometabolic risk score; CRP = C-reactive protein; CVD = cardiovascular disease; DBP = diastolic blood pressure (mmHg); DPP = Diabetes
Prevention Program; DPPOS = Diabetes Prevention Program Outcomes Study; eGFR = estimated glomerular filtration rate; GM = geometric mean; HDL = high-density
lipoprotein cholesterol; HELIX = Human Early-Life Exposome; IQR = Interquartile range; HOME = Health Outcomes and Measures of the Environment; HR = hazard ratio;
INMA = INfancia y Medio Ambiente (Environment and Childhood) Project; LDL = low-density lipoprotein cholesterol; LVEDD = left ventricular end-diastolic diameter (mm);
LVMI = left ventricular mass index (g/m2); MP AH = 2-(N-methyl-PFOSA) acetate; MVD = microvascular disease; NHANES = National Health and Nutrition Examination
Survey; OR = odds ratio; PFOA = perfluorooctanoic acid; PFDE = perfluorodecanoic acid; PFHxS = perfluorohexane sulfonic acid; PFNA = perfluorononanoic acid;
D-135
-------
APRIL 2024
PFUnDA = perfluoroundecanoic acid; PIVUS = Prospective Investigation of the Vasculature in Uppsala Seniors; POUNDS = Preventing Overweight Using Novel Dietary
Strategies; PWV = pulse wave velocity; RR = risk ratio; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; RWT = relative wall thickness; SBP = systolic blood
pressure (mmHg); SD = standard deviation; SE = standard error; TFF1 = Tromse Fit Futures 1; WTCHR = World Trade Center Health Registry; yr = years(s).
a Exposure reported as median (25th-75th percentile) in ng/mL unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D.5.2 Serum Lipids
Table D-14. Associations Between PFOS Exposure and Serum Lipid Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Children
Li et al. (2021a) United States Cohort
Pregnant women
Maternal
Levels (mg/dL) of
Regression
Triglycerides
High for 2003-2006
and their children
serum
triglycerides and HDL;
coefficient per
Gestation: 0.0 (-0.2,
gestation, birth,
followed up at
Gestation:
triglycerides to HDL
log2-unit IQR
0.2)
and childhood
birth and ages 3,
12.9 (8.9-
ratio
increase in PFOS
At birth: 0.1 (-0.1,0.3)
exposures (3-yr
8, and 12 yrfrom
18.0)
Age 3:-0.1 (-0.3,0.1)
and 8-yr)
HOME study
Age 8: 0.1 (-0.1, 0.3)
Medium for
Gestation:
Cord serum
Age 12:0.1 (-0.1,0.3)
exposure at 12-yr
N = 203
At birth: 4.2
follow-up
At birth: N= 124
(3.0-6.5)
HDL
Age 3: N = 137
Gestation: 0.9 (-2.3,
Age 8: N = 165
Serum
4.1)
Age 12: N = 190
At age 3: 6.2
At birth: 0.9 (-2.6, 4.3)
(4.5-9.9)
Age 3: 0.4 (-3.5,4.4)
At age 8: 3.6
Age 8: 3.8 (-0.2, 7.7)
(2.8-4.7)
Age 12: 6.0 (1.9, 10)
At age 12: 2.4
(1.8-3.2)
Triglycerides to HDL
ratio
Gestation: 0.0 (-0.2,
0.2)
At birth: 0.1 (-0.1,0.3)
Age 3:—0.1 (-0.3,0.1)
Age 8: 0.1 (-0.1, 0.3)
Age 12: 0.1 (-0.1,0.3)
D-136
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
HOME = Health Outcomes and Measures of the Environment
Confounding: visit, visit x PFAS, maternal age, maternal education, maternal pre-pregnancy BMI, gestational serum cotinine
concentrations, and parity; and child age, sex, race, and pubertal stage. Additional confounding for analyses at age 3, age 8, and age 12:
Breastfeeding duration.
Lin et al. (2009) United States Cross-sectional Adolescents ages Serum
Metabolic syndrome OR per loglO- Metabolic syndrome
Medium
Nelson et al.
(2010)
Medium
1999-2000
and 2003-
2004
12-20 yrfrom
NHANES
N = 474
Mean
(SEM) = 3.11
(0.05) loglO-
ng/mL
HDL cholesterol and
metabolic syndrome
triglycerides
umt increase in
PFOS
HDL cholesterol
Model 4: 0.89 (0.51,
1.55)
Model 5: 1.38(0.61,
3.14)
Metabolic syndrome
triglycerides
Model 4: 0.95 (0.50,
1.80)
Model 5: 0.78 (0.41,
1.49)
Outcome: Metabolic syndrome HDL cholesterol defined as HDL < 1.04 mmol/L; metabolic syndrome triglycerides defined as
triglycerides > 1.24 mmol/L.
Confounding: Model 4: Age, sex, race, health behaviors (smoking status, alcohol intake, and household income), measurement data
(CRP and HOMA/insulin) and medications; additional confounding for model 5: Other components of the metabolic syndrome.
United States
2003-2004
Cross-sectional
Adolescent girls
ages 12-19 yr
from NHANES
N not reported
Serum Level (mg/dL) of HDL Regression HDL
Level not coefficient by Q4: 3.7 (-0.5, 7.9)
reported quartiles
Results: Lowest quartile used as the reference group,
only.
Confounding: Not reported.
Quartile analyses discussed in-text only and quantitative values provided for Q4
Geiger et al.
(2014a)
Medium
United States
1999-2008
Cross-sectional
Adolescents ages
12-18 yrfrom
NHANES
N = 815
Plasma
Mean
(SE) = 17.7
(0.7)
Levels (mg/dL) of TC,
LDL, HDL, and
triglycerides; elevated
TC; elevated LDL;
depressed HDL;
elevated triglycerides
Lipid levels: TC: 0.06 (0.02, 0.1)
Regression T2: 3.37 (-1.39, 8.13)
coefficient per In- T3: 5.85 (0.1, 11.61)
unit increase in p-trend = 0.051
PFOS, Mean
change by tertiles HDL
T2: 1.62 (-0.54, 3.78)
D-137
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Elevated or
depressed: OR
per ln-unit
increase in PFOS,
or by tertiles
T3: -0.01 (-2.06,2.04)
p-trend = 0.970
LDL: 4.28 (1.6, 6.95)
T2: 2.7 (-1.39, 6.78)
T3: 6.99 (1.99, 11.98)
p-trend = 0.0081
TG: -1.85 (-5.61, 1.91)
T2: -4.79 (-11.09, 1.5)
T3: -5.55 (-12.26, 1.16)
p-trend = 0.110
Elevated TC: 1.35 (1.11,
1.64)
T2: 1.35 (0.94, 1.95)
T3: 1.53 (1.07,2.19)
p-trend = 0.018
Depressed HDL: 1.03
(0.7, 1.53)
T2: 0.88 (0.52, 1.5)
T3: 0.99 (0.58, 1.7)
p-trend = 0.987
Elevated LDL: 1.48
(1.15, 1.9)
T2: 1.43 (0.91,2.24)
T3: 1.76 (1.1,2.82)
p-trend = 0.018
Elevated TG: 0.9 (0.56,
1.43)
T2: 0.82 (0.46, 1.45)
T3: 0.64 (0.3, 1.37)
p-trend = 0.242
D-138
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Outcome: Elevated TC defined as TC > 170 mg/dL; elevated LDL defined as LDL >110 mg/dL; depressed HDL defined as
HDL < 40 mg/dL; elevated triglycerides defined as triglycerides > 150 mg/dL.
Results: Lowest tertile used as the reference group. Regression coefficient for continuous analysis of HDL not reported.
Confounding: Age, sex, race-ethnicity, BMI categories, annual household income categories, activity level, and serum cotinine
Frisbee et al. United States Cross-sectional
Children and
Serum
Abnormal TC,
OR by quintiles Abnormal TC
(2010) 2005-2006
adolescents ages
Mean
abnormal HDL,
Q2: 1.3 (1.1, 1.4)
Medium for TC,
1.0 to 17.9 yr in
(SD) = 22.7
abnormal LDL, and
Q3: 1.3 (1.2, 1.5)
GDL-C, fasting
the C8 Health
(12.6)
abnormal fasting
Q4: 1.3 (1.2, 1.6)
TG; low for LDL
Project
triglycerides
Q5: 1.6 (1.4, 1.9)
N = 12,470
Abnormal HDL
Q2: 0.9 (0.8, 1.1)
Q3: 0.8 (0.7, 1.0)
Q4: 0.8 (0.7, 0.9)
Q5: 0.7 (0.6,0.9)
Abnormal LDL
Q2: 1.2 (1.0, 1.5)
Q3: 1.2 (1.0, 1.5)
Q4: 1.3 (1.1, 1.6)
Q5: 1.6 (1.3, 1.9)
Abnormal fasting
triglycerides
Q2: 1.3 (0.9, 1.8)
Q3: 1.0 (0.7, 1.4)
Q4: 1.1 (0.7, 1.6)
Q5: 1.2 (0.8, 1.5)
Outcomes: Abnormal TC defined as TC > 170 mg/dL; abnormal HDL defined as HDL < 40 mg/dL; abnormal LDL calculated for
participants with a triglyceride level < 400 mg/dL regardless of fasting status and defined as LDL >110 mg/dL; fasting triglycerides
defined as self-reported fasting > 6 hr before phlebotomy, and abnormal fasting triglycerides defined as fasting triglycerides >150 mg/dL.
Results: Lowest quintile used as the reference group.
Confounding: Age, estimated time of fasting, BMI z-score, sex, regular exercise
Timmermann et
al. (2014)
Medium
Denmark Cross-sectional Children Plasma Triglycerides (mmol/L) Percent change
1997 ages 8-10 from per 10-unit
Danish increase PFOS
Normal weight: -0.5
(-3.2, 2.4),
p-value = 0.75
D-139
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
component of 41.5
Overweight: 8.6 (1.2,
EYHS (Range = 6.2-
16.5), p-value = 0.02
N = 400 normal 132.5)
weight, N = 59
p-value for PFOS-BMI
overweight
interaction = 0.02
Confounding: Sex, age, ethnicity, paternal income, fast-food consumption, and fitness
Maisonet et al.
United Case-control Pregnant women Serum Levels (mg/dL) of TC,
Regression
TC
(2015b)
Kingdom and their 20.5 LDL, HDL, and
coefficient per
Age 7
Medium for TC
1991-1992 daughters (Range = 7.6- triglycerides (In-
unit increase in
Tl: 0.30 (-3.10, 3.70)
and HDL at age 7
followed up at 38.2) mg/dL)
PFOS in each
T2: 2.09 (-0.64, 4.82)
and all lipids at
ages 7 and 15
tertile of
T3:-0.10 (-0.73, 0.54)
age 15
from ALSPAC
exposure
Age 15
Low for
Age 7: N = 111
Tl: 1.64 (-2.20,5.48)
Triglycerides and
Age 15: N = 88
T2: 3.41 (0.37, 6.45)
LDL at age 7
T3: -0.77 (-1.40, -0.13)
LDL
Age 7
Tl: 0.37 (-2.34, 3.08)
T2: 1.02 (-1.15,3.19)
T3: 0.02 (-0.48,0.53)
Age 15
Tl: 1.91 (-1.34, 5.17)
T2: 2.09 (-0.50, 4.67)
T3: -0.54 (-1.08,
-0.003)
HDL
Age 7
Tl: 0.76 (-0.79,2.31)
T2: 0.22 (-1.03, 1.46)
T3: -0.04 (-0.33,0.25)
Age 15
Tl: -0.55 (-2.34, 1.24)
T2: 1.15 (-0.27,2.57)
D-140
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
T3:-0.18 (-0.47, 0.12)
Triglycerides
Age 7
Tl: -0.031 (-0.085,
0.023)
T2: 0.008 (-0.035,
0.052)
T3:-0.004 (-0.015,
0.006)
Age 15
Tl: 0.012 (-0.032,
0.056)
T2: 0.016 (-0.019,
0.051)
T3:-0.004 (-0.011,
0.004)
ALSPAC = Avon Longitudinal Study of Parents and Children
Confounding: Previous live births, maternal education, and maternal age at delivery
Zeng et al.
(2015)
Medium
Taiwan
2009-2010
Cross-sectional
Children
ages 12-15
N = 225
Serum
Median = 28.8
among males,
29.9 among
females
Levels (ng/dL) of TC,
LDL, HDL, and
triglycerides
Regression TC: 0.31 (0.18, 0.45)
coefficient per In- p-value <0.001
unit increase LDL: 0.28 (0.18, 0.38)
PFOS p-value < 0.001
HDL: -0.01 (-0.07,
0.05)
p-value = 0.72
Triglycerides: 0.19 (0,
0.38)
p-value = 0.05
Confidence: Results for TG and LDL considered low confidence because of a lack of fasting prior to blood sample collection.
Confounding: Age, gender, BMI, parental education level, exercise, environmental tobacco smoke exposure0
Domazet et al. Denmark Cohort
Members of the
Plasma
Levels (mmol/L) of TG Percent change in
Age 9 to 15:
(2016) 1997-2009
EYHS evaluated
Median at
TG at age 15 or
-0.7 (-5.03, 3.77)
Medium
at ages 9 and 15
9 = 44.5
21 per 10 unit
Age 9 to 21: -1.98
(N = 260), 9 and
(-8.17,4.75)
D-141
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
21 (N= 175), or
15 and 21
(N = 171)
(male) or 39.9
(female)
Median at
15 = 22.3
(male) or 20.8
(female)
Median at
21 = 11.9
(male) or 9.1
(female)
increase in PFOS Age 15 to 21: 0.77
at age 9 or 15 (-8.28,10.71)
Confounding: Sex, age, and TG levels at baseline age; ethnicity, maternal parity, and maternal income in 1997 (9 yr of age). Waist
circumference was adjusted for height in order to account for body size.
Manzano-
Salgado et al.
(2017b)
Medium
Jain et al. (2018)
Medium
Spain
2003-2008
Cohort
Pregnant women Maternal Levels (z-score) of TC,
and their children plasma during LDL, HDL, and TG
(age 4) from 1st trimester
INMA study GM = 5.80
N = 627
Regression TC: 0.02 (-0.10, 0.15)
coefficient per LDL: 0.02 (-0.10, 0.15)
log2-unit increase HDL: -0.03 (-0.14,
PFOS 0.09)
TG: 0.05 (-0.06,0.17)
Confidence: Results for TG and LDL considered low confidence because of a lack of fasting prior to blood sample collection.
Confounding: Maternal region of residence, country of birth, previous breastfeeding, age, pre-pregnancy BMI; age/sex of child
United States
2013-2014
Cross-sectional
Children ages 6-
11
N = 458
Serum Levels (loglO-mg/dL) Regression
GM = 2.67 for of TC, HDL, and non- coefficient per
linear PFOS, HDL loglO-unit
1.35 for lm- increase PFOS
PFOS
Linear PFOS
TC: 0.02738
p-value = 0.03
Non-HDL: -0.00357
p-value = 0.4
HDL: 0.04631
p-value = 0.1
lm-PFOS
TC: 0.01241
p-value = 0.22
Non-HDL: -0.00661
p-value = 0.04
HDL: 0.04612
p-value = 0.05
Confounding: Gender, race/ethnicity, age, poverty-income ratio, BMI percentiles, fasting time, and exposure to secondhand smoke
D-142
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
Kang et al.
Korea Cross-sectional Children aged 3- Serum Levels of TC (mg/dL),
Regression
TC: -0.45 (-10.67,
(2018)
2012-2014 18 from Korea Median = 5.68 LDL (mg/dL), and TG
coefficient per ln-
9.77)
Medium
Environmental (ln-mg/dL)
unit increase
LDL: 2.51 (-6.88,
Health Survey in
PFOS
11.89)
Children and
TG:-0.020 (-0.19,
Adolescents
0.15)
(KorEHS-C)
All p-value >0.5
t"-
-t
II
£
Results: LDL and TG evaluated at ages 7-18 only (N = 117)
Confounding: Age, sex, BMI z-score, household income, secondhand smoking
Mora et al.
United States Cohort and Pregnant women Prenatal Levels (mg/dL) of TC,
Regression
Prenatal:
(2018)
1999-2010 cross-sectional and their children maternal HDL, LDL, and TG
coefficient per
TG: -1.4 (-4.6, 1.8)
Medium
from Project Viva plasma
IQR increase in
Boys: 1.0 (-2.2, 4.2)
N = 512 prenatal, Median = 24.6
PFOS
Girls: -4.2 (-9.2, 0.8)
596 mid-
p-value for interaction
childhood Mid-
by sex = 0.04
Jensen et al.
(2020a)
Medium
childhood
plasma
Median = 6.2
Mid-childhood:
TC: 1.8 (-0.2, 3.7)
HDL: 1.5 (0.4, 2.5)
TG: -2.5 (-4.3, -0.6)
Boys: 0.5 (-1.8, 2.9)
Girls: 4.0 (0.3, 7.8)
No other statistically
significant associations
Confounding: maternal education, prenatal smoking, gestational age at blood draw (for prenatal data), and child's sex, race/ethnicity, and
age at lipids/ALT measurements
Denmark
2010-2012
Cohort
Pregnant women Maternal Levels (standard Regression
and their children serum deviation score) of TC, coefficient per
assessed at 3 mo Median = 8.04 LDL, HDL, and TG unit increase in
and 18 mo PFOS
N = 260 at 3 mo,
83 at 18 mo
All associations were
between -0.07 and 0.05,
all with p-values > 0.05
Confounding: Maternal age, parity, pre-pregnancy BMI, pre-pregnancy BMI2, education, smoking, sex, and lipid outcome at 3 mo
D-143
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
Spratlen et al.
(2020b)
Medium
United States
2001-2002
Cross-sectional
Pregnant women
and their children
from the
Columbia
University World
Trade Center
birth cohort
N = 222
Cord blood
Median = 6.32
Levels (mg/dL) of TC,
total lipids, and TG in
cord blood
Percent change
per 1% increase
in PFOS
TC: 0.062
(-0.004,0.13)
Total lipids: 0.067
(0.005,0.129)
p-value < 0.05
TG: 0.086
(-0.036, 0.21)
Confounding: Maternal age, child sex, maternal education, maternal race, parity, pre-pregnancy BMI, marital status, family smoking, and
gestational age
Averina et al.
(2021)
Medium
Norway
2010-2011
Cross-sectional
First level high
school students
ages 15-19 yr
from TFF1
N = 940
Serum
Girls: GM
(IQR) = 5.71
(2.64)
Boys: GM
(IQR) = 6.52
(3.09)
Levels (mmol/L) of
TC, HDL, LDL, and
TG
Regression
coefficient per
loglO-unit
increase in PFOS
TC: 0.38(0.10,0.66),
p-value = 0.008
HDL: 0.08 (-0.03,
0.20), p-value = 0.152
LDL: 0.30 (0.05, 0.55),
p-value = 0.021
TG: 0.006 (-0.18,0.20),
p-value = 0.947
TFF1 = Tromso Fit Futures 1
Confounding: Sex, age, BMI, and lifestyle and diet variables
Blomberg et al.
Faroe Islands
Cohort and
Children from the
Serum
Levels (mmol/L) of
Regression
TC, age 9 (PFOS age 9)
(2021)
Recruitment:
cross-sectional
Faroese Birth
Birth: 2.87
TC, HDL
coefficient per
0.15 (0.025,0.27),
Medium for HDL
2007-2009
Cohort 5 at birth,
(2.13-4.04)
log2-unit increase p-value < 0.05
and TC
18 mo, and 9 yr
Female: 2.82
in PFOS
Females: 0.25 (0.077,
Low for LDL and
Birth: N = 459
(2.04-3.86)
0.43), p-value < 0.05
TG
(219 female, 240
Male: 2.93
Males: 0.05 (-0.12,
male)
(2.19-4.10)
0.22)
18 mo: N= 334
p-value for interaction
9 yr: N = 366
18 mo: 6.81
by sex = 0.104
(4.38-9.82)
D-144
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
9 yr: 3.08
(2.42-4.31)
Levels at 5 yr
and by sex at
18 mo and
9 yr not
reported
HDL, age 9 (PFOS
age 9)
0.077 (0.03,0.12),
p-value < 0.05
Females: 0.07 (0.0017,
0.14), p-value < 0.05
Males: 0.083 (0.018,
0.15), p-value < 0.05
p-value for interaction
by sex = 0.788
Confounding: Child sex and maternal education; analyses except PFAS at 9 yr additionally adjusted for maternal smoking during
pregnancy, maternal pre-pregnancy BMI, and parity
Canova et al.
(2021)
Medium for TC,
HDL
Low for LDL,
TG
Italy
2017-2019
Cross-sectional
Adolescents aged
14 to 19 yr and
children aged 8 to
11 yr from health
surveillance
program in
Veneto Region
Adolescents:
N = 6,669
Children:
N = 2,693
Serum
Adolescents:
3.3 (2.2-4.9)
Children: 2.2
(1.6-3.0)
Levels (ng/mL) of TC,
HDL, LDL,
triglycerides
Regression TC
coefficient per In- Adolescents: 3.32 (2.20,
unit increase in 4.45)
PFOS Children: 6.22 (4.32,
8.13)
HDL
Adolescents: 1.17 (0.71,
1.63)
Children: 1.91 (1.10,
2.73)
LDL
Adolescents: 2.66 (1.70,
3.62)
Children: 4.52 (2.80,
6.23)
Triglycerides
Adolescents: -0.02
(-0.04, 0.00)
Children: -0.01 (-0.04,
0.02)
D-145
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Confounding: Age, gender, country of birth, data on food consumption, degree of physical activity, salt intake, smoking status (for
adolescents only), time lag between the beginning of the study and the date of enrollment.
Papadopoulou et
al. (2021)
Medium
United
Kingdom,
France, Spain,
Lithuania,
Norway,
Greece
Recruitment
1999-2010,
Follow-up:
2013-2015
Cohort
Mother-child
pairs from the
HELIX Project,
children followed
up around age 8
(range 6-12)
N = 1,101
Maternal
plasma
(prenatal)
6.15 (3.99-
9.16)
Plasma
(childhood)
1.93 (1.22-
3.11)
Levels (z-scores) of
HDL, LDL, and
triglycerides
Regression
coefficient per
doubling in
PFOS, or by
quartiles
HDL
Maternal PFOS: 0.06
(-0.06,0.18)
Q2:-0.13 (-0.33, 0.07)
Q3: -0.06 (-0.29,0.17)
Q4:-0.18 (-0.47, 0.12)
p-trend = 0.577
Childhood PFOS: 0.00
(-0.08, 0.08)
Q2: 0.23 (0.04, 0.41)
Q3: 0.33 (0.11, 0.54)
Q4: 0.37 (0.11, 0.63)
p-trend = 0.009
LDL
Maternal PFOS: -0.03
(-0.15,0.09)
Q2: -0.05 (-0.26,0.15)
Q3: -0.11 (-0.35,0.12)
Q4: 0.09 (-0.21, 0.39)
p-trend = 0.990
Childhood PFOS: 0.05
(-0.03,0.13)
Q2: 0.06 (-0.13,0.25)
Q3: 0.15 (-0.06,0.37)
Q4: 0.12 (-0.14, 0.38)
p-trend = 0.210
Triglycerides
Maternal PFOS: -0.07
(-0.19, 0.05)
Q2: -0.07 (-0.27,0.14)
Q3:-0.19 (-0.43, 0.04)
Q4: -0.14 (-0.44,0.16)
D-146
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
p-trend = 0.191
Childhood PFOS: 0.04
(-0.04,0.12)
Q2: 0.02 (-0.17,0.21)
Q3: 0.03 (-0.19,0.24)
Q4: 0.13 (-0.14, 0.39)
p-trend = 0.256
Comparison: Maternal PFOS quartiles are defined as follows: Ql: 0.28-3.98; Q2: 3.99-6.15; Q3: 6.15-9.15; Q4: 9.16-47.98; childhood
PFOS quartiles are defined as follows: Ql: 0.00-1.22; Q2: 1.22-1.92; Q3: 1.93-3.10; Q4: 3.11-33.83.
Results: Lowest quartile used as the reference group.
Confounding: Maternal age and education, pre-pregnancy BMI, parity, cohort, child ethnicity, age, child gender, PFHxS, PFNA, PFOA
Tian et al. (2020) China
Medium 2012
Cohort
Pregnant women Maternal
and their newborn plasma
children from the 10.5 (7.3 7-
S-MBCS 16.3)
N = 306
Levels (ln-mg/dL) of
TC, LDL, HDL, and
triglycerides
Regression TC
coefficient per In- Per ln-unit: -0.10
umt increase in
PFOS, or by
tertile
(-0.18, -0.02),
p-value = 0.018
T2: -0.09 (-0.20, 0.03)
T3: -0.15 (-0.27,
-0.03), p-value < 0.05
p-trend < 0.05
LDL
Per ln-unit: -0.07
(-0.18,0.03),
p-value = 0.164
T2:-0.12 (-0.27, 0.03)
T3: -0.09 (-0.24,0.06)
HDL
Per ln-unit: -0.11
(-0.21,-0.02),
p-value = 0.021
T2: -0.11 (-0.25,0.03)
T3: -0.17 (-0.31,
-0.031), p-value < 0.05
p-trend < 0.05
D-147
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Triglycerides
Per ln-unit: -0.05
(-0.14,0.04),
p-value = 0.287
T2: -0.08 (-0.21,0.06)
T3: -0.02 (-0.16,0.11)
Results: Lowest tertile used as reference group.
Confounding: Maternal age, pre-pregnancy BMI, household income, infant sex, gestational age.
Pregnant Women
Starling et al.
Norway
Cross-sectional Women in mid
Plasma
Levels (mg/dL) of TC,
Regression
TC
(2014b)
2003-2004
pregnancy
13.03 (10.31-
HDL, LDL, and
coefficient per ln-
Per ln-unit: 8.96 (1.70,
Medium for TC,
(median = 18 wk
16.60)
triglycerides (ln-
unit or IQR
16.22)
HDL, and LDL
of gestation) from
mg/dL)
increase in PFOS,
Per IQR: 4.25 (0.81,
Low for
MoBa
or by quartiles
7.69)
Triglycerides
N = 891
Q2: -3.35 (-10.34,
3.64)
Q3: 3.06 (-4.93, 11.05)
Q4: 7.59 (-0.42, 15.60)
HDL
Per ln-unit: 4.39 (2.37,
6.42)
Per IQR: 2.08(1.12,
3.04)
Q2: 1.96 (-0.39,4.31)
Q3: 2.49 (0.00,4.97)
Q4: 4.45 (2.04, 6.86)
LDL
Per ln-unit: 6.48 (-0.07,
13.03)
Per IQR: 3.07 (-0.03,
6.18)
Q2: -3.23 (-9.28, 2.83)
Q3: 2.60 (-4.49, 9.70)
D-148
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
Skuladottir et al.
(2015)
Medium
Matilla-
Santander et al.
(2017)
Medium
Starling et al.
(2017)
Medium
Q4: 5.51 (-1.62, 12.64)
Triglycerides
Per ln-unit: -0.02
(-0.09, 0.04)
Per IQR: -0.01 (-0.04,
0.02)
Q2: 0.00 (-0.06, 0.07)
Q3: -0.03 (-0.10,0.05)
Q4: 0.00 (-0.09, 0.04)
Results: Lowest quartile used as reference group.
Confounding: Age, pre-pregnant BMI, nulliparous or interpregnancy interval, duration of breastfeeding previous child, education
completed, current smoking at mid-pregnancy, gestational weeks at blood draw, and oily fish consumed daily.
Denmark
1988-1989
Cross-sectional
Pregnant women
N = 854
Serum Levels (mmol/L) of TC Regression
Mean =22.3 coefficient by
quintile
Q2: 0.24 (-0.04, 0.53)
Q3: 0.22 (-0.07,0.50)
Q4: 0.35 (0.06, 0.64)
Q5: 0.44 (0.15,0.74)
p-trend = 0.004
Results: Lowest quintile used as reference group.
Confounding: Age, parity, education, smoking and pre-pregnancy BMI, total caloric intake, and intake of vegetables, meat, and meat
products
Spain
2003-2008
Cohort
Pregnant women
from the Spanish
INMA birth
cohort
N = 1240
Plasma Levels of TC (mg/dL), Percent change in TC: 0.88 (-0.53, 2.37)
Median = 6.05 TG (loglO-mg/dL), and median lipid TG: -5.86 (-9.91,
C-reactive protein level per loglO- -1.63)
(loglO-mg/dL) unit increase in
PFOS
Confidence: TG results considered low confidence because of a lack of fasting prior to blood sample collection.
Confounding: Sub-cohort, country of birth, pre-pregnancy BMI, previous breastfeeding, parity, gestational week at blood extraction,
physical activity, and relative Mediterranean Diet Score
United States
2009-2014
Cohort
Pregnant women
ages 16^15 from
the Healthy Start
study
N = 598
Serum
Median = 2.4
Levels of HDL
(mg/dL) and TG (ln-
mg-dL)
Regression HDL: 0.79 (-0.68, 2.27)
coefficient per In- TG: 0.004 (-0.033,
unit increase 0.041)
PFOS
Confounding: Maternal age, race/ethnicity, pre-pregnancy BMI, education, gravidity, smoking, and gestational age at
D-149
-------
APRIL 2024
„ , T 4.' Population, Exposure
Reference, Location, ¦' . _ a .
„ Design Ages, Matrix, Outcome Comparison Select Results
Commence Years T , '
N Levels3
blood draw
Yang et al.
(2020b)
Medium
China
2013-2014
Cohort
Pregnant women
ages 20^10 yr in
early pregnancy
N = 436
Serum
6.78 (5.08-
9.60)
Levels (ln-mmol/L) of Regression
TC
TC, triglycerides,
HDL, and LDL;
LDL/HDL ratio
coefficient per In- Per ln-unit: -0.090
umt increase in
PFOS, or by
quartiles
(-0.274, 0.093)
Q2
Q3
Q4
0.26 (-0.33, 0.85)
-0.04 (-0.44, 0.36)
-0.10 (-0.52,0.32)
p-trend = 0.832
Triglycerides
Per ln-unit: -0.084
(-0.307,0.138)
Q2: -0.03 (-0.48, 0.42)
Q3: 0.07 (-0.38,0.52)
Q4: 0.09 (-0.35,0.53)
p-trend = 0.478
HDL
Per ln-unit: 0.025
(-0.030, 0.081)
Q2: 0.06 (-0.05,0.17)
Q3: 0.00 (-0.05,0.17)
Q4: 0.04 (-0.06,0.14)
p-trend = 0.600
LDL
Per ln-unit: -0.116
(-0.262, 0.027)
Q2: 0.02 (-0.22, 0.26)
Q3: -0.05 (-0.28,0.18)
Q4: -0.11 (-0.36,0.14)
p-trend = 0.532
LDL/HDL ratio
Per ln-unit: -0.039
(-0.084, 0.007)
D-150
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Q2
Q3
Q4
-0.02 (-0.08, 0.04)
0.00 (-0.07, 0.07)
-0.08 (-0.18,0.02)
p-trend = 0.240
Results: Lowest quartile as reference group.
Confounding: Age, BMI at baseline, husband smoking, GDM, parity (nulliparous, muciparous), education, career, income, energy intake
and physical activity in the late term of pregnancy, gestational weeks, carbohydrate, protein, SFA, MUFA, and PUFA intake in the late
term of pregnancy.
Dalla Zuanna et Italy
al. (2021) 2017-2020
Medium for TC
HDL
Low for LDL
Cross-sectional
Pregnant women
ages 18-44 from
an area exposed
to PFAS through
drinking water
N = 319
I Trimester:
N = 101
II Trimester:
N = 88
III Trimester:
N = 130
Serum
2.7 (1.9-3.8)
I Trimester:
2.9 (2.2-3.9)
II Trimester:
2.5 (1.8-3.5)
III Trimester:
2.9 (1.8-4.2)
Levels (mg/dL) of TC,
HDL, and LDL
Regression TC
coefficient per In- Perln-unit: 3.01 (-4.51,
umt increase in
PFOS, or by
quartiles
10.53)
Q2: 4.42 (-8.21, 17.05)
Q3: -1.65 (-13.80,
10.50)
Q4: 9.89 (-2.82, 22.59)
HDL
Perln-unit: 4.84 (2.15,
7.54), p-value < 0.05
Q2: 8.60 (4.07, 13.14),
p-value < 0.05
Q3: 4.81 (0.49, 9.14),
p-value < 0.05
Q4: 9.20 (4.65, 13.76),
p-value < 0.05
LDL
Per ln-unit: -2.50
(-8.99, 3.98)
Q2: -2.76 (-13.73,
8.21)
Q3: -5.10 (-15.63,
5.43)
Q4: 0.01 (-11.04, 11.06)
First Trimester
D-151
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
TC: 15.34 (-1.08,
31.78)
HDL: 8.31 (1.07, 15.55),
p-value < 0.05
LDL: 6.65 (-5.90,
19.20)
Second Trimester
TC: -2.86 (-17.86,
12.13)
HDL: 3.76 (-3.35,
10.87)
LDL: -3.51 (-14.72,
7.69)
Third Trimester
TC: -4.51 (-18.13,
9.09)
HDL: 4.25 (0.26, 8.24),
p-value < 0.05
LDL: -10.05 (-22.71,
2.61)
Results: Lowest quartile as the reference group.
Confounding: Age, number of previous deliveries, BMI, physical activity, smoking habits, country of birth, education level, laboratory in
charge of the analyses of serum lipids, gestation weeks and reported fish consumption (in tertiles)
General Population
Lin et al. (2009)
Medium
United States
1999-2000
and 2003-
2004
Cross-sectional
Adults ages 20+
years from
NHANES
N = 969
Serum
Mean
(SEM) = 3.19
(0.04) loglO-
ng/mL
Metabolic syndrome
HDL cholesterol and
metabolic syndrome
triglycerides
OR per loglO-
unit increase in
PFOS
Metabolic syndrome
HDL cholesterol
Model 4: 1.47(1.07,
2.00), p-value < 0.05
Model 5: 1.61 (1.15,
2.26), p-value < 0.05
Metabolic syndrome
triglycerides
Model 4: 0.97 (0.73,
1.27)
D-152
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Nelson et al.
(2010)
Medium
Model 5: 0.86 (0.65,
1.16)
Outcome: Metabolic syndrome HDL cholesterol defined as HDL < 1.03 mmol/L in men and HDL < 1.29 mmol/L in women; metabolic
syndrome triglycerides defined as triglycerides > 1.69 mmol/L.
Confounding: Model 4: Age, sex, race, health behaviors (smoking status, alcohol intake, and household income), measurement data
(CRP and HOMA/insulin) and medications; additional confounding for model 5: Other components of the metabolic syndrome.
United States
2003-2004
Cross-sectional
Adults ages 20-
Serum Levels (mg/dL) of TC,
Regression
TC
80 yr from
21.0 HDL, non-HDL, LDL
coefficient per
Per unit increase: 0.27
NHANES
(Range = 1.4-
unit increase in
(0.05, 0.48)
N = 860
392.0)
PFOS, or by
Q4: 13.4 (3.8, 23.0)
quartiles
p-trend by
quartiles = 0.01
HDL
Per unit increase: 0.02
(-0.05, 0.09)
Non-HDL
Per unit increase: 0.25
(0.00, 0.50)
LDL
Per unit increase: 0.12
(-0.17,0.41)
Results: Lowest quartile used as the reference group.
Confounding: Age, sex, race/ethnicity, SES, saturated fat intake, exercise, time in front of a TV or computer, BMI, alcohol consumption,
and smoking.
Liu et al. (2018b) United States Cross-sectional Adults ages 18+ Serum
Medium 2013-2014 fromNHANES GM = 5.28
N = 1871
Levels of TC (mg/dL),
LDL (mg/dL), HDL
(mg/dL), TG (ln-
mg/dL)
Regression
coefficient (SE)
per ln-unit
increase in PFOS
TC: 1.22(1.91)
LDL: 0.88 (1.75)
HDL: 0.91 (0.70)
TG: -0.08 (0.05)
Confounding: Age, gender, ethnicity, smoking status, alcohol intake, household income, waist circumference, and medications
(antihypertensive, anti-hyperglycemic, and anti-hyperlipidemic agents)
Dong et al.
(2019)
United States
2003-2014
Cross-sectional
Adults age 20-80
from NHANES
Serum
Mean= 15.6
Levels (mg/dL) of TC,
LDL, HDL
Regression
coefficient per
TC all cycles: 0.4 (0.06,
0.6)
D-153
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Medium
N = 8,814
unit increase
PFOS
p-value < 0.05
Inconsistent associations
with LDL or HDL
across NHANES cycles.
Confounding: Age, gender, race, family income index, BMI, waist circumference, physical activities, diabetes status, smoking status,
number of alcoholic drinks per day
Jain et al.
United States Cross-sectional
Members of
Serum Levels (mg/dL) of TC,
Regression
TC: No clear
(2019d)
2004-2015
NHANES
GMs: LDL, HDL, TG
coefficient per
associations
Medium
Non-obese
Female = 7.4
loglO-unit
LDL
N = 1053 females
Male = 11.5
increase PFOS
OF: 0.0375 (0.0024,
(NF) and 1,237
0.0727)
males (NM)
p-value = 0.04
Obese N = 699
No clear associations in
females (OF) and
NF. NM, or OM
640 males (OM)
HDL: No clear
associations
TG
OF:-0.0912 (-0.153,
-0.0294)
p-value <0.01
No clear associations in
NF, NM, or OM
Confounding: Race/ethnicity, smoking status, age, poverty-income ratio (PIR), fasting time, use of lipid-lowering medicine, physical
exercise, survey year, daily dietary intake of total cholesterol, daily intake of total saturated fat, calories, caffeine, alcohol, protein intake
Fan et al. (2020) United States
Medium 2011-2014
Cross-sectional Adults age 20+ Serum Levels (mg/dL) of TC,
from NHANES Median = 5.14 LDL, HDL, and TG
N = 1,067 ng/mL
Regression
coefficient per
loglO-unit
increase in PFOS
TC: 3.85 (1.27,6.42)
p-value = 0.003
LDL: 3.02 (0.75, 5.29)
p-value = 0.009
HDL: 1.24 (0.32,2.16)
p-value = 0.009
TG: -0.01 (-0.04, 0.02)
p-value = 0.505
Confounding: Age, gender, race, education level, PIR, BMI, smoking status, alcohol use, energy intake levels, screen time
Jain and United States Cross-sectional Adults age 20+ Serum
Ducatman (2020) 2007-2014 from NHANES
Apolipoprotein B
(loglO-mg/dL)
Regression
coefficient per
Apolipoprotein B
D-154
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Medium
Non-diabetic
non-LLM users:
N = 2,872
Diabetic non-
LLM users:
N = 316
Non-diabetic
LLM users:
N = 519
Diabetic LLM
users: N = 293
Levels not
reported
loglO-unit
increase in PFOS
Non-diabetic non-LLM
users: 0.02027,
p-value = 0.02
Diabetic non-LLM
users: 0.01547,
p-value = 0.41
Non-diabetic LLM
users: -0.01327,
p-value = 0.40
Diabetic LLM users:
0.02001, p-value = 0.19
Steenland et al.
(2009)
Medium for TC,
HDL
Low for TG,
LDL
Confounding: Gender, age, age squared, race/ethnicity, PIR, fasting time in hours, loglO-transformed BMI, smoking status, survey year,
daily intake of cholesterol, caffeine, alcohol, total calories, total protein, and total fat
United States
2005-2006
Cross-sectional
Adults ages 18+
from the C8
Health Project,
current or former
residents from
areas supplied
with
contaminated
water
N = 46,494
Serum
19.6 (Range:
0.25-759.2)
Levels (ln-mg/dL) of
TC, LDL, HDL, non-
HDL cholesterol, and
triglycerides; TC/HDL
ratio; high TC
Lipid levels,
ratios:
Regression
coefficient per In- HDL
TC
0.0266 (SD = 0.0014)
umt increase in
PFOS
High TC:
OR by PFOS
quartiles
0.00355 (SD = 0.00173)
LDL
0.04172 (SD = 0.00221)
Triglycerides
0.01998 (SD = 0.00402)
TC/HDL ratio
0.02290 (SD = 0.00202)
Non-HDL
0.03476 (SD = 0.0019)
High TC
Q2: 1.14 (1.05, 1.23)
D-155
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Q3: 1.28 (1.19, 1.39)
Q4: 1.51 (1.40, 1.64)
p-trend < 0.0001
Outcome: High TC defined as > 240 mg/dL.
Results: Lowest quartile used as the reference group; lowest decile used as the reference group.
Confounding: Age, male gender, smoking status, education level, drinks alcohol, currently exercises, and BMI
Chateau-Degat et Canada
al. (2010) 2004
Medium
Cross-sectional
Nunavik Inuit
adults
Quartile analyses:
N = 716 (395
Plasma
GM (95%
confidence
interval): 18.6
women, 325 men) (17.8-19.5)
TC, TC/HDL
ratio: N = 663
LDL: N = 651
Non-HDL:
N = 670
HDL: N = 384
women, 309 men
Triacylglycerols:
N = 365 women,
284 men
Levels (mmol/L) of
TC, LDL, HDL, non-
HDL cholesterol, and
triacylglycerols;
TC/HDL ratio
Regression
coefficient per
unit increase in
PFOS or adjusted
mean by quartiles
TC
0.0009, p-value = 0.086
Ql: 4.781 (4.704, 4.864)
Q2: 4.869 (4.804, 4.940)
Q3: 4.969 (4.901,5.041)
Q4:5.301 (5.221,5.381)
p-trend < 0.0001
LDL
-0.002, p-value = 0.242
Ql: 2.750 (2.680,2.819)
Q2: 2.780 (2.730, 2.830)
Q3: 2.831 (2.770, 2.891)
Q4: 2.871 (2.801, 2.942)
p-trend = 0.58
HDL
Women: 0.0042,
p-value = 0.001
Men: 0.0016,
p-value <0.001
Ql: 1.539 (1.510, 1.572)
Q2: 1.619(1.580, 1.660)
Q3: 1.630 (1.580, 1.660)
Q4: 1.831 (1.788, 1.868)
p-trend < 0.0001
Non-HDL
-0.0011,
p-value = 0.315
D-156
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix,
N Levels3
Outcome
Comparison
Select Resultsb
Ql: 3.241 (3.160, 3.321)
Q2: 3.241 (3.182, 3.301)
Q3: 3.341 (3.271, 3.412)
Q4: 3.469 (3.388, 3.549)
p-trend = 0.09
Triacylglycerols
Women: -0.0014,
p-value = 0.04
Men: -0.0009,
p-value = 0.162
Ql: 1.051 (1.009, 1.092)
Q2: 1.067 (1.038, 1.096)
Q3: 0.941 (0.910, 0.970)
Q4: 1.000 (0.968, 1.030)
p-trend = 0.42
TC/HDL ratio
-0.0035,
p-value <0.001
Ql: 3.250 (3.181,3.320)
Q2: 3.210 (3.140, 3.281)
Q3: 3.240 (3.170,3.311)
Q4: 3.130 (3.049, 3.211)
p-trend = 0.75
Results: Adjusted means presented with lower and upper bounds of standard error in parentheses.
Confounding: Means adjusted for age, gender, BMI, and smoking status. All regression analyses adjusted for lipid-lowering drugs.
Additional regression analyses adjustments: TC: gender, smoking status, age and n-3 PUFAs; LDL: age, BMI, smoking status, and
insulinaemia; HDL: PFOS and n-3 PUFAs; non-HDL cholesterol: smoking status, age and gender; triacylglycerols: PFOS, smoking
status, BMI, stratified by gender; TC/HDL ratio: smoking status and gender
Eriksenetal. Denmark Cross-sectional Adults ages 50- Plasma Levels of TC (mg/dL) Regression 4.6(0.8,8.5)
(2013) 1993-1997 65 fromDCH Mean =36.1 coefficient per p-value = 0.02
Medium N = 753 IQR increase in
PFOS
D-157
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Confounding: Sex, education, age, BMI, smoking status, intake of alcohol, egg, and animal fat and physical activity
Fisher et al.
(2013)
Medium
Canada
2007-2009
Cross-sectional Adults ages 18- Plasma
Levels (ln-mmol/L) of Lipid levels,
TC
74 yr from
CHMS, cycle 1
N = 2,700
TC, HDL, Non-
HDL, TC/HDL
ratio: N = 2,345
LDL,
triglycerides:
N = 1,168
High cholesterol:
N = 1,042
GM
(SD) = 8.40
(2.04)
TC, HDL, LDL, non- TC/HDL ratio: 0.014 (-0.019, 0.05)
HDL, triglycerides;
TC/HDL ratio (In-
transformed); high
cholesterol
Regression
coefficient per ln-
unit increase in
PFOS
High cholesterol:
OR per ln-unit
increase in PFOS,
or by quartiles
HDL
-0.02 (-0.07, 0.02)
LDL
0.02 (-0.03, 0.08)
Non-HDL
0.03 (-0.11,0.07)
Triglycerides
-0.02 (-0.12,0.07)
TC/HDL ratio
0.04 (-0.008, 0.08)
High cholesterol
per ln-unit increase: 1.15
(0.89, 1.59)
Q2: 0.97 (0.58, 1.62)
Q3: 0.94 (0.58, 1.54)
Q4: 1.36(0.87,2.12)
p-trend = 0.13
Outcome: High cholesterol defined as TC > 5.2 mmol/L.
Results: Lowest quartile used as the reference group.
Confounding: Lipid levels, TC/HDL ratio: Age, sex, marital status,
cholesterol: Age, gender and alcohol consumption
BMI alcohol, smoking status and physical activity index; High
Fitz-Simon et al.
(2013)
Medium for TC,
HDL
United States
Baseline:
2005-2006;
Follow-up:
2010
Cohort
Adults ages 20-
60 from C8
Short-Term
Follow-up Study
living in West
Serum
Baseline GM
(SD) = 18.5
(13.5)
Levels (mg/dL) of TC,
LDL, HDL, and
triglycerides
Percentage TC: 3.20 (1.63, 4.76)
decrease (loglO R2 = 0.04
of final and initial LDL: 4.99 (2.46, 7.44)
ratio change per R2 = 0.07
HDL: 1.28 (-0.59,3.12)
D-158
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Low for TG,
LDL
Virginia and Ohio Follow-up
with PFOA- GM
contaminated (SD) = 8.2
drinking water (7.1)
N = 560 (N = 521
for LDL analysis)
log 10 of ratio
change in PFOS)
R2 = 0.04
Triglycerides: 2.49
(-2.88, 7.57)
R2 = 0.08
Confounding: Age, sex, interval between measurements, and fasting status
Donat-Vargas et
al. (2019b)
Medium
Sweden
1990-2013
Cohort
Non-diabetic Plasma
adults ages 30-60 Baseline
at baseline in median = 20
Vasterbotten Median at 10-
Intervention yr follow-
Programme (VIP) up = 15
N = 187
Levels (mmol/L) of TC Regression
and TG coefficient per 1-
SD change PFOS
or by tertiles
Per change in PFOS
TC
Baseline: -0.21 (-0.39,
-0.04)
Follow-up: 0.01 (-0.19,
0.21)
Prospective: 0.05
(-0.15,0.21)
TG
Baseline: -0.05 (-0.16,
0.06)
Follow-up: -0.15
(-0.28, -0.03)
Prospective: -0.14
(-0.27, -0.02)
Confounding: Gender, age, education, sample year, BMI, smoking habit, alcohol consumption, physical activity and healthy diet score
Lin et al. (2019)
Medium
United States Cohort and Prediabetic adults Plasma Levels (mg/dL) of TC,
1996-2014 cross-sectional age 25+from the Median = 27.2 LDL, HDL,
triglycerides, non-
DPP and
Outcomes Study
(DPPOS)
N = 940 (888 not
on metformin)
HDL, and very low-
density lipids (VLDL);
hypercholesterolemia,
hypertriglyceridemia
Regression
coefficient per
doubling PFOS
HR or OR for
hypercholesterole
mia or
hypertriglyceride
mia per doubling
of PFOS
Cross-sectional
TC: 2.53 (-0.10, 5.16)
LDL: 1.38 (-1.02, 3.77)
HDL:-0.40 (-1.19,
0.39)
Triglycerides: 7.75
(0.63, 14.88)
VLDL: 1.57 (0.24, 2.89)
Hypercholesterolemia at
baseline OR: 1.02 (0.85,
1.21)
D-159
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Hypertriglyceridemia at
baseline OR: 1.23 (1.03,
1.46)
Prospective
Hypercholesterolemia
HR: 1.01 (0.91, 1.12)
Hypertriglyceridemia
HR: 1.09 (0.93, 1.27)
Greater effect in the
placebo group
Confounding: Age, sex, race and ethnicity, marital status, educational attainment, drinking, smoking, percent of daily calorie from fat
intake, daily fiber intake, physical activity level, and waist circumference at baseline
Canova et al.
(2020)
Medium
Italy
2017-2019
Cross-sectional
Residents of
PFAS "Red
Area" with
contaminated
public water
supply ages 20-
39
N = 15720 (7,620
female, 8100
male)
Serum
Median = 3.7
Female = 3
Male = 4.8
Levels (mg/dL) of TC,
LDL, HDL, non-HDL,
and triglycerides
Regression TC
coefficient per In- 4.99 (4.12,5.86)
umt increase
PFOS or by
quartile, or by
decile
p-value for interaction
by sex = 0.39
Consistently increased
associations by deciles,
from 4.33 to 11.77
LDL
3.97 (3.21,4.73)
Males: 5.07 (3.87, 6.27)
Females: 2.43 (1.47,
3.39)
p-value for interaction
by sex = 0.003
Associations for deciles
2-10 consistently
increase from 2.94 to
9.67
HDL
1.43 (1.1, 1.76)
Males: 0.91 (0.47, 1.36)
D-160
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Females: 1.95 (1.46,
2.45)
p-value for
associations = 0.001
Associations for deciles
2-10 moderately
increase from 1.13 to
3.43
Triglycerides
0 (-0.01, 0.01)
p-value for
associations = 0.954
Associations for deciles
2-10 inconsistently vary
from 0 to 0.02
Results: Lowest quartile or decile used as reference group.
Confounding: Age, BMI, time lag between enrollment and beginning of study, physical activity, smoking habits, country of birth,
alcohol consumption, education level, laboratory in charge of analyses, reported food consumption
Lin et al. (2020c) Taiwan
Medium 2016-2017
Cross-sectional
Adults aged 55 to
75 that resided in
the study area for
more than 10 yr
and not taking
lipid-lowering
medication
N = 352
Serum Levels (mg/dL) of TC, Regression
16.2(10.1- HDL, LDL, and coefficient by
24.1) triglycerides quartiles
TC
Q2: 15.06 (4.66, 25.46),
p-value < 0.05
Q3: 11.47 (1.03,21.91),
p-value < 0.05
Q4: 10.18 (-0.59, 20.94)
p-trend = 0.11
HDL
Q2: 3.23 (-0.79, 7.24)
Q3: 1.92 (-2.11,5.95)
Q4: -2.68 (-6.84, 1.47)
p-trend = 0.19
LDL
Q2: 13.43 (4.05, 22.80),
p-value < 0.05
D-161
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Q3: 12.32 (2.91,21.73),
p-value < 0.05
Q4: 15.29 (5.59, 24.99),
p-value < 0.05
p-trend = 0.004
Triglycerides
Q2: 8.93 (-9.74, 27.59)
Q3: 7.58 (-11.16, 26.31)
Q4: 6.76 (-12.55, 26.07)
p-trend = 0.53
Results: Lowest quartile used as the reference group.
Confounding: Age, sex, smoking status, and drinking status
Liu et al. (2020a) United States Randomized Adults from Plasma
Medium 2004-2007 clinical trial POUNDS Lost 23.5
study ages 20+
N = 326
Levels (mg/dL) of TC,
triglycerides, and
apolipoproteins loglO-
ApoB, ApoE, and
ApoC-III
Least-squared
means (LSM) by
tertile PFOS
TC
Tl: 180.9 (8.0)
T2: 189.3 (7.9)
T3: 190.7 (7.3)
p-trend = 0.21
Triglycerides
Tl: 126.8 (11.6)
T2: 132.4 (11.4)
T3: 126.1 (10.5)
p-trend = 0.80
Results: LSM are presented with standard error in parentheses.
Confounding: Age, sex, race, educational attainment, smoking status, alcohol consumption, physical activity, BMI, regular lipid-
lowering medication use, dietary intervention groups
Han et al. (2021)
Medium
China
2016-2017
Case-control
Adults ages 25 to
74 including type
2 diabetes cases
and healthy
controls
N = 304
Serum
Cases: 7.60
(4.47-10.55)
Controls: 8.45
(5.40-11.95)
Levels (loglO-mmol/L) Regression
of TC, HDL, LDL, and coefficient per
triglycerides loglO-unit
increase in PFOS
TC: 0.06 (-0.01,0.12)
HDL -0.02 (-0.09,
0.05)
LDL: 0.12(0.03,0.21),
p-value < 0.05
Triglycerides: 0.03
(-0.13,0.18)
Confounding: Age, sex, BMI.
D-162
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix,
Levels"
Outcome
Comparison
Select Resultsb
Jeddi et al.
(2021a)
Medium
Italy
2017-2019
Cross-sectional
Residents aged
20-39 from the
PFAS-
contaminated
Veneto region
N = 15,876
Serum
GM (range):
4.54 ( 175 mg/dL.
Confounding: Age, gender, time lag between the beginning of the study and blood sampling center where BP has been measured,
education, number of deliveries, physical activity, country of birth, diet, alcohol intake, and smoking status, and other components of
metabolic syndrome
Occupational Populations
Olsen et al.
(2003a)
Medium
United States, Cross-sectional Current and
Serum
Levels of cholesterol Comparison of No significant
Belgium
1994-2000
former workers at Antwerp
two
fluorochemical
production plants
Male
N = 421,
Female
N = 97,
Regression
analysis
N = 174
Mean
(SD) = 0.96 p
pm (0.97);
Decatur = 1.4
0 ppm (1.15)
(ln-mg/dL), HDL
(mg/dL)
mean outcome by
PFOS exposure
quartile
Regression
coefficient per
unit increase in
PFOS
differences between
mean cholesterol or
HDL by quartile among
male and female
employees
Cholesterol
0.01 (-0.005,0.025)
Confounding: Age, BMI, drinks/day, cigarettes/day, location, entry period, baseline years worked
Notes: ALSPAC = Avon Longitudinal Study of Parents and Children; ALT = alanine aminotransferase; APFO = ammonium perfluorooctanoate; ApoB = Apolipoprotein B;
ApoE = Apolipoprotein E; ApoC-III = Apolipoprotein C-III; BMI = body mass index; CHMS = Canadian Health Measures Survey; DCH = Diet, Cancer and Health;
DPPOS = Diabetes Prevention Program and Outcomes Study; EYHS = European Youth Study; GDM = gestational diabetes; GM = geometric mean; HDL = high-density lipids;
HELIX = Human Early-Life Exposome; HOME = Health Outcomes and Measures of the Environment; hr = hours; HR = hazard ratio; INMA = INfancia y Medio Ambiente
(Environment and Childhood) Project; IQR = interquartile range; KorEHS-C = Korea Environmental Health Survey in Children and Adolescents; LDL = low-density lipids;
LLM = lipid lowering medication; mo = months; MoBa = Norwegian Mother and Child Cohort Study; MUFA = monounsaturated fatty acid; NF = non-obese female;
NHANES = National Health and Nutrition Examination Survey; NM = non-obese male; OF = obese female; OM = obese male; OR = odds ratio; PFHxS = perfluorohexane
sulfonic acid; PFNA = perfluorononanoic acid; PIR = poverty income ratio; POUNDS = Preventing Overweight Using Novel Dietary Strategies; PUFA = polyunsaturated fatty
acid; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; Q5 = quartile 5; S-MBCS = Shanghai-Minhang Birth Cohort Study; SD = standard deviation;
SE = standard error; SEM = serum mean; SES = socioeconomic status; SFA = saturated fatty acid; T1 = tertile 1; T2 = tertile 2; T3 = tertile 3; TC = total cholesterol;
TFF1 = Tromse Fit Futures 1; TG = triglycerides; VIP = Vasterbotten Intervention Programme; VLDL = very low-density lipoprotein; wk = weeks; yr = year(s).
D-163
-------
APRIL 2024
a Exposure reported as median (25th-75th percentile) in ng/mL unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D.6 Endocrine
Table D-15. Associations Between PFOS Exposure and Endocrine Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels Outcome
(ng/mL)a
Comparison
Select Resultsb
General Population
Lebeaux et al.
(2020)
High for cord
serum thyroid
hormones;
Medium for
maternal thyroid
hormones
United States
2003-2007
Cohort
Mother-infant Cord serum
pairs from
Health Outcome
Measures of the
Environment
(HOME) Study
N = 256 for
cord serum
N = 185 for
14.3
Maternal serum
5.5
Levels of TSH
(jjIU/L), TT4
(|ig/dL). TT3
(ng/dL), FT4
(ng/dL), and FT3
(pg/mL)
Regression Cord serum
coefficient per TSH: 0.09 (-0.06, 0.25)
log2-unit TT4: 0.01 (-0.04, 0.07)
increase in TT3:-0.02 (-0.10, 0.06)
PFOS FT4: -0.02 (-0.06, 0.02)
FT3: -0.03 (-0.07,0.02)
Maternal serum
TSH: 0.02 (-0.24, 0.28)
maternal serum
TT4
0.02 (-0.08, 0.08)
TT3
-0.02 (-0.07, 0.03)
FT4
0.02 (-0.02, 0.07)
FT3
-0.03 (-0.06, 0.00)
Confounding: Individual PFAS, maternal age at delivery, race/ethnicity, marital status at baseline, maternal education level, household
income, mean loglO-transformed cotinine, maternal alcohol usage during pregnancy, nulliparity, maternal BMI based on pre-pregnancy
weight in pounds, child's sex, gestational week at blood draw for PFAS measurement, and (for cord serum only) delivery mode
Blake et al. Fernand, Ohio, Cohort FCC
(2018) USA Median age
Medium 1991-2008 38yrat
enrollment,
N = 122 for
TSH
measurements;
47 male and 75
female
N = 144 for TT4
measurements;
Drinking water
Serum
28.4
Levels of
TSH (ln-|iIU/mL),
TT4 (ln-|ig/dL)
Percent TSH
change per 9.75 (1.72, 18.4), p-value = 0.02
IQR increase Males: 21.4 (6.55, 38.3)
in PFOS p-value = 0.01
Females: 5.13 (-5.29, 16.7)
p-value = 0.36
TT4
-0.51 (-4, 3.1), p-value = 0.78
Males: -5.29 (-10.1,-0.26),
p-value = 0.04
D-164
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
63 males and 81
females
Females: 1.69 (-3.28,6.91),
p-value = 0.52
Confounding: Age, year of measurement, sex, education, income, marital status, BMP
Jain and
Ducatman
(2019b)
Medium
United States
2007-2012
Cross-sectional
Adults from
NHANES aged
20+
GF status:
GF-1 = 1,653
GF-2 = 720
GF-3A = 114
GF-3B/4 = 62
Serum
Levels not
reported
Levels of
TSH (log-
(iIU/mL),
TGN (log-ng/mL),
TT4 (log-ng/dL),
FT4 (log-ng/dL),
TT3 (log-ng/dL),
FT3 (log-pg/mL)
Regression TT4
coefficient per GF-1: 0.002, p-value = 0.76
loglO-unit GF-2: -0.008, p-value = 0.47
increase in GF-3A: 0.058, p-value = 0.02
PFOS GF-3B/4: -0.002, p-value = 0.94
GF Stages: GF-1: GFR > 90 mL/min/1.73 m2; GF-2: GFR between 60 and
60 mL/min/1.73 m2; GF-3B/4: GFR between 15 and 45 mL/min/1.73 m2
Confounding: Gender, race/ethnicity, iodine deficiency status, age, BMI, fasting time.
smoking status, use of drugs
90 mL/min/1.73 m2; GF-3A: GFR between 45 and
PIR, total calories consumed during the last 24 hr,
Jain (2013)
Low
United States
2007-2008
Cohort
Adults and
children from
NHANES aged
12+
N= 1,540
including
children
Serum
Total cohort
Levels of
TSH (|iIU/L),
FT3 (pg/L),
TT3 (fg/dL),
FT4 (pg/L),
TT4 (pg/L),
TGN
Regression TSH, FT3, FT4, TT3, TT4, TGN:
coefficient per No statistically significant
loglO-unit associations
increase in
PFOS, or by
tertiles
Results: Lowest tertile used as the reference group.
Confounding: Gender, race, age, iodine deficiency, iodine replete
Lewis et al. United States Cross-sectional Men and Serum Levels of Percent TSH
(2015) 2011-2012 womenfrom Males 20-40: TSH ((iIU/mL), change per Males
Low NHANES ages 7.75 TT3 (ng/dL), doubling of 20 to < 40:-2.9 (-8.6,3.2)
20-80 Males 40-60: FT3 (pg/mL), PFOS 40 to < 60:-1.3 (-8.9,7.1)
699 men 9.28 TT4 (|ig/mL), 60 to 80: -2.3 (-9.4,5.3)
680 women Males 60-80: FT4 (ng/dL) Females
11.1 20 to <40:-1.0 (-7.9, 6.4)
Females 20-40: 40 to < 60: 0.0 (-7.1, 7.7)
4.20 60 to 80:-1.5 (-9.6, 7.3)
Females 40-60:
4.93 FT4
D-165
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Females 60-80:
9.50
Females
20 to <40: 2.2 (0.5,3.9)
p-value < 0.05
40 to <60: 1.3 (-0.5,3.2)
60 to 80:-0.5 (-2.5, 1.5)
Males: No statistically significant
associations
TT3, FT3, TT4: No statistically
significant associations
Confounding: Age, BMI, PIR, serum cotinine, and race/ethnicity
Li et al. (2017) China
Low 2013-2014
Cross-sectional
Residents of Serum
Southern China, 1.3
ages 1 mo to
90 yr, 70% with
thyroid
condition
N = 202
Levels of Regression TSH: 0.41 (0.05, 0.76),
TSH ((iIU/mL), coefficient per p-value = 0.024
FT3 (pmol/L), log-unit IQR FT3: -0.14 (-0.24, -0.04),
FT4 (pmol/L) increase in p-value = 0.007
PFOS FT4: -0.13 (-0.22, -0.04),
p-value = 0.004
Comparison: Logarithm base not specified.
Confounding: Age, sex
Byrne et al.
(2018)
Low
St. Lawrence
Island, Alaska,
USA
2013-2014
Cross-sectional
Alaska Natives,
aged 18-45
N = 85
38 men
47 women
Serum
4.55
Males: 6.81
Females: 3.35
Levels of
TSH (ln-|iIU/mL),
TT3 (pg/mL),
FT3 (ng/dL),
TT4 (ng/dL),
FT4 (ng/dL)
Regression TSH
coefficient per Males: -0.06 (-0.62, 0.51),
ln-unit
increase in
PFOS
p-value = 0.085
Females: No association
TT3
Males: -10.54 (-22.28, 1.20),
p-value = 0.08
Females: No association
FT3
Males: -0.30 (-0.53, 0.07),
p-value = 0.01
Females: 0.35 (0.05, 0.65)
p-value for sex interaction = 0.02
D-166
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Select Resultsb
TT4, FT4: No statistically
significant associations
Confounding: Age, sex, smoking status
Zhang et al.
(2018b)
Low
China
2013-2016
Cross-sectional Women aged Plasma
20-40 yr, with
(cases) or
without
(controls) POI
N= 120
Cases: 8.18
Controls: 6.02
Levels (ng/mL) of Regression TSH
TSH, FT3, FT4 coefficient per POI cases: 1.57 (0.65, 2.5)
log-unit POI controls: 0.67 (0.08, 1.26)
increase in
PFOS FT3
POI cases -0.88 (-1.64, -0.09)
FT4
POI cases -2.99 (-4.52, -1.46)
FT3 and FT4 in POI controls: No
associations
Comparison: Logarithm base not specified.
Confounding: Age, BMI, education, income, sleep, and parity
Children
Xiao et al.
(2019)
High
Faroe Islands,
Denmark
1994-1995
Cohort
Pregnant
women and
their infant
children
N = 172 and
153 for
measurements
in maternal and
cord serum,
respectively
Maternal blood
Geometric
mean = 20.86 |i
Cord serum levels
of TSH (log-
IU/L),
T4 (log-pmol/L),
FT3 (log-pmol/L),
FT4, (log-pmol/L)
FT3 resin uptake,
FT4 index (FTI)
(log-IU/L)
Regression TSH
coefficient per All children: 39.7 (7.9, 80.9)
log2-unit
increase in
PFOS
Boys: 39.5 (0.4, 94.1)
Girls: 39.9 (-4.1, 104.2)
FTI
All children: 6.7 (-1.5, 15.6)
Boys: 2.1 (-7.7, 13)
Girls: 13.2(0.9, 27.1)
T4, FT3, FT4, FT3 resin uptake: No
statistically significant associations
Confounding: Child sex (in detailed results), parity,
alcohol during pregnancy, total PCB, mercury
maternal BMI, maternal height, maternal education, maternal age, smoking and drinking
D-167
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Select Resultsb
Kim et al.
(2020a)
High
South Korea
2012-2017
Cohort
Children, aged
2, 4, 6 yr
N = 511 for age
6 (268 boys)
Serum
Age 2: 4.530
Age 4: 4.050
Age 6: 3.980
Levels of
TSH (ln-jjIU/mL),
FT4 (ln-ng/dL),
and T3 (ln-ng/dL)
at age 6
Subclinical
hypothyroidism
Regression
coefficient per
ln-unit
increase in
PFOS
Subclinical
hypothyroidis
m: OR per
increase in
PFOS
T3 at age 6
All: 0.04 (0.017), p-value < 0.05
Boys: 0.04 (0.018), p-value < 0.05
No interaction with sex
Subclinical hypothyroidism at age 6
All: 0.36(0.41,0.96)
Boys: 0.24 (0.07, 0.92)
No interaction with sex
TSH, FT4: No statistically
significant associations between or
within age groups
Results: Comparisons for T3 are presented with standard error in parentheses.
Confounding: Age, sex, dietary iodine intake
Kato et al.
Japan Cross-sectional Pregnant
Maternal serum Levels of
Regression
TSH
(2016)
2002-2005 women and
Male: 5.2 TSH (loglO-
coefficient per All infants: 0.18, p-value = 0.001
Medium
their children
Female: 5.3 (iU/mL),
loglO-unit
Increasing trend in LSM by
N = 392
FT4 (loglO-
increase in
quartiles p-trend = 0.024
Male
ng/mL)
PFOS
Males: 0.21, p-value = 0.014
children = 180
Females: 0.17, p-value = 0.021
Female
LSMby
children = 212
quartile
FT4: No statistically significant
associations
Confounding: Maternal age at delivery, BMI, parity, educational level, thyroid antibody, intake of seaweed, blood sampling period
before/after delivery for PFOS and PFOA, and gestational week at which blood sampling was obtained for TSH and FT4
Preston et al.
(2018)
Medium
United States Cohort
Pregnant
Maternal plasma Levels of
Regression
T4, all neonates:
1999-2002
women and
23.5
T4 (ng/dL)
coefficient by
Q2
-0.63 (-1.64, 0.37)
their children
quartiles
Q3
-0.36 (-1.36, 0.67)
N = 465
Q4
-1.1 (-2.13,-0.07)
neonates (236
male, 229
T4, males:
female)
Q2
Q3
Q4
-1.56 (-3.04, -0.08)
-1.7 (-3.28, -0.12)
-2.2 (-3.74, -0.66)
D-168
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Select Resultsb
No associations in newborn females
Results: Lowest quartile used as the reference group.
Confounding: Maternal age, race/ethnicity, smoking status, fish intake, parity, and gestational week at blood draw
Aimuzi et al.
China Cross-sectional Pregnant Cord blood
Levels of
Regression TSH
(2019)
2012-2013 women and 2.51
TSH (ln-mlU/L),
coefficient per All children: -0.05 (-0.08, -0.02)
Medium
their children
FT3 (pmol/L),
ln-unit Boys: -0.047 (-0.097, 0.003)
N = 567
FT4 (pmol/L)
increase in Girls: -0.048 (-0.093, -0.003)
Male
PFOS
children =305
Female
children = 262
Confounding: Maternal age, fish intake, parity infant sex, gestational age at delivery, and maternal pre-pregnancy BMI
Itoh et al.
Japan Cohort Pregnant Plasma
Levels of
Regression TSH
(2019)
2003-2005 women and 6.21
TSH (ln-|iU/mL),
coefficient per All boys: 0.23 (0.07, 0.39),
Medium
their children
FT3 (ln-pg/mL),
ln-unit p-value = 0.004
365 male
FT4 (ln-pg/mL),
increase in Boys with TA-negative mothers:
children
TPOAb (ln-
PFOS 0.39 (0.12, 0.66), p-value = 0.005
336 female
IU/mL),
children
TgAb (ln-IU/mL)
No significant association among
TA-positive mother-infant pairs
Confounding: Age at delivery, parity, educational level, alcohol consumption, smoking during pregnancy, pre-pregnancy BMI, logFT4
Tsai (2017)
Taiwan Cross-sectional Newborns from Cord blood
Levels of
Regression TSH, all newborns:
Low
2004-2005 Taiwan Birth Mean = 7.24
TSH ((iIU/mL),
coefficient by Q2: 0.21 (-0.20, 0.63)
Panel Study
T3 (ln-|ig/dL).
quartiles or Q3:0.19 (-0.22, 0.61)
(TBPS)
T4 (ng/dL)
per ln-unit Q4: 0.65 (0.02, 1.28)
N= 118(64
increase in Per increase: 0.35 (0.10, 0.59)
boys, 54 girls)
PFOS
TSH, boys:
Q2: 0.63 (0.04, 1.22)
Q3: 0.30 (-0.33, 0.94)
Q4: 0.75 (0.13, 1.62)
Per increase: 0.33 (0.01, 0.68)
D-169
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
T4, all newborns:
Q2
Q3
Q4
-0.50 (-1.29, 0.29)
-0.28 (-1.08,0.51)
-1.03 (-2.17,-0.12)
Per increase: -0.46 (-0.92, -0.001)
T4, boys:
Q2
Q3
Q4
-0.30 (-1.40, 0.80)
0.19 (-0.99, 1.36)
-2.12 (-3.62,-0.618)
Per increase: -0.67 (-1.28, -0.05)
Results: Lowest quartile used as the reference group.
Confounding: Maternal age at delivery, newborn sex, maternal BMI, maternal education, gestational age, and delivery type
Pregnant Women
Dreyer et al.
(2020)
High
Denmark
2010-2012
Cohort
Pregnant
women from
Odense Child
Cohort (OCC)
N = 1,048
Serum
7.64
Levels of diurnal
urinary (dU)
Cortisol (nmol/24-
hr), dU-cortisone
(nmol/24-hr), dU-
cortisol/cortisone,
serum Cortisol
(nmol/L)
Percent dU-cortisone: -9.1 (-14.7, -3.0),
change per 2- p-value < 0.05
fold increase T2: -5.7 (-14.7, 4.2)
inPFOS T3:-16.0 (-23.9,-7.2),
p-value < 0.05
p-trend < 0.01
dU-cortisol/cortisone: 9.3 (3.3,
15.6), p-value < 0.05
T2: 11.0(1.8, 21.1), p-value <0.05
T3: 16.6 (6.9, 27.1), p-value < 0.05
p-trend < 0.01
dU-cortisol and serum Cortisol: No
statistically significant associations
Confounding: Age, parity, and offspring sex
Xiao et al
(2019)
High
Faroe Islands, Cross-sectional Pregnant Maternal blood Maternal serum Regression TSH in maternal serum
Denmark women and Geometric levels of TSH coefficient per All children: 16.4 (-7.5,46.5)
1994-1995 their children mean = 20.86 (i (log-IU/L), log2-unit Boys:-6 (-29.6,25.4)
g/g T4 (log-pmol/L), Girls: 54.2 (11.3, 113.8)
D-170
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Select Resultsb
Maternal age 28
FT3 (log-pmol/L),
increase in
(SD = 5.6)
FT4 (log-pmol/L)
PFOS
T4, FT3, FT4, FT3 resin uptake,
FT4 index: No statistically
N = 172 and
FT3 resin uptake
significant associations
153 for
FT4 index
measurements
in maternal and
cord serum,
respectively
Confounding: Child sex (in detailed results), parity, maternal BMI, maternal height, maternal education, maternal age, smoking and drinking
alcohol during pregnancy, total PCB, mercury
Berg (2017)
Norway Cohort Pregnant
Serum
Levels of
Regression
TSH
Medium
2007-2009 or women and
8.03
TSH (mlU/L),
coefficient by
Q2: 0.04 (-0.03,0.11)
until 3 d after children from
FT3 (pmol/L),
quartiles
Q3: 0.08 (0.01,0.15)
birth the Norway
T3 (nmol/L),
Q4: 0.10 (0.02, 0.17)
Mother and
FT4 (pmol/L),
Child
T4 (nmol/L)
T3,T4, FT3, or FT4: No
Contaminant
statistically significant associations
Cohort Study
(MISA)
N = 370
Results: Lowest quartile used as reference group.
Confounding: Parity, t-uptake
Preston et al.
United States Cross-sectional Pregnant
Maternal plasma Levels of
Percent
TSH among TPOAb-positive
(2018)
1999-2002 women and
24.0
TSH (mlU/mL),
difference in
mothers: -16.4 (-29.8, -0.38)
Medium
their children
T4 (ng/dL),
hormone level
p-value for effect modification by
FT4 index
per IQR
TPOAb status = 0.05
N = 718 women
increase in
(98 TPOAb-
PFOS
FT4, TT4: No statistically
positive and 620
significant associations
TPOAb-
negative)
Confounding: Maternal age, race/ethnicity, smoking status, fish intake, parity, and gestational week at blood draw
Reardon et al.
Canada Cohort Pregnant
Maternal blood
Levels of
Regression
TSH, linear PFOS
(2019)
2019-2012 women
coefficient per Main effect: 0.01 (-0.03, 0.04)
D-171
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Select Resultsb
Medium
recruited prior Total PFOS:
to 18 wk of
gestation
N = 478
4.77
Linear PFOS:
2.49
XBr-PFOS:
1.08
TSH (log-
mlU/mL),
FT3 (log-pmol/L),
FT4 (log-pmol/L)
by gestation status
and 3 mo
postpartum
unit increase
in total,
linear, or 1 HI-
PPOS
3 mo postpartum: 0.06 (0.01, 0.12)
TSH, XBr-PFOS
Main effect: 0.29 (0.02, 0.56)
FT3, FT4: No statistically
significant associations
Confounding: Maternal age, ethnicity, history of smoking, history of drug and alcohol use
Kato et al. Japan Cross-sectional Pregnant
(2016) 2002-2005 women and
Low their children
N = 392
Male
children = 180
Female
children = 212
Maternal serum Levels of
Male: 5.2 TSH (loglO-
Female: 5.3
(iU/mL),
FT4 (loglO-
ng/mL)
Regression TSH
coefficient per All mothers: -0.21, p-value < 0.001
loglO-unit Decreasing trend in LSM by
increase quartiles: p-trend < 0.001
PFOS Male: -0.25, p-value = 0.002
Female: -0.21, p-value = 0.005
LSM by
quartile FT4: No statistically significant
associations
Confounding: Maternal age at delivery, BMI, parity, educational level, thyroid antibody, intake of seaweed, blood sampling period
before/after delivery for PFOS and PFOA, and gestational week at which blood sampling was obtained for TSH and FT4
Notes: BMI = body mass index; d = day(s); FCC = Fernald Community Cohort; FTI = free thyroxine index; GF = glomerular filtration; GFR = glomerular filtration rate;
HOME = Health Outcomes and Measures of the Environment; IQR = interquartile range; LSM = least square means; MtSA = Norway Mother and Child Contaminant Cohort
Study; NHANES = National Health and Nutrition Examination Survey; OCC = Odense Child Cohort; TSH = thyroid stimulating hormone; T3 = triiodothyronine; T4 = thyroxine;
FT3 = free triiodothyronine; FT4 = free thyroxine; hr = hour(s); mo = month(s); PCBs = polychlorinated biphenyls; PIR = poverty income ratio; POI = premature ovarian
insufficiency; POUNDS = Preventing Overweight Using Novel Dietary Strategies; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; T3 = tertile 3; T4 = tertile ; TBPS = Taiwan
Birth Panel Study; TgAb = thyroglobulin antibody; TPOAb = thyroid peroxidase antibody; TT3 = total triiodothyronine; TT4 = total thyroxine; TGN = thyroglobulin;
USA = United States of America; wk = week(s); yr = years.
a Exposure levels are reported as median unless otherwise noted.
b Results reported as effect estimate (95% confidence interval), unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D-172
-------
APRIL 2024
D.7 Metabolic/Systemic
Table D-16. Associations Between PFOS Exposure and Metabolic Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
„ , Exposure
°')U a I"n' Matrix, Levels Outcome
geS' (ng/mL)a
Comparison
Resultsb
Children and Adolescents
Ashley-Martin
etal. (2017)
High
Canada,
Recruitment
2008-2011
Cohort
Pregnant
women and
their children,
from the
MIREC Study
N = 1,175
Maternal blood Adiponectin,
4.6 leptin
Regression
coefficient per
loglO-unit
increase in
PFOS
Adiponectin, leptin: No statistically
significant associations
Confounding: Maternal age, pre-pregnancy BMI, sex, and parity0
Buck et al.
(2018)
High
Chen et al.
(2019b)
High
United States,
2003-2006
Cohort
Pregnant
women and
their children in
the HOME
study
N = 230
Maternal serum
14
Adiponectin,
leptin
Percent change
per doubling of
PFOS
Adiponectin, leptin: No statistically
significant associations
Confounding: Maternal age, race, education, income, parity, maternal BMI, serum cotinine, delivery mode, and infant sex
China, Cohort Infants followed Cord blood
2012-2017 up at age 5, 2.44
N = 404
BMI, WC, body Regression BMI, waist circumference, body
fat, waist-to- coefficient per fat, waist-to-height ratio: No
height ratio ln-unit increase statistically significant association
in PFOS, or by
tertile
Confounding: Maternal age,
pregnancy, and parity
maternal pre-pregnancy BMI, gestational week at delivery, maternal education, paternal smoking during
Jensen et al.
(2020a)
High
Denmark,
2010-2012
Cohort
Pregnant
women and
their infants
assessed at
birth, 3 mo, and
18 mo, Odense
Child Cohort
N = 593
Maternal serum
8.04
BMI z-score,
WC
Regression
coefficient per
unit increase in
PFOS
BMI z-score, WC: No statistically
significant associations
Confounding: Maternal age, parity, pre-pregnancy BMI, pre-pregnancy BMI2, education, smoking, sex, visit, adiposity marker at birth
D-173
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Minatoya et al.
(2017)
High
Alderete et al.
(2019)
Medium
Japan,
2002-2005
Cohort
Pregnant
women and
their children
N = 168
Serum
5.1
Adiponectin,
leptin
Regression
coefficient per
loglO-unit
increase in
maternal serum
PFOS
Adiponectin: 0.12 (0.01, 0.22),
p-value = 0.028
Leptin: No statistically significant
association
Confounding: Maternal BMI, parity, smoking during pregnancy, blood sampling period, gestational age, infant sex
United States, Cohort Obese Hispanic Plasma
2001-2012 children (8- 12.22
14 yr), SOLAR
Project
N = 38
Blood glucose,
insulin, 2-hr
glucose
(mg/dL)), 2-hr
insulin, insulin
resistance,
insulin levels
Regression
coefficient per
ln-unit increase
in PFOS
Glucose (2-hr)
6.2 (-2.3, 14.8)
Blood glucose, insulin, 2-hr insulin,
insulin resistance, insulin levels: No
statistically significant associations
Confounding: Sex, baseline social position (categorical), baseline outcome, baseline and change in age at follow-up, pubertal status
(categorical), baseline and change in body fat percent at follow-up.
Braun et al.
(2016)
Medium
United States,
2003-2006,
follow-up at age
Cohort
Pregnant
women and
their children in
the HOME
study
N = 204
Maternal serum
13
Overweight,
obesity, BMI z-
score, waist
circumference,
body fat
Percent change Overweight, obesity, BMI z-score,
per doubling of waist circumference, body fat: No
PFOS statistically significant associations
Confounding: Maternal age, race, education, income, parity, marital status, employment, depressive symptoms, BMI at 16 wk gestation,
fruit/vegetable consumption, fish consumption, prenatal vitamin use, maternal serum cotinine concentrations, and child age in months
Conway et al. United States, Cross-Sectional Children
(2016) 2005-2006 working or
Medium living in six
PFOS-
contaminated
water districts,
C8 Health
Project
N = 47
Serum Type 1 Diabetes OR per ln-unit
Mean =86.5 increase in
PFOS
Children with T1D: 0.52 (0.54,
0.87)
Confounding: Age, sex, race, BMI, eGFR, hemoglobin, iron
D-174
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Domazet et al. Denmark,
(2016) 1997-2009
Medium
Cohort
Children from
EYHS followed
through ages 9,
15, and 21,
N = 176
Plasma
Age 21
Males: 11.9
Females: 9.1
Age 15
Males: 22.3
Females: 20.8
Age 9
Males: 44.5
Females: 39.9
WC, HOMA-
Beta, HOMA-
IR, insulin,
glucose,
skinfold
thickness, BMI
Percent change
at 15 or 21 yr
old per 10-unit
increase in
PFOS at 9 yr
old
WC:
Age 15 from age 9:
1.18 (0.42, 1.84)
Age 21 from age 9:
1.52 (0.05,2.91)
Skinfold thickness:
Age 15 from age 9:
4.03 (1.33,6.67)
Age 21 from age 9:
5.67 (0.6, 10.93)
BMI:
Age 15 from age 9:
1.54 (0.62, 2.4)
HOMA-Beta age 21, BMI age 21,
HOMA-IR, insulin, glucose: No
statistically significant associations
Confounding: Sex, age, and outcome levels at baseline (9 yr of age), and ethnicity, maternal parity, and maternal income in 1997 (9 yr of
age). Waist circumference was adjusted for height in order to account for body size.
Domazet et al.
(2020)
Medium
Denmark, 1997 Cross-sectional
Children from Plasma
EYHS, 9-yr-old Boys: 42.9
N = 242 Girls: 42.0
Body fat (mm),
adiponectin
(ng/mL), leptin
(pg/mL)
Percent change
per 10%
increase in
PFOS
Body fat: -0.59 (-2.88, 1.24),
p-value = 0.552
Adiponectin: 0.24 (-1.70, 2.21),
p-value = 0.811
Leptin: -3.65 (-8.23, 1.16),
p-value = 0.134
Confounding (Adiponectin and leptin): Sex, age, parity, maternal income level
Confounding (Body fat): Sex, age, accelerometer wear time, parity, maternal income level
Gyllenhammar Sweden, Cohort Mothers and Maternal serum BMI z-score Regression BMI z-score:
etal. (2018b) 1996-2011, their children 13 coefficient per Ages 36 Non-significant positive
Medium children from the IQR increase in association (numeric results not
followed up at POPUP Study maternal PFOS provided)
age 5 N=381 Ages 48 and 60 mo: Positive
statistically significant associations.
D-175
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: Sampling year, maternal age, pre-pregnancy BMI, maternal weight gain during pregnancy, maternal weight loss after
delivery, years of education, and total time of breastfeeding
Hartman et al.
(2017)
Medium
United
Kingdom,
recruitment
1991-1992
Cohort
Pregnant
Maternal serum Waist
women and 19.£
their daughters,
ALSPAC
N = 319
circumference
(WC)(cm),
Trunk fat (%),
BMI (kg/m2),
Total body fat
(%) per high,
medium, and
low educational
status
Regression
coefficient per
unit increase in
PFOS
WC:-0.12 (-0.20, -0.04),
p-value = 0.005
Trunk fat:
-0.06 (-0.12, 0.01), p-value = 0.02
BMI:
-0.04 (-0.07, 0.0), p-value = 0.03
Total body fat (%), WC, Trunk fat,
and BMI for overall, low, and
medium education status: No
statistically significant associations
Confounding: Sampling design, pre-pregnancy BMI (kg/m2) and maternal educational status
Kang et al.
(2018)
Medium
Korea, Cross-sectional Children from
2012-2014 KorEHS-C
Seoul and
Gyeonggi, 3-
18 yr of age,
N = 147
Plasma Fasting blood Regression
5.68 glucose (mg/dL) coefficient per
ln-unit increase
in PFOS
Blood glucose:
0.707 (-1.921,3.336),
p-value = 0.595
Confounding: Age, sex, BMI z-score, household income, secondhand smoking
Karlsen et al.
(2017)
Medium
Faroe Islands,
recruited 2007-
2009 (at birth);
follow-up at
child ages
18 mo, 5 yr
Cohort Children, 5 yr
(BMI)
N = 349
Children, 5 yr
(overweight)
N = 371
Children, 18 mo
(overweight)
N = 444
Serum,
Maternal serum
5 yr: 4.7
18 mo: 8.25
BMI z-score,
Overweight
RR (OW), or
Regression
coefficient per
loglO-unit
increase in
maternal PFOS,
or by tertiles
(BMI)
BMI z-score
18 mo: 0.2 (0.1, 0.4),
p-value < 0.05
OW
18 mo:1.29 (1.01,
p-value < 0.05
1.64),
Results: Lowest tertile used as reference.
D-176
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: Maternal nationality, age at delivery, pre-pregnancy BMI, smoking during pregnancy, child sex, exclusive breastfeeding
duration, child's fish intake at age 5 yr
Kobayashi et al.
(2017)
Medium
-1.07 (-1.79, -0.36),
p-value = 0.004
Japan, Cross-sectional Children from Maternal serum Ponderal index Regression
2002-2005 Hokkaido Study 5.3 coefficient per
on Environment ln-unit increase
and Children's in PFOS
Health
N = 176
Confounding: Maternal age, pre-pregnancy BMI, parity, maternal education, maternal smoking during pregnancy, gestational age, infant sex,
and maternal blood sampling period
Lauritzen et al.
(2018)
Medium
Norway and
Sweden,
Recruitment
1986-1988
Cohort
Pregnant
women and
their children at
5-yr follow-up
N = 412
Serum
Norway: 9.62
Sweden: 16.3
BMI, triceps
skin fold,
subscapular
skinfold,
overweight
Regression
coefficient or
OR per ln-unit
increase in
maternal PFOS
Regression coefficient
BMI: 0.18(0.01,0.35)
Triceps skinfold: 0.15 (0.02, 0.27)
Odds ratio
Overweight: 2.04 (1.11, 3.74)
Subscapular skinfold: No
statistically significant association
Confounding: Age, education, smoking at conception, pre-pregnancy BMI, weight gain at 17 wk, interpregnancy interval, previous
breastfeeding duration and country of residence
Lopez-Espinosa United States, Cohort
etal. (2016) 2005-2006
Medium
Children, ages
6-9 yr from the
C8 Health
Project
N = 1,123 girls
and 1,169 boys
Serum
Girls: 20.9
Boys: 22.4
Insulin-like
growth factor 1
(IGF-1) (ln-
ng/mL)
Percent
difference
for 75th vs. 25th
percentile of
ln(PFOS), or by
quartiles
IGF-1
Girls: -5.6 (-8.2, -2.9)
Q4: -11.4 (-16.5,-6.0)
Boys: -5.9 (-8.3, -3.3)
Q3: -6.3 (-11.6,-0.6)
Q4: -11.5 (-16.6,-6.1)
Boys Q2; Girls Q2, Q3: No
statistically significant associations
Results: Lowest quartile used as reference.
Confounding: Age and month of sampling
D-177
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Manzano-
Salgado et al.
(2017b)
Medium
Spain,
Recruitment
2003-2008
Cohort
Mother-child
pairs, followed
for 8 yr, INMA
Study
N = 1230
Maternal blood
GM = 5.80
BMI, WC,
overweight,
waist-to-hip
ratio
Regression
coefficient per-
log2-unit
increase in
PFOS
BMI, waist circumference,
overweight, waist-to-hip ratio: No
statistically significant associations
Confounding: Maternal characteristics (i.e., region of residence, country of birth, previous breastfeeding, age, pre-pregnancy BMI), age of
child
Martinsson et al.
(2020)
Medium
Sweden,
2003-2008
Case-control
Pregnant
women and
their children at
age 4, Southern
Sweden
Maternity
Cohort
N = 1,048
Serum
16.6
Overweight OR by quartiles OW
Q4: 1.57 (1.07,2.3)
Q2 and Q3: No statistically
significant association
Results: Lowest quartile used as reference
Confounding: Risk strata, difference from strata-specific mean, sex
Mora et al.
(2017)
Medium
United States,
1999-2002
Cohort
Early childhood
N = 992
Mid-childhood
N = 871
Maternal
Plasma
Early childhood:
24.8
Mid-childhood:
24.7
WC (cm), Sum
of subscapular
and triceps
skinfold
thickness (mm),
BMI, waist-to-
hip ratio,
obesity,
overweight,
total fat mass
index, total fat-
free mass index
Regression
coefficient per
IQR increase in
PFOS
All:
Sum of subscapular and triceps
skinfold thickness:
-0.41 (-0.77, -0.05)
Boys:
Waist-to-hip ratio:
-0.76 (-1.47, -0.05)
Early childhood:
BMI, obesity, overweight, total fat
mass index, total fat-free mass
index: No statistically significant
association
Mid-childhood:
Waist circumference (cm), Sum of
subscapular and triceps skinfold
thickness (mm), BMI, waist-to-hip
D-178
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
ratio, obesity, overweight, total fat
mass index, total fat-free mass
index: No statistically significant
association.
Confounding: Maternal age, race/ethnicity, education, parity, pre-pregnancy BMI, timing of blood draw, household income, child sex, age at
outcome assessment
Scinicariello et United States,
al. (2020a) 2013-2014
Medium
Cross-sectional Children aged
3-11 yr from
NHANES
N = 600
Serum
GM = 3.90
(SE = 0.17)
Girls:
GM = 3.69
(SE = 0.15)
Boys:
GM = 4.12
(SE = 0.27)
BMI z-score Regression
(BMIZ), height- coefficient per
for-age z-score ln-unit increase
(HAZ), WAZ in PFOS or by
tertiles
BMIZ:-0.09 (-0.30,0.13)
T2:-0.19 (-0.41, 0.03)
T3: -0.21 (-0.53,0.11)
p-value for trend = 0.17
Girls: -0.20 (-0.48, 0.07)
Boys: -0.02 (-0.29, 0.24)
HAZ: -0.29 (-0.49,-0.10)
T2: -0.32 (-0.60, -0.04)
T3:-0.39 (-0.72,-0.06)
p-value for trend = 0.06
Girls: -0.34 (-0.73, 0.05)
Boys: -0.22 (-0.41, -0.03)
T3: -0.28 (-0.53,-0.03)
WAZ: -0.25 (-0.47, -0.03)
T2: -0.32 (-0.60, -0.04)
T3: -0.40 (-0.76, -0.04)
p-value for trend = 0.06
Girls: -0.35 (-0.72, 0.03)
Boys:-0.17 (-0.37, 0.03)
No other statistically significant
associations or trends by quartiles
stratified by sex
NHANES = National Health and Nutrition Examination Survey
Results: Lowest tertile used as reference
Confounding: Age, quadratic age, race/ethnicity, PIR, serum cotinine, birthweight, maternal smoking during pregnancy, hematocrit, sex
D-179
-------
APRIL 2024
Reference,
Confidence
Location, __ .
Design
Years
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Fleisch et al.
United States, Cohort
Pregnant
Plasma
Leptin,
Percent change
HOMA-IR:
(2017)
Medium for
Pregnant
women
women and
their children
GM = 6.2
Adiponectin,
HOMA-IR
per IQR
increase in
Per IQR increase -10.1% (-16.4,
-3.3)
metabolic
recruited 1999-
from
PFOS, or by
Q4: -24.7 (-37.8, -8.8)
function
2002, outcome
Project Viva
quartiles
Females:
Low for
assessed at mid-
N = 584
-16.7 (-25.7, -6.7)
HOMA-IR
childhood
follow-up
Median age at
follow-
up = 7.7 yr
Q4: -30.7 (-47.5, -8.4)
Leptin, adiponectin: No statistically
significant associations
Results: Lowest quartile used as reference; Q4 (9.8-51.4 ng/mL), Q1 (<0.1—4.2 ng/mL) PFOS.
Confounding: Characteristics of child (age, sex, race/ethnicity), mother (age, education), and neighborhood census tract at mid-childhood
(median household income, percent below poverty)
Pregnant Women
Jensen et al.
Denmark, Cohort
Pregnant
Serum
Blood glucose,
Percent change
Blood glucose, insulin, c-peptide,
(2018)
High
recruitment
2010-2012,
women, Odense
Child Cohort
8.37
insulin, c-
peptide, 2-hr
per log2-unit
increase in
2-hr glucose, insulin resistance,
beta-cell function, insulin
outcome
assessed 12-
N = 158
glucose, insulin
resistance, beta-
PFOS
sensitivity: No statistically
significant association
20 wk later
cell function,
insulin
sensitivity
Confounding: Age, parity, education level, pre-pregnancy BMI
Mitro et al.
United States, Cohort
Pregnant
Plasma
WC (cm),
Percent
Skinfold thickness
(2020)
High
Recruitment
1999-2002
women,
Project Viva
N = 786
24.8
BMI (kg/m2),
Adiponectin
((ig/mL),
Skinfold
thickness, Arm
circumference,
HbAlc, Leptin
difference per
log2-unit
increase in
PFOS
All: 1.2 (0.1, 2.2), p-value < 0.05
Women < 35 at pregnancy: 1.5
(0.1, 3), p-value < 0.05
WC, BMI, Adiponectin, arm
circumference, HbAlc, leptin: No
statistically significant associations
Confounding: Age, pre-pregnancy BMI, marital status, race/ethnicity, education, income, smoking, parity, breastfeeding in a prior pregnancy
D-180
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Preston et al.
(2020)
High
United States,
1999-2002
Cohort
Pregnant
women from
Project Viva
N = 1,533
Serum
25.7
Gestational
diabetes,
glucose
tolerance,
hyperglycemia,
glucose blood
level
Regression Glucose blood level,
coefficient by All
quartiles Q4: 4.3 (0.5, 8.0)
< 35 yr
Q4: 6.5 (2.1, 10.9)
Q3: 5.2 (0.8, 9.7)
Q2: 5.2 (0.8, 9.6)
Gestational diabetes, glucose
tolerance, hyperglycemia: No
statistically significant association
Results: Lowest quartile used as reference; Q1 (0.1-18.8 ng/mL), Q2 (18.9-25.7 ng/mL), Q3 (25.8-34.9 ng/mL), Q4 (35.0-185.0 ng/mL).
Confounding: Pre-pregnancy BMI, prior history of gestational diabetes/parity, race/ethnicity, smoking, and education, maternal age (Full
group only)
Starling et al.
(2017)
High
United States,
2009-2014
Cohort
Pregnant
women and
their children in
the Healthy
Start study
N = 628
Maternal serum Maternal
2.4 glucose
Regression
coefficient per
unit increase in
PFOS and by
tertile
Maternal glucose: No statistically
significant associations
Confounding: Maternal age, pre-pregnancy BMI, race/ethnicity, education, smoking during pregnancy, gravidity, and gestational age at blood
draw
Ashley-Martin
Canada, Cohort
Pregnant
Serum
GWG (kg)
Regression
Underweight/normal BMI: 0.39
et al. (2016)
Pregnant
women from
0.15
coefficient per
(0.02, 0.75)
Medium
women
MIREC
log2-unit
recruited 2008-
N = 1,609
increase in
Overweight and obese BMI: No
2011, outcome
PFOS
statistically significant association
assessed at birth
Confounding: Age, income, parity
Jaacks et al.
United States, Cohort
Pregnant
Serum
GWG (kg)
Regression
GWG
(2016)
2005-2007
women
Mean= 14.81
coefficient and
0.26 (-0.66, 1.18)
Medium
N = 218
OR per SD-unit
OR for excessive GWG: 1.01
increase in
(0.72, 1.4)
PFOS
D-181
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: Pre-pregnancy non-fasting serum lipids, BMI
Liu et al. (2019) China, 2013-
Medium 2015
Case-control
Marks et al.
(2019)
Medium
Rahman et al.
(2019)
Medium
Pregnant Serum
women without 3.13
history or
family history
of diabetes
N = 189
Gestational Regression
diabetes coefficient per
(GDM), glucose ln-unit increase
homeostasis or by tertiles
sum m-PFOS or
L-PFOS
GDM:
m-PFOS
Per ln-unit increase: 1.36(0.
2.11)
T2: 1.53 (0.7, 3.34)
T3: 1.23 (0.56,2.72)
L-PFOS
Per ln-unit increase: 1.58 (0.89,
2.79)
T2: 1.34 (0.62, 2.93)
T3: 1.37 (0.62, 3.02)
Glucose homeostasis: No
statistically significant association
Results: Lowest tertile used as reference.
Confounding: Maternal age, BMI in early pregnancy, fetal sex, serum triglyceride, total cholesterol
United
Kingdom
1991-1992
Cohort
Mothers from
ALSPAC
N = 905
Serum
Mothers of
sons: 13.8
Mothers of
daughters: 19.8
GWG (absolute) Regression
coefficient per
10% increase in
log-unit PFOS
GWG: No statistically significant
associations
Comparison: Logarithm base not specified.
Confounding: Maternal education, prenatal smoking, maternal age at delivery, parity, pre-pregnancy BMI, gestational age at delivery,
gestational age at sample
United States,
2009-2013
Cohort
Pregnant
women with
singleton
pregnancies
N = 2,292
Plasma
GM = 5.21
GDM
RR per SD-unit
increase in
PFOS
GDM: No statistically significant
associations
Confounding: Maternal age, enrollment BMI, education, parity, race/ethnicity, serum cotinine
Renetal. China, 2012 Cross-sectional Pregnant Plasma Glucose (1 hr, Regression Glucose (1 hr tolerance test): 0.31
(2020) women, 10.7 fasting) coefficient per (0.11,0.50), p-value = 0.003
D-182
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Medium
Shanghai-
Minhang Birth
Cohort Study
N = 705
ln-unit increase
in PFOS
Glucose after fasting, glucose after
1 hr tolerance test by gestational
weeks: No statistically significant
association
Confounding: Maternal age at enrollment, pre-pregnancy BMI, per capita household income, education level, passive smoking, pregnancy
complication, history of abortion and stillbirth, parity
Shapiro et al.
(2016)
Medium
Canada,
2008-2011
Cohort
Pregnant
women
N = 1,195
Urine
Normal glucose
GM = 4.58
Gestational
impaired
glucose
tolerance
GM = 4.29
Women with
GDM
GM = 4.74
GDM,
gestational
impaired
glucose
tolerance
OR per quartile Gestational diabetes, gestational
PFOS impaired glucose tolerance: No
statistically significant association
Confounding: Maternal age, race, pre-pregnancy BMI, and education
Valvi et al. Faroe Islands, Cohort Pregnant Maternal serum Gestational OR per Gestational diabetes:
(2017) 1997-2000 women and 27.2 diabetes doubling of Per doubling: 0.86 (0.43, 1.7)
Medium their children PFOS, or by T2: 0.85 (0.43, 1.7)
N = 604 tertiles T3: 0.56 (0.26, 1.19)
Results: Lowest tertile used as the reference group
Confounding: Maternal age at delivery, education, parity, pre-pregnancy BMI, smoking during pregnancy
Wang et al.
China
Case-control Pregnant
Serum
Fasting blood
Fasting blood
Fasting blood glucose
(2018c)
2013
women with
n-PFOS
glucose, GDM
glucose: OR by
n-PFOS
Medium
(cases) and
Cases: 2.70
tertiles of PFOS
T2: 1.94 (1.05,3.58),
without
Controls: 2.81
isomer
p-value < 0.05
(controls) GDM
lm-PFOS
GDM: OR per
T3: 1.59 (0.85,2.96)
N = 242
Cases: 0.14
unit increase in
lm-PFOS
Controls: 0.14
PFOS isomer
T2: 1.86 (1.00, 3.48),
3m+4m-PFOS
p-value < 0.05
Cases: 0.44
T3: 2.07 (1.09, 3.93),
Controls: 0.42
p-value < 0.05
D-183
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
5m-PFOS
Cases: 0.36
Controls: 0.36
6m-PFOS
Cases: 0.29
Controls: 0.31
3m + 4m-PFOS
T2: 1.81 (0.98, 3.33)
T3: 1.88 (1.00,3.52),
p-value < 0.05
5m-PFOS
T2: 1.94 (1.05,3.80),
p-value < 0.05
T3: 2.45 (1.24, 4.64),
p-value < 0.05
6m-PFOS
T2: 1.24 (0.67,2.28)
T3: 1.42 (0.83,2.77)
GDM: No statistically significant
associations
Results: Lowest tertile used as reference.
Confounding: Fasting blood glucose: BMI, age, GDM status; GDM: BMI, GWG, ethnic groups, maternal education, parity, maternal
drinking during pregnancy, household income
Wang et al.
China, Cohort Pregnant
Serum
Fasting blood
LSM by tertiles
Fasting blood glucose:
(2018a)
2013-2014 women aged
5.4
glucose, fasting
T2: 1.47 (1.45, 1.48),
Medium
20-40
insulin, HOMA-
p-value < 0.05
IR, gestational
T3: 1.47 (1.45, 1.48),
N = 385
diabetes, oral
p-value < 0.05
glucose
tolerance
Oral glucose tolerance: 1.88 (1.84,
1.91), p-value < 0.05
Fasting insulin, HOMA-IR,
gestational diabetes: No statistically
significant association
Results: Lowest tertile used as reference.
Confounding: Pregnant age, diabetes mellitus history of relatives, husband smoking status, family per capita income, baby sex, averaged
intake of meat, vegetable, and aquatic products, averaged physical activity, and averaged energy intake, pre-pregnant maternal BMI
Xu et al. (2020b) China, Nested case- Pregnant
Serum
GDM
OR per unit
GDM
Medium
2017-2019 control women
Cases: 6.69
increase in
Q2: 0.69 (0.34, 2.07)
Controls: 6.45
PFOS; OR per
Q3: 0.72 (0.48, 1.90)
D-184
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
N = 165 cases,
330 controls
loglO-unit Q4: 1.07 (0.51, 1.32)
increase in p-trend = 0.27
PFOS logPFOS: 0.61 (0.42, 1.65),
p-value = 0.21
Confounding: Maternal age, sampling time, parity, BMI, educational level, and serum lipids
General Population
Cardenas et al.
(2017)
High
United States, Cohort Adults at high Plasma Adiponectin Regression
Recruitment risk of Type 2 GM = 26.38 (|ig/mL). coefficient per
July 1996-May diabetes HbAlc (%), doubling of
1999, outcome N= 956 Insulin (fasting) PFOS
assessed ((iU/mL),
annually until Glucose
May 2001 (fasting)
((iU/mL),
HOMA-IR,
Insulin (30 min,
(iU/mL),
Proinsulin
(fasting, pM),
HOMA-B,
Insulin
(corrected
response),
Insulinogenic
index, Diabetes,
HOMA-IR,
glucose
(30 min),
glucose (2 hr),
BMI
HbAlc: 0.03 (0.002, 0.07),
p-value = 0.04
Insulin (fasting): 1.37 (0.41,2.34),
p-value = 0.005
Glucose (fasting): 0.55 (0.03, 1.06),
p-value = 0.04
HOMA-IR: 0.39 (0.13, 0.66),
p-value = 0.004
Insulin (30 min): 4.63 (0.89, 8.36),
p-value = 0.02
Proinsulin (fasting): 1.37 (0.5,
2.25), p-value = 0.002
HOMA-B: 9.62 (1.55, 17.7),
p-value = 0.02
Diabetes, glucose (30 min), glucose
(2 hr), BMI, adiponectin, insulin
(corrected), insulinogenic index:
No statistically significant
association
Confounding: Sex, race/ethnicity, BMI, age, marital status, education, smoking history.
Blake et al.
(2018)
United States,
1991-2008
Cohort
Adults living in
a community
Serum
28.4
BMI
Percent change BMI: No statistically significant
per IQR associations
D-185
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Medium
with water
supply from a
PFAS-
contaminated
aquifer
N = 192
increase in
PFOS
Confounding: Age, year of measurement, sex, education, income, marital status, and BMI
Cardenas et al. United States,
(2019) 1996-2014
Medium
Controlled trial
Adults older
than 25 without
diabetes and
with elevated
fasting and
postload
glucose, DPP
N = 956
Plasma
GM = 26.38
T2D
Hazard ratio per T2D:
log2-unit
increase in
baseline PFOS
and by PFOS
tertiles
HR: 1.05 (0.94, 1.18)
T2: 0.94 (0.75, 1.17)
T3: 0.94 (0.75, 1.18)
Confounding: Sex, race/ethnicity, baseline age, marital status, education, income, smoking history, BMI, maternal diabetes, paternal diabetes,
treatment assignment
Christensen et
al. (2016b)
Medium
Conway et al.
(2016)
Medium
United States,
2011-2013
Cross-sectional
Male anglers
N = 154
Serum
19.0
Diabetes, pre- OR per unit in Diabetes, pre-diabetes: No
diabetes PFOS statistically significant associations.
Confounding: Age, BMI, employment status, number of alcoholic drinks consumed per month
United States,
2005-2006
Cross-sectional All individuals Serum
T1D,
OR per ln-unit
working or
living in six
PFOS-
contaminated
water districts
with diabetes
N = 6,460
All participants T2D, increase in
mean = 86.5 Uncategorized PFOS
Diabetes
T1D: 0.73 (0.67, 0.79)
T2D: 0.92 (0.88, 0.96)
Children with T1D: 0.52 (0.54,
0.87)
Adults with TID: 0.77 (0.71, 0.84)
Uncategorized diabetes: No
statistically significant association
Confounding: Age, sex, race, BMI, eGFR, hemoglobin, iron
Donat-Vargas et
al. (2019a)
Medium
Sweden,
1990-2003,
2001-2012
Case-control
Adults with Plasma
(cases) and Cases:
without 19.0
(controls) type 2 Controls:
diabetes living 20.0
in Sweden
T2D
OR per SD
loglO-unit
increase in
baseline PFOS,
or by tertiles
T2D
OR: 0.7 (0.47, 1.03)
T2: OR: 0.79 (0.34, 1.87)
HOMA-B and HOMA-IR: No
statistically significant associations
D-186
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
N = 248
Results: Lowest tertile used as reference; T1 (13, 11-16 ng/mL), T2 (21, 19-23 ng/mL).
Confounding: Gender, age, sample year, red and processed meat intake, fish intake, BMI
Duan et al.
China, 2017 Cross-sectional
Adults, 19 to
Serum
Fasting glucose
Regression
HbAlc 55+: 0.02819 (0.00557,
(2020)
87 yr old
14.24
(nmol/L),
coefficient per
0.04965)
Medium
N = 252
HbAlc
1% increase in
serum PFOS
HbAlc < 55, fasting glucose: No
statistically significant association
Confounding: Sex, age, BMI, smoking and alcohol-drinking status, exercising status, education level, and family history of diabetes
Jain et al.
United States, Cohort
Adults from
Serum
Obesity
Comparison of
Obesity: p-value = 0.01
(2019e)
2011-2014
NHANES, 20
Non-obese
GM of PFOS
Medium
and older
GM = 2.2
levels for non-
N = 2,883
Obese GM = 2.0
obese vs. obese
Confounding: Not reported
Jeddy et al.
England, Nested case-
Pregnant
Maternal serum
Fat mass
Regression
Fat mass: No statistically
(2018)
mothers control studies
mothers and
20.2
coefficient per
significant association
Medium
recruited 1991—
their 17-yr old
unit increase in
2002, outcome
daughters,
PFOS
assessed at
ALSPAC
age 17
N = 221
Confounding: Maternal pre-pregnancy BMI, maternal education, maternal age at delivery, gestational age at sample collection, and ever
breastfed status at 15 mo
Liu et al.
Boston, Controlled Trial
Overweight and
Plasma, glucose
Body weight
Partial
Spearman correlations
(2018a)
Massachusetts
obese patients
Males: 27.2
(kg), Resting
Spearman
Body weight: 0.8, p-value < 0.05
Medium for
and Baton
from the
Females: 22.3
metabolic rate
correlation with
adiposity/weight
Rouge,
POUNDS Lost
(RMR)
baseline PFOS
Body weight, months 6-24
change
Louisiana,
Trial, Ages 30-
(kcal/24 hr),
(insulin, leptin)
All:
Uninformative
2004-2007
70,
HbAlc, insulin,
Tl: 1.5, p-trend = 0.007
for insulin
N = 621
glucose, fat
Regression
T2: 3.5, p-trend = 0.007
resistance
mass, WC,
coefficient per
T3: 3.2, p-trend = 0.007
leptin, HOMA-
loglO-unit
Women:
IR
increase in
Tl: 2.1, p-trend = 0.01
PFOS, or
T2: 4.1, p-trend = 0.01
by tertile
T3: 4.0, p-trend = 0.01
D-187
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Per loglO-umit increase in PFOS
0.8, p-value < 0.05
RMR
First 6 mo, all
Tl: -5.0, p-trend = 0.005
T2: -24.7, p-trend = 0.005
T3: -45.4, p-trend = 0.005
Months 6-24, all
Tl: 94.6, p-trend< 0.001
T2: 67.3, p-trend < 0.001
T3: 0.9, p-trend < 0.001
First 6 mo, women
Tl: -19.2, p-trend = 0.01
T2: -29.7, p-trend = 0.01
T3: -60.4, p-trend = 0.01
Months 6-24, men
Tl: 46.8, p-trend = 0.05
T2: 60.8, p-trend = 0.05
T3: -40.2, p-trend = 0.05
Months 6-24, women
Tl: 141.6, p-trend = 0.001
T2: 90.1, p-trend = 0.001
T3: 47.7, p-trend = 0.001
Liu et al.
(2018b)
Medium
HbAlc, glucose, fat mass, WC,
leptin: No statistically significant
association
Results: Lowest tertile used as reference; Tertile 1 (<19.2 ng/mL), tertile 2 (19.2-32.1 ng/mL), tertile 3 (> 32.1 ng/mL) PFOS.
Confounding: Age, sex, race, education, smoking status, alcohol consumption, physical activity, menopausal status (women only), hormone
replacement therapy (women only), and dietary intervention groups.
United States,
2013-2014
Cross-sectional
Adults from
NHANES
N = 1,871
Serum
GM =
5.28
Fasting blood Regression
glucose, 2-hr coefficient per
glucose, HbAlc, ln-unit increase
insulin levels, in PFOS
HOMA-IR,
Fasting blood glucose: 1.96
(SE = 0.79)
2-hr glucose, HbAlc, insulin levels,
HOMA-IR, beta-cell function,
D-188
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
beta-cell
function,
metabolic
syndrome, WC
metabolic syndrome, WC: No
statistically significant associations
Confounding: Age, gender, ethnicity, smoking status, alcohol intake, household income, WC, and medications (antihypertensive, anti-
hyperglycemic, and anti-hyperlipidemic agents)
Mancini et al.
(2018)
Medium
Su et al. (2016)
Medium
France,
1990-2012
Cohort
Women aged
40-60, E3N
Cohort
N = 71,294
Food T2D
Mean = 0.49 ng/
kg body
weight/day
Hazard ratio per T2D: No statistically significant
decile PFOS association
Confounding: Smoking status, physical activity, education level, hypertension, hypercholesterolemia, family history of diabetes, energy
intake, alcohol intake, adherence to the Western diet and adherence to the Mediterranean diet, water consumption, dairy product consumption
Taiwan,
2009-2011
Cross-Sectional Adults aged 20- Plasma
60 living in
Taiwan
N = 571
5.0
Diabetes,
Fasting blood
glucose
(ng/mL),
blood glucose
(120 min) (In)
(ng/mL),
glucose AUC
(ng/mL),
HbAlc (In) (%)
OR and GM Diabetes:
ratio (GMR) per OR: 2.39 (1.52, 3.76)
doubling of
PFOS, or by
quartiles
OR Q4: 3.37 (1.18, 9.56)
Glucose (Fasting):
GMR: 1.03 (1.01, 1.04)
GMRQ4: 1.05 (1.02, 1.09)
Glucose (120 min)
GMR: 1.08 (1.05, 1.12)
GMRQ4: 1.17 (1.08, 1.25)
Glucose AUC:
GMR: 1.06 (1.04, 1.09)
GMRQ4: 1.12 (1.06, 1.19)
Results: Lowest quartile used as reference; Q1 (<2.4 ng/mL); Q4 (> 4.8 ng/mL).
Confounding (Diabetes): Age, sex, education, smoking (ever vs. never), alcohol (ever vs. never), BMI, hypertension, total cholesterol,
regular exercise
Confounding (Other): Age, sex, education, smoking, alcohol, BMI, hypertension, total cholesterol, regular exercise
Sun et al. (2018) United States, Case-control Female nurses Plasma T2D Regression
Medium recruitment drawn from the Cases: hemoglobin, coefficient SD
1989, blood Nurses' Health 35.7 insulin, loglO-unit
sample Controls: adiponectin
T2D
Per SD increase: 1.15 (0.98, 1.35),
p-value = 0.008
OR for T2: 1.63 (1.25,2.12)
D-189
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
collection 1995—
2000, outcome
assessed during
biennial follow-
up through June
2011
Study II cohort 33.1
study,
N = 1,586
increase in OR for T3: 1.62 (1.09, 2.41)
PFOS
Partial Spearman correlation
OR by tertiles coefficient for hemoglobin, insulin,
and adiponectin: No statistically
significant association
Results: Lowest tertile used as reference.
Confounding: Age, month of sample collection, fasting status, menopausal status, postmenopausal hormone use, family history of diabetes,
oral contraceptive use, breastfeeding duration at blood draw, number of children delivered after 1993, states of residence, smoking status,
alcohol intake, physical activity, baseline BMI, and Alternative Healthy Eating Index (AHEI) score
Chen et al. Croatia
Cross-sectional Residents of
Plasma
BMI, fasting
Metabolic
Metabolic syndrome: 1.89 (0.93,
(2019a) 2007-2008
Hvar ages 44-
GM = 8.91
insulin
syndrome: OR
3.86); p-value = 0.08
Medium for
56 yr
(Range: 2.36-
((iIU/mL),
per ln-unit
metabolic
N = 122
33.67)
fasting plasma
increase in
All other outcomes: No statistically
syndrome
glucose
PFOS
significant associations
Low for all other
(mmol/L),
outcomes
glycated HbAlc
All other
(%), hip
outcomes:
circumference
regression
(cm),
coefficient per
homeostatic
ln-unit increase
model
in PFOS
assessment of
beta-cell
function
(HOMA-(3),
homeostatic
model
assessment of
insulin
resistance
(HOMA-IR),
metabolic
syndrome
defined by the
ATP III criteria,
D-190
-------
APRIL 2024
Reference,
Confidence
Location, Design Population, Matrix, Levels Outcome Comparison
Years Ages,N (ng/mL)a
Resultsb
waist
circumference
(cm)
Confounding: Age, sex, education, socioeconomic status, smoking, dietary pattern, and physical activity
Notes: AHEI = Alternative Healthy Eating Index; ALSPAC = Avon Longitudinal Study of Parents and Children; AUC = area under the curve; BMI = body mass index;
BMIZ = BMI z-score; DM = diabetes mellitus; DPP = Diabetes Prevention Program; EYHS = European Youth Heart Study; GDM = gestational diabetes mellitus;
GM = geometric mean; GWG = gestational weight gain; HAZ = height-for-age z-score; HbAlc = Hemoglobin Ale; HOMA-Beta = homeostatic model assessment of P-cell
function; HOMA-IR = homeostatic model assessment for insulin resistance; HOME = Health Outcomes and Measures of the Environment; hr = hour; IGF = insulin-like growth
factor; INMA = INfancia y Medio Ambiente (Environment and Childhood) Project; IQR = interquartile range; IR = insulin resistance; KorEHS-C: Korea Environmental Health
Survey in Children and Adolescents; LSM = least square mean; min = minutes; MIREC = Maternal-Infant Research on Environmental Chemicals; mo = months;
NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; OW = overweight; PIR = poverty income ratio; POPUP = Persistent Organic Pollutants in
Uppsala Primiparas; POUNDS = Preventing Overweight Using Novel Dietary Strategies; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; RMR = resting metabolic rate;
RR = risk ratio; SD = standard deviation; SE = standard error; SOLAR = Study of Latino Adolescents at Risk of Type 2 Diabetes; T1 = tertile 1; T2 = tertile 2; T3 = tertile 3;
T1D = type 1 diabetes; T2D = type 2 diabetes; vs. = versus; WC = waist circumference; wk = weeks; yr = years.
a Exposure levels are reported as median in ng/mL unless otherwise noted.
b Results are reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.8 Nervous
Table D-17. Associations Between PFOS Exposure and Neurological Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population, Exposure
Ages, Matrix, Levels Outcome
N (ng/mL)a
Comparison
Resultsb
Children and Adolescents
Harris et al.
(2018)
High
United States,
Recruitment:
1999-2002;
Follow-up at
early- and mid-
childhood
Cohort
Pregnant
Plasma
Both age
Mean difference Visual-Motor
women and Maternal: 24.9 groups: Wide by quartiles of Mid-childhood (maternal plasma)
their children
from Project
Viva
N = 853
(18.4-34.4)
Child: 6.2 (4.2-
9.7)
Range
Assessment of
Visual Motor
Abilities
(WRAVMA)
score
PFOS exposure
Q2
Q3
Q4
-1.6 (-4.7, 1.6)
-1.4 (-4.7, 1.8)
-3.2 (-6.6, 0.2)
Mid-childhood (child plasma)
Q2
Q3
Q4
-1.6 (-5.5, 2.2)
-4.6 (-8.7, -0.5)
-2.0 (-6.3, 2.2)
D-191
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Early childhood
only: Peabody
Picture
Vocabulary Test
(PPVT-III)
score
Mid-childhood
only: Kaufman
Brief
Intelligence Test
Second Edition
(KBIT-2)
nonverbal and
verbal IQ,
(WRAML2)
design memory
and picture
memory
Nonverbal IQ
Mid-childhood (maternal plasma)
Q2
Q3
Q4
-0.7 (-3.8, 2.3)
-1.8 (-5.0, 1.4)
1.6 (-1.8,4.9)
Mid-childhood (child plasma)
Q2
Q3
Q4
-0.4 (-4.0, 3.2)
1.6 (-2.3, 5.4)
-0.1 (-4.1, 3.8)
Verbal IQ
Mid-childhood (maternal plasma)
Q2
Q3
Q4
-2.1 (-4.5,0.2)
-1.7 (-4.2, 0.7)
0.8 (-1.8, 3.4)
Mid-childhood (child plasma)
Q2
Q3
Q4
0.9 (-2, 3.8)
-0.4 (-3.4, 2.7)
-0.2 (-3.4, 3.0)
Design memory
Mid-childhood (maternal plasma)
Q2:-0.1 (-0.7,0.4)
Q3: 0.3 (-0.3,0.8)
Q4: 0.6 (0, 1.2)
Mid-childhood (child plasma)
Q2: 0.1 (-0.5,0.7)
Q3: 0.1 (-0.6,0.7)
Q4: -0.2 (-0.9, 0.5)
Picture memory
Mid-childhood (maternal plasma)
Q2
Q3
Q4
-0.3 (-0.9, 0.2)
-0.1 (-0.7,0.5)
0.4 (-0.2, 1.0)
Mid-childhood (child plasma)
D-192
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Q2:-0.1 (-0.8,0.5)
Q3: 0.1 (-0.6,0.9)
Q4: 0 (-0.7, 0.8)
Early childhood: No statistically
significant associations
Results: Lowest quartile used as reference.
Confounding: Year of pregnancy blood collection gestational age at time of pregnancy blood collection, estimated glomerular filtration rate at
blood draw, maternal race/ethnicity, age, education, KBIT-2 score, pre-pregnancy BMI, smoking status, paternal education, annual household
income in mid-childhood, HOME-SF score, child's sex and age at mid-childhood cognitive testing, proxy for breastfeeding of a prior child0
Niu et al. (2019) China, Cohort
Pregnant
Maternal serum ASQ-3 skill
RR per ln-unit
Communication
High Recruitment:
women and
10.8 (7.6-15.8) scales:
increase in
Overall: 1.01 (0.77, 1.34)
2012; Follow-up
their children
communication,
PFOS and by
Females: 1.04 (0.65, 1.68)
at age 4 yr
from the
gross motor,
tertiles
T2: 0.52 (0.26, 1.04); p-value <0.10
Shanghai-
fine motor,
T3: 1.10 (0.63, 1.92)
Minhang Birth
problem
Males: 1.00 (0.70, 1.44)
Cohort
solving,
T2: 1.16(0.76, 1.77)
N = 533 (236
personal-social
T3: 0.89 (0.53, 1.51)
Females; 297
p-value for interaction by
Males)
sex = 0.350
Gross Motor
1.22 (0.79, 1.89)
No statistically significant
associations, trends, or interactions
by sex
Fine Motor
Overall: 1.25 (0.79, 1.96)
No statistically significant
associations, trends, or interactions
by sex
Problem Solving
Overall: 1.02 (0.71, 1.47)
Females: 1.16(0.63,2.15)
D-193
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
T2: 0.55 (0.15, 2.07)
T3: 2.00 (0.77, 5.17)
Males: 0.93 (0.59, 1.47)
T2: 1.21 (0.65,2.28)
T3: 0.66 (0.29, 1.48)
p-value for interaction by
sex = 0.010
Personal-Social Skills
Overall: 1.34 (0.91, 1.96)
Females: 2.56 (1.2, 5.45)
T2: 0.32 (0.04, 2.77)
T3: 2.97 (0.90,9.84);
p-value <0.10
p-trend <0.10
Males: 1.05 (0.67, 1.64)
T2: 1.47 (0.76,2.84)
T3: 1.18 (0.57,2.44)
p-value for interaction by
sex = 0.039
Outcome: Neuropsychological problems defined as scores < 10th percentile.
Results: Lowest tertile used as reference
Confounding: Maternal age at enrollment, pre-pregnancy BMI, maternal education, paternal education, parity, per capita household income,
maternal passive smoking, maternal prenatal depressive symptoms, gestational age, child's sex
Oulhote et al.
Faroe Islands, Cohort
Children at 5 yr
Serum
Strengths and
Mean difference
SDQ total score
(2016)
Recruitment:
(N = 508) and
Maternal: 27.35
Difficulties
(autism,
Prenatal: 0.46 (-0.78, 1.7),
High
1997-2000,
7 yr (N = 491)
(23.19-33.13)
Questionnaire
internalizing,
p-value = 0.47
Follow-up at
5 yr: 16.78
(SDQ) scores:
externalizing,
5-yr serum: 0.51 (-0.5, 1.52),
ages 5 and 7
(13.52-21.05)
Total score
total) or mean
p-value = 0.32
7 yr: 15.26
(hyperactivity/in
ratio
7-yr serum: 0.18 (-0.95, 1.31),
(12.38-18.99)
attention,
(hyperactivity/in
p-value = 0.76
conduct
attention,
problems, peer
conduct, peer
Hyperactivity /Inattention
relationship
relationship,
Prenatal: 1.03 (0.80, 1.31),
problems,
emotional,
p-value = 0.84
emotional
prosocial) per
D-194
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
symptoms), doubling of 5-yr serum: 1.05 (0.86, 1.29),
prosocial PFOS p-value = 0.64
behavior, 7-yr serum: 0.88 (0.70, 1.11),
internalizing p-value = 0.27
problem,
externalizing Conduct
problems, Prenatal: 1.03 (0.81, 1.32),
autism p-value = 0.80
screening (peer 5-yr serum: 1.00 (0.81, 1.23),
problems minus p-value = 0.98
prosocial) 7-yr serum: 1.01 (0.80, 1.26),
p-value = 0.95
Peer Relationship
Prenatal: 1.31 (0.87, 1.96),
p-value = 0.19
5-yr serum: 1.28 (0.91, 1.80),
p-value = 0.15
7-yr serum: 1.17 (0.82, 1.69),
p-value = 0.39
Emotional
Prenatal: 1.10(0.84, 1.44),
p-value = 0.49
5-yr serum: 1.14 (0.90, 1.45),
p-value = 0.26
7-yr serum: 1.22 (0.94, 1.58),
p-value = 0.13
Prosocial
Prenatal: 1.00 (0.91, 1.09),
p-value = 0.96
5-yr serum: 0.98 (0.91, 1.06),
p-value = 0.70
7-yr serum: 1.01 (0.92, 1.10),
p-value = 0.88
D-195
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Internalizing
Prenatal: 0.35 (-0.35, 1.05),
p-value = 0.32
5-yr serum: 0.44 (-0.15, 1.02),
p-value = 0.15
7-yr serum: 0.48 (-0.16, 1.13),
p-value = 0.14
Externalizing
Prenatal: 0.11 (-0.68,0.89),
p-value = 0.79
5-yr serum: 0.08 (-0.58, 0.73),
p-value = 0.82
7-yr serum: -0.31 (-1.03, 0.42),
p-value = 0.41
Autism screening
Prenatal: 0.2 (-0.37, 0.77),
p-value = 0.49
5-yr serum: 0.33 (-0.14, 0.8),
p-value = 0.17
7-yr serum: 0.06 (-0.46, 0.58),
p-value = 0.82
Confounding: Age, sex, maternal age, pre-pregnancy BMI, parity, socioeconomic status, alcohol, and smoking during pregnancy
Braun et al.
(2014)
Medium
United States,
Recruitment:
2003-2006;
Follow-up at
ages 4-5 yr
Cohort
Pregnant
women and
their children
from the HOME
study
N = 175 (80
Females; 95
Males)
Maternal Serum
13 (9.3-18)
Social
Responsiveness
Scale (SRS)
total score
Regression
coefficient per
loglO-unit/2SD
increase in
PFOS
SRS
2.1 (0.2,3.9)
Females: 0.9 (-1.5, 3.3)
Males: 3.8 (1.3, 6.3)
p-value for interaction by
sex = 0.08
Confounding: Maternal race, maternal age, maternal education, marital status, annual household income, maternal depressive symptoms,
maternal IQ, child sex, caregiving environment score, maternal serum
D-196
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population, Exposure
Ages, Matrix, Levels
N (ng/mL)a
Outcome
Comparison
Resultsb
Chen et al.
Taiwan,
Cohort
Pregnant Cord blood
CDI skill
Regression
Cognitive: -0.8 (-2.8, 1.1)
(2013)
Recruitment:
women and Mean = 7.0
quotients:
coefficient per
Fine Motor: -1.8 (-3.8, 0.1)
Medium
2004-2005;
their children (SD = 5.8)
cognitive, fine
IQR increase in
Gross Motor: -3.7 (-6.0, -1.5)
Follow-up at
from the Taiwan
motor, gross
ln-transformed
Language: -0.9 (-2.9, 1.2)
age 2 yr
Birth Panel
motor,
PFOS
Self Help: -2.2 (-4.8, 0.3)
Study
language, self-
Social: -1.0 (-3.7, 1.6)
N = 239
help, social,
Whole Test: -2.1 (-4.1,-0.2)
whole test
Confounding: Maternal education, family income, infant sex and gestational age, breastfeeding, HOME score at 24 mo of age, cord blood
cotinine levels, postnatal environmental tobacco smoke exposure
Ghassabian et
al. (2018)
Medium
United States,
2008-2010
Cohort
Children aged
7 yr from
Upstate KIDS
Study
N = 788
Blood
1.74
(IQR = 1.33)
SDQ scores:
total behavioral
difficulties-total
score,
borderline
problems;
hyperactivity,
conduct, peer,
or emotional
problems;
difficulties in
prosocial
behavior
Regression
coefficient (total
behavioral
difficulties,
problem scores)
and OR
(borderline
behavioral
difficulties,
problem scores,
difficulties in
prosocial
behavior) per
log-SD increase
in PFOS and by
quartiles
Total Behavioral Difficulties ((3)
0.04 (-0.02,0.10)
Q2: 0.14 (-0.01, 0.28)
Q3: 0.04 (-0.11,0.19)
Q4: 0.17 (0.01, 0.32)
Conduct problems (OR)
1.22 (0.97, 1.52)
Q2: 1.78 (0.97, 3.27)
Q3: 0.86 (0.43, 1.74)
Q4: 2.22 (1.18, 4.15)
Conduct problems ((3)
0.02 (-0.08,0.13)
Q2: 0.14 (-0.10, 0.39)
Q3: -0.07 (-0.33,0.19)
Q4: 0.19 (-0.07, 0.46)
Emotional problems (OR)
1.31 (1.04, 1.63)
Q2: 2.08 (1.13, 3.80)
Q3: 0.89 (0.47, 1.68)
Q4: 2.28 (1.24, 4.18)
Emotional problems ((3)
0.09(0,0.18)
Q2: 0.24 (0.03, 0.45)
Q3: 0.01 (-0.20, 0.22)
D-197
-------
APRIL 2024
Reference, Location, Population, Exposure
„ .... Design Ages, Matrix, Levels Outcome Comparison Results
Confidence Years , .
s N (ng/mL)a
Q4: 0.27 (0.05, 0.49)
Borderline Behavioral Difficulties
(OR)
1.30 (1.03, 1.65)
Q2: 1.67 (0.84, 3.34)
Q3: 1.73 (0.87, 3.43)
Q4: 2.47 (1.29, 4.72)
Difficulties in Prosocial Behavior
(OR)
1.26 (0.92, 1.72)
Q2: 0.86 (0.35,2.15)
Q3: 1.72 (0.65,4.52)
Q4: 1.87 (0.70, 4.98)
Hyperactivity problems, peer
problems: No statistically
significant associations
Comparison: Logarithm base not specified.
Results: Lowest quartile used as reference.
Confounding: Child's age and sex, maternal age, pre-pregnancy BMI, race/ethnicity, education, marital status, history of smoking in
pregnancy, having private insurance, parity, and infertility treatment
Goudarzi et al.
Japan,
Cohort
Pregnant
Maternal serum
Bayley Scales
Regression
MDI
(2016b)
2002-2005
women and
5.7 (4.4-7.4)
of Infant
coefficient
6 Months: 0.018 (-4.52, 5.59)
Medium
their infants at 6
Development,
loglO-unit
Females: 0.072 (-5.19, 9.38)
and 18 mo from
Second Edition
increase in
Males:-0.141 (-11.26,3.45)
the Hokkaido
(B SID-II)
PFOS
18 Months: 0.052 (-9.91, 16.66)
Study on
Mental
Environment
Development
PDI
and Children's
Index (MDI),
6 Months: 0.039 (-6.38, 10.37)
Health
Psychomotor
Females: 0.031 (-11.66, 15.09)
N = 173 (90
Development
Males: 0.120 (-5.24, 15.60)
Females; 83
Index (PDI)
18 Months: -0.023 (-13.45, 10.72)
Males)
D-198
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: Gestational age, parity, maternal age, smoking during pregnancy, alcohol consumption during pregnancy, caffeine intake
during pregnancy, maternal education level, blood sampling period, breast feeding, total dioxin levels
Jeddy et al.
(2017)
Medium
Great Britain.
Recruitment:
1991-1992;
Follow-up at
ages 15 and
18 mo
Cohort
Mothers and
daughters aged
15 and 38 mo
from ALSPAC
N = 353
Maternal serum
19.8(15.0-
24.95)
MacArthur
Communicative
Development
Inventories
(MCDI):
communicative,
intelligibility,
language,
nonverbal
communication,
social
development,
verbal
comprehension,
and vocabulary
comprehension
scores
Regression
coefficient per
unit increase in
PFOS
Nonverbal, 15 mo.: 0.02 (-0.01,
0.05)
Social, 15 mo.: 0.02 (-0.03, 0.08)
Verbal, 15 mo.: 0.03 (0.01, 0.05)
Maternal age < 30: No statistically
significant associations
Maternal age > 30: 0.04 (0.01,
0.08)
Vocabulary, 15 mo.: 0.02 (-0.39,
0.44)
Communicative, 38 mo.: 0 (-0.01,
0.01)
Intelligibility, 38 mo.: -0.01
(-0.01,0)
Maternal age < 25: 0.02 (0.01,
0.03)
Maternal age > 25: No statistically
significant associations
Language, 38 mo.: -0.29 (-0.54,
-0.05)
Nonverbal, social, vocabulary,
communicative, language: No
statistically significant associations
stratified by maternal age at
delivery
Confounding: Parity, maternal age, maternal education, maternal smoking status, gestational age at sample collection, total maternal Crown-
Crisp Experiential Index
D-199
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Liew et al.
(2015)
Medium
Liew et al.
(2018)
Medium
Long et al.
(2019)
Medium
Denmark,
Recruitment:
1996-2002;
Follow-up at
average age
10.7 yr
Case-control
Mother-child
pairs from
Danish National
Birth Cohort
215 Cases (39
Females; 176
Males)
545 Controls
(33 Females;
180 Males)
Maternal plasma ADHD, ASD
Cases: 25.40
(18.73-32.40)
Controls: 27.40
(20.40-35.60)
RRandOR
(stratified by
quartile or by
sex) per ln-unit
increase in
PFOS or by
quartiles
ADHD: 0.87 (0.74, 1.02)
Q4: 0.79 (0.64, 0.98)
ASD: 0.92 (0.69, 1.22)
No other statistically significant
associations by quartiles or by sex
Results: Lowest quartile used as reference
Confounding: Maternal age at delivery, SES, parity, smoking and drinking during pregnancy, psychiatric illnesses, gestational week of blood
drawn, child's sex, birth year
Denmark,
Recruitment:
1996-2002;
Follow-up at
age 5 yr
Cohort
Pregnant
women and
their children
from the Danish
National Birth
Cohort
N = 1,592
Maternal plasma Wechsler
28.10(21.60- Primary and
35.80) Preschool
Scales of
Intelligence-
Revised
(WPPSI-R) full-
scale IQ,
performance IQ,
Regression Full-Scale IQ
coefficient for Q2: -0.4 (-3.2, 2.5)
mean difference Q3: 1.1 (-1.8, 4.0)
per ln-unit Q4: -0.5 (-3.5, 2.6), p-trend = 0.87
increase in Performance IQ
PFOS and by Q2: 0.6 (-2.3, 3.5)
quartiles Q3: 1.6 (-1.2, 4.5)
Q4: -0.1 (-3.1, 2.8), p-trend = 0.93
Verbal IQ
verbal IQ
Q2:
-1.0 (-3.9, 1.9)
Q3:
-0.2 (-3.3, 2.9)
Q4:
-0.7 (-3.9, 2.4), p-trend = 0.76
Results: Lowest quartile used as reference.
Confounding: Maternal age at childbirth, parity, maternal socioeconomic status, maternal IQ, maternal smoking during pregnancy, maternal
alcohol consumption during pregnancy, maternal pre-pregnancy BMI, gestational week of blood draw
Denmark,
Recruitment:
1982-1999;
Follow-Up:
1993-2009
Case-control
Pregnant
women and
their children
from the
Historic Birth
Cohort at
Amniotic fluid ASD
Cases: 0.61
(Range: 0.61-
2.98)
ORperunit 0.410(0.174,0.967),
increase in p-value = 0.042
PFOS Females: 0.027 (0, 4.755),
p-value = 0.171
Males: 0.586 (0.192, 1.782),
p-value = 0.346
D-200
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Lyall et al.
(2018)
Medium
Statens Serum
Institute
37 Cases (7
Females; 29
Males)
50 Controls (15
Females; 35
Males)
Controls: 1.44
(Range: 0.61-
4.22)
Confounding: Child's birth year, child sex, mother's age at delivery, father age at childbirth, birth weight, gestational week at sampling,
gestational age at birth, Apgar score, parity, congenital malformation
United States,
2007-2009
Case-control
Children and
adolescents
aged 4.5-9 yr
from EMA
study
985 (553 Cases;
432 Controls)
Maternal serum
Cases:
GM = 17.5
(95%
CI = 16.8-18.3)
Controls:
GM = 17.9
(95%
CI = 17.0-18.7)
ASD measured
by Diagnostic
and Statistical
Manual of
Mental
Disorders,
Fourth Edition
(DSM-IV-TR),
intellectual
disability
OR per ln-unit
increase in
PFOS and by
quartiles
ASD: 0.77 (0.58, 1.01)
Q2: 0.85 (0.58, 1.23)
Q3: 0.74 (0.50, 1.09)
Q4: 0.64 (0.43, 0.97),
p-trend = 0.03
Intellectual Disability: 0.67 (0.45,
0.98)
1.05)
1.38)
1.09),
Q2: 0.61 (0.36,
Q3: 0.80 (0.46,
Q4: 0.59 (0.32,
p-trend = 0.17
Results: Lowest quartile used as reference.
Confounding: Matching factors, parity, maternal age, race/ethnicity,
weight at sample collection, and maternal birthplace
Oulhote et al.
(2019)
Medium
Faroe Islands,
Recruitment:
1997-2000;
Follow-up at
age 7 yr
Cohort
Children
N = 419
Blood
Maternal: 27.69
(23.22-33.35)
5 Years: 16.8
(13.5-21.13)
Boston Naming
Test with and
without cues,
SDQ total score
Regression
coefficient per
IQR increase in
PFOS
Boston Naming Test
With Cues
Prenatal: -0.11 (-0.27,0.01)
5-yr serum: 0.00 (-0.08, 0.07)
Without Cues
Prenatal: -0.04 (-0.19, 0.06)
5-yr serum: 0.00 (-0.06, 0.06)
SDQ
Prenatal: 0.15 (0.08, 0.23)
5-yr serum: 0.02 (-0.03, 0.08)
D-201
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Confounding: None reported
Quaak et al.
(2016)
Medium
Netherlands,
Recruitment:
2011-2013;
Follow-up
through age
18 mo
Cohort
Pregnant
women and
their children
from the LINC
cohort
54 (20 Females;
34 Males)
Cord blood
1,600.0 ng/L
(Range: 570-
3,200 ng/L)
Child Behavior
Checklist 1.5-5
(CBCL 1.5-5)
measures of
ADHD,
externalizing
behavior
Regression
coefficient by
tertiles
Results: Lowest tertile used as reference.
Confounding: Alcohol use, smoking, family history of ADHD, education
ADHD
T2: -0.33 (-1.75, 1.17),
p-value = 0.66
T3: -0.87 (-2.06,0.42),
p-value = 0.19
Females
T2: 0.17 (-1.50, 1.67),
p-value = 0.85
T3: -0.73 (-2.36,0.90),
p-value = 0.43
Males
T2: -0.55 (-2.84, 1.57),
p-value = 0.64
T3: -0.99 (-3.03,0.92),
p-value = 0.35
Externalizing Behavior
T2: -1.23 (-5.68, 3.85),
p-value = 0.62
T3: -2.43 (-6.55, 1.93),
p-value = 0.31
Females
T2: -2.63 (-8.21, 4.33),
p-value = 0.44
T3: -2.98 (-8.08,2.23),
p-value = 0.31
Males
T2: 0.72 (-5.77, 6.59),
p-value = 0.81
T3: -0.94 (-6.72,5.12),
p-value = 0.74
Shin et al.
(2020)
United States, Case-Control
Mother-child Maternal serum ASD measured OR per increase
pairs from 5.81 (3.86-9.11) by Autism (ln-transformed
By modeled prenatal exposure
ln-transformed: 1.18 (0.77, 1.80)
D-202
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Medium
Recruitment:
2002-2009;
Follow-up:
2009-2017
CHARGE with
children aged 2-
5 yr
N = 453 (239
Cases; 214
Controls; 88
Females; 365
Males)
Diagnostic or linear scale) No statistically significant
Interview- in modeled, associations or interactions by sex
Revised (ADI- maternal, Linear: 1.03 (0.99, 1.08);
R) prenatal PFOS p-value < 0.10
or measured, Females: 0.96 (0.85, 1.08)
maternal, Males: 1.05 (1.00, 1.10),
postnatal PFOS p-value < 0.05
and by quartiles Interaction p-value = 0.38
By measured postnatal levels
ln-transformed: 1.21 (0.80, 1.83)
Linear: 1.05 (0.97, 1.13);
p-value <0.10
No statistically significant
associations or trends by quartiles
Confounding: Child's age, child's sex, regional center, child's birth year, parity, gestational age at delivery, maternal race/ethnicity, maternal
birthplace, mother's age at delivery, maternal pre-pregnancy BMI, periconceptional maternal vitamin intake, homeownership, breastfeeding
duration
Skogheim et al.
Norway, Cohort
Mother-child
Maternal plasma Nonverbal and
Regression
Nonverbal Working Memory
(2019)
Recruitment:
pairs from
11.51 (8.77- Verbal Working
coefficient per
Q2: 0.06 (-0.14,0.26)
Medium
1999-2008;
MoBa
14.84) Memory
unit increase in
Q3: -0.10 (-0.30,0.10)
Follow-up:
N = 943
measured by
PFOS and by
Q4: -0.02 (-0.22,0.18)
2007-2011
Stanford Binet
quintiles
Q5: -0.26 (-0.48, -0.06)
Intelligence
Scales
Verbal Working Memory
Q2
-0.05 (-0.27,0.17)
Q3
0.09 (-0.14,0.31)
Q4
0.10 (-0.12, 0.33)
Q5
-0.01 (-0.24, 0.22)
Results: Lowest quintile used as reference.
Confounding: Maternal education, age, parity, fish intake, child sex, child age at testing, maternal ADHD symptoms
Spratlen et al. United States, Cohort Pregnant Cord blood BSID-II scores: Regression MDI
(2020a) Recruitment: women and GM = (Range:) MDIandPDI), coefficient of Year 1:-0.61 (-3.17, 1.95)
Medium 2001-2001; their children FullIQ, mean difference Year 2:2.36 (-1.23,5.94)
D-203
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Follow-up at
age 1, 2, and
3 yr
from the
Columbia
University Birth
Cohort
N= 302 (150
Females; 152
Males)
Performance IQ, per log-unit Females: 5.52 (0.64, 10.4)
Verbal IQ increase in Males:-1.35 (-7.09, 4.39)
maternal PFOS Interaction p-value = 0.04
Year 3: 1.96 (-1.24, 5.16)
PDI
Year 1
Year 2
Year 3
-0.07 (-4.56, 4.43)
-1.34 (-4.26, 1.57)
-0.55 (-5.34, 4.23)
Full IQ
Year 4: -0.41 (-4.25,3.43)
Year 6: 2.81 (-1.84,7.46)
Performance IQ
Year 4: -0.05 (-4.56,4.46)
Year 6: 2.81 (-2.29,7.91)
Verbal IQ
Year 4: -0.19 (-4.50, 4.12)
Year 6: 2.67 (-2.56,7.90)
Strom et al.
(2014)
Medium
No other statistically significant
associations or interactions by sex
Comparison: Logarithm base not specified.
Confounding: Maternal age, material hardship, parity, pre-pregnancy BMI, maternal IQ, maternal race, maternal education, family smoking
status, child age at testing, child's gestational age at birth, maternal demoralization, trimester on 9/11, child's sex, child's breastfeeding history
Denmark
Recruitment:
1988-1999
Follow-up: 2010
Cohort
Pregnant
women and
their children,
from the
DaF088 cohort
N = 876
Maternal serum Depression,
Median = 21.4 ADHD,
(IQR = 9.0) scholastic
achievement
Depression, Depression
ADHD: Hazard T2: 1.61 (0.99, 2.61)
ratio (depression T3: 1.16 (0.69, 1.95)
and ADHD) by p-value = 0.14
tertile
ADHD
Scholastic T2: 1.05 (0.43, 2.53)
achievement: T3: 0.54 (0.19, 1.53)
Regression p-value = 0.38
D-204
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Vuong et al.
(2016)
Medium
Vuong et al.
(2018b)
Medium
coefficient per
unit increase in
PFOS and by
tertiles
Scholastic Achievement: -0.01
(-0.03,0.01), p-value = 0.57
T3: -0.11 (-0.50,0.28),
p-trend = 0.59
Results: Lowest tertile used as reference.
Confounding: Maternal age, pre-pregnancy BMI, parity, maternal smoking during pregnancy, maternal education, maternal cholesterol,
maternal triglycerides, offspring sex
United States, Cohort
Children ages 5
Serum
BRIEF
All outcomes:
Recruitment:
and 8 yr from
13.2 (8.£
>-17.8) measures of
OR for
2003-2006;
the HOME
behavioral
score > 60 per
Follow-up at
study
regulation,
unit increase in
ages 5 and 8 yr
N = 218
metacognition,
PFOS
global executive
composite
Index and
indices, inhibit,
compositive
shift, emotional
scores only:
control, working
Regression
memory,
coefficient per
plan/organize,
ln-unit increase
initiate,
in PFOS and by
organization of
quartiles
materials,
monitor
Behavioral Regulation: 3.14 (0.68,
5.61)
Metacognition: 3.10 (0.62, 5.58)
Global Executive Function: 3.38
(0.86, 5.90)
No statistically significant
interactions by age; no statistically
significant trends by quartiles
Inhibit: 2.59 (1.23, 5.41)
Shift: 1.50 (0.72, 3.11)
Emotional control: 1.97 (0.84, 4.64)
Working memory: 1.87 (1.01, 3.48)
Plan/organize: 3.54 (1.65, 7.60)
Initiate: 1.89 (0.80, 4.45)
Organization: 1.84 (0.82, 4.13)
Monitor: 3.39 (1.42, 8.08)
Confounding: Maternal age, race, education, income, maternal serum cotinine, maternal depression, HOME score, maternal IQ, marital
status, child sex
United States,
Recruitment:
2003-2006;
Follow-up at
age 3 and 8 yr
Cohort
Children from
the HOME
study
N = 204
Serum
3 yr: 6.2 (4.5-
10.0)
8 yr: 3.6 (2.7-
4.9)
BRIEF
measures of
behavioral
regulation,
metacognition,
global executive
OR per ln-unit Behavioral Regulation
increase in 3 yr: 0.66 (0.29, 1.51)
PFOS 8 yr: 0.40 (0.14, 1.14)
Metacognition
3 yr: 0.83 (0.42, 1.63)
D-205
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
composite
indices
8 yr: 1.53 (0.67, 3.52)
Global Executive Function
3 yr: 0.95 (0.45,2.01)
8 yr: 1.04 (0.41,2.68)
Confounding: Maternal age, race/ethnicity, household income, maternal smoking status, maternal alcohol consumption, maternal depression,
HOME Score, marital status, maternal marijuana use, maternal IQ, maternal serum PCBs, maternal blood lead levels, child sex
Vuong et al.
(2018a)
Medium
United States,
Recruitment:
2003-2006;
Follow-up at
ages 3 and 8 yr
Cohort
Mother-child
dyads from the
HOME study
204
Serum
Conners'
Regression
Conners'
Prenatal: 12.9
Continuous
coefficient per
Commissions
(8.8-17.6)
Performance
ln-unit increase
Prenatal: -0.1 (-2.0, 1.8)
3 yr: 6.2 (4.5-
Test-II
in PFOS
3 Years: 1.0 (-1.5, 3.5)
9.9)
commissions t-
8 Years: 1.3 (-1.0,3.6)
8 yr: 3.6 (2.7-
score, omissions
Omissions
4.8)
t-score, hit
Prenatal: -0.8 (-5.2, 3.5)
reaction time,
3 Years: -0.1 (-4.4, 4.2)
tau (ms)
8 Years: -0.8 (-5.3, 3.8)
Females: 4.3 (-1.2, 9.9)
Virtual Morris
Males: -7.3 (-13.0, -1.7)
Water Maze
Hit reaction time
(VMWM)
Prenatal: -1.5 (-4.2, 1.2)
scores for
3 yr: -0.4 (-3.2, 2.5)
visual-spatial
8 yr: -2.5 (-6.0, 1.1)
learning
Tau
distance (pool
Prenatal: 6.0 (-23.2, 35.2)
units), learning
3 yr: 13.4 (-9.8, 36.5)
time (s),
8 yr: 5.8 (-22.1, 33.7)
memory
retention
Visual-spatial scores (VMWM)
distance (%),
Learning distance
and memory
Prenatal: 0.2 (-1.6, 1.7)
retention time
3 yr: -0.7 (-2.2, 0.7)
(s)
8 yr: -0.2 (-1.7, 1.3)
Learning time
Prenatal: -0.1 (-2.8, 2.6)
3 yr: -1.1 (-3.5, 1.2)
8 yr: -2.1 (-4.9,0.6)
D-206
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Memory retention distance
Prenatal: 2.8 (-1.3, 6.8)
3 yr: 0.3 (-4.7, 5.4)
8 yr: 2.1 (-2.9,7.0)
Memory retention time
Prenatal: 0.4 (-1.1, 1.9)
3 yr: -0.4 (-2.1, 1.3)
8 yr: 0.5 (-1.3,2.3)
Confounding: Maternal age, race/ethnicity, household income, maternal smoking status, maternal alcohol consumption, maternal depression,
HOME Score, marital status, maternal marijuana use, maternal IQ, maternal serum IPCBs. maternal blood lead levels, child sex
Vuong et al. United States, Cohort Pregnant
(2019) Recruitment: women and
Medium 2003-2006; their children
Follow-up at from the HOME
ages 3 and 8 yr study
N = 221
Serum Wechsler Regression
Maternal: Intelligence coefficient per
GM =12.4 Scale for ln-unit increase
8 Years: Children-Fourth in PFOS
GM = 3.9 Edition (WISC-
IV): full-scale
IQ, perceptual
reasoning,
processing
speed, verbal
comprehension,
working
memory
Full-Scale IQ
Prenatal: 2.2 (-0.9, 5.2)
3 Years: 0.8 (-2.4, 4.0)
8 Years: 1.6 (-2.7,5.8)
Perceptual Reasoning
Prenatal: 1.4 (-1.8, 4.7)
3 Years: 1.0 (-2.6,4.5)
8 Years: 2.8 (-2.1, 7.7)
Processing Speed
Prenatal: 1.3 (-2.0, 4.7)
3 Years: 1.6 (-1.9,5.1)
8 Years: 3.7 (-1.2,8.5)
Verbal Comprehension
Prenatal: 1.4 (-1.7, 4.5)
3 Years: 0.1 (-3.3, 3.5)
8 Years: -1.7 (-5.2, 1.8)
Working Memory
Prenatal: 2.6 (-0.8, 5.9)
3 Years:-0.1 (-3.4,3.2)
8 Years: 2.9 (-0.8, 6.5)
Confounding: Maternal age, race/ethnicity, household income, maternal marijuana use, maternal blood lead, maternal serum IPCBs and
cotinine, maternal depression, vitamin use, maternal IQ, marital status, HOME score, child sex, breastfed
D-207
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Vuong et al.
(2020a)
Medium
Wang et al.
(2015b)
Medium
United States,
Recruitment:
2003-2006;
Follow-up at
age 8 yr
Cohort
Mother-child Maternal serum Wide Range Regression
7.0 (-2.9, 16.9)
pairs with
children aged
8 yr from the
HOME study
N = 161
Mean= 13.9 Achievement coefficient per
(SD = 7.9) Test 4 (WRAT- loglO-unit
4) reading increase in
composite score PFOS
Confounding: Maternal age, race/ethnicity, education, household income, marital status, maternal depression, maternal serum cotinine,
maternal blood lead levels, maternal fish consumption, maternal IQ, child sex, HOME score
Taiwan,
Recruitment:
2000-2001;
Follow-up at
ages 5 yr
Cohort
Pregnant
women and
their children
aged 5 and 8 yr
from TMICS
N = 120
Serum
5 Years: 13.25
(9.75-17.50)
8 Years: 12.28
(9.50-16.30)
Full-Scale IQ,
Performance IQ,
Verbal IQ
Regression
coefficient per
log2-unit
increase in
PFOS
Full-Scale IQ
5 Years: -1.9 (-4.3, 0.5)
8 Years: -1.9 (-4.3, 0.4)
Performance IQ
5 Years: -2.2 (-4.7, 0.3)
8 Years: -1.6 (-4, 0.7)
Verbal IQ
5 Years: -1.7 (-4, 0.7)
8 Years: -1.3 (-3.6, 1.1)
Confounding: Maternal education, family annual income, children's age, sex, HOME score at IQ assessment
Zhang et al.
United States, Cohort
Pregnant
Serum
Basic reading,
Regression
Basic Reading
(2018a)
Recruitment:
women and
Maternal: 13.0
brief reading,
coefficient per
Maternal Serum: 3.2 (-2.0, 8.3)
Medium
2003-2006;
their children
(9.1-17.8)
letter word
ln-unit increase
Year 3 Serum: 1.1 (-4.8, 7.0)
Follow-up at
aged 3, 5, and
3 yr: 6.6 (4.6-
identification,
PFOS
ages 3, 5, and
7 yr from the
10.2)
passage
Brief Reading
7 yr
HOME study
8 yr: 3.6 (2.7-
comprehension
Maternal Serum: 2.9 (-2.2, 8.1)
N = 167
4.9)
measured by
Year 3 Serum: 3.2 (-2.6,9.1)
Woodcock
Johnson Test of
Letter Word Identification
Achievement-Ill
Maternal Serum: 2.0 (-2.7, 6.8)
(WJ-III)
Year 3 Serum: 2.1 (-3.4,7.5)
Reading
Passage Comprehension
composite, word
Maternal Serum: 1.7 (-1.9, 5.3)
reading,
Year 3 Serum: 3.5 (-0.5, 7.6)
D-208
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
sentence
Comprehension
measured by
Wide Range
Achievement
Test 4 (WRAT-
4)
Word Attack
Maternal Serum: 4.1 (-1.2, 9.5)
Year 3 Serum: 2.8 (-2.8, 8.4)
Reading Composite
Maternal Serum: 3.1 (-1.3, 7.5)
Year 3 Serum: 1.6 (-3.1, 6.4)
Year 8 Serum: 2.6 (-1.7, 6.9)
Word Reading
Maternal Serum: 3.1 (-1.0, 7.3)
Year 3 Serum: -0.3 (-4.8, 4.3)
Year 8 Serum: 4.4 (0.3, 8.4)
Sentence Comprehension
Maternal Serum: 3.2 (-1.8, 8.2)
Year 3 Serum: 2.5 (-3.1,8.1)
Year 8 Serum: 1.6 (-3.3, 6.5)
Confounding: Maternal age, race, education, household income, parity, smoking (serum cotinine concentration), maternal IQ, breastfeeding
duration, HOME score
General Population
Ding and Park
(2020)
Medium
United States,
2003-2016
Cross-sectional
Adults aged 20-
69 yr from
NHANES
N = 2,731
Serum
6.2 (3.5-10.5)
High and low
frequency
hearing
impairment
(HFHI and
LFHI)
OR per log2-
unit increase in
PFOS and
for > 90th
percentile
vs. < 90th
percentile
HFHI
OR (per doubling): 0.96 (0.85,
1.10)
OR (90th percentiles): 1.31 (0.75,
2.27)
LFHI
OR (per doubling): 0.87 (0.73,
1.03)
OR (90th percentiles): 0.72 (0.29,
1.75)
Confounding: Age, age square, sex, race/ethnicity, education level, PIR, smoking status, BMI, noise exposures (occupational, recreational,
firearm noise), NHANES cycles
D-209
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Gallo et al.
(2013)
Medium
Lenters et al.
(2019)
Medium
Li (2020)
Medium
United States,
2005-2006
Cross-sectional
Adults aged 50+ Serum
years from the Range = 0.25-
C8 Health 759.2
Project
N = 21,024
Memory OR per
impairment doubling of
(self-reported) PFOS and by
quintiles
0.93 (0.90, 0.96)
Q2: 0.96 (0.87, 1.07)
Q3: 0.86 (0.78,0.96)
Q4: 0.87 (0.78, 0.96)
Q5: 0.85 (0.76, 0.94)
p-trend < 0.001
Comparison: Logarithm base not specified.
Results: Lowest quartile used as reference.
Confounding: Age, ethnicity, gender and school level, household income, physical activity, alcohol consumption, cigarette smoking
Norway,
Recruitment:
2003-2009;
Follow-up:
2008-2016
Cohort
Children and
adults from
HUMIS
N = 1,199
Breast milk
117.732 ng/L
(80.000-
160.000 ng/L)
ADHD
OR per IQR
increase in ln-
unit PFOS
1.75 (1.11, 2.76), p-value = 0.017
Confounding: Maternal age, childbirth year, maternal education, parity, smoking during pregnancy, small-for-gestational age, preterm birth,
maternal pre-pregnancy BMI, single mother around perinatal period, maternal fish intake
United States,
1999-2016
Cross-sectional
Adults aged
20+ years from
NHANES
N = 2,525
Serum
8.00 (Range:
0.14-392)
Hearing
threshold > 25 d
B by frequency
OR by quartiles
2,000 Hz
Q2: 0.70 (0.46, 1.06)
Q3: 1.12 (0.76, 1.65)
Q4: 1.60 (1.09, 2.37),
p-trend < 0.0001
3,000 Hz
Q2: 0.76 (0.53,
Q3: 1.00 (0.71,
Q4: 1.20 (0.85,
p-trend = 0.02
1.08)
1.41)
1.71),
4,000 Hz
Q2: 0.69 (0.50, 0.97)
Q3: 0.89 (0.65, 1.24)
Q4: 1.02 (0.73, 1.44),
p-trend = 0.14
Results: Lowest quartile used as reference.
Confounding: Age, sex, BMI, education, ethnicity group, family income, sample weights
D-210
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Shrestha et al. United States,
(2017) 2000-2002
Medium
Cross-sectional
Residents aged
55-74 yr who
lived adjacent to
Hudson River
N = 126
Serum Affective state:
33.7 (23.3-50.8) Beck
Depression
Inventory (BDI)
total score,
State-Trait
Anxiety
Inventory state
and trait t-scores
Attention: Trail
making test Part
A (In-
transformed
time to
complete)
Executive
function: Stroop
color word test
t-score, Trail
making test part
B (In-
transformed
time to
complete),
Wisconsin Card
Sorting Test
preservative In-
transformed
error and
response
Memory and
learning:
California
Regression
coefficient per
IQR increase in
ln-unit PFOS
Depression:
0.25 (-0.77, 1.26), p-value = 0.63
CVLT-Total score:
-0.14 (-0.59, 0.31)
Wisconsin Card Sorting Test
Perseverative Error:
-0.14 (-0.30, 0.02), p-value = 0.09
Perseverative Response:
-0.16 (-0.34, 0.01), p-value = 0.07
Wechsler Memory Scale
Logical Memory
Immediate Recall: -0.7 (-1.92,
0.52), p-value = 0.26
Delayed Recall: -0.14 (-1.29,
1.01), p-value = 0.81
Visual Reproduction
Immediate Recall: 0.56 (-0.16,
1.29), p-value = 0.13
Delayed Recall: 0.79 (0.03, 1.55),
p-value = 0.04
No statistically significant
associations: State-Trait Anxiety
Inventory, Stroop color word test,
trail-making tests, motor function
outcomes, visuospatial outcomes
D-211
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Verbal Learning
Test total and
subscores,
Wechsler
Memory Scale
logical memory
and visual
reproduction
immediate and
delayed recall
scores
Motor function
(dominant and
non-dominant
hands): finger
tapping test
average scores,
grooved
pegboard test
ln-transformed
times to
completion,
static motor
steadiness test
ln-transformed
total numbers of
contacts and
times touching
Dominant hand
reaction time
Visuospatial
function:
Wechsler Adult
D-212
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Resultsb
Intelligence
Scale-Revised
total scores for
block design
and digit
symbol coding
Confounding: Age, sex, education, serum total PCB
Pregnant Women
Vuong et al.
(2020b)
Medium
United States
Recruitment:
2003-2006
Follow-up:
~20 wk
gestation and
postpartum
(4 wk, 1, 2, 3,
5, and 8 yr)
Cohort Pregnant Maternal serum Beck Relative risk
women from the 13.3(9.0-17.9) Depression and OR per ln-
HOME study Inventory-II unit increase in
N = 355 (BDI-II) PFOS
Medium Score Trajectory: 0.9 (0.6,
1.5)
High Score Trajectory: 0.6 (0.3,
1.2)
OR for score >13 from pregnancy
to 8 yr postpartum: 1.0 (0.7, 1.5)
Confounding: Age, race/ethnicity, household income, maternal marijuana use, serum cotinine and PCBs, IQ, marital status, parity
Notes: ADHD = attention deficit hyperactivity disorder; ADI-R = Autism Diagnostic Interview-Revised; ALSPAC = Avon Longitudinal Study of Parents and Children;
ASD = autism spectrum disorder; ASQ-3 = Ages and Stages Questionnaire-3; BDI = Beck Depression Inventory; BDI-II = Beck Depression Inventory II; BMt = body mass
index; BRIEF = Behavior Rating Inventory of Executive Function; BSID-II = Bayley Scales of Infant Development, Second Edition; CDI = Comprehensive Developmental
Inventory; CHARGE = Childhood Autism Risk from Genetics and Environment; CI = confidence interval; CVLT P = California Verbal Learning Test; DaF088 = Danish Fetal
Origins 1988; CRP = C-reactive protein; DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; EMA = Early Markers for Autism;
GM = geometric mean; HFHI = high frequency hearing impairment; HOME = Health Outcomes and Measures of the Environment; HUMIS = Human Milk Study;
ID = intellectual disability; IQ = intelligence quotient; IQR = interquartile range; KBIT-2 = Kaufman Brief Intelligence Test Second Edition; LINC = Linking Maternal Nutrition
to Child Health; LFHI = low frequency hearing impairment; MCDI = MacArthur Communicative Development Inventories; MDI = Mental Development Index; mo = months;
MoBa = Mother, Father, and Child Cohort Study; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; PCBs = polychlorinated biphenyls;
PDI = Psychomotor Development Index; PFOS = perfluorooctane sulfonic acid; PPVT-III = Peabody Picture Vocabulary Test; RR = risk ratio; SD = standard deviation;
SDQ = Strengths and Difficulties Questionnaire; SES = socioeconomic status; TMtCS = Taiwan Maternal and Infant Cohort Study; VMWM = Virtual Morris Water Maze;
WISC-IV = Wechsler Intelligence Scale for Children-Fourth Edition; SRS = Social Responsiveness Scale; T2 = tertile 2; T3 = tertile 3; WJ-III = Woodcock Johnson Test of
Achievement-Ill; WPPSI-R = Wechsler Primary and Preschool Scales of Intelligence-Revised; WRAML2 = Wide Range Assessment of Memory and Learning Second Edition;
WRAT-4 = Wide Range Achievement Test 4; WRAVMA = Wide Range Assessment of Visual Motor Abilities; yr = year(s).
a Exposure levels are reported as median unless otherwise noted.
b Results reported as effect estimate (95% confidence interval), unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D-213
-------
APRIL 2024
D.9 Renal
Table D-18. Associations Between PFOS Exposure and Renal Effects in the General Population
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels Outcome
(ng/mL)a
Comparison
Select Resultsb
General Population
Blake et al.
United States, Cohort Adults and Serum eGFR
Percent change
All:
(2018)
1991-2008 children from 28.4 (21.6-35.7)
per IQR
Repeated measures model: -0.68
Low
FCC
increase in
(-1.9, 0.54); p-value = 0.27
N = 192(115
PFOS
Latent model: -1.72 (-3.29, -0.15);
females, 77
p-value = 0.03
males)
Females: -1.32 (-3.37, 0.73),
p-value = 0.64
Males: 0.71 (-2.75,4.16),
p-value = 0.69
p-value for interaction by
sex = 0.46
Confounding: Age, year of measurement, sex, education, income, marital status, and BMP
Lin et al. (2013)
Taiwan, Cross-sectional Adolescents and Serum Uric acid
Mean
< 25th percentile: 6.09 (0.13)
Low
2006-2008 young adults 8.65(5.41- (mg/dL)
concentration by 25th-50th: 6.13 (0.13)
fromYOTA 13.52)
PFOS
50th-75th: 6.04 (0.13)
study, 12-30 yr,
percentiles
> 75th: 6.12(0.13)
N = 644
p-value for trend = 0.891
Results: Effect estimates are provided with standard error in parentheses.
Confounding: Age, gender, smoking status, alcohol drinking, BMI
Conway et al.
(2018)
Low
United States,
2005-2006
Cohort
Adults, C8
Health Project,
Diabetic = 5,21
0, non-
diabetic = 48,44
0
Serum
Diabetic: 21.2
(13.7-31.4)
Non-diabetic:
20.2 (13.6-29.1)
CKD (eGFR of
<60 mL/min/1.7
3 m2)
OR per ln-unit
increase in
PFOS
Diabetics: 0.81 (0.73, 0.9)
Non-diabetic: 1.09 (1.03, 1.16)
Confounding: Age, sex, BMI, HDLc, LDLc, white blood cell count, CRP, hemoglobin, and iron
Liu et al.
(2018b)
Low
United States,
2013-2014
Cross-sectional
Adults from
NHANES,
18+ years,
Serum
GM = 5.28
(SE= 1.02)
Total protein
(g/dL)
Regression
coefficient per
0.05 (SE = 0.02); p-value < 0.01
D-214
-------
APRIL 2024
Reference,
Confidence
Population, Exposure
(•cation, Design Ages, Matrix, Levels Outcome Comparison Select Resultsb
YCftrs
N (ng/mL)a
N=1871 ln-unit increase
in PFOS
Confounding: Age, gender, ethnicity, smoking status, alcohol intake, household income, waist circumference, and medications
(antihypertensive, anti-hyperglycemic, and anti-hyperlipidemic agents)
Arrebola et al.
(2019)
Low
Spain, Cross-sectional Adults, Serum Uric acid OR Uric acid
2009-2010 BIOAMBIENT. 7.23(5.14- (mg/dL), (hyperuricemia), Wet-basis and lipid-basis models:
ES study 10.11) hyperuricemia orregression 0.06 (-0.03,0.16); p-value = 0.192
N = 342 coefficient per Wet-basis model with adjustment
log-unit increase for serum lipids:
in PFOS 0.06 (-0.03,0.157);
p-value = 0.207
Hyperuricemia
Wet-basis and lipid-basis models:
1.70 (0.86, 3.49); p-value = 0.138
Wet-basis model with adjustment
for serum lipids:
1.67 (0.84, 3.41); p-value = 0.151
Outcome: Hyperuricemia defined as at least one of a) serum uric acid levels > 7.0 mg/dL in males or > 6.0 mg/dL in females, at recruitment
or in previous screenings, b) had been prescribed any pharmacological treatment for lowering uric acid levels, and/or c) had been diagnosed
with gout by a clinician.
Comparison: Logarithm base not specified.
Confounding: Sex, age, BMI, weight loss during the last 6 mo, region of recruitment, smoking habit, alcohol consumption, education, place
of residence
Chen et al.
(2019a)
Low
Croatia, Cross-sectional Adults, 44- Plasma Uric acid Regression Uric acid: -4.87 (-25.63, 15.89)
2007-2008 56 yr GM = 8.91 (|imol/L). coefficient per Creatinine:-3.36 (-7.96, 1.24)
N = 122 (range = 2.36- creatinine ln-unit increase
33.67) (|imol/L) in PFOS
Confounding: Age, sex, education, socioeconomic status, smoking, dietary pattern, and physical activity
Jain and
Ducatman
(2019c)
Low
United States, Cross-sectional Adults from Serum Levels of Regression Albumin in urine
2005-2014 NHANES, Levels not albumin in urine coefficient per Per loglO-unit increase: -0.08
>20yr reported (logl0-|ig/mL). loglO-unit p-value <0.01
N = 8,220 creatinine in increase in Negative associations across eGFR
urine (loglO- PFOS, or stages
mg/dL), percent change
D-215
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
albumin-to-
creatinine ratio
in urine (log 10-
mg/g), albumin
in serum (loglO-
mg/dL),
creatinine in
serum (loglO-
mg/dL)
per 10% Percent change per 10% increase:
increase in -0.75, p-value < 0.05
PFOS p-value for gender and
race/ethnicity interaction =0.10
Creatinine in urine
Per loglO-unit increase: 0.04
p-value = 0.01
Positive associations across eGFR
stages
Percent change per 10% increase:
0.38
p-value < 0.05
p-value for gender and
race/ethnicity interaction = 0.02
Albumin-to-creatinine ratio in urine
Per loglO-unit increase: -0.12
p-value <0.01
Negative associations across eGFR
stages
Percent change per 10%
increase: -1.13
p-value < 0.05
p-value for gender and
race/ethnicity interaction = 0.73
Albumin in serum
Per loglO-unit increase: 0.01
p-value <0.01
Positive associations across eGFR
stages
Percent change per 10% increase:
0.11
p-value < 0.05
D-216
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
p-value for gender and
race/ethnicity interaction = 0.68
Creatinine in serum
Per loglO-unit increase: 0.01
p-value = 0.01
Positive associations in GF-1, GF-
2, GF-3A
Negative association in GF-3B/4
Percent change per 10% increase:
0.11
p-value < 0.05
p-value for gender and
race/ethnicity interaction < 0.01
GF Stages: GF-1: GFR> 90 mL/min/1.73 m2; GF-2: GFR between 60 and 90 mL/min/1.73 m2; GF-3A: GFR between 45 and
60 mL/min/1.73 m2; GF-3B/4: GFR between 15 and 45 mL/min/1.73 m2.
Confounding: Gender, race/ethnicity, age, loglO(BMI), Iogl0(serum cotinine), PIR, NHANES survey period
Jain and
United States, Cross-sectional Adults from Serum
Uric acid
Regression
Males
Ducatman
2007-2014 NHANES, Males:
(mg/dL) by
coefficient per
GF-1: 0.01, p-value = 0.01
(2019a)
> 20 yr, GM= 10.51
glomerular
loglO-unit
GF-2: 0.02, p-value = 0.05
Low
Males = 3,330, (9.88-11.18)
filtration (GF)
increase in
GF-3A: -0.01, p-value = 0.66
females = 3,506
stage
PFOS
GF-3B: -0.04, p-value < 0.01
Females:
GM = 6.58
Females
(6.22-6.96)
GF-1: 0.02, p-value = 0.04
GF-2: 0.01, p-value = 0.52
GF-3A: 0.04, p-value <0.01
GF-3B: 0.01, p-value = 0.64
GF Stages: GF-1: eGFR > 90 mL/minper 1.73 m2; GF-2: 60 < eGFR < 90 mL/minper 1.73 m2; GF-3A: 45 < eGFR < 60 mL/minper
1.73 m2; GF-3B/4: 15 < eGFR < 45 mL/minper 1.73 m2.
Confounding: Gender, race/ethnicity, age, loglO(BMI), logl0(serum cotinine), PIR, NHANES survey period
Wang et al.
China, 2015- Cross-sectional Adults, Isomers Serum
CKD, eGFR
OR (CKD) or
CKD (OR)
(2019b)
2016 of C8 Health 24.22(14.62-
regression
Per ln-unit increase: 1.71 (0.92,
Low
Project 37.19)
coefficient per
1.49), p-value = 0.205
ln-unit increase
Q2: 1.19 (0.67,2.09)
D-217
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Exposure
Design Ages, Matrix, Levels
N (ng/mL)a
Outcome Comparison
Select Resultsb
N = 1612
(males = 1204,
females = 408)
in PFOS, or by
quartiles
Q3: 1.42 (0.82,2.47)
Q4: 1.34 (0.77, 2.33)
p-value for trend = 0.617
eGFR
Per ln-unit increase:
All: -0.91 (-1.83, 0),
p-value = 0.05
Males: -0.73 (-1.82, 0.37)
p-value = 0.193
Females: -0.62 (-0.24, 1.15)
p-value = 0.491
p-value for interaction by
sex = 0.419
Q2
Q3
Q4
-1.25 (-3.14,0.63)
-1.59 (-3.53, 0.35)
-1.77 (-3.74,0.19)
p-value for trend = 0.086
Outcome: CKD defined as eGFR < 60 mL/min per 1.73 m2.
Results: Lowest quartile used as reference group.
Confounding: Age, sex, BMI, education, annual income, regular exercise, cigarette smoking, drinking alcohol, family history of CKD, total
cholesterol
Zeng et al.
China,
Cross-sectional Adults, Isomers
Serum
Hyperuricemia,
OR
Hyperuricemia
(2019c)
2015-2016
of C8 Health
24.22 (14.62-
uric acid
(hyperuricemia)
All: 1.17 (0.99, 1.39)
Low
Project
37.19)
(mg/dL)
or regression
Males: 1.11 (0.92, 1.34)
N = 1612
coefficient (uric
Females: 1.27 (0.8, 2)
(males = 1204,
acid) per loglO-
p-value for interaction by
females = 408)
unit increase in
sex = 0.118
PFOS
Uric acid
All: 0.1 (0.02,0.18),
p-value = 0.017
Males: 0.07 (-0.03,0.18)
Females: 0.11 (-0.01, 0.18)
p-value for interaction by
sex = 0.209
D-218
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Outcome: Hyperuricemia defined as serum uric acid levels > 7.0 mg/dL in males or > 6.0 mg/dL in females.
Confounding: Age, sex, BMI, income, drinking, smoking, career, exercise, offal consumption, fish and seafood consumption, serum
creatinine
Scinicariello et United States,
al. (2020b) 2009-2014
Low
Cross-sectional
Adults from
NHANES
N = 4915 (no
CKD = 4,103;
CKD = 874)
Serum
GM = 6.98
(SE = 0.23)
Uric acid
(mg/dL),
hyperuricemia,
gout
OR
(hyperuricemia,
gout), or
regression
coefficient (uric
acid) by
quartiles
Uric acid
Overall population
Q2: 0.13 (0.01, 0.24)
Q3: 0.21 (0.05,0.37)
Q4: 0.29 (0.14,0.44)
p-value for trend = 0.003
Participants with CKD
Q2: 0.6 (0.15, 1.05)
Q3: 0.31 (-0.02,0.7)
Q4: 0.38 (0.06, 0.83)
p-value for trend = 0.08
Participants without CKD
Q2: 0.03 (-0.1,0.15)
Q3: 0.13 (-0.02, 0.28)
Q4: 0.2 (0.06, 0.34)
p-value for trend = 0.02
Hyperuricemia
Overall population
Q2: 1.1 (0.84, 1.45)
Q3: 1.27(0.92, 1.76)
Q4: 1.45 (1.03,2.03)
p-value for trend = 0.15
Participants with CKD
Q2: 1.93 (0.91, 4.06)
Q3: 0.85 (0.4, 1.77)
Q4: 1.15 (0.53,2.5)
p-value for trend = 0.12
Participants without CKD
Q2: 0.94 (0.68, 1.3)
Q3: 1.26 (0.89, 1.79)
Q4: 1.35 (0.92, 1.99)
p-value for trend = 0.19
D-219
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Gout
Overall population
Q2: 1.17 (0.54,2.53)
Q3: 1.23 (0.54,2.53)
Q4: 1.46 (0.67, 3.16)
p-value for trend = 0.79
Participants with CKD
Q2: 0.88 (0.26, 2.92)
Q3: 1.08 (0.38,3.07)
Q4: 1.08 (0.39, 2.94)
p-value for trend = 0.97
Participants without CKD
Q2: 1.73 (0.6, 4.94)
Q3: 1.56 (0.51,4.78)
Q4: 1.93 (0.71, 5.22)
p-value for trend = 0.58
Outcomes: CKD defined as eGFR < 60 mL/min per 1.73 m2 and/or albuminuria. Hyperuricemia defined as serum uric acid levels
> 7.0 mg/dL in males or > 6.0 mg/dL in females. Gout was self-reported diagnosis from a health professional.
Results: Lowest quartile used as reference group.
Confounding: Race/ethnicity, age, sex, education, alcohol, smoking, serum cotinine, BMI, diabetes, hypertension, CKD
Children and Adolescents
Geiger et al.
United States,
Cross-sectional Children and
Serum
Hyperuricemia,
OR
Hyperuricemia
(2013)
1999-2000;
adolescents
Mean= 18.4
uric acid
(hyperuricemia)
Per In increase: 1.37 (1.06, 1.76)
Low
2003-2008
from NHANES,
(SE = 0.5)
(mg/dL)
or regression
Q2: 1.17(0.8, 1.72)
12-18 yr,
coefficient (uric
Q3: 1.18 (0.74, 1.87)
N = 1,772
acid) per ln-unit
Q4: 1.65 (1.1,2.49)
increase in
p-value for trend = 0.022
PFOS or by
quartiles
Uric acid
Per 1-ln increase: 0.09 (0.02, 0.17)
Q2: 0.03 (-0.1,0.16)
Q3: 0.09 (-0.04,0.21)
Q4: 0.12 (-0.01, 0.26)
p-value for trend = 0.058
D-220
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Kataria et al.
(2015)
Low
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Outcome: Hyperuricemia defined as serum uric acid levels > 6 mg/dL.
Results: Lowest quartile as reference group.
Confounding: Age, sex, race/ethnicity, BMI, annual household income, moderate activity, total cholesterol, serum cotinine
United States,
2003-2010
Cross-sectional
Children and
adolescents
from NHANES,
12-19 yr,
NHANES
N = 1,962
Serum
3.5 (2.5-4.7)
eGFR Regression
(min/mL/1.73 m coefficient by
2), uric acid quartiles
(mg/dL),
creatinine
(mg/dL)
eGFR
Q2: -5.24 (-9.75, -0.73),
p-value < 0.05
Q3:-7.21 (-12.21,-2.21),
p-value <0.01
Q4: -9.47 (-14.68, -4.25),
p-value <0.001
Uric acid
Q2: 0.095 (-0.081, 0.27)
Q3: 0.046 (-0.1,0.19)
Q4: 0.19 (0.032,0.34),
p-value < 0.05
Creatinine
Q2: 0.021 (-0.007, 0.049)
Q3: 0.038 (0.008, 0.068),
p-value < 0.05
Q4: 0.04 (0.01, 0.071),
p-value <0.01
Results: Lowest quartile as reference group.
Confounding: Sex, PIR, caregiver education, serum cotinine, prehypertension, insulin resistance, BMI, hypercholesterolemia, race/ethnicity
categories
Qinetal. (2016) Taiwan,
Cross-sectional Children from
Serum Uric acid
Regression
Uric acid
Low 2009-2010
GBCA Study,
All: 28.9 (14.1- (mg/dL),
coefficient per
All: 0.05 (-0.03,0.13)
12-15 yr,
43.0) hyperuricemia
ln-unit increase
Boys: 0.05 (-0.04,0.15)
N = 225 (123
Boys: 29.9
in PFOS (uric
Girls: 0.01 (-0.14, 0.16)
girls, 102 boys)
(13.0-43.8)
acid); OR scaled
Girls: 28.8
with increasing
Hyperuricemia (OR)
(14.8-42.6)
quartiles
All: 1.35 (0.95, 1.93)
(hyperuricemia)
Boys: 1.4 (0.88, 2.21)
Girls: 1.51 (0.79,2.89)
D-221
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Outcome: Hyperuricemia defined as uric acid level > 6 mg/dL.
Results: Lowest quartile used as the reference group.
Confounding: Age, gender, BMI, parental education level, exercise, environmental tobacco smoke exposure, and serum creatinine
Khalil et al.
United States Cross-sectional
Obese children,
Serum
Creatinine
Regression
0 (-0.02, 0.03)
(2018)
2016
8-12 yr
2.79
(mg/dL)
coefficient per
Low
N = 40
(IQR = 2.10)
unit increase in
PFOS
Confounding: Age, sex, race
Pregnant Women
Nielsen et al.
Sweden, Cohort
Pregnant
Serum
eGFR:
Spearman's
Cross-sectional correlations
(2020)
2009-2014
women,
Early
LMrev, CKD-
correlation
consistently weak and non-
Low
PONCH study
pregnancy: 5.6
EP Icreatinine,
coefficient
significant
N = 73
(5th-95th
CAPA, CKD-
Early to late pregnancy changes:
percentile = 2.6
EP Icystatin C,
No significant associations
-11.5)
mean of LMrev
Late pregnancy:
and CAPA,
eGFR:
4.8 (5th-95th
mean of CKD-
LMrev: 0.02, p-value = 0.85
percentile =1.9
EP Icreatinine and
CKD -EP Icreatinine: 0.02,
-8.4)
CKD -EP Icystatin C
p-value = 0.87
CAPA: -0.04, p-value = 0.73
Glomerular pore
CKD-EPIcystatmc: —0.05,
size
p-value = 0.66
mean of LMrev and CAPA: -0.04,
p-value = 0.76
mean of CKD-EPIcreatmine and CKD-
EPICyStatinc: -0.06, p-value = 0.63
Glomerular pore size:
CAPA/LMrev: -0.05,
p-value = 0.68
CKD-EP Icystatin c/CKD-EPIcreatinine:
-0.06, p-value = 0.63
Outcome: Glomerular pore size is estimated as the ratio between eGFRcystatmc and eGFRcreatmme and was calculated by the two ratios provided.
Confounding: Number of days between sampling, pregnancy-induced change in BMI
Occupational Populations
D-222
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome Comparison
Select Resultsb
Rotander et al. Australia, 2013 Cross-sectional Firefighters with Serum Uric acid Regression
(2015) past exposure to 66 (range = 3.1- ((imol/L) coefficient per
Low AFFF, 17-66 yr 391) loglO-unit
old increase in
N = 137 (97% PFOS
male)
0.045 (SE = 0.047), p-value = 0.342
Confounding: Age, sex, BMI, smoking status, total serum protein, PFOA, PFHxS
Notes: AFFF = aqueous film-forming foam; BMI = body mass index; CAPA = Caucasian Asian Pediatric Adult; CKD = chronic kidney disease; CKD-EPI = Chronic Kidney
Disease Epidemiology Collaboration study; CRP = C-reactive protein; eGFR = estimated glomerular filtration rate (mL/min per 1.73 m2); FCC = Fernald Community Cohort;
GBCA = Genetic Biomarkers Study for Childhood Asthma; GF = glomerular filtration; GM = geometric mean; HDLc = high-density lipoprotein cholesterol; IQR = interquartile
range; LDLc = low-density lipoprotein cholesterol; LMrev = Lund Malmo Revised; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio;
PFFIxS = perfluorohexane sulfonic acid; PIR = poverty income ratio; PONCFI = Pregnancy Obesity Nutrition and Child Flealth study; Q2 = quartile 2; Q3 = quartile 3;
Q4 = quartile 4; SD = standard deviation; SE = standard error; YOTA = Young Taiwanese Cohort Study; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.10 Hematological
Table D-19. Associations Between PFOS Exposure and Hematological Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
„ , . Exposure Matrix,
Population, Ages, Levels
N (ng/mL)
Outcome
Comparison
Select Results"
General Population
Etzel et al. (2019) United States,
Medium 2003-2010
Cross-sectional Children and adults Serum,
fromNHANES, Median = 15.1
> 12 yr of age, (9.1-23.8)
N = 7,040
Vitamin D
deficiency
(<50 ng/mL), 25-
hydroxy Vitamin
D (25(OH)D,
nmol/L)
Regression
coefficient or
prevalence OR
(POR) per
doubling of
PFOS, or by
quintiles
Per doubling of PFOS:
Vitamin D deficiency
POR: 1.05 (0.97, 1.13)
25-hydroxy Vitamin D
-0.9 (-1.5, -0.2)
Q5: -2.8 (-4.7, -0.8)
60+years: -1.7 (-2.9,
-0.5)
D-223
-------
APRIL 2024
„ , . Exposure Matrix,
Reference, Location, . Population, Ages, T , . c . , D a
„ Design i^eveis Outcome Comparison Select Results
Commence Years N
(ng/mL)
No other statistically
significant associations or
trends
Results: Lowest quintile used as reference group.
Confounding: Gender, race/ethnicity, age, BMI category, vitamin D supplement use, poverty to income ratio, smoking status, 6-mo
examination time period0
Chen et al. (2019a) Croatia
Medium 2007-2008
Jain (2020a)
Medium
Cross-sectional Adults, 44-56 yr of Plasma, Calcium in serum Regression -0.05 (-0.09,-0.01),
age, N= 122 GM=8.91(min= (mmol/L) coefficient per p-value<0.05
2.36, max = 33.67) ln-unit
increase in
PFOS
Confounding: Age, sex, education, socioeconomic status, smoking, dietary pattern, and physical activity
United States
2003-2016
Cross-sectional
Adults from
NHANES, > 20 yr
of age, N = 11,251
Serum,
Non-anemic
males: GM = 12.0
(95% CI: 11.5,
12.7)
Non-anemic
females: GM = 8.1
(95% CI: 7.7, 8.5)
anemic males:
GM = 10.7 (95%
CI: 9.2, 12.5)
anemic females:
GM = 5.0 (95%
CI: 4.4, 5.8)
Whole blood
hemoglobin
(WBHGB)
(logl0-g/dL)
Regression Non-anemic males: 0.009,
coefficient per p-value <0.01
loglO-unit
increase in
PFOS
Non-anemic
females:0.006,
p-value <0.01
Anemic males: 0.023,
p-value <0.01
Anemic females: 0.024,
p-value <0.01
Confounding: Age, BMI, PIR, serum cotinine, survey year, daily alcohol intake
D-224
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population, Ages,
N
Exposure Matrix,
Levels
(ng/mL)
Outcome Comparison
Select Results"
Khalil et al. (2018) United States,
Low 2016
Cross-sectional Children with Serum, 25-hydroxy
obesity, 8-12 yr of Median = 2.79 vitamin D
age, N = 47 (IQR = 2.10) (ng/mL)
Regression -0.10 (-1.54, 1.33)
coefficient per
unit increase
inPFOS
Confounding: Age, sex, race
van den Dungen et
al. (2017)
Low
The
Netherlands,
2015
Cross-sectional
Dutch men, 40-
70 yr of age, with
habitual eel
consumption of at
least one portion
a month, N = 37
Serum,
Median = 40 ng/g
wet weight (15-
93)
Hemoglobin
(Hb), Hematocrit
(Ht), Retinol
(units not
provided)
Regression Hb: -0.112 (-0.477,
coefficient 0.250)
Ht: -0.095 (-0.455, 0.263)
Retinol: 0.205 (-0.146,
0.561)
Confounding: Age, waist-to-hip ratio
Notes: aPTT = activated partial thromboplastin time; BMI = body mass index; CI = confidence interval; GM = geometric mean; HIV = human immunodeficiency virus;
Hb = hemoglobin; Ht = hematocrit; IQR = interquartile range; mo = month; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; POR = prevalence
odds ratio; PPT = prothrombin time; WBHGB = whole blood hemoglobin; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval) unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.ll Respiratory
Table D-20. Associations Between PFOS Exposure and Respiratory Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design
Population,
Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Agier et al.
(2019)
Medium
France, Greece, Cohort
Lithuania,
Norway, Spain,
United
Kingdom
2003-2009
Pregnant women
and their
children, ages 6-
12 yr,
N= 1,033
Maternal and child's FEV1
serum, plasma, or whole
blood
Prenatal (maternal)
Median =6.6
(IQR = 5.8)
Regression
coefficient per
log2-unit
increase in
PFOS
Prenatal: 0.1
(-1.1, 1.3), p-value = 0.89
Postnatal: 0.5
(-0.6, 1.6),
p-value = 0.38
D-225
-------
APRIL 2024
Reference, Location, __ . Population, Exposure Matrix, „ ^ _ ,, h
„ „ Design . T , , , T,a Outcome Comparison Results"
Commence Years Ages, N Levels (ng/mL)a
Postnatal (child)
Median = 2.1
(IQR = 1.9)
Confounding: Center of recruitment, child's sex, child's age, child's height, parental country of birth, breastfeeding duration, season of
conception, presence of older siblings, parental education level, maternal age, maternal pre-pregnancy BMI, postnatal passive smoking status,
prenatal maternal active, passive smoking status0
Gaylord et al.
(2019)
Medium
New York, U.S. Cross-
2014-2016 sectional
Adolescents and
young adults,
ages 13-22 yr,
N = 287
Serum,
Comparison group:
median = 2.75 (range
0.60, 27.80)
WTCHR group:
median = 3.72 (range:
1.01, 14.20)
FEV1
FVC
FEV1/FVC
TLC
RV
FRC
Resistance at an
oscillation
frequency of
5 Hz, 5-20 Hz,
20 Hz
Regression
No statistically significant
coefficient per differences observed between
log-unit increase groups for the measured
in PFOS outcomes, p-values > 0.05
Comparison: Logarithm base not specified.
Confounding: Sex, race/ethnicity, age, BMI, tobacco smoke exposure
Impinen et al. Norway Cohort Infants followed
(2018) 1992-2002 up at 2 yr and 10,
Medium N = 641
Cord blood,
Median =5.2 (4.0, 6.6)
Oslo Severity
Score (1-5 vs. 0)
Oslo Severity
Score (6-12 vs.
0)
Reduced lung
function at birth
OR per log2-
unit increase in
PFOS
1.71 (1.16,2.53),
p-value = 0.007
1.15 (0.71, 1.84),
p-value = 0.576
0.86 (0.43, 1.72),
p-value = 0.680
Outcome: Reduced lung function at birth: Lung function (tPTEF/tE) with standardized z-score, and binary variable of decreased lung function
(cutoff <0.20).
Confounding: Sex
D-226
-------
APRIL 2024
Reference, Location, __ . Population, Exposure Matrix, „ ^ _ ,, h
„ „ Design . T , , , T,a Outcome Comparison Results"
Commence Years Ages, N Levels (ng/mL)a
Manzano-
Spain
Cohort
Pregnant women
Maternal blood,
FEV1,
Regression
No statistically significant
Salgado et al.
2003-2015
and their
Median =6.06 (4.52,
FVC
coefficient per
associations for the measured
(2019)
children,
7.82)
FEV1/FVC,
log2-unit
outcomes
Medium
followed up at
FEF25%-75%
increase in
ages 1.5, 4, and
PFOS
7 yr,
N = 503 (4 yr)
N = 992 (7 yr)
Confounding: Maternal age at delivery, parity, previous breastfeeding, pre-pregnancy BMI, region of residence, and country of birth
Qin et al.
Taiwan,
Case-
Children with
Serum,
FEV1
Regression
Statistically significant
(2017)
2009-2010
control
asthma and
Children with asthma:
FVC
coefficient per
associations in children with
Medium
without asthma,
Median =31.51 (19.60,
FEF25%-75%
ln-unit increase
asthma:
ages 10-15,
91.69)
PEF
in PFOS
N = 132 (with
FEV1: -0.06 (-0.10,-0.02),
asthma)
Children without asthma:
p-value < 0.05
N = 168 (without Median = 28.83 (12.39,
asthma) 42.02) FVC:-0.06 (-0.10,-0.01),
p-value < 0.05
Confounding: Age, sex, BMI, parental education level, exercise, environmental tobacco smoke exposure, and month of survey
Notes: BMI = body mass index; IQR = Interquartile range; FEF25%-75% = Forced Expiratory Flow at 25%-75%; FEV1 = Forced Expiratory Volume in 1 s; FRC = Functional
Residual Capacity; FVC = Forced Vital Capacity; IQR = interquartile range; OR = odds ratio; PEF = Peak Expiratory Flow rate; RV = residual volume; TLC = total lung
capacity; U.S. = United States; WTCUR = World Trade Center Health Registry; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise noted.
b Results reported as effect estimate (95% confidence interval), unless otherwise noted.
c Confounding indicates factors the models presented adjusted for.
D.12 Musculoskeletal
Table D-21. Associations Between PFOS Exposure and Musculoskeletal Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Population, Exposure
Study Design Ages, Matrix, Levels" Outcome
N (ng/mL)
Comparison
Resultsb
Children and Adolescents
D-227
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels"
(ng/mL)
Outcome Comparison
Resultsb
Jeddy et al.
(2018)
Medium
England,
2009
1991- Cohort
Females from Maternal serum Area adjusted Regression
the ALSPAC 20.2 (15.6-25.5) BMC (g), bone coefficient per
Height:
-0.11 (-0.19, -0.02)
Study,
area (cm2), unit increase in
Total lean mass:
Age 17,
BMC (g), BMD, PFOS
-75.61 (-131.12,-20.1)
N = 221
cortical bone
area (cm2),
Bone area: -4.07 (-7.38, -0.76)
cortical BMC
(mg), cortical
BMC: -5.94 (-10.96, -0.92)
BMD (mg/cm2),
cortical
No other statistically significant
thickness (mm),
associations
endosteal
circumference
(mm), height
(cm), periosteal
circumference
(mm), total
femoral neck
BMD (g/cm2),
total hip BMD
(g/cm2), total
lean mass (g)
Confounding: Maternal pre-pregnancy BMI, maternal education, maternal age at delivery, gestational age at sample collection, ever breastfed
status at 15 mo°
Cluett et al.
(2019)
Medium
United States,
1999-2010
Cross-sectional
Children from
Project Viva,
Ages 6-10,
N = 531
Plasma
6.4 (IQR =5.6)
Areal bone
mineral density
(aBMD) z-
score, BMC z-
score
Regression
coefficient per
log2-unit
increase in
PFOS
aBMD z-score
-0.08 (-0.16, -0.01)
No statistically significant
associations or interactions by sex
BMC z-score: No statistically
significant associations
Confounding: Maternal age, education, census tract median household income, individual household income, and child age, sex,
race/ethnicity, year of blood draw, dairy intake, physical activity
Khalil et al.
(2018)
United States
2016
Cross-sectional
Obese children,
ages 8-12
Serum
BMD measured
as broadband
Regression
coefficient per
BMD (broadband ultrasound
attenuation)
D-228
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels"
(ng/mL)
Outcome Comparison
Resultsb
Low
N = 23
2.79 ultrasound unit increase in -1.03 (-5.35,3.29)
(IQR = 2.10) attenuation PFOS
(dB/MHz) and BMD (speed of sound)
speed of sound -5.22 (-11.2, 0.79)
(m/s), stiffness
index (%) Stiffness index
-2.15 (-5.56, 1.26)
Confounding: Age, sex, race
Di Nisio et al.
(2019)
Low
Italy
2017-2018
Cross-sectional
Male high
school students
N = 100 (50
controls, 50
exposed)
Serum
Controls: 0.82
(0.4-1.3)
Exposed: 1.11
(0.8-1.3)
Semen
Controls: 0.11
(0.08-0.13)
Exposed: 0.11
(0.01-0.14)
Arm span (cm) Mann-Whitney Arm span
test (Exposed Controls: 182.75 (178.0, 185.8)
vs. Controls) Exposed: 179.00 (174.2, 187.0)
Adjusted p-value for comparison of
medians = 0.738
Results: Values for each outcome are reported as median (25th, 75th percentile).
Confounding: None reported
General Population
Uhl et al. (2013) United States,
Medium 2003-2008
Cross-sectional
Adults from
NHANES,
Ages 20-84,
N= 3,809,
Females
N = 1,921
Serum
Adults:
Weighted
mean = 21.23
Females:
Weighted
mean = 18.17
Osteoarthritis OR per ln-unit Adults 20-84
increase in 1.15 (0.94, 1.40)
PFOS orby Q2: 1.04 (0.58, 1.85)
quartiles Q3: 1.99 (1.14, 3.49),
p-value < 0.05
Q4: 1.77 (1.05, 2.96),
p-value < 0.05
Females 20-49
2.37 (1.35, 4.16), p-value < 0.01
Q2: 0.65 (0.19,2.20)
Q3: 1.11 (0.29,4.30)
D-229
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels"
(ng/mL)
Outcome Comparison
Resultsb
Q4: 4.99 (1.61, 15.4),
p-value <0.01
No other statistically significant
associations
Results: Lowest quartile used as the reference group.
Confounding: Age, race/ethnicity, SES, smoking, BMI, vigorous reactional activity, prior wrist, hip, or spine fracture
Lin et al. (2014) United States,
Medium 2005-2006,
2007-2008
Cross-sectional
Adults from
NHANES
Ages > 20,
Males
N = 1,192,
Females
N = 842,
Females in
menopause
N = 305
Serum
GM= 15.32
(SD = 17.58)
Total BMD Regression
(g/cm2) in hip or coefficient per
lumbar spine; ln-unit increase
fractures in hip, in PFOS
wrist, spine, or
all types
Total BMD in lumbar spine
Women not in menopause: -0.022
(-0.038, -0.007), p-value = 0.006
Other outcomes: No statistically
significant associations
Confounding: Age, race/ethnicity, BMI, smoking, drinking, treatment for osteoporosis, use of prednisone or Cortisol daily
Khalil et al.
(2016)
Medium
United States,
2009-2010
Cross-sectional
Adolescents and
adults from
NHANES, Ages
12-80,
Males N = 956,
Females
N = 958
Serum
Mean= 12.7
(SE= 1.20)
BMD (g/cm2) of BMD:
total femur, Regression
femoral neck, coefficient per
lumbar spine; ln-unit increase
Osteoporosis in PFOS and by
among females quartiles
Osteoporosis:
OR per ln-unit
increase in
PFOS and by
quartiles
Total femur
Females: -0.018 (-0.034, -0.002),
p-value < 0.05
Q2: -0.007 (-0.038, 0.023)
Q3:-0.009 (-0.037,0.019)
Q4: -0.044 (-0.074, -0.014),
p-value < 0.05
Males: Not statistically significant
Femoral neck
Females: -0.016 (-0.029, -0.002),
p-value < 0.05
Q2
Q3
Q4
0.001 (-0.019, 0.019)
-0.001 (-0.025, 0.025)
-0.034 (-0.059, -0.009),
p-value < 0.05
D-230
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels"
(ng/mL)
Outcome Comparison
Resultsb
Males: -0.013 (-0.024, -0.002),
p-value < 0.05
Q2
Q3
Q4
-0.036 (-0.077, 0.006)
-0.027 (-0.063, 0.009)
-0.046 (-0.078, -0.015),
p-value < 0.05
Lumbar spine, osteoporosis: No
statistically significant associations
Results: Lowest quartile used as the reference group.
Confounding: Age, ethnicity, BMI, serum cotinine, physical activity, milk consumption, blood lead concentration
Hu et al. (2019) United States,
Medium 2004-2007
Cohort and
cross-sectional
Adults from the
POUNDS Lost
study,
Ages 30-70,
N = 294
Plasma
Mean= 32.2
(16.8-43.1)
BMD and 2-yr
ABMD (g/cm2)
of spine, total
hip, femoral
neck, hip
trochanter, hip
intertrochanteric
area, and
Ward's triangle
area
Regression
coefficient per
SD increase in
PFOS
Spine BMD analyses
Cross-sectional: -0.02 (-0.037,
-0.003)
Total hip BMD analyses
2-yr ABMD: -0.005
(-0.009, -0.001), p-value < 0.05
Hip intertrochanteric area BMD
analyses
2-yr ABMD: -0.008 (-0.013,
-0.003), p-value < 0.05
Femoral neck, hip trochanter,
Ward's triangle area: no
statistically significant associations
No statistically significant
associations or interactions by sex
Confounding: For cross-sectional, age, sex, race, alcohol consumption, physical activity, BMI, dietary intervention group; For cohort, age,
sex, race, alcohol consumption, physical activity, BMI, dietary intervention group, baseline BMD, 2-yr weight change
Notes: aBMD = areal bone mineral density; ALSPAC = Avon Longitudinal Study of Parents and Children; BMC = bone mineral content; BMD = bone mineral density;
BMI = body mass index; GM = geometric mean; IQR = interquartile range; mo = months; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio;
POUNDS Lost = Prevention of Obesity Using Novel Dietary Strategies Lost clinical trial; Q1 = quartile 1; Q2 = quartile 2; Q3 = quartile 3; Q4 = quartile 4; SD = standard
deviation; SE = standard error; SES = socioeconomic status; vs. = versus; yr = year.
D-231
-------
APRIL 2024
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D.13 Gastrointestinal
Table D-22. Associations Between PFOS Exposure and Gastrointestinal Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Population, Exposure
Study Design Ages, Matrix, Levels Outcome Comparison
N (ng/mL)a
Resultsb
Timmermanet Guinea-Bissau Cohort Children aged Serum Diarrhea OR per
al. (2020) 2012-2015 <2 yr previously 0.77 (0.53-1.02) doubling of
Medium enrolled in a PFOS at
RCT for inclusion or 9-
measles mo visit
vaccination
N = 236 (113
girls, 123 boys)
At inclusion: 1.14(0.66, 1.96)
At 9 mo: 1.2(0.62,2.31)
No statistically significant
associations or interactions by sex
Confounding: Weight and age at inclusion, sex, maternal education, breastfeeding without solids0
Dalsager et al. Denmark Cohort Pregnant Serum Diarrhea, Incidence rate
(2016) 2010-2015 women and 8.07 (Range: vomiting ratio (number
Low their children 2.36-25.10) (number of days of days) or OR
from the Odense with symptom (proportion
Child Cohort, or proportion of days) by
Ages 1-4 yr of days tertiles of PFOS
N = 346 under/above exposure
median)
Diarrhea
Number of days with symptom
T2: 1.41 (0.79,2.51)
T3: 1.19(0.67,2.12)
Proportion of days under/above
median
T2: 0.89 (0.51, 1.56)
T3: 1.04 (0.59, 1.82)
Vomiting
Number of days with symptom
T2: 1.18(0.8, 1.74)
T3: 0.87 (0.58, 1.31)
Proportion of days under/above
median
T2: 1.47 (0.86, 2.54)
T3: 0.78 (0.45, 1.35)
Results: Lowest tertile used as reference.
Confounding: Maternal age, maternal educational level, parity, and child age
D-232
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Hammer et al.
(2019)
Low
Faroe Islands
Enrollment:
1986-2009;
follow-up until
2017
Cohort
Children and
adults from
CHEF
N = 2,843
Blood
Low exposure:
GM = 2.33
(1.93-2.90)
High exposure:
GM = 26.88
(21.90-32.24)
Inflammatory
bowel disease
Incidence rate
ratio for highest
vs. lowest tertile
of PFOS
exposure
0.30 (0.08, 1.07)
Confounding: Age, calendar period
Xu et al.
(2020d)
Low
Sweden
2014-2016
Cohort
Residents of
Ronneby
municipality
Ronneby panel
study: N = 57
Ronneby
resampling:
N = 113
Karlshamn:
N = 19
Serum
Ronneby panel
study: 216
(118-300)
Ronneby
resampling: 271
(147-449)
Karlshamn: 5
(4-7)
Inflammatory
bowel disease
(ln-ng/mL levels
of calprotectin PFOS
or zonulin)
Regression Calprotectin
coefficient per Panel study: -0.0008 (-0.0033,
unit increase in 0.0018)
Resampling: -0.0006 (-0.0016,
0.0005)
Karlshamn: -0.045 (-0.14, 0.05)
Zonulin
Panel study: 0.0007 (-0.0012,
0.0025)
Resampling: -0.0001 (-0.0008,
0.0005)
Karlshamn: -0.019 (-0.1, 0.063)
Confounding: Age, BMI, gender
Notes: BMI = body mass index; CHEF = Children's Health and the Environment in the Faroes; GM = geometric mean; mo = month(s); OR = odds ratio; PFOS = perfluorooctane
sulfonate; RR = risk ratio; RCT = randomized controlled trial; T2 = tertile 2; T3 = tertile 3; vs. = versus; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results reported as effect estimate (95% confidence interval) unless otherwise specified.
c Confounding indicates factors the models presented adjusted for.
D-233
-------
APRIL 2024
D.14 Dental
Table D-23. Associations Between PFOS Exposure and Dental Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Population,
Study Design Ages,
N
Exposure
Matrix, Levels Outcome
(ng/mL)a
Comparison
Resultsb
Puttige Ramesh
etal. (2019)
Medium
United States
1999-2002
Cross-sectional Adolescents
from NHANES
aged 12-19 yr
N = 2,869
Serum Dental caries
Median =13
(7.2-22)
OR per log2-
unit increase in
PFOS and by
quartiles
0.99 (0.92, 1.07)
Q2: 0.91 (0.72, 1.16)
Q3: 1.02 (0.81, 1.31)
Q4: 0.92 (0.72, 1.17)
Results: Lowest quartile used as reference
Confounding: Gender, race, education level of parent/guardian, family-poverty-to-income ratio, blood lead level, serum cotinine level0
Wiener and
Waters (2019)
Medium
United States
2013-2014
Cross-sectional Children from
NHANES aged
3-1 lyr
N = 629
Serum Dental caries
GM = 3.88 experience
(95% CI: 3.53,
4.27)
OR per IQR
increase in
PFOS
1.41 (0.97, 2.05); p-value = 0.069
Confounding: Age, sex, race/ethnicity, ratio of family-income-to-poverty guidelines, tooth brushing frequency, dental visit, percentages of
sugar in the diet, fluoride in the water
Notes: CI = confidence interval; IQR = interquartile range; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; Q2 = quartile 2; Q3 = quartile 3;
Q4 = quartile 4; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results are reported as effect estimate (95% confidence interval).
c Confounding indicates factors the models presented adjusted for.
D.15 Ocular
Table D-24. Associations Between PFOS Exposure and Ocular Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Design Population,
M-J 1- A -m T
" Ages, N
Exposure Matrix,
Levels (ng/mL)a
Outcome
Comparison
Resultsb
Zeeshan et al.
China,
Cross-sectional Adults from the
Serum
Visual impairment,
OR per ln-unit
Visual impairment
(2020)
2016
Isomers of C8
Median = 24.07
synechia, macula
increase in PFOS
3.11 (2.3,4.2);
Medium
Health Project,
(14.13-36.41)
disorder, corneal
p-value < 0.05
ages 22-96 yr,
pannus, shallow
N = 1,202
anterior chamber,
Eye disease, combined
vitreous disorder,
< 65 yr: 1.52(1.21, 1.91);
retinal disorder,
p-value < 0.05
D-234
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Design
Population, Exposure Matrix,
Ages, N Levels (ng/mL)a
Outcome
Comparison
Resultsb
lens opacity,
conjunctival
disorder, combined
eye disease
> 65 yr: 0.91 (0.55, 1.51)
All other outcomes: No
statistically significant
associations
Confounding: Age, sex, BMI, education, income, career, exercise time, drinking, smoking0
Notes: BMI = body mass index; OR = odds ratio; yr = years.
a Exposure levels reported as median (25th-75th percentile) unless otherwise specified.
b Results are reported as effect estimate (95% confidence interval).
c Confounding indicates factors the models presented adjusted for.
D.16 Dermal
Table D-25. Associations Between PFOS Exposure and Dermal Health Effects in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Study Design
Population,
Ages,
N
Exposure
Matrix, Levels
(ng/mL)a
Outcome
Comparison
Resultsb
Ernst et al.
Denmark
Cohort
Pregnant
Maternal blood
Acne, age at
Regression
Girls: -1.73 (-5.24, 1.77)
(2019)
1999-2017
women and
(1st trimester)
occurrence
coefficient per
T2: 0.09 (-4.69, 4.87)
Medium
their children
Girls Sample 1:
(months)
log2-unit
T3: -1.96 (-6.89,2.97)
from the
32.3 (19.3-50.8)
increase in
Puberty Cohort
Girls Sample 2:
PFOS, and by
Boys:-1.52 (-4.52, 1.48)
within the
27.9 (16.5-42.2)
tertiles
T2:-1.33 (-5.02, 2.36)
DNBC
Boys Sample 1:
T3:-0.7 (-4.75, 3.35)
N = 555 girls,
31.9(19.2-51.2)
565 boys
Boys Sample 2:
27.2 (16.7-45.2)
Results: Lowest tertile used as a reference group.
Confounding: Highest social class of parents, maternal age at menarche, maternal age at delivery, parity, pre-pregnancy BMI, daily number
of cigarettes smoked in first trimester0
Notes: DNBC = Danish National Birth Cohort; T2 = tertile 2; T3 = tertile 3.
a Exposure levels reported as median (10th-90th percentile).
b Results reported as effect estimate (95% confidence interval).
c Confounding indicates factors the models presented adjusted for.
D-235
-------
APRIL 2024
D.17 Cancer
Table D-26. Associations Between PFOS Exposure and Cancer in Recent Epidemiologic Studies
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Grice et al. (2007)
Medium
United States Cohort Employees of a
1997-1998 PFOS-based
chemical and film
manufacturing
plant, 2,083
Modeled
Non-exposed:
GM
range = 0.11—
0.29 ppm;
Low-exposed:
GM
range = 0.39-
0.89 ppm;
High-exposed:
GM
range = 1.30—
1.97 ppm
Cancers: colon,
melanoma, and
prostate
OR by PFOS
exposure category
Colon cancer:
Ever exposed: 1.21 (0.51, 2.87)
Low or high-exposed: 1.37
(0.57, 3.30)
High-exposed: 1.69 (0.68, 4.17)
Melanoma:
Ever exposed: 1.08 (0.31, 3.72)
Low or high-exposed: 0.90
(0.24, 3.43)
High-exposed: 1.01 (0.25, 4.11)
Prostate cancer:
Ever exposed: 1.34 (0.62, 2.91)
Low or high-exposed: 1.36
(0.61, 3.02)
High-exposed: 1.08 (0.44, 2.69)
Results: Non-exposed used as the reference group
Confounding: Age and gender
Eriksen et al.
Denmark
Cohort
Adults with no
Serum
Cancers: IRR per unit increase Prostate cancer:
(2009)
1993-2006
previous cancer
Mean (5th-95th prostate, bladder, in PFOS, or by
Q2: 1.35 (0.97, 1.87)
Medium
diagnosis,
percentile):
pancreatic, liver quartiles
Q3: 1.31 (0.94, 1.82)
Ages 50-65 at
Cases, men:
Q4: 1.38 (0.99, 1.93)
enrollment,
35.1 (17.4-
Per unit increase: 1.05
(0.97,
Prostate cancer,
60.9);
1.14)
1,393;
Controls, men:
Bladder cancer,
35.0 (16.8-
Bladder cancer:
1,104;
62.4);
Q2: 0.76 (0.50, 1.16)
Pancreatic cancer,
Cases, women:
Q3: 0.93 (0.61, 1.41)
900;
32.1 (14.0-
Q4: 0.70 (0.46, 1.07)
Liver cancer, 839
58.1);
Per unit increase: 0.93
(0.83,
1.03)
D-236
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Controls,
women: 29.3
(14.2-55.6)
Pancreatic cancer:
Q2: 1.02 (0.57, 1.84)
Q3: 1.24 (0.67,2.31)
Q4: 0.91 (0.51, 1.65)
Per unit increase: 0.99 (0.86,
1.14)
Liver cancer:
Q2: 0.62 (0.29, 1.33)
Q3: 0.72 (0.33, 1.56)
Q4: 0.59 (0.27, 1.27)
Per unit increase: 0.97 (0.79,
L19)
Results: Lowest quartile used as the reference group
Confounding: Prostate cancer: years of school attendance, BMI, dietary fat intake, and vegetable intake; Bladder cancer: smoking status,
smoking intensity, smoking duration, years of school attendance, occupation associated with risk for bladder cancer; Pancreatic cancer:
smoking status, smoking intensity, smoking duration, dietary fat intake, and fruit and vegetable intake; Liver cancer: smoking status, years of
school attendance, alcohol intake, and occupation associated with risk for liver cancer
Bonefeld-
Greenland Case-control Greenlandic Inuit Plasma Breast cancer
OR per ln-unit
1.030(1.001, 1.070),
Jorgensen et al.
2000-2003 women with and Cases: 45.6
increase in PFOS
p-value = 0.05
(2011)
without breast (Range = 11.6—
Medium
cancer, 76 124)
Controls: 21.9
(Range =1.5-
172)
Confounding: Age, BMI, pregnancy, cotinine, breastfeeding, and menopausal status
Ghisari et al.
Greenland Case-control Women of Serum Breast cancer
OR (for high serum
CYP1A1; Ilc/Val + Val/Val:
(2014)
2000-2003 Greenland Inuit
PFOS vs low)
12.1 (1.29, 115); p = 0.029
Medium
descent aged 18-80
years. Cases were
CYP IB 1; Leu/Leu:
diagnosed with
11.2 (1.8, 71.1); p = 0.011
breast cancer, 100
COMT; Val/Met + Met/Met
16.8(1.68, 167); p = 0.016
D-237
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
CYP17; A1/A2 + A2/A2:
18.2 (1.67, 198.8); p = 0.017
CYP19 CT; CC:
9.6(1.48, 62.4); p = 0.018
CYP19 TTTA; (TTTA)8-10:
29.3 (2.89, 298); p = 0.004
Confounding: Age and cotinine.
Ducatman et al.
(2015)
Medium
United States Cross-
Men from C8
Serum
Prostate-specific Regression
Age 20-49
2005-2006
sectional Health Study,
Mean (SD): antigen (PSA) coefficient ((3) per In- (3=1, p-value = 0.71;
Ages 20-49, 9,169; 22.18 (1.97)
Ages 50-69, 3,819
level
unit increase in PFOS GMR = 0.95 (0.71, 1.28)
GM ratio (GMR)
(PSA < 4.0 ng/mL vs. Age 50-69
PSA > 4.0 ng/mL) (3=1, p-value = 0.99;
GMR = 1.1 (0.98, 1.23)
Confounding: Age, smoking status, average alcohol intake, and BMP
Ghisari et al. Denmark Nested case- Adult women, 283 Serum
(2017) 1996-2002 control Cases: 27.80
Medium Controls: 28.77
Breast cancer
Cohort: 1.15 (0.64,2.08)
CYP19 CC: 6.42 (1.08, 38.3),
p-value < 0.05
Relative risk ratio
(RR) per ln-unit
increase in PFOS,
compared across
genotypes:
CYP1A1 (Ile462Val), No significant associations
CYP IB 1 observed for remaining
(Leu432Val), COMT genotypes
(Vall58Met), CYP 17
(-34T > C), CYP19
(C>T)
Results: Lowest tertile used as the reference group
Confounding: Age at blood draw, BMI before pregnancy, total number of gravidities, oral contraceptives use, age of menarche, smoking
status and alcohol intake during pregnancy, physical activity, maternal education.
D-238
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Hurley et al. (2018) California, U.S. Nested case- Adult women,
Medium 2011-2015 control 1,760
Mancini et al.
(2020)
Medium
Serum
Median (min-
max): Cases:
6.695 (0.046-
39.400)
Controls: 6.950
(0.046-99.800)
Breast cancer
(invasive)
OR per loglO-unit
increase inPFOS, or
by tertiles
T3: 0.898 (0.695, 1.161)
T2: 0.883 (0.691, 1.129)
Per unit increase: 0.934 (0.683,
1.277), p-value = 0.67
Confounding: Age at baseline enrollment, race/ethnicity, region of residence, date of blood draw, season of blood draw, total smoking pack-
years, BMI, family history of breast cancer, age at first full-term pregnancy, menopausal status at blood draw, and pork consumption
Cohn et al. (2020)
Medium
United States Nested case- Adult daughters of Perinatal serum Breast cancer
1959-2013 control women in CHDS Cases: 30.5
cohort, 310 (14.1-55.8)
controls, 102 cases Controls: 32.1
(14.9-58.2)
OR per log2-unit
increase in PFOS
0.3 (0.1,0.9), p-value = 0.02
Confounding: Maternal: cholesterol, age at pregnancy, history of breast cancer, primiparity, overweight at first prenatal visit, serum levels
of DDTs and metabolite DDE, African American status, whether daughter was breastfed
France
1990-2013
Nested case-
control
Postmenopausal
women,
Ages 40-65 in
1990, 194 cases,
194 controls
Serum Breast cancer ORs by quartiles, and Overall:
17.51 (5.83- by estrogen (ER) or Q2: 1.94 (1, 3.78)
85.26) progesterone receptor Q3: 2.03 (1.02, 4.04)
(PR) status Q4: 1.72 (0.88, 3.36)
p-trend = 0.25
ER positive:
ORs of 1.8-2.4
p-trend = 0.04
ER negative:
ORs of 4.7-15
p-trend = 0.72
PR positive:
ORs of 1.8-2.7
p-trend = 0.02
PR negative: ORs of 1.7-3.5
p-trend = 0.93
Results: Lowest quartile used as the reference group
D-239
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Fry and Power
(2017)
Medium
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Confounding: Total serum lipids, BMI, smoking status, physical activity, education level, personal history of benign breast disease, family
history of breast cancer, parity/age at first full-term pregnancy, total breastfeeding duration, age at menarche, age at menopause, use of oral
contraceptives, current use of menopausal hormone therapy
Shearer et al. United States Nested case- Adults, 55-74, 648 Serum
(2021) 1993-2002 control Ages 55-59, 190 38.4 (26.3-
Medium Ages 60-65, 224 49.9)
Ages 65+, 234
Males 432
Females 216
Renal cell
carcinoma
ORs per log2-unit
increase in PFOS or
by quartiles (total
cohort only)
Q2: 1.67 (0.84, 3.3)
Q3: 0.92 (0.45, 1.88)
Q4: 2.51 (1.28, 4.92)
p-trend = 0.009
Per doubling increase: 1.39
(1.04, 1.86)
Results: Lowest quartile used as the reference group
Confounding: BMI, smoking, history of hypertension, estimated glomerular filtration rate, previous freeze-thaw cycle, calendar year of
blood draw; sex, race and ethnicity, study year of blood draw, study center
U.S. NHANES Cohort
2003-2006
Adults,
Ages 60+, 1,036
Serum
Median (SE):
4.3 (0.2) ng/g
lipid
Cancer mortality Hazard ratio per SD- 1.01 (0.86, 1.19),
unit increase in PFOS p-value = 0.88
Confounding: Age, gender, race/ethnicity, and smoking status
Goodrich et al.
(2022)
Medium
United States Nested case- Adults, 100 (50 Plasma
Hepatocellular OR for >54.9 (ig/L 4.50(1.20,16.00),
MEC Study
Recruitment:
1993-1996
control
cases, 50 controls)
GM (GSD):
Cases: 29.2
(2.37)
Controls: 29.2
(1-95)
carcinoma
vs. < 54.9 (ig/L-
PFOS, or per SD
increase in PFOS
p-value = 0.02
Per SD increase:
1.20 (0.91, 1.60),
p-value = 0.18
Results: PFOS cutoff of 54.9 |ig/L is the 85th percentile of PFOS in the study, and corresponds to the 90th percentile of PFOS exposures in
the 1999-2000 NHANES
Confounding: Age, sex, race, and study site
Christensen et al.
(2016a)
Low
Wisconsin,
U.S., 2012-
2013
Cross-
sectional
Male anglers,
Ages 50+, 154
Serum
19.00 (9.80-
28.00)
Cancer (any) OR per unit increase 0.99(0.96,1.01)
in PFOS
Confounding: Age, BMI, work status, alcohol consumption
Lin et al. (2020a)
Low
China
2014-2017
Case-control Children, <16, 84 Serum
Germ cell tumors OR per unit increase 1.08 (0.96, 1.21)
in PFOS
D-240
-------
APRIL 2024
Reference,
Confidence
Location, Population, Ages, Exposure
v Design Matrix, Levels Outcome Comparison
YearS N (ng/m L)1'
Select Resultsb
4.47 (2.48-
8.26)
Confounding: Infectious disease, cosmetics usage, barbecued food consumption, filtered water use, indoor decorating, living near farmland
(maternal behaviors/factors during pregnancy)
Tsai et al. (2020)
Low
Taiwan Case-control Adult women, 239 Plasma Breast cancer ORperln-unit
2014-2016 Age 50 oryounger, Mean (GM): increase inPFOS
120 5.64 (4.77)
Age over 50, 119
Total cohort: 1.07 (0.64, 1.79)
Age 50 oryounger: 2.34 (1.02,
5.38), p-value < 0.05
ER+: 3.25 (1.29, 8.23)
Age over 50: 0.62 (0.29, 1.29),
p-value > 0.05
Confounding: Pregnancy history, oral contraception use, abortion, BMI, menopause, and education level
Itoh et al. (2021)
Low
Japan Case-control Adult women, Serum Breast cancer OR by quartiles
2001-2005 Ages 20-74, 14.27(10.24-
802 (401 breast 19.24)
cancer cases, 401
controls)
Q2: 0.38 (0.18, 0.82),
p-value < 0.05
Q3: 0.31 (0.14,0.69),
p-value < 0.05
Q4: 0.15 (0.06, 0.39),
p-value < 0.05
p-trend = 0.0001
Results: Lowest quartile used as the reference group
Confounding: Age, residential area, BMI, height, menopausal status, age at menopause, age at first childbirth, family history of breast
cancer, smoking status, strenuous physical activity in the past five years, moderate physical activity in the past five years, age at menarche,
number of births, breastfeeding duration, alcohol intake, isoflavone intake, education level, serum total concentrations of PCBs, fish and
shellfish intake, vegetable intake, and calendar year of blood sampling
Liu et al. (2021)
Low
China Case-control Adult men, 96 Serum Thyroid cancer OR by quartiles
2016-2017 Adult women, 223 Case: 5.5 (3.6-
8.8); Control:
7.5 (4.7-10.8)
Total
Q2: 0.81 (0.42, 1.53)
Q3: 0.26 (0.12,0.57)
Q4: 0.28 (0.12,0.66)
p-trend = 0.001
Male:
Q2: 1.13 (0.30,4.23)
Q3: 0.15 (0.02, 1.04)
D-241
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Q4: 0.62 (0.15,2.65)
p-trend = 0.284
Female:
Q2: 1.10 (0.52,2.34)
Q3: 0.33 (0.13, 0.80)
Q4: 0.24 (0.09, 0.64)
p-trend = 0.001
Results: Lowest quartile used as the reference group
Confounding: Age, sex, and diabetes status
Omoike et al.
(2021)
Low
United States
2005-2012
Cross-
sectional
Adults from
NHANES,
Ages > 20 yr,
6,652
Serum
11.40 (6.45-
19.68)
Cancers: ovarian, OR per unit increase
breast, uterine,
and prostate
in PFOS, or by
quartiles
Ovarian cancer:
Q2: 0.08 (0.08, 0.084)
Q3: 1.64(1.62, 1.66)
Q4: 2.25 (2.22, 2.28)
p-trend <0.001
Per unit increase: 1.012 (1.012,
1.013)
Breast cancer:
Q2: 0.87 (0.86, 0.89)
Q3: 1.06 (1.05, 1.06)
Q4: 1.47 (1.46, 1.48)
p-trend <0.001
Per unit increase:
1.011 (1.011, 1.011)
Uterine cancer:
Per unit increase:
0.945 (0.944, 0.945)
Prostate cancer:
Per unit increase:
0.994 (0.994, 0.994)
Results: Lowest quartile used as the reference group
Confounding: Age, sex, education, race/ethnicity, PIR, BMI, and serum cotinine
D-242
-------
APRIL 2024
Reference,
Confidence
Location,
Years
Population, Ages, Exposure
Design Matrix, Levels
^ (ng/mL)a
Outcome
Comparison
Select Resultsb
Cao et al. (2022) China
Low 2019-2021
Case-control Adults and Serum
children, Cases: 7.2 (3.8-
Ages 12-84 yr, 406 15)
(203 cases, 203 Controls: 5.5
controls) (3.0-11)
Liver cancer
OR per log-ng/mL
increase in PFOS
2.609(1.179,4.029)
p-trend = 0.001
Results: Logarithm base not specified
Confounding: Age, education level, BMI, annual household income, sex, smoking habit, and medical history
Notes: BMI = body mass index; CHDS = The Child Health and Development Studies; DDE = dichlorodiphenyldichloroethylene; DDT = dichloro-diphenyl-trichloroethane;
ER = progesterone receptor; GM = geometric mean; GMR = geometric mean ratio; GSD = geometric standard deviation; IRR = incidence rate ratio; MEC = Multiethnic Cohort
study; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; PR = progesterone receptor; PSA = Prostate-specific antigen; Q2 = quartile 2;
Q3 = quartile 3; Q4 = quartile 4; RR = risk ratio; SD = standard deviation; T2 = tertile 2; T3 = tertile 3; U.S. = United States; yr = years.
a Exposure levels reported as median (25th-75th percentile) in ng/mL unless otherwise noted.
b Results reported as effect estimate (95% confidence interval).
c Confounding indicates factors the models presented adjusted for.
D-243
-------
APRIL 2024
Appendix E. Benchmark Dose Modeling
E.l Epidemiology Studies
For the epidemiological studies considered for dose-response assessment, the U.S.
Environmental Protection Agency (EPA) used multiple modeling approaches to determine points
of departure (PODs), depending upon the health outcome and the data provided in the studies.
For the developmental, hepatic, and serum lipid dose-response studies, EPA used a hybrid
modeling approach that involves estimating the prevalence of the outcome above or below a
level considered to be adverse and determining the probability of responses at specified exposure
levels above the control (U.S. EPA, 2012) because EPA was able to define a level considered
clinically adverse for these outcomes. Details are provided in the following sections. In addition,
EPA re-expressed the reported regression (P) coefficients when modeling results for decreased
birthweight when regression coefficients were reported per log-transformed units of exposure
(see details in Section E.l.2). Sensitivity analyses to evaluate the potential impact of re-
expression in a hybrid approach when modeling hepatic and serum lipid studies for
perfluorooctane sulfonic acid (PFOS) showed little impact on benchmark doses (lower
confidence limit) (BMDLs) (see Sections E.l.3 and E.1.4).
EPA also performed benchmark dose (BMD) modeling and provided study lowest-observed-
adverse-effect levels/no-observed-adverse-effect levels (LOAELs/NOAELs) for the hepatic and
serum lipid dose-response studies as sensitivity analyses of the hybrid approach. For the immune
studies, where a clinically defined adverse level is not well defined, EPA used the results from
the multivariate models provided in the studies and determined a benchmark response (BMR)
according to EPA guidance to calculate BMDs and BMDLs (U.S. EPA, 2012) (see Section
E. 1.1). For specific approaches used to determine PODs please see Table E-l.
Table E-l. Summary of Modeling Approaches for POD Derivation from Epidemiological
Studies
„ , . , .. „ Reported Result or LosP Re-Expression . , BMR (SD
Endpoint Studies ,, , _T ® , Approach „ ,v
F Beta (Units) FOS (Yes/No) 11 Cutoff)
Anti-tetanus and
Budtz-
BMD = log2(l~BMR) Yes
No
BMD modeling
0.5 SD and
anti-diphtheria
Jargensen
/P
BMD = log2(l~
1 SD
antibody
and
log2(tetanus or
BMR)/p
response
Grandjean
diphtheria) per ng/mL
(2018a)
PFOS
Anti-tetanus and
Timmerman et Percent difference
No
No
BMD modeling
0.5 SD and
anti-diphtheria
al. (2021)
=(10P—1)*100
BMD = logio(l
1 SD
antibody
logio(tetanus or
-BMR)/p
response
diphtheria) per ng/mL
PFOS
Anti-rubella
Granum et al.
Rubella (IU/mL) per
No
No
BMD modeling
0.5 SD and
antibody
(2013)
ng/mL PFOS
1SD
response
Anti-rubella
Zhang et al.
Percent
Yes
No; sensitivity
BMD modeling
0.5 SD and
antibody
(2023c)
difference = (2.7 IP—1)
analysis with re-
1 SD
response
*100
expressed
ln(rubella) per
values
ln(ng/mL) PFOS
E-l
-------
APRIL 2024
Endpoint
Studies3
Reported Result or
Beta (Units)
LogP
FOS
Re-Expression
(Yes/No)
Approach
BMR (SD or
Cutoff)
Decreased birth
weight
Wikstrom et
al. (2020)
Chu et al.
(2020)
Sagiv et al.
(2018)
Starling et al.
(2017)
Darrow et al.
(2013)
Yao et al.
(2021)
BW per ln(ng/mL)
PFOS or per IQR
PFOS
Yes
Yes
Hybrid
5% and 10%
Elevated ALT
Nian et al.
(2019)
Percent
difference = (e13-!)*!
00
ln(ALT) per
ln(ng/mL) PFOS
Yes
No; sensitivity
analysis with re-
expressed
values
Hybrid
5% and 10%
Elevated ALT
Gallo et al.
(2012)
ln(ALT) per
ln(ng/mL) PFOS
Yes
No
Hybrid
5% and 10%
Increased total
cholesterol
Dong et al.
(2019)
TC per ng/mL PFOS
No
No
Hybrid
5% and 10%
Increased total
cholesterol
Steenland et
al. (2009)
ln(TC) per ln(ng/mL)
PFOS
Yes
No; sensitivity
analysis with re-
expressed
values
Hybrid
Sensitivity
analyses:
LOAEL,
BMDS
5% and 10%
Increased total
cholesterol
Lin et al.
(2019)
mean difference in TC No
(mg/dL) per quartile
of PFOS (ng/mL)
No
BMDS
0.5 SD and
1 SD
Notes: ALT = alanine transaminase; BMD = benchmark dose; BMDS = Benchmark Dose Software; BMR = benchmark
response; BW = birth weight; IQR = interquartile range; IU = international units; LOAEL = lowest-observed-adverse-effect
levels; POD = point of departure; SD = standard deviation; TC = total cholesterol.
a Bolded study name identifies study result that advanced as the POD or selected approach.
E.l.l Modeling Results for Immunotoxicity
E. 1.1.1 Modeling Results for Decreased Tetanus Antibody
Concentrations
E.l.l.1.1 Budtz-J0rgensen and Grandjean (2018a) Results for Decreased Tetanus
Antibody Concentrations at 7 Years of Age and PFOS Exposure Measured at
5 Years of Age
Budtz-j0rgensen and Grandjean (2018a) fit multivariate models of PFOS measured at age
5 years, against log2-transformed anti-tetanus antibody concentrations measured at the 7-year-old
examination controlling for sex, exact age at the 7-year-old examination, and booster type at age
5 years. Models were evaluated with additional control for perfluorooctanoic acid (PFOA) (as
log2(PFOA)) (also called multi-PFAS models), and without PFOA (also called single-PFAS
models). Three model shapes were evaluated by Budtz-j0rgensen and Grandjean (2018a) using
E-2
-------
APRIL 2024
likelihood ratio tests: a linear model, a piecewise-linear model with a knot at the median PFOS
concentration, and a logarithmic function. The logarithmic functions did not fit better than the
piecewise-linear functions (Budtz-j0rgensen and Grandjean, 2018a). The piecewise-linear model
did not fit better than the linear model for the PFOS exposure without adjustment for PFOA
using a likelihood ratio test (p = 0.60; see Budtz-j0rgensen and Grandjean (2018a) Table 3), or
for the model that did adjust for PFOA (log2(PFOA)) (p = 0.71).
Table E-2 summarizes the results from Budtz-j0rgensen and Grandjean (2018a) for PFOS at age
5 years and tetanus antibodies at age 7 years. These regression coefficients (P) and their standard
errors (SE) were calculated by EPA from the authors Budtz-j0rgensen and Grandjean (2022,
2018a). As Budtz-j0rgensen and Grandjean (2018a) log2-transformed the outcome variable, the
BMR measured in unit of log2(tetanus antibody concentration) was
log2(l-0.05) = 0.074 log2(IU/mL)).
Table E-2. Results Specific to the Slope from the Linear Analyses of PFOS Measured at
Age 5 Years and Log2(Tetanus Antibody Concentrations) Measured at Age 7 Years from
Table 1 in Budtz-Jorgensen and Grandjean (2018a) in a Single-PFAS Model3 and in a
Multi-PFAS Modelb
Exposure
M ,,c. PFOA
Model Shape ... , ,
Adjusted
Slope (P) per
ng/mL
SE(P)
ng/mL
Slope (P) Fit
Lower Bound
Slope (Plb)
ng/mL
PFOS at Age 5
Linear Noa
-0.0274
0.0176
p = 0.12
-0.0565
PFOS at Age 5
Linear Yesb
-0.0039
0.0198
p = 0.84
-0.0365
Notes: SE = standard error.
a Single-PFAS model: adjusted for a single PFAS (i.e., PFOS), and sex, exact age at the 7-year-old examination, and booster type
at age 5 years.
b Multi-PFAS model: adjusted for PFOS and PFOA, and sex, exact age at the 7-year-old examination, and booster type at age
5 years.
Interpretation of results in Table E-2:
• PFOS is a non-significant predictor in the single-PFAS model (P = -0.0274; p = 0.12).
• Effects of PFOS in the single-PFAS model are attenuated when log2(PFOA) is included in
the model (P = -0.0039; p = 0.84).
• Nevertheless, these data can be used to estimate a BMDL for completeness and to allow
comparisons across PFAS.
E.1.1.1.1.1 Selection of the Benchmark Response
The BMD approach involves dose-response modeling to obtain BMDs, i.e., dose levels
corresponding to specific response levels near the low end of the observable range of the data
and the BMDLs to serve as potential PODs for deriving quantitative estimates below the range of
observation (U.S. EPA, 2012). Selecting a BMR to estimate the BMDs and BMDLs involves
making judgments about the statistical and biological characteristics of the dataset and about the
applications for which the resulting BMDs and BMDLs will be used. An extra risk of 10% is
recommended as a standard reporting level for quantal data for toxicological data. Biological
considerations may warrant the use of a BMR of 5% or lower for some types of effects as the
basis of the POD for a reference value. However, a BMR of 1% has typically been used for
quantal human data from epidemiology studies (U.S. EPA, 2012), although this is more typically
E-3
-------
APRIL 2024
used for epidemiologic studies of cancer mortality within large cohorts of workers, which can
support the statistical estimation of small BMRs.
In the 2021 Proposed Approaches draft (U.S. EPA, 2021c) reviewed by the SAB PFAS Review
Panel, EPA relied on the BMDL modeling approach published in Budtz-j0rgensen and
Grandjean (2018a), described above. During validation of the modeling, EPA reevaluated the
approach chosen by Budtz-j0rgensen and Grandjean (2018a) and determined that a different
approach should be used to be consistent with EPA guidance (U.S. EPA, 2012), which
recommends the use of a 1 or V2 SD change in cases where there is no accepted definition of an
adverse level of change or clinical cutoff for the health outcome.
A blood concentration for tetanus antibodies of 0.1 IU/mL is sometimes cited in the tetanus
literature as a 'protective level' and (Grandjean et al., 2017b) noted that the Danish vaccine
producer Statens Serum Institut recommended the 0.1 IU/mL "cutoff level "to determine
whether antibody concentrations could be considered protective," and Galazka and
Kardymowicz (1989) mentions the same concentration. However, the 2018 WHO update (WHO,
2018) argues that:
"...the minimum amount of circulating antitoxin that in most cases ensures immunity
to tetanus is assay specific. Within in vivo neutralization tests; modified ELISAsor
bead-based immunofluorescence assays¦, concentrations at or exceeding 0.01 IU/mL
are usually considered protective against disease, whereas antitoxin concentrations of
at least 0.1-0.2 IU/mL are defined as positive when ELISA techniques are used for the
assessment. Cases of tetanus have been documentedhowever¦, in persons with
antitoxin concentrations above these thresholds. Hence, a "protective antibody
concentration" may not be considered a guarantee of immunity under all
circumstances."
In the absence of a clear definition of an adverse effect for a continuous endpoint like antibody
concentrations, a default BMR of 1 or V2 SD change from the control mean may be selected (U.S.
EPA, 2012). As noted above, a lower BMR can also be used if it can be justified on a biological
and/or statistical basis. Figure E-l replicates a figure in the Technical Guidance (page 23) (U.S.
EPA, 2012) to show that in a control population where 1.4% are considered to be at risk of
having an adverse effect, a downward shift in the control mean of 1 SD results in a -10% extra
risk of being at risk of having an adverse effect.
E-4
-------
APRIL 2024
Standard deviation units
Figure E-l. Difference in Population Tail Probabilities Resulting from a One Standard
Deviation Shift in the Mean from a Standard Normal Distribution, Illustrating the
Theoretical Basis for a Baseline BMR of 1 SD
BMR = benchmark response; SD = standard deviation.
Statistically, the Technical Guidance additionally suggests that studies of developmental effects
can support lower BMRs. Consistent with EPA's Benchmark Dose Technical Guidance (U.S.
EPA, 2012), EPA typically selects a 5% or 0.5 standard deviation (SD) benchmark response
(BMR) when performing dose-response modeling of data from an endpoint resulting from
developmental exposure. Because Budtz-Jorgensen and Grandjean (2018a) assessed antibody
response after PFAS exposure during childhood, this is considered a developmental study (U.S.
EPA, 1991) based on EPA's Guidelines for Developmental Toxicity Risk Assessment, which
states that a developmental effect "may result from exposure prior to conception (either parent),
during prenatal development, or postnatally to the time of sexual maturation" and can be
"detected at any point in the lifespan of the organism."
Biologically, a BMR of V2 SD is a reasonable choice as anti-tetanus antibody concentrations
prevent against tetanus, which is a rare, but severe and sometimes fatal infection, with a case-
fatality rate in the United States of 13% during 2001-2008 (CDC, 2011). The case-fatality rate
can be more than 80% for early lifestage cases (Patel and Mehta, 1999). Selgrade (2007)
E-5
-------
APRIL 2024
suggests that specific immunotoxic effects observed in children may be broadly indicative of
developmental immunosuppression impacting these children's ability to protect against a range
of immune hazards—which has the potential to be a more adverse effect that just a single
immunotoxic effect. Thus, decrements in the ability to maintain effective levels of tetanus
antitoxins following immunization may be indicative of wider immunosuppression in these
children exposed to PFOS. By contrast, a BMR of 1 SD may be more appropriate for an effect
that would be considered 'minimally adverse.' A BMR smaller than V2 SD is generally selected
for severe effects (e.g., 1% extra risk of cancer mortality); decreased antibody concentrations
offer diminished protection from severe effects but are not themselves severe effects.
Following the technical guidance (U.S. EPA, 2012), EPA derived BMDs and BMDLs associated
with both a 1 SD change in the distribution of log2(tetanus antibody concentrations) and V2 SD
change in the distribution of log2(tetanus antibody concentrations) (Table E-3). The SD of the
log2(tetanus antibody concentrations) at age 7 years was estimated from the distributional data
presented in Grandjean et al. (2012) as follows: the 25th and 75th percentiles of the tetanus
antibody concentrations at age 7 years in IU/mL were (0.65, 4.6). Log2-tranforming these values
provides the 25th and 75th percentiles in log2(IU/mL) as (-0.62, 2.20). Assuming that these log2-
transformed values are reasonably represented by a normal distribution, the IQR (which is the
difference between the 75th and 25th percentiles) is approximately 1.35 SDs (Rosner, 2015).
Thus, SD = IQR/1.35, and the SD of tetanus antibodies in log2(IU/mL) is (2.20-
(-0.62))/1.35 = 2.09 log2(IU/mL).
While there was not a clear definition of the size of an adverse effect for a continuous endpoint
like antibody concentrations, the value of 0.1 IU/mL is sometimes cited. As a check, EPA
evaluated how much extra risk would have been associated with a BMR set at a cutoff value of
0.1 IU/mL. Using the observed distribution of tetanus antibodies at age 7 years in log2(IU/mL),
EPA calculated that 2.8% of those values would be below the cutoff value of 0.1 IU/mL (i.e.,
-3.32 log2(IU/mL)). A BMR of V2 SD resulted in 7.9% of the values being below that cutoff,
which is 5.1%) extra risk. This demonstrates the generic guidance that a BMR of V2 SD can
provide a reasonably good estimate of 5% extra risk. Figure E-2 shows an example of this.
E-6
-------
APRIL 2024
0.25-
Control distribution
Distribution for group with mean response 1/2 SD below control mean
LWY1 5.1% excess risk below control distribution percentile
0.20-
1/2 SD shift in mean [1.05 Log2(IU/ml)]
0.15
si
CD
-Q
O
0.10-
0.05-
0.00
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Tetanus antibody concentrations in Log2(IU/ml)
Figure E-2. Difference in Population Tail Probabilities Resulting from a Vz Standard
Deviation Shift in the Mean from an Estimation of the Distribution of Log2(Tetanus
Antibody Concentrations at Age 7 Years)
IU = international units; SD =standard deviation.
Table E-3. BMDs and BMDLs for Effect of PFOS at Age 5 Years on Anti-Tetanus
Antibody Concentrations at age 7 Years (Budtz-Jorgensen and Grandjean, 2018a) Using a
BMR of Vz SD Change in Log2(Tetanus Antibodies Concentration) and a BMR of 1 SD
Change in Log2(Tetanus Antibodies Concentration)
Estimated Without Control of PFOA
Estimated With Control of PFOA
BMR BMD (ng/mL) BMDL (ng/mL) BMD (ng/mL) BMDL (ng/mL)
P = -0.274 per ng/mL 0lb = -0.0565 per ng/mL p = -0.0039 per ng/mL 0lb = -0.0365 per ng/mL
!/> SD 38.1 18.5a 268 28.6
1 SD 762 310 536 573
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD = standard deviation.
a Denotes the selected POD.
The lowest serum PFOS concentration measured at age 5 years was 3.3 ng/mL, the 5th percentile
was 9.5 ng/mL, and the 10th percentile was 10.7 ng/mL (Grandjean and Bateson, 2021) so the
E-7
-------
APRIL 2024
estimated BMDL for a BMR of V2 SD (BMDL', sd = 18.5 ng/mL) in the single-PFAS model is
well within the observed range. No information was available to judge the fit of the model in the
range of the BMDLs, but the BMD and BMDL were both within the range of observed values.
The BMDy2 sd estimate from the multi-PFAS models is 7-fold higher than the BMD1, sd estimate
from the models with just PFOS, and the BMDL sd estimates is 55% higher. The change in
BMD estimates may, or may not, reflect control for any potential confounding of the regression
effect estimates. While it is not clear which PFAS model provided the 'better' estimate of the
point estimate of the effect of PFOS considering potential confounding, the two BMDL', sd
estimates are 55% different (18.5 ng/mL vs. 28.6 ng/mL). EPA advanced the derivation based on
results that did not control for PFOA because this model appeared to fit PFOS better (p = 0.12
vs. 0.84) and there was moderate uncertainty due to potential confounding in the BMDL.
However, confidence was diminished by the non-significant fit for PFOS (p = 0.12) and stronger
potential confounding in the main effect—even though there was moderate confounding of the
BMDL.
For immunotoxicity related to tetanus associated with PFOS exposure measured at age
5 years, the POD is based on a BMR of V2 SD and a BMDL^ sd of 18.5 ng/mL.
£ 1.1.1.2 Budtz-J0rgensen and Grandjean (2018a) Results for Decreased Tetanus
Antibody Concentrations at 5 Years of Age and PFOS Exposure Measured
Perinatally
Budtz-j0rgensen and Grandjean (2018a) fit multivariate models of PFOS measured perinatally in
maternal serum, against log2-transformed anti-tetanus antibody concentrations measured at the 5-
year-old examination controlling for sex, and exact age at the 5-year-old examination, cohort,
and interaction terms between cohort and sex, and between cohort and age. Models were
evaluated with additional control for PFOA (as log2(PFOA)), and without PFOA. Three model
shapes of PFOS were evaluated by Budtz-j0rgensen and Grandjean (2018a) using likelihood
ratio tests: a linear model, a piecewise-linear model with a knot at the median, and a logarithmic
function. The logarithmic functions did not fit better than the piecewise-linear functions Budtz-
J0rgensen and Grandjean (2018a). Compared with the linear model, the piecewise-linear model
did not fit better than the linear model for either the PFOS exposure without adjustment for
PFOA using a likelihood ratio test (p = 0.43; see Budtz-j0rgensen and Grandjean (2018a)
Table 3), or for the model that did adjust for PFOA (log2(PFOA)) (p = 0.98).
Table E-4 summarizes the results from Budtz-j0rgensen and Grandjean (2018a) for tetanus in
this exposure window. These regression coefficients (P) and their standard errors (SE) were
obtained by EPA from the authors (Budtz-j0rgensen and Grandjean, 2022, 2018a).
Table E-4. Results of the Linear Analyses of PFOS Measured Perinatally and Tetanus
Antibodies Measured at Age 5 Years from Budtz-Jorgensen and Grandjean (2018b) in a
Single-PFAS Model and in a Multi-PFAS Model
Exposure
Model Shape
PFOA
Adjusted
Slope (P) per
ng/mL
SE(P)
ng/mL
Slope (P) Fit
Lower Bound
Slope (Plb)
ng/mL
Perinatal PFOS
Perinatal PFOS
Linear
Linear
Noa
Yesb
-0.0102
0.0021
0.0095
0.0107
p = 0.28
p = 0.85
-0.0259
-0.0156
E-8
-------
APRIL 2024
Notes: SE = standard error.
a Single-PFAS model: adjusted for a single PFAS (i.e., PFOS), and sex, exact age at the 5-year-old examination, cohort,
interaction terms between cohort and sex, and between cohort and age.
b Multi-PFAS model: adjusted for PFOS and PFOA, and sex, exact age at the 7-year-old examination cohort, and interaction
terms between cohort and sex, and between cohort and age.
Interpretation of results in Table E-4:
• PFOS is a non-significant predictor in the single-PFAS model (P = -0.0102; p = 0.28).
• Effects are attenuated when log2(PFOA) are included in the model (P = 0.0021; p = 0.85).
• Nevertheless, these data can be used to estimate a BMDL for completeness and to allow
comparisons across PFAS.
E. 1.1.1.2.1 Selection of the Benchmark Response
In the 2021 Proposed Approaches draft (U.S. EPA, 2021c) reviewed by the SAB PFAS Review
Panel, EPA relied on the BMDL modeling approach published in Budtz-j0rgensen and
Grandjean (2018a), described above. During validation of the modeling, EPA reevaluated the
approach chosen by Budtz-j0rgensen and Grandjean (2018a) and determined that a different
approach should be used to be consistent with EPA guidance (U.S. EPA, 2012), which
recommends the use of a 1 or V2 SD change in cases where there is no accepted definition of an
adverse level of change or clinical cutoff for the health outcome. Additionally, consistent with
EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012), EPA typically selects a 5% or
0.5 SD benchmark response (BMR) when performing dose-response modeling of data from an
endpoint resulting from developmental exposure. Because Budtz-j0rgensen and Grandjean
(2018a) assessed antibody response after PFAS exposure during childhood, this is considered a
developmental study (U.S. EPA, 1991) based on EPA's Guidelines for Developmental Toxicity
Risk Assessment, which states that a developmental effect "may result from exposure prior to
conception (either parent), during prenatal development, or postnatally to the time of sexual
maturation" and can be "detected at any point in the lifespan of the organism."
Following the technical guidance (U.S. EPA, 2012), EPA derived BMDs and BMDLs associated
with a 1 SD change in the distribution of log2(tetanus antibody concentrations) and V2 SD change
in the distribution of log2(tetanus antibody concentrations). The SD of the log2(tetanus antibody
concentrations) at age 5 years was estimated from two sets of distributional data presented from
two different cohorts of 5-year-olds that were pooled in Budtz-j0rgensen and Grandjean (2018a).
Grandjean et al. (2012) reported on 587 five-year-olds from the cohort of children born during
1997-2000 and Grandjean et al. (2017b) reported on 349 five-year-olds from the cohort of
children born during 2007-2009. The means and SDs were computed separately by the authors.
EPA then pooled the summary statistics to describe the common SD. The 25th and 75th
percentiles of the tetanus antibody concentrations in the earlier birth cohort at age 5 years in
IU/mL were (0.10, 0.51). Log2-tranforming these values provides the 25th and 75th percentiles
in log2(IU/mL) as (-3.32, -0.97). Assuming that these log2-transformed values are similar to the
normal distribution, the IQR is approximately 1.35 SDs, thus SD = IQR/1.35, and the SD of
tetanus antibodies in log2(IU/mL) is (-0.97-(-3.32))/1.35 = 1.74 log2(IU/mL).
The 25th and 75th percentiles of the tetanus antibody concentrations in the later birth cohort at
age 5 years in IU/mL was (0.1, 0.3). Log2-tranforming these values provides the 25th and 75th
percentiles in log2(IU/mL) as (-3.32, -1.74), and the SD of tetanus antibodies in log2(IU/mL) is
(—1.74—(—3.32))/l .35 = 1.17 log2(IU/mL). The pooled variance is a weighted sum of the
E-9
-------
APRIL 2024
independent SDs, and the pooled SD was estimated as 1.55 log2(IU/mL).8 To show the impact of
the BMR on these results, Table E-5 presents the BMDs and BMDLs at BMRs of V2 SD and
1 SD.
Table E-5. BMDs and BMDLs for Effect of PFOS Measured Perinatally and Anti-Tetanus
Antibody Concentrations at Age 5 Years (Budtz-Jorgensen and Grandjean, 2018a)
Estimated without control of PFOA
Estimated with control of PFOA
BMR
BMD (ng/mL)
BMDL (ng/mL)
BMD (ng/mL)
BMDL (ng/mL)
P = -0.0102 per ng/mL
Plb = -0.0259 per ng/mL
P = 0.00207 per ng/mL
Plb = -0.0156 per ng/mL
ViSD
75.9
29.9a
_b
49.8
1 SD
151.8
59.8
-
99.7
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD =standard deviation.
a Denotes the POD that corresponds to the analyses of PFOS concentrations perinatally and tetanus antibodies at age 5 years.
b Values cannot be determined.
The lowest perinatal maternal serum PFOS concentration measured was 9.4 ng/mL, the 5th
percentile was 17.1 ng/mL, and the 10th percentile was 19.1 ng/mL (Grandjean and Bateson,
2021) so the estimated BMDLs for a BMR of V2 SD (BMDL~_ sd = 29.9 ng/mL) in the single-
PFAS model is well within the observed range. No information was available to judge the fit of
the model in the range of the BMDLs, but the BMD and BMDL were both within the range of
observed values. The BMDL sd estimate from the single-PFAS models was 29.9 ng/mL. The
BMDL estimates from the multi-PFAS models were about 67% higher than for the single-PFAS
model.
Confidence is diminished by the low quality of the model fit for PFOS in either model compared
with the PFOS results from tetanus in the 5-year to 7-year exposure-outcome window of time
and there is some uncertainty regarding potential confounding.
For immunotoxicity related to tetanus, associated with PFOS measured perinatally, the POD is
based on a BMR of V2 SD and a BMDL sd of 29.9 ng/mL. Note that this result is based on a
poorly fit PFOS regression parameter (P) estimated as -0.0102 per ng/mL (90% CI: -0.0259,
0.0055; p = 0.28) (Budtz-j0rgensen and Grandjean, 2018b).
For immunotoxicity related to tetanus associated with PFOS exposure measured at age
5 years, the POD estimated for comparison purposes were based on a BMR of Vz SD and a
BMDLi/2 sd of 29.9 ng/mL.
E.l.1.1.3 Timmerman etol. (2021)
Timmerman et al. (2021) analyzed data from Greenlandic children ages 7-12 and fit multivariate
models of PFOS and loglO-transformed anti-tetanus antibody concentrations measured at the
same time as PFOS, controlling for time since vaccine booster/estimated time since vaccine
booster, and duration of being breastfed (<6 months, 6-12 months, >1 year) and area of
residence (Nuuk, Maniitsoq, Sisimiut, Ilulissat, Aasiaat, Qeqertarsuaq, Tasiilaq) and including
children with known tetanus-diphtheria booster date only. Estimates from the linear regression
8 Pooled variance for tetanus in five-year-olds = [(502-l)(1.74)A2+(298-l)(1.17)A2]/[502+298-2] = 2.41. The pooled SD is the
square root of 2.41 which is 1.55 log2(IU/mL).
E-10
-------
APRIL 2024
models were subsequently back-transformed to express the percent difference in antibody
concentrations at each ng/mL increase in serum PFOS concentrations in children, which was -3
(95% CI: -8, 3) (Table 4, Timmerman et al. (2021)). Using the equation provided below, EPA
estimated the regression slope as -0.013 (95% CI: -0.036, 0.013).
Percent Difference = (10^ — 1) x 100
Following the approach described previously for Budtz-j0rgensen and Grandjean (2018a), EPA
derived BMDs and BMDLs were derived for both a one SD change in the distribution of logio
(tetanus antibody concentrations) as a standard reporting level, and V2 SD change in the
distribution of logio (tetanus antibody concentrations) (Table E-6). The SD of the logio (tetanus
antibody concentrations) was estimated from the median (25th, 75th percentiles) of 0.92 (0.25,
2.20) tetanus antibody concentrations in IU/mL (Table 1 in Timmerman et al. (2021)). Logio -
transforming these values results in 25th and 75th percentiles in logio (IU/mL) as -0.60 and 0.34,
respectively. Assuming that these logio -transformed values are reasonably represented by a
normal distribution, the IQR is approximately 1.35 SDs. Thus, SD = IQR/1.35, and the SD of
tetanus antibodies in logio (IU/mL) is (0.34 - (—0.60))/l .35 = 0.70 logio (IU/mL).
Table E-6. BMDs and BMDLs for Effect of PFOS on Anti-Tetanus Antibody
Concentrations (Timmermann et al., 2021) Using a BMR of Vz SD Change in Logio(Tetanus
Antibodies Concentration) and a BMR of 1 SD Change in Logio(Tetanus Antibodies
Concentration)
BMR
BMD (ng/mL)
BMDL (ng/mL)
P = -0.013 per ng/mL
P = -0.036 per ng/mL
V2SD
26.4
9.66
1 SD
52.9
19.3
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD =standard deviation.
As a check, EPA evaluated how much extra risk would have been associated with a BMR set at a
cutoff value of 0.1 IU/mL. Using the observed distribution of tetanus antibodies in logio (IU/mL),
a BMR of V2 SD resulted in 10.6% extra risk. This suggests that, in this case, a BMR of V2 SD
may not be a reasonably good estimate of 5% extra risk.
Note that this BMDL is based on a poorly fit PFOS regression parameter (P) estimated as -0.013
(95% CI: -0.036, 0.013) (Timmermann et al., 2021).
For immunotoxicity related to tetanus associated with PFOS exposure measured at ages 5
to 10 years old, the POD estimated for comparison purposes was based on a BMR of Vz SD
and a BMDLh sd of 9.7 ng/mL.
E. 1.1.1.4 Summary of Modeling Results for Decreased Tetanus Antibody
Concentrations
Table E-7 summarizes the PODs resulting from the modeling approaches for decreased tetanus
antibody concentrations. The selected and comparison PODs were based on a BMR of V2 SD,
resulting in BMDLs ranging from 9.7 to 29.9, with the selected POD of 18.5 also representing
the median of the BMDLs.
E-ll
-------
APRIL 2024
Table E-7. BMDLs for Effect of PFOS on Anti-Tetanus Antibody Concentrations Using a
BMR of V2 SD (Timmermann et al., 2021)
Study
Effect
BMDLy, sd (ng/mL)
'/:SD
Budtz-Jorgensen
PFOS at age 5 years and anti-tetanus antibody
18.5
1.05 log2 (IU/mL)
and Grandjean
concentrations at age 7 years
(2018a)
Budtz-Jorgensen
PFOS perinatally and anti-tetanus antibody
29.9
0.78 log2 (IU/mL)
and Grandjean
concentrations at age 7 years
(2018a)
Timmerman et al.
PFOS and anti-tetanus antibody concentrations at
9.66
0.35 logio (IU/mL)
(2021)
ages 7-12 years
Notes: BMDL = benchmark dose lower limit; BMR = benchmark response; IU = international units; SD =standard deviation.
E.l.1.2 Modeling Results for Decreased Diphtheria Antibody
Concentrations
E. 1.1.2.1 Budtz-J0rgensen and G ret adject n (2018a) Results for Decreased
Diphtheria Antibody Concentrations at 7 Years of Age and PFOS Exposure
Measured at 5 Years of Age
Budtz-j0rgensen and Grandjean (2018a) fit multivariate models of PFOS measured at age
5 years, against log2-transformed anti-diphtheria antibody concentrations measured at the 7-year-
old examination controlling for sex, exact age at the 7-year-old examination, and booster type at
age 5 years. Models were evaluated with additional control for PFOA (as log2(PFOA)), and
without PFOA. Three model shapes were evaluated by Budtz-j0rgensen and Grandjean (2018a)
using likelihood ratio tests: a linear model of PFOS, a piecewise-linear model with a knot at the
median, and a logarithmic function. The logarithmic functions did not fit better than the
piecewise-linear functions (Budtz-j0rgensen and Grandjean, 2018a). The piecewise-linear model
did not fit better than the linear model for the PFOS exposure without adjustment for PFOA
using a likelihood ratio test (p = 0.30; see Budtz-j0rgensen and Grandjean (2018a) Table 3), or
for the model that did adjust for PFOA (log2(PFOA)) (p = 0.34). Table E-8 summarizes the
results from Budtz-j0rgensen and Grandjean (2018a) for diphtheria in this exposure window.
These P and their SE were obtained by EPA from study authors (Budtz-j0rgensen and
Grandjean, 2022, 2018a).
Table E-8. Results Specific to the Slope from the Linear Analyses of PFOS Measured at
Age 5 Years and Log2(Diphtheria Antibodies) Measured at Age 7 Years from Table 1 in
Budtz-Jorgensen and Grandjean (2018a) in a Single-PFAS Model3 and in a Multi-PFAS
Modelb
Exposure
M ,,c. PFOA
Model Shape ... , ,
Adjusted
Slope (P) per
ng/mL
SE(P)
ng/mL
Slope (P) Fit
Lower Bound
Slope (Plb)
ng/mL
PFOS at Age 5
Linear Noa
-0.0322
0.0163
p = 0.05
-0.0591
PFOS at Age 5
Linear Yesb
-0.0207
0.0184
p = 0.26
-0.0510
Notes: SE = standard error.
a Single-PFAS model: adjusted for a single PFAS (i.e., PFOS), and sex, exact age at the 7-year-old examination, and booster type
at age 5 years.
E-12
-------
APRIL 2024
b Multi-PFAS model: adjusted for PFOS and PFOA, and sex, exact age at the 7-year-old examination, and booster type at age
5 years.
Interpretation of results in Table E-8:
• PFOS is a significant predictor in the single-PFAS model (P = -0.0322; p = 0.05).
• Effects are attenuated when log2(PFOA) are included in the model (P = -0.0207;
p = 0.26).
• The point estimate results for PFOS are potentially confounded by PFOA since there was
a 36% reduction in the effect size for PFOS from -0.0322 to -0.0207 when controlling for
PFOA.
• One explanation is that PFOA was a confounder of the PFOS effect.
• Another possibility is physiological confounding, which can arise when biomarkers
measured from the same blood test are more highly correlated due to individual's
physiological processes. Physiological confounding can therefore induce confounding bias
by the inclusion of co-measured co-exposures in regression models.
• The reasons for the change in main effect size are not known and remain an uncertainty
because it is not known whether the change in estimate was induced by physiologic
confounding or was the result of controlling for classical confounding. For this reason,
there is uncertainty in knowing which estimate is the best representation of any effect of
PFOS.
• The uncertainty from potential confounding does not have much impact on the RfD,
which is defined as allowing for an order of magnitude (10-fold or 1,000%) uncertainty in
the estimate. This is because there is only 36% difference in the BMD and 16% difference
in the BMDL when PFOS is included in the model.
E. 1.1.2.1.1 Selection of the Benchmark Response
In the 2021 Proposed Approaches draft (U.S. EPA, 2021c) reviewed by the SAB PFAS Review
Panel, EPA relied on the BMDL modeling approach published in Budtz-j0rgensen and
Grandjean (2018a), described above. During validation of the modeling, EPA reevaluated the
approach chosen by Budtz-j0rgensen and Grandjean (2018a) and determined that a different
approach should be used to be consistent with EPA guidance (U.S. EPA, 2012), which
recommends the use of a 1 or V2 SD change in cases where there is no accepted definition of an
adverse level of change or clinical cutoff for the health outcome. Additionally, consistent with
EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012), EPA typically selects a 5% or
0.5 SD benchmark response (BMR) when performing dose-response modeling of data from an
endpoint resulting from developmental exposure. Because Budtz-j0rgensen and Grandjean
(2018a) assessed antibody response after PFAS exposure during childhood, this is considered a
developmental study (U.S. EPA, 1991) based on EPA's Guidelines for Developmental Toxicity
Risk Assessment, which state that a developmental effect "may result from exposure prior to
conception (either parent), during prenatal development, or postnatally to the time of sexual
maturation" and can be "detected at any point in the lifespan of the organism."
Following the technical guidance (U.S. EPA, 2012), EPA derived BMDs and BMDLs associated
with a 1 SD change in the distribution of log2(diphtheria antibody concentrations), and V2 SD
change in the distribution of log2(diphtheria antibody concentrations). A blood concentration for
diphtheria antibodies of 0.1 IU/mL is sometimes cited in the diphtheria literature as a 'protective
E-13
-------
APRIL 2024
level' (Grandjean et al. (2017b) noted that the Danish vaccine producer Statens Serum Institut
recommended the 0.1 IU/mL 'cutoff level; and Galazka et al. (1993) mentions the same
concentration, but also argues:
"However¦, it has also been shown that there is no sharply defined level of antitoxin that
gives complete protection from diphtheria (Ipsen, 1946). A certain range of variation
must be accepted; the same degree of antitoxin may give an unequal degree of
protection in different persons. Other factors may influence the vulnerability to
diphtheria including the dose and virulence of the diphtheria bacilli and the general
immune status of the person infected (Christenson and Bottiger, 1986). Thus, an
antibody concentration between 0.01 and 0.09 IU/mL may be regarded as giving basic
immunity, whereas a higher titer may be needed for full protection. In some studies
that used in vitro techniques, a level of 0.1 IU/mL was considered protective (Cellesi et
al, 1989; Galazka and Kardymowicz, 1989)."
Statistically, the Technical Guidance suggests that studies of developmental effects can support
lower BMRs. Biologically, a BMR of V2 SD is a reasonable choice as anti-diphtheria antibody
concentrations prevent against diphtheria, which is very rare in the United States, but can cause
life-threatening airway obstruction, or cardiac failure (Collier, 1975). Among 13 cases reported
in the United States during 1996-2016, no deaths were mentioned (Liang et al., 2018). However,
diphtheria remains a potentially fatal disease in other parts of the world (Galazka (1993)
mentions a case-fatality rate of 5%—10%) and PFOS-related changes in anti-diphtheria antibody
concentrations cannot be considered 'minimally adverse' given the historic lethality of diphtheria
in the absence of vaccination. Selgrade (2007) suggests that specific immunotoxic effects
observed in children may be broadly indicative of developmental immunosuppression impacting
these children's ability to protect against a range of immune hazards—which has the potential to
be a more adverse effect that just a single immunotoxic effect.
Following the technical guidance (U.S. EPA, 2012), EPA derived BMDs and BMDLs associated
with a 1 SD change in the distribution of log2(diphtheria antibody concentrations) as a standard
reporting level, and V2 SD change in the distribution of log2(diphtheria antibody concentrations).
The SD of the log2(diphtheria antibody concentrations) at age 7 years was estimated from the
distributional data presented in Grandjean et al. (2012) as follows: the 25th and 75th percentiles
of the diphtheria antibody concentrations at age 7 years in IU/mL were (0.4, 1.6). Log2-
tranforming these values provides the 25th and 75th percentiles in log2(IU/mL) as (-1.32, 0.68).
Assuming that these log2-transformed values are similar to the normal distribution, the IQR is
approximately 1.35 SDs, thus SD = IQR/1.35, and the SD of tetanus antibodies in log2(IU/mL) is
(0.68—(—1,32))/l .35 = 1.48 log2(IU/mL). To show the impact of the BMR on these results, Table
E-9 presents the BMDs and BMDLs at BMRs of V2 SD and 1 SD.
Table E-9. BMDs and BMDLs for Effect of PFOS at Age 5 Years on Anti-Diphtheria
Antibody Concentrations at Age 7 Years (Budtz-Jorgensen and Grandjean, 2018a) Using a
E-14
-------
APRIL 2024
BMR of V2 SD Change in Log2(Diphtheria Antibodies Concentration) and a BMR of 1 SD
Log2(Diphtheria Antibodies Concentration)
Estimated Without Control of PFOA
Estimated With Control of PFOA
BMR
BMD (ng/mL) BMDL (ng/mL)
P = -0.0322 per ng/mL 0lb = -0.0592 per ng/mL
BMD (ng/mL) BMDL (ng/mL)
P = -0.0207 per ng/mL 0lb = -0.0510 per ng/mL
V2SD
1 SD
23.0 12.5a
46.0 25.0
35.8 14.5
71.7 29.0
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD =standard deviation.
a Denotes the selected POD.
The lowest serum PFOS concentration measured at age 5 years was 3.3 ng/mL, the 5th percentile
was 9.5 ng/mL, and the 10th percentile was 10.7 ng/mL (Grandjean and Bateson, 2021) so the
estimated BMDL for a BMR of V2 SD (BMDL1, sd = 12.5 ng/mL) in the single-PFAS model is
well within the observed range. No information was available to judge the fit of the model in the
range of the BMDLs, but the BMD and BMDL were both within the range of observed values
and the model fit PFOS well (p = 0.05).
The BMD1, sd estimate from the multi-PFAS models is 56% higher than the BMD1, sd estimate
from the model with just PFOS, and the BMDL sois 16% higher. This may, or may not, reflect
control for any potential confounding of the regression effect estimates. While it is not clear
which PFAS model provided the 'better' estimate of the point estimate of the effect of PFOS in
light of potential confounding, the two BMDL sd estimates that serve as the PODs are
comparable (12.5 ng/mL vs. 14.5 ng/mL). EPA advanced POD based on results that did not
controls for PFOA because this model appeared to fit PFOS data better (p = 0.05 vs. 0.26) and
there was low uncertainty due to potential confounding in the BMDL. However, confidence was
diminished by the potential confounding in the main effect—even though there was low
confounding of the BMDL.
For immunotoxicity related to diphtheria, associated with PFOS measured at age 5 years,
the POD is based on a BMR of V2 SD and a BMDL^ sd of 12.5 ng/mL.
£ 1.1.2.2 Budtz-J0rgensen and Grandjean (2018a) Results for Decreased
Diphtheria Antibody Concentrations at 5 Years of Age and PFOS Exposure
Measured Perinatally
Budtz-j0rgensen and Grandjean (2018a) fit multivariate models of PFOS measured perinatally,
against log2-transformed anti-diphtheria antibody concentrations measured at the 5-year-old
examination controlling for sex and age. Models were evaluated with additional control for
PFOA (as log2(PFOA)), and without PFOA. Three model shapes were evaluated by Budtz-
J0rgensen and Grandjean (2018a) using likelihood ratio tests: a linear model of PFOS, a
piecewise-linear model with a knot at the median, and a logarithmic function. The logarithmic
functions did not fit better than the piecewise-linear functions Budtz-j0rgensen and Grandjean
(2018a). Compared with the linear model, the piecewise-linear model did not fit better than the
linear model for either the PFOS exposure without adjustment for PFOA using a likelihood ratio
test (p = 0.55; see Budtz-j0rgensen and Grandjean (2018a) Table 3), or for the model that did
adjust for PFOA (log2(PFOA)) (p = 0.84). Table E-10 summarizes the results from Budtz-
E-15
-------
APRIL 2024
J0rgensen and Grandjean (2018a) for diphtheria in this exposure window. These P and their SE
were obtained by EPA from the study authors (Budtz-j0rgensen and Grandjean, 2022, 2018a).
Table E-10. Results of the Linear Analyses of PFOS Measured Perinatally and Diphtheria
Antibodies Measured at age 5 Years from Budtz-Jorgensen and Grandjean (2018b) in a
Single-PFAS Model3 and in a Multi-PFAS Modelb
Exposure
Model Shape
PFOA
Adjusted
Slope (P) per
ng/mL
SE (P)
Slope (P) Fit
Lower Bound
Slope (Plb)
Perinatal PFOS
Linear
Noa
-0.0310
0.0100
p = 0.002
0.04V 5
Perinatal PFOS
Linear
Yesb
-0.0241
0.0113
p = 0.033
0.042V
Notes: SE = standard error.
a Single-PFAS model: adjusted for a single PFAS (i.e., PFOS), and sex, and exact age at the 5-year-old examination.
b Multi-PFAS model: adjusted for PFOS and PFOA and sex, and exact age at the 5-year-old examination.
Interpretation of results in Table E-10:
• PFOS is a significant predictor in the single-PFAS model (P = -0.0310; p = 0.002).
• Effects of PFOS are attenuated when PFOA is in the model (P = -0.0241; p = 0.033).
• Results for PFOS are potentially confounded by PFOA since there was a 22% change in
the effect size for PFOS from -0.0310 to -0.0241 when controlling for PFOA.
• One explanation is that PFOA was a confounder of the PFOS effect.
• Another possibility is physiological confounding, which can arise when biomarkers
measured from the same blood test are more highly correlated due to individual's
physiological processes. Physiological confounding can therefore induce confounding bias
by the inclusion of co-measured co-exposures in regression models.
• The reasons for the change in main effect size are not known and remain an uncertainty
because it is not known whether the change in estimate was induced by physiologic
confounding or was the result of controlling for classical confounding. For this reason,
there is uncertainty in knowing which estimate is the best representation of any effect of
PFOS.
• The uncertainty from potential confounding does not have much impact on the RfD,
which is defined as allowing for an order of magnitude (10-fold or 1,000%) uncertainty in
the estimate. This is because there is only a 22% difference in the BMD and 11%
difference in the BMDL when PFOS is included in the model.
E.1.1.2.2.1 Selection of the Benchmark Response
In the 2021 Proposed Approaches draft (U.S. EPA, 2021c) reviewed by the SAB PFAS Review
Panel, EPA relied on the BMDL modeling approach published in Budtz-j0rgensen and
Grandjean (2018a), described above. During validation of the modeling, EPA reevaluated the
approach chosen by Budtz-j0rgensen and Grandjean (2018a) and determined that a different
approach should be used to be consistent with EPA guidance (U.S. EPA, 2012), which
recommends the use of a 1 or V2 SD change in cases where there is no accepted definition of an
adverse level of change or clinical cutoff for the health outcome. Additionally, consistent with
EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012), EPA typically selects a 5% or
0.5 SD benchmark response (BMR) when performing dose-response modeling of data from an
endpoint resulting from developmental exposure. Because Budtz-j0rgensen and Grandjean
(2018a) assessed antibody response after PFAS exposure during childhood, this is considered a
E-16
-------
APRIL 2024
developmental study (U.S. EPA, 1991) based on EPA's Guidelines for Developmental Toxicity
Risk Assessment, which states that a developmental effect "may result from exposure prior to
conception (either parent), during prenatal development, or postnatally to the time of sexual
maturation" and can be "detected at any point in the lifespan of the organism."
Following the technical guidance (U.S. EPA, 2012), EPA derived BMDs and BMDLs associated
with a 1 SD change in the distribution of log2(tetanus antibody concentrations) as a standard
reporting level, and V2 SD change in the distribution of log2(tetanus antibody concentrations).
The SD of the log2(diphtheria antibody concentrations) at age 5 years was estimated from two
sets of distributional data presented from two different birth cohorts of 5-year-olds that were
pooled in Budtz-j0rgensen and Grandjean (2018a). Grandjean et al. (2012) reported on 587 5-
year-olds from the cohort of children born during 1997-2000 and Grandjean et al. (2017b)
reported on 349 5-year-olds from the cohort of children born during 2007-2009. The means and
SDs were computed separately by the author. EPA then pooled the summary statistics to describe
the common SD. The 25th and 75th percentiles of the diphtheria antibody concentrations in the
earlier birth cohort at age 5 years in IU/mL were (0.05, 0.4). Log2-tranforming these values
provides the 25th and 75th percentiles in log2(IU/mL) as (-4.32, -1.32). Assuming that these
log2-transformed values are similar to the normal distribution, the IQR is approximately
1.35 SDs, thus SD = IQR/1.35, and the SD of diphtheria antibodies in log2(IU/mL) is
(—1.32—(—4.32))/l .35 = 2.22 log2(IU/mL).
The 25th and 75th percentiles of the diphtheria antibody concentrations in the later birth cohort
at age 5 years in IU/mL were (0.1, 0.3). Log2-tranforming these values provides the 25th and
75th percentiles in log2(IU/mL) as (-3.32, -1.74), and the SD of diphtheria antibodies in
log2(IU/mL) is (—1.74—(—3,32))/l .35 = 1.17 log2(IU/mL). The pooled variance is a weighted sum
of the independent SDs, and the pooled SD was estimated as 1.90 log2(IU/mL).9 To show the
impact of the BMR on these results, Table E-l 1 presents the BMDs and BMDLs at BMRs of
Vi SD and 1 SD.
Table E-ll. BMDs and BMDLs for Effect of PFOS Measured Perinatally and Anti-
Diphtheria Antibody Concentrations at age 5 Years (Budtz-Jorgensen and Grandjean,
2018a)
Estimated Without Control of PFOA
Estimated With Control of PFOA
BMR
BMD (ng/mL) BMDL (ng/mL)
P = -0.031 per ng/mL 0lb = -0.0475 per ng/mL
BMD (ng/mL)
P = -0.0241 per ng/mL
BMDL (ng/mL)
Plb = -0.0427per ng/mL
ViSD
1 SD
30.6 20.0a
61.3 40.0
39.4
78.9
22.3
44.5
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD =standard deviation.
a Denotes the selected POD.
The lowest serum PFOS concentration measured perinatally was 9.4 ng/mL, the 5th percentile
was 17.1 ng/mL, and the 10th percentile was 19.1 ng/mL (Grandjean and Bateson, 2021) so the
estimated BMD for a BMR of V2 SD (BMDL', sd = 20.0 ng/mL) in the single-PFAS model is
well within the observed range. No information was available to judge the fit of the model in the
9 Pooled variance for diphtheria in 5-year-olds = [(502-l)(2.22)A2+(298-l)(1.17)A2]/[502+298-2] = 3.60. The pooled SD is the
square root of 3.60 which is 1.90 log2(IU/mL).
E-17
-------
APRIL 2024
range of the BMDLs, but the BMD and BMDL were both within the range of observed values
and the model fit PFOS well (p = 0.002).
The BMD1, sd estimate from the multi-PFAS models is 29% higher than the BMD1, sd estimated
from the model with just PFOS, and the BMDL sois 12% higher. This may, or may not, reflect
control for any potential confounding of the regression effect estimates. The BMDLs that serve
as the PODs are comparable (20.0 ng/mL vs. 22.3 ng/mL) and EPA advanced the derivation
based on results that did not control for PFOA because this model appeared to fit PFOS well
(p = 0.002 vs. 0.031) and there was low uncertainty due to potential confounding in the BMD
and moderate uncertainty in the BMDL.
For immunotoxicity related to diphtheria, associated with PFOS measured at age 5 years,
the POD is based on a BMR of Vi SD and a BMDLh sd of 20.0 ng/mL.
E.l.1.2.3 Timmerman etol. (2021)
Timmerman et al. (2021) analyzed data from Greenlandic children ages 7-12 and fit multivariate
models of PFOS against loglO-transformed anti-diphtheria antibody concentrations measured at
the same time as PFOS, controlling for time since vaccine booster/estimated time since vaccine
booster, and duration of being breastfed (<6 months, 6-12 months, >1 year) and area of
residence (Nuuk, Maniitsoq, Sisimiut, Ilulissat, Aasiaat, Qeqertarsuaq, Tasiilaq) and including
children with known tetanus-diphtheria booster date only. Estimates from the linear regression
models were subsequently back-transformed to express the percent difference in antibody
concentrations at each ng/mL increase in serum PFOS concentrations in children, which was -9
(95% CI: -16, 2) (Table 4, Timmerman et al. (2021)). Using the equation provided below, EPA
estimated the regression slope as -0.04 (95% CI: -0.08, 0.01).
Percent Difference = (10^ — 1) x 100
Following the description provided for Budtz-j0rgensen and Grandjean (2018a), EPA derived
BMDs and BMDLs for both a one SD change in the distribution of logio (diphtheria antibody
concentrations) as a standard reporting level, and V2 SD change in the distribution of logio
(diphtheria antibody concentrations) (Table E-12). The SD of the logio (diphtheria antibody
concentrations) was estimated from the distributional data presented in Table 1 as follows: the
25th and 75th percentiles of the diphtheria antibody concentrations in IU/mL were 0.02 and 0.28,
respectively. Logio -transforming these values provides the 25th and 75th percentiles in logio
(IU/mL) as (-1.7, -0.55). Assuming that these logio -transformed values are reasonably
represented by a normal distribution, the IQR is approximately 1.35 SDs. Thus, SD = IQR/1.35,
and the SD of tetanus antibodies in logio(IU/mL) is 0.85 logio(IU/mL).
Table E-12. BMDs and BMDLs for Effect of PFOS on Anti-Diphtheria Antibody
Concentrations (Timmermann et al., 2021) Using a BMR of Vz SD Change in Logio(Tetanus
Antibodies Concentration) and a BMR of 1 SD Change in Logio(Tetanus Antibodies
Concentration)
BMR
BMD (ng/mL)
BMDL (ng/mL)
P = -0.11 per ng/mL
P = -0.28 per ng/mL
'/jSD
10.4
5.61
1 SD
20.7
11.2
E-18
-------
APRIL 2024
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD = standard deviation.
As a check, EPA evaluated how much extra risk would have been associated with a BMR set at a
cutoff value of 0.1 IU/mL. Using the observed distribution of diphtheria antibodies in logio
(IU/mL), EPA calculated that 57% of those values would be below the cutoff value of
0.1 IU/mL. A BMR of '/2 SD resulted in 75% of the values being below that cutoff, which is 18%
extra risk. This suggest that in this case a BMR of V2 SD may not be a reasonably good estimate
of 5% extra risk.
Note that this result is based on a poorly fit PFOS regression parameter (P) estimated as -0.04
(95% CI: -0.08, 0.01) (Timmermann et al., 2021).
For immunotoxicity related to tetanus associated with PFOS exposure measured at ages 5
to 10 years old, the POD estimated for comparison purposes were based on a BMR of Vz SD
and a BMDL^ sd of 5.6 ng/mL.
E. 1.1.2.4 Summary of Modeling Results for Decreased Diphtheria Antibody
Concentrations
Table E-13 summarizes the PODs resulting from the modeling approaches for decreased tetanus
antibody concentrations. The selected and comparison PODs were based on a BMR of V2 SD,
resulting in BMDLs ranging from 5.6 ng/mL to 20.0 ng/mL with the selected POD of 12.5 also
representing the median of the BMDLs. The comparison PODs are considered low confidence
because they are based on a poorly fit PFOS regression parameters.
Table E-13. BMDLs for Effect of PFOS on Anti-Diphtheria Antibody Concentrations
Using a BMR of Vz SD (Timmermann et al., 2021)
Study Name
Effect
BMDL (ng/mL)
HSD
Budtz-Jorgcnscn and
Grandjean (2018a)
PFOS at age 5 years on anti-diphtheria
antibody concentrations at age 7 years
12.5
0.74 log2(IU/mL)
Budtz-Jorgcnscn and
Grandjean (2018a)
PFOS perinatally on anti-diphtheria antibody
concentrations at age 7 years
20.0
0.95 log2(IU/mL)
Timmerman et al. (2021)
PFOS and anti-diphtheria antibody
concentrations at ages 7-12 years
5.6
0.48 logio(IU/mL)
Notes: BMDL = benchmark dose lower limit; IU = international units; SD = standard deviation.
E.1.1.3 Modeling Results for Decreased Rubella Antibody
Concentrations
E.1.1.3.1 Granum etal. (2013)
Granum et al. (2013) investigated the association between prenatal exposure to perfluorinated
compounds and vaccination responses and clinical health outcomes in early childhood in the
BraMat subcohort of the Norwegian Mother and Child Cohort Study. A total of 56 mother-child
pairs with maternal blood samples at delivery and blood samples from the children at 3 years of
age were evaluated. Antibody titers specific to rubella were measured in 50 serum samples.
Prenatal exposure to PFOS (mean = 5.6 ng/mL) was inversely associated with rubella antibody
levels at age 3. Granum et al. (2013) fit multivariate linear regression models of maternal PFOS
E-19
-------
APRIL 2024
against antibody concentrations in units of optical density (OD) adjusted for maternal allergy,
paternal allergy, maternal education, child's gender, and/or age at 3-year follow-up. The
estimated regression coefficient and 95% confidence interval was -0.08, 95% CI: -0.14, -0.02
(Table 4, Granum et al., 2013). The summary statistics for rubella antibody levels at the age of 3
in units of OD were median = 1.9; 25th, 75th percentiles: 1.5, 2.1. Study authors were contacted
to provide these summary statistics in units of IU/mL (median = 60.6; 25th, 75th percentiles:
41.8, 80.2), and the corresponding regression coefficient and 95% confidence interval: -5.1, 95%
CI: -9.0,-1.1 (Table E-14).
Table E-14. Levels of Rubella Vaccine-Induced Antibodies at the Age of 3 Years (Adapted
from Table 3 in Granum et al. (2013))
Parameter Optical Density (OD) IU/mLa
25th percentile 1.5 41.8
Median 1.9 60.6
75 th percentile 2.1 80.2
Min-Max 0.8-2.4 15.0-120.0
Mean 1.7 61.6
0.5 SD 0.22 14.3
1 SD 0.44 28.6
(3 (95% CI) for PFOS -0.08 (-0.14, -0.02) -5.1 (-9.0,-1.1)
Notes: IU = international units; OD = optical density; SD = standard deviation.
a Authors were contacted to provide summary statistics for rubella antibody levels in IU/mL (n = 50).
Following the technical guidance (U.S. EPA, 2012) and the approach described previously for
Budtz-j0rgensen and Grandjean (2018, 5083631; see Section E.l.1.1.1) and accounting for the
fact that here the outcome variable is not log-transformed, EPA derived BMDs and BMDLs for
both a 1 and V2 SD change from the control mean in the distribution of rubella antibody
concentrations. However, rubella differs from diphtheria and tetanus in that several levels for
rubella antibody have been cited in the literature as "protective levels," representing a clinically
significant cutoff for an adverse response. These levels vary depending on geography and study,
ranging from 4 IU/mL in Finland (Davidkin et al., 2008), to 11 IU/mL in Iran (Honarvar et al.,
2013), or 15 IU/mL in the United States (Tosh et al., 2009). However, 10 IU/mL appears to be
the most widely accepted standard for rubella immunity. For example, Skenzdel et al. (1996)
noted:
"...The Rubella Subcommittee of the National Committee for Clinical Laboratory
Standards has proposed lowering the breakpoint to define rubella immunity from 15
to 10 IU/mL. This recommendation stems from epidemiologic studies on vaccinated
persons with low levels of antibody and anecdotal reports. Additional support comes
from Centers for Disease Control and Prevention studies and reports. The effectiveness
of rubella vaccination is well documented and the 10 IU/mL antibody level is
protective in the vast majority of persons... The Subcommittee, recognizing that
sporadic and conflicting reports may suggest a relationship between antibody levels
and protection against the rubella virus, did not advocate lowering the breakpoint
<10 IU/mL"
Charlton et al. (2016), provides further context:
E-20
-------
APRIL 2024
"...the level of rubella IgG antibody is used as a surrogate marker for protection. In
1985, the Rubella Subcommittee of the National Committee on Clinical Laboratory
Standards (NCCLS) set a level of >15 IU/mlfor rubella IgG antibodies as the indicator
of immunity. In light of further epidemiological investigations, and additional studies
indicating that individuals with low levels of antibody (<15 IU/ml) produced a
secondary immune response upon vaccine challenge rather than a primary immune
response, these cut offs were revised by the Subcommittee from 15 IU/ml to 10 IU/ml
in 1992. However, since 1992, the rubella cutoffs have not been assessed."
As noted by Charlton et al. (2016) and the other literature cited above, the geographical
variability, lack of consensus, and relatively dated assessment of this cutoff precludes its use as
the basis of the BMR.
In the absence of a clear definition of an adverse effect for a continuous endpoint like antibody
concentrations, a default BMR of 1 or V2 SD change from the control mean may be selected (U.S.
EPA, 2012). The SD of the rubella antibody concentrations in OD units was estimated from the
distributional data provided in Table 3 in Granum et al. (2013): the 25th and 75th percentiles of
the rubella antibody concentrations in OD units were 1.5 and 2.1, respectively. Assuming that
these values are reasonably represented by a normal distribution, the IQR is approximately
1.35 SDs. Thus, SD = IQR/1.35, and the SD of rubella antibodies in OD is 0.44. The SD of
rubella antibodies in IU/mL units was provided by study authors and was 28.6. Table E-15
presents the BMDs and BMDLs at BMRs of V2 SD and 1 SD. Note that the estimated
BMD/BMDLs were the same regardless of the units (OD or IU/mL) used in the analysis.
As an additional check, EPA evaluated how much extra risk would have been associated with a
BMR set at a 10 IU/mL cutoff value for rubella seropositivity, given the uncertainty in definitive
cutoffs for rubella in OD or IU/mL units discussed above. Because rubella antibody levels were
reported in OD units and IU/mL units, EPA investigated the extra risk using both units.
First, the extra risk was investigated using the distributional data in OD units and the BMR
cutoff value of 0.990 or 0.927 OD, which were used to determine rubella seropositivity in
Granum et al. (2013). Communications with the study authors confirmed that in Granum et al.
(2013), two different OD cutoffs were used for rubella seropositivity in two different runs:
>0.990 OD or >0.927 OD (St0levik, 2012). Of the 50 samples, 47 samples were seropositive.
The remaining three samples were equivocal (i.e., between 0.590-0.990 or 0.553-0.927 OD).
None of the 50 samples were considered seronegative (i.e., <0.590 or <0.553 OD) for rubella.
All participants were vaccinated for rubella, and Granum et al. (2013) noted that "[cjhildren not
following the Norwegian Childhood Vaccination Program (n = 4) were excluded from the
statistical analyses regarding vaccination responses."
Using these BMR cutoffs and the distribution of rubella antibodies in OD, EPA calculated that
1.4-2.0% of the values would be below the cutoffs. A BMR of V2 SD resulted in 4.6% or 6.1% of
the values being below the cutoffs of 0.927 or 0.990 OD, respectively, which is -4% extra risk.
A BMR of 1 SD resulted in 12% or 15% of the values being below the cutoffs of 0.927 or
0.990 OD, respectively, which is -12.7% extra risk. This suggests that in this case, BMRs of V2
or 1 SD provide reasonably good estimates of 5% and 10% extra risk.
E-21
-------
APRIL 2024
Then, using the distributional data of rubella antibodies in IU/mL and a cutoff of 10 IU/mL,
which was considered as the protective antibody level for rubella, EPA calculated that 3.8% of
the values would be below the cutoff. A BMR of '/2 SD resulted in 10% of the values being
below the cutoff, which is —6.3% extra risk. A BMR of 1 SD resulted in 21.8% of the values
being below the cutoff, which is —18% extra risk. This further suggests that in this case, BMRs
of V2 or 1 SD provide reasonably good estimates of 5% and 10% extra risk.
Table E-15. BMDs and BMDLs for Effect of Maternal Serum PFOS on Anti-Rubella
Antibody Concentrations in Children Using a BMR of Vz SD Change in Rubella Antibodies
Concentration and a BMR of 1 SD Change in Rubella Antibodies Concentration (Granum
et al., 2013)
BMD (ng/mL)
BMDL (ng/mL)
BMR
P = -0.08 per ng/mL (For Units of OD)
P = -0.14 per ng/mL (For Units of OD)
P = -5.1 per ng/mL (For Units of IU/mL)
P = -9.0 per ng/mL (For Units of IU/mL)
V2SD
2.8
1.6
1 SD
5.7
3.2
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; IU = international units;
OD = optical density; SD = standard deviation.
For immunotoxicity related to Rubella associated with PFOS exposure measured at age three
years old, the POD estimated for comparison purposes were based on a BMR of V2 SD and a
BMDLV2 SD of 1.6 ng/mL.
E. 1.1.3.2 Zhang et a I. (2023c)
Zhang et al. (2023c) investigated the association between exposure to PFAS and vaccination
responses in children aged 12 to 19 years. A total of 819 children in the United States were
evaluated from the 2003-2004 and 2009-2010 cycles of NHANES. Antibody titers specific to
rubella, mumps, and measles were measured in serum, and rubella antibody levels were inversely
associated with PFOS serum levels (mean = 12.44 ng/mL).
Zhang et al. (2023c) fit multivariate regression models of natural log (In) transformed serum
PFOS concentrations against ln-transformed anti-rubella antibody levels in children, adjusting
for age, sex, race, income-poverty ratio, BMI, serum cotinine concentrations, survey cycle, and
dietary intake of milk and milk products, eggs, and meat. Estimates from the linear regression
models for the total population were then back-transformed to express the results as percent
difference in rubella antibody concentrations per each 2.7-fold increase in serum PFAS
concentration, which was -8.16 (95% CI: -13.67, -2.31, Table 2, Zhang et al. (2023c).
Using the equation provided below, EPA estimated the regression slope as -0.085 (95% CI:
-0.15, -0.02).
Percent Difference = (2.71^ — 1) x 100
Zhang et al. (2023c) also reported summary statistics of PFOS concentration in ng/mL (GM:
12.44; 25th, 75th percentiles: 7.35, 21.90) and of rubella antibody levels in IU/mL (GM: 45.21;
25th, 75th percentiles: 31.25, 64.52) for the total population of 819 participants. All participants
had detectable levels of PFOS in serum.
E-22
-------
APRIL 2024
As a sensitivity analysis, EPA also re-expressed the reported P coefficients in terms of per
ng/mL, according to Dzierlenga et al. (2020). Then EPA used the re-expressed P and lower limit
on the confidence interval to estimate BMD and BMDL.
EPA considered a similar approach to those described above for decreased tetanus antibody
concentrations in Budtz-j0rgensen and Grandjean (2018, 5083631; see Section E.l.1.1.1), to
estimate the BMD/BMDL associated with decreased rubella antibody concentrations in Zhang et
al. (2023c). In the absence of a clear definition of an adverse effect for a continuous endpoint
like antibody concentrations, a default BMR of 1 or V2 SD change from the control mean may be
selected (U.S. EPA, 2012). Table E-16 presents the BMDs and BMDLs at BMRs of V2 SD and
1 SD.
As an additional check, EPA evaluated how much extra risk would have been associated with a
BMR set at a 10 IU/mL cutoff value for rubella seropositivity. EPA calculated that 0.25% of the
values would be below the cutoff. A BMR of V2 SD resulted in 1.1% of the values being below
the cutoff, which is —0.8% extra risk. A BMR of 1 SD resulted in 3.5% of the values being below
the cutoff, which is -3.3% extra risk. The Benchmark Dose Technical Guidance (U.S. EPA,
2012) explains that in a control population where 1.4% are considered to be at risk of having an
adverse effect, a downward shift in the control mean of 1 SD results in about 10% extra risk of
having an adverse effect. However, the BMR cutoff value of 10 IU/mL in the observed
distribution of rubella antibodies in ln(IU/mL) resulted in only 0.25% of the control population at
risk of having an adverse effect, a value much smaller than 1.4% recommended by the technical
guidance, suggesting that, in this case, a BMR of 1 SD or V2 SD may not be a reasonably good
estimate of 10% and 5% extra risk, respectively.
This may be due to the way the study population was restricted to only seropositive adolescents.
The 886 NHANES children with complete data had a rubella seropositivity rate of 96.39%.
Participants without detectable antibodies were excluded and only the 819 children with
detectable antibody serum levels to both measles and rubella (as a proxy for having
measles-mumps-rubella (MMR) vaccination, to reduce confounding by vaccination and health
consciousness) were included in the final study population. This makes it likely that the children
in the study all had antibody rubella levels above the hypothesized clinical threshold of 10
IU/mL.
Table E-16. BMDs and BMDLs for Effect of PFOS on Anti-Rubella Antibody
Concentrations in Adolescents (Zhang et al., 2023c) Using a BMR of V2 SD Change in
Ln(Rubella Antibodies Concentration) and a BMR of 1 SD Change in Ln(Rubella
Antibodies Concentration)
BMD (ln(ng/mL)) BMDL (ln(ng/mL)) BMD (ng/mL) BMDL (ng/mL)
BMR
P = -0.085 per ln(ng/mL) p = -0.147 per ln(ng/mL) p = -0.006 per ng/mL p = -0.011 per ng/mL
Vi SD 32 L8 4L9 243
1 SD 63 3/7 818 48i>
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; SD = standard deviation.
For immunotoxicity related to rubella associated with PFOS exposure measured at ages 12
to 19 years old, the POD estimated for comparison purposes were based on a BMR of Vz SD
and a BMDLh sd of 1.8 In (ng/mL) or 24.3 ng/mL.
E-23
-------
APRIL 2024
E.l.1.3.3 Summary of Modeling Results for Decreased Rubella Antibody
Concentrations
Table E-17 presents the BMDs and BMDLs for all studies considered for POD derivation, with
and without accounting for re-expression of the reported P coefficients in terms of per ng/mL
when necessary.
Table E-17. BMDs and BMDLs in ng/mL for Effect of PFOS on Anti-Rubella Antibody
Concentrations
Study
Exposure
Reported p
(95% CI)
Units
Re-Expressed p
BMR =
'/2SD
BMR = 1 SD
Mean
(VSvo CI)
Ln(IU/mL)/(ng/mL)
BMD
BMDL
BMD
BMDL
Granum et
al. (2013)
5.6
-5.1 (-9.0,-1.1)
(IU/mL)/ng/mL
NA
2.8
1.6
5.7
3.2
Zhang et
al. (2023c)
12.4
-8.16 (-13.67,
-2.31)
(IU/mL)/ln(ng/m
L)
NA
23.4
6.2
549.3
38.6
Zhang et
al. (2023c)
12.4
-8.16 (-13.67,
-2.31)
(IU/mL)/ln(ng/m
L)
-0.0006 (-0.0111,
-0.0018)
41.9
24.3
83.9
48.6
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; CI = confidence interval;
IU = international units; NA = not applicable; SD =standard deviation; SE = standard error.
Table E-18 summarizes the PODs resulting from the modeling approaches for decreased rubella
antibody concentrations. The selected and comparison PODs were based on a BMR of V2 SD,
resulting in BMDLs ranging from 1.6 ng/mL to 24.3 ng/mL with the selected POD of 1.6 ng/mL.
Table E-18. BMDLs for Effect of PFOS on Anti-Rubella Antibody Concentrations Using a
BMR of 5%
Study Name
Effect
BMDL (ng/mL)
Granum et al. (2013)
PFOS prenatally on anti-rubella antibody concentrations at age three
1.6
years
Zhang et al. (2023c)
PFOS and anti-rubella antibody concentrations at ages 12-19 years
24.3
Notes: BMDL = benchmark dose lower limit; BMR = benchmark response.
E.1.2 Modeling Results for Decreased Birthweight
Six high confidence studies (Yao et al., 2021; Chu et al., 2020; Wikstrom et al., 2020; Sagiv et
al., 2018; Starling et al., 2017; Darrow et al., 2013) reported decreased birth weight in infants
whose mothers were exposed to PFOS. These candidate studies offer a variety of PFOS exposure
measures across the fetal and neonatal window. All six studies reported their exposure metric in
units of ng/mL and reported the P coefficients per ng/mL or ln(ng/mL), along with 95%
confidence intervals, estimated from linear regression models. The logarithmic transformation of
E-24
-------
APRIL 2024
exposure yields a negative value for small numbers, which can result in implausible results from
dose-response modeling (i.e., estimated risks are negative and unable to determine the responses
at zero exposure). EPA first re-expressed the reported P coefficients in terms of per ng/mL, if
necessary, according to Dzierlenga et al. (2020). Then EPA used the re-expressed P and lower
limit on the confidence interval to estimate BMD and BMDL values using the general equation^
= mx + b, where y is birth weight and x is exposure, substituting the re-expressed P values from
these studies for m. The intercept b represents the baseline value of birth weight in an unexposed
population and it can be estimated through y using an average birth weight from an external
population asy, an average exposure as x and re-expressed P from the studies as m.
The CDC Wonder site (https://wonder.cdc.eov/natality.html) provides vital statistics for babies
born in the United States. There were 3,791,712 all live births in the United States in 2018
according to final natality data. The mean and standard deviation of birth weight were
3,261.6 ± 590.7 g (7.19 ± 1.30 lb), with 8.27% of live births falling below the public health
definition of low birth weight (i.e., 2500 g, or 5.5 lb). The full natality data for the United States
data on birth weight was used as it is more relevant for deriving toxicity values for the U.S.
general public than the study-specific birthweight data. Also, the CDC Wonder database may be
queried to find the exact percentage of the population falling below the cutoff value for clinical
adversity. America's Children and the Environment (ACE) Biomonitoring on
Perfluorochemicals (https://www.epa.gov/americaschildrenenvironment/data-tables-
biomonitoring-perfluorochemicals-pfcs) provides in Table B6b the median blood serum levels of
PFOS of 2.6 ng/mL in 2015-2016 in woman ages 16 to 49, using NHANES as data source.
These values are assumed to be representative of women of reproductive age and are
subsequently used in the estimation of BMD and BMDL values from the available four
epidemiological studies.
E.l.2.1 Chu et al. (2020)
Chu et al. (2020) reported a P coefficient of-83.3 g (95%CI: -133.2, -33.4) per ln(ng/mL)
increase for the association between birth weight and maternal PFOS serum concentrations
(collected within 3 days of delivery) in a China cohort. The reported P coefficient can be re-
expressed in terms of per ng/mL according to Dzierlenga et al. (2020). Given the reported study-
specific median (7.2 ng/mL) and the 25th and 75th percentiles (4.4 and 11.9 ng/mL) of the
exposure from Chu et al. (2020), EPA estimated the distribution of exposure by assuming the
exposure follows a lognormal distribution with mean and standard deviation as:
PL = ln(q50) = ln{72) = 1.97 (1)
o- = ln(q75/q2 5)/1.349 = Zn(11.9/4.4)/1.349 = 0.75 (2)
Then, EPA estimated the 25th-75th percentiles at 10 percentile intervals of the exposure
distribution and corresponding responses of reported P coefficient. The re-expressed P
coefficient is determined by minimizing the sum of squared differences between the curves
generated by the re-expressed P and the reported p. Doing so results in a re-expressed P
coefficient of-11.0 g (95% CI: -17.6, -4.4) per ng/mL.
E-25
-------
APRIL 2024
Typically, for continuous data, the preferred definition of the BMR is to have a basis for what
constitutes a minimal level of change in the endpoint that is biologically significant. For birth
weight, there is no accepted percent change that is considered adverse. However, there is a
clinical measure for what constitutes an adverse response. Babies born weighing less than
2,500 g are considered to have low birth weight, and further, low birth weight is associated with
a wide range of health conditions throughout life (Tian et al., 2019a; Reyes and Manalich, 2005;
Hack et al., 1995). Given this clinical cutoff for adversity and that 8.27% of all live births in the
United States in 2018 fell below this cutoff, the hybrid approach can be used to define the BMR.
The hybrid approach harmonizes the definition of the BMR for continuous data with that for
dichotomous data, and therefore is an advantageous approach10. Essentially, the hybrid approach
involves the estimation of the dose that increases the percentile of responses falling below (or
above) some cutoff for adversity in the tail of the response distribution. Application of the hybrid
approach requires the selection of an extra risk value for BMD estimation. In the case of birth
weight, an extra risk of 5% is selected given that this level of response is typically used when
modeling developmental responses from animal toxicology studies, and that low birthweight
confers increased risk for adverse health effects throughout life, thus supporting a BMR lower
than the standard BMR of 10% extra risk.
Therefore, given a background response and a BMR = 5% extra risk, the BMD would be the
dose that results in 12.86% of the responses falling below the 2,500 g cutoff value:
Extra Risk(HR) = (P(d) - P(0)) / (1 - P(0))
P(cl) = ER(1 - P(0)) + P(0) = 0.05(1 - 0.0827) + 0.0827 = 0.1286
Using the mean birth weight for all births in the United States in 2018 of 3,261.6 g with a
standard deviation of 590.7 g, EPA calculated the mean response that would be associated with
the 12.86th percentile of the distribution falling below 2,500 g. In this case, the mean birth
weight would be 3,169.2 g. Given the median exposure of 2.6 ng/mL from ACE Biomonitoring
on Perfluorochemicals as x, the mean birth weight in the United States as_y and the re-expressed
P as m term, the intercept b can be estimated as:
h = y^ml= 3261.6g - (-11.0S©"1) 2.6^ = 3290.3g (3)
The BMD was calculated by rearranging the equation y = mx + b and solving for x, using
3,290.3 g for the b term and 1 1.0 for the/« term. Doing so results in a value of 11.0 ng/mL:
x = (y- b)/m = (3169.2 g — 3290.3 g)/(—11.0 = 11.0 ng/mL
mL
To calculate the BMDL, the method is essentially the same except that the lower limit (LL) on
the P coefficient (Pll= -17.6) is used for the m term. However, Chu et al. (2020) reported a two-
10 While the explicit application of the hybrid approach is not commonly used in IRIS do se/ concentration/expo sure-response
analyses, the more commonly used SD-definition of the BMR for continuous data is simply one specific application of the hybrid
approach. The SD-definition of the BMR assumes that the cutoff for adversity is the 1,4th percentile of a normally distributed
response and that shifting the mean of that distribution by one standard deviation approximates an extra risk of 10%.
E-26
-------
APRIL 2024
sided 95% confidence interval for the P coefficient, meaning that the LL of that confidence
interval corresponds to a 97.5% one-sided LL. The BMDL is defined as the 95% LL of the BMD
(i.e., corresponds to a two-sided 90% confidence interval), so the corresponding LL on the P
coefficient needs to be calculated before calculating the BMDL. First, the standard error of the P
coefficient can be calculated as:
Upper Limit — Lower Limit 4.4 1 ( ') na .
SE = — = ^i^= 3.37 q(—Yl
3.92 3,92
Then the corresponding 95% one-sided lower bound on the P coefficient can be calculated as:
95% one - sided LL = fi - 1.645(57; (/?)) = -11 g(r^r)~l - 1-645 (3.37
v J mL \ mL /
=-m5«0~l
Using this value for the m term results in a BMDL value of 7.3 ng/mL maternal serum
concentration.
E.l.2.2 Wikstrom etoL (2020)
Wikstrom et al. (2020) reported a P coefficient of-46.0 g (95%CI: -88.0, -3.0) per ln(ng/mL)
for the association between birth weight and maternal PFOS serum concentrations (collected
during 9 weeks to 10 weeks of pregnancy with a median of 10 weeks) in a Swedish cohort.
Given the reported study-specific median (5.4 ng/mL) and the 25th and 75th percentiles (4.0,
7.6 ng/mL) of the exposure, EPA estimated the mean (1.68) and standard deviation (0.48) of the
log normally distributed exposure. The re-expressed P coefficient is -8.4 g (95%CI: -16.0, -0.5)
per ng/mL and the intercept b is 3,283.4 g. The 95% one-sided LL for the re-expressed P
coefficient is -14.8 g per ng/mL. The values of the BMD and BMDL are 13.7 ng/mL and
7.7 ng/mL, respectively.
E.l.2.3 Sogiv et al. (2018)
Sagiv et al. (2018) reported a P coefficient of-17.9 g (95% CI: -40.9, 5.1) per IQR increase in
PFOS (ng/mL), corresponding to a P coefficient of-1.1 g (95%CI: -2.6, 0.3) per ng/mL
increase, for the association between birth weight and maternal PFOS serum concentrations
(collected during 5 weeks to 19 weeks of pregnancy with a median of 9 weeks) in a U.S. cohort.
The intercept b is 3,264.5 g based on the P coefficient of-1.1 g per ng/mL. A BMD of
85.2 ng/mL is calculated from Sagiv et al. (2018) using the same approach as above with the
same values for the mean birth weight in the United States.
To calculate the BMDL, the same procedure as above is used to calculate the corresponding 95%
one-sided LL for the P coefficient from the LL on the 95% two-sided confidence interval of
-2.6 g per ng/mL. Using the corresponding LL (-2.3 g per ng/mL), a BMDL of 41.0 ng/mL is
calculated.
E-27
-------
APRIL 2024
E.1.2.4 Starling et al. (2017)
Starling et al. (2017) reported a P coefficient of-13.8 g (95%CI: -53.8, 26.3) per ln(ng/mL) for
the association between birth weight and maternal PFOS serum concentrations (collected during
20 to 34 weeks of pregnancy with a median of 27 weeks) in a U.S. cohort. Given the reported
study-specific median (2.4 ng/mL) and the 25th and 75th percentiles (1.5, 3.7 ng/mL) of the
exposure, EPA estimated the mean (0.88) and standard deviation (0.67) of the log normally
distributed exposure. The re-expressed P coefficient is -5.5 g (95%CI: -21.4, 10.5) per ng/mL
and the intercept b is 3,275.9 g. The 95% one-sided LL for the re-expressed P coefficient is
-18.9 g per ng/mL. The values of the BMD and BMDL are 19.4 ng/mL and 5.7 ng/mL,
respectively.
E.1.2.5 Darrow et al. (2013)
Darrow et al. (2013) reported a P coefficient of-49.0 g (95%CI: -90.0, -8.0) per ln(ng/mL) for
the association between birth weight and maternal PFOS serum concentrations in a U.S. cohort.
Given the reported study-specific median (13.9 ng/mL) and the 25th and 75th percentiles (9.5,
19.7 ng/mL) of the exposure, EPA estimated the mean (2.63) and standard deviation (0.54) of the
log normally distributed exposure. The re-expressed P coefficient is -3.4 g (95%CI: -6.3, -0.6)
per ng/mL and the intercept b is 3,270.5 g. The 95% one-sided LL for the re-expressed P
coefficient is -5.8 g per ng/mL. The values of the BMD and BMDL are 29.6 ng/mL and
17.4 ng/mL, respectively.
E.l.2.6 Yao etal. (2021)
Yao et al. (2021) reported a P coefficient of-32.3 g (95%CI: -116.2, 51.6) per ln(ng/mL) for the
association between birth weight and maternal PFOS serum concentrations (collected within
3 days of delivery) in a China cohort. Given the cohort-specific median (4.6 ng/mL) and the 25th
and 75th percentiles (3.2, 5.9 ng/mL) of the exposure reported in Han et al. (2018b), EPA
estimated the mean (1.52) and standard deviation (0.45) of the log normally distributed exposure.
The re-expressed P coefficient is -6.9 g (95%CI: -25.0, 11.1) per ng/mL and the intercept b is
3,279.7 g. The 95% one-sided LL for the re-expressed P coefficient is -22.1 g per ng/mL. The
values of the BMD and BMDL are 15.9 ng/mL and 5.0 ng/mL, respectively.
E.1.2.7 Summary of Modeling Results for Decreased Birthweight
For all of the above calculations, EPA used the exact percentage (8.27%) of live births in the
United States in 2018 that fell below the cutoff of 2,500 g as the tail probability to represent the
probability of extreme ("adverse") response at zero dose (P(0)). However, this exact percentage
of 8.27%) was calculated without accounting for the existence of background PFOS exposure in
the U.S. population (i.e., 8.27% is not the tail probability of extreme response at zero dose).
Thus, EPA considers an alternative control-group response distribution (A'(//c, crc)), using the
study-specific intercept b obtained through equation (3) (representing the baseline value of birth
weight in an unexposed population) as fie and the standard deviation of U.S. population as ac, to
estimate the tail probability that fells below the cutoff of 2,500 g. EPA estimated the study-
specific tail probability of live births falling below the public health definition of low birth
weight (2,500 g) as:
E-28
-------
APRIL 2024
1 ^2500 , (x-b)2^ -J (.2500 ^ (x-b)2 y
Pro) = — I e 2ac dx = -=. I e(-~2*59o.72^ dx
-------
APRIL 2024
Table E-19. BMDs and BMDLs in ng/mL for Effect of PFOS on Decreased Birth Weight, by Using the Exact Percentage
(8.27%) of Live Births Falling Below the Public Health Definition of Low Birth Weight, or Alternative Study-Specific Tail
Study
Exposure
Median
Exposure
Distribution
Reported p
(95% CI)
Re-expressed p
(95% CI)
Intercept
SE(P)
Pll
Exact Percentage
(P(0) = 8.27%)
Alternative Tail Probability
a
(25th-75th
Percentiles)
(H, a)
units
g/(ng/mL)
b
BMD
BMDL
P(0)
BMD
BMDL
Chuetal. (2020)
7.2(4.4-11.9)
(1.97,0.75)
-83.3
(-133.2, -33.4)
g/ln(ng/mL)
-11.0 (-17.6,-4.4)
3,290.3
3.37
-16.5
11.0
7.3
9.05%
12.8
8.5
Sagiv et al.
(2018)
25.7(18.9-
34.9)
(3.25,0.45)
-17.9 (-40.9, 5.1)
g/IQR (ng/mL)
-1.1 (-2.6,0.3)
3,264.5
0.73
-2.3
85.2
41.0
9.78%
119.8
57.6
Starling et al.
(2017)
2.4(1.5-3.7)
(0.88, 0.67)
-13.8 (-53.8, 26.3)
g/ln(ng/mL)
-5.5 (-21.4,10.5)
3,275.9
8.14
-18.9
19.4
5.7
9.45%
25.0
7.3
Wikstrom et al.
(2020)
5.4 (4.0-7.6)
(1.68,0.48)
-46.0 (-88.0, -3.0)
g/ln(ng/mL)
-8.4 (-16.0, -0.5)
3,283.4
3.94
-14.8
13.7
7.7
9.24%
16.7
9.4
Darrow et al.
(2013)
13.9(9.5-19.7)
(2.63, 0.54)
-49.0 (-90.0, -8.0)
g/ln(ng/mL)
-3.4 (-6.3, -0.6)
3,270.5
1.46
-5.8
29.6
17.4
9.60%
40.0
23.3
Yaoetal. (2021)
4.6 (3.2-5.9)
(1.52,0.45)
-32.3 (-116.2,
51.6) g/ln(ng/mL)
-6.9 (-25.0,11.1)
3,279.7
9.22
-22.1
15.9
5.0
9.34%
19.9
6.3
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; CI = confidence interval; IQR = interquartile range; SE = standard error.
a The alternative study-specific tail probability of live births falling below the public health definition of low birth weight based on normal distribution with intercept b as mean and
standard deviation of 590.7 based on the U.S. population.
E-30
-------
APRIL 2024
ACE Biomonitoring on Perfluorochemicals also provides the median blood serum levels of
PFOS among women ages 16 to 49 in 1999-2000 (23.8 ng/mL), in 2009-2010 (5.7 ng/m) and in
2013-2014 (3.0 ng/mL). EPA performed a sensitivity analysis by estimating BMD and BMDL
using these values as background exposures. The results for each study considered for POD
derivation, presented in Table E-20, demonstrate the robustness of EPA's approaches with
alternative assumptions on background exposures.
Table E-20. BMDs and BMDLs for Effect of PFOS on Decreased Birth Weight by
Background Exposure, Using the Exact Percentage of the Population (8.27%) of Live
Births Falling Below the Public Health Definition of Low Birth Weight, or Alternative Tail
Probability
Study Background Intercept Exact Percentage Alternative Tail Probabilityb
Exposurea /, (P(0) = 8.27%)
BMD
BMDL
m
BMD
BMDL
(ng/mL)
(ng/mL)
(ng/mL)
(ng/mL)
Wikstrom
2.6
3,283.4
13.7
7.7
9.24%
16.7
9.4
et al.
3.0
3,286.7
14.1
7.9
9.14%
16.8
9.5
(2020)
5.7
3,309.2
16.8
9.4
8.53%
17.6
9.9
23.8
3,460.4
34.9
19.6
5.20%
24.1
13.6
Chu et al.
2.6
3,290.3
11.0
7.3
9.05%
12.8
8.5
(2020)
3.0
3,294.7
11.4
7.6
8.93%
13.0
8.6
5.7
3,324.4
14.1
9.4
8.14%
13.8
9.2
23.8
3,523.6
32.2
21.4
4.16%
20.9
13.9
Darrow et
2.6
3,270.5
29.6
17.4
9.60%
39.7
23.3
al. (2013)
3.0
3,271.9
30.0
17.6
9.56%
39.8
23.4
5.7
3,281.1
32.7
19.2
9.30%
40.5
23.8
23.8
3,343.1
50.8
29.9
7.67%
46.0
27.1
Sagiv et
2.6
3,264.5
85.2
41.0
9.78%
119.8
57.6
al. (2018)
3.0
3,265.0
85.6
41.2
9.76%
119.9
57.7
5.7
3,268.0
88.3
42.5
9.68%
120.7
58.1
23.8
3,288.3
106.4
51.2
9.10%
125.8
60.5
Starling et
2.6
3,275.9
19.4
5.7
9.45%
25.0
7.3
al. (2017)
3.0
3,278.1
19.8
5.8
9.39%
25.1
7.3
5.7
3,293.0
22.5
6.6
8.97%
25.9
7.5
23.8
3,392.4
40.6
11.8
6.54%
31.8
9.3
Yao et al.
2.6
3,279.7
15.9
5.0
9.34%
19.9
6.3
(2021)
3.0
3,282.5
16.3
5.1
9.26%
20.0
6.3
5.7
3,301.2
19.0
6.0
8.75%
20.8
6.5
23.8
3,427.0
37.1
11.7
5.83%
27.0
8.5
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit;.
a Assumptions on background exposure for the estimation of intercept using Equation (3).
b The tail probability of live births falling below the public health definition of low birth weight based on normal distribution.
For decreased birth weight associated with PFOS exposure, the POD selected from the available
epidemiologic literature is 7.7 ng/mL maternal serum concentration, based on birth weight data
E-31
-------
APRIL 2024
from Wikstrom et al. (2020). Of the six individual studies, Sagiv et al. (2018) and Wikstrom et
al. (2020) assessed maternal PFOS serum concentrations primarily or exclusively in the first
trimester, minimizing concerns surrounding bias due to pregnancy-related hemodynamic effects.
Therefore, the PODs from these two studies were considered further for POD selection. The
POD from Wikstrom et al. (2020) was ultimately selected as the reported PFOS exposure
concentrations were more representative of current U.S. exposure levels compared with the
levels reported in Sagiv et al. (2018), and it was the lowest POD from these two studies.
E.1.3 Modeling Results for Liver Toxicity
This updated review indicated that PFOS is associated with increases in the liver enzyme ALT
(See Toxicity Assessment, (U.S. EPA, 2024)). Three medium confidence studies were selected as
candidates for POD derivation. One of the largest studies of PFOS and ALT in adults is Gallo et
al. (2012) conducted in 47,092 adults from the C8 Study Project (for detailed descriptions of the
study and findings see Toxicity Assessment, (U.S. EPA, 2024) and Appendix D). Two additional
studies (Nian et al., 2019; Lin et al., 2010) were considered by EPA for POD derivation because
they reported significant association in general populations in the United States and a high
exposed population China, respectively. In an NHANES adult population, Lin et al. (2010)
observed elevated ALT levels per log-unit increase in PFOS in the models adjusted for age,
gender, and race/ethnicity, but not in the fully adjusted models or in the models additionally
adjusted for PFOA, PFHxS, and PFNA. While this is a large nationally representative
population, several methodological limitations preclude its use for POD derivation. Limitations
include lack of clarity about base of logarithmic transformation applied to PFOS concentrations
in regression models, and the choice to model ALT as an untransformed variable, a departure
from the typically lognormality assumed in most of the ALT literature.
Nian et al. (2019) examined 1,605 adults in Shenyang (one of the largest fluoropolymer
manufacturing centers in China) part of the Isomers of C8 Health Project and observed
significant increases in ln-transformed ALT per each ln-unit increase in PFOS, as well
significant increases in odds ratios of elevated ALT. Median serum PFOS concentrations were
24.22 ng/mL.
E.1.3.1 Nian et al. (2019)
No-observed-adverse-effect concentration/lowest-observed-adverse-effect concentration
(NOAEC/LOAEC) method. Significant positive linear trends were observed for branched PFOS
with ORs of elevated ALT across quartiles of exposure (p-value = 0.04). However, categorical
data, which can be used to develop NOAECs, were not available for total PFOS from the peer-
reviewed publication.
Hybrid method. The previously described hybrid method was implemented using data from Nian
et al. (2019). The regression model adjusted for age, sex, career, income, education, drink,
smoke, giblet and seafood consumption, exercise, and BMI. The percentage change in In ALT
for ln-unit increase in PFOS was 4.1 (95% CI: 0.6, 7.7) (Table 3, Nian et al. (2019). The reported
regression coefficient P, which is also referred to as /??, was calculated from the reported percent
change expressed as (ep-l)*100, resulting in a slope of 0.04 (95% CI: 0.01, 0.07) In ALT (IU/L)
per In ng/mL PFOS. The estimated BMDs and BMDLs are presented in Table E-21.
E-32
-------
APRIL 2024
For increased ALT associated with PFOS exposure, the POD is based on the data Nian et al.
(2019), a BMR of 5% and a BMDLs of 15.1 ng/mL.
Table E-21. BMD and BMDL for Effect of PFOS (ng/mL) on Increased ALT in Nian et al.
(2019), for 5% and 10% Extra Risk
Time Period
1999-2018
1999-2018
2003-2018
2003-2018
2017-2018
2017-2018
Sex
Male
Female
Male
Female
Male
Female
BMR = 5%, P(0) Empirical
BMD
36.82
25.93
41.00
24.89
19.58
10.97
BMDL
22.29
15.12
23.49
14.57
11.73
6.84
BMR = 5%, P(0) Lognormal
BMD
69.49
43.37
68.30
40.87
34.44
20.81
BMDL
32.30
20.42
31.64
19.46
16.32
9.94
BMR = 10%, P(0) Empirical
BMD
206.25
134.66
225.92
126.14
105.81
57.11
BMDL
60.98
39.58
63.63
37.58
31.43
17.93
BMR = 10%, P(0) Lognormal
BMD
352.86
206.31
347.61
190.43
171.58
97.41
BMDL
83.44
50.78
81.84
47.80
41.68
24.50
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response.
E.l.3.2 Gallo et al. (2012)
Gallo et al. (2012) evaluated the relationship between PFOS and ALT using two general types of
analyses. In the first, subjects were divided into deciles of PFOS exposure, and linear regression
models were used to compare mean ALT levels by each non-reference quantile versus mean
ALT level in the lowest decile. In the second type of analysis, a logistic regression evaluated
ORs for having an ALT level above a certain cutoff for each non-reference deciles compared
with the lowest (reference) deciles. The cutoff values used to define elevated ALT levels in this
study were 45 IU/L for men and 34 IU/L for women, clinically based value recommended by the
International Federation of Clinical Chemistry and Laboratory Medicine (Schumann et al.,
2002), and were approximately the 90th percentile of all ALT values in this study.
E. 1.3.2.1 Elevated ALT
E.l.3.2.1.1 Hybrid Method
The hybrid method used the regression slope from the linear regression model of ln-transformed
ALT and In PFOS concentrations adjusted for age, sex, alcohol consumption, socioeconomic
status, fasting status, race, month of blood sample collection, smoking status, body mass index,
physical activity, and insulin resistance. The reported regression coefficient P, which is also
referred to as m, was 0.02 (95% CI: 0.014, 0.026) of In ALT (IU/L) per In ng/mL PFOS (Table 2,
Gallo et al. (2012), model 3).
Using a normal approximation, the standard error of the regression coefficient is estimated as
Upper Limit — Lower Limit 0.026 — 0.014
SE = — —— = — = 0.0025
3.92 3.92
E-33
-------
APRIL 2024
Elevated ALT is a biomarker of acute liver disease. For the following analyses, the adverse
effect level of ALT for liver disease was chosen to be C = 42 IU/L for males and C = 30 IU/L for
females, based on the sex-specific upper reference limits found in Valenti et al. (2021). These are
slightly lower and more health protective than the cutoff values used in the original study
(45 IU/L for men and 34 IU/L for women). These cutoffs are also slightly higher than the
American College of Gastroenterology (ACG) cutoffs, which considers that "true healthy normal
ALT level ranges from 29 to 33 IU/L for males, 19 to 25 IU/L for females" (Kwo et al., 2017).
They are the most updated clinical consensus cutoffs, which update the American Association
for the Study of Liver Diseases (AASLD) journal Clinical Liver Disease recommended values of
30 IU/L for males, and 19 IU/L for females (Ducatman et al., 2023; Kasarala and Tillmann,
2016). Valenti et al. (2021) determined the updated values using the same approach at the same
center but using an updated standardized method.
These analyses were for the periods 1999-2018, 2003-2018, and 2017-2018, separately for
males and females ages 18 and over, assuming that the Gallo regression model coefficient
developed for the C8 Health Project data in Ohio starting in 2005 and 2006 can be applied to the
alternative NHANES periods. These analyses used the NHANES-recommended regression
model adjustment to correct the 2017-2018 ALT data to match the earlier laboratory method.
EPA used the NHANES PFOS data for each NHANES period including data adjustments to
stored biospecimen data collected in 1999-2000 and 2013-2014 that were publicly released in
April 2022. NHANES survey weights were applied.
Using the NHANES data for each period and sex, EPA estimated the mean and standard
deviation of In ALT and the estimated mean In PFOS (Table E-22). The unrounded values were
used in the calculations:
Table E-22. NHANES Mean and Standard Deviation of Ln(ALT) (In IU/L) and Mean
PFOS (Ln ng/mL)
Time Period
1999-2018
1999-2018
2003-2018
2003-2018
2017-2018
2017-2018
Sex
Male
Female
Male
Female
Male
Female
Mean ln ALT (ln IU/L) (y)
3.28
2.96
3.28
2.96
3.29
2.96
Standard Deviation ln ALT (ln
0.46
0.41
0.46
0.41
0.48
0.42
IU/L) (S)
Mean ln PFOS (ln ng/mL) (X)
2.40
1.96
2.37
1.93
1.74
1.26
Notes: ALT = alanine transaminase; IU = international units; NHANES = National Health and Nutrition Examination Survey.
For the BMD analyses, the response of interest is elevated ALT, defined as ALT greater than or
equal to an adverse effect threshold C IU/L defined as 42 IU/C for males and 30 IU/L for
females. EPA estimated P(0), the prevalence of population with elevated ALT using two
approaches. First, the empirical estimate of P(0), "P(0) Empirical," was calculated as the
proportion of the population with ALT greater than or equal to C, using the NHANES survey
weights. Second, the lognormal estimate of P(0), "P(0) Lognormal," was calculated assuming
that ALT is lognormally distributed using the equation:
E-34
-------
APRIL 2024
,, . (ln(C) — mean(\n ALT))
P(0) Lognormal = 1 - ® | j
where €> is the normal cumulative distribution function.
The selected BMR is an extra risk of either 5% or 10%. The extra risk of high ALT is given by
the equation
P(cl) - P(0)
Extra Risk = ——
1 - P(0)
where P(d) is the probability of ALT greater than or equal to C (IU/L) for a given PFOS dose d.
Thus
P(cl) = {1 - P(0)} x Extra Risk + P(0)
The values of C, P(0) Empirical, /'(d) Empirical, /'(d) Lognormal for Extra Risk 5% or 10%, and
P(d) Lognormal for Extra Risk 5% or 10% are shown in Table H-23.
Table E-23. Prevalence of Elevated ALT
Time Period
1999-2018
1999-2018
2003-2018
2003-2018
2017-2018
2017-2018
Sex
Male
Female
Male
Female
Male
Female
Adverse effect level C (IU/L)
42
30
42
30
42
30
P(0) Empirical
0.14
0.13
0.15
0.13
0.16
0.13
P(d) Empirical, Extra Risk 5%
0.19
0.17
0.19
0.17
0.20
0.17
P(d) Empirical, Extra Risk 10%
0.23
0.21
0.23
0.21
0.24
0.22
P(0) Lognormal
0.16
0.14
0.16
0.14
0.17
0.15
P(d) Lognormal, Extra Risk 5%
0.20
0.18
0.20
0.18
0.22
0.19
P(d) Lognormal, Extra Risk 10%
0.24
0.23
0.24
0.23
0.26
0.23
Notes: ALT = alanine transaminase; IU = international units.
The mean In ALT y for a In PFOS dose x is given by the equation
y = mx + b
where m is the slope, P, (from the Gallo regression model) and b is the intercept. The intercept b
is the mean In ALT for a population exposed to 1 ng/mL PFOS. For the U.S. population, the
mean In ALT is j7 (tabulated above) and the mean In PFOS is x (tabulated above) so the intercept
is given by the equation
b = y — mx
For a given group and dose, the probability of ALT greater than or equal to C is
P(d) = P(ALT > C) = P(ln ALT > In C) =1-0 ^ J)
E-35
-------
APRIL 2024
where
-------
APRIL 2024
E. 1.3.2.1.2 NOAEC/LOAEC Method
The results of the logistic regression analysis of elevated ALT across deciles of PFOS are
presented in Table E-25. The mean, median and ranges of PFOS concentrations in each decile
were not provided with the OR results in the publication. EPA obtained these from author
correspondence, and they are illustrated in Table E-25. The NOAEC is bolded and is the mean
PFOS serum concentration in the highest decile of PFOS that did not show a statistically
significant OR of elevated ALT, which in this case is the 2nd decile, compared with the
reference category (the lowest decile of PFOS). The NOAEC based on the elevated ALT data
from Gallo et al. (2012) is 10.6 ng/mL.
Table E-25. Odds Ratios for Elevated ALT by Decile of PFOS Serum Concentrations
(ng/mL) from Gallo et al. (2012)
Participants Participants
Decile
Minimum
(ng/mL)
Maximum
(ng/mL)
Median
(ng/mL)
Mean
(ng/mL)
OR
95% CI
Without
Elevated
ALT
With
Elevated
ALT
Total
(N)
0
0.25
00
00
6.4
5.751386
1
reference
4,119
427
4,546
1
8.9
12.2
10.7
10.63289
1.09
0.94,1.26
4,264
446
4,710
2
12.3
14.9
13.6
13.60556
1.19
1.03, 1.37
4,113
459
4,572
3
15
17.5
16.3
16.26427
1.26
1.09, 1.45
4,104
500
4,604
4
17.6
20.2
18.9
18.88567
1.40
1.22, 1.62
4,115
545
4,660
5
20.3
23.3
21.7
21.74935
1.39
1.21, 1.60
4,181
571
4,752
6
23.4
27
25.1
25.11534
1.31
1.14, 1.52
4,099
561
4,660
7
27.1
32
29.3
29.38941
1.42
1.23, 1.64
4,071
586
4,657
8
32.1
40.4
35.6
35.76743
1.40
1.21, 1.62
4,068
547
4,615
9
40.5
585.2
49.7
56.12528
1.54
1.33, 1.78
4,124
552
4,676
Notes: ALT = alanine transaminase; CI = confidence interval; NOAEC = no-observed-adverse-effect concentration.
The NOAEC is bolded.
E.1.3.2.1.3 BMD Method
EPA applied BMDS to calculate a BMD. In addition, EPA performed a sensitivity analysis using
the generalized least-squares for trend (gist) method (Greenland and Longnecker, 1992), which
assumes a linear relationship between exposure and log-transformed ORs, and accounts for
covariance between estimates. These analyses were performed in STATA vl7.0 (StataCorp,
2021). Through author correspondence, EPA obtained the number of participants with and
without elevated ALT for each decile of PFOS (Table E-25).
Applying BMDS v3.3rcl0 using a BMR of 10% and 5% the data for all 10 deciles did not result
in any viable models. Applying BMDS v3.3rcl0 to the data for all first five deciles did result in
viable models. The data associated with the first five deciles was also run using a no intercept
approach in which the lowest dose was subtracted out, subsequently referred to as an adjusted
dose. The results of this modeling using both the mean and median doses are summarized in
Table E-26, Table E-27, Table E-28, Table E-29. This modeling approach results in BMD and
BMDL values higher than the maximum dose included in the modeled dataset. The BMD and
BMDL values were inside the range of mean exposure values when considering all 10 deciles.
E-37
-------
APRIL 2024
Table E-26. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et al. (2012) Using the Unadjusted
Mean PFOS Serum Concentration
Model3
Goodness of Fit
Scaled Residual
Dose Group Dose Group Control Dose
Near BMDio Near BMDs Group
p-value AIC
BMDio
(ng/mL)
BMDLio
(ng/mL)
BMDs
(ng/mL)
BMDLs
(ng/mL)
Dichotomous Hill
_b
-
-
-
-
-
-
-
-
Gamma
0.92
15,296.47
-0.11
-0.11
0.16
28.37
25.58
22.69
20.63
Log-Logistic
0.91
15,296.50
-0.11
-0.11
0.17
27.68
22.17
22.50
20.19
Weibull
0.98
15,294.50
-0.11
-0.11
0.17
27.47
23.26
22.46
20.46
Logistic
0.52
15,296.80
0.67
0.67
0.83
43.97
33.33
25.48
19.53
Log-Probit
0.94
15,296.44
-0.10
-0.10
0.14
29.51
22.98
22.98
20.39
Probit
0.51
15,296.87
0.69
0.69
0.83
45.41
34.13
25.66
19.47
Quantal Linear
0.45
15,297.26
0.80
0.80
0.82
54.66
38.95
26.61
18.96
Notes: AIC = Akaike information criterion; ALT = alanine transaminase; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level corresponding to a
10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence limit for a 10% response level; BMD5 = dose level corresponding to a
5%) response level; BMDL5 = lower bound on the dose level corresponding to the 95% lower confidence limit for a 5% response level.
a Selected model in bold.
bBMD Computation failed.
E-38
-------
APRIL 2024
Table E-27. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et al. (2012) Using the Adjusted, No
Intercept Mean PFOS Serum Concentration
Goodness of Fit Scaled Residual
BMDio BMDLio BMDs BMDLs
D-value AIC Dose Group Dose Group Control (ng/mL) (ng/mL) (ng/mL) (ng/mL)
Near BMDio Near BMDs Dose Group
b
Dichotomous Hill
_b
-
-
-
-
-
-
-
-
Gamma
0.95
15,296.40
-0.09
-0.09
0.12
24.22
18.67
17.44
15.03
Log-Logistic
0.95
15,296.41
-0.09
-0.09
0.14
23.67
16.76
17.30
14.58
Weibull
0.94
15,296.42
-0.09
-0.09
0.14
23.39
17.63
17.25
14.87
Logistic
0.52
15,296.80
0.67
0.67
0.83
41.00
30.25
23.47
17.42
Log-Probit
0.97
15,296.36
-0.07
-0.07
0.10
26.47
17.71
17.96
14.79
Probit
0.51
15,296.87
0.69
0.69
0.83
42.78
31.38
23.92
17.64
Quantal Linear
0.45
15,297.26
0.80
0.80
0.82
54.66
38.95
26.61
18.96
Notes: AIC = Akaike information criterion; ALT = alanine transaminase; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level corresponding to a
10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence limit for a 10% response level; BMD5 = dose level corresponding to a
5%) response level; BMDL5 = lower bound on the dose level corresponding to the 95% lower confidence limit for a 5% response level.
a Selected model in bold.
bBMD Computation failed.
E-39
-------
APRIL 2024
Table E-28. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et al. (2012) Using the Unadjusted,
Median PFOS Serum Concentration
Goodness of Fit Scaled Residual
BMDio BMDLio BMDs BMDLs
D-value AIC DoseGrouP Dose Group Control (ng/mL) (ng/mL) (ng/mL) (ng/mL)
Near BMDio Near BMDs Dose Group
nmrrnc tTill b
Dichotomous Hill
_ b
-
-
-
-
-
-
-
-
Gamma
0.93
15,296.46
-0.10
-0.10
0.16
28.47
25.68
22.71
20.60
Log-Logistic
0.92
15,296.49
-0.10
-0.10
0.17
27.80
22.17
22.53
20.20
Weibull
0.98
15,294.49
-0.10
-0.10
0.17
27.60
23.80
22.49
20.44
Logistic
0.59
15,296.40
0.59
0.59
0.79
42.06
32.11
24.42
18.86
Log-Probit
0.94
15,296.43
-0.10
-0.10
0.14
29.59
22.97
23.01
20.40
Probit
0.58
15,296.47
0.61
0.61
0.79
43.34
32.79
24.53
18.75
Quantal Linear
0.52
15,296.83
0.72
0.72
0.79
51.43
36.76
25.04
17.89
Notes: AIC = Akaike information criterion; ALT = alanine transaminase; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level corresponding to a
10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence limit for a 10% response level; BMD5 = dose level corresponding to a
5%) response level; BMDL5 = lower bound on the dose level corresponding to the 95% lower confidence limit for a 5% response level.
a Selected model in bold.
bBMD Computation failed.
E-40
-------
APRIL 2024
Table E-29. Summary of Benchmark Dose Modeling Results for Elevated ALT in Gallo et al. (2012) Using the Adjusted, No
Intercept Median PFOS Serum Concentration
Goodness of Fit Scaled Residual
BMDio BMDLio BMDs BMDLs
D-value AIC DoseGrouP Dose Group Control (ng/mL) (ng/mL) (ng/mL) (ng/mL)
Near BMDio Near BMDs Dose Group
nmrrnc tTill ^
Dichotomous Hill
_b
-
-
-
-
-
-
-
-
Gamma
0.96
15,296.38
-0.08
-0.08
0.12
23.95
18.49
16.91
14.37
Log-Logistic
0.95
15,296.40
-0.08
-0.08
0.13
23.44
16.17
16.78
13.96
Weibull
0.95
15,296.40
-0.08
-0.08
0.13
23.14
16.75
16.73
14.27
Logistic
0.59
15,296.40
0.59
0.59
0.79
38.74
28.66
22.18
16.50
Log-Probit
0.98
15,296.34
-0.06
-0.06
0.09
26.43
17.13
17.48
14.18
Probit
0.58
15,296.47
0.61
0.61
0.79
40.40
29.72
22.58
16.70
Quantal Linear
0.52
15,296.83
0.72
0.72
0.79
51.43
36.75
25.04
17.89
Notes: AIC = Akaike information criterion; ALT = alanine transaminase
; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level corresponding to a 10% response level; BMDLio = lower bound on the dose level
corresponding to the 95% lower confidence limit for a 10% response level; BMD5 = dose level corresponding to a 5% response level; BMDL5 = lower bound on the dose level
corresponding to the 95% lower confidence limit for a 5% response level.
a Selected model in bold.
bBMD Computation failed.
E-41
-------
APRIL 2024
E.1.3.3 Summary of Modeling Results for Liver Toxicity
Table E-30. BMDs and BMDLs in ng/mL for Effect of PFOS on Serum Ln(ALT) in Females
Exposure
Median (25th-
75th Percentiles)
Reported p
(95% CI)
Units
Re-Expressed p
(95% CI)
Ln(IU/L)/(ng/mL)
Exact Percentage, P(0) = 13.0%
Study
SE(P)
PlJL
BMR =
5%
BMR
= 10%
BMD
BMDL
BMD
BMDL
Gallo et
al. (2012)
20.3 (13.7-29.4)
0.02 (0.014, 0.026)
ln(IU/L)/ln(ng/mL)
NA
0.0030612
0.025
95.88
56.79
2,624
799.02
Nian et
al. (2019)
25.7 (18.9-34.9)
0.0401818 (0.00598, 0.0741794)
ln(IU/L)/ln(ng/mL)
0.00158
(0.00023527,
0.00292)
0.0006842
0.00235
44.4
30.69
86.28
51.15
Nian et
al. (2019)
25.7 (18.9-34.9)
0.0401818 (0.00598, 0.0741794)
ln(IU/L)/ln(ng/mL)
NA
0.017397806
0.07
25.93
15.12
134.66
39.58
Notes: ALT = alanine transaminase; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; CI = confidence interval; IU = international
units; NA = not applicable; SE = standard error; TBD = to be determined.
E-42
-------
APRIL 2024
Table E-31 summarizes the PODs resulting from the modeling approaches for increased ALT.
The selected PODs were based on a BMR of 5%, resulting in BMDLs ranging from 15.12 ng/mL
to 56.79 ng/mL, with a selected POD of 15.12 ng/mL.
Table E-31. BMDLs for Effect of PFOS on Serum ALT Using a BMR of 5%
Study Name
BMDL (ng/mL)
Gallo et al. (2012)
56.79
Nian et al. (2019)
15.12
Notes: ALT = alanine transaminase; BMDL = benchmark dose lower limit; BMR = benchmark response.
E.1.4 Modeling Results for Increased Cholesterol
This updated review indicated that there was as association between increases in PFOS and
increases in total cholesterol (TC) in adults. Three medium confidence studies were considered
for POD derivation (Dong et al., 2019; Lin et al., 2019; Steenland et al., 2009). These candidate
studies offer a variety of PFOS exposure measures across various populations. Dong et al. (2019)
investigated an NHANES population (2003-2014), while Steenland et al. (2009) investigated
effects in a high-exposure community (the C8 Health Project study population). Lin et al. (2019)
collected data from prediabetic adults from the Diabetes Prevention Program (DPP) and DPP
Outcomes Study at baseline (1996-1999).
E.1.4.1 Dong et al. (2019)
Using data from NHANES (2003-2014) on 8,948 adults, Dong et al. (2019) calculated a BMD
for PFOS and TC using a hybrid model (Crump, 1995). The cutoff for adverse response (i.e.,
elevated TC) was set at the upper 5th percentile of TC values in the lowest PFOS exposure group
(the actual TC value at this cutoff point was not provided), and the BMR was defined as a 10%
increase in the number of people with TC values above this level. Using this method, Dong et al.
(2019) reported a BMDio and BMDLio of 44.2 ng/mL and 24.1 ng/mL, respectively. Key
variables or other results such as the cutoff point used to define elevated TC or model fit
parameters were not provided.
Although the hybrid approach has several advantages (Crump, 1995), few details were provided
in Dong et al. (2019) on several important aspects of this approach or on other key issues,
including the definition of the unexposed reference group, the distribution of PFOS or TC values
in this group, model fit (e.g., the fit of linear vs. non-linear models), the impact of potential
confounders, or the potential role of reverse causality.
EPA re-analyzed the data using the regression models from the Dong et al. (2019) study,
together with updated NHANES data, applied to a modified hybrid model to develop BMD and
BMDL estimates for various time periods and assumptions. The BMD values for a BMR of 5%
ranged from 15.84 ng/mL for the period 1999-2018, excluding adults taking cholesterol
medications, up to 36.20 ng/mL for the period 2017-2018, for all adults. The BMDL values for a
BMR of 5% ranged from 9.34 ng/mL for the period 1999-2018, excluding adults taking
cholesterol medications, up to 21.35 ng/mL for the period 2017-2018, for all adults. The BMD
values for a BMR of 10% ranged from 35.79 ng/mL for the period 1999-2018, excluding adults
taking cholesterol medications, up to 55.71 ng/mL for the period 2017-2018, for all adults. The
BMDL values for a BMR of 10% ranged from 21.11 ng/mL for the period 1999-2018, excluding
E-43
-------
APRIL 2024
adults taking cholesterol medications, up to 32.86 ng/mL for the period 2017-2018, for all
adults.
An important caveat is that these calculations assume that Dong's regression model is still
applicable, or at least a good approximation, for all the time periods, for all adults and for adults
taking cholesterol medications, and for the recently updated NHANES data.
Dong et al. (2019) reported a regression coefficient P, which is also referred to as /??, of
0.4 mg/dL TC per ng/mL PFOS (95% CI: 0.06, 0.6). From correspondence with the author, EPA
obtained an updated estimated coefficient of 0.35 (95% CI: 0.06, 0.64) mg/dL TC per ng/mL
PFOS, which EPA used for these analyses. The regression model applies to all adults 20 to
80 years old and was adjusted for age, gender, race, poverty income ratio, body mass index,
waist circumference, physical activity level, diabetes status, smoking status, and number of
alcoholic drinks per day. Using a normal approximation, the standard error of the regression
coefficient is estimated as
Upper Limit — Lower Limit 0.64 — 0.06 mo nq .
SE = — = = 0.148 — (—r1
3.92 3.92 dL VmLJ
These analyses were for the periods 1999-2008, 2003-2014, 2003-2018, and 2017-2018,
assuming that regression model coefficient developed for the period 2003-2014 in the Dong et
al. (2019) study can be applied to the alternative NHANES periods. These analyses used the
NHANES-recommended reference method data for TC. EPA used the NHANES PFOS data for
each NHANES period including data adjustments to stored biospecimen data collected in 1999-
2000 and 2013-2014 that were publicly released in April 2022. Alternative analyses were for all
adults ages 20 and over, and for adults ages 20 and over that reported not taking prescribed
cholesterol medications. NHANES survey weights were applied.
EPA estimated the distribution of TC assuming a normal distribution and also estimated the
mean PFOS. The means and standard deviations for each group are shown in Table E-32.
Table E-32. NHANES Mean and Standard Deviation of Total Cholesterol (mg/dL) and
Mean PFOS (ng/mL)
Time Period
1999-
2018
1999-
2018
2003-
2014
2003-
2014
2003-
2018
2003-
2018
2017-
2018
2017-2018
Taking prescribed cholesterol
No
No
No
No
medication?
Mean TC (y)
196.17
197.89
196.36
198.01
194.86
196.96
189.01
192.12
Standard Deviation TC (S)
41.99
41.47
41.84
41.39
41.80
41.28
40.57
39.67
Mean PFOS (X)
13.73
13.73
15.64
15.64
13.21
13.21
6.13
6.13
Notes: NHANES = National Health and Nutrition Examination Survey; TC = total cholesterol.
For the BMD analyses, the response of interest is having elevated serum cholesterol, defined as
greater than or equal to 240 mg/dL. The baseline probability of such a response is P(0), estimated
as 11.5%, for adults aged 20 and older in 2015-2018, as reported by the CDC Health, United
States, 2019 Data Finder (NCHS, 2019).
E-44
-------
APRIL 2024
The selected BMR is an extra risk of either 5% or 10%. The extra risk of high serum cholesterol
is given by the equation
P(cl) - P(0)
Extra Risk = ——
1 - P(0)
where P(d) is the probability of serum cholesterol greater than or equal to 240 mg/dL for a given
PFOS dose d. Thus
P(cl) = {1 — P(0)} x Extra Risk + P(0)
P(cl) = {1 - 0.115} x Extra Risk + 0.115
P(d) = 0.1593 for 5% extra risk and P(d) = 0.2035 for 10% extra risk.
The mean serum cholesterol y for a PFOS dose x is given by the equation
y = mx + b
where m is the slope, P, (from the Dong regression model) and b is the intercept. The intercept b
is the mean serum cholesterol for an unexposed population. For the U.S. population, the mean
TC is_y (tabulated above) and the mean PFOS is x (tabulated above) so the intercept is given by
the equation
b = y — mx
For a given group and dose, the probability of serum cholesterol greater than or equal to
240 mg/dL is
/240 — y\
P(d) = P(TC > 240) = 1 - is the normal cumulative distribution function. Thus, the mean serum cholesterol y is
the solution of the last equation, i.e., y = 240 - S x 1{ I - where "' is the inverse of the
normal cumulative distribution function.
The BMD is the corresponding dose x such that v = mx + b. Thus
y - b
BMD =
m
For the BMDL, the lower bound of the dose is calculated, so that in the last equation, instead of
m the 95th upper limit for P is used, which is given by
P95 = 95th Upper limit for /?=/? + 1.645 x se(/?)
Thus
y — b
BMDL = V-—
(395
E-45
-------
APRIL 2024
Note that P95 is different from the upper bound of the 95% confidence interval, since that
number is the 97.5th percentile. The estimated BMDs and BMDLs are presented in Table E-33:
Table E-33. BMDs and BMDLs for Effect of PFOS on Increased Cholesterol in Dong et al.
(2019)
Time Period
1999-
2018
1999-
2018
2003-
2014
2003-
2014
2003-
2018
2003-
2018
2017-
2018
2017-
2018
Taking prescribed
No
No
No
No
cholesterol medication?
BMR = 5%
BMD (ng/mL)
19.28
15.84
21.08
17.63
23.07
18.54
36.20
29.86
BMDL (ng/mL)
11.37
9.34
12.44
10.40
13.61
10.93
21.35
17.61
BMR = 10%
BMD (ng/mL)
39.48
35.79
41.21
37.54
43.18
38.39
55.71
48.95
BMDL (ng/mL)
23.29
21.11
24.31
22.14
25.47
22.65
32.86
28.87
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response.
Given the potential impact of taking cholesterol medication on the true association between PFOS and
increased TC, the results based on the data excluding such possibility is considered higher confidence. As
illustrated in
Table E-33 there was a decline over time in PFOS levels based on NHANES data, suggesting
that reliance on distributional data based on the most recent NHANES cycle available (2017-
2018) might be more reflective of recent exposure levels. However, given the chronic nature of
both exposure and increased TC development, a higher confidence might be the given to
estimates based on the largest period available (1999-2018).
For increased cholesterol associated with PFOS exposure, the POD is based on the data
Dong et al. (2019) excluding people taking cholesterol medication, the longest period
available, a BMR of 5% and a BMDLs of 9.3 ng/mL.
E.1.4.2 Steenland et al. (2009)
The above hybrid approach was also applied to Steenland et al. (2009) using log-transformed
values. In Table 4, Steenland et al. (2009) reported a linear regression coefficient for change in
ln-transformed TC per ln(PFOS): 0.02660 with a standard deviation of 0.00140. The NHANES
data used in this approached is summarized in Table E-34 and BMD/BMDL values are presented
in Table E-35.
Table E-34. NHANES Mean and Standard Deviation of Ln(TC) (Ln(mg/dL)) and Mean
Ln(PFOS) (Ln(ng/mL))
Time Period
1999-
2018
1999-
2018
2003-
2014
2003-
2014
2003-
2018
2003-
2018
2017-
2018
2017-
2018
Taking prescribed
No
No
No
No
cholesterol medication?
Mean ln(TC) (y)
5.26
5.27
5.26
5.27
5.25
5.26
5.22
5.24
Standard Deviation
0.21
0.21
0.21
0.21
0.21
0.21
0.22
0.21
ln(TC) (S)
Mean ln(PFOS) (X)
2.17
2.17
2.36
2.36
2.14
2.14
1.50
1.50
Notes: NHANES = National Health and Nutrition Examination Survey; TC = total cholesterol.
E-46
-------
APRIL 2024
Table E-35. BMDs and BMDLs for Effect of PFOS on Increased Cholesterol in Steenland
et al. (2009)
Time Period
1999-
2018
1999-
2018
2003-
2014
2003-
2014
2003-
2018
2003-
2018
2017-
2018
2017-
2018
Taking prescribed
No
No
No
No
cholesterol medication?
BMR = 5%
BMD (ng/mL)
14.16
11.58
16.77
13.48
17.21
13.23
26.36
18.88
BMDL (ng/mL)
11.46
9.52
13.39
10.95
13.72
10.77
20.31
14.94
BMR = 10%
BMD (ng/mL)
54.05
43.02
63.79
50.20
66.14
49.34
102.98
69.54
BMDL (ng/mL)
39.33
31.88
45.81
36.75
47.36
36.17
71.18
49.59
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response.
Mean serum TC
EPA also conducted dose-response modeling using mean serum TC reported across PFOS
deciles from Table 3 in Steenland et al. (2009). The associated standard error terms were found
through author correspondence. BMDS 3.3rcl0 was used to fit the dose-response data using all
deciles, no viable models were identified. To further investigate, BMDS 3.3rcl0 was used to fit
the dose-response data in the lowest five deciles and regression coefficients for the mean change
of ln-transformed serum TC (Table 3 in Steenland et al. (2009)), summarized in Table E-36.
BMRs of a change in the mean equal to V2 and 1 SDs from the control mean were chosen. The
BMD modeling results are summarized in Table E-37.
Table E-36. Regression Results for Serum Total Cholesterol by Deciles of Serum PFOS
from Steenland et al. (2009)
Decile
Dose
(ng/mL)
N
Regression Coefficient3
(SD)
1
6.37
4,629
0.00 (0.192)
2
10.60
4,629
0.01 (0.192)
3
13.65
4,629
0.01 (0.192)
4
16.19
4,629
0.03 (0.192)
5
18.79
4,629
0.03 (0.192)
Notes: SD = standard deviation.
a Regression coefficient, change in the natural log of total cholesterol
E-47
-------
APRIL 2024
Table E-37. Summary of Benchmark Dose Modeling Results for Increase Mean Serum Total Cholesterol in Steenland et al.
(2009)
Goodness of Fit Scaled Residual
Model3
p-value
AIC
Dose Group
Near BMDisd
Dose Group
Near
BMDo.ssd
BMDisd
Control Dose (n2/mL)
Group
BMDLisd
(ng/mL)
BMDo.ssd
(ng/mL)
BMDLo.ssd
(ng/mL)
Exponential 3
<0.0001
-10,350.92
0.00
-1.16
-1.54 0.76
0.00
25.38
24.66
Exponential 5
-
-
-
-
-
-
-
-
Hill
-
-
-
-
-
-
-
-
Polynomial Degree 3
0.00
-10,588.86
-0.78
-0.78
0.00 45.95
33.33
31.36
26.15
Polynomial Degree 2
0.00
-10,588.82
-0.71
-
47.85
39.78
-
-
Power
0.00
-10,588.89
-0.75
-0.75
0.02 48.56
47.46
32.31
29.22
Linear
0.01
-10,589.87
-0.23
-0.23
0.51 74.49
62.75
37.24
31.37
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDisd = dose level corresponding to a change in the mean equal to
one standard deviation from the control mean; BMDLisd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to one
standard deviation from the control mean. BMDo.5sd = dose level corresponding to a change in the mean equal to 0.5 standard deviations from the control mean;
BMDLo.5sd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 0.5 standard deviation from the control mean.
aNo viable models. No model was selected.
bBMD Computation failed
E-48
-------
APRIL 2024
E. 1.4.2.1 Elevated TC
In additional to modeling the regression coefficients, dichotomous models using BMDS 3.3rcl0
were used to fit the ORs of elevated TC from Steenland et al. (2009) as shown in Table E-38.
Sample sizes, mean PFOS concentrations in each quartile and prevalence of elevated TC in each
exposure group were obtained from Dr. Kyle Steenland. A BMR of 10% and 5% extra risk were
both included. The BMD modeling results are summarized in Table E-39.
Table E-38. Odds Ratios for Elevated Serum Total Cholesterol by Quartiles of Serum
PFOS from Steenland et al. (2009)
Quartile
Dose
(ng/mL)
N
Incidence
OR
95% CI
1
6.6
11,534
1,479
1
Ref
2
16.4
11,587
1,634
1.14
1.05, 1.23
3
23.8
11,441
1,795
1.28
1.19, 1.39
4
50.55
11,400
2,158
1.51
1.40, 1.64
Notes'. CI = confidence interval; OR = odds ratio.
E-49
-------
APRIL 2024
Table E-39. Summary of Benchmark Dose Modeling Results for Elevated Total Cholesterol in Steenland et al. (2009)
Goodness of Fit Scaled Residual
Model"
p-value
AIC
Dose Group
Near BMD10
Dose Group
Near BMDs
Control Dose
Group
BMD10
(ng/mL)
BMDLio
(ng/mL)
BMDs
(ng/mL)
BMDLs
(ng/mL)
Dichotomous Hill
_b
-
3.56 x 10-6
-
-
-
-
31.08
26.59
Gamma
0.53
39,272.57
-0.28
-0.28
-0.14
63.00
55.89
30.67
27.21
Log-Logistic
0.57
39,272.40
-0.24
-0.24
-0.05
63.18
55.91
29.93
26.39
Multistage Degree 3
0.01
39,282.00
-0.58
-0.58
-1.57
62.48
0.00
40.96
40.29
Multistage Degree 2
0.53
39,272.57
-0.28
-0.28
-0.14
63.00
55.88
30.67
27.20
Multistage Degree 1
0.53
39,272.57
-0.28
-0.28
-0.14
63.00
55.89
30.67
27.20
Weibull
0.53
39,272.57
-0.28
-0.28
-0.14
63.00
55.89
30.67
27.21
Logistic
0.27
39,274.11
-0.42
-0.42
-0.62
62.30
56.70
34.49
31.47
Log-Probit
0.35
39,274.11
-0.10
-0.10
0.16
66.02
57.02
29.71
14.27
Probit
0.31
39,273.81
-0.40
-0.40
-0.55
62.43
56.61
33.93
30.84
Quantal Linear
0.53
39,272.57
-0.28
-0.28
-0.14
63.00
55.89
30.67
27.21
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level corresponding to a 10% response level;
BMDLio = lower bound on the dose level corresponding to the 95% lower confidence limit for a 10% response level; BMD5 = dose level corresponding to a 5% response level;
BMDL5 = lower bound on the dose level corresponding to the 95% lower confidence limit for a 5% response level.
a Selected model in bold.
bBMD Computation failed
E-50
-------
APRIL 2024
Given the potential impact of taking cholesterol medication on the true association between
PFOS and increased TC, the results based on the data excluding such possibility is considered
higher confidence. As illustrated in Table E-34 there was a dramatic decline over time in PFOS
levels based on NHANES data, suggesting that reliance on distributional data based on the most
recent NHANES cycle available (2017-2018) might be more reflective or current impacts.
However, given the chronic nature of both exposure and increased TC development, a higher
confidence might be the given to estimates based on the largest period available (1999-2018).
For increased cholesterol associated with PFOS exposure, the POD is based on the data
from Steenland et al. (2009) excluding people taking cholesterol medication, the longest
period available, a BMR of 5% and a BMDLs of 9.52 ng/mL. A comparison BMDL of
14.9 ng/mL based on the most recent period available supports the selected POD.
E.1.4.3 Lin et al. (2019)
Lin et al. (2019) collected data from prediabetic adults from the DPP and DPP Outcomes Study
at baseline (1996-1999). This study included 888 prediabetic adults who were recruited from 27
medical centers in the United States. Median PFOS levels at baseline were comparable to those
from NHANES 1999-2000, 27.2 (25th, 75th percentiles: 18.0 ng/mL, 40.4 ng/mL). The study
presented both cross-sectional and prospective analyses. The cross-sectional analyses evaluated
associations between baseline PFAS and baseline lipid levels. The prospective analysis evaluated
whether baseline PFAS levels predicted higher risk of incident hypercholesterolemia and
hypertriglyceridemia, but in the placebo and the lifestyle intervention groups, separately.
EPA conducted dose-response modeling using mean serum TC reported across PFOS quartiles
from Table S5 in Lin et al. (2019). For its POD calculations, EPA used the results from the
cross-sectional analysis because they were presented in a format that was more amendable to
dose-response analysis.
BMDS 3.3rcl0 was used to fit the dose-response data for the adjusted mean difference in lipid
levels (mg/dL) per quartile of baseline plasma PFOS concentrations (ng/mL), summarized in
Table E-40. BMRs of a change in the mean equal to 0.5 SD and 1 SD from the control mean
were used. The BMD modeling results are summarized in Table E-41. However, the PODs
derived from this study are considered lower confidence since they are based on a poorly fir
PFOS association (adjusted mean difference = 2.53, 95% CI: -0.10, 5.16).
Table E-40. Adjusted Mean Differences in Serum Total Cholesterol by Quartiles of Serum
PFOS (ng/mL) from Lin et al. (2019)
Dose
(ng/mL)
N
Adjusted Mean
Difference TC (95% CI)
(mg/dL)
Mean TCa'b
12.8
212
Ref
0.00 ±35.48
21.7
224
1.13 (-5.50,7.77)
1.13 ±35.33
32.7
230
5.05 (-1.55, 11.66)
5.05 ±35.39
53
222
5.13 (-1.58, 11.86)
5.13 ±35.70
Notes: CI = confidence interval; TC = total cholesterol.
a Mean ± standard deviation.
b Adjusted mean difference in lipid levels (mg/dL) per quartile of baseline plasma PFOS concentration (ng/mL)
E-51
-------
APRIL 2024
Table E-41. Summary of Benchmark Dose Modeling Results for Increase Mean Serum Total Cholesterol Lin et al. (2019)
Goodness of Fit Scaled Residual
BMDisd BMDLisd BMDo.ssd BMDLo.ssd
p-value AIC 1VD()SC,^I"('U|) lvD()SCD?ir"u|) Coi^ro1 Dose (ng/mL) (ng/mL) (ng/mL) (ng/mL)
Near BMDisd Near BMDo.ssd Group
Exponential 3
0.23
8,863.69
-0.21
-0.21
-0.60
108.34
61.19
88.53
57.34
Exponential 5
_b
-
-
-
-
-
-
-
-
Hill
-
-
-
-
-
-
-
-
-
Polynomial Degree 3
0.65
8,861.12
-0.35
-0.35
-0.21
261.96
86.09
130.98
66.43
Polynomial Degree 2
0.65
8,861.12
-0.34
-
-
262.61
100.07
-
-
Power
0.65
8,861.12
-0.34
-0.34
-0.21
262.62
58.47
131.31
66.54
Linear
0.65
8,861.12
-0.34
-0.34
-0.21
262.62
133.07
131.31
66.54
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDisd = dose level corresponding to a change in the mean equal to
one standard deviation from the control mean; BMDLisd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to one
standard deviation from the control mean. BMDo.ssd = dose level corresponding to a change in the mean equal to 0.5 standard deviations from the control mean;
BMDLo.ssd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 0.5 standard deviation from the control mean.
a Selected model in bold.
bBMD Computation failed
E.1.4.4 Summary of Modeling Results for Increased Cholesterol
Table E-42. BMDs and BMDLs in ng/mL for Effect of PFOS on Serum Total Cholesterol
Reported p
Re-Expressed P (95% CI)
Ln(mg/dL)/(ng/mL)
Exact Percentage, P(0) =
11.5%
Study
Exposure
Mean (SD)
(95% CI)
SE(P)
PlJL
BMR
= 5% BMR = 10%
Units
BMD
BMDL BMD
BMDL
Steenland et al.
(2009)
22.4 (14.8)
0.0266 (0.0243, 0.0289)
ln(mg/dL)/ln(ng/mL)
0.00137(0.00125,0.00149)
0.0000605
0.00147
7.58
7.07 33.04
30.08
Steenland et al.
(2009)
22.4 (14.8)
0.0266 (0.0243, 0.0289)
ln(mg/dL)/ln(ng/mL)
NA
3.02415E-08
0.03
11.58
9.52 43.02
31.88
Dong et al.
(2019)
15.6(17.8)
0.35 (95% CI: 0.06,
0.64) mg/dL/ng/mL
NA
0.15
0.59
15.84
9.34 38.39
22.65
Notes: BMD = benchmark dose; BMDL = benchmark dose lower limit; BMR = benchmark response; CI = confidence interval; NA = not applicable; SD =standard deviation;
SE = standard error.
E-52
-------
APRIL 2024
Table E-43 summarizes the PODs resulting from the modeling approaches for increased
cholesterol. The selected and comparison PODs were based on a BMR of 5%, resulting in
BMDLs ranging from 9.3 ng/mL to 66.5 ng/mL with the selected POD of 9.35 also representing
the median of the BMDLs. The comparison POD based on the data from Lin et al. (2019) is
considered low confidence because it is based on a poorly fit PFOS regression parameter.
Table E-43. BMDLs for Effect of PFOS on Serum Total Cholesterol Using a BMR of 5%
Study Name Effect BMDL (ng/mL)
Dong et al. (2019) Exclude those prescribed cholesterol medication, 1999-2018 9.34
Steenland et al. (2009) Exclude those prescribed cholesterol medication 9.52
Lin et al. (2019) Diabetic adults 66.5
Notes: BMDL = benchmark dose lower limit; BMR = benchmark response
E.2 Toxicology Studies
E.2.1 Butenhoff et al. (2012)/Thomford (2002)
EPA conducted dose-response modeling of the Butenhoff et al. (2012)/Thomford (2002) study
using the BMDS 3.2 program. This study addresses incidence of adenomas and/or carcinomas in
the liver and pancreas in male rats and the liver and thyroid in female rats, and individual cell
necrosis in the liver in female Sprague-Dawley Crl:CD(SD)IGS BR rats.
E.2.1.1 Hepatocellular Adenomas in Males
Increased incidence of hepatocellular adenomas was observed in male rats. Dichotomous models
were used to fit dose-response data. Multistage models were used consistent with the
longstanding practice of EPA to prefer multistage models to fit tumor dose-response data and a
BMR of 10% extra risk was chosen per EPA's Benchmark Dose Technical Guidance (U.S. EPA,
2012). The dose and response data used for the modeling are listed in Table E-44. The area under
the curve (AUC) normalized per day (AUCavg), equivalent to the mean serum concentration over
the duration of the study, was selected as the dose metric for modeling cancer endpoints (see the
Toxicity Assessment, (U.S. EPA, 2024)). BMD analysis was conducted using both the number of
animals at the start of the study and the number of animals alive at the time of first tumor.
Table E-44. Dose-Response Modeling Data for Hepatocellular Adenomas in Male Rats
Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002)
Administered Dose
(ppm)
Internal Dose
(mg/L)
Number per Group
at Start of Study
Number per Group
at Time of First
Tumor11
Incidence
0
0.0
50
41
0
0.5
1.4
50
42
3
2
5.9
50
47
3
5
14.3
50
44
1
20
57.8
50
43
7
E-53
-------
APRIL 2024
Notes:
a The time of first occurrence of this tumor was day 512 in males.
BMD modeling results for hepatocellular adenomas following exposure to PFOS for the number
of animals at the start of the study and the number of animals alive at the time of first tumor are
summarized in Table E-45 and Figure E-3 and Figure E-4. The best fitting model was the
Multistage Degree 4 model based on adequate p-values (greater than 0.1), the BMDLs were
sufficiently close (less than threefold difference) among adequately fitted models, and the
Multistage Degree 4 model had the lowest Akaike information criterion (AIC). The lower bound
on the dose level corresponding to the 95% lower confidence limit for a 10% response level
BMDLio from the selected Multistage Degree 4 model for the number of animals at the start of
the study is 29.3 mg/L and for the number of animals alive at the time of first tumor is
25.6 mg/L. The number of animals alive at the time of first tumor ensures the potency is not
underestimated by mortality of animals prior to tumor occurrence. The relatively small
difference in the two BMDLio values supports using these values and the selected value is based
on the number of animals alive at the time of first tumor, 25.6 mg/L.
Table E-45. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas in Male Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford,
2002)
Goodness of
Fit
Scaled Residual
Model3
p- Dose Group Control
value Near BMD Dose Group
BMDio BMDLio
(mg/L) (mg/L)
Basis for Model
Selection
Multistage 0.260 105.2
Degree 4
Multistage
Degree 3
Animals at
the start of Multistage
the study Degree 2
Multistage
Degree 1
0.254 105.2
0.235 105.4
0.192 105.7
0.004
0.017
0.065
0.204
-1.35
-1.34
-1.32
-1.19
56.6
56.3
55.9
54.5
29.3
29.1
28.5
27.6
EPA selected the
Multistage Degree 4
model. All multistage
models had adequate
fit (p-values greater
than 0.1), the BMDLs
were sufficiently
close (less than
threefold difference),
and the Multistage
Degree 4 model had
the lowest AIC.
Multistage
0.281
100.9
0.005
-1.31
54.2
25.6
EPA selected the
Degree 4
Multistage Degree 4
model. All multistage
Multistage
0.275
101.0
0.018
-1.31
53.2
25.4
models had adequate
Animals
Degree 3
fit (p-values greater
alive at the
Multistage
0.252
101.2
0.071
-1.29
51.4
24.9
than 0.1), the BMDLs
time of
Degree 2
were sufficiently
first tumor
close (less than
Multistage
0.196
101.6
0.238
-1.16
46.8
23.7
threefold difference),
Degree 1
and the Multistage
Degree 4 model had
the lowest AIC.
Notes: AIC = Akaike information criterion; BMD =
corresponding to a 10% response level; BMDLio:
limit for a 10% response level.
benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
lower bound on the dose level corresponding to the 95% lower confidence
E-54
-------
APRIL 2024
a Selected model in bold.
1
0.9
0.8
0.7
8 °-6
I 0.5
(u
ce 0.4
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-3. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Male Rats Following Exposure to PFOS,
for Number of Animals Per Group at Start of Study (Butenhoff et al., 2012; Thomford,
2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
1
0.9
0.8
0.7
8 °-6
I 0.5
(u
ce 0.4
0.3
0.1
r-*
1
J
'LA
T~
10
20
30
Dose
40
50
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-4. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas in Male Rats Following Exposure to PFOS,
for Number of Animals Per Group at Time of First Tumor
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E-55
-------
APRIL 2024
E.2.1.2 Pancreas Islet Cell Carcinomas in Males
Increased incidence of islet cell carcinomas was observed in male rats. Dichotomous models
were used to fit dose-response data. A BMR of 10% extra risk per EPA's Benchmark Dose
Technical Guidance (U.S. EPA, 2012). The dose and response data used for the modeling are
listed in Table E-46. The AUCavg, equivalent to the mean serum concentration over the duration
of the study, was selected as the dose metric for modeling cancer endpoints (see the Toxicity
Assessment, (U.S. EPA, 2024)). BMD analysis was conducted using both the number of animals
at the start of the study and the number of animals alive at the time of first tumor.
Table E-46. Dose-Response Modeling Data for Incidence of Islet Cell Carcinomas in Male
Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002)
Administered Dose
(ppm)
Internal Dose
(mg/L)
Number per
Group at Start of
Study
Number per
Group at Time of
First Tumor3
Incidence
0
0.0
50
38
1
0.5
1.4
50
41
2
2
5.9
50
44
2
5
14.3
50
44
5
20
57.8
50
40
5
Notes:
a The time of first occurrence of this tumor was day 542 in males.
The BMD modeling results for incidence of islet cell carcinomas following exposure to PFOS
for the number of animals at the start of the study and the number of animals alive at the time of
first tumor are summarized in Table E-47 and Figure E-5 and Figure E-6. The best fitting model
was the Multistage Degree 1 model based on adequate p-values (greater than 0.1), the BMDLs
were sufficiently close (less than threefold difference) among adequately fitted models, and the
higher degree Multistage models estimated parameters at the zero boundary and reduced to the
Multistage Degree 1 model. The BMDLio from the selected Multistage Degree 1 model for the
number of animals at the start of the study is 29.7 mg/L and for the number of animals alive at
the time of first tumor is 26.1 mg/L. The number of animals alive at the time of first tumor
ensures the potency is not underestimated by mortality of animals prior to tumor occurrence. The
relatively small difference in the two BMDLio values supports using these values and the
selected value is based on the number of animals alive at the time of first tumor, 26.1 mg/L.
Table E-47. Summary of Benchmark Dose Modeling Results for Incidence of Islet Cell
Carcinomas in Male Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford,
2002)
Goodness of
Fit
Scaled Residual
Model"
p- Dose Group Control
value Near BMD Dose Group
BMDio BMDLio
(mg/L) (mg/L)
Basis for Model
Selection
Multistage 0.526 114.5 -0.434 -0.633 67.6 29.7 EPA selected the
Degree 4 Multistage Degree 1
E-56
-------
APRIL 2024
Goodness of
Fit
Scaled Residual
Model"
BMDio
BMDLio
Basis for Model
P-
AIC
Dose Group
Control
(mg/L)
(mg/L)
Selection
value
Near BMD
Dose Group
Animals at
Multistage
0.526
114.5
-0.434
-0.633
67.6
29.7
model. All
the start of
Degree 3
multistage models
the study
had adequate fit (p-
Multistage
0.526
114.5
-0.434
-0.633
67.6
29.7
values greater than
Degree 2
0.1), the BMDLs
Multistage
were sufficiently
0.526
114.5
-0.434
-0.633
67.6
29.7
close (less than
Degree 1
threefold difference),
and higher degree
models reduced to
the Multistage
Degree 1 model.
Multistage
0.554
111.2
-0.417
-0.590
58.5
26.1
EPA selected the
Degree 4
Multistage Degree 1
model. All
Multistage
0.554
111.2
-0.417
-0.590
58.5
26.1
multistage models
Degree 3
had adequate fit (p-
Animals
Multistage
values greater than
alive at the
0.554
111.2
-0.417
-0.590
58.5
26.1
0.1), the BMDLs
time of
Degree 2
were sufficiently
first tumor
Multistage
Degree 1
0.554
111.2
-0.417
-0.590
58.5
26.1
close (less than
threefold difference),
and higher degree
models reduced to
the Multistage
Degree 1 model.
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
corresponding to a 10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence
limit for a 10% response level.
a Selected model in bold.
E-57
-------
APRIL 2024
1
0.9
0.8
0.7
d) 0.6
LO
I 0.5
(u
0.4
0.3
0.2
i
-
'J
10
20
30
Dose
40
50
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-5. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Incidence of Islet Cell Carcinomas in Male Rats Following Exposure to
PFOS, for Number of Animals Per Group at Start of Study (Butenhoff et al., 2012;
Thomford, 2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
1
0.9
0.8
0.7
-------
APRIL 2024
E.2.1.3 Pancreas Combined Islet Cell Adenomas and Carcinomas in
Males
Increased incidence of combined islet cell adenomas and carcinomas was observed in male rats.
Dichotomous models were used to fit dose-response data. A BMR of 10% extra risk per EPA's
Benchmark Dose Technical Guidance (U.S. EPA, 2012). The dose and response data used for the
modeling are listed in Table E-48. The AUCavg, equivalent to the mean serum concentration over
the duration of the study, was selected as the dose metric for modeling cancer endpoints (see
Toxicity Assessment, (U.S. EPA, 2024)). BMD analysis was conducted using both the number of
animals at the start of the study and the number of animals alive at the time of first tumor.
Table E-48. Dose-Response Modeling Data for Combined Incidence of Islet Cell Adenomas
and Carcinomas in Male Rats Following Exposure to PFOS (Butenhoff et al., 2012;
Thomford, 2002)
Administered Dose
(ppm)
Internal Dose
(mg/L)
Number per Group
at Start of Study
Number per Group
at Time of First
Tumor3
Incidence
0
0.0
50
44
5
0.5
1.4
50
45
5
2
5.9
50
48
6
5
14.3
50
46
8
20
57.8
50
44
9
Notes:
a The time of first occurrence of this tumor was day 465 in males.
The BMD modeling results for combined incidence of islet cell adenomas and carcinomas
following exposure to PFOS for the number of animals at the start of the study and the number of
animals alive at the time of first tumor are summarized in Table E-49 and Figure E-7 and Figure
E-8. The best fitting model was the Multistage Degree 1 model based on adequate p-values
(greater than 0.1), the benchmark dose lower limits (BMDLs) were sufficiently close (less than
threefold difference) among adequately fitted models, and the higher degree Multistage models
estimated parameters at the zero boundary and reduced to the Multistage Degree 1 model. The
BMDLio from the selected Multistage Degree 1 model for the number of animals at the start of
the study is 25.1 mg/L and for the number of animals alive at the time of first tumor is
21.7 mg/L. The number of animals alive at the time of first tumor ensures the potency is not
underestimated by mortality of animals prior to tumor occurrence. The relatively small
difference in the two BMDLio values supports using these values and the selected value is based
on the number of animals alive at the time of first tumor, 21.7 mg/L.
E-59
-------
APRIL 2024
Table E-49. Summary of Benchmark Dose Modeling Results for Combined Incidence of
Islet Cell Adenomas and Carcinomas in Male Rats Following Exposure to PFOS
(Butenhoff et al., 2012; Thomford, 2002)
Model"
Goodness of Fit
Scaled Residual
p-value AIC
Dose Group Control Dose
Near BMD Group
BMDio
(mg/L)
BMDLio
(mg/L)
Basis for Model Selection
Multistage
0.909
197.34
-0.191
-0.214
63.8
25.1
EPA selected the
Degree 4
Multistage Degree 1
Multistage
0.909
197.34
-0.191
-0.214
63.8
25.1
model. All multistage
models had adequate fit
Degree 3
(p-values greater than 0.1),
Multistage
Degree 2
0.909
197.34
-0.191
-0.214
63.8
25.1
the BMDLs were
sufficiently close (less
than threefold difference),
Multistage
0.909
197.34
-0.191
-0.214
63.8
25.1
and higher degree models
Degree 1
reduced to the Multistage
Degree 1 model.
Multistage
0.938
190.0
-0.162
-0.130
53.6
21.7
EPA selected the
Degree 4
Multistage Degree 1
Multistage
0.938
190.0
-0.162
-0.130
53.6
21.7
model. All multistage
models had adequate fit
Degree 3
(p-values greater than 0.1),
Multistage
Degree 2
0.938
190.0
-0.162
-0.130
53.6
21.7
the BMDLs were
sufficiently close (less
than threefold difference),
Multistage
0.938
190.0
-0.162
-0.130
53.6
21.7
and higher degree models
Degree 1
reduced to the Multistage
Degree 1 model.
Notes: AIC = Akaike information criterion; BMD =
corresponding to a 10% response level; BMDLio =
limit for a 10% response level.
a Selected model in bold.
benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
lower bound on the dose level corresponding to the 95% lower confidence
E-60
-------
APRIL 2024
1
0.9
0.8
0.7
8 0.6
I 0.5
(u
ce 0.4
10
20
30
Dose
40
50
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-7. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Combined Incidence of Islet Cell Adenomas and Carcinomas in Male
Rats Following Exposure to PFOS, for Number of Animals Per Group at Start of Study
(Butenhoff et al., 2012; Thomford, 2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
1
0.9
0.8
0.7
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-8. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Combined Incidence of Islet Cell Adenomas and Carcinomas in Male
Rats Following Exposure to PFOS, for Number of Animals Per Group at Time of First
Tumor (Butenhoff et al., 2012; Thomford, 2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E-61
-------
APRIL 2024
E.2.1.4 Hepatocellular Adenomas in Females
Increased incidence of hepatocellular adenomas was observed in female rats. Dichotomous
models were used to fit dose-response data. A BMR of 10% extra risk per EPA's Benchmark
Dose Technical Guidance (U.S. EPA, 2012). The doses and response data used for the modeling
are listed in Table E-50. The AUCavg, equivalent to the mean serum concentration over the
duration of the study, was selected as the dose metric for modeling cancer endpoints (See
Toxicity Assessment, (U.S. EPA, 2024)). BMD analysis was conducted using both the number of
animals at the start of the study and the number of animals alive at the time of first tumor.
Table E-50. Dose-Response Modeling Data for Hepatocellular Adenomas in Female Rats
Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002)
Administered Dose
(ppm)
Internal Dose
(mg/L)
Number per
Group at Start of
Study
Number per
Group at Time of
First Tumor3
Incidence
0
0.0
50
28
0
0.5
1.6
50
26
1
2
6.6
49
15
1
5
16.1
50
28
1
20
65.2
50
31
5
Notes:
a The time of first occurrence of this tumor was day 653 in females.
The BMD modeling results for hepatocellular adenomas following exposure to PFOS for the
number of animals at the start of the study and the number of animals alive at the time of first
tumor are summarized in Table E-51 and Figure E-9 and Figure E-10. The best fitting model was
the Multistage Degree 1 model based on adequate p-values (greater than 0.1), the BMDLs) were
sufficiently close (less than threefold difference) among adequately fitted models, and the
Multistage Degree 1 model had the lowest AIC. The BMDLio from the selected Multistage
Degree 1 model for the number of animals at the start of the study is 37.2 mg/L and for the
number of animals alive at the time of first tumor is 21.8 mg/L. The number of animals alive at
the time of first tumor ensures the potency is not underestimated by mortality of animals prior to
tumor occurrence. The relatively small difference in the two BMDLio values supports using
these values and the selected value is based on the number of animals alive at the time of first
tumor, 21.8 mg/L.
Table E-51. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas in Female Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford,
2002)
Goodness of
Fit
Scaled Residual
Model3
BMD io BMDLio
Dose Control (mg/L) (mg/L)
p-value AIC Group Dose
Near BMD Group
Basis for Model Selection
Multistage 0.601 69.2 0.00105
Degree 4b
-0.668 68.3 37.4 EPA selected the
Multistage Degree 1
E-62
-------
APRIL 2024
Goodness of
Fit
Scaled Residual
Model"
p-value AIC
Dose
Group
Near BMD
Control
Dose
Group
BMD 10 BMDLio
(mg/L) (mg/L)
Basis for Model Selection
Animals at Multistage
the start of Degree 3b
0.598 69.3
the study
Multistage
Degree 2
Multistage
Degree 1
0.586
0.761
69.3
67.3
0.00722
0.02918
0.08232
-0.665
-0.655
-0.608
69.0
70.5
73.0
37.4
37.3
37.2
model. All multistage
models had adequate fit
(p-values greater than 0.1),
the BMDLs were
sufficiently close (less than
threefold difference), the
Multistage Degree 1 model
had the lowest AIC.
Multistage
0.449
59.8
0.0024
-0.719
46.7
21.8
EPA selected the
Degree 4
Multistage Degree 1
model. All multistage
Multistage
0.447
59.8
0.0094
-0.713
45.4
21.8
models had adequate fit
Degree 3
(p-values greater than 0.1),
Animals
the BMDLs were
alive at the
Multistage
0.654
57.8
0.0228
-0.701
43.9
21.8
sufficiently close (less than
time of first
Degree 2°
threefold difference), and
tumor
the Multistage Degree 1
Multistage
Degree 1
0.654
57.8
0.0228
-0.701
43.9
21.8
model had the lowest AIC
(the Degree 2 model
estimated parameters at the
zero boundary and reduced
to the Degree 1 model).
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
corresponding to a 10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence
limit for a 10% response level.
a Selected model in bold.
b Degree 3 and 4 models estimated parameters at the zero boundary and reduced to the Multistage Degree 2 model.
c Degree 2 model estimated parameters at the zero boundary and reduced to the Multistage Degree 1 model.
E-63
-------
APRIL 2024
-------
APRIL 2024
E.2.1.5 Hepatocellular Combined Adenomas and Carcinomas in Females
Increased incidence of hepatocellular adenomas and carcinomas was observed in female rats.
Dichotomous models were used to fit dose-response data. A BMR of 10% extra risk per EPA's
Benchmark Dose Technical Guidance (U.S. EPA, 2012). The dose and response data used for the
modeling are listed in Table E-52. The AUCavg, equivalent to the mean serum concentration over
the duration of the study, was selected as the dose metric for modeling cancer endpoints (See
Toxicity Assessment, (U.S. EPA, 2024)). BMD analysis was conducted using both the number of
animals at the start of the study and the number of animals alive at the time of first tumor.
Table E-52. Dose-Response Modeling Data for Hepatocellular Adenomas and Carcinomas
in Female Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002)
Administered Dose
(ppm)
Internal Dose
(mg/L)
Number per Group
at Start of Study
Number per Group at
Time of First Tumor3
Incidence
0
0.0
50
28
0
0.5
1.6
50
29
1
2
6.6
49
16
1
5
16.1
50
31
1
20
65.2
50
32
6
Notes:
aThe time of first occurrence of this tumor was day 653 in females.
The BMD modeling results for hepatocellular adenomas following exposure to PFOS for the
number of animals at the start of the study and the number of animals alive at the time of first
tumor are summarized in Table E-53 and Figure E-l 1 and Figure E-12. The best fitting model
was the Multistage Degree 1 model based on adequate p-values (greater than 0.1), the BMDLs
were sufficiently close (less than threefold difference) among adequately fitted models, and the
Multistage Degree 1 model had the lowest AIC. The BMDLio from the selected Multistage
Degree 1 model for the number of animals at the start of the study is 32.7 mg/L and for the
number of animals alive at the time of first tumor is 19.8 mg/L. The number of animals alive at
the time of first tumor ensures the potency is not underestimated by mortality of animals prior to
tumor occurrence. The relatively small difference in the two BMDLio values supports using
these values and the selected value is based on the number of animals alive at the time of first
tumor, 19.8 mg/L.
Table E-53. Summary of Benchmark Dose Modeling Results for Data for Hepatocellular
Adenomas and Carcinomas in Female Rats Following Exposure to PFOS (Butenhoff et al.,
2012; Thomford, 2002)
Goodness of
Fit
Scaled Residual
Model"
P-
value
AIC
Dose
Group
Near BMD
Control
Dose
Group
BMD io
(mg/L)
BMDLi
(mg/L)
Basis for Model
Selection
Multistage 0.600 73.4 0.0021 -0.668 61.8 33.2 EPA selected the
Degree 4 Multistage Degree 1
Multistage 0.597 73.4 0.0081 -0.667 61.2 33.2 model. All multistage
Degree 3 models had adequate fit
Animals at
the start of
the study
E-65
-------
APRIL 2024
Goodness of
Fit
Scaled Residual
Model3
P-
value
AIC
Dose
Group
Near BMD
Control
Dose
Group
BMD 10
(mg/L)
BMDLi
(mg/L)
Basis for Model
Selection
Multistage
Degree 2
Multistage
Degree 1
0.581 73.5
0.723 71.6
0.0331
0.1462
-0.663
-0.565
60.6
60.3
33.0
32.7
(p-values greater than
0.1), the BMDLs were
sufficiently close (less
than threefold difference),
the Multistage Degree 1
model had the lowest AIC.
Animals
Multistage
Degree 4
Multistage
Degree 3
alive at the Multistage
time of Degree 2b
first tumor Multistage
Degree 1
0.466 63.£
0.461 63.£
0.449 63.£
0.643 61.8
0.0029
0.0109
0.0415
-0.613
-0.716
-0.711
-0.694
-0.630
47.5
45.2
41.7
37.2
20.0
20.0
19.9
19.8
EPA selected the
Multistage Degree 1
model. All multistage
models had adequate fit
(p-values greater than
0.1), the BMDLs were
sufficiently close (less
than threefold difference),
and the Multistage Degree
1 model had the lowest
AIC.
benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
lower bound on the dose level corresponding to the 95% lower confidence
Notes: AIC = Akaike information criterion; BMD
corresponding to a 10% response level; BMDLio
limit for a 10% response level.
a Selected model in bold.
b Degree 2 model estimated parameters at the zero boundary and reduced to the Multistage Degree 1 model.
1
0.9
0.8
0.7
® 0.6
m
I 0.5
w
W
cc 0,4
0,3
0,2
0,1
o C
0
10
20
30
Dose
• Estimated Probability
• Response at BMD
Linear Extrapolation
O Data
BMD
BMDL
Figure E-ll. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas and Carcinomas in Female Rats Following
Exposure to PFOS, for Number of Animals Per Group at Start of Study (Butenhoff et al.,
2012; Thomford, 2002)
E-66
-------
APRIL 2024
BMD = benchmark dose; BMDL = benchmark dose lower limit.
Estimated Probability
Response at BMD
— — Linear Extrapolation
O Data
BMD
BMDL
Figure E-12. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 4 Model for Hepatocellular Adenomas and Carcinomas in Female Rats Following
Exposure to PFOS, for Number of Animals Per Group at Time of First Tumor (Butenhoff
et al., 2012; Thomford, 2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.1.6 Individual Cell Necrosis in the Liver in Females
Increased incidence of individual cell necrosis in the liver was observed in female Sprague-
Dawley Crl:CD(SD)IGS BR rats. Dichotomous models were used to fit dose-response data. A
BMR of 10% extra risk was chosen per EPA's Benchmark Dose Technical Guidance (U.S. EPA,
2012). The doses and response data used for the modeling are listed in Table E-54. As described
in the Toxicity Assessment (U.S. EPA, 2024), the average concentration over the final week of
study Ciastzavg, was selected for all non-developmental studies to provide a consistent internal
dose for use across chronic and subchronic study designs where steady state may or may not
have been reached and to allow extrapolation to the human pharmacokinetic (PK) model.
Table E-54. Dose-Response Modeling Data for Individual Cell Necrosis in the Liver in
Female Sprague-Dawley Crl:CD(SD)IGS BR Rats Following Exposure to PFOS (Butenhoff
et al., 2012; Thomford, 2002)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Incidence
0
0
50
3
0.029
1.8
50
4
0.120
7.4
50
4
0.299
18.0
50
5
1.251
72.5
50
9
l
0.9
0.8
0.7
-------
APRIL 2024
BMD modeling results for individual cell necrosis in the liver are summarized in Table E-55 and
Figure E-13. The Log-Logistic model was selected based on adequate p-values (greater than 0.1)
and had the lowest AIC among adequately fitting with BMD/BMDL ratios less than 3. The
BMDLio from the selected Log-Logistic model is 27.0 mg/L.
Table E-55. Summary of Benchmark Dose Modeling Results for Individual Cell Necrosis in
the Liver in Female Sprague-Dawley Crl:CD(SD)IGS BR Rats Following Exposure to
PFOS (Butenhoff et al., 2012; Thomford, 2002)
Goodness of Fit Scaled Residual
Model"
BMDio
BMDLio
Basis for Model
p-value
AIC
Dose Group
Control Dose
(mg/L)
(mg/L)
Selection
Near BMD
Group
Dichotomous
0.947
164.2
0.003
-0.201
57.1
9.4
EPA selected the
Hill
Log-Logistic
Gamma
0.990
162.2
-0.024
-0.239
59.2
29.0
model. All models
Log-Logistic
0.990
162.2
-0.017
-0.226
58.5
27.0
had adequate fit (p-
Multistage
Degree 4
0.990
162.2
-0.024
-0.239
59.2
29.0
values greater than
0.1). The
Dichotomous Hill
Multistage
Degree 3
0.990
162.2
-0.024
-0.239
59.2
29.0
and Log-Probit
were the only
Multistage
0.990
162.2
-0.024
-0.239
59.2
29.0
models that did not
Degree 2
have BMD/BMDL
Multistage
0.990
162.2
-0.024
-0.239
59.2
29.0
ratio <3. Of the
Degree 1
remaining models,
Weibull
0.990
162.2
-0.024
-0.239
59.2
29.0
the Log-Logistic
Logistic
0.981
162.3
-0.040
-0.334
64.2
41.8
model had the
Log-Probit
0.938
164.2
0.022
-0.208
57.0
0.6
lowest AIC.
Probit
0.983
162.3
-0.041
-0.322
63.5
39.9
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
corresponding to a 10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence
limit for a 10% response level.
a Selected model in bold.
E-68
-------
APRIL 2024
1
0.9
0.8
0.7
qj
8> 0.6
£
o
Q.
0.5
& 0.4
0.3
Frequentist Log-Logistic Model with BMR of 10% Extra Risk for
the BMD and 0.95 Lower Confidence Limit for the BMDL
0.2 T T
01 cM-
10
20
30
Dose
40
50
^^Estimated Probability
^^Response at BMD
O Data
BMD
BMDL
Figure E-13. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Log-Logistic
Model for Individual Cell Necrosis in the Liver in Female Sprague-Dawley Crl:CD(SD)IGS
BR Rats Following Exposure to PFOS (Butenhoff et al., 2012; Thomford, 2002)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.2 Lee et al. (2015)
EPA conducted dose-response modeling of the Lee et al. (2015) study using the BMDS 3.2
program. This study addresses fetal body weight in Fi male and female CD-I mice.
E.2.2.1 Fetol Body Weight
Decreased mean response of fetal body weight was observed in Fi male and female CD-I mice.
Continuous models were used to fit dose-response data. BMR of a 5% change was chosen and a
change in the mean equal to 0.5 standard deviations from the control mean was also modeled for
comparison purposes. The doses and response data used for the modeling are listed in Table
E-56. For developmental endpoints, a dose metric that represents the average concentration
normalized per day (Cavg) during the relevant exposure window used for the study (i.e., gestation
(Cavg.pup.gest), lactation (Cavg.PuP.iact), or gestation and lactation (Cavg.pup.gesUact)). See the Toxicity
Assessment (U.S. EPA, 2024) for additional details. For decreased fetal body weight,
the Cavg.pup.gest metric was selected because pups were exposed during gestation only.
Table E-56. Dose-Response Modeling Data for Fetal Body Weight in Fi Male and Female
CD-I Mice Following Exposure to PFOS (Lee et al., 2015)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Mean Response (g)a
0
0
10
1.7 ±0.2
0.5
0.9
10
1.5 ±0.1
2
3.5
10
1.3 ±0.1
8
14.0
10
1.1 ± 0.2
E-69
-------
APRIL 2024
Notes:
aData are presented as mean ± standard deviation.
Tests for constant and nonconstant variance failed. In such cases, it is not recommended to
model the dataset. Significance testing for constant variance models assumes that the model
errors (or residuals) have constant variance; if this assumption is violated the p-values from the
model are no longer reliable. Similarly, significance testing for nonconstant models assumes that
the model errors (or residuals) have nonconstant variance; if this assumption is violated the p-
values from the model are no longer reliable (Breusch and Pagan, 1979). For modeling endpoints
where tests for constant and nonconstant variance failed, it is thus not recommended to model the
dataset, therefore, a NOAEL approach was taken for such endpoints.
E.2.3 Luebker et oi. (2005b)
EPA conducted dose-response modeling of the Luebker et al. (2005b) study using the BMDS 3.2
program. This study addresses pup body weight relative to the litter at LD 1 and LD 5 in Fi male
and female Sprague-Dawley rats.
E.2.3.1 Pup Body Weight Relative to Litter at LD 5
Decreased mean response of pup body weight relative to the litter at LD 5 was observed in Fi
male and female Sprague-Dawley rats. Continuous models were used to fit dose-response data.
A BMR of a 5% change from the control mean was selected and a BMR of a 0.5 standard
deviation change from the mean is provided for comparison purposes. The doses and response
data used for the modeling are listed in Table E-57. For developmental endpoints, a dose metric
that represents the average concentration normalized per day (Cavg) during the relevant exposure
window used for the study (i.e., gestation (Cavg,pup,gest), lactation (Cavg,pup,iact), or gestation and
lactation (Cavg,pup,gest,iact)). See the Toxicity Assessment (U.S. EPA, 2024) for additional details.
For decreased pup weight at LD 5, the Cavg,pup,gest,iact metric was selected because pups were
exposed during gestation and lactation.
Table E-57. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 5) in Fi Male and Female Sprague-Dawley Rats Following Exposure to PFOS
(Luebker et al., 2005b)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Mean Response (g)a
0
0
17
9.8 ± 2.1b
0.4
15.9
17
8.6 ± 1.9
0.8
31.9
17
8.5 ±2.8
1
39.8
17
8.1 ±2.5
1.2
47.8
17
7.5 ±2.7
1.6
63.7
17
7.2 ±2.7
2
79.6
17
7.3 ±7.3
Notes:
aData are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
E-70
-------
APRIL 2024
The dose-response data for the highest dose group was removed prior to modeling as the
variance surrounding the mean response for this group was large and dropping this dose group
ensured adequate model fit in the low-dose range (U.S. EPA, 2012). Figure E-14 shows the best
viable model (Polynomial Degree 6) when the highest dose group is included in modeling for
visual comparison of fit. The BMD modeling results for pup body weight relative to the litter at
LD 5 are summarized in Table E-58 and Figure E-15. The Exponential 5 model was selected as it
had the lowest BMDL among the viable models (the Hill model was questionable in this run).
The BMDLs from the selected Exponential 5 model is 2.3 mg/L.
E-71
-------
APRIL 2024
Table E-58. Summary of Benchmark Dose Modeling Results for Pup Body Weight Relative to the Litter (LD 5) in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Constant Variance) (Luebker et al., 2005b)
Goodness of Fit Scaled Residual
BMDo.ssd BMDLo.ssd BMDs BMDLs Basis for Model
Model" p. Dose Group Dose Group Control (mg/L) (mg/L) (mg/L) (m„/L) Selection
value Near BMDo s Near BMDs Dose Group
Exponential 2
0.951
474.7
0.4
-0.6
0.3
28.2
17.6
10.8
7.3
EPA selected the
Exponential 3
0.951
474.7
0.4
-0.6
0.3
28.2
17.6
10.8
7.3
Exponential 5
model. All viable
Exponential 4
0.881
476.6
0.4
-0.5
0.2
25.5
6.9
9.5
2.3
models had
adequate fit (p-
Exponential 5
0.881
476.6
0.4
-0.5
0.2
25.6
6.9
9.5
2.3
values greater
Hill
0.882
476.6
0.4
-0.5
0.2
25.0
3.9
9.2
1.1
than 0.1), and the
Exponential 5
Polynomial
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
8.7
model was
Degree 5
selected as it had
Polynomial
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
9.2
the lowest BMDL
Degree 4
among the viable
Polynomial
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
8.7
models.
Degree 3
Polynomial
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
8.7
Degree 2
Power
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
8.7
Linear
0.941
474.8
0.3
-0.6
0.4
30.6
20.5
12.2
8.7
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDo.ssd = dose level corresponding to a change in the mean equal to
0.5 standard deviations from the control mean; BMDLo.ssd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 0.5
standard deviations from the control mean; BMD5 = dose level corresponding to a 5% change in the mean from the control mean; BMDL5 = lower bound on the dose level
corresponding to the 95% lower confidence limit for a 5% change in the mean from the control mean.
a Selected model in bold
E-72
-------
APRIL 2024
12
Frequentist Polynomial Degree 6 Model with BMR of 0.05 Rel. Dev.
for the BMD and 0.95 Lower Confidence Limit for the BMDL
^^Estimated Probability
^^Response at BMD
O Data
BMD
BMDL
10
20
30
40
Dose
50
60
70
Figure E-14. Plot of Mean Response by Dose (Including Highest Dose) with Fitted Curve
for the Polynomial 6 Model for Pup Body Weight Relative to the Litter at LD 5 in Fi Male
and Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker et al., 2005b)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
12
Frequentist Exponential Degree 5 Model with BMR of 0.05 Rel. Dev.
for the BMD and 0.95 Lower Confidence Limit for the BMDL
^^Estimated Probability
^^Response at BMD
O Data
BMD
BMDL
10
20
30
Dose
40
50
60
Figure E-15. Plot of Mean Response by Dose with Fitted Curve for the Selected
Exponential 5 Model for Pup Body Weight Relative to the Litter at LD 5 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker et al., 2005b)
E-73
-------
APRIL 2024
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.3.2 Pup Body Weight Relative to Litter at LD1
Decreased mean response of pup body weight relative to the litter at LD 1 (i.e., day of birth) was
observed in Fi male and female Sprague-Dawley rats. Continuous models were used to fit dose-
response data. A BMR of a 5% change from the control mean was selected and a BMR of a 0.5
standard deviation change from the mean is provided for comparison purposes. The doses and
response data used for the modeling are listed in Table E-59. For developmental endpoints, a
dose metric that represents the average concentration normalized per day (Cavg) during the
relevant exposure window used for the study (i.e., gestation (Cavg,pup,gest), lactation (Cavg,pup,iact), or
gestation and lactation (Cavg,pup,gest,iact)). See the Toxicity Assessment (U.S. EPA, 2024) for
additional details. For decreased pup weight at LD 1, the Cavg,pup,gest metric was selected because
pups were exposed during gestation only.
Table E-59. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 1) in Fi Male and Female Sprague-Dawley Rats Following Exposure to PFOS
(Luebker et al., 2005b)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Mean Response (g)a
0
0
17
6.4 ± 0.8b
0.4
15.4
17
6.0 ± 1.2
0.8
30.8
17
6.0 ± 1.2
1
38.5
17
5.9 ± 1.6
1.2
46.1
17
5.7 ± 1.2
1.6
61.5
17
5.4 ± 1.0
2
76.9
17
5.4 ± 1.0
Notes:
aData are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
The BMD modeling results for pup body weight relative to the litter at LD 1 are summarized in
Table E-60 and Figure E-16. The Exponential 3 model was selected as it had the lowest AIC
among the viable models. The BMDLs from the selected Exponential 3 model is 14.7 mg/L.
E-74
-------
APRIL 2024
Table E-60. Summary of Benchmark Dose Modeling Results for Pup Body Weight Relative to the Litter (LD 1) in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Nonconstant Variance) (Luebker et al., 2005b)
Goodness of Fit Scaled Residual
BMDo.ssd BMDLo.ssd BMDs BMDLs Basis for Model
Model" p. Dose Group Dose Group Control (mg/L) (mg/L) (mg/L) (m„/L) Selection
value Near BMDo s Near BMDs Dose Group
Exponential 2
0.950
375.8
0.3
-0.5
0.2
40.5
23.8
22.6
14.7
EPA selected the
Exponential 3
0.950
375.8
0.3
-0.5
0.2
40.5
23.8
22.6
14.7
Exponential 3
model. All
Exponential 4
0.888
377.8
0.3
-0.5
0.2
40.3
0
22.5
0
models had
Exponential 5
0.887
377.8
0.3
-0.5
0.2
40.4
0
22.6
0
adequate fit (p-
Hill
0.974
378.8
0.4
0.4
-0.2
11.0
0
8.6
0
values greater
than 0.1), and the
Polynomial
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
17.0
Exponential 3
Degree 6
model was
Polynomial
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.4
selected as it had
Degree 5
the lowest AIC
Polynomial
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.3
among the viable
Degree 4
models.
Polynomial
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.3
Degree 3
Polynomial
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.3
Degree 2
Power
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.3
Linear
0.946
375.8
-0.1
0.3
0.2
42.7
26.3
24.0
16.3
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDo.ssd = dose level corresponding to a change in the mean equal to
0.5 standard deviations from the control mean; BMDLo.ssd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 0.5
standard deviations from the control mean; BMD5 = dose level corresponding to a 5% change in the mean from the control mean; BMDL5 = lower bound on the dose level
corresponding to the 95% lower confidence limit for a 5% change in the mean from the control mean.
a Selected model in bold
E-75
-------
APRIL 2024
8
Estimated Probabilit*
Response at BMD
O Data
BMD
—-BMDL
1
0
0 10 20 30 40 50 60 70
Dose
Figure E-16. Plot of Mean Response by Dose with Fitted Curve for the Selected
Exponential 3 Model for Pup Body Weight Relative to the Litter at LD 1 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Luebker et al., 2005b)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.4 NTP (2019)
EPA conducted dose-response modeling of the NTP (2019) study using the BMDS 3.2 program.
This study addresses extramedullar hematopoiesis in the spleen in male and female Sprague-
Dawley rats.
E.2.4.1 Extramedullar)/ Hematopoiesis in the Spleen in Male Sprague-
Dawley Rats
Increased incidence of extramedullar hematopoiesis in the spleen was observed in male
Sprague-Dawley rats. Dichotomous models were used to fit dose-response data. A BMR of 10%
extra risk was chosen per EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012). The
doses and response data used for the modeling are listed in Table E-61. As described in the
Toxicity Assessment (U.S. EPA, 2024), the Ciast7,avg was selected for all non-developmental
studies rather than alternate metrics such as Cmax to provide a consistent internal dose for use
across chronic and subchronic study designs where steady state may or may not have been
reached and to allow extrapolation to the human PK model.
Table E-61. Dose-Response Modeling Data for Extramedullar Hematopoiesis in Male
Sprague-Dawley Rats Following Exposure to PFOS (NTP, 2019)
Administered Dose
Internal Dose
(mg/kg/day)
(mg/L)
Number per Group
Incidence
0
0
10
1
E-76
-------
APRIL 2024
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Incidence
0.312
10.2
10
1
0.625
20.4
10
2
1.25
40.8
10
7
2.5
81.6
10
8
5
162.7
10
10
The BMD modeling results for extramedullary hematopoiesis in the spleen are summarized in
Table E-62 and Figure E-17. The best fitting model was the Logistic model based on adequate p-
values (greater than 0.1), the BMDLs were sufficiently close (less than threefold difference)
among adequately fitted models, and the Logistic model had the lowest AIC. The BMDLio from
the selected Logistic model is 9.6 mg/L.
Table E-62. Summary of Benchmark Dose Modeling Results for Extramedullary
Hematopoiesis in Male Sprague-Dawley Rats Following Exposure to PFOS (NTP, 2019)
Goodness of Fit
Scaled Residual
BMDio
BMDLio
Basis for Model
Model3
p-value
AIC
Dose Group
Control Dose
(mg/L)
(mg/L)
Selection
Near BMD
Group
Dichotomous
0.646
53.0
-0.3
0.2
15.7
7.1
EPA selected the
Hill
Logistic model. All
Gamma
0.594
53.2
-0.3
0.2
13.8
4.6
models had
Log-Logistic
0.646
53.0
-0.3
0.2
15.7
7.1
adequate fit (p-
values greater than
0.1), the BMDLs
Multistage
0.487
53.7
-0.5
0.3
10.9
4.2
Degree 5
were sufficiently
Multistage
0.487
53.7
-0.5
0.3
10.9
4.2
close (less than
Degree 4
threefold
Multistage
0.487
53.7
-0.5
0.3
10.9
4.3
difference), and the
Degree 3
Logistic model had
the lowest AIC.
Multistage
0.487
53.7
-0.5
0.3
10.9
4.3
Degree 2
Multistage
0.475
53.4
-1.0
0.6
5.4
3.7
Degree 1
Weibull
0.549
53.4
-0.4
0.3
12.1
4.4
Logistic
0.558
52.2
-0.6
-0.1
14.0
9.6
Log-Probit
0.676
52.8
-0.4
0.2
16.0
7.5
Probit
0.558
52.3
-0.6
0.0
13.4
9.5
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDio = dose level
corresponding to a 10% response level; BMDLio = lower bound on the dose level corresponding to the 95% lower confidence
limit for a 10% response level.
a Selected model in bold.
E-77
-------
APRIL 2024
Figure E-17. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Logistic
Model for Extramedullar Hematopoiesis in the Spleen in Male Sprague-Dawley Rats
Following Exposure to PFOS (NTP, 2019)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.4.2 Extra medullary Hematopoiesis in the Spleen in Female Sprague-
Dawley Rats
Increased incidence of extramedullar hematopoiesis in the spleen was observed in female
Sprague-Dawley rats. Dichotomous models were used to fit dose-response data. A BMR of 10%
extra risk was chosen per EPA's Benchmark Dose Technical Guidance (U.S. EPA, 2012). The
doses and response data used for the modeling are listed in Table E-63. As described in the
Toxicity Assessment (U.S. EPA, 2024), the Ciastzavg was selected for all non-developmental
studies rather than alternate metrics such as Cmax to provide a consistent internal dose for use
across chronic and subchronic study designs where steady state may or may not have been
reached and to allow extrapolation to the human PK model.
Table E-63. Dose-Response Modeling Data for Extramedullary Hematopoiesis in the
Spleen in Female Sprague-Dawley Rats Following Exposure to PFOS (NTP, 2019)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Incidence
0
0
10
2
0.312
10.0
10
3
0.625
20.0
10
3
1.25
40.0
10
8
2.5
80.0
10
10
5
159.6
10
10
E-78
-------
APRIL 2024
The BMD modeling results for extramedullary hematopoiesis in the spleen are summarized in
Table E-64 and Figure E-18. The Multistage Degree 1 model was selected based on adequate p-
values (greater than 0.1), the BMDLs were sufficiently close (less than threefold difference)
among adequately fitted models, and the Multistage Degree 1 model had the lowest BMDL. The
BMDLio from the selected Multistage Degree 1 model is 2.3 mg/L.
Table E-64. Summary of Benchmark Dose Modeling Results for Extramedullary
Hematopoiesis in the Spleen in Female Sprague-Dawley Rats Following Exposure to PFOS
(NTP, 2019)
Basis for Model
Selection
Goodness of Fit
Scaled Residual
BMDio
(mg/L)
BMDLio
(mg/L)
Model3
p-value
AIC
Dose Group
Near BMD
Control Dose
Group
Dichotomous
0.849
52.8
0.2
-0.5
26.4
9.1
Hill
Gamma
0.966
50.7
0.0
-0.4
21.8
5.7
Log-Logistic
0.956
50.8
0.2
-0.4
25.7
9.1
Multistage
0.989
50.6
-0.2
-0.1
16.1
3.4
Degree 5
Multistage
0.981
50.6
-0.2
-0.1
16.5
3.4
Degree 4
Multistage
0.959
50.8
-0.3
-0.2
16.5
3.5
Degree 3
Multistage
0.948
49.2
0.3
0.1
11.5
3.6
Degree 2
Multistage
0.448
53.0
0.6
0.6
3.5
2.3
Degree 1
Weibull
0.990
48.7
-0.2
-0.2
18.0
5.0
Logistic
0.877
49.8
0.3
0.5
7.6
5.1
Log-Probit
0.963
50.8
0.1
-0.4
22.5
8.8
Probit
0.888
49.7
0.2
0.5
7.2
5.0
EPA selected the
Multistage Degree
1 model. All
models had
adequate fit (p-
values greater than
0.1), the BMDLs
were sufficiently
close (less than
threefold
difference), and the
Multistage Degree
1 model had the
lowest BMDL.
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL
corresponding to a 10% response level; BMDLio = lower bound on the dose
limit for a 10% response level.
a Selected model in bold.
= benchmark dose lower limit; BMDio = dose level
level corresponding to the 95% lower confidence
E-79
-------
APRIL 2024
Figure E-18. Plot of Incidence Rate by Dose with Fitted Curve for the Selected Multistage
Degree 1 Model for Extramedullar Hematopoiesis in the Spleen in Female Sprague-
Dawley Rats Following Exposure to PFOS (NTP, 2019)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.5 Zhong et al. (2016)
EPA conducted dose-response modeling of the Zhong et al. (2016) study using the BMDS 3.2
program. This study addresses plaque-forming cell (PFC) response of splenic cells in Fi male
C57BL/6 mice at PNW 4.
E.2.5.1 Ploque-Forming Cell Response of Splenic Cells in Fi Mole
C57BL/6 Mice
Decreased mean response of PFC response of splenic cells was observed in Fi male C57BL/6
mice at PNW 4. Continuous models were used to fit dose-response data. A BMR of a change in
the mean equal to one standard deviation from the control mean was chosen per EPA's
Benchmark Dose Technical Guidance (U.S. EPA, 2012). The doses and response data used for
the modeling are listed in Table E-65. For developmental endpoints, a dose metric that represents
the average concentration normalized per day (Cavg) during the relevant exposure window used
for the study (i.e., gestation (Cavg.PuP.gest), lactation (Cavg.PuP.iact), or gestation and lactation
(Cavg,PuP,gest,iact)). See the Toxicity Assessment (U.S. EPA, 2024) for additional details. For
decreased PFC response at PNW 4, the Cavg.PuP.gest.iact metric was selected because pups were
exposed during gestation and lactation.
E-80
-------
APRIL 2024
Table E-65. Dose-Response Modeling Data for PFC Response of Splenic Cells in Fi Male
C57BL/6 Mice at PNW 4 Following Exposure to PFOS (Zhong et al., 2016)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Mean Response (# Cells
per 106 Spleen Cells)3
0
0.0
12
465.7 ±78.5b
0.1
1.7
12
423.0 ±60.4
1
16.8
12
398.7 ±72.5
5
84.1
12
340.1 ±54.4
Notes:
a Data are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
BMD modeling results for PFC response of splenic cells are summarized in Table E-66 and
Figure E-19. The best fitting model was the Hill model based on adequate p-values (greater than
0.1), the BMDLs were sufficiently close (less than threefold difference) among adequately fitted
models, and the Hill model had the lowest BMDL. The BMDLisd from the selected Hill model
is 1.8 mg/L.
Table E-66. Summary of Benchmark Dose Modeling Results for Plaque-Forming Cell
Response of Splenic Cells in Fi Male C57BL/6 Mice at PNW 4 Following Exposure to
PFOS (Constant Variance) (Zhong et al., 2016)
Goodness of Fit
Scaled Residual
Model3
p-value
. T„ Dose Group
Near BMD
Control Dose
Group
BMDisd
(mg/L)
BMDLisd
(mg/L)
Exponential 2
0.181
545.3
0.2
1.4
51.3
34.4
Exponential 3
0.181
545.3
0.2
1.4
51.3
34.4
Exponential 4
0.174
545.7
0.2
0.9
22.3
6.6
Exponential 5
0.174
545.7
0.2
0.9
22.2
6.6
Hill
0.190
545.6
0.3
0.8
20.6
1.8
Polynomial
0.161
545.5
0.2
1.4
55.1
38.9
Degree 3
Polynomial
0.161
545.5
0.2
1.4
55.1
38.9
Degree 2
Power
0.161
545.5
0.2
1.4
55.1
38.9
Linear
0.161
545.5
0.2
1.4
55.1
38.9
Basis for Model
Selection
EPA selected the
Hill model. All
models had
adequate fit (p-
values greater than
0.1), the BMDLs
were sufficiently
close (less than
threefold
difference), and the
Hill model had the
lowest BMDL.
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDisd = dose
level corresponding to a change in the mean equal to 1 standard deviation from the control mean; BMDLisd = lower bound on
the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 1 standard deviation from the
control mean.
a Selected model in bold.
E-81
-------
APRIL 2024
^—Estimated Probability
^—Response at BMD
O Data
BMD
BMDL
Figure E-19. Plot of Mean Response by Dose with Fitted Curve for the Selected Hill Model
for PFC Response of Splenic Cells in Fi Male C57BL/6 Mice at PNW 4 Following Exposure
to PFOS (Zhong et al., 2016)
BMD = benchmark dose; BMDL = benchmark dose lower limit.
E.2.6 Lou et al. (2003)
EPA conducted dose-response modeling of the Lau et al. (2003) study using the BMDS 3.2
program. This study addresses offspring survival in Fi male and female Sprague-Dawley rats at
PND 5 and PND 22.
E.2.6.1 Pup Survival at PND 5
Decreased mean response of number of surviving offspring at PND 5 was observed in Fi male
and female Sprague-Dawley rats. Continuous models were used to fit dose-response data. A
BMR of a change in the mean equal to 0.1 and 0.5 standard deviations from the control mean
were chosen. The doses and response data used for the modeling are listed in Table E-67. For
developmental endpoints, a dose metric that represents the average concentration normalized per
day (Cavg) during the relevant exposure window used for the study (i.e., gestation (Cavg.PuP.gest),
lactation (Cavg.PuP.iact), or gestation and lactation (Cavg.PuP.gest.iact)). See the Toxicity Assessment
(U.S. EPA, 2024) for additional details. For decreased pup survival at PND 5, the Cavg.PuP.gest.iact
metric was selected because pups were exposed during gestation and lactation. The Cavg.PuP.gest.iact
was selected for this model.
600
Dose
E-82
-------
APRIL 2024
Table E-67. Dose-Response Modeling Data for Pup Survival at PND 5 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Lau et al., 2003)
Administered Dose
(mg/kg/day)
Internal Dose
C avg,pup,gest,lact
(mg/L)
Number per
Group
Mean Response (Percent
Survival per Litter)3
0
0
18
90 ± 8.9b
1
13.0
12
86 ± 27.7
2
25.9
9
79 ±20.1
3
38.9
17
45 ±37.1
5
64.9
17
4 ± 10.3
Notes:
aData are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
Tests for constant and nonconstant variance failed. In such cases, it is not recommended to
model the dataset. Significance testing for constant variance models assumes that the model
errors (or residuals) have constant variance; if this assumption is violated the p-values from the
model are no longer reliable. Similarly, significance testing for nonconstant models assumes that
the model errors (or residuals) have nonconstant variance; if this assumption is violated the p-
values from the model are no longer reliable (Breusch and Pagan, 1979). For modeling endpoints
where tests for constant and nonconstant variance failed, it is thus not recommended to model the
dataset, therefore, a NOAEL approach was taken for such endpoints.
E.2.6.2 Pup Survival at PND 22
Decreased mean response of number of surviving offspring at PND 22 was observed in Fi male
and female Sprague-Dawley rats. Continuous models were used to fit dose-response data. A
BMR of a change in the mean equal to 0.1 and 0.5 standard deviations from the control mean
were chosen. The doses and response data used for the modeling are listed in Table E-68. For
developmental endpoints, a dose metric that represents the average concentration normalized per
day (Cavg) during the relevant exposure window used for the study (i.e., gestation (Cavg,pup,gest),
lactation (Cavg,pup,iact), or gestation and lactation (Cavg,pup,gest,iact)). See the Toxicity Assessment
(U.S. EPA, 2024) for additional details. For decreased pup survival at PND 22, the Cavg,pup,gest,iact
metric was selected because pups were exposed during gestation and lactation. The Cavg,pup,gest,iact
was selected for this model.
Table E-68. Dose-Response Modeling Data for Pup Survival at PND 22 in Fi Male and
Female Sprague-Dawley Rats Following Exposure to PFOS (Lau et al., 2003)
Administered Dose
(mg/kg/day)
Internal Dose
C avg,pup,gest,lact
(mg/L)
Number per Group
Mean Response (Percent
Survival per Litter)3
0
17.3
34.6
51.9
18
12
9
17
78 ± 14.8b
74 ±28.8
61 ±40.2
34 ±33
E-83
-------
APRIL 2024
5 86.4 17 2 ±8.7
Notes: PND =postnatal day.
aData are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
Tests for constant and nonconstant variance failed. In such cases, it is not recommended to
model the dataset. Significance testing for constant variance models assumes that the model
errors (or residuals) have constant variance; if this assumption is violated the p-values from the
model are no longer reliable. Similarly, significance testing for nonconstant models assumes that
the model errors (or residuals) have nonconstant variance; if this assumption is violated the p-
values from the model are no longer reliable (Breusch and Pagan, 1979). For modeling endpoints
where tests for constant and nonconstant variance failed, it is thus not recommended to model the
dataset, therefore, a NOAEL approach was taken for such endpoints.
E.2.7 Luebker et al. (2005a)
EPA conducted dose-response modeling of the Luebker et al. (2005a) study using the BMDS 3.2
program. This study addresses pup body weight relative to the litter observed on LD 1 (i.e., day
of birth) in Fi male and female Sprague-Dawley rats.
E.2.7.1 Pup Body Weight Relative to Litter at LD 1
Decreased mean response of pup body weight relative to the litter at LD 1 (i.e., day of birth) was
observed in Fi male and female Sprague-Dawley rats. Continuous models were used to fit dose-
response data. A BMR of a 5% change from the control mean was selected and a BMR of a 0.5
standard deviation change from the mean is provided for comparison purposes. The doses and
response data used for the modeling are listed in Table E-69. For developmental endpoints, a
dose metric that represents the average concentration normalized per day (Cavg) during the
relevant exposure window used for the study (i.e., gestation (Cavg,pup,gest), lactation (Cavg,pup,iact), or
gestation and lactation (Cavg,pup,gest,iact)). See the toxicity assessment (U.S. EPA, 2024) for
additional details. For decreased pup weight at LD 1, the Cavg,pup,gest metric was selected because
pups were exposed during gestation only.
Table E-69. Dose-Response Modeling Data for Pup Body Weight Relative to the Litter
(LD 1) in Fi Male and Female Sprague-Dawley Rats Following Exposure to PFOS
(Luebker et al., 2005a)
Administered Dose
(mg/kg/day)
Internal Dose
(mg/L)
Number per Group
Mean Response (g)a
0
0
23
6.6 ± 0.6b
0.1
3.8
25
6.6 ±0.5
0.4
15.3
22
6.4 ±0.7
1.6
61.6
20
5.7 ±0.5
3.2
131.0
20
5.3 ±0.4
Notes:
aData are presented as mean ± standard deviation.
b Standard deviations were calculated from standard errors.
E-84
-------
APRIL 2024
The BMD modeling results for pup body weight relative to the litter at LD 1 are summarized in
Table E-70 and Figure E-20. The Exponential 4 model was selected as it had the lowest AIC
among the viable models. The BMDLs from the selected Exponential 4 model is 11.3 mg/L.
E-85
-------
APRIL 2024
Table E-70. Summary of Benchmark Dose Modeling Results for Pup Body Weight Relative to the Litter at LD 1 in Fi Male
and Female Sprague-Dawley Rats Following Exposure to PFOS (Nonconstant Variance) (Luebker et al., 2005a)
Goodness of Fit Scaled Residual
Model"
p-value
AIC
Dose
Group
Near
BMDo.s
Dose Group
Near BMDs
Control Dose
Group
BMDo.ssd
(mg/L)
BMDLo.ssd
(mg/L)
BMDs
(mg/L)
BMDLs
(mg/L)
Basis for Model
Selection
Exponential 2
0.259
184.4
0.04
0.04
0.3
27.1
21.5
29.5
25.3
EPA selected the
Exponential 3
0.259
184.4
0.04
0.04
0.3
27.1
21.5
29.5
25.3
Exponential 4
Exponential 4
0.675
183.1
0.36
0.36
-0.4
15.9
9.9
17.7
11.3
model. All models
Exponential 5
0.969
184.4
-0.29
-0.29
0.1
24.1
10.6
25.9
12.1
had adequate fit
Hill
0.926
184.4
-0.29
-0.29
0.1
23.7
9.9
25.3
11.3
(p-values greater
Polynomial
0.164
185.5
0.03
0.03
0.4
30.0
24.2
32.5
28.4
man u. ij, ine
BMDLs were
Degree 4
sufficiently close
Polynomial
0.164
185.5
0.03
0.03
0.4
30.0
24.2
32.5
28.4
(less than
Degree 3
threefold
Polynomial
0.164
185.5
0.03
0.03
0.4
30.0
24.2
32.5
28.4
difference), and
Degree 2
the Exponential 4
Power
0.164
185.5
0.03
0.03
0.4
30.0
24.2
32.5
28.4
model had the
Linear
0.164
185.5
0.03
0.03
0.4
30.0
24.2
32.5
28.4
lowest AIC.
Notes: AIC = Akaike information criterion; BMD = benchmark dose; BMDL = benchmark dose lower limit; BMDo.5sd = dose level corresponding to a change in the mean equal to
0.5 standard deviations from the control mean; BMDLo.5sd = lower bound on the dose level corresponding to the 95% lower confidence limit for a change in the mean equal to 0.5
standard deviations from the control mean; BMD5 = dose level corresponding to a 5% change in the mean from the control mean; BMDL5 = lower bound on the dose level
corresponding to the 95% lower confidence limit for a 5% change in the mean from the control mean.
a Selected model in bold
E-86
-------
APRIL 2024
8
6
C
D
-------
APRIL 2024
Appendix F. Pharmacokinetic Modeling
For the animal pharmacokinetic model, model predictions from Wambaugh et al. (2013) were
evaluated by comparing each predicted final serum concentration to the serum value in the
supporting animal studies (training data set) and to animal studies published since the publication
of Wambaugh et al. (2013) (test data set). The predictions to these two data sets were generally
similar to the experimental values. There were no systematic differences between the
experimental data and the model predictions across species, strain, or sex, and median model
outputs uniformly appeared to be biologically plausible despite the uncertainty reflected in some
of the 95th percentile confidence intervals (CIs). The application of the model outputs in the
derivation of a human RfD can be found in the main perfluorooctane sulfonic acid (PFOS)
document.
F.l Comparison of Fits to Training Datasets Used in Wambaugh
et al. (2013)
The following figures show comparisons of the model predicted serum concentrations to the data
used for model training. Fits are also presented in supplemental material of Wambaugh et al.
(2013).
io2 -
cn
E
- io1 -
c
o
u
E
-------
APRIL 2024
60 80 100
Time (days)
60 80 100 120 140
Time (days)
Figure F-2. Experimentally Observed Serum Concentrations (Chang et al., 2012) and
Median Prediction for a Single Oral Dose of 1 or 20 mg/kg PFOS to Male CD1 Mice
A) Fits to observed male data using male-specific model. B) Fits to observed male data using female-specific model parameters.
One mg/kg represented by the squares and solid line; 20 mg/kg represented by the circles and dashed line.
Time (days)
Figure F-3. Experimentally Observed Serum Concentrations (Chang et al., 2012) and
Median Prediction for a Single IV Dose of 2 mg/kg or a Single Oral Dose of 2 or 15 mg/kg
PFOS to Male Sprague-Dawley Rats
Two mg/kg intravenous (IV) dose represented by the upward triangles and solid line; 2 mg/kg oral dose represented by the
squares and solid line; 15 mg/kg oral dose represented by the circles and dashed line.
F-2
-------
APRIL 2024
Literature reported concentration (mg/L)
Figure F-4. Model Prediction Summary for PFOS Training Data
Model predictions on the training data result in a mean squared log error (MSLE) of 0.174. Dashed lines represent +/- one-half
logio.
A local, one-at-a-time sensitivity analysis was conducted to examine how parameter sensitivity
varied across the adult and developmental models (Figure F-5). For each parameter/dose metric
pair, sensitivity coefficients were calculated to describe the relative change in a dose metric
relative to the proportional change in a parameter value. A sensitivity coefficient of 1 describes
the situation where a 1% increase in a parameter resulted in a 1% increase in the dose metric.
F-3
-------
APRIL 2024
Vcc_i
Vcc
¦
Vcc
Qfilc
¦
¦
free
Qfilc
1
¦
Rv2v1
free
1
k12
Cavg pup qest
k12
0)
0)
I KT
9
Cavg pup lact
Cavg pup gest lact
Cavg pup diet
Cavg pup total
ra Ry2v1
5
Cavg
Vfilc
KT
P_milk
halfjife
¦
¦
Vfilc
Tmc
Tmc
ka
ka
r_f_m
-0.8 -0.6 -0.4 -0.2 0
value
0
-1.0 -0.5 0.0 0.5 1.0
value
Adult model
Developmental model
Figure F-5. PFOS Sensitivity Coefficients of the Adult Model and Developmental Model
As demonstrated in Figure F-5, the volume of distribution (Vd) represents the most sensitive
parameter for average concentrations in the adult animal. Because of the long half-life and high
degree of plasma protein binding, renal resorption parameters that impact the effective half-life
of PFOS are not as sensitive when compared to PFOA which has a shorter net half-life.
Comparatively, the four one-compartment parameters for the infant (volume of distribution, half-
life, serum:milk partition coefficient, and fetal:maternal ratio) are all sensitive to the
gestational/lactational dose metrics. However, once the pup transitions to the adult model
(Wambaugh model), PFOS transfer during gestation/lactation does not impact the average
concentration during the post-weaning phase (Cavg-PuP-diet). This is because the steady state
concentration for the pup exposed to PFOS in the diet during growth is much larger than the
steady state concentration during the 21 days of lactational exposure.
F.2 Visual Inspection of Test Datasets not Used for Initial Fitting
The following figures show a comparison between model predictions and data from more
recently published studies that were not part of the Wambaugh et al. (2013) parameterization.
F-4
-------
APRIL 2024
102 l
=!
f 101 :
u I
c
o
u
| iff3 -
U
•/I
in
O
_
CL-
IO'1 -
10"3 10"2 10"1 10° 101 102 103
Time (days)
Figure F-6. mentally Observed Serum Concentrations (Huang et al., 2021) and Median
Predictions for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 or 20 mg/kg PFOS to Male
Sprague-Dawley Rats
Two mg/kg intravenous (IV) dose represented by the squares and solid line; 2 mg/kg oral dose represented by the circles and
solid line; 20 mg/kg oral dose represented by the downward triangles and dashed line.
Time (days) Time (days)
Figure F-7. Experimentally Observed Serum Concentrations (Huang et al., 2021) and
Median Predictions for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 or 20 mg/kg PFOS
to Female Sprague-Dawley Rats
A) Fits to observed female data using female-specific model parameters. B) Fits to observed female data using male-specific
model parameters.
Two mg/kg intravenous (IV) dose represented by the squares and solid line; 2 mg/kg oral dose represented by the circles and
solid line; 20 mg/kg oral dose represented by the downward triangles and dashed line.
F-5
-------
APRIL 2024
E
_ 10°
10"
o ° °
° ~ ~
O
~ O
o
~ 2 mg/kg, IV
O 2 mg/kg, oral
10"
10"
10°
Time (days)
1Q1
102
Figure F-8. Experimentally Observed Serum Concentrations (Kim et al., 2016b) and
Median Prediction for a Single IV Dose of 2 mg/kg or an Oral Dose of 2 mg/kg PFOS to
Male Sprague-Dawley Rats
Two mg/kg intravenous (IV) dose represented by the squares; 2 mg/kg oral dose represented by the circles.
|
£
| 10°
~°
r\
o°o
~
O
B
^°a00°a
O
\
i
£
T 10°
u
C
o
/o A
O
o
~
£
E
V
M
s
fe 10-'
o
~
2 mg/kg, IV
~ 2 mg/kg, IV
o
2 mg/kg, oral
O 2 mg/kg, oral
10"
10"
10°
Time (days)
101
10*
10~
10"
10°
Time (days)
101
102
Figure F-9.Experimentally Observed Serum Concentrations (Kim et al., 2016b) and
Median Prediction for a Single IV Dose of 2 mg/kg an Oral Dose of 2 mg/kg PFOS to
Female Sprague-Dawley Rats
A) Fits to observed female data using female-specific model parameters. B) Fits to observed female data using male-specific
model parameters.
Two mg/kg intravenous (IV) dose represented by the squares; 2 mg/kg oral dose represented by the circles.
F-6
-------
APRIL 2024
cone mean
Figure F-10. Model Prediction Summary for PFOS Test Data
Model predictions on the adult, single-dose test data result in a mean squared log error (MSLE) of 0.384. Dashed lines
represent +/- one-half logio. Developmental pharmacokinetic summary results not shown as only one study (presented in main
text) is available for comparison.
F.3 Human Model Validation
As mentioned in the Toxicity Assessment (U.S. EPA, 2024), the human model was implemented
in R/MCSim from the original AcslX model (Verner et al., 2016). Comparison with model
output from the original model shows that, with the original parameters, the R model exactly
replicates the original model (Figure F-l 1). The only difference remaining was that the start of
pregnancy occurs at slightly different times in the two models, but this does not affect
predictions outside of that very narrow time. Validation figures shown in this section include
data for PFOA as well as PFOS. This is because model validation and decisions related to model
structure were made for both chemicals together due to the preference for a similar model
structure for the two chemicals.
F-7
-------
APRIL 2024
o
Maternal PFAS_TK
— - Child PFAS_TK
Maternal Verner
— Child Verner
Maternal Rep.
— Child Rep.
T"
1.0
Age (yr)
Age (yr)
Figure F-ll. Model Comparison
Comparison of the original AcslX model output (red, "Verner" label), the R model output with original model parameters (blue,
"Rep." label), and the R model output with updated parameters (black, "PFAS_TK" label). Note that the red lines are almost
entirely obscured by the blue lines.
The updated parameters result in lower serum concentrations for both the maternal and child. This is mainly due to lower half-
lives selected during the parameter update.
Application of the updated parameters to predictions of serum levels in children showed good
agreement between model predictions and reported values (Figure F-l l;Figure F-12). This
simulation was performed using mean breastmilk consumption estimates rather than the 95th
percentile values from EPA's Exposure Factors Handbook (U.S. EPA, 201 lb). Exposure in the
validation scenario was assumed to be constant relative to body weight and was the same in the
mother and child. This exposure was set such that predicted maternal serum level at delivery
matched the reported value. Unlike the version of the model applied for human exposure
prediction, validation was performed using the age-dependent mean breastmilk consumption
estimates. The main application of the model used the 95th quantile of breastmilk consumption
to provide a health-protective estimate of exposure. Each validation scenario was customized
based on information about the length of breastfeeding typical in that cohort. As a reminder, the
default modeling scenario consisted of 1 year of breastfeeding, with an instantaneous transition
to non-breastfeeding exposure (i.e., with exposure to other PFAS sources at weaning). One year
is more typical of total (exclusive and partial) breastfeeding, as opposed to exclusive
breastfeeding which typically lasts up to around 6 months of age.
For the simulation of the Fromme et al. (2010) cohort, information on breastfeeding status was
only available 6 months after birth. At this point 37 of 50 participants were exclusively breastfed,
6 predominantly breastfed, 6 partially breastfed, and 1 received no breast milk. As in the analysis
by Verner et al. (2016), this scenario was modeled as exclusive breastfeeding to 6 months of age
at which point the constant per bodyweight exposure starts equivalent to maternal exposure. For
the cohort of the MOBA study (Granum et al., 2013), the average breast-feeding duration was
12.8 months. Because breastfeeding parameters were only developed in the model up to 1 year,
F-8
-------
APRIL 2024
and the information used to inform the model only extended to 1 year, the simulation for this
scenario used the default 1 year of breastfeeding. In the Mogensen et al. (2015b) study, the
median length of exclusive breastfeeding was 4.5 months, and the median length of partial
breastfeeding was 4.0 months so 8.5 months was chosen as the breastfeeding duration for
simulation of this study.
Fromme, 2010
MOBA Cohort
Mogensen, 2015
Fromme, 2010
MOBA Cohort
Mogensen, 2015
Age (yr) Age (yr)
Figure F-12. Predicted Child Serum Levels Compared to Reported Values
These values were calculated using the updated parameters with constant Vd and exposure relative to body weight.
Observed PFOA (ng/ml) Observed PFOS (ng/ml)
Figure F-13. Comparison of Predicted and Observed Child Serum Concentration
Dashed guidelines represent a 2-fold difference between observed and predicted concentration.
F-9
-------
APRIL 2024
Local, one-at-a-time sensitivity analysis was performed to examine how parameter sensitivity
varied across age and between maternal and child serum (Figure F-13). Sensitivity coefficients
describe the change in a dose metric, in this case serum concentration, relative to the
proportional change in a parameter value, in this case a 1% increase. A sensitivity coefficient of
1 describes the situation where a 1% increase in a parameter resulted in a 1% increase in serum
concentration. Half-life and Vd were sensitive for every dose metric because they govern the
distribution and excretion in all life stages and have a synergistic effect on child levels because
they influence the serum levels in children directly as well as the indirect exposure to the child
early in life through maternal exposure.
For maternal serum at delivery, only the half-life and the Vd influenced the serum concentration.
This was expected as the other parameters evaluated govern distribution of PFOS to the child
and are not in play at this point. For cord blood, a similar effect is observed from Vd and half-life
as in the maternal serum, because cord blood levels are based on maternal levels in the model,
but a high sensitivity is also seen on the cord blood:maternal serum ratio parameter. This was not
unexpected but emphasizes the importance of this parameter for this endpoint. The 1-year
timepoint occurs at the peak serum concentration associated with the end of breastfeeding.
Consistent with this observation, the parameters that govern lactational transfer of PFOS (i.e.,
breastmilk intake and the milk:maternal serum ratio) have high sensitivity coefficients.
Additionally, sensitivity to Vd is high because that governs the relationship between exposure
and serum levels by accounting for the amount of PFOS distributed to tissues. At the 5-year
timepoint the sensitivity to parameters associated with lactational exposure has decreased. The
sensitivity to Vd is somewhat lower compared with the value at 1 year, and the sensitivity to half-
life has slightly increased. This reflects the increased importance of excretion relative to the
distribution of incoming PFOS during the time period following lactational exposure.
F-10
-------
APRIL 2024
Maternal at Delh
Breastmilk Intake per kg BW ! I
Cord Blood:Maternal Serum Ratio j I
Milk:Maternal Serum Ratio j I
Volume of Distribution j I
Half-life |
i 1 F=—
-2-10
Child at 1 yr
i I
Breastmilk Intake per kg BW !
I
I
Cord Blood:Maternal Serum Ratio
Milk:Maternal Serum Ratio
Volume of Distribution
Half-life
i 1 1 1 i 1 1 1
-2-1012 -2-1012
Figure F-14. Sensitivity Coefficients
Sensitivity coefficients from a local sensitivity analysis of maternal serum at delivery, cord blood at delivery, and child serum at
1 and 5 years old. The child was female. Results for a male child were similar (not shown).
BW = body weight; yr = year.
A model developed by the Minnesota Department of Health (MDH model) (Goeden et al., 2019)
was also considered for application to this assessment. This model has a similar model structure
to the chosen model, with single compartments to represent the mother and child and excretion
handled by first-order clearance.
To evaluate the effect of Vd in children, the Vd scaling in the MDH model was integrated into
model shown in Figure F-14. The main effect is to reduce the peak serum levels in children that
occurs due to exposure through breastmilk. Based on root mean squared error, it was determined
that the model with constant Yd had better performance (Table F-l).
/ery
Cord Blood
-2
-1 0 1
Child at 5 yr
~
I1
F-ll
-------
APRIL 2024
Table F-l. Root mean squared error comparison between the baseline model (as applied in
the main risk assessment) and alternative models with features inspired by the MDH
model.
Root Mean Squared Error
Chemical
Baseline Model
Model with Variable Vd Model with Drinking Water
Exposure
PFOA
0.65
1.59 1.27
PFOS
2.48
5.06 4.82
Fromme, 2010
MOBA Cohort
Mogensen, 2015
Fromme, 2010
MOBA Cohort
Mogensen, 2015
Age (yr) Age (yr)
Figure F-15. Predicted Child Serum Levels Compared to Reported Values with Increased
Volume of Distribution in Children as was Implemented in the Minnesota Department of
Health Model
MOBA = Norwegian Mother, Father, and Child Cohort Study.
EPA also implemented exposure based on drinking water consumption in the modified Verner
model to examine the effect on model predictions and especially on the results of the risk
assessment (Figure F-15). Based on root mean squared error, it was determined that the model
with constant exposure relative to bodyweight had better performance than a model that
explicitly adjusts for drinking water consumption (Table F-l). An MCLG based on constant
exposure does not greatly underestimate the risk to populations with greater water consumption
per body weight (e.g., children and lactating women) because the method for calculating the
MCLG from a RfD that assumes constant exposure accounts for the greater drinking water
consumption in these populations.
F-12
-------
APRIL 2024
Age (yr) Age (yr)
Figure F-16. Predicted Child Serum Levels Compared to Reported Values with Constant
Volume of Distribution and Variable Exposure Based on Drinking Water Intake
MOBA = Norwegian Mother, Father, and Child Cohort Study.
F-13
-------
APRIL 2024
Appendix G. Relative Source Contribution
G.l Background
The EPA applies an RSC to the RfD when calculating an MCLG based on noncancer effects or
for carcinogens that are known to act through a nonlinear mode of action to account for the
fraction of an individual's total exposure allocated to drinking water (U.S. EPA, 2000). The EPA
emphasizes that the purpose of the RSC is to ensure that the level of a chemical allowed by a
criterion (e.g., the MCLG for drinking water) or multiple criteria, when combined with other
identified sources of exposure (e.g., diet, ambient and indoor air) common to the population of
concern, will not result in exposures that exceed the RfD. In other words, the RSC is the portion
of total daily exposure equal to the RfD that is attributed to drinking water ingestion (directly or
indirectly in beverages like coffee tea or soup, as well as from transfer to dietary items prepared
with drinking water) relative to other exposure sources; the remainder of the exposure equal to
the RfD is allocated to other potential exposure sources. For example, if for a particular
chemical, drinking water were to represent half of total exposure and diet were to represent the
other half, then the drinking water contribution (or RSC) would be 50%. The EPA considers any
potentially significant exposure source when deriving the RSC.
The RSC is derived by applying the Exposure Decision Tree approach published in the EPA's
Methodology for Deriving Ambient Water Quality Criteria for the Protection of Human Health
(U.S. EPA, 2000). The Exposure Decision Tree approach allows flexibility in the RfD
apportionment among sources of exposure and considers several characteristics of the
contaminant of interest, including the adequacy of available exposure data, levels of the
contaminant in relevant sources or media of exposure, and regulatory agendas (i.e., whether there
are multiple health-based criteria or regulatory standards for the contaminant). The RSC is
developed to reflect the exposure to the U.S. general population or a sensitive population within
the U.S. general population and may be derived qualitatively or quantitatively, depending on the
available data.
A quantitative RSC determination first requires "data for the chemical in question...
representative of each source/medium of exposure and... relevant to the identified population(s)"
(USEPA, 2000). The term "data" in this context is defined as ambient sampling measurements in
the media of exposure, not internal human biomonitoring metrics. More specifically, the data
must adequately characterize exposure distributions including the central tendency and high-end
exposure levels for each source and 95% confidence intervals for these terms (U.S. EPA, 2000).
Frequently, an adequate level of detail is not available to support a quantitative RSC derivation.
When adequate quantitative data are not available, the agency relies on the qualitative
alternatives of the Exposure Decision Tree approach. A qualitatively-derived RSC is an estimate
that incorporates data and policy considerations and thus, is sometimes referred to as a "default"
RSC (U.S. EPA, 2000). Both the quantitative and qualitative approaches recommend a "ceiling"
RSC of 80%) and a "floor" RSC of 20% to account for uncertainties including unknown sources
of exposure, changes to exposure characteristics over time, and data inadequacies (U.S. EPA,
2000).
In cases in which there is a lack of sufficient data describing environmental monitoring results
and/or exposure intake, the Exposure Decision Tree approach results in a recommended RSC of
20%. In the case of MCLG development, this means that 20% of the exposure equal to the RfD
G-l
-------
APRIL 2024
is allocated to drinking water and the remaining 80% is reserved for other potential sources, such
as diet, air, consumer products, etc. This 20% RSC value can be replaced if sufficient data are
available to develop a scientifically defensible alternative value. If scientific data demonstrating
that sources and routes of exposure other than drinking water are not anticipated for a specific
pollutant, the RSC can be raised as high as 80% based on the available data, allowing the
remaining 20% for other potential sources (U.S. EPA, 2000). Applying a lower RSC (e.g., 20%)
is a more conservative approach to public health and results in a lower MCLG.
G.2 Literature Review
In 2019, EPA's Office of Research and Development (ORD) conducted a literature search to
evaluate evidence for pathways of human exposure to PFOA and PFOS. This search was not date
limited and spanned the information collected across the Web of Science, PubMed, and
ToxNet/ToxLine (now ProQuest) databases. An updated literature search was conducted and
captured relevant literature published through March 2021. Literature captured by this search is
housed in EPA's HERO database (https://hero.epa.eov/).
Results of this broad literature search were further distilled to address two questions. First, a
systematic review was conducted to investigate evidence for important PFAS exposure pathways
from indoor environment media including consumer products, household articles, cleaning
products, personal care products, and indoor air and dust (Deluca et al., 2022a). Literature that
reported exposure measures from household media paired with occupant PFAS concentrations in
blood serum was identified. Second, systematic evidence mapping was conducted for literature
reporting measured occurrence of PFAS in exposure media (Holder et al., 2023). This review
focused on real-world occurrences (measured concentrations) primarily in media commonly
related to human exposure (outdoor and indoor air, indoor dust, drinking water, food, food
packaging, articles and products, and soil).
G.2.1 Systematic Review
Deluca et al. (2022b) investigated evidence for important PFAS exposure pathways from indoor
environment media including consumer products, household articles, cleaning products, personal
care products, and indoor air and dust. The authors adapted existing systematic review
methodologies and study evaluation tools to identify and screen exposure studies that presented
concordant data on PFAS occurrence in indoor media and PFAS concentrations in blood or
serum. Studies included in the systematic review report exposure measures from household
media paired with occupant PFAS concentrations in blood serum, focusing on PFOS and seven
other frequently measured PFAS (PFOA, perfluorobutanoic acid (PFBA), perfluorobutane
sulfonate (PFBS), PFDA, PFHxA, PFHxS, and PFNA). Machine learning approaches were used
during the literature scoping and title/abstract screening to prioritize exposure pathways of
interest by automated tagging and to select studies for inclusion using an iterative predictive
screening model. Title/Abstract screening for the PECO criteria identified 486 studies for full
text screening; only 6 studies fully addressed the protocol requirements (Balk et al., 2019; Kim et
al., 2019; Poothong et al., 2019; Byrne et al., 2017; Makey et al., 2017; Wu et al., 2014). The
extraction of exposure measurement data and study characteristics from each included study was
performed in DistillerSR software. Exposure intake calculations were used to estimate a
percentage of occupant serum concentrations that could be attributed to indoor exposure
pathways other than drinking water and diet. The included studies were evaluated using an
G-2
-------
APRIL 2024
approach modified from EPA's IRIS Handbook (U.S. EPA, 2022c). Along with providing
evidence for an estimated range of indoor exposure media's contribution to serum PFAS
concentrations, this systematic review highlights the limited availability of concordant
measurement data from indoor exposure media and participant serum.
The Deluca and coworkers review (2022a) described above focused on indoor pathways and
therefore excluded non-indoor pathways such as surface water or soil. Ninety-seven articles fell
into this excluded group (i.e., PFOS was measured in non-indoor environmental medium). These
97 papers were reviewed for this effort, though are not fully described in this appendix.
G.2.2 Evidence Mapping
Holder et al. (2023) investigated evidence for important pathways of exposure to PFAS by
reviewing literature reporting measured occurrence of PFAS in exposure media. The review
focused on eight PFAS (PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, and PFNA) and
their real-world occurrences primarily in human matrices and media commonly related to human
exposure (outdoor and indoor air, indoor dust, drinking water, food, food packaging, articles and
products, and soil). The initial review identified 3,622 peer-reviewed papers matching these
criteria that were published between 2003-2020. ICF's litstream™ software was used to conduct
title-abstract (TiAb) and full-text screening, and to extract relevant primary data into a
comprehensive evidence database. Parameters of interest included: sampling dates and locations
(focused on locations in the United States, Canada, and Europe), numbers of collection sites and
participants, analytical methods, limits of detection and detection frequencies, and occurrence
statistics.
Detailed data on PFAS occurrence in high-priority household and environmental media from 210
studies were extracted, as well as limited data on human matrices from 422 additional papers.
Published studies of PFAS occurrence became numerous after about 2005 and were most
abundant for PFOA and PFOS. Co-measurements for PFAS occurrence in human matrices plus
other media, while relatively infrequent, were typically for occurrence in food and drinking
water. Most studies found detectable levels of PFAS, and half or more of the limited studies of
indoor air and products detected PFAS in 50% or more of their samples. Levels of PFOS in these
media ranged widely.
Literature search results were categorized into 7 types of exposure pathway categories, including
environmental media, home products/articles/building materials, cleaning products, food
packaging, personal care products, clothing, and specialty products. The environmental media
pathway category included the sub-categories of food, water, air, dust, soil, wastewater, and
landfill. The identified studies were reviewed for this effort, though are not fully described in this
appendix.
G.3 Summary of Potential PFOS Sources
PFOS is a synthetic, fully fluorinated, organic compound that is used in many types of consumer
products and is resistant to metabolic and environmental degradation (U.S. EPA, 2016c). It has
been associated with releases from manufacturing sites, industrial sites, fire/crash training areas,
and industrial and municipal waste sites. PFOS is one of a large group of perfluoroalkyl
substances that are widely used in consumer and industrial products to improve their resistance
G-3
-------
APRIL 2024
to stains, grease, and water. PFOS was a major component of AFFF which were used to
extinguish petroleum-based fires. Most PFOS production in the United States was voluntarily
phased out by its primary manufacturer (3M) between 2000 and 2002. In 2002 and 2007 EPA
took regulatory action under the Toxic Substances Control Act (TSCA) to require that EPA be
notified prior to any future domestic manufacture or importation of PFOS and 270 related PFAS
(U.S. EPA, 2016a). Exposure to PFOS can occur through food, including fish and shellfish,
house dust, air, and contact with consumer products (U.S. EPA, 2016c).
G.3.1 Dietary Sources
Ingestion of food is a potentially significant source of exposure to PFOS and is often claimed to
be the dominant source of exposure for the general population based on early studies that
modeled the relative contributions of various sources among the general populations of North
America and Europe (Fromme et al., 2009; Vestergren and Cousins, 2009; Trudel et al., 2008).
The exposure among adults in western countries is typically estimated to be about 1 ng/kg/day,
but studies on the dietary exposure among the U.S. population are limited (East et al., 2021;
Domingo and Nadal, 2017). The dominance of the food ingestion pathway is attributed to
bioaccumulation in food from environmental emissions, relatively large amounts of foods being
consumed, and high GI uptake (Trudel et al., 2008). However, the estimates are highly uncertain
due to limited data availability, relatively low detection frequencies, and relatively large
differences in composition of diets across geographic locations (EFSA, 2020; Domingo and
Nadal, 2017).
There is currently no comprehensive, nationwide Total Diet Study (TDS) for PFOS that can be
used to draw conclusions about the occurrence and potential risk of PFOS in the U.S. food
supply for the general population. In 2021, the U.S. Food and Drug Administration (FDA)
released PFAS testing results from their first survey of nationally distributed processed foods,
including several baby foods. Results of the survey showed that 164 of the 167 foods tested had
no detectable levels of the PFAS measured. Three food samples had detectable levels of PFAS:
fish sticks (PFOS (33 parts per trillion (ppt)) and PFNA), canned tuna (PFOS (76 ppt) and
PFDA), and protein powder (PFOS (140 ppt)). In another recent FDA study, PFOS was detected
in one sample (baked cod, 98 ppt) out of 94 food samples collected nationally (FDA, 2021). In a
2019 national survey of produce, meats, dairy and grain products, PFOS was detected in three of
the 179 food samples tested (two samples of tilapia, one sample of turkey) (FDA, 2019a, b).
PFOS was also detected in produce samples (collard greens and lettuce) in a 2018 focused study
near a PFAS production plant in the Fayetteville, North Carolina area (FDA, 2018). The sample
size in all of these studies is limited, and thus, the results cannot be used to draw definitive
conclusions about the levels of PFAS in the U.S. food supply more generally (FDA, 2021). In a
2010 study of 31 types of food collected from 5 grocery stores in Texas, PFOS was not detected
in any of the samples (Schecter et al., 2010).
As a component of a scientific evaluation on the risks to human health related to PFAS in food,
the European Food Safety Authority (EFSA) conducted an exposure assessment using
consumption data from the EFSA Comprehensive Food Consumption Database and 69,433
analytical results for 26 PFAS in 1,528 samples of food and beverages obtained from 16
European countries (EFSA, 2020). Samples were collected between the years 2000 and 2016
(74% after 2008), mainly from Norway, Germany, and France. With 92% of the analytical
G-4
-------
APRIL 2024
results below the LOD or LOQ, lower bound dietary exposure estimates were obtained by
assigning zero to values below LOD/LOQ. Median chronic dietary exposures of PFOS for
children and adults were estimated as 1.02 and 0.58 ng/kg-body weight/day, respectively. The
most important contributors for PFOS were "Fish and other seafood," "Eggs and egg products,"
and "Meat and meat products." It is unclear whether the contribution from food contact material
is reflected in the data. The authors determined diet to be the major source of PFAS exposure for
most of the population but noted that dust ingestion and indoor air inhalation may provide
substantial contributions for some individuals.
The 2020 EFSA report highlighted a recent study of aggregate exposure to PFAS from diet,
house dust, indoor air, and dermal contact among Norwegian adults (Poothong et al., 2020).
Dietary exposures were estimated for 61 study participants using food diaries and data on
concentrations from an extensive Norwegian database of concentrations in sixty-eight different
food and drinks (including drinking water). For PFOS, the authors concluded that dietary intake
was by far the greatest contributor to aggregate exposure (contributing 95% of total estimated
PFOS intake), but intake from ingestion of house dust represented the dominant pathway for
some of the top 20% most highly exposed individuals. While the authors did not separately
quantify intake from food and drinking water, an earlier article from the same research group
(Papadopoulou et al., 2017) reported measured concentrations in duplicate diets with median
estimated intake of PFOS approximately 150 times higher from solid food than from liquids.
De Felip et al. (2015) investigated correlations of blood concentrations of PFOS with dietary
intake among Italian women. They estimated daily intake of PFOS based on the reported food
consumption frequencies of specific food items and found strongly significant correlations of
blood levels with consumption of beef, pork, and vegetables (p < 0.01), and moderate correlation
with consumption of fish (p < 0.05).
EPA's Emerging Issues in Food Waste Management Persistent Chemical Contaminants (U.S.
EPA, 2021b) further describes global PFOS and other PFAS occurrence in food items, waste,
and compost, as well as food contact materials, described below (Section G.3.1.2).
G.3.1.1 Fish and Shellfish
PFOS has been shown to bioaccumulate and biomagnify with increasing trophic level in a
variety of freshwater ecosystems (Penland et al., 2020; Xu et al., 2014; Kannan et al., 2005;
Martin et al., 2004) and saltwater ecosystems (Loi et al., 2011; de Vos et al., 2008; Powley et al.,
2008; Houde et al., 2006; Tomy et al., 2004) in North America, Europe, and Asia. PFOS is often
the most abundant PFAS in aquatic organisms, and this high relative abundance is at least
partially explained by the biotransformation of PFOS precursor chemicals into PFOS (Kelly et
al., 2009; Haukas et al., 2007; Kannan et al., 2005; Martin et al., 2004; Tomy et al., 2004).
Higher trophic level organisms have a greater capacity to metabolize PFOS precursor chemicals,
which have been found in lower concentrations in increasing trophic level (Fang et al., 2014;
Kannan et al., 2005; Martin et al., 2004).
Global distribution of PFAS chemicals in tissues of aquatic species has been demonstrated in
studies conducted in freshwater and marine environments across every continent, including
remote regions far from direct sources, such as the high arctic, Antarctica, and oceanic islands
(Houde et al., 2006; Giesy and Kannan, 2001).
G-5
-------
APRIL 2024
EPA collaborates with federal agencies, states, tribes, and other partners to conduct freshwater
fish contamination studies as part of a series of statistically based surveys to produce information
on the condition of U.S. lakes, streams, rivers, and coastal waters. PFOS was detected in nearly
all freshwater fish fillet samples collected during several national studies in rivers and the Great
Lakes (Table G-l).
Table G-l. Summary of EPA national fish tissue monitoring results for PFOS
Reference
Most Commonly
Sampled Species
Site Description
Results
U.S. EPA (2010)
Smallmouth bass
162 urban river sites across
PFOS was the most
Largemouth bass
the United States
commonly detected PFAS
Channel catfish
(out of 13 PFAS).
PFOS was detected in 77
percent of samples.
Maximum detected
concentration 127 ng/g.
U.S. EPA (2015)
Largemouth bass
349 urban and nonurban
PFOS was the most
Smallmouth bass
river sites across the United
commonly detected PFAS
Black crappie
States.
(out of 13 PFAS).
White crappie
PFOS was detected in 99
Walleye/sauger
percent of samples.
Yellow perch
Maximum detected
White bass
concentration 283 ng/g.
Northern pike
Lake trout
Brown trout
Rainbow trout
Brook trout
U.S. EPA (2011a)
Lake trout
157 nearshore sites along
PFOS was the most
Smallmouth bass
the U.S. shoreline of the
commonly detected PFAS
Walleye
Great Lakes
(out of 13 PFAS).
PFOS was detected in 100
percent of samples.
Maximum detected
concentration 80 ng/g;
median 15 ng/g.
U.S. EPA (2016e)
Freshwater Drum
152 nearshore sites along
PFOS was the most
Longnose Sucker
the U.S. shoreline of the
commonly detected PFAS
White Sucker
Great Lakes
(out of 13 PFAS).
Lake Whitefish
PFOS was detected in 100
Northern Pike
percent of samples.
Channel Catfish
Maximum detected
Burbot
concentration 64 ng/g;
Smallmouth Bass
median 11 parts per billion
White Perch
(ppb).
White Bass
Coho Salmon
Rainbow Trout
Chinook Salmon
Yellow Perch
Brown Trout
Lake Trout
Walleye
G-6
-------
APRIL 2024
Guo et al. (2012) measured PFOS in lake trout muscle tissues in Canadian waters of Lake
Superior, Huron, Erie, and Ontario. Average PFOS concentrations correlated with watershed
urbanization, and were 0.85 ng/g, 8.3 ng/g, 27 ng/g, and 46 ng/g wet weight (ww), respectively.
Delinsky et al. (2010) measured PFOS in bluegill, black crappie, and pumpkinseed muscle tissue
in 59 lakes in Minnesota, including four lakes in the Minneapolis-St. Paul metropolitan area.
PFOS was detected in muscle tissues of fish collected in 13 of the 59 lakes, and concentrations
ranged from 1.08 ng/g ww to 52.4 ng/g ww in lakes where it was detected. In the four lakes in
the Minneapolis-St. Paul metropolitan area, PFOS concentrations in fish muscle tissues ranged
from 4.39 ng/g ww to 47.3 ng/g ww.
Penland et al. (2020) measured PFAS concentrations in invertebrates and vertebrates along the
Yadkin - Pee Dee River, in North Carolina and South Carolina in 2015. PFOS was measured in
whole body tissues of snails (6.47 ng/g ww) but was not detected whole body tissues of in Asian
clam, unionid mussels, or crayfish. The highest concentrations in invertebrates were measured in
aquatic insect whole body samples (132.8 ng/g ww) and was hypothesized to result from dietary
uptake of aquatic biofilms. PFOS was measured in muscle tissue of all 11 sampled fish species
and ranged from 11.42 ng/g ww in channel catfish to 37.36 ng/g ww in whitefin shiner. The
highest PFOS concentration that Penland et al. (2020) measured was 482.9 ng/g ww, from the
eggs of a redhorse fish sample.
Houde et al. (2006) measured whole body PFOS in six fish species in Charleston Harbor, South
Carolina, and whole body PFOS in zooplankton and five fish species in Sarasota Bay, Florida.
Charleston Harbor was the more developed of the two sites and had higher overall PFOS
concentrations. Average PFOS concentrations in Charleston Harbor ranged from 19 ng/g in
pinfish to 92 ng/g in spot. In Sarasota Bay, PFOS concentrations averaged 0.2 ng/g in
zooplankton, and ranged from 3.1 ng/g in pigfish to 8.8 ng/g in spotted seatrout, suggesting
evidence of trophic biomagnification.
Zafeiraki et al. (2019) analyzed about 250 samples of marine fish, farmed fish, crustaceans,
bivalves and European eel, caught in Dutch waters or purchased at Dutch markets between 2012
and 2018. Of the 16 PFAS that were analyzed, PFOS was generally detected at a higher
frequency and concentration across the tested species. Shrimps and seabass had the highest
average concentrations of PFOS (each over 4 ng/g ww). PFOS was also detected in mussels,
brown crab, eel (100% detection, ranging from 3.3 to 67 ng/g ww) and several farmed and
marine fish species.
Ruffle et al. (2020) analyzed marine and freshwater finfish and shellfish from four regions of the
United States and seven countries with significant imports to the United States. A total of 70
samples were analyzed for 26 PFAS. PFOS represented 80% to 100% of total PFAS measured in
all but one sample. The highest PFOS concentrations (1.2 ng/g ww to 19.1 ng/g ww) were found
in whitefish, walleye, and yellow perch from the Great Lakes region.
In seafood samples collected for the FDA 2021-22 seafood survey, Young et al. (2022),
analyzed concentrations of 20 PFAS, including PFOS, in 8 of the most highly consumed seafood
products in the U.S. PFOS was detected most frequently (100% of samples; n=10) and at the
highest average concentrations (422.9 ppt) in clams. The study also reported detections in crab
G-7
-------
APRIL 2024
(45.5% of samples; n=l 1; 151.6 ppt average concentration in samples with detections), tuna
(50% of samples; n=10; 86.8 ppt average concentration in samples with detections), tilapia (20%
of samples; n=10; 57.5 ppt average concentration in samples with detections), and cod (60% of
samples; n=10; 62.5 ppt average concentration in samples with detections). PFOS was not
detected above the method detection limits (39 or 45 ppt) in salmon, shrimp, or pollock.
Based on National Oceanic and Atmospheric Administration (NOAA) National Centers for
Ocean and Coastal Science, National Status and Trends Data, PFOS concentrations (in ww) were
not detected in mussels, oysters, and fish liver samples. However, PFOS was detected in marine
fish fillet samples, up to 75.1 ppb (NOAA, 2017).
PFOS concentrations in aquatic biota tend to be higher in areas with known PFAS
manufacturing, industrial use, and/or application of AFFF, which also tend to be more populated
areas and where recreational and subsistence fishing is more common. Several states have
developed fish consumption advisories for PFOS (e.g., Alabama, Wisconsin, Minnesota,
Michigan).
G.3.1.2 Food Contact Materials
The FDA has authorized the use of PFAS in food contact substances due to their non-stick and
grease, oil, and water-resistant properties since the 1960s. There are four categories of products
that may contain PFAS:
• "Non-stick cookware: PFAS may be used as a coating to make cookware non-stick.
• Gaskets, O-Rings, and other parts used in food processing equipment: PFAS may be used
as a resin in forming certain parts used in food processing equipment that require chemical
and physical durability.
• Processing aids: PFAS may be used as processing aids for manufacturing other food
contact polymers to reduce build-up on manufacturing equipment.
• Paper/paperboard food packaging: PFAS may be used as grease-proofing agents in fast-
food wrappers, microwave popcorn bags, take-out paperboard containers, and pet food
bags to prevent oil and grease from foods from leaking through the packaging." (FDA,
2020)
Paper products used for food packaging are often treated with PFAS for water and grease
resistance. In previous testing, sandwich wrappers, french-fry boxes, and bakery bags were all
been found to contain PFAS (Schreder and Dickman, 2018). Older generation PFAS (e.g.,
PFOA, PFOS) were manufactured and used in products for decades, and the bulk of the
information available on PFAS toxicity relates to the older compounds. However, because
newer-generation PFAS are more mobile than their predecessors, they migrate more readily into
food. In 2016, the FDA deauthorized the remaining uses of long-chain "C8" PFAS in food
packaging, which are therefore, no longer used in food contact applications sold in the United
States (FDA, 2020).
Under FDA rules, there are dozens of PFAS chemicals still approved for food contact materials.
In 2020, Safer Chemicals Healthy Families and Toxic-Free Future co-published a report where
78 samples of food packaging including take-out containers and deli or bakery paper, among
others, were collected from 20 stores in 12 states (Schreder and Dickman, 2018). An independent
laboratory tested the samples for fluorine. The utility of measuring fluorine content is limited
G-8
-------
APRIL 2024
because it does not allow for identification and quantification of individual PFAS; however, this
method can be used to determine if a food-packaging material has been treated with PFAS. Over
10% of 78 samples tested contained PFAS. The sample size was not large enough to indicate
how widespread the use of PFAS in food packaging is at this time. However, the study
demonstrated that PFAS in food packaging is still a concern, especially for fiber bowls and trays.
Several other relatively recent studies found PFAS in fast-food packaging collected in the United
States, China, or Europe. The data from the references described below and other publications
likely contributed to the recent regulatory actions of the FDA and a number of states to ban or
restrict the presence of PFAS in food contact materials (Keller & Heckman LLP, 2021). Schaider
at al. (2017) collected 407 samples of food contact papers, beverage containers, and paperboard
boxes from locations throughout the United States. As was the case with Schreder & Dickman
(2018), inorganic fluoride was the analyte for the initial analysis. 56% of the dessert and bread
wrappers were positive for fluoride, 38% of the sandwich and burger wrappers, and 20% of the
paper-board containers. None of the 30 (hot/cold) paper beverage cups tested positive in contrast
to 16% of beverage containers (milk/juice) made from other materials. Generally, food contact
papers had higher fluoride detection frequencies than food contact paperboard.
An analysis of popcorn bags, snack bags, and sandwich bags purchased in 2018 from
international vendors and grocery stores in the United States found no evidence of PFOS at
concentrations above the LOD (0.63 ng/g paper) (Monge Brenes et al., 2019). The authors
presented these results as evidence of a reduction in PFOS concentrations in microwave
packaging between 2005 and 2018. In an analysis of microwave popcorn bags from around the
world, Zabaleta et al. (2017) reported no measurable concentrations of any PFSA, including
PFOS, in any of the samples. In a second study, Zabaleta et al. (2020) looked at PFAS in 25
paper- and paperboard packaging materials primarily collected in Spain. Again, no PFSAs,
including PFOS, were found above the level of detection. The packaging materials with the
largest number of detectable analytes was a popcorn bag from China and the inside paper lining
from three individual pet food products, which contained a spectrum of C-3 to C10
perfluorinated carboxylates.
G.3.2 Consumer Product Uses
An early investigation of consumer exposure to PFOS by Trudel et al. (2008) used mechanistic
modeling together with information on product-use habits to estimate exposures from mill-
treated carpets and impregnated clothing. The authors concluded that contact with consumer
products represents less than 1% of total exposure to PFOS, but also pointed out that because
carpets have a relatively long lifetime, the exposure is expected to continue long after cessation
of use of PFOS in carpet treatments. Liu et al. (2014) also investigated trends in PFAS content of
household goods between 2007 and 2011. They reported a decrease in the availability of
consumer products that contain PFOS is declining but were still able to find products that
contained PFOS. In an analysis of 52 European products collected between 2014-2016, Borg
and Ivarrson (2017) reported that PFSAs were rarely detected in the samples; PFOS was the only
PFSA detected and was only present in one sample, a microwave popcorn bag. Notably, the
authors specifically targeted products that were known or suspected to contain PFAS in their
analyses.
G-9
-------
APRIL 2024
In contrast, Kotthoff et al. (2015) reported broad detection of PFOS in a 2010 sampling effort
that collected 115 European consumer products, including carpets, leather, outdoor materials,
cooking materials, and others. PFOS was detected in all but two sample types, often at the
highest median concentration compared to other PFSAs. However, PFSAs were detected at
concentrations often several orders of magnitude lower than perfluorinated carboxylic acids
(PFCAs) and fluorotelomers. The products with the highest concentrations of total PFAS
included ski wax (median concentration of 1.6 |ig/kg), leather products (maximum concentration
of 5.6 |ig/m2), and outdoor materials (median concentration of 9.5 |ig/m2), PFOS was the most
frequently and abundantly detected PFAS in paper-based cooking materials. PFOS has also been
detected in textile samples of outdoor apparel from Europe and Asia (van der Veen et al., 2020;
Gremmel et al., 2016). PFOS was detected in one-third of the jackets tested by Gremmel et al.
(2016) at relatively low concentrations ranging from 0.01 [j,g/m2-0.59 (J,g/m2. Interestingly, while
the concentrations of almost all individual PFAS and total PFAS concentrations increased when
the textiles were subjected to weathering (i.e., increased ultraviolet light radiation, temperature,
and humidity for 300 hours to mimic the average lifespan of outdoor apparel), PFOS
concentrations declined after weathering in the one sample that exceeded European Commission
restrictions on PFOS content of coated materials (1 |ig/m2) (van der Veen et al., 2020).
G.3.3 Indoor Dust
Several studies suggest that PFOS and its precursors in indoor dust may be an important
exposure source for some individuals (Poothong et al., 2020; NJDWQI, 2018; Gebbink et al.,
2015; Shoeib et al., 2011). PFOS is generally a dominant ionic PFAS constituent in household
dust, frequently occurring above detection limits and at relatively high concentrations in all or
most samples (Poothong et al., 2020; Kim et al., 2019; Byrne et al., 2017; Makey et al., 2017;
Wu et al., 2014; Fraser et al., 2013; Shoeib et al., 2011).
PFOS was measured at the second highest concentrations (geometric mean concentrations
ranging from 29.0 ng/g-34.6 ng/g) and frequencies (ranging from 85%-87% detected) in dust
sampled from Californian households. Similarly, PFOS was found at the highest levels (mean
concentration of 3.06 ng/g) of 15 PFAS measured in dust samples taken from households in
Seoul, Republic of Korea (Kim et al., 2019). One study of Alaska Natives noted that PFOS was
the predominant compound in dust samples (Byrne et al., 2017).
G.3.4 Ambient Air
Air concentrations of PFOS in the atmosphere vary widely across the globe. Areas near
wastewater treatment facilities, waste incinerators, and landfills can be point sources of PFOS to
air (Ahrens et al., 2011). In an urban area in Albany, NY, perfluorinated acids were measured in
air samples in both the gas and particulate phase in May and July 2006 (Kim and Kannan, 2007).
PFOS in the gas phase had a mean concentration of 1.70 pg/m3 (range: 0.94-3.0 pg/m3) and in
the particulate phase had a mean concentration of 0.64 pg/m3 (range: 0.35-1.16 pg/m3).
However, at Lake Ontario, concentrations of PFOS in the particulate phase measured in air
samples over the lake were higher (Boulanger et al., 2005). The mean concentration of PFOS at
Lake Ontario was 6.4 ±3.3 pg/m3; with a range of concentrations from detected to 8.1 pg/m3. In
an urban area in Minneapolis, Minnesota, PFOS was measured in both the particulate and gas
G-10
-------
APRIL 2024
phase (MPCA, 2008). PFOS in the particulate phase ranged from 2.1 pg/m3 -7.9 pg/m3 and the
gas phase ranged from 1.8 pg/m3-5.0 pg/m3 across the five samples.
In Canada, PFOS air concentrations measured in 2009 showed widespread distribution with
remote sites having similar concentrations to urban sites (ECCC, 2018). Using passive samplers,
PFOS concentrations were detected in Toronto, Ontario (8 pg/m3), an agricultural site in
Saskatchewan (5 pg/m3), Whistler, British Columbia (4 pg/m3), and Alert, N Nunavut (2 pg/m3)
(ECCC, 2018).
Other reported concentrations of PFOS in air samples from Sydney, Florida (3.4 pg/m3), Tudor
Hill, Bermuda (6.1 pg/m3), Malin Head, Ireland (3.3 pg/m3), and Hilo, Hawaii (6.6 pg/m3) are
similar to the concentrations reported in Canada (ECCC, 2018) and Japan (Sasaki et al., 2003).
The annual geometric mean concentration of PFOS in air samples collected monthly from 2001-
2002 in the town of Oyamazaki and Fukuchiyama City were 5.3 and 0.6 pg/m3, respectively
(Sasaki et al., 2003).
Across Europe, PFOS air concentrations were reported to be variable. In the particulate phase
PFOS concentrations ranged from <1.8 pg/m3-46 pg/m3 (Martin et al., 2004). Most locations
had low (~1 pg/m3-2 pg/m3) to less than the reported Minimum Detection Limit (MDL) and
included Hazelrigg, United Kingdom, Kjeller Norway, and Mace Head, Ireland (Barber et al.,
2007). The highest concentrations were reported in Manchester, United Kingdom. Similarly,
high concentrations, 150 pg/m3 for were reported Paris, France (ECCC, 2018).
Even in the Arctic, PFOS, its precursors, and degradation products, have been detected in air
samples in Resolute Bay, Nunavut, Canada, during the summer of 2004 (Stock et al., 2007).
PFOS in the filter samples were 1-2 orders of magnitude greater than other compounds, with a
mean concentration of 5.9 pg/m3. These concentrations are greater than PFOS concentrations
measured in the particle phase of air samples measured in Zeppelinstasjonen, Svalbard, Norway
(Butt et al., 2010). PFOS was measured in September and December, 2006 and August and
December, 2007, with mean concentrations of 0.11 pg/m3 (range: 0.03 pg/m3-0.50 pg/m3) and
0.18 pg/m3 (range: 0.02 pg/m3-0.97 pg/m3), respectively.
G.3.5 Other Possible Exposure Sources
PFOS has also been detected in soils and dust from carpets and upholstered furniture in homes,
offices, and vehicles. Incidental exposure from soils and dust is an important exposure route,
particularly for small children because of their increased level of hand-to-mouth behaviors
compared to adults. Also, the levels in soils and surface waters can affect the concentrations in
local produce, meat/poultry, dairy products, fish, and particulates in the air.
G.4 Recommended RSC
EPA followed the Exposure Decision Tree approach to determine the RSC for PFOS, as outlined
in Figure G-l (U.S. EPA, 2000). EPA first identified several potential populations of concern
(Box 1): pregnant women and their developing fetuses, infants, children, lactating women, and
women of childbearing age. However, limited information was available regarding specific
exposure of these populations to PFOS in different environmental media. EPA considered
exposures in the general U.S. population as likely being applicable to the majority of these
G-ll
-------
APRIL 2024
populations. Second, EPA identified several relevant PFOS exposures and pathways (Box 2),
including dietary consumption, incidental oral, inhalation, or dermal exposure via dust, consumer
products, and soil, and inhalation exposure via ambient air. Several of these may be potentially
significant exposure sources. Third, EPA determined that there was inadequate quantitative data
to describe the central tendencies and high-end estimates for all of the potentially significant
sources (Box 3). For example, studies from the U.S. indicate that dust may be a significant
source of exposure to PFOS. Although several studies report PFOS detections in consumer
products, most examined samples from specific locations that may not be nationally
representative. Therefore, the agency does not have adequate quantitative data to describe the
central tendency and high-end estimate of exposure for this potentially significant source in the
U.S. population. However, the agency determined there were sufficient data, physical/chemical
property information, fate and transport information, and/or generalized information available to
characterize the likelihood of exposure to relevant sources (Box 4). Notably, based on the studies
summarized in the sections above, there are significant known or potential uses/sources of PFOS
other than drinking water (Box 6), though there is not information available on each source to
make a characterization of exposure (Box 8A). For example, there are several studies from the
U.S., Canada, and Europe indicating that PFOS may occur in multiple food products, most
notably, seafood. The physico-chemical properties of PFOS indicate that it is likely to
bioaccumulate. However, the available evidence about the occurrence of PFOS in other food
types (e.g., eggs, meats, vegetables, fruit) is less substantive; the majority of studies examined
very few samples (i.e., n=l-5) of various food products and a nationally representative total diet
study does not exist. Therefore, it is not possible to determine whether food or other types of
media can be considered a major or minor contributor to total PFOS exposure. Given these
G-12
-------
APRIL 2024
considerations, following recommendations of the Exposure Decision Tree (U.S. EPA, 2000),
EPA recommends an RSC of 20% (0.20) for PFOS.
Figure G-l. Application of the Exposure Decision Tree (U.S. EPA, 2000) for PFOS
Green highlighted boxes indicate selections made at each branch of the Decision Tree.
POD = point of departure; RiD = reference dose; TIF = uncertainty factor.
G-13
-------
APRIL 2024
References
3M. (2000). Determination of serum half-lives of several fluorochemicals, Interim report #1, June 8, 2000 [TSCA Submission],
In TSCA 8(e) Supplemental Notice for Sulfonate-based and Carboxylic-based Fluorochemicals-DocketNumbers
8EHQ-1180-373; 8EHQ-1180-374; 8EHQ-0381- 0394; 8EHQ-0598-373. (8EHQ-80-373. 8EHQ-0302-00373.
89(811844Q). AR226-0611).
3M. (2002). Determination of serum half-lives of several fluorochemicals, Interim report #2, January 11, 2002 [TSCA
Submission], In TSCA 8(e) Supplemental Notice for Sulfonate-based and Carboxylic-based Fluorochemicals-
DocketNumbers 8EHQ-1180-373; 8EHQ-1180-374; 8EHQ-0381- 0394; 8EHQ-0598-373. (8EHQ-80-373. 8EHQ-
0302-00373. 89(811844Q). AR226-1086).
Abdullah Soheimi, SS; Abdul Rahman, A; Abd Latip, N; Ibrahim, E; Sheikh Abdul Kadir, SH. (2021). Understanding the impact
of perfluorinated compounds on cardiovascular diseases and their risk factors: A meta-analysis study [Review].
International Journal of Environmental Research and Public Health 18: 8345. http://dx.doi.org/10.3390/iierphl8168345
Abraham, K; Mielke, H; Fromme, H; Volkel, W; Menzel, J; Peiser, M, et al. (2020). Internal exposure to perfluoroalkyl
substances (PFASs) and biological markers in 101 healthy 1-year-old children: associations between levels of
perfluorooctanoic acid (PFOA) and vaccine response. Archives of Toxicology 94: 2131-2147.
Agier, L; Basagana, X; Maitre, L; Granum, B; Bird, PK; Casas, M, et al. (2019). Early-life exposome and lung function in
children in Europe: an analysis of data from the longitudinal, population-based HELIX cohort. The Lancet Planetary
Health 3: e81-e92.
Ahrens, L. (2011). Polyfluoroalkyi compounds in the aquatic environment: a review of their occurrence and fate [Review],
Journal of Environmental Monitoring 13: 20-31. littp: //dx. doi. org/1.0.1.03 9/c0em00373e
Ahrens, L; Shoeib, M; Harner, T, om; Lee, S; Guo, R, ui; Reiner, EJ. (2011). Wastewater Treatment Plant and Landfills as
Sources of Polyfluoroalkyi Compounds to the Atmosphere. Environmental Science and Technology 45: 8098-8105.
http://dx.doi.org/10.1021/esl036173
Aimuzi, R; Luo, K; Chen, Q; Wang, H; Feng, L; Ouyang, F; Zhang, J. (2019). Perfluoroalkyl and polyfluoroalkyi substances and
fetal thyroid hormone levels in umbilical cord blood among newborns by prelabor caesarean delivery. Environment
International 130: 104929.
Aimuzi, R; Luo, K; Huang, R; Huo, X; Nian, M; Ouyang, F, et al. (2020). Perfluoroalkyl and polyfluroalkyl substances and
maternal thyroid hormones in early pregnancy. Environmental Pollution 264: 114557.
Ait Bamai, Y; Goudarzi, H; Araki, A; Okada, E; Kashino, I; Miyashita, C; Kishi, R. (2020). Effect of prenatal exposure to per-
and polyfluoroalkyi substances on childhood allergies and common infectious diseases in children up to age 7 years:
The Hokkaido study on environment and children's health. Environment International 143: 105979.
Alam, MN; Han, X; Nan, B; Liu, L; Tian, M; Shen, H; Huang, Q. (2021). Chronic low-level perfluorooctane sulfonate (PFOS)
exposure promotes testicular steroidogenesis through enhanced histone acetylation. Environmental Pollution 284:
117518. http://dx.doi.org/10.1.016/i.envpol.2021.. 1.1.751.8
Alderete, TL; Jin, R; Walker, DI; Valvi, D; Chen, Z; Jones, DP, et al. (2019). Perfluoroalkyl substances, metabolomic profiling,
and alterations in glucose homeostasis among overweight and obese Hispanic children: A proof-of-concept analysis.
Environment International 126: 445-453.
Alexander, BH; Olsen, GW. (2007). Bladder cancer in perfluorooctanesulfonyl fluoride manufacturing workers. Annals of
Epidemiology 17: 471-478. http://dx.doi.Org/10.lQ16/j.amiepidem.2QQ7.Ql.Q36
Alexander, BH; Olsen, GW; Burris, JM; Mandel, JH; Mandel, JS. (2003). Mortality of employees of a perfluorooctanesulphonyl
fluoride manufacturing facility. Occupational and Environmental Medicine 60: 722-729.
littp: //dx. doi. org/1.0.1136/oeni. 60.1.0.'722
Ammitzbell, C; Bornsen, L; Petersen, ER; Oturai, AB; Sendergaard, HB; Grandjean, P; Sellebjerg, F. (2019). Perfluorinated
substances, risk factors for multiple sclerosis and cellular immune activation. Journal of Neuroimmunology 330: 90-95.
Anand-Ivell, R; Cohen, A; Nargaard-Pedersen, B; Jonsson, BAG; Bonde, JP; Hougaard, DM, et al. (2018). Amniotic fluid INSL3
measured during the critical time window in human pregnancy relates to cryptorchidism, hypospadias, and phthalate
load: a large case-control study. Frontiers in Physiology 9: 406.
Andersen, CS; Fei, C; Gamborg, M; Nohr, EA; Serensen, TI; Olsen, J. (2010). Prenatal exposures to perfluorinated chemicals
and anthropometric measures in infancy. American Journal of Epidemiology 172: 1230-1237.
littp: //dx. doi. org/1.0.1.093/aj e/kwq2 8 9
Andersen, ME; Clewell, HJ; Tan, YM; Butenhoff, JL; Olsen, GW. (2006). Pharmacokinetic modeling of saturable, renal
resorption of perfluoroalkylacids in monkeys—probing the determinants of long plasma half-lives. Toxicology 227:
156-164. http://dx.doi.Org/10.1016/j.tox.2006.08.004
Andersson, AG; Lundgren, A; Xu, Y; Nielsen, C; Lindh, CH; Pineda, D, et al. (2023). High Exposure to Perfluoroalkyl
Substances and Antibody Responses to SARS-CoV-2 mRNA Vaccine-an Observational Study in Adults from
Ronneby, Sweden. Environmental Health Perspectives 131: 87007. http://dx.doi.org/10.1.289/.EHP1.1.847
Apelberg, BJ; Goldman, LR; Calafat, AM; Herbstman, JB; Kuklenyik, Z; Heidler, J, et al. (2007a). Determinants of fetal
exposure to polyfluoroalkyi compounds in Baltimore, Maryland. Environmental Science and Technology 41: 3891-
3897. http: //dx. dot org/1.0.1.021. /es070091.1.
R-l
-------
APRIL 2024
Apelberg, BJ; Witter, FR; Herbstman, IB; Calafat, AM; Halden, RU; Needham, LL; Goldman, LR. (2007b). Cord serum
concentrations of perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) in relation to weight and size at
birth. Environmental Health Perspectives 115: 1670-1676. http://dx.doi.org/10.1289/ehp. 1.0334
Arbuckle, T; Kubwabo, C; Walker, M; Davis, K; Lalonde, K; Kosarac, I, et al. (2013). Umbilical cord blood levels of
perfluoroalkyl acids and polybrominated flame retardants. International Journal of Hygiene and Environmental Health
216: 184-194.
Arbuckle, TE; Macpherson, S; Foster, WG; Sathyanarayana, S; Fisher, M; Monnier, P, et al. (2020). Prenatal Perfluoroalkyl
Substances and Newborn Anogenital Distance in a Canadian Cohort. Reproductive Toxicology 94: 31-39.
Argus Research Laboratories. (2000). Support: Oral (stomach tube) developmental toxicity study of PFOS in rabbits with
attachments and cover letter dated 0425000. (EPA/OTS; Doc #8EHQ-0800-0373).
Arrebola, JP; Ramos, JJ; Bartolome, M; Esteban, M; Huetos, O; Canas, Al, et al. (2019). Associations of multiple exposures to
persistent toxic substances with the risk of hyperuricemia and subclinical uric acid levels in BIOAMBIENT.ES study.
Environment International 123: 512-521.
Ashley-Martin, J; Dodds, L; Arbuckle, TE; Bouchard, MF; Fisher, M; Morriset, AS, et al. (2017). Maternal concentrations of
perfluoroalkyl substances and fetal markers of metabolic function and birth weight. American Journal of Epidemiology
185: 185-193.
Ashley-Martin, J; Dodds, L; Arbuckle, TE; Morisset, AS; Fisher, M; Bouchard, MF, et al. (2016). Maternal and Neonatal Levels
of Perfluoroalkyl Substances in Relation to Gestational Weight Gain. International Journal of Environmental Research
and Public Health 13: 146.
ATSDR. (2021). Toxicological profile for perfluoroalkyls [ATSDR Tox Profile], Atlanta, GA: U.S. Department of Health and
Human Services, Public Health Service. http://dx.doi.Org/10.15620/cdc:59198
Attanasio, R. (2019). Sex differences in the association between perfluoroalkyl acids and liver function in US adolescents:
Analyses ofNHANES 2013-2016. Environmental Pollution 254: 113061.
Audet-Delage, Y; Ouellet, N; Dallaire, R; Dewailly, E; Ayotte, P. (2013). Persistent organic pollutants and transthyretin-bound
thyroxin in plasma of inuit women of childbearing age. Environmental Science and Technology 47: 13086-13092.
Averina, M; Brox, J; Huber, S; Furberg, AS. (2021). Exposure to perfluoroalkyl substances (PFAS) and dyslipidemia,
hypertension and obesity in adolescents. The Fit Futures study. Environmental Research 195: 110740.
http://dx.doi.org/10.1016/i.envres.2021.. 1.1.0740
Averina, M; Brox, J; Huber, S; Furberg, AS; Serensen, M. (2019). Serum perfluoroalkyl substances (PFAS) and risk of asthma
and various allergies in adolescents. The Tromse study Fit Futures in Northern Norway. Environmental Research 169:
114-121.
Aylward, LL; Hays, SM; Kirman, CR; Marchitti, SA; Kenneke, JF; English, C, et al. (2014). Relationships of chemical
concentrations in maternal and cord blood: a review of available data [Review], Journal of Toxicology and
Environmental Health, Part B: Critical Reviews 17: 175-203. http://dx.doi.org/10.1080/10937404.2Q14.884956
Bach, C; Matthiesen, B; Olsen; Henriksen, B. (2018). Conditioning on parity in studies of perfluoroalkyl acids and time to
pregnancy: an example from the danish national birth cohort. Environmental Health Perspectives 126: 117003.
Bach, CC; Bech, BH; Nohr, EA; Olsen, J; Matthiesen, NB; Bonefeld-Jergensen, EC, et al. (2016). Perfluoroalkyl acids in
maternal serum and indices of fetal growth: The Aarhus Birth Cohort. Environmental Health Perspectives 124: 848-
854.
Bach, CC; Liew, Z; Bech, BH; Nohr, EA; Fei, C; Bonefeld-Jorgensen, EC, et al. (2015). Perfluoroalkyl acids and time to
pregnancy revisited: An update from the Danish National Birth Cohort. Environmental Health: A Global Access
Science Source 14: 59. http://dx.doi.Org/10.l 1.86/s 1.2940-01.5-0040-9
Balk, FGP; Winkens Pixtz, K; Ribbenstedt, A; Gomis, MI; Filipovic, M; Cousins, IT. (2019). Children's exposure to
perfluoroalkyl acids - a modelling approach. Environmental Science: Processes & Impacts 21: 1875-1886.
http://dx.doi.org/10.1039/c9em00323a
Bangma, J; Eaves, LA; Oldenburg, K; Reiner, JL; Manuck, T; Fry, RC. (2020). Identifying Risk Factors for Levels of Per- and
Polytluoroalkyl Substances (PFAS) in the Placenta in a High-Risk Pregnancy Cohort in North Carolina. Environmental
Science and Technology 54: 8158-8166.
Bao, WW; Qian, ZM; Geiger, SD; Liu, E; Liu, Y; Wang, SQ, et al. (2017). Gender-specific associations between serum isomers
of perfluoroalkyl substances and blood pressure among Chinese: Isomers of C8 Health Project in China. Science of the
Total Environment 607-608: 1304-1312.
Barber, JL; Berger, U; Chaemfa, C; Huber, S; Jahnke, A; Temme, C; Jones, KC. (2007). Analysis of per- and polyfluorinated
alkyl substances in air samples from Northwest Europe. Journal of Environmental Monitoring 9: 530-541.
http://dx.doi.org/10.1039/b701417a
Bartell, SM; Calafat, AM; Lyu, C; Kato, K; Ryan, PB; Steenland, K. (2010). Rate of Decline in Serum PFOA Concentrations
after Granular Activated Carbon Filtration at Two Public Water Systems in Ohio and West Virginia. Environmental
Health Perspectives 118: 222-228. fattp: //dx. dot org/1.0.1.2 89/ehp.0901.252
Batzella, E; Girardi, P; Russo, F; Pitter, G; Da Re, F; Fletcher, T; Canova, C. (2022a). Perfluoroalkyl substance mixtures and
cardio-metabolic outcomes in highly exposed male workers in the Veneto Region: A mixture-based approach.
Environmental Research 212: 113225. http://dx.doi.Org/10.1016/i.envres.2022.l 1.3225
R-2
-------
APRIL 2024
Batzella, E; Zare Jeddi, M; Pitter, G; Russo, F; Fletcher, T; Canova, C. (2022b). Associations between Mixture of Perfluoroalkyl
Substances and Lipid Profile in a Highly Exposed Adult Community in the Veneto Region. International Journal of
Environmental Research and Public Health 19. http://dx.doi.org/10.3390/iierphl91912421
Beach, SA; Newsted, JL; Coady, K; Giesy, JP. (2006). Ecotoxicological evaluation of perfluorooctanesulfonate (PFOS)
[Review], Reviews of Environmental Contamination and Toxicology 186: 133-174. http://dx.doi. org/1.0.1.007/0-3 87-
32883-1. 5
Beck, IH; Timmermann, CAG; Nielsen, F; Schoeters, G; Jehnk, C; Kyhl, HB, et al. (2019). Association between prenatal
exposure to perfluoroalkyl substances and asthma in 5-year-old children in the Odense Child Cohort. Environmental
Health: A Global Access Science Source 18: 97.
Beesoon, S; Martin, JW. (2015). Isomer-Specific Binding Affinity of Perfluorooctanesulfonate (PFOS) and Perfluorooctanoate
(PFOA) to Serum Proteins. Environmental Science and Technology 49: 5722-5731.
Beesoon, S; Webster, GM; Shoeib, M; Harner, T, om; Benskin, JP; Martin, JW. (2011). Isomer Profiles of Perfluorochemicals in
Matched Maternal, Cord, and House Dust Samples: Manufacturing Sources and Transplacental Transfer.
Environmental Health Perspectives 119: 1659-1664. http://dx.doi.org/10.1289/ehp. 1.003265
Bell, EM; Yeung, EH; Ma, W; Kannan, K; Sundaram, R; Smarr, MM; Buck Louis, GM. (2018). Concentrations of endocrine
disrupting chemicals in newborn blood spots and infant outcomes in the upstate KIDS study. Environment International
121: 232-239.
Benskin, JP; Muir, DCG; Scott, BF; Spencer, C; De Silva, AO; Kylin, H, et al. (2012). Perfluoroalkyl Acids in the Atlantic and
Canadian Arctic Oceans. Environmental Science and Technology 46: 5815-5823. http://dx.doi.org/10.1.021./es300S78x
Berg, V; Nest, TH; Hansen, S; Elverland, A; Veyhe, AS; Jorde, R, et al. (2015). Assessing the relationship between
perfluoroalkyl substances, thyroid hormones and binding proteins in pregnant women; a longitudinal mixed effects
approach. Environment International 77: 63-69. http://dx.doi.org/10.1.01.6/j.eiiviiit.201.5.01. .007
Berg, V; Nest, TH; Pettersen, RD; Hansen, S; Veyhe, AS; Jorde, R, et al. (2017). Persistent organic pollutants and the association
with maternal and infant thyroid homeostasis: a multipollutant assessment. Environmental Health Perspectives 125:
127-133.
Berk, M; Williams, LJ; Andreazza, A; Pasco, JA; Dodd, S; Jacka, FN, et al. (2014). Pop, heavy metal and the blues: secondary
analysis of persistent organic pollutants (POP), heavy metals and depressive symptoms in the NHANES National
Epidemiological Survey. British Medical Journal Open 4: e005142.
Biesemeier, J; Harris, D. (1974). Eye and skin irritation report on sample T-1117. (Project No. 4102871). Madison, WI: WARF
Institute Inc.
Bjerregaard-Olesen, C; Bach, CC; Long, M; Wielsee, M; Bech, BH; Henriksen, TB, et al. (2019). Associations of Fetal Growth
Outcomes with Measures of the Combined Xenoestrogenic Activity of Maternal Serum Perfluorinated Alkyl Acids in
Danish Pregnant Women. Environmental Health Perspectives 127: 17006.
Blake, BE; Pinney, SM; Hines, EP; Fenton, SE; Ferguson, KK. (2018). Associations between longitudinal serum perfluoroalkyl
substance (PFAS) levels and measures of thyroid hormone, kidney function, and body mass index in the Fernald
Community Cohort. Environmental Pollution 242: 894-904.
Blomberg, AJ; Shih, YH; Messerlian, C; Jergensen, LH; Weihe, P; Grandjean, P. (2021). Early-life associations between per-
and polyfluoroalkyi substances and serum lipids in a longitudinal birth cohort. Environmental Research 200: 111400.
http://dx.doi.org/10.1.01.6/j. en.vres.2021.. 1.1.1.400
Bogdanska, J; Borg, D; Sundstrom, M; Bergstrom, U; Halldin, K; Abedi-Valugerdi, M, et al. (2011). Tissue distribution of 35S-
labelled perfluorooctane sulfonate in adult mice after oral exposure to a low environmentally relevant dose or a high
experimental dose. Toxicology 284: 54-62. http://dx .doi.org/1.0.1.01.6/j.tox.201.1..03.01.4
Bonefeld-Jorgensen, EC; Long, M; Bossi, R; Ayotte, P; Asmund, G; Krixger, T, et al. (2011). Perfluorinated compounds are
related to breast cancer risk in Greenlandic Inuit: a case control study. Environmental Health: A Global Access Science
Source 10: 88. http: //dx. dot org/1.0.1.1.86/ 1476-069X-1.0-88
Bonefeld-Jergensen, EC; Long, M; Fredslund, SO; Bossi, R; Olsen, J. (2014). Breast cancer risk after exposure to perfluorinated
compounds in Danish women: a case-control study nested in the Danish National Birth Cohort. Cancer Causes and
Control 25: 1439-1448. http://dx.doi.org/10.1007/slQ552-014-0446-7
Borg, D; Bogdanska, J; Sundstrom, M; Nobel, S; Hakansson, H; Bergman, A, et al. (2010). Tissue distribution of (35)S-labelled
perfluorooctane sulfonate (PFOS) in C57B1/6 mice following late gestational exposure. Reproductive Toxicology 30:
558-565. http://dx.doi.Org/10.1016/i.reprotox.2010.07.004
Borg, D; Ivarsson, J. (2017). Analysis of PFASs and TOF in Products. (TemaNord 2017:543). Nordic Council of Ministers.
https://norden.diva-portal.Org/smash/get/diva2:l 118439/FULLTEXTOl.pdf
Borghese, MM; Liang, CL; Owen, J; Fisher, M. (2022). Individual and mixture associations of perfluoroalkyl substances on liver
function biomarkers in the Canadian Health Measures Survey. Environmental Health 21: 85.
http://dx.doi.org/10.1186/sl2940-022-00892-6
Borghese, MM; Walker, M; Helewa, ME; Fraser, WD; Arbuckle, TE. (2020). Association of perfluoroalkyl substances with
gestational hypertension and preeclampsia in the MIREC study. Environment International 141: 105789.
Boulanger, B; Peck, AM; Schnoor, JL; Hornbuckle, KC. (2005). Mass budget of perfluorooctane surfactants in Lake Ontario.
Environmental Science and Technology 39: 74-79. http://dx.doi.org/10.1.021./es049044o
Braun, JM; Chen, A; Romano, ME; Calafat, AM; Webster, GM; Yolton, K; Lanphear, BP. (2016). Prenatal perfluoroalkyl
substance exposure and child adiposity at 8 years of age: The HOME study. Obesity 24: 231-237.
R-3
-------
APRIL 2024
Braun, JM; Kalkbrenner, AE; Just, AC; Yolton, K; Calafat, AM; Sjodin, A, et al. (2014). Gestational exposure to endocrine-
disrupting chemicals and reciprocal social, repetitive, and stereotypic behaviors in 4- and 5-year-old children: the
HOME study. Environmental Health Perspectives 122: 513-520.
Breusch, TS; Pagan, AR. (1979). A simple test for heteroscedasticity and random coefficient variation. 47: 1287-1294.
http://dx.doi.org/10.2307/1911963
Brochot, C; Casas, M; Manzano-Salgado, C; Zeman, FA; Schettgen, T; Vrijheid, M; Bois, FY. (2019). Prediction of maternal and
foetal exposures to perfluoroalkyl compounds in a Spanish birth cohort using toxicokinetic modelling. Toxicology and
Applied Pharmacology 379: 114640.
Buck, CO; Eliot, MN; Kelsey, KT; Calafat, AM; Chen, A; Ehrlich, S, et al. (2018). Prenatal exposure to perfluoroalkyl
substances and adipocytokines: the HOME Study. Pediatric Research 84: 854-860.
Buck Louis, GM; Chen, Z; Schisterman, EF; Kim, S; Sweeney, AM; Sundaram, R, et al. (2015). Perfluorochemicals and human
semen quality: The LIFE Study. Environmental Health Perspectives 123: 57-63. http://dx.doi.org/10.1289/efap. 1.307621.
Buck Louis, GM; Sapra, KJ; Barr, DB; Lu, Z; Sundaram, R. (2016). Preconception perfluoroalkyl and polyfluoroalkyl substances
and incident pregnancy loss, LIFE Study. Reproductive Toxicology 65: 11-17.
Buck Louis, GM; Zhai, S; Smarr, MM; Grewal, J; Zhang, C; Grantz, KL, et al. (2018). Endocrine disruptors and neonatal
anthropometry, NICHD Fetal Growth Studies - Singletons. Environment International 119: 515-526.
Buck, RC; Franklin, J; Berger, U; Conder, JM; Cousins, IT; de Voogt, P, et al. (2011). Perfluoroalkyl and polyfluoroalkyl
substances in the environment: terminology, classification, and origins [Review], Integrated Environmental Assessment
and Management 7: 513-541. http://dx.doi.org/1.0.1.002/ieain.258
Buck, RC; Korzeniowski, SH; Laganis, E; Adamsky, F. (2021). Identification and classification of commercially relevant per-
and poly-fluoroalkyl substances (PFAS). Integrated Environmental Assessment and Management 17: 1045-1055.
http: //dx. doi. org/1.0.1.002/ieam. ¦4450
Budtz-Jergensen, E; Grandjean, P. (2018a). Application of benchmark analysis for mixed contaminant exposures: Mutual
adjustment of perfluoroalkylate substances associated with immunotoxicity. PLoS ONE 13: e0205388.
http: //dx. doi. org/1.0.1.3'71. /j oumal .pone. 0205388
Bulka, CM; Avula, V; Fry, RC. (2021). Associations of exposure to perfluoroalkyl substances individually and in mixtures with
persistent infections: Recent findings from NHANES 1999-2016. Environmental Pollution 275: 116619.
http://dx.doi.org/10.1016/i.envpol.2021.. 1.1.661.9
Buser, MC; Scinicariello, F. (2016). Perfluoroalkyl substances and food allergies in adolescents. Environment International 88:
74-79.
Butenhoff, JL; Chang, SC; Olsen, GW; Thomford, PJ. (2012). Chronic dietary toxicity and carcinogenicity study with potassium
perfluorooctanesulfonate in Sprague Dawley rats. Toxicology 293: 1-15. http://dx.doi.Org/10.1016/i.tox.2012.01.003
Butenhoff, JL; Ehresman, DJ; Chang, SC; Parker, GA; Stump, DG. (2009). Gestational and lactational exposure to potassium
perfluorooctanesulfonate (K+PFOS) in rats: developmental neurotoxicity. Reproductive Toxicology 27: 319-330.
Butt, CM; Berger, U; Bossi, R; Tomy, GT. (2010). Levels and trends of poly- and perfluorinated compounds in the arctic
environment [Review], Science of the Total Environment 408: 2936-2965.
http://dx.doi.Org/10.1016/i.scitotenv.2010.03.015
Byrne, S; Seguinot-Medina, S; Miller, P; Waghiyi, V; von Hippel, FA; Buck, CL; Carpenter, DO. (2017). Exposure to
polybrominated diphenyl ethers and perfluoroalkyl substances in a remote population of Alaska Natives.
Environmental Pollution 231: 387-395. http://dx.doi.Org/10.1016/i.envpol.2017.08.020
Byrne, SC; Miller, P; Seguinot-Medina, S; Waghiyi, V; Buck, CL; von Hippel, FA; Carpenter, DO. (2018). Exposure to
perfluoroalkyl substances and associations with serum thyroid hormones in a remote population of Alaska Natives.
Environmental Research 166: 537-543.
Cai, A; Portengen, L; Govarts, E; Martin, LR; Schoeters, G; Legler, J, et al. (2023). Prenatal exposure to persistent organic
pollutants and changes in infant growth and childhood growth trajectories. Chemosphere 314: 137695.
http://dx.doi.Org/10.1016/i.chemosphere.2022.137695
Cai, D; Li, QQ; Chu, C; Wang, SZ; Tang, YT; Appleton, AA, et al. (2020). High trans-placental transfer of perfluoroalkyl
substances alternatives in the matched maternal-cord blood serum: Evidence from a birth cohort study. Science of the
Total Environment 705: 135885.
Cakmak, S; Lukina, A; Karthikeyan, S; Atlas, E; Dales, R. (2022). The association between blood PFAS concentrations and
clinical biochemical measures of organ function and metabolism in participants of the Canadian Health Measures
Survey (CHMS). Science of the Total Environment 827: 153900. http://dx.doi.Org/10.1016/i.scitotenv.2022.153900
Calafat, AM; Kato, K; Hubbard, K; Jia, T; Botelho, JC; Wong, LY. (2019). Legacy and alternative per- and polyfluoroalkyl
substances in the U. S. general population: Paired serum-urine data from the 2013-2014 National Health and Nutrition
Examination Survey. Environment International 131: 105048. http://dx.doi.org/1.0.1.01.6/j.eiwint.201.9.1.05048
Calafat, AM; Wong, LY; Kuklenyik, Z; Reidy, JA; Needham, LL. (2007). Polyfluoroalkyl chemicals in the US population: Data
from the National Health and Nutrition Examination Survey (NHANES) 2003-2004 and comparisons with NHANES
1999-2000. Environmental Health Perspectives 115: 1596-1602. http://dx.doi.org/10.1289/ehp, 1.0598
CalEPA. (2021). Public Health Goals: Perfluorooctanoic Acid and Perfluorooctane Sulfonic Acid in Drinking Water (First Public
Review Draft ed.). California Environmental Protection Agency, Office of Environmental Health Hazard Assessment,
Pesticide and Environmental Toxicology Branch.
https://oehha.ca.gov/sites/default/files/media/downloads/cmr/pfoapfosphgdraft061021.pdf
R-4
-------
APRIL 2024
Callan, AC; Rotander, A; Thompson, K; Heyworth, J; Mueller, IF; Odland, J0; Hinwood, AL. (2016). Maternal exposure to
perfluoroalkyl acids measured in whole blood and birth outcomes in offspring. Science of the Total Environment 569-
570: 1107-1113.
Campbell, S; Raza, M; Pollack, AZ. (2016). Perfluoroalkyl substances and endometriosis in US women in NHANES 2003-2006.
Reproductive Toxicology 65: 230-235.
Canova, C; Barbieri, G; Zare Jeddi, M; Gion, M; Fabricio, A; Dapra, F, et al. (2020). Associations between perfluoroalkyl
substances and lipid profile in a highly exposed young adult population in the Veneto Region. Environment
International 145: 106117. littp://dx.doi.org/1.0.1.01.6/j.envint.2020.1.061.1.7
Canova, C; Di Nisio, A; Barbieri, G; Russo, F; Fletcher, T; Batzella, E, et al. (2021). PFAS Concentrations and Cardiometabolic
Traits in Highly Exposed Children and Adolescents. International Journal of Environmental Research and Public
Health 18. http://dx.doi.org/10.3390/iierphl82412881
Cao, L; Guo, Y; Chen, Y; Hong, J; Wu, J; Hangbiao, J. (2022). Per-/polyfluoroalkyl substance concentrations in human serum
and their associations with liver cancer. Chemosphere 296: 134083.
http://dx.doi.org/10.1016/icheiiiosphere.2022.134083
Cao, T; Qu, A; Li, Z; Wang, W; Liu, R; Wang, X, et al. (2021). The relationship between maternal perfluoroalkylated substances
exposure and low birth weight of offspring: a systematic review and meta-analysis. Environmental Science and
Pollution Research 28: 67053-67065. fattp: //dx. dot ore/10.1.007/s 1.1.3 56-021. -1.5061. -4
Cao, W; Liu, X; Liu, X; Zhou, Y; Zhang, X; Tian, H, et al. (2018). Perfluoroalkyl substances in umbilical cord serum and
gestational and postnatal growth in a Chinese birth cohort. Environment International 116: 197-205.
Cardenas, A; Gold, DR; Hauser, R; Kleinman, KP; Hivert, MF; Calafat, AM, et al. (2017). Plasma concentrations of per- and
polyfluoroalkyi substances at baseline and associations with glycemic indicators and diabetes incidence among high-
risk adults in the Diabetes Prevention Program trial. Environmental Health Perspectives 125: 107001.
Cardenas, A; Hivert, MF; Gold, DR; Hauser, R; Kleinman, KP; Lin, PD, et al. (2019). Associations of perfluoroalkyl and
polyfluoroalkyi substances with incident diabetes and microvascular disease. Diabetes Care 42: 1824-1832.
Cariou, R; Veyrand, B; Yamada, A; Berrebi, A; Zalko, D; Durand, S, et al. (2015). Perfluoroalkyl acid (PFAA) levels and
profiles in breast milk, maternal and cord serum of French women and their newborns. Environment International 84:
71-81.
Caron-Beaudoin, E; Ayotte, P; Laouan Sidi, EA; Simon, CoL; Nation, CoWLPF; Nutashkuan, CTKo, et al. (2019). Exposure to
perfluoroalkyl substances (PFAS) and associations with thyroid parameters in First Nation children and youth from
Quebec. Environment International 128: 13-23.
Caserta, D; Pegoraro, S; Mallozzi, M; Di Benedetto, L; Colicino, E; Lionetto, L; Simmaco, M. (2018). Maternal exposure to
endocrine disruptors and placental transmission: a pilot study. Gynecological Endocrinology 34: 1-4.
CDC. (2011). Tetanus surveillance — United States, 2001-2008. MMWR Morbidity and Mortality Weekly Report 60: 365-369.
Cellesi, C; Michelangeli, C; Rossolini, GM; Giovannoni, F; Rossolini, A. (1989). Immunity to diphtheria, six to 15 years after a
basic three-dose immunization schedule. Journal of Biological Standardization 17: 29-34.
http://dx.doi.ore/10.1.01.6/0092-1.1.57(89)90025-5
Chan, E; Burstyn, I; Cherry, N; Bamforth, F; Martin, JW. (2011). Perfluorinated acids and hypothyroxinemia in pregnant women.
Environmental Research 111: 559-564. http://dx.doi.org/1.0.1.01.6/j.eiivres.201.1.01..01.1.
Chang, CJ; Barr, DB; Ryan, PB; Panuwet, P; Smarr, MM; Liu, K, et al. (2022). Per- and polyfluoroalkyi substance (PFAS)
exposure, maternal metabolomic perturbation, and fetal growth in African American women: A meet-in-the-middle
approach. Environment International 158: 106964. http://dx.doi.Org/10.1016/i.envint.2021.106964
Chang, S; Allen, BC; Andres, KL; Ehresman, DJ; Falvo, R; Provencher, A, et al. (2017). Evaluation of serum lipid, thyroid, and
hepatic clinical chemistries in association with serum perfluorooctanesulfonate (PFOS) in cynomolgus monkeys after
oral dosing with potassium PFOS. Toxicological Sciences 156: 387-401.
Chang, SC; Ehresman, DJ; Bjork, JA; Wallace, KB; Parker, GA; Stump, DG; Butenhoff, JL. (2009). Gestational and lactational
exposure to potassium perfluorooctanesulfonate (K+PFOS) in rats: toxicokinetics, thyroid hormone status, and related
gene expression. Reproductive Toxicology 27: 387-399. http://dx.doi.Org/10.1016/i.reprotox.2009.01.005
Chang, SC; Noker, PE; Gorman, GS; Gibson, SJ; Hart, JA; Ehresman, DJ; Butenhoff, JL. (2012). Comparative pharmacokinetics
of perfluorooctanesulfonate (PFOS) in rats, mice, and monkeys. Reproductive Toxicology 33: 428-440.
http://dx.doi.Org/10.1016/i.reprotox.2011.07.002
Charlton, CL; Lai, FY; Dover, DC. (2016). How to determine protective immunity in the post-vaccine era. 12: 903-906.
http://dx.doi.org/10.1.080/2164551.5.201.5.1.1.28600
Chateau-Degat, ML; Pereg, D; Dallaire, R; Ayotte, P; Dery, S; Dewailly, E. (2010). Effects of perfluorooctanesulfonate exposure
on plasma lipid levels in the Inuit population of Nunavik (Northern Quebec). Environmental Research 110: 710-717.
http://dx.doi.Org/10.1016/i.envres.2010.07.003
Chen, A; Jandarov, R; Zhou, L; Calafat, AM; Zhang, G; Urbina, EM, et al. (2019a). Association of perfluoroalkyl substances
exposure with cardiometabolic traits in an island population of the eastern Adriatic coast of Croatia. Science of the
Total Environment 683: 29-36.
Chen, F; Yin, S; Kelly, BC; Liu, W. (2017a). Chlorinated polyfluoroalkyi ether sulfonic acids in matched maternal, cord, and
placenta samples: A study of transplacental transfer. Environmental Science and Technology 51: 6387-6394.
Chen, F; Yin, S; Kelly, BC; Liu, W. (2017b). Isomer-specific transplacental transfer of perfluoroalkyl acids: Results from a
survey of paired maternal, cord sera, and placentas. Environmental Science and Technology 51: 5756-5763.
R-5
-------
APRIL 2024
Chen, L; Tong, C; Huo, X; Zhang, J; Tian, Y. (2021). Prenatal exposure to perlluoroalkyl and polylluoroalkyl substances and
birth outcomes: A longitudinal cohort with repeated measurements. Chemosphere 267: 128899.
http: //dx. doi. org/1.0.1.01.6/j. chemosphere.2020.1.28899
Chen, MH; Ha, EH; Liao, HF; Jeng, SF; Su, YN; Wen, TW, et al. (2013). Perlluorinated compound levels in cord blood and
neurodevelopment at 2 years of age. Epidemiology 24: 800-808.
Chen, MH; Ha, EH; Wen, TW; Su, YN; Lien, GW; Chen, CY, et al. (2012a). Perlluorinated compounds in umbilical cord blood
and adverse birth outcomes. PLoS ONE 7: e42474. http://dx.doi.org/10.1371/joumal.pone.0042474
Chen, MH; Ng, S; Hsieh, CJ; Lin, CC; Hsieh, WS; Chen, PC. (2017c). The impact of prenatal perlluoroalkyl substances exposure
on neonatal and child growth. Science of the Total Environment 607-608: 669-675.
Chen, Q; Huang, R; Hua, L; Guo, Y; Huang, L; Zhao, Y, et al. (2018). Prenatal exposure to perlluoroalkyl and polylluoroalkyl
substances and childhood atopic dermatitis: A prospective birth cohort study. Environmental Health: A Global Access
Science Source 17: 1-12.
Chen, Q; Zhang, X; Zhao, Y; Lu, W; Wu, J; Zhao, S, et al. (2019b). Prenatal exposure to perlluorobutanesulfonic acid and
childhood adiposity: A prospective birth cohort study in Shanghai, China. Chemosphere 226: 17-23.
Chen, T; Zhang, L; Yue, JQ; Lv, ZQ; Xia, W; Wan, YJ, et al. (2012b). Prenatal PFOS exposure induces oxidative stress and
apoptosis in the lung of rat off-spring. Reproductive Toxicology 33: 538-545.
http://dx.doi.Org/10.1016/j.reprotox.2011.03.003
Chen, YM; Guo, LH. (2009). Fluorescence study on site-specific binding of perlluoroalkyl acids to human serum albumin.
Archives of Toxicology 83: 255-261. http: //dx. doi. org/1.0.1.007/s00204-008-03 5 9-x
Cheng, W; Ng, CA. (2018). Predicting relative protein affinity of novel per- and polylluoroalkyl substances (PFASs) by an
efficient molecular dynamics approach. Environmental Science and Technology 52: 7972-7980.
Cheng, X; Wei, Y; Zhang, Z; Wang, F; He, J; Wang, R, et al. (2022). Plasma PFOA and PFOS Levels, DNA Methylation, and
Blood Lipid Levels: A Pilot Study. Environmental Science & Technology 56: 17039-17051.
http://dx.doi.org/10.1021/acs.est.2cQ4107
Christensen, KY; Maisonet, M; Rubin, C; Holmes, A; Calafat, AM; Kato, K, et al. (2011). Exposure to polylluoroalkyl chemicals
during pregnancy is not associated with offspring age at menarche in a contemporary British cohort. Environment
International 37: 129-135. http://dx.doi.Org/10.1016/i.envint.2010.08.007
Christensen, KY; Raymond, M; Meiman, J. (2019). Perlluoroalkyl substances and metabolic syndrome. International Journal of
Hygiene and Environmental Health 222: 147-153.
Christensen, KY; Raymond, M; Thompson, BA; Anderson, HA. (2016a). Perlluoroalkyl substances in older male anglers in
Wisconsin. Environment International 91: 312-318.
Christensen, KY; Raymond, MR; Thompson, BA; Anderson, HA. (2016b). Fish consumption, levels of nutrients and
contaminants, and endocrine-related health outcomes among older male anglers in Wisconsin. Journal of Occupational
and Environmental Medicine 58: 668-675. http://dx.doi.org/10.1097/JOM.000000000000Q758
Christenson, B; Bottiger, M. (1986). Serological immunity to diphtheria in Sweden in 1978 and 1984. Scandinavian Journal of
Infectious Diseases 18: 227-233. http://dx.doi.org/10.3109/00365548609Q32331
Chu, C; Zhou, Y; Li, QQ; Bloom, MS; Lin, S; Yu, YJ, et al. (2020). Are perlluorooctane sulfonate alternatives safer? New
insights from a birth cohort study. Environment International 135: 105365.
Cluett, R; Seshasayee, SM; Rokoff, LB; Rifas-Shiman, SL; Ye, X; Calafat, AM, et al. (2019). Per- and Polylluoroalkyl Substance
Plasma Concentrations and Bone Mineral Density in Midchildhood: A Cross-Sectional Study (Project Viva, United
States). Environmental Health Perspectives 127: 87006.
Cohn, BA; La Merrill, MA; Krigbaum, NY; Wang, M; Park, JS; Petreas, M, et al. (2020). In utero exposure to poly- and
perlluoroalkyl substances (PFASs) and subsequent breast cancer. Reproductive Toxicology 92: 112-119.
Collier, RJ. (1975). Diphtheria toxin: mode of action and structure. 39: 54-85. http://dx.doi.org/10.1.1.28/br.39.1.54-85.1975
Conley, JM; Lambright, CS; Evans, N; Medlock-Kakaley, E; Dixon, A; Hill, D, et al. (2022a). Cumulative maternal and neonatal
effects of combined exposure to a mixture of perlluorooctanoic acid (PFOA) and perlluorooctane sulfonic acid (PFOS)
during pregnancy in the Sprague-Dawley rat. Environment International 170: 107631.
http://dx.d0i.0rg/l 0.1.01.6/j .envint.2022.1.07631.
Conley, JM; Lambright, CS; Evans, N; Medlock-Kakaley, E; Hill, D; McCord, J, et al. (2022b). Developmental toxicity of
Nafion byproduct 2 (NBP2) in the Sprague-Dawley rat with comparisons to hexalluoropropylene oxide-dimer acid
(HFPO-DA or GenX) and perlluorooctane sulfonate (PFOS). Environment International 160: 107056.
http: //dx. doi. org/1.0.1.01.6/j. envint .2021.. 1.07056
Convertino, M; Church, TR; Olsen, GW; Liu, Y; Doyle, E; Elcombe, CR, et al. (2018). Stochastic pharmacokinetic-
pharmacodynamic modeling for assessing the systemic health risk of perlluorooctanoate (pfoa). Toxicological Sciences
163: 293-306.
Conway, B; Innes, KE; Long, D. (2016). Perlluoroalkyl substances and beta cell deficient diabetes. Journal of Diabetes and its
Complications 30: 993-998.
Conway, BN; Badders, AN; Costacou, T; Arthur, JM; Innes, KE. (2018). Perlluoroalkyl substances and kidney function in
chronic kidney disease, anemia, and diabetes. Diabetes, Metabolic Syndrome and Obesity 11: 707-716.
Costa, O; Iniguez, C; Manzano-Salgado, CB; Amiano, P; Murcia, M; Casas, M, et al. (2019). First-trimester maternal
concentrations of polylluoroalkyl substances and fetal growth throughout pregnancy. Environment International 130:
104830. littp://dx.doi.org/1.0.1.01.6/j.envint.201.9.05.024
R-6
-------
APRIL 2024
Costello, E; Rock, S; Stratakis, N; Eckel, SP; Walker, DI; Valvi, D, et al. (2022). Exposure to per- and Polyfluoroalkyl
Substances and Markers of Liver Injury: A Systematic Review and Meta-Analysis [Review], Environmental Health
Perspectives 130: 46001. http://dx.doi.org/10.1.289/EHP1.0092
Crawford, NM; Fenton, SE; Strynar, M; Hines, EP; Pritchard, DA; Steiner, AZ. (2017). Effects of perfluorinated chemicals on
thyroid function, markers of ovarian reserve, and natural fertility. Reproductive Toxicology 69: 53-59.
Crissman, JW; Goodman, DG; Hildebrandt, PK; Maronpot, RR; Prater, DA; Riley, JH, et al. (2004). Best practices guideline:
Toxicologic histopathology. Toxicologic Pathology 32: 126-131. http://dx.doi.org/10.1080/01926230490268756
Crump, KS. (1995). Calculation of benchmark doses from continuous data. Risk Analysis 15: 79-89.
http://dx.doi.Org/10.llll/i.1539-6924.1995.tbQQQ95.x
Cui, L; Liao, CY; Zhou, QF; Xia, TM; Yun, ZJ; Jiang, GB. (2010). Excretion of PFOA and PFOS in male rats during a
subchronic exposure. Archives of Environmental Contamination and Toxicology 58: 205-213.
http://dx.doi.org/10.10Q7/sQQ244-QQ9-9336-5
Cui, L; Zhou, QF; Liao, CY; Fu, JJ; Jiang, GB. (2009). Studies on the toxicological effects of PFOA and PFOS on rats using
histological observation and chemical analysis. Archives of Environmental Contamination and Toxicology 56: 338-
349. http://dx.doi.org/10.10Q7/sQQ244-QQ8-9194-6
Cui, Q; Pan, Y; Wang, J; Liu, H; Yao, B; Dai, J. (2020). Exposure to per- and polyfluoroalkyl substances (PFASs) in serum
versus semen and their association with male reproductive hormones. Environmental Pollution 266 Pt. 2: 115330.
Curran, I; Hierlihy, SL; Liston, V; Pantazopoulos, P; Nunnikhoven, A; Tittlemier, S, et al. (2008). Altered fatty acid homeostasis
and related toxicologic sequelae in rats exposed to dietary potassium perfluorooctanesulfonate (PFOS). Journal of
Toxicology and Environmental Health, Part A: Current Issues 71: 1526-1541.
http://dx.doi.org/10.lQ8Q/1528739Q8Q2361763
DAlessandro, ML; Ellis, DA; Carter, JA; Stock, NL; March, RE. (2013). Competitive binding of aqueous perfluoroctanesulfonic
acid and ibuprofen with bovine serum albumin studied by electrospray ionization mass spectrometry. International
Journal of Mass Spectrometry 345: 28-36. http://dx.doi.Org/lQ.lQ16/i.ijms.2Q12.12.Q12
Dalla Zuanna, T; Savitz, DA; Barbieri, G; Pitter, G; Zare Jeddi, M; Dapra, F, et al. (2021). The association between
perfluoroalkyl substances and lipid profile in exposed pregnant women in the Veneto region, Italy. Ecotoxicology and
Environmental Safety 209: 111805. littp ://dx.doi.org/1.0.1.016/iecoeii¥.202Q. 1.1.1.805
Dalsager, L; Christensen, N; Halekoh, U; Timmermann, CAG; Nielsen, F; Kyhl, HB, et al. (2021). Exposure to perfluoroalkyl
substances during fetal life and hospitalization for infectious disease in childhood: A study among 1,503 children from
the Odense Child Cohort. Environment International 149: 106395. http://dx.doi.Org/10.1016/i.enviiit.2021.106395
Dalsager, L; Christensen, N; Husby, S; Kyhl, H; Nielsen, F; Hast, A, et al. (2016). Association between prenatal exposure to
perfluorinated compounds and symptoms of infections at age l-4years among 359 children in the Odense Child Cohort.
Environment International 96: 58-64.
Dangudubiyyam, SV; Mishra, JS; Song, R; Kumar, S. (2022). Maternal PFOS exposure during rat pregnancy causes
hypersensitivity to angiotensin II and attenuation of endothelium-dependent vasodilation in the uterine arteries. Biology
of Reproduction 107: 1072-1083. littp ://dx.doi. org/10.1093/biolre/ioac 141
Darrow, LA; Howards, PP; Winquist, A; Steenland, K. (2014). PFOA and PFOS serum levels and miscarriage risk.
Epidemiology 25: 505-512. http://dx.doi.org/10.1097/EDE.00000000000001Q3
Darrow, LA; Stein, CR; Steenland, K. (2013). Serum perfluorooctanoic acid and perfluorooctane sulfonate concentrations in
relation to birth outcomes in the Mid-Ohio Valley, 2005-2010. Environmental Health Perspectives 121: 1207-1213.
http://dx.doi.org/10.1.289/ehp. 1.206372
Dassuncao, C; Hu, XC; Nielsen, F; Weihe, P; Grandjean, P; Sunderland, EM. (2018). Shifting Global Exposures to Poly- and
Perfluoroalkyl Substances (PFASs) Evident in Longitudinal Birth Cohorts from a Seafood-Consuming Population.
Environmental Science and Technology 52: 3738-3747.
Davidkin, I; Jokinen, S; Broman, M; Leinikki, P; Peltola, H. (2008). Persistence of measles, mumps, and rubella antibodies in an
MMR-vaccinated cohort: a 20-year follow-up. Journal of Infectious Diseases 197: 950-956.
http://dx.doi.org/10.1086/528993
de Cock, M; de Boer, MR; Lamoree, M; Legler, J; van De Bor, M. (2014). First Year Growth in Relation to Prenatal Exposure to
Endocrine Disruptors - A Dutch Prospective Cohort Study. International Journal of Environmental Research and Public
Health 11: 7001-7021.
de Cock, M; De Boer, MR; Lamoree, M; Legler, J; Van De Bor, M. (2016). Prenatal exposure to endocrine disrupting chemicals
and birth weight-A prospective cohort study. Journal of Environmental Science and Health, Part A: Toxic/Hazardous
Substances & Environmental Engineering 51: 178-185.
De Felip, E; Abballe, A; Albano, FL; Battista, T; Carraro, V; Conversano, M, et al. (2015). Current exposure of Italian women of
reproductive age to PFOS and PFOA: A human biomonitoring study. Chemosphere 137: 1-8.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere.201.5.03.046
de Vos, MG; Huijbregts, MA; van den Heuvel-Greve, MJ; Vethaak, AD; Van de Vijver, KI; Leonards, PE, et al. (2008).
Accumulation of perfluorooctane sulfonate (PFOS) in the food chain of the Western Scheldt estuary: Comparing field
measurements with kinetic modeling. Chemosphere 70: 1766-1773.
http://dx.doi.Org/10.1016/i.chemosphere.2007.08.038
Dean, WP; Jessup, DC; Thompson, G; Romig, G; Powell, D. (1978). Fluorad Fluorochemical Surfactant FC-95 Acute Oral
Toxicity (LD50) Study in Rats. (Study No. 137-083). International Research and Development Corporation.
R-7
-------
APRIL 2024
Deji, Z; Liu, P; Wang, X; Zhang, X; Luo, Y; Huang, Z. (2021). Association between maternal exposure to perlluoroalkyl and
polylluoroalkyl substances and risks of adverse pregnancy outcomes: A systematic review and meta-analysis [Review],
Science of the Total Environment 783: 146984. http://dx.doi.Org/10.1016/i.scitotenv.2021.146984
Delinsky, AD; Strynar, MJ; Mccann, PJ; Varns, JL; Mcmillan, L; Nakayama, SF; Lindstrom, AB. (2010). Geographical
distribution of perfluorinated compounds in fish from Minnesota lakes and rivers. Environmental Science and
Technology 44: 2549-2554. http://dx.doi.org/10.1.021./es903777s
Deluca, NM; Angrish, M; Wilkins, A; Thayer, K; Cohen Hubal, EA. (2022a). Human exposure pathways to poly- and
perlluoroalkyl substances (PFAS) from indoor media: A systematic review protocol. Environment International 146:
106308. http://dx.doi.Org/10.1016/i.enviiit.2020.106308
Deluca, NM; Minucci, JM; Mullikin, A; Slover, R; Cohen Hubal, EA. (2022b). Human exposure pathways to poly- and
perlluoroalkyl substances (PFAS) from indoor media: A systematic review. Environment International 162: 107149.
littp: //dx. doi. org/1.0.1.01.6/j. envint .2022.1.071.49
Dhingra, R; Winquist, A; Darrow, LA; Klein, M; Steenland, K. (2017). A study of reverse causation: Examining the associations
of perlluorooctanoic acid serum levels with two outcomes. Environmental Health Perspectives 125: 416-421.
Di Nisio, A; Sabovic, I; Valente, U; Tescari, S; Rocca, MS; Guidolin, D, et al. (2019). Endocrine disruption of androgenic
activity by perlluoroalkyl substances: clinical and experimental evidence. Journal of Clinical Endocrinology and
Metabolism 104: 1259-1271.
Ding, N; Harlow, SD; Randolph, JF; Calafat, AM; Mukheijee, B; Batterman, S, et al. (2020). Associations of perlluoroalkyl
substances with incident natural menopause: The study of women's health across the nation. Journal of Clinical
Endocrinology and Metabolism 105: E3169-E3182.
Ding, N; Karvonen-Gutierrez, CA; Mukherjee, B; Calafat, AM; Harlow, SD; Park, SK. (2022). Per- and Polylluoroalkyl
Substances and Incident Hypertension in Multi-Racial/Ethnic Women: The Study of Women's Health Across the
Nation. Hypertension 79: 101161HYPERTENSIONAHA12118809.
http://dx.doi.org/10.1.1.61. /HYPERTEN SKIN AHA. 1.21.. 1.8809
Ding, N; Park, SK. (2020). Perlluoroalkyl substances exposure and hearing impairment in US adults. Environmental Research
187: 109686.
Dinglasan-Panlilio, MJ; Prakash, SS; Baker, JE. (2014). Perfluorinated compounds in the surface waters of Puget Sound,
Washington and Clayoquot and Barkley Sounds, British Columbia. Marine Pollution Bulletin 78: 173-180.
littp: //dx. doi. org/1.0.1.01.6/j .marpolbul .201.3.1.0.046
Domazet, SL; Grantved, A; Timmermann, AG; Nielsen, F; Jensen, TK. (2016). Longitudinal associations of exposure to
perfluoroalkylated substances in childhood and adolescence and indicators of adiposity and glucose metabolism 6 and
12 years later: The European Youth Heart Study. Diabetes Care 39: 1745-1751.
Domazet, SL; Jensen, TK; Wedderkopp, N; Nielsen, F; Andersen, LB; Grantved, A. (2020). Exposure to perfluoroalkylated
substances (PFAS) in relation to fitness, physical activity, and adipokine levels in childhood: The European youth heart
study. Environmental Research 191: 110110.
Domingo, JL; Nadal, M. (2017). Per- and Polylluoroalkyl Substances (PFASs) in Food and Human Dietary Intake: A Review of
the Recent Scientific Literature [Review]. Journal of Agricultural and Food Chemistry 65: 533-543.
littp: //dx. doi. org/1.0.1.021. /ac s. j afc. 6b04683
Donat-Vargas, C; Bergdahl, IA; Tornevi, A; Wennberg, M; Sommar, J; Kiviranta, H, et al. (2019a). Perlluoroalkyl substances
and risk of type II diabetes: A prospective nested case-control study. Environment International 123: 390-398.
Donat-Vargas, C; Bergdahl, IA; Tornevi, A; Wennberg, M; Sommar, J; Koponen, J, et al. (2019b). Associations between
repeated measure of plasma perlluoroalkyl substances and cardiometabolic risk factors. Environment International 124:
58-65.
Dong, GH; Liu, MM; Wang, D; Zheng, L; Liang, ZF; Jin, YH. (2011). Sub-chronic effect of perfluorooctanesulfonate (PFOS) on
the balance of type 1 and type 2 cytokine in adult C57BL6 mice. Archives of Toxicology 85: 1235-1244.
http://dx.d0i.0rg/l 0.1.007/s00204-01.1. -0661. -x
Dong, GH; Tung, KY; Tsai, CH; Liu, MM; Wang, D; Liu, W, et al. (2013). Serum polylluoroalkyl concentrations, asthma
outcomes, and immunological markers in a case-control study of Taiwanese children. Environmental Health
Perspectives 121: 507-513, 513e501-508. http://dx.doi.org/10.1289/ehp. 1.205351.
Dong, GH; Zhang, YH; Zheng, L; Liu, W; Jin, YH; He, QC. (2009). Chronic effects of perfluorooctanesulfonate exposure on
immunotoxicity in adult male C57BL/6 mice. Archives of Toxicology 83: 805-815. http://dx.doi.org/10.10Q7/sQQ204-
009-0424-0
Dong, Z; Wang, H; Yu, YY; Li, YB; Naidu, R; Liu, Y. (2019). Using 2003-2014 U.S. NHANES data to determine the
associations between per- and polylluoroalkyl substances and cholesterol: Trend and implications. Ecotoxicology and
Environmental Safety 173: 461-468.
Donley, GM; Taylor, E; Jeddy, Z; Namulanda, G; Hartman, TJ. (2019). Association between in utero perlluoroalkyl substance
exposure and anti-Miillerian hormone levels in adolescent females in a British cohort. Environmental Research 177:
108585.
Dreyer, AF; Jensen, RC; Glintborg, D; Schmedes, AV; Brandslund, I; Nielsen, F, et al. (2020). Perlluoroalkyl substance exposure
early in pregnancy was negatively associated with late pregnancy cortisone levels. Journal of Clinical Endocrinology
and Metabolism 105: E2834-E2844.
R-8
-------
APRIL 2024
Duan, Y; Sun, H; Yao, Y; Meng, Y; Li, Y. (2020). Distribution of novel and legacy per-/polyfluoroalkyl substances in serum and
its associations with two glycemic biomarkers among Chinese adult men and women with normal blood glucose levels.
Environment International 134: 105295.
Ducatman, A; Tan, Y; Nadeau, B; Steenland, K. (2023). Perfluorooctanoic Acid (PFOA) Exposure and Abnormal Alanine
Aminotransferase: Using Clinical Consensus Cutoffs Compared to Statistical Cutoffs for Abnormal Values. Toxics 11:
449. http://dx.doi.org/1.0.3390/toxics 1.1.050449
Ducatman, A; Zhang, J; Fan, H. (2015). Prostate-specific antigen and perfluoroalkyl acids in the C8 health study population.
Journal of Occupational and Environmental Medicine 57: 111-114.
Dufour, P; Pirard, C; Seghaye, MC; Charlier, C. (2018). Association between organohalogenated pollutants in cord blood and
thyroid function in newborns and mothers from Belgian population. Environmental Pollution 238: 389-396.
Dunder, L; Lind, PM; Salihovic, S; Stubleski, J; Karrman, A; Lind, L. (2022). Changes in plasma levels of per- and
polyfluoroalkyl substances (PFAS) are associated with changes in plasma lipids - A longitudinal study over 10 years.
Environmental Research 211: 112903. http://dx.doi.Org/10.1016/i.envies.2022.l 1.2903
Dzierlenga, M, .W.; Crawford, L,.; Longnecker, M, .P. (2020). Birth weight and perfluorooctane sulfonic acid: a random-effects
meta-regression analysis. Environmental Epidemiology 4: e095. http://dx.doi.org/10.1097/EE9.0000000000000Q95
E, L; Zhang, S; Jiang, X. (2023). Association between perfluoroalkyl substances exposure and the prevalence of nonalcoholic
fatty liver disease in the different sexes: a study from the National Elealth and Nutrition Examination Survey 2005-
2018. Environmental Science and Pollution Research International 30: 44292-44303. http://dx.doi.org/10.1007/sl 1.356-
023-25258-4
East, A; Egeghy, PP; Hubal, E; Slover, R; Vallero, DA. (2021). Computational estimates of daily aggregate exposure to
PFOA/PFOS from 2011 to 2017 using a basic intake model. Journal of Exposure Science and Environmental
Epidemiology Online ahead of print, http://dx.doi.org/10.1038/s41370-021-00374-w
Ebert, A; Allendorf, F; Berger, U; Goss, KU; Ulrich, N. (2020). Membrane/water partitioning and permeabilities of
perfluoroalkyl acids and four of their alternatives and the effects on toxicokinetic behavior. Environmental Science and
Technology 54: 5051-5061.
ECCC. (2018). Canadian Environmental Protection Act, 1999. Federal Environmental Quality Guidelines. Perfluorooctane
Sulfonate (PFOS). h ttps://www.canada.ca/en/environment-climate-change/services/evaluating-existing-
substances/federal-environmental-quality-guidel ines-perfluorooctane-sulfonate.html
EFSA. (2020). Risk to human health related to the presence ofperfluoroalkyl substances in food. EFSA Journal 18.
http://dx.doi.Org/https://doi.org/10.2903/j.efsa.2020.6223
Egeghy, PP; Lorber, M. (2011). An assessment of the exposure of Americans to perfluorooctane sulfonate: A comparison of
estimated intake with values inferred from NHANES data. Journal of Exposure Science and Environmental
Epidemiology 21: 150-168. http://dx.doi.org/10.1038/jes.2009.73
Ehresman, DJ; Froehlich, JW; Olsen, GW; Chang, SC; Butenhoff, JL. (2007). Comparison of human whole blood, plasma, and
serum matrices for the determination of perfluorooctanesulfonate (PFOS), perfluorooctanoate (PFOA), and other
fluorochemicals. Environmental Research 103: 176-184. http://dx.doi.org/10.1016/ieiiYres.2006,06,008
Eick, SM; Demicco, E; Valeri, L; Woodruff, TJ; Morello-Frosch, R; Horn Thepaksorn, EK, et al. (2020). Associations between
prenatal maternal exposure to per- and polyfluoroalkyl substances (PFAS) and polybrominated diphenyl ethers
(PBDEs) and birth outcomes among pregnant women in San Francisco. Environmental Health 19: 100-100.
http://dx.doi.org/10.1186/sl2940-020-0Q654-2
Eriksen, KT; Raaschou-Nielsen, O; Mclaughlin, JK; Lipworth, L; Tjenneland, A; Overvad, K; Serensen, M. (2013). Association
between plasma PFOA and PFOS levels and total cholesterol in a middle-aged Danish population. PLoS ONE 8:
e56969. http://dx.doi.org/10.1371/joumal.pone.0056969
Eriksen, KT; Serensen, M; Mclaughlin, JK; Lipworth, L; Tjenneland, A; Overvad, K; Raaschou-Nielsen, O. (2009).
Perfluorooctanoate and perfluorooctanesulfonate plasma levels and risk of cancer in the general Danish population.
Journal of the National Cancer Institute 101: 605-609. http: //dx. doi. org/1.0.1.093/j nci/dj p041.
Ernst, A; Brix, N; Lauridsen, LLB; Olsen, J; Parner, ET; Liew, Z, et al. (2019). Exposure to perfluoroalkyl substances during
fetal life and pubertal development in boys and girls from the danish national birth cohort. Environmental Health
Perspectives 127: 17004.
Erol, E; Kumar, LS; Cline, GW; Shulman, GI; Kelly, DP; Binas, B. (2004). Liver fatty acid binding protein is required for high
rates of hepatic fatty acid oxidation but not for the action of PPARalpha in fasting mice. FASEB Journal 18: 347-349.
http: //dx. doi. org/1.0.1.096/fj.03-033Ofj e
Eryasa, B; Grandjean, P; Nielsen, F; Valvi, D; Zmirou-Navier, D; Sunderland, E, et al. (2019). Physico-chemical properties and
gestational diabetes predict transplacental transfer and partitioning of perfluoroalkyl substances. Environment
International 130: 104874.
Espindola-Santos, AD; Meyer, A; Dabkiewicz, VE; Camara, VDM; Asmus, CIR, F. (2021). Serum levels of perfluorooctanoic
acid and perfluorooctane sulfonic acid in pregnant women: Maternal predictors and associations with birth outcomes in
the PIPA Project. Journal of Obstetrics and Gynaecology Research 47: 3107-3118. http://dx.doi.org/10.1.1.1.1/iog. 1.4883
Etzel, TM; Braun, JM; Buckley, JP. (2019). Associations of serum perfluoroalkyl substance and vitamin D biomarker
concentrations in NHANES, 2003-2010. International Journal of Hygiene and Environmental Health 222: 262-269.
R-9
-------
APRIL 2024
Fan, Y; Lu, C; Li, X; Xu, Q; Zhang, Y; Yang, X, et al. (2020). Serum albumin mediates the effect of multiple per- and
polyfluoroalkyl substances on serum lipid levels. Environmental Pollution 266 Pt 2: 115138.
fattp: //dx. doi. org/1.0.1.01.6/j. envpol.2020.1.1.51.3 8
Fang, S; Chen, X; Zhao, S; Zhang, Y; Jiang, W; Yang, L; Zhu, L. (2014). Trophic magnification and isomer fractionation of
perfluoroalkyl substances in the food web of Taihu Lake, China. Environmental Science and Technology 48: 2173-
2182. fattp: //dx. dot ore/1.0.1.021. /es40501.8b
Fassler, CS; Pinney, SE; Xie, C; Biro, FM; Pinney, SM. (2019). Complex relationships between perfluorooctanoate, body mass
index, insulin resistance and serum lipids in young girls. Environmental Research 176: 108558.
FDA. (2018). Analytical Results for PFAS in 2018 Produce Sampling (Parts Per Trillion). Retrieved from
https://www.fda.gov/media/127848/dowaload
FDA. (2019a). Analytical Results for PFAS in 2019 Total Diet Study Sampling (Parts Per Trillion)—Dataset 1: U. S. Food and
Drug Administration (FDA). Retrieved from https://www.fda.gov/media/127852/download
FDA. (2019b). Analytical Results for PFAS in 2019 Total Diet Study Sampling (Parts Per Trillion)—Dataset 2: U.S. Food and
Drug Administration (FDA). Retrieved from https://www.fda.gov/media/133693/download
FDA. (2020). Authorized Uses of PFAS in Food contact Applications, https://www.fda.gov/food/chemical-contaminants-
food/authorized-uses-pfas-food-contact-applications
FDA. (2021). Analytical Results of Testing Food for PFAS from Environmental Contamination. Retrieved from
https://www.fda.gov/food/chemical-contaminants-food/analvtical-results-testing-food-pfas-environmental-
contamination
Fei, C; Mclaughlin, JK; Lipworth, L; Olsen, J. (2008a). Prenatal exposure to perfluorooctanoate (PFOA) and
perfluorooctanesulfonate (PFOS) and maternally reported developmental milestones in infancy. Environmental Health
Perspectives 116: 1391-1395. http://dx.doi.org/10.1289/ehp, 1.1.277
Fei, C; Mclaughlin, JK; Lipworth, L; Olsen, J. (2010). Prenatal exposure to PFOA and PFOS and risk of hospitalization for
infectious diseases in early childhood. Environmental Research 110: 773-777.
fattp: //dx. doi. org/1.0.1.01.6/j. envres.201.0.08.004
Fei, C; Mclaughlin, JK; Tarone, RE; Olsen, J. (2007). Perfluorinated chemicals and fetal growth: A study within the Danish
National Birth Cohort. Environmental Health Perspectives 115: 1677-1682. http://dx.doi.org/10.1289/ehp. 1.0506
Fei, C; Mclaughlin, JK; Tarone, RE; Olsen, J. (2008b). Fetal growth indicators and perfluorinated chemicals: a study in the
Danish National Birth Cohort. American Journal of Epidemiology 168: 66-72. http://dx.doi.org/10.1.093/aie/kwnO95
Fei, C; Olsen, J. (2011). Prenatal exposure to perfluorinated chemicals and behavioral or coordination problems at age 7 years.
Environmental Health Perspectives 119: 573-578. http://dx.doi.org/10.1289/ehp. 1.002026
Fei, CY; Mclaughlin, JK; Lipworth, L; Olsen, J. (2009). Maternal levels of perfluorinated chemicals and subfecundity. Human
Reproduction 24: 1200-1205. http://dx.doi.org/10.1093/humrep/den49Q
Feng, X; Long, G; Zeng, G; Zhang, Q; Song, B; Wu, KH. (2022a). Association of increased risk of cardiovascular diseases with
higher levels of perfluoroalkylated substances in the serum of adults. Environmental Science and Pollution Research
29: 89081-89092. http://dx.doi.org/10.1007/sll356-022-22021-z
Feng, Y; Bai, Y; Lu, Y; Chen, M; Fu, M; Guan, X, et al. (2022b). Plasma perfluoroalkyl substance exposure and incidence risk of
breast cancer: A case-cohort study in the Dongfeng-Tongji cohort. Environmental Pollution 306: 119345.
http: //dx. doi. org/1.0.1.01.6/j. envpol.2022.1.1.9345
Fisher, M; Arbuckle, TE; Wade, M; Haines, DA. (2013). Do perfluoroalkyl substances affect metabolic function and plasma
lipids?—Analysis of the 2007-2009, Canadian Health Measures Survey (CHMS) Cycle 1. Environmental Research 121:
95-103. http://dx.doi.Org/10.1016/i.envres.2012.ll.006
Fitz-Simon, N; Fletcher, T; Luster, MI; Steenland, K; Calafat, AM; Kato, K; Armstrong, B. (2013). Reductions in serum lipids
with a 4-year decline in serum perfluorooctanoic acid and perfluorooctanesulfonic acid. Epidemiology 24: 569-576.
http://dx.doi.Org/1.0.1.097/.EDE.0b01.3e31.829443ee
Fleisch, AF; Rifas-Shiman, SL; Mora, AM; Calafat, AM; Ye, X; Luttmann-Gibson, H, et al. (2017). Early-life exposure to
perfluoroalkyl substances and childhood metabolic function. Environmental Health Perspectives 125: 481-487.
Forns, J; Iszatt, N; White, RA; Mandal, S; Sabaredzovic, A; Lamoree, M, et al. (2015). Perfluoroalkyl substances measured in
breast milk and child neuropsychological development in a Norwegian birth cohort study. Environment International
83: 176-182.
Forsthuber, M; Kaiser, AM; Granitzer, S; Hassl, I; Hengstschlager, M; Stangl, H; Gundacker, C. (2020). Albumin is the major
carrier protein for PFOS, PFOA, PFHxS, PFNA andPFDA in human plasma. Environment International 137: 105324.
Fraser, AJ; Webster, TF; Watkins, DJ; Strynar, MJ; Kato, K; Calafat, AM, et al. (2013). Polyfluorinated compounds in dust from
homes, offices, and vehicles as predictors of concentrations in office workers' serum. Environment International 60:
128-136. http://dx.doi.Org/10.1016/i.envint.2013.08.012
Frisbee, SJ; Shankar, A; Knox, SS; Steenland, K; Savitz, DA; Fletcher, T; Ducatman, AM. (2010). Perfluorooctanoic acid,
perfluorooctanesulfonate, and serum lipids in children and adolescents: results from the C8 Health Project. Archives of
Pediatrics and Adolescent Medicine 164: 860-869. http://dx.doi.org/10.1001/archpediatrics.2010.163
Fromme, H; Mosch, C; Morovitz, M; Alba-Alejandre, I; Boehmer, S; Kiranoglu, M, et al. (2010). Pre- and postnatal exposure to
perfluorinated compounds (PFCs). Environmental Science and Technology 44: 7123-7129.
http://dx.d0i.0rg/l 0.1.021/es 1.01.1.84f
R-10
-------
APRIL 2024
Fromme, H; Tittlemier, SA; Volkel, W; Wilhelm, M; Twardella, D. (2009). Perfluorinated compounds - Exposure assessment for
the general population in western countries [Review], International Journal of Hygiene and Environmental Health 212:
239-270. http://dx.doi.Org/10.1016/i.iiheh.2008.04.007
Fry, K; Power, MC. (2017). Persistent organic pollutants and mortality in the United States, NHANES 1999-2011.
Environmental Health: A Global Access Science Source 16: 105.
Fu, J; Gao, Y; Cui, L; Wang, T; Liang, Y; Qu, G, et al. (2016). Occurrence, temporal trends, and half-lives of perfluoroalkyl
acids (PFAAs) in occupational workers in China. Scientific Reports 6: 38039.
Fu, Y; Wang, T; Fu, Q; Wang, P; Lu, Y. (2014). Associations between serum concentrations of perfluoroalkyl acids and serum
lipid levels in a Chinese population. Ecotoxicology and Environmental Safety 106: 246-252.
littp: //dx. doi. org/1.0.1.01.6/j. ecoenv.201.4.04.03 9
Fuentes, S; Colomina, MT; Rodriguez, J; Vicens, P; Domingo, JL. (2006). Interactions in developmental toxicology: concurrent
exposure to perfluorooctane sulfonate (PFOS) and stress in pregnant mice. Toxicology Letters 164: 81-89.
littp: //dx. doi. org/1.0.1.01.6/j .toxlet .2005.11.01.3
Fuentes, S; Colomina, MT; Vicens, P; Franco-Pons, N; Domingo, JL. (2007a). Concurrent exposure to perfluorooctane sulfonate
and restraint stress during pregnancy in mice: effects on postnatal development and behavior of the offspring.
Toxicological Sciences 98: 589-598. http://dx.doi.org/10. K)93/toxsci/kfiiil21.
Fuentes, S; Vicens, P; Colomina, MT; Domingo, JL. (2007b). Behavioral effects in adult mice exposed to perfluorooctane
sulfonate (PFOS). Toxicology 242: 123-129. http://dx.doi.Org/10.1016/i.tox.2007.09.012
Gabrielsson, J; Weiner, D. (2000). Pharmacokinetic and pharmacodynamic data analysis: concepts and applications (3rd ed.).
Stockholm: Swedish Pharmaceutical Press.
Galazka, A; Kardymowicz, B. (1989). Immunity against diphtheria in adults in Poland. Epidemiology and Infection 103: 587-
593. http://dx.doi.Org/1.0.1.01.7/s0950268800030983
Galazka, AM; Milstien, JB; Robertson, SE; Cutts, FT. (1993). The immunological basis for immunization module 2 : Diphtheria.
(WHO/EPI/Gen/93.11-18). Galazka, AM; Milstien, JB; Robertson, SE; Cutts, FT.
http://apps.wlio.int/iris/bitstreaiii/faaiidle/1066S/S8891/WHO~EPI-GEN~93.12-iiiod2-
eng. pdf? sequence=3 8&isA llowed=y
Gallo, V; Leonardi, G; Brayne, C; Armstrong, B, en; Fletcher, T. (2013). Serum perfluoroalkyl acids concentrations and memory
impairment in a large cross-sectional study. British Medical Journal Open 3.
Gallo, V; Leonardi, G; Genser, B; Lopez-Espinosa, MJ; Frisbee, SJ; Karlsson, L, et al. (2012). Serum perfluorooctanoate (PFOA)
and perfluorooctane sulfonate (PFOS) concentrations and liver function biomarkers in a population with elevated
PFOA exposure. Environmental Health Perspectives 120: 655-660. http://dx.doi.org/10.1289/ehp, 1.1.04436
Gao, B; He, X; Liu, W; Zhang, H; Saito, N; Tsuda, S. (2015). Distribution of perfluoroalkyl compounds in rats: Indication for
using hair as bioindicator of exposure. Journal of Exposure Science and Environmental Epidemiology 25: 632-638.
Gao, K, e; Zhuang, T; Liu, X; Fu, J; Zhang, J; Fu, J, ie, et al. (2019). Prenatal Exposure to Per- and Polyfluoroalkyi Substances
(PFASs) and Association between the Placental Transfer Efficiencies and Dissociation Constant of Serum Proteins-
PFAS Complexes. Environmental Science and Technology 53: 6529-6538.
Gao, X; Ni, W; Zhu, S; Wu, Y; Cui, Y; Ma, J, et al. (2021). Per- and polyfluoroalkyi substances exposure during pregnancy and
adverse pregnancy and birth outcomes: A systematic review and meta-analysis. Environmental Research 201: 111632.
http://dx.doi.org/10.1.01.6/j.envres.2021.. 1.1.1.632
Gardener, H; Sun, Q; Grandjean, P. (2021). PFAS concentration during pregnancy in relation to cardiometabolic health and birth
outcomes. Environmental Research 192: 110287. http://dx.doi.org/10.1016/i.envres.2020.1.1.0287
Gay lord, A; Berger, KI; Naidu, M; Attina, TM; Gilbert, J; Koshy, TT, et al. (2019). Serum perfluoroalkyl substances and lung
function in adolescents exposed to the World Trade Center disaster. Environmental Research 172: 266-272.
Gaylord, A; Trasande, L; Kannan, K; Thomas, KM; Lee, S; Liu, M; Levine, J. (2020). Persistent organic pollutant exposure and
celiac disease: A pilot study. Environmental Research 186: 109439.
Gebbink, WA; Berger, U; Cousins, IT. (2015). Estimating human exposure to PFOS isomers and PFCA homologues: the relative
importance of direct and indirect (precursor) exposure. Environment International 74: 160-169.
littp: //dx. doi. org/1.0.1.01.6/j. eiivint .2 01.4,1.0.01.3
Geiger, SD; Xiao, J; Ducatman, A; Frisbee, S; Innes, K; Shankar, A. (2014a). The association between PFOA, PFOS and serum
lipid levels in adolescents. Chemosphere 98: 78-83. http://dx.doi.Org/10.1016/i.chemosphere.2013.10.005
Geiger, SD; Xiao, J; Shankar, A. (2013). Positive association between perfluoroalkyl chemicals and hyperuricemia in children.
American Journal of Epidemiology 177: 1255-1262.
Geiger, SD; Xiao, J; Shankar, A. (2014b). No association between perfluoroalkyl chemicals and hypertension in children.
Integrated Blood Pressure Control 7: 1-7. http://dx.doi.org/10.2147/IBPC.S47660
Gennings, C; Wolk, A; Hakansson, N; Lindh, C; Bornehag, CG. (2020). Contrasting prenatal nutrition and environmental
exposures in association with birth weight and cognitive function in children at 7 years. 3: 162-171.
littp: //dx. doi. org/1.0.1.1.36/biiij nph-2020-000099
Genuis, SJ; Birkholz, D; Ralitsch, M; Thibault, N. (2010). Human detoxification of perfluorinated compounds. Public Health
124: 367-375. http://dx.doi.Org/10.1016/i.puhe.2010.03.002
Genuis, SJ; Liu, Y; Genuis, QI; Martin, JW. (2014). Phlebotomy treatment for elimination of perfluoroalkyl acids in a highly
exposed family: a retrospective case-series. PLoS ONE 9: el 14295.
R-ll
-------
APRIL 2024
Ghassabian, A; Bell, EM; Ma, WL; Sundaram, R; Kannan, K; Buck Louis, GM; Yeung, E. (2018). Concentrations of
perfluoroalkyl substances and bisphenol A in newborn dried blood spots and the association with child behavior.
Environmental Pollution 243: 1629-1636.
Ghisari, M; Eiberg, H; Long, M; Bonefeld-Jergensen, EC. (2014). Polymorphisms in phase I and phase II genes and breast
cancer risk and relations to persistent organic pollutant exposure: a case-control study in Inuit women. Environmental
Health: A Global Access Science Source 13: 19. http://dx.doi.org/10.1.1.86/1476-069X-1.3-1.9
Ghisari, M; Long, M; R0ge, DM; Olsen, J; Bonefeld-Jergensen, EC. (2017). Polymorphism in xenobiotic and estrogen
metabolizing genes, exposure to perfluorinated compounds and subsequent breast cancer risk: A nested case-control
study in the Danish National Birth Cohort. Environmental Research 154: 325-333.
Giesy, JP; Kannan, K. (2001). Global distribution of perfluorooctane sulfonate in wildlife. Environmental Science and
Technology 35: 1339-1342. http://dx.doi.org/10.1021/es001834k
Gleason, JA; Post, GB; Fagliano, JA. (2015). Associations of perfluorinated chemical serum concentrations and biomarkers of
liver function and uric acid in the US population (NHANES), 2007-2010. Environmental Research 136: 8-14.
http: //dx. doi. org/1.0.1.01.6/j. envres.2014.10.004
Glynn, A; Berger, U; Bignert, A; Ullah, S; Aune, M; Lignell, S; Darnerud, PO. (2012). Perfluorinated alkyl acids in blood serum
from primiparous women in Sweden: serial sampling during pregnancy and nursing, and temporal trends 1996-2010.
Environmental Science and Technology 46: 9071-9079. http://dx.doi.org/10.1021/es301168c
Goeden, HM; Greene, CW; Jacobus, JA. (2019). A transgenerational toxicokinetic model and its use in derivation of Minnesota
PFOA water guidance. Journal of Exposure Science and Environmental Epidemiology 29: 183-195.
http://dx.doi.org/10.1.038/s41.370-01.8-01.10-5
Goldenthal, E; Jessup, DC; Geil, RG; Mehring, JS. (1978). Ninety-day subacute rhesus monkey toxicity study: Fluorad"
Fluorochemical FC-143. (Study No. 137-090). St. Paul,MN: Report prepared for 3M by Institutional Research and
Devlopment Corporation (Mattawan, MN).
Goldenthal, EI; Jessup, DC; Geil, RB; Mehring, JS. (1979). Ninety-Day Subacute Rhesus Monkey Toxicity Study. (Study No.
137-087). Mattawan, MI: International Research and Development Corporation. http://intemal-
pdf://074771.9523/1.529 3M-EPA-00070027 native.pdf
Gomis, MI; Vestergren, R; Macleod, M; Mueller, JF; Cousins, IT. (2017). Historical human exposure to perfluoroalkyl acids in
the United States and Australia reconstructed from biomonitoring data using population-based pharmacokinetic
modelling. Environment International 108: 92-102. http://dx.doi.Org/10.1016/i.envint.2017.08.002
Goodrich, JA; Walker, D; Lin, X; Wang, H; Lim, T; Mcconnell, R, et al. (2022). Exposure to perfluoroalkyl substances and risk
of hepatocellular carcinoma in a multiethnic cohort. 4: 100550. http://dx.doi.org/1.0.1.01.6/j .jhepr.2022.1.00550
Goudarzi, H; Araki, A; Itoh, S; Sasaki, S; Miyashita, C; Mitsui, T, et al. (2017a). The association of prenatal exposure to
perfluorinated chemicals with glucocorticoid and androgenic hormones in cord blood samples: The Hokkaido study.
Environmental Health Perspectives 125: 111-118.
Goudarzi, H; Miyashita, C; Okada, E; Kashino, I; Chen, CJ; Ito, S, et al. (2017b). Prenatal exposure to perfluoroalkyl acids and
prevalence of infectious diseases up to 4 years of age. Environment International 104: 132-138.
Goudarzi, H; Miyashita, C; Okada, E; Kashino, I; Kobayashi, S; Chen, CJ, et al. (2016a). Effects of prenatal exposure to
perfluoroalkyl acids on prevalence of allergic diseases among 4-year-old children. Environment International 94: 124-
132.
Goudarzi, H; Nakajima, S; Ikeno, T; Sasaki, S; Kobayashi, S; Miyashita, C, et al. (2016b). Prenatal exposure to perfluorinated
chemicals and neurodevelopment in early infancy: The Hokkaido Study. Science of the Total Environment 541: 1002-
1010.
Govarts, E; Iszatt, N; Trnovec, T; de Cock, M; Eggesba, M; Palkovicova Murinova, L, et al. (2018). Prenatal exposure to
endocrine disrupting chemicals and risk of being born small for gestational age: Pooled analysis of seven European
birth cohorts. Environment International 115: 267-278.
Govarts, E; Remy, S; Bruckers, L; Den Hond, E; Sioen, I; Nelen, V, et al. (2016). Combined effects of prenatal exposures to
environmental chemicals on birth weight. International Journal of Environmental Research and Public Health 13: n/a.
Graber, JM; Alexander, C; Laumbach, RJ; Black, K; Strickland, PO; Georgopoulos, PG, et al. (2019). Per and polyfluoroalkyl
substances (PFAS) blood levels after contamination of a community water supply and comparison with 2013-2014
NHANES. Journal of Exposure Science and Environmental Epidemiology 29: 172-182.
Grandjean, P; Andersen, EW; Budtz-Jergensen, E; Nielsen, F; Malbak, K; Weihe, P; Heilmann, C. (2012). Serum vaccine
antibody concentrations in children exposed to perfluorinated compounds. JAMA: Journal of the American Medical
Association 307: 391-397. http://dx.doi.org/10.1001/iama.2011.2034
Grandjean, P; Bateson, T. (2021). RE: Benchmark analysis for PFAS immunotoxicity. Available online
Grandjean, P; Heilmann, C; Weihe, P; Nielsen, F; Mogensen, UB; Budtz-Jergensen, E. (2017a). Serum Vaccine Antibody
Concentrations in Adolescents Exposed to Perfluorinated Compounds. Environmental Health Perspectives 125:
077018.
Grandjean, P; Heilmann, C; Weihe, P; Nielsen, F; Mogensen, UB; Timmermann, A; Budtz-Jergensen, E. (2017b). Estimated
exposures to perfluorinated compounds in infancy predict attenuated vaccine antibody concentrations at age 5-years.
Journal of Immunotoxicology 14: 188-195.
Grandjean, P; Timmermann, CAG; Kruse, M; Nielsen, F; Vinholt, PJ; Boding, L, et al. (2020). Severity of COVID-19 at elevated
exposure to perfluorinated alkylates. PLoS ONE 15: e0244815. http://dx.doi.org/1.0.1371 /joumat.pone.024481.5
R-12
-------
APRIL 2024
Granum, B; Haug, LS; Namork, E; Stelevik, SB; Thomsen, C; Aaberge, IS, et al. (2013). Pre-natal exposure to perfluoroalkyl
substances may be associated with altered vaccine antibody levels and immune-related health outcomes in early
childhood. Journal of Immunotoxicology 10: 373-379. http://dx.doi.org/10.3109/1547691X.2012.75558Q
Grasty, RC; Bjork, JA; Wallace, KB; Lau, CS; Rogers, JM. (2005). Effects of prenatal perfluorooctane sulfonate (PFOS)
exposure on lung maturation in the perinatal rat. Birth Defects Research, Part B: Developmental and Reproductive
Toxicology 74: 405-416. littp://dx.doi.org/1.0.1.002/bdrb.20059
Grasty, RC; Wolf, DC; Grey, BE; Lau, CS; Rogers, JM. (2003). Prenatal window of susceptibility to perfluorooctane sulfonate-
induced neonatal mortality in the Sprague-Dawley rat. Birth Defects Research, Part B: Developmental and
Reproductive Toxicology 68: 465-471. http://dx.doi.org/10.1002/bdrb. 1.0046
Gray, LE; Ostby, J; Sigmon, R; Ferrell, J; Rehnberg, G; Linder, R, et al. (1988). The development of a protocol to assess
reproductive effects of toxicants in the rat [Review], Reproductive Toxicology 2: 281-287.
http://dx.doi.org/10.1.01.6/0890-6238(88)90032-9
Greenland, S; Longnecker, MP. (1992). Methods for trend estimation from summarized dose-response data, with applications to
meta-analysis. American Journal of Epidemiology 135: 1301-1309.
Gremmel, C; Fromel, T; Knepper, TP. (2016). Systematic determination of perfluoroalkyl and polyfluoroalkyl substances
(PFASs) in outdoor jackets. Chemosphere 160: 173-180. littp://dx.doi.org/1.0.1.01.6/j.clieniosphere,201.6,06,043
Grice, MM; Alexander, BH; Eloffbeck, R; Kampa, DM. (2007). Self-reported medical conditions in perfluorooctanesulfonyl
fluoride manufacturing workers. Journal of Occupational and Environmental Medicine 49: 722-729.
http://dx.doi.org/10.1097/JOM.0b013e3180582Q43
Gross, RS; Ghassabian, A; Vandyousefi, S; Messito, MJ; Gao, C; Kannan, K; Trasande, L. (2020). Persistent organic pollutants
exposure in newborn dried blood spots and infant weight status: A case-control study of low-income Elispanic mother-
infant pairs. Environmental Pollution 267: 115427. http://dx.doi.Org/10.1016/i.envpol.2020.l 1.5427
Gui, SY; Chen, YN; Wu, KJ; Liu, W; Wang, WJ; Liang, EtR, et al. (2022a). Association Between Exposure to Per- and
Polyfluoroalkyl Substances and Birth Outcomes: A Systematic Review and Meta-Analysis. Frontiers in Public Elealth
10: 855348. http://dx.doi.org/10.3389/fpubh.2022.855348
Gui, SY; Qiao, JC; Xu, KX; Li, ZL; Chen, YN; Wu, KJ, et al. (2022b). Association between per- and polyfluoroalkyl substances
exposure and risk of diabetes: a systematic review and meta-analysis [Review], Journal of Exposure Science &
Environmental Epidemiology. http://dx.doi.org/10.1038/s41370-022-0Q464-3
Gump, BB; Hill, DT; Robinson, M; Kannan, K; Eleffernan, K; Atallah-Yunes, NH, et al. (2023). Perfluoroalkyl substances
(PFAS) and lead (Pb) as "cardiovascular disruptors" in 9-11-year-old children living in Syracuse, New York, United
States. Environmental Research 236: 116758. http://dx.doi.Org/10.1016/i.envres.2023.l 1.6758
Guo, R; Reiner, EJ; Bhavsar, SP; Elelm, PA; Mabury, SA; Braekevelt, E; Tittlemier, SA. (2012). Determination of
polyfluoroalkyl phosphoric acid diesters, perfluoroalkyl phosphonic acids, perfluoroalkyl phosphinic acids,
perfluoroalkyl carboxylic acids, and perfluoroalkane sulfonic acids in lake trout from the Great Lakes region.
Analytical and Bioanalytical Chemistry 404: 2699-2709. http://dx.doi.org/10.1007/sQ0216-012-6125-l
Gyllenhammar, I; Benskin, JP; Sandblom, O; Berger, U; Ahrens, L; Lignell, S, et al. (2018a). Perfluoroalkyl acids (PFAAs) in
serum from 2-4-month-old infants: Influence of maternal serum concentration, gestational age, breast-feeding, and
contaminated drinking water. Environmental Science and Technology 52: 7101-7110.
Gyllenhammar, I; Benskin, JP; Sandblom, O; Berger, U; Ahrens, L; Lignell, S, et al. (2019). Perfluoroalkyl Acids (PFAAs) in
Children's Serum and Contribution from PFAA-Contaminated Drinking Water. Environmental Science and
Technology 53: 11447-11457.
Gyllenhammar, I; Diderholm, B; Gustafsson, J; Berger, U; Ridefelt, P; Benskin, JP, et al. (2018b). Perfluoroalkyl acid levels in
first-time mothers in relation to offspring weight gain and growth. Environment International 111: 191-199.
Hack, M; Klein, NK; Taylor, HG. (1995). Long-term developmental outcomes of low birth weight infants. The Future of
Children 5: 176-196.
Hall, AP; Elcombe, CR; Foster, JR; Harada, T; Kaufmann, W; Knippel, A, et al. (2012). Liver hypertrophy: a review of adaptive
(adverse and non-adverse) changes—conclusions from the 3rd International ESTP Expert Workshop [Review],
Toxicologic Pathology 40: 971-994. http://dx.doi.org/10.1177/0192623312448935
Hall, SM; Zhang, S; Hoffman, K; Miranda, ML; Stapleton, HM. (2022). Concentrations of per- and polyfluoroalkyl substances
(PFAS) in human placental tissues and associations with birth outcomes. Chemosphere 295: 133873.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere.2022.1.3 3873
Hamm, MP; Cherry, NM; Chan, E; Martin, JW; Burstyn, I. (2010). Maternal exposure to perfluorinated acids and fetal growth.
Journal of Exposure Science and Environmental Epidemiology 20: 589-597. http://dx.doi.org/10.1038/ies.2009.57
Hammer, T; Lophaven, SN; Nielsen, KR; Petersen, MS; Munkholm, P; Weihe, P, et al. (2019). Dietary risk factors for
inflammatory bowel diseases in a high-risk population: Results from the Faroese IBD study. 7: 924-932.
http://dx.doi.org/10.1177/2050640619852244
Han, R; Zhang, F; Wan, C; Liu, L; Zhong, Q; Ding, W. (2018a). Effect of perfluorooctane sulphonate-induced Kupffer cell
activation on hepatocyte proliferation through the NF-KB/TNF-a/IL-6-dependent pathway. Chemosphere 200: 283-294.
Han, W; Gao, Y; Yao, Q; Yuan, T; Wang, Y; Zhao, S, et al. (2018b). Perfluoroalkyl and polyfluoroalkyl substances in matched
parental and cord serum in Shandong, China. Environment International 116: 206-213.
http://dx.doi.Org/10.1016/i.envint.2018.04.025
R-13
-------
APRIL 2024
Han, X; Meng, L; Zhang, G; Li, Y; Shi, Y; Zhang, Q; Jiang, G. (2021). Exposure to novel and legacy per- and polyfluoroalkyl
substances (PFASs) and associations with type 2 diabetes: A case-control study in East China. Environment
International 156: 106637. http://dx.doi.org/1.0.1.01.6/j.envint.2021.1.06637
Hanssen, L; Dudarev, AA; Huber, S; Odland, J0; Nieboer, E; Sandanger, TM. (2013). Partition of perlluoroalkyl substances
(PFASs) in whole blood and plasma, assessed in maternal and umbilical cord samples from inhabitants of arctic Russia
and Uzbekistan. Science of the Total Environment 447: 430-437.
Hanssen, L; Rollin, H; Odland, J0; Moe, MK; Sandanger, TM. (2010). Perfluorinated compounds in maternal serum and cord
blood from selected areas of South Africa: results of a pilot study. Journal of Environmental Monitoring 12: 1355-
1361. http: //dx. dot org/10.1.039/b924420d
Harada, KH; Hashida, S; Kaneko, T; Takenaka, K; Minata, M; Inoue, K, et al. (2007). Biliary excretion and cerebrospinal fluid
partition of perfluorooctanoate and perfluorooctane sulfonate in humans. Environmental Toxicology and Pharmacology
24: 134-139. http://dx.doi.Org/10.1016/i.etap.2007.04.003
Hardell, E; Karrman, A; van Bavel, B; Bao, J; Carlberg, M; Hardell, L. (2014). Case-control study on perfluorinated alkyl acids
(PFAAs) and the risk of prostate cancer. Environment International 63: 35-39.
http: //dx. doi. org/1.0.1.01.6/j. eiivint .201.3.1.0.005
Harris, MH; Oken, E; Rifas-Shiman, SL; Calafat, AM; Ye, X; Bellinger, DC, et al. (2018). Prenatal and childhood exposure to
per- and polyfluoroalkyl substances (PFASs) and child cognition. Environment International 115: 358-369.
Hartman, TJ; Calafat, AM; Holmes, AK; Marcus, M; Northstone, K; Flanders, WD, et al. (2017). Prenatal exposure to
perlluoroalkyl substances and body fatness in girls. Childhood Obesity 13: 222-230.
Haug, LS; Huber, S; Becher, G; Thomsen, C. (2011). Characterisation of human exposure pathways to perfluorinated
compounds—comparing exposure estimates with biomarkers of exposure. Environment International 37: 687-693.
http: //dx. doi. org/1.0.1.01.6/j. eiivint .2 01.1.. 01.. 01.1.
Haukas, M; Berger, U; Hop, H; Gulliksen, B; Gabrielsen, GW. (2007). Bioaccumulation of per- and polyfluorinated alkyl
substances (PFAS) in selected species from the Barents Sea food web. Environmental Pollution 148: 360-371.
http: //dx. doi. org/1.0.1.01.6/j. envpol.2006,09.021.
He, X; Liu, Y; Xu, B; Gu, L; Tang, W. (2018). PFOA is associated with diabetes and metabolic alteration in US men: National
Health and Nutrition Examination Survey 2003-2012. Science of the Total Environment 625: 566-574.
Heffernan, AL; Cunningham, TK; Drage, DS; Aylward, LL; Thompson, K; Vijayasarathy, S, et al. (2018). Perfluorinated alkyl
acids in the serum and follicular fluid of UK women with and without polycystic ovarian syndrome undergoing fertility
treatment and associations with hormonal and metabolic parameters. International Journal of Hygiene and
Environmental Health 221: 1068-1075.
Hill, AB. (1965). The environment and disease: Association or causation? Proceedings of the Royal Society of Medicine 58: 295-
300.
Hjermitslev, MH; Long, M; Wielsee, M; Bonefeld-Jergensen, EC. (2019). Persistent organic pollutants in Greenlandic pregnant
women and indices of foetal growth: The ACCEPT study. Science of the Total Environment 698: 134118.
Hoffman, K; Webster, TF; Weisskopf, MG; Weinberg, J; Vieira, VM. (2010). Exposure to polyfluoroalkyl chemicals and
attention deficit/hyperactivity disorder in U.S. children 12-15 years of age. Environmental Health Perspectives 118:
1762-1767. http://dx.doi.org/10.1289/ehp.1001.898
Holder, C; Deluca, NM; Luh, J; Alexander, P; Minucci, JM; Vallero, D, et al. (2023). Systematic Evidence Mapping of Potential
Exposure Pathways for Per- and Polyfluoroalkyl Substances Based on Measured Occurrence in Multiple Media.
Environmental Science & Technology 57: 5107-5116.
http://dx.doi.Org/https://pubs.acs.org/doi/abs/10.1021/acs.est.2c07185
Honarvar, B; Moghadami, M; Moattari, A; Emami, A; Odoomi, N; Bagheri Lankarani, K. (2013). Seroprevalence of anti-rubella
and anti-measles IgG antibodies in pregnant women in Shiraz, Southern Iran: outcomes of a nationwide measles-rubella
mass vaccination campaign. PLoS ONE 8: e55043. http://dx.doi.org/10.1371/ioumal.pone.0055043
Houde, M; Martin, JW; Letcher, RJ; Solomon, KR; Muir, DCG. (2006). Biological Monitoring of Polyfluoroalkyl Substances: A
Review [Review], Environmental Science and Technology 40: 3463-3473. http://dx.doi.org/10.1.021./es0S2S80b
Howard, BE; Phillips, J; Miller, K; Tandon, A; Mav, D; Shah, MR, et al. (2016). SWIFT-Review: a text-mining workbench for
systematic review. Systematic Reviews 5: 87. http://dx.doi.org/10.1186/sl3643-016-0263-z
Hayer, BB; Ramlau-Hansen, CH; Obel, C; Pedersen, HS; Hernik, A; Ogniev, V, et al. (2015). Pregnancy serum concentrations of
perfluorinated alkyl substances and offspring behaviour and motor development at age 5-9 years—a prospective study.
Environmental Health: A Global Access Science Source 14: 2. http://dx.doi.org/10.1.1.86/1.476-069X-1.4-2
Hu, Y; Liu, G; Rood, J; Liang, L; Bray, GA; de Jonge, L, et al. (2019). Perlluoroalkyl substances and changes in bone mineral
density: A prospective analysis in the POUNDS-LOST study. Environmental Research 179: 108775.
Huang, H; Yu, K; Zeng, X; Chen, Q; Liu, Q; Zhao, Y, et al. (2020). Association between prenatal exposure to perlluoroalkyl
substances and respiratory tract infections in preschool children. Environmental Research 191: 110156.
http: //dx. doi. org/1.0.1.01.6/j. envres.2020.1.1.01.56
Huang, M; Jiao, J; Zhuang, P; Chen, X; Wang, J; Zhang, Y. (2018). Serum polyfluoroalkyl chemicals are associated with risk of
cardiovascular diseases in national US population. Environment International 119: 37-46.
Huang, MC; Dzierlenga, AL; Robinson, VG; Waidyanatha, S; Devito, MJ; Eifrid, MA, et al. (2019a). Toxicokinetics of
perfluorobutane sulfonate (PFBS), perfluorohexane-l-sulphonic acid (PFHxS), and perfluorooctane sulfonic acid
R-14
-------
APRIL 2024
(PFOS) in male and female Hsd:Sprague Dawley SD rats after intravenous and gavage administration. Toxicology
Reports 6: 645-655.
Huang, MC; Dzierlenga, AL; Robinson, VG; Waidyanatha, S; Devito, MJ; Eifrid, MA, et al. (2021). Corrigendum to
"Toxicokinetics of perfluorobutane sulfonate (PFBS), perfluorohexane-l-sulphonic acid (PFHxS), and perfluorooctane
sulfonic acid (PFOS) in male and female Hsd:Sprague Dawley SD rats after intravenous and gavage administration"
[Toxicol. Rep. 6 (2019) 645-655] [Erratum], Toxicology Reports 8: 365. http://dx.doi.Org/10.1016/i.toxrep.2021.02.001
Eluang, R; Chen, Q; Zhang, L; Luo, K; Chen, L; Zhao, S, et al. (2019b). Prenatal exposure to perfluoroalkyl and polyfluoroalkyl
substances and the risk of hypertensive disorders of pregnancy. Environmental Elealth: A Global Access Science
Source 18: 5.
Elumblet, O; Diaz-Ramirez, LG; Balmes, JR; Pinney, SM; Eliatt, RA. (2014). Perfluoroalkyl chemicals and asthma among
children 12-19 years of age: NE1ANES (1999-2008). Environmental Elealth Perspectives 122: 1129-1133.
littp: //dx. doi. org/1.0.1.2 89/ehp. 1.306606
Eluo, X; Eluang, R; Gan, Y; Luo, K; Aimuzi, R; Nian, M, et al. (2020). Perfluoroalkyl substances in early pregnancy and risk of
hypertensive disorders of pregnancy: A prospective cohort study. Environment International 138: 105656.
Elurley, S; Goldberg, D; Wang, M; Park, JS; Petreas, M; Bernstein, L, et al. (2018). Breast cancer risk and serum levels of per-
and poly-fluoroalkyi substances: a case-control study nested in the California Teachers Study. Environmental Elealth: A
Global Access Science Source 17: 83.
Elutcheson, R; Innes, K; Conway, B. (2020). Perfluoroalkyl substances and likelihood of stroke in persons with and without
diabetes. Diabetes and Vascular Disease Research 17: 1-8.
Impinen, A; Longnecker, MP; Nygaard, UC; London, SJ; Ferguson, KK; Elaug, LS; Granum, B. (2019). Maternal levels of
perfluoroalkyl substances (PFASs) during pregnancy and childhood allergy and asthma related outcomes and infections
in the Norwegian Mother and Child (MoBa) cohort. Environment International 124: 462-472.
Impinen, A; Nygaard, UC; Ledrup Carlsen, KC; Mowinckel, P; Carlsen, KH; Haug, LS; Granum, B. (2018). Prenatal exposure to
perfluoralkyl substances (PFASs) associated with respiratory tract infections but not allergy- and asthma-related health
outcomes in childhood. Environmental Research 160: 518-523.
Inoue, K; Okada, F; Ito, R; Kato, S; Sasaki, S; Nakajima, S, et al. (2004). Perfluorooctane sulfonate (PFOS) and related
perfluorinated compounds in human maternal and cord blood samples: assessment of PFOS exposure in a susceptible
population during pregnancy. Environmental Health Perspectives 112: 1204-1207. littp://dx.doi.org/1.0.1.289/ehp.6864
Inoue, K; Ritz, B; Andersen, SL; Ramlau-Hansen, CH; Hayer, BB; Bech, BH, et al. (2019). Perfluoroalkyl Substances and
Maternal Thyroid Hormones in Early Pregnancy; Findings in the Danish National Birth Cohort. Environmental Health
Perspectives 127: 117002.
Ipsen, J. (1946). Circulating antitoxin at the onset of diphtheria in 425 patients. Journal of Immunology 54: 325-347.
Ishida, K; Tsuyama, Y; Sanoh, S; Ohta, S; Kotake, Y. (2017). Perfluorooctane sulfonate induces neuronal vulnerability by
decreasing GluR2 expression. Archives of Toxicology 91: 885-895.
Itoh, H; Harada, KH; Kasuga, Y; Yokoyama, S; Onuma, H; Nishimura, H, et al. (2021). Serum perfluoroalkyl substances and
breast cancer risk in Japanese women: A case-control study. Science of the Total Environment 800: 149316.
http://dx.doi.Org/10.1016/i.scitotenv.2021.149316
Itoh, S; Araki, A; Mitsui, T; Miyashita, C; Goudarzi, H; Sasaki, S, et al. (2016). Association of perfluoroalkyl substances
exposure in utero with reproductive hormone levels in cord blood in the Hokkaido Study on Environment and
Children's Health. Environment International 94: 51-59.
Itoh, S; Araki, A; Miyashita, C; Yamazaki, K; Goudarzi, H; Minatoya, M, et al. (2019). Association between perfluoroalkyl
substance exposure and thyroid hormone/thyroid antibody levels in maternal and cord blood: The Hokkaido Study.
Environment International 133: 105139.
Iwabuchi, K; Senzaki, N; Mazawa, D; Sato, I; Hara, M; Ueda, F, et al. (2017). Tissue toxicokinetics of perfluoro compounds with
single and chronic low doses in male rats. Journal of Toxicological Sciences 42: 301-317.
Jaacks, LM; Boyd Barr, D; Sundaram, R; Grewal, J; Zhang, C; Buck Louis, GM. (2016). Pre-Pregnancy Maternal Exposure to
Persistent Organic Pollutants and Gestational Weight Gain: A Prospective Cohort Study. International Journal of
Environmental Research and Public Health 13.
Jackson-Browne, MS; Eliot, M; Patti, M; Spanier, AJ; Braun, JM. (2020). PFAS (per- and polyfluoroalkyl substances) and
asthma in young children: NHANES 2013-2014. International Journal of Hygiene and Environmental Health 229:
113565.
Jain, R. (2013). Association between thyroid profile and perfluoroalkyl acids: Data from NHNAES 2007-2008. Environmental
Research 126: 51-59.
Jain, RB. (2019). Concentration of selected liver enzymes across the stages of glomerular function: The associations with PFOA
and PFOS. Heliyon 5: e02168.
Jain, RB. (2020a). Associations between selected perfluoroalkyl acids in serum and hemoglobin in whole blood, a biomarker of
anemia: Impact of deteriorating kidney function. Environmental Pollution 263: 114458.
Jain, RB. (2020b). Impact of the co-occurrence of obesity with diabetes, anemia, hypertension, and albuminuria on
concentrations of selected perfluoroalkyl acids. Environmental Pollution 266 Pt. 2: 115207.
Jain, RB; Ducatman, A. (2018). Associations between lipid/lipoprotein levels and perfluoroalkyl substances among US children
aged 6-11 years. Environmental Pollution 243: 1-8.
R-15
-------
APRIL 2024
Jain, RB; Ducatman, A. (2019a). Dynamics of associations between perfluoroalkyl substances and uric acid across the various
stages of glomerular function. Environmental Science and Pollution Research 26: 12425-12434.
Jain, RB; Ducatman, A. (2019b). Perfluoroalkyl acids and thyroid hormones across stages of kidney function. Science of the
Total Environment 696: 133994.
Jain, RB; Ducatman, A. (2019c). Perfluoroalkyl acids serum concentrations and their relationship to biomarkers of renal failure:
Serum and urine albumin, creatinine, and albumin creatinine ratios across the spectrum of glomerular function among
US adults. Environmental Research 174: 143-151.
Jain, RB; Ducatman, A. (2019d). Roles of gender and obesity in defining correlations between perfluoroalkyl substances and
lipid/lipoproteins. Science of the Total Environment 653: 74-81.
Jain, RB; Ducatman, A. (2019e). Selective associations of recent low concentrations of perfluoroalkyl substances with liver
function biomarkers: nhanes 2011 to 2014 data on us adults aged >20 years. Journal of Occupational and
Environmental Medicine 61: 293-302.
Jain, RB; Ducatman, A. (2020). Associations between apolipoprotein B and selected perfluoroalkyl substances among diabetics
and nondiabetics. Environmental Science and Pollution Research International 2020: 13819-13828.
http://dx.doi.org/10.1007/sll356-02Q-11593-3
Jeddy, Z; Hartman, TJ; Taylor, EV; Poteete, C; Kordas, K. (2017). Prenatal concentrations of perfluoroalkyl substances and early
communication development in British girls. Early Human Development 109: 15-20.
Jeddy, Z; Tobias, JH; Taylor, EV; Northstone, K; Flanders, WD; Hartman, TJ. (2018). Prenatal concentrations of perfluoroalkyl
substances and bone health in British girls at age 17. Archives of Osteoporosis 13: 84.
Jensen, RC; Andersen, MS; Larsen, PV; Glintborg, D; Dalgard, C; Timmermann, CAG, et al. (2020a). Prenatal Exposures to
Perfluoroalkyl Acids and Associations with Markers of Adiposity and Plasma Lipids in Infancy: An Odense Child
Cohort Study. Environmental Health Perspectives 128: 77001.
Jensen, RC; Glintborg, D; Gade Timmermann, CA; Nielsen, F; Kyhl, HB; Frederiksen, H, et al. (2020b). Prenatal exposure to
perfluorodecanoic acid is associated with lower circulating concentration of adrenal steroid metabolites during mini
puberty inhuman female infants. The odense child cohort. Environmental Research 182: 109101.
Jensen, RC; Glintborg, D; Timmermann, CAG; Nielsen, F; Kyhl, HB; Andersen, HR, et al. (2018). Perfluoroalkyl substances and
glycemic status in pregnant Danish women: The Odense Child Cohort. Environment International 116: 101-107.
Jensen, TK; Andersen, LB; Kyhl, HB; Nielsen, F; Christesen, HT; Grandjean, P. (2015). Association between Perfluorinated
Compound Exposure and Miscarriage in Danish Pregnant Women. PLoS ONE 10: e0123496.
Ji, J; Song, L; Wang, J; Yang, Z; Yan, H; Li, T, et al. (2021). Association between urinary per- and poly-fluoroalkyl substances
and COVID-19 susceptibility. Environment International 153: 106524. http://dx.doi.Org/10.1016/i.envint.2021.106524
Jia, J; Duan, L; Dong, B; Dong, Q; Liu, Y; Yu, W, et al. (2023). Perfluoroalkyl and polyfluoroalkyl substances in cord serum of
newborns and their potential factors. Chemosphere 313: 137525. http://dx.doi.Org/10.1016/i.chemosphere.2022.137525
Jiang, H; Liu, H; Liu, G; Yu, J; Liu, N; Jin, Y, et al. (2022). Associations between Polyfluoroalkyl Substances Exposure and
Breast Cancer: A Meta-Analysis [Review], Toxics 10. http://dx.doi.org/10.3390/toxicsl0060318
Jiang, W; Zhang, Y; Zhu, L; Deng, J. (2014). Serum levels of perfluoroalkyl acids (PFAAs) with isomer analysis and their
associations with medical parameters in Chinese pregnant women. Environment International 64: 40-47.
Jin, H; Zhang, Y; Jiang, W; Zhu, L; Martin, JW. (2016). Isomer-Specific Distribution of Perfluoroalkyl Substances in Blood.
Environmental Science and Technology 50: 7808-7815.
Jin, R; Mcconnell, R; Catherine, C; Xu, S; Walker, DI; Stratakis, N, et al. (2020). Perfluoroalkyl substances and severity of
nonalcoholic fatty liver in Children: An untargeted metabolomics approach. Environment International 134: 105220.
Joensen, UN; Bossi, R; Leffers, H; Jensen, AA; Skakkebask, NE; Jergensen, N. (2009). Do perfluoroalkyl compounds impair
human semen quality? Environmental Health Perspectives 117: 923-927. http://dx.doi.org/10.1289/ehp.Q80Q5 17
Joensen, UN; Veyrand, B; Antignac, JP; Jensen, MB; Petersen, JH; Marchand, P, et al. (2014). PFOS (perfluorooctanesulfonate)
in serum is negatively associated with testosterone levels, but not with semen quality, in healthy men. Human
Reproduction 29: 1600-1600. http://dx.doi.org/lQ.1093/humrep/deulQ4
Johansson, N; Fredriksson, A; Eriksson, P. (2008). Neonatal exposure to perfluorooctane sulfonate (PFOS) and perfluorooctanoic
acid (PFOA) causes neurobehavioural defects in adult mice. NeuroToxicology 29: 160-169.
littp: //dx. doi. org/1.0.1.01.6/j .neuro .2007.1.0.008
Johnson, JD; Gibson, SJ; Ober, RE. (1984). Cholestyramine-enhanced fecal elimination of c-14 in rats after administration of
ammonium [c-14] perfluorooctanoate or potassium [c-14]perfluorooctanesulfonate. Fundamental and Applied
Toxicology 4: 972-976. http://dx.doi.org/10.1016/0272-0590(84)90235-5
Jones, LE; Ghassabian, A; Lawrence, DA; Sundaram, R; Yeung, E; Kannan, K; Bell, EM. (2022). Exposure to perfluoroalkyl
substances and neonatal immunoglobulin profiles in the upstate KIDS study (2008-2010). Environmental Pollution
308: 119656. http://dx.doi.Org/10.1016/i.envpol.2022.1.1.9656
Jergensen, KT; Specht, IO; Lenters, V; Bach, CC; Rylander, L; Jonsson, BA, et al. (2014). Perfluoroalkyl substances and time to
pregnancy in couples from Greenland, Poland and Ukraine. Environmental Health: A Global Access Science Source
13: 116. http://dx.doi.org/10.1.1.86/1.476-069X-1.3-1.1.6
Jusko, TA; Oktapodas, M; Palkovicova Murinova, L; Babinska, K; Babjakova, J; Verner, MA, et al. (2016). Demographic,
reproductive, and dietary determinants of perfluorooctane sulfonic (PFOS) and perfluorooctanoic acid (PFOA)
concentrations in human colostrum. Environmental Science and Technology 50: 7152-7162.
littp: //dx. doi. org/1.0.1.021. /ac s. est. 6b001.9 5
R-16
-------
APRIL 2024
Kang, H; Ding, N; Karvonen-Gutierrez, CA; Mukherjee, B; Calafat, AM; Park, SK. (2023). Per- and Polyfluoroalkyl Substances
(PFAS) and Lipid Trajectories in Women 45-56 Years of Age: The Study of Women's Health Across the Nation.
Environmental Health Perspectives 131: 87004. http://dx.doi.org/1.0.1.289/EHP1.2351.
Kang, H; Lee, HK; Moon, HB; Kim, S; Lee, J; Ha, M, et al. (2018). Perfluoroalkyl acids in serum of Korean children:
Occurrences, related sources, and associated health outcomes. Science of the Total Environment 645: 958-965.
Kang, Q; Gao, F; Zhang, X; Wang, L; Liu, J; Fu, M, et al. (2020). Nontargeted identification of per- and polyfluoroalkyl
substances in human follicular fluid and their blood-follicle transfer. Environment International 139: 105686.
Kannan, K; Tao, L; Sinclair, E; Pastva, SD; Jude, DJ; Giesy, JP. (2005). Perfluorinated compounds in aquatic organisms at
various trophic levels in a Great Lakes food chain. Archives of Environmental Contamination and Toxicology 48: 559-
566. http://dx.doi.Org/10.1.007/s00244-004-01.33-x
Karlsen, M; Grandjean, P; Weihe, P; Steuerwald, U; Oulhote, Y; Valvi, D. (2017). Early-life exposures to persistent organic
pollutants in relation to overweight in preschool children. Reproductive Toxicology 68: 145-153.
Karrman, A; Domingo, JL; Llebaria, X; Nadal, M; Bigas, E; van Bavel, B; Lindstrom, G. (2010). Biomonitoring perfluorinated
compounds in Catalonia, Spain: concentrations and trends in human liver and milk samples. Environmental Science
and Pollution Research 17: 750-758. http://dx.doi.org/10.1007/sll356-009-0178-5
Karrman, A; van Bavel, B; Jarnberg, U; Hardell, L; Lindstrom, G. (2006). Perfluorinated chemicals in relation to other persistent
organic pollutants in human blood. Chemosphere 64: 1582-1591. http://dx.doi.Org/10.1016/i.chemosphere.2005.l 1.040
Kasarala, G; Tillmann, III.. (2016). Standard liver tests. Clinical Liver Disease 8: 13-18. fattp: //dx .doi. org/1.0.1.002/ctd, 562
Kashino, I; Sasaki, S; Okada, E; Matsuura, H; Goudarzi, H; Miyashita, C, et al. (2020). Prenatal exposure to 11 perfluoroalkyl
substances and fetal growth: A large-scale, prospective birth cohort study. Environment International 136: 105355.
Kataria, A; Trachtman, H; Malaga-Dieguez, L; Trasande, L. (2015). Association between perfluoroalkyl acids and kidney
function in a cross-sectional study of adolescents. Environmental Health: A Global Access Science Source 14: 89.
Kato, K; Wong, LY; Chen, A; Dunbar, C; Webster, GM; Lanphear, BP; Calafat, AM. (2014). Changes in serum concentrations
of maternal poly- and perfluoroalkyl substances over the course of pregnancy and predictors of exposure in a
multiethnic cohort of Cincinnati, Ohio pregnant women during 2003-2006. Environmental Science and Technology 48:
9600-9608.
Kato, S; Itoh, S; Yuasa, M; Baba, T; Miyashita, C; Sasaki, S, et al. (2016). Association of perfluorinated chemical exposure in
utero with maternal and infant thyroid hormone levels in the Sapporo cohort of Hokkaido Study on the Environment
and Children's Health. Environmental Health and Preventive Medicine 21: 334-344.
Kaur, K; Lesseur, C; Chen, L; Andra, SS; Narasimhan, S; Pulivarthi, D, et al. (2023). Cross-sectional associations of maternal
PFAS exposure on SARS-CoV-2 IgG antibody levels during pregnancy. Environmental Research 219: 115067.
http://dx.doi.Org/10.1016/i.envres.2022.l 1.5067
Kawamoto, K; Sato, I; Tsuda, S; Yoshida, M; Yaegashi, K; Saito, N, et al. (2011). Ultrasonic-induced tonic convulsion in rats
after subchronic exposure to perfluorooctane sulfonate (PFOS). Journal of Toxicological Sciences 36: 55-62.
http://dx.doi.org/10.2131/its.36.55
Keller & HeckmanLLP. (2021). Attack on PFAS extends to food packaging. National Law Review X.
Kelly, BC; Ikonomou, MG; Blair, JD; Surridge, B; Hoover, D; Grace, R; Gobas, FA. (2009). Perfluoroalkyl contaminants in an
Arctic marine food web: trophic magnification and wildlife exposure. Environmental Science and Technology 43:
4037-4043. http://dx.doi.org/10.1021/es9003894
Kerstner-Wood, C; Coward, L; Gorman, G; Southern Research Institute. (2003). Protein binding of perfluorobutane sulfonate,
perfluorohexane sulfonate, perfluorooctane sulfonate and perfluorooctanoate to plasma (human, rat, and monkey), and
various human-derived plasma protein fractions. Study ID 9921.7 [TSCA Submission], (8EHQ-04-15845A;
88040000364). St. Paul, MN: 3M Company.
Khalil, N; Chen, A; Lee, M; Czerwinski, SA; Ebert, JR; Dewitt, JC; Kannan, K. (2016). Association of Perfluoroalkyl
Substances, Bone Mineral Density, and Osteoporosis in the US Population in NHANES 2009-2010. Environmental
Health Perspectives 124: 81-87.
Khalil, N; Ebert, JR; Honda, M; Lee, M; Nahhas, RW; Koskela, A, et al. (2018). Perfluoroalkyl substances, bone density, and
cardio-metabolic risk factors in obese 8-12 year old children: A pilot study. Environmental Research 160: 314-321.
Kim, DH; Kim, UJ; Kim, HY; Choi, SD; Oh, JE. (2016a). Perfluoroalkyl substances in serum from South Korean infants with
congenital hypothyroidism and healthy infants - Its relationship with thyroid hormones. Environmental Research 147:
399-404. http://dx.doi.org/10.1.01.6/i.eiivres.201.6.02.037
Kim, DH; Lee, JH; Oh, JE. (2019). Assessment of individual-based perfluoroalkly substances exposure by multiple human
exposure sources. Journal of Hazardous Materials 365: 26-33. http: //dx. doi. org/1.0.1.01.6/i. ihazma t.201.8.1.0.066
Kim, HY; Kim, KN; Shin, CH; Lim, YH; Kim, JI; Kim, BN, et al. (2020a). The relationship between perfluoroalkyl substances
concentrations and thyroid function in early childhood: A prospective cohort study. Thyroid 30: 1556-1565.
Kim, MJ; Moon, S; Oh, BC; Jung, D; Ji, K; Choi, K; Park, YJ. (2018). Association between perfluoroalkyl substances exposure
and thyroid function in adults: A meta-analysis. PLoS ONE 13: eOl 97244.
http: //dx. doi. org/1.0.1371/j oumal .pone. 01.97244
Kim, OJ; Kim, S; Park, EY; Oh, JK; Jung, SK; Park, S, et al. (2023). Exposure to serum perfluoroalkyl substances and
biomarkers of liver function: The Korean national environmental health survey 2015-2017. Chemosphere 322: 138208.
http: //dx. doi. org/1.0.1.01.6/i. chemosphere.202 3.1.38208
R-17
-------
APRIL 2024
Kim, S; Choi, K; Ji, K; Seo, J; Kho, Y; Park, J, et al. (2011). Trans-placental transfer of thirteen perfluorinated compounds and
relations with fetal thyroid hormones. Environmental Science and Technology 45: 7465-7472.
littp: //dx. doi. org/1.0.1.021./es202408a
Kim, SJ; Heo, SH; Lee, DS; Hwang, IG; Lee, YB; Cho, HY. (2016b). Gender differences in pharmacokinetics and tissue
distribution of 3 perfluoroalkyl and polyfhioroalkyl substances in rats. Food and Chemical Toxicology 97: 243-255.
Kim, SK; Kannan, K. (2007). Perfluorinated acids in air, rain, snow, surface runoff, and lakes: relative importance of pathways to
contamination of urban lakes. Environmental Science and Technology 41: 8328-8334.
http://dx.doi.org/10.1021/es072107t
Kim, YR; White, N; Braunig, J; Vijayasarathy, S; Mueller, JF; Knox, CL, et al. (2020b). Per- and poly-fluoroalkyl substances
(PFASs) in follicular fluid from women experiencing infertility in Australia. Environmental Research 190: 109963.
Kishi, R; Nakajima, T; Goudarzi, H; Kobayashi, S; Sasaki, S; Okada, E, et al. (2015). The association of prenatal exposure to
perfluorinated chemicals with maternal essential and long-chain polyunsaturated fatty acids during pregnancy and the
birth weight of their offspring: the hokkaido study. Environmental Health Perspectives 123: 1038-1045.
Klamt, A; Huniar, U; Spycher, S; Keldenich, J. (2008). COSMOmic: a mechanistic approach to the calculation of membrane-
water partition coefficients and internal distributions within membranes and micelles. Journal of Physical Chemistry B
112: 12148-12157. http://dx.doi.org/10.1021/ro801.736k
Knox, SS; Jackson, T; Javins, B; Frisbee, SJ; Shankar, A; Ducatman, AM. (2011). Implications of early menopause in women
exposed to perfluorocarbons. Journal of Clinical Endocrinology and Metabolism 96: 1747-1753.
http://dx.doi.org/10.121.0/ic.201.0-2401.
Kobayashi, S; Azumi, K; Goudarzi, H; Araki, A; Miyashita, C; Kobayashi, S, et al. (2017). Effects of prenatal perfluoroalkyl acid
exposure on cord blood IGF2/H19 methylation and ponderal index: The Hokkaido Study. Journal of Exposure Science
and Environmental Epidemiology 27: 251-259.
Kobayashi, S; Sata, F; Ikeda-Araki, A; Miyashita, C; Goudarzi, H; Iwasaki, Y, et al. (2022). Relationships between maternal
perfluoroalkyl substance levels, polymorphisms of receptor genes, and adverse birth outcomes in the Hokkaido birth
cohort study, Japan. Reproductive Toxicology 107: 112-122. littp://dx.doi.org/1.0.1.01.6/i .reprotox.2021.12.004
Koponen, J; Winkens, K; Airaksinen, R; Berger, U; Vestergren, R; Cousins, IT, et al. (2018). Longitudinal trends of per- and
polyfhioroalkyl substances in children's serum. Environment International 121: 591-599.
littp: //dx. doi. org/1.0.1.01.6/i. eiivint .201.8.09.006
Koshy, TT; Attina, TM; Ghassabian, A; Gilbert, J; Burdine, LK; Marmor, M, et al. (2017). Serum perfluoroalkyl substances and
cardiometabolic consequences in adolescents exposed to the World Trade Center disaster and a matched comparison
group. Environment International 109: 128-135.
Koskela, A; Koponen, J; Lehenkari, P; Viluksela, M; Korkalainen, M; Tuukkanen, J. (2017). Perfluoroalkyl substances in human
bone: concentrations in bones and effects on bone cell differentiation. Scientific Reports 7: 6841.
Kotlarz, N; Mccord, J; Collier, D; Lea, CS; Strynar, M; Lindstrom, AB, et al. (2020). Measurement of Novel, Drinking Water-
Associated PFAS in Blood from Adults and Children in Wilmington, North Carolina. Environmental Health
Perspectives 128: 77005.
Kotthoff, M; Mtiller, J; Jtirling, H; Schlummer, M; Fiedler, D. (2015). Perfluoroalkyl and polyfhioroalkyl substances in consumer
products. Environmental Science and Pollution Research 22: 14546-14559. http://dx.doi.org/10.1007/sl 1356-015-
4202-7
Kummu, M; Sieppi, E; Koponen, J; Laatio, L; VaaHaaKangas, K; Kiviranta, H, et al. (2015). Organic anion transporter 4 (OAT
4) modifies placental transfer of perfluorinated alkyl acids PFOS and PFOA in human placental ex vivo perfusion
system. Placenta 36: 1185-1191.
Kvalem, HE; Nygaard, UC; Ledrup Carlsen, KC; Carlsen, KH; Haug, LS; Granum, B. (2020). Perfluoroalkyl substances, airways
infections, allergy and asthma related health outcomes - Implications of gender, exposure period and study design.
Environment International 134: 105259.
Kvist, L; Giwercman, YL; Jonsson, BA; Lindh, CH; Bonde, JP; Toft, G, et al. (2012). Serum levels of perfluorinated compounds
and sperm Y:X chromosome ratio in two European populations and in Inuit from Greenland. Reproductive Toxicology
34: 644-650.
Kwo, PY; Cohen, SM; Lim, JK. (2017). ACG Clinical Guideline: Evaluation of Abnormal Liver Chemistries. American Journal
of Gastroenterology 112: 18-35. http://dx.doi.org/10.1.038/aig.201.6,517
Kwon, EJ; Shin, JS; Kim, BM; Shah-Kulkarni, S; Park, H; Kho, YL, et al. (2016). Prenatal exposure to perfluorinated
compounds affects birth weight through GSTM1 polymorphism. Journal of Occupational and Environmental Medicine
58: el98-e205.
Lai, KP; Lee, JC; Wan, HT; Li, JW; Wong, AY; Chan, TF, et al. (2017a). Effects of in Utero PFOS Exposure on Transcriptome,
Lipidome, and Function of Mouse Testis. Environmental Science and Technology 51: 8782-8794.
Lai, KP; Li, JW; Cheung, A; Li, R; Billah, MB; Chan, TF; Wong, CK. (2017b). Transcriptome sequencing reveals prenatal PFOS
exposure on liver disorders. Environmental Pollution 223: 416-425.
Lai, KP; Ng, AH; Wan, HT; Wong, AY; Leung, CC; Li, R; Wong, CK. (2018). Dietary exposure to the environmental chemical,
pfos on the diversity of gut microbiota, associated with the development of metabolic syndrome. Frontiers in
Microbiology 9: 2552.
Lau, C; Thibodeaux, JR; Hanson, RG; Rogers, JM; Grey, BE; Stanton, ME, et al. (2003). Exposure to perfluorooctane sulfonate
during pregnancy in rat and mouse. II: postnatal evaluation. Toxicological Sciences 74: 382-392.
R-18
-------
APRIL 2024
Lauritzen, HB; Larose, TL; 0ien, T; Sandanger, TM; Odland, J0; van de Bor, M; Jacobsen, GW. (2017). Maternal serum levels
of perfluoroalkyl substances and organochlorines and indices of fetal growth: a Scandinavian case-cohort study.
Pediatric Research 81: 33-42.
Lauritzen, HB; Larose, TL; 0ien, T; Sandanger, TM; Odland, J0; van de Bor, M; Jacobsen, GW. (2018). Prenatal exposure to
persistent organic pollutants and child overweight/obesity at 5-year follow-up: a prospective cohort study.
Environmental Health: A Global Access Science Source 17: 9.
Lebeaux, RM; Doherty, BT; Gallagher, LG; Zoeller, RT; Hoofnagle, AN; Calafat, AM, et al. (2020). Maternal serum
perfluoroalkyl substance mixtures and thyroid hormone concentrations in maternal and cord sera: The HOME Study.
Environmental Research 185: 109395.
Lee, CK; Kang, SG; Lee, JT; Lee, SW; Kim, JH; Kim, DH, et al. (2015). Effects of perfluorooctane sulfuric acid on placental
PRL-family hormone production and fetal growth retardation in mice. Molecular and Cellular Endocrinology 401: 165-
172.
Lee, S; Kim, S; Park, J; Kim, HJ; Choi, G; Choi, S, et al. (2017). Perfluoroalkyl substances (PFASs) in breast milk from Korea:
Time-course trends, influencing factors, and infant exposure. Science of the Total Environment 612: 286-292.
Lefebvre, DE; Curran, I; Armstrong, C; Coady, L; Parenteau, M; Liston, V, et al. (2008). Immunomodulatory effects of dietary
potassium perfluorooctane sulfonate (PFOS) exposure in adult Sprague-Dawley rats. Journal of Toxicology and
Environmental Health, Part A: Current Issues 71: 1516-1525. fattp://dx. doi.org/1.0.1.080/1.5287390802391.943
Lenters, V; Iszatt, N; Forns, J; Cechova, E; Kocan, A; Legler, J, et al. (2019). Early-life exposure to persistent organic pollutants
(OCPs, PBDEs, PCBs, PFASs) and attention-deficit/hyperactivity disorder: A multi-pollutant analysis of a Norwegian
birth cohort. Environment International 125: 33-42.
Lenters, V; Portengen, L; Rignell-Hydbom, A; Jonsson, BA; Lindh, CH; Piersma, AH, et al. (2016). Prenatal phthalate,
perfluoroalkyl acid, and organochlorine exposures and term birth weight in three birth cohorts: multi-pollutant models
based on elastic net regression. Environmental Health Perspectives 124: 365-372.
Leter, G; Consales, C; Eleuteri, P; Uccelli, R; Specht, IO; Toft, G, et al. (2014). Exposure to perfluoroalkyl substances and sperm
DNA global methylation in Arctic and European populations. Environmental and Molecular Mutagenesis 55: 591-600.
Lewis, RC; Johns, LE; Meeker, JD. (2015). Serum Biomarkers of Exposure to Perfluoroalkyl Substances in Relation to Serum
Testosterone and Measures of Thyroid Function among Adults and Adolescents fromNHANES 2011-2012.
International Journal of Environmental Research and Public Health 12: 6098-6114.
Li, H; Chen, J; Jingchao, L; Yang, J; Tan, Z; Li, L, et al. (2023). Association of exposure to perfluoroalkyl substances and risk of
the acute coronary syndrome: A case-control study in Shijiazhuang Hebei Province. Chemosphere 313: 137464.
fattp: //dx. doi. org/1.0.1.01.6/j .chemosphere.2022.1.37464
Li, J; Cai, D; Chu, C; Li, QQ; Zhou, Y; Hu, LW, et al. (2020a). Transplacental Transfer of Per- and Polyfluoroalkyl Substances
(PFASs): Differences between Preterm and Full-Term Deliveries and Associations with Placental Transporter mRNA
Expression. Environmental Science and Technology 54: 5062-5070.
Li, MC. (2020). Serum Per- and Polyfluoroalkyl Substances Are Associated with Increased Hearing Impairment: A Re-Analysis
of the National Health and Nutrition Examination Survey Data. International Journal of Environmental Research and
Public Health 17.
Li, N; Liu, Y; Papandonatos, GD; Calafat, AM; Eaton, CB; Kelsey, KT, et al. (2021a). Gestational and childhood exposure to
per- and polyfluoroalkyl substances and cardiometabolic risk at age 12 years. Environment International 147: 106344.
http://dx.doi.Org/10.1016/i.enviiit.2020.106344
Li, X; Li, T; Wang, Z; Wei, J; Liu, J; Zhang, Y; Zhao, Z. (2021b). Distribution of perfluorooctane sulfonate in mice and its effect
on liver lipidomic. Talanta 226: 122150. http://dx.doi.org/1.0.1.01.6/j.talanta.2021.. 1.221.50
Li, X; Song, F; Liu, X; Shan, A; Huang, Y; Yang, Z, et al. (2022). Perfluoroalkyl substances (PFASs) as risk factors for breast
cancer: a case-control study in Chinese population. Environmental Health 21: 83. fattp://dx.doi.org/1.0.1.1.86/s 1.2940-
022-00895-3
Li, X; Ye, L; Ge, Y; Yuan, K; Zhang, Y; Liang, Y, et al. (2016). In utero perfluorooctane sulfonate exposure causes low body
weights of fetal rats: A mechanism study. Placenta 39: 125-133.
Li, Y; Cheng, Y; Xie, Z; Zeng, F. (2017). Perfluorinated alkyl substances in serum of the southern Chinese general population
and potential impact on thyroid hormones. Scientific Reports 7: 43380.
Li, Y; Fletcher, T; Mucs, D; Scott, K; Lindh, CH; Tallving, P; Jakobsson, K. (2018). Half-lives of PFOS, PFHxS and PFOA after
end of exposure to contaminated drinking water. Occupational and Environmental Medicine 75: 46-51.
Li, Y; Yu, N; Du, L; Shi, W; Yu, H; Song, M; Wei, S. (2020b). Transplacental Transfer of Per- and Polyfluoroalkyl Substances
Identified in Paired Maternal and Cord Sera Using Suspect and Nontarget Screening. Environmental Science and
Technology 54: 3407-3416.
Liang, JL; Tiwari, T; Moro, P; Messonnier, NE; Reingold, A; Sawyer, M; Clark, TA. (2018). Prevention of pertussis, tetanus, and
diphtheria with vaccines in the United States: Recommendations of the Advisory Committee on Immunization
Practices (ACIP). US Centers for Disease Control and Prevention Morbidity and Mortality Weekly Report
Recommendations and Report 67: 1-44. http://dx.doi.org/10.15585/mmwr.rr6702al
Liao, Q; Tang, P; Song, Y; Liu, B; Huang, H; Liang, J, et al. (2022). Association of single and multiple prefluoroalkyl substances
exposure with preterm birth: Results from a Chinese birth cohort study. Chemosphere 307: 135741.
fattp: //dx. doi. org/1.0.1.01.6/j. chemosphere.2022.1.35741.
R-19
-------
APRIL 2024
Liao, S; Yao, W; Cheang, I; Tang, X; Yin, T; Lu, X, et al. (2020). Association between perfluoroalkyl acids and the prevalence
of hypertension among US adults. Ecotoxicology and Environmental Safety 196: 110589.
Lien, GW; Huang, CC; Shiu, JS; Chen, MH; Hsieh, WS; Guo, YL; Chen, PC. (2016). Perfluoroalkyl substances in cord blood
and attention deficit/hyperactivity disorder symptoms in seven-year-old children. Chemosphere 156: 118-127.
Liew, Z; Luo, J; Nohr, EA; Bech, BH; Bossi, R; Arah, OA; Olsen, J. (2020). Maternal Plasma Perfluoroalkyl Substances and
Miscarriage: A Nested Case-Control Study in the Danish National Birth Cohort. Environmental Health Perspectives
128: 47007.
Liew, Z; Ritz, B; Bach, CC; Asarnow, RF; Bech, BH; Nohr, EA, et al. (2018). Prenatal exposure to perfluoroalkyl substances and
iq scores at age 5; a study in the danish national birth cohort. Environmental Health Perspectives 126: 067004.
Liew, Z; Ritz, B; Bonefeld-Jergensen, EC; Henriksen, TB; Nohr, EA; Bech, BH, et al. (2014). Prenatal exposure to
perfluoroalkyl substances and the risk of congenital cerebral palsy in children. American Journal of Epidemiology 180:
574-581. http://dx.doi.org/10.1093/aie/kwul79
Liew, Z; Ritz, B; von Ehrenstein, OS; Bech, BH; Nohr, EA; Fei, C, et al. (2015). Attention deficit/hyperactivity disorder and
childhood autism in association with prenatal exposure to perfluoroalkyl substances: A nested case-control study in the
Danish National Birth Cohort. Environmental Health Perspectives 123: 367-373.
Lin, CY; Chen, PC; Lin, YC; Lin, LY. (2009). Association among serum perfluoroalkyl chemicals, glucose homeostasis, and
metabolic syndrome in adolescents and adults. Diabetes Care 32: 702-707. fattp:/klx.doiorg/1.0.2337/dc08-1.81.6
Lin, CY; Chen, PC; Lo, SC; Torng, PL; Sung, FC; Su, TC. (2016). The association of carotid intima-media thickness with serum
Level of perfluorinated chemicals and endothelium-platelet microparticles in adolescents and young adults.
Environment International 94: 292-299.
Lin, CY; Lee, HL; Chen, CW; Wang, C; Sung, FC; Su, TC. (2022). Global DNA methylation mediates the association between
serum perfluorooctane sulfonate and carotid intima-media thickness in young and middle-aged Taiwanese populations.
Ecotoxicology and Environmental Safety 241: 113782. http://dx.doi.org/1.0.1016/i.ecoenv.2022.1.1.3782
Lin, CY; Lin, LY; Chiang, CK; Wang, WJ; Su, YN; Hung, KY; Chen, PC. (2010). Investigation of the Associations Between
Low-Dose Serum Perfluorinated Chemicals and Liver Enzymes in US Adults. American Journal of Gastroenterology
105: 1354-1363. http://dx.doi.org/10.1038/aig.2009.707
Lin, CY; Lin, LY; Wen, TW; Lien, GW; Chien, KL; Hsu, SH, et al. (2013). Association between levels of serum perfluorooctane
sulfate and carotid artery intima-media thickness in adolescents and young adults. International Journal of Cardiology
168: 3309-3316.
Lin, HW; Feng, FIX; Chen, L; Yuan, XJ; Tan, Z. (2020a). Maternal exposure to environmental endocrine disrupters during
pregnancy is associated with pediatric germ cell tumors. Nagoya Journal of Medical Science 82: 323-333.
Lin, LY; Wen, LL; Su, TC; Chen, PC; Lin, CY. (2014). Negative association between serum perfluorooctane sulfate
concentration and bone mineral density in US premenopausal women: NHANES, 2005-2008. Journal of Clinical
Endocrinology and Metabolism 99: 2173-2180.
Lin, P; Cardenas, A; Hauser, R; Gold, DR; Kleinman, K; Hivert, MF, et al. (2019). Per- and polyfluoroalkyl substances and
blood lipid levels in pre-diabetic adults-longitudinal analysis of the diabetes prevention program outcomes study.
Environment International 129: 343-353.
Lin, PD; Cardenas, A; Hauser, R; Gold, DR; Kleinman, KP; Hivert, MF, et al. (2020b). Per- and polyfluoroalkyl substances and
blood pressure in pre-diabetic adults-cross-sectional and longitudinal analyses of the diabetes prevention program
outcomes study. Environment International 137: 105573.
Lin, TW; Chen, MK; Lin, CC; Chen, MH; Tsai, MS; Chan, DC, et al. (2020c). Association between exposure to perfluoroalkyl
substances and metabolic syndrome and related outcomes among older residents living near a Science Park in Taiwan.
International Journal of Hygiene and Environmental Health 230: 113607. http://dx.doi.Org/10.1016/i.iiheh.2020.l 1.3607
Linakis, MW; Gustafson, P; Allen, BC; Bachand, AM; Van Landingham, C; Keast, DR; Longnecker, MP. (2022). Is the
cholesterol-perfluoroalkyl substance association confounded by dietary fiber intake?: a Bayesian analysis of NHANES
data with adjustment for measurement error in fiber intake. Environmental Health 21: 114.
http://dx.doi.org/10.1186/sl2940-022-0Q923-2
Lind, DV; Priskorn, L; Lassen, TH; Nielsen, F; Kyhl, HB; Kristensen, DM, et al. (2017a). Prenatal exposure to perfluoroalkyl
substances and anogenital distance at 3 months of age in a Danish mother-child cohort. Reproductive Toxicology 68:
200-206.
Lind, L; Zethelius, B; Salihovic, S; van Bavel, B; Lind, PM. (2014). Circulating levels of perfluoroalkyl substances and prevalent
diabetes in the elderly. Diabetologia 57: 473-479.
Lind, PM; Salihovic, S; van Bavel, B; Lind, L. (2017b). Circulating levels of perfluoroalkyl substances (PFASs) and carotid
artery atherosclerosis. Environmental Research 152: 157-164.
Liu, G; Dhana, K; Furtado, JD; Rood, J; Zong, G; Liang, L, et al. (2018a). Perfluoroalkyl substances and changes in body weight
and resting metabolic rate in response to weight-loss diets: A prospective study. PLoS Medicine 15: el002502.
Liu, G; Zhang, B; Hu, Y; Rood, J; Liang, L; Qi, L, et al. (2020a). Associations of Perfluoroalkyl substances with blood lipids and
Apolipoproteins in lipoprotein subspecies: the POUNDS-lost study. Environmental Health: A Global Access Science
Source 19: 5.
Liu, H; Pan, Y; Jin, S; Li, Y; Zhao, L; Sun, X, et al. (2020b). Associations of per-/polyfluoroalkyl substances with
glucocorticoids and progestogens in newborns. Environment International 140: 105636.
R-20
-------
APRIL 2024
Liu, HS; Wen, LL; Chu, PL; Lin, CY. (2018b). Association among total serum isomers of perlluorinated chemicals, glucose
homeostasis, lipid profiles, serum protein and metabolic syndrome in adults: NHANES, 2013-2014. Environmental
Pollution 232: 73-79.
Liu, J; Li, J; Liu, Y; Chan, HM; Zhao, Y; Cai, Z; Wu, Y. (2011). Comparison on gestation and lactation exposure of
perlluorinated compounds for newborns. Environment International 37: 1206-1212.
littp: //dx. doi. org/1.0.1.01.6/j. eiivint .201.1. .05.001.
Liu, JJ; Cui, XX; Tan, YW; Dong, PX; Ou, YQ; Li, QQ, et al. (2022). Per- and perlluoroalkyl substances alternatives, mixtures
and liver function in adults: A community-based population study in China. Environment International 163: 107179.
http://dx.doi.Org/10.1016/i.enviiit.2022.107179
Liu, L; Liu, W; Song, J; Yu, H; Jin, Y; Oami, K, et al. (2009). A comparative study on oxidative damage and distributions of
perfluorooctane sulfonate (PFOS) in mice at different postnatal developmental stages. Journal of Toxicological
Sciences 34: 245-254. http://dx.doi.org/10.2131/its.34.245
Liu, M; Zhang, G; Meng, L; Han, X; Li, Y; Shi, Y, et al. (2021). Associations between novel and legacy per- and polyfluoroalkyl
substances in human serum and thyroid cancer: A case and healthy population in Shandong Province, East China.
Environmental Science & Technology, http://dx.doi.org/10.1.021/acs.est. 1.c028S0
Liu, WS; Lai, YT; Chan, HL; Li, SY; Lin, CC; Liu, CK, et al. (2018c). Associations between perfluorinated chemicals and serum
biochemical markers and performance status in uremic patients under hemodialysis. PLoS ONE 13: e0200271.
Liu, X; Guo, Z; Krebs, KA; Pope, RH; Roache, NF. (2014). Concentrations and trends of perfluorinated chemicals in potential
indoor sources from 2007 through 2011 in the US. Chemosphere 98: 51-57.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere.201.3.1.0.001.
Liu, X; Zhang, L; Chen, L; Li, J; Wang, Y; Wang, J, et al. (2019). Structure-based investigation on the association between
perfluoroalkyl acids exposure and both gestational diabetes mellitus and glucose homeostasis in pregnant women.
Environment International 127: 85-93.
Liu, Y; Cao, Z; Zong, W; Liu, R. (2017). Interaction rule and mechanism of perfluoroalkyl sulfonates containing different carbon
chains with human serum albumin. RSC Advances 7: 24781-24788.
Loccisano, AE; Campbell, JL; Butenhoff, JL; Andersen, ME; Clewell, HJ. (2012a). Evaluation of placental and lactational
pharmacokinetics of PFOA and PFOS in the pregnant, lactating, fetal and neonatal rat using a physiologically based
pharmacokinetic model. Reproductive Toxicology 33: 468-490. http://dx.doi.Org/10.1016/i.reprotox.2011.07.003
Loccisano, AE; Campbell, JL, Jr; Butenhoff, JL; Andersen, ME; Clewell, HJ, III. (2012b). Comparison and evaluation of
pharmacokinetics of PFOA and PFOS in the adult rat using a physiologically based pharmacokinetic model.
Reproductive Toxicology 33: 452-467. littp://dx.doi.org/1.0.1.01.6/j.reprotox.201.1..04.006
Loi, EIH; Yeung, LWY; Taniyasu, S; Lam, PKS; Kannan, K; Yamashita, N. (2011). Trophic Magnification of Poly- and
Perfluorinated Compounds in a Subtropical Food Web. Environmental Science and Technology 45: 5506-5513.
littp: //dx. doi. org/1.0.1.021./es200432n
Long, M; Ghisari, M; Kjeldsen, L; Wielsee, M; Nergaard-Pedersen, B; Mortensen, EL, et al. (2019). Autism spectrum disorders,
endocrine disrupting compounds, and heavy metals in amniotic fluid: a case-control study. Molecular autism 10: 1.
Long, Y; Wang, Y; Ji, G; Yan, L; Hu, F; Gu, A. (2013). Neurotoxicity of perfluorooctane sulfonate to hippocampal cells in adult
mice. PLoS ONE 8: e54176. http://dx.doi.org/10.1371/ioumal.pone.0054176
Looker, C; Luster, MI; Calafat, AM; Johnson, VJ; Burleson, GR; Burleson, FG; Fletcher, T. (2014). Influenza vaccine response
in adults exposed to perfluorooctanoate and perfluorooctanesulfonate. Toxicological Sciences 138: 76-88.
littp: //dx. doi. org/1.0.1.093/tox sci/kf "t2 6 9
Lopez-Doval, S; Salgado, R; Fernandez-Perez, B; Lafuente, A. (2015). Possible role of serotonin and neuropeptide Y on the
disruption of the reproductive axis activity by perfluorooctane sulfonate. Toxicology Letters 233: 138-147.
Lopez-Doval, S; Salgado, R; Pereiro, N; Moyano, R; Lafuente, A. (2014). Perfluorooctane sulfonate effects on the reproductive
axis in adult male rats. Environmental Research 134: 158-168. http://dx.doi.Org/10.1016/i.envres.2014.07.006
Lopez-Espinosa, MJ; Carrizosa, C; Luster, MI; Margolick, JB; Costa, O; Leonardi, GS; Fletcher, T. (2021). Perfluoroalkyl
substances and immune cell counts in adults from the Mid-Ohio Valley (USA). Environment International 156:
106599. http://dx.doi.org/1.0.1.01.6/j.envint.2021.106599
Lopez-Espinosa, MJ; Fletcher, T; Armstrong, B, en; Genser, B; Dhatariya, K; Mondal, D, et al. (2011). Association of
Perfluorooctanoic Acid (PFOA) and Perfluorooctane Sulfonate (PFOS) with Age of Puberty among Children Living
near a Chemical Plant. Environmental Science and Technology 45: 8160-8166. http://dx.doi.org/1.0.1.021/esl038694
Lopez-Espinosa, MJ; Mondal, D; Armstrong, B; Bloom, MS; Fletcher, T. (2012). Thyroid function and perfluoroalkyl acids in
children living near a chemical plant. Environmental Health Perspectives 120: 1036-1041.
http://dx.doi.org/10.1289/ehp. 1.1.04370
Lopez-Espinosa, MJ; Mondal, D; Armstrong, BG; Eskenazi, B; Fletcher, T. (2016). Perfluoroalkyl Substances, Sex Hormones,
and Insulin-like Growth Factor-1 at 6-9 Years of Age: A Cross-Sectional Analysis within the C8 Health Project.
Environmental Health Perspectives 124: 1269-1275.
Lorber, M; Eaglesham, GE; Hobson, P; Toms, LM; Mueller, JF; Thompson, JS. (2015). The effect of ongoing blood loss on
human serum concentrations of perfluorinated acids. Chemosphere 118: 170-177.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere.201.4.07.093
Luebker, DJ; Case, MT; York, RG; Moore, JA; Hansen, KJ; Butenhoff, JL. (2005a). Two-generation reproduction and cross-
foster studies of perfluorooctanesulfonate (PFOS) in rats. Toxicology 215: 126-148.
R-21
-------
APRIL 2024
Luebker, DJ; Hansen, KJ; Bass, NM; Butenhoff, JL; Seacat, AM. (2002). Interactions of fluorochemicals with rat liver fatty acid-
binding protein. Toxicology 176: 175-185. http://dx.doi.org/10.1.016/S03Q0-483X(02)0Q081.-1.
Luebker, DJ; York, RG; Hansen, KJ; Moore, JA; Butenhoff, JL. (2005b). Neonatal mortality from in utero exposure to
perfluorooctanesulfonate (PFOS) in Sprague-Dawley rats: dose-response, and biochemical and pharamacokinetic
parameters. Toxicology 215: 149-169.
Lum, KJ; Sundaram, R; Barr, DB; Louis, TA; Buck Louis, GM. (2017). Perfluoroalkyl Chemicals, Menstrual Cycle Length, and
Fecundity: Findings from a Prospective Pregnancy Study. Epidemiology 28: 90-98.
Luo, D; Wu, W; Pan, Y; Du, B; Shen, M; Zeng, L. (2021). Associations of prenatal exposure to per- and polyfluoroalkyl
substances with the neonatal birth size and hormones in the growth hormone/insulin-like growth factor axis.
Environmental Science & Technology 55: 11859-11873. http://dx.doi.org/10.1021/acs.est.lc02670
Luo, J; Ramlau-Hansen, CH; Kesmodel, US; Xiao, J; Vasiliou, V; Deziel, NC, et al. (2022). Prenatal Exposure to Per- and
Polyfluoroalkyl Substances and Facial Features at 5 Years of Age: A Study from the Danish National Birth Cohort.
Environmental Health Perspectives 130: 17006. fattp://dx.doi.org/1.0.1.289/EHP9478
Lv, QY; Wan, B; Guo, LH; Yang, Y; Ren, XM; Zhang, H. (2015). In vivo immunotoxicity of perfluorooctane sulfonate in
BALB/c mice: Identification of T-cell receptor and calcium-mediated signaling pathway disruption through gene
expression profiling of the spleen. Chemico-Biological Interactions 240: 84-93.
http://dx.doi.Org/10.1016/i.cbi.2015.07.015
Lv, Z; Li, G; Li, Y; Ying, C; Chen, J; Chen, T, et al. (2013). Glucose and lipid homeostasis in adult rat is impaired by early-life
exposure to perfluorooctane sulfonate. Environmental Toxicology 28: 532-542. http://dx.doi.org/10.1002/tox.20747
Lyall, K; Yau, VM; Hansen, R; Kharrazi, M; Yoshida, CK; Calafat, AM, et al. (2018). Prenatal maternal serum concentrations of
per- and polyfluoroalkyl substances in association with autism spectrum disorder and intellectual disability.
Environmental Health Perspectives 126: 017001.
Lyngse, J; Ramlau-Hansen, CH; Hayer, BB; Stevring, H; Bonde, JP; Jonsson, BA, et al. (2014). Menstrual cycle characteristics
in fertile women from Greenland, Poland and Ukraine exposed to perfluorinated chemicals: a cross-sectional study.
Human Reproduction 29: 359-367.
Ma, S; Xu, C; Ma, J; Wang, Z; Zhang, Y; Shu, Y; Mo, X. (2019). Association between perfluoroalkyl substance concentrations
and blood pressure in adolescents. Environmental Pollution 254: 112971.
Ma, X; Fisher, JA; Vopham, T; Vasiliou, V; Jones, RR. (2023). Associations between per- and polyfluoroalkyl substances, liver
function, and daily alcohol consumption in a sample of U.S. adults. Environmental Research 235: 116651.
fattp: //dx. doi. org/1.0.1.01.6/i .envres.202 3.116651.
Maestri, L; Negri, S; Ferrari, M; Ghittori, S; Fabris, F; Danesino, P; Imbriani, M. (2006). Determination of perfluorooctanoic
acid and perfluorooctanesulfonate in human tissues by liquid chromatography/single quadrupole mass spectrometry.
Rapid Communications in Mass Spectrometry 20: 2728-2734. http://dx.doi.org/lQ.lQQ2/rcm.2661
Maisonet, M; Calafat, AM; Marcus, M; Jaakkola, JJ; Lashen, H. (2015a). Prenatal exposure to perfluoroalkyl acids and serum
testosterone concentrations at 15 years of age in female ALSPAC study participants. Environmental Health
Perspectives 123: 1325-1330.
Maisonet, M; Nayha, S; Lawlor, DA; Marcus, M. (2015b). Prenatal exposures to perfluoroalkyl acids and serum lipids at ages 7
and 15 in females. Environment International 82: 49-60. fattp://dx. doi.org/1.0.1.01.6/i.envint.201.5.05.001.
Maisonet, M; Terrell, ML; Mcgeehin, MA; Christensen, KY; Holmes, A; Calafat, AM; Marcus, M. (2012). Maternal
concentrations of polyfluoroalkyl compounds during pregnancy and fetal and postnatal growth in British girls.
Environmental Health Perspectives 120: 1432-1437. http://dx.doi.org/10.1289/ehp. 1.003096
Makey, CM; Webster, TF; Martin, JW; Shoeib, M; Harner, T; Dix-Cooper, L; Webster, GM. (2017). Airborne precursors predict
maternal serum perfluoroalkyl acid concentrations. Environmental Science and Technology 51: 7667-7675.
fattp: //dx. doi. org/1.0.1.021. /ac s. est.71)0061.5
Mamsen, LS; Bjorvang, RD; Mucs, D; Vinnars, MT; Papadogiannakis, N; Lindh, CH, et al. (2019). Concentrations of
perfluoroalkyl substances (PFASs) in human embryonic and fetal organs from first, second, and third trimester
pregnancies. Environment International 124: 482-492.
Mamsen, LS; Jonsson, BAG; Lindh, CH; Olesen, RH; Larsen, A; Ernst, E, et al. (2017). Concentration of perfluorinated
compounds and cotinine in human foetal organs, placenta, and maternal plasma. Science of the Total Environment 596-
597: 97-105.
Mancini, FR; Cano-Sancho, G; Gambaretti, J; Marchand, P; Boutron-Ruault, MC; Severi, G, et al. (2020). Perfluorinated
alkylated substances serum concentration and breast cancer risk: Evidence from a nested case-control study in the
French E3N cohort. International Journal of Cancer 146: 917-928.
Mancini, FR; Rajaobelina, K; Praud, D; Dow, C; Antignac, JP; Kvaskoff, M, et al. (2018). Nonlinear associations between
dietary exposures to perfluorooctanoic acid (PFOA) or perfluorooctane sulfonate (PFOS) and type 2 diabetes risk in
women: Findings from the E3N cohort study. International Journal of Hygiene and Environmental Health 221: 1054-
1060.
Manzano-Salgado, CB; Casas, M; Lopez-Espinosa, MJ; Ballester, F; Basterrechea, M; Grimalt, JO, et al. (2015). Transfer of
perfluoroalkyl substances from mother to fetus in a Spanish birth cohort. Environmental Research 142: 471-478.
Manzano-Salgado, CB; Casas, M; Lopez-Espinosa, MJ; Ballester, F; Iniguez, C; Martinez, D, et al. (2017a). Prenatal exposure to
perfluoroalkyl substances and birth outcomes in a Spanish birth cohort. Environment International 108: 278-284.
R-22
-------
APRIL 2024
Manzano-Salgado, CB; Casas, M; Lopez-Espinosa, MJ; Ballester, F; Iniguez, C; Martinez, D, et al. (2017b). Prenatal exposure to
perlluoroalkyl substances and cardiometabolic risk in children from the Spanish INMA birth cohort study.
Environmental Health Perspectives 125: 097018.
Manzano-Salgado, CB; Granum, B; Lopez-Espinosa, MJ; Ballester, F; Iniguez, C; Gascon, M, et al. (2019). Prenatal exposure to
perlluoroalkyl substances, immune-related outcomes, and lung function in children from a Spanish birth cohort study.
International Journal of Hygiene and Environmental Health 222: 945-954.
Maranhao Neto, GA; Polcrova, AB; Pospisilova, A; Blaha, L; Klanova, J; Bobak, M; Gonzalez-Rivas, JP. (2022). Associations
between Per- and Polyfluoroalkyl Substances (PFAS) and Cardiometabolic Biomarkers in Adults of Czechia: The
Kardiovize Study. International Journal of Environmental Research and Public Health 19: 13898.
http://dx.doi.org/10.3390/iierphl 921.1.3898
Marks, KJ; Jeddy, Z; Flanders, WD; Northstone, K; Fraser, A; Calafat, AM, et al. (2019). Maternal serum concentrations of
perlluoroalkyl substances during pregnancy and gestational weight gain: The Avon Longitudinal Study of Parents and
Children. Reproductive Toxicology 90: 8-14.
Martin, JW; Smithwick, MM; Braune, BM; Hoekstra, PF; Muir, DC; Mabury, SA. (2004). Identification of long-chain
perlluorinated acids in biota from the Canadian Arctic. Environmental Science and Technology 38: 373-380.
http://dx.doi.org/10.1021/es034727+
Martin, MT; Brennan, RJ; Hu, W; Ayanoglu, E; Lau, C; Ren, H, et al. (2007). Toxicogenomic study of triazole fungicides and
perlluoroalkyl acids in rat livers predicts toxicity and categorizes chemicals based on mechanisms of toxicity.
Toxicological Sciences 97: 595-613. http://dx.doi.org/10.lQ93/toxsci/kfmQ65
Martinsson, M; Nielsen, C; Bjork, J; Rylander, L; Malmqvist, E; Lindh, C; Rignell-Hydbom, A. (2020). Intrauterine exposure to
pertluorinated compounds and overweight at age 4: A case-control study. PLoS ONE 15: e0230137.
Matilla-Santander, N; Valvi, D; Lopez-Espinosa, MJ; Manzano-Salgado, CB; Ballester, F; Ibarluzea, J, et al. (2017). Exposure to
Pertluoroalkyl Substances and Metabolic Outcomes in Pregnant Women: Evidence from the Spanish INMA Birth
Cohorts. Environmental Health Perspectives 125: 117004.
Mattsson, K; Rignell-Hydbom, A; Holmberg, S; Thelin, A; Jonsson, BA; Lindh, CH, et al. (2015). Levels of perlluoroalkyl
substances and risk of coronary heart disease: Findings from a population-based longitudinal study. Environmental
Research 142: 148-154.
Mccoy, JA; Bangma, JT; Reiner, JL; Bowden, JA; Schnorr, J; Slowey, M, et al. (2017). Associations between pertluorinated
alkyl acids in blood and ovarian follicular fluid and ovarian function in women undergoing assisted reproductive
treatment. Science of the Total Environment 605-606: 9-17.
Melzer, D; Rice, N; Depledge, MH; Henley, WE; Galloway, TS. (2010). Association between serum pertluorooctanoic acid
(PFOA) and thyroid disease in the U.S. National Health and Nutrition Examination Survey. Environmental Health
Perspectives 118: 686-692. http://dx.doi.org/10.1289/ehp.Q9Q1584
Meng, Q; Inoue, K; Ritz, B; Olsen, J; Liew, Z. (2018). Prenatal exposure to pertluoroalkyl substances and birth outcomes; an
updated analysis from the danish national birth cohort. International Journal of Environmental Research and Public
Health 15: 1832.
Mi, X; Lin, SQ; Zhang, XF; Li, JJ; Pei, LJ; Jin, F, et al. (2022). Maternal perlluorinated compound exposure and risk of early
pregnancy loss: A nested case-control study [Letter], Biomedical and Environmental Sciences 35: 174-179.
http://dx.doi.org/10.3967/bes2Q22.Q26
Midasch, O; Drexler, H; Hart, N; Beckmann, MW; Angerer, J. (2007). Transplacental exposure of neonates to
perfhiorooctanesulfonate and perlluorooctanoate: a pilot study. International Archives of Occupational and
Environmental Health 80: 643-648. http://dx.doi.org/lQ.lQQ7/sQQ42Q-QQ6-Q165-9
Minatoya, M; Itoh, S; Miyashita, C; Araki, A; Sasaki, S; Miura, R, et al. (2017). Association of prenatal exposure to
pertluoroalkyl substances with cord blood adipokines and birth size: The Hokkaido Study on environment and
children's health. Environmental Research 156: 175-182.
Mitro, SD; Sagiv, SK; Fleisch, AF; Jaacks, LM; Williams, PL; Rifas-Shiman, SL, et al. (2020). Pregnancy per- and
polyfluoroalkyl substance concentrations and postpartum health in project viva: A prospective cohort. Journal of
Clinical Endocrinology and Metabolism 105: e3415-e3426.
Mobacke, I; Lind, L; Dunder, L; Salihovic, S; Lind, PM. (2018). Circulating levels of perlluoroalkyl substances and left
ventricular geometry of the heart in the elderly. Environment International 115: 295-300.
Mogensen, UB; Grandjean, P; Heilmann, C; Nielsen, F; Weihe, P; Budtz-Jergensen, E. (2015a). Structural equation modeling of
immunotoxicity associated with exposure to pertluorinated alkylates. Environmental Health: A Global Access Science
Source 14: 47.
Mogensen, UB; Grandjean, P; Nielsen, F; Weihe, P; Budtz-Jergensen, E. (2015b). Breastfeeding as an Exposure Pathway for
Pertluorinated Alkylates. Environmental Science and Technology 49: 10466-10473.
Mondal, D; Weldon, RH; Armstrong, BG; Gibson, LJ; Lopez-Espinosa, MJ; Shin, HM; Fletcher, T. (2014). Breastfeeding: a
potential excretion route for mothers and implications for infant exposure to pertluoroalkyl acids. Environmental
Health Perspectives 122: 187-192.
Monge Brenes, AL; Curtzwiler, G; Dixon, P; Harrata, K; Talbert, J; Vorst, K. (2019). PFOA and PFOS levels in microwave
paper packaging between 2005 and 2018. Food Additives and Contaminants: PartB: Surveillance 12: 191-198.
http://dx.doi.org/10.lQ8Q/1939321Q.2Q19.1592238
R-23
-------
APRIL 2024
Monroy, R; Morrison, K; Teo, K; Atkinson, S; Kubwabo, C; Stewart, B; Foster, WG. (2008). Serum levels of perfluoroalkyl
compounds in human maternal and umbilical cord blood samples. Environmental Research 108: 56-62.
http://dx.doi.Org/10.1016/i.envies.2008.06.001
Mora, AM; Fleisch, AF; Rifas-Shiman, SL; Woo Baidal, JA; Pardo, L; Webster, TF, et al. (2018). Early life exposure to per- and
polyfluoroalkyl substances and mid-childhood lipid and alanine aminotransferase levels. Environment International
111: 1-13.
Mora, AM; Oken, E; Rifas-Shiman, SL; Webster, TF; Gillman, MW; Calafat, AM, et al. (2017). Prenatal exposure to
perfluoroalkyl substances and adiposity in early and mid-childhood. Environmental Flealth Perspectives 125: 467-473.
MPCA. (2008). PFCs in Minnesota's Ambient Environment: 2008 Progress Report.
https://www.pca.state.mn.us/sites/default/files/c-pfcl-02.pdf
Mshaty, A; Flaijima, A; Takatsuru, Y; Ninomiya, A; Yajima, H; Kokubo, M, et al. (2020). Neurotoxic effects of lactational
exposure to perfluorooctane sulfonate on learning and memory in adult male mouse. Food and Chemical Toxicology
145: 111710.
Nakayama, S; Strynar, MJ; Flelfant, L; Egeghy, P; Ye, X; Lindstrom, AB. (2007). Perfluorinated compounds in the Cape Fear
Drainage Basin in North Carolina. Environmental Science and Technology 41: 5271-5276.
http://dx.doi.org/10.1021/es070792v
Narizzano, AM; Lent, EM; Hanson, JM; East, AG; Bohannon, ME; Quinn, MJ, Jr. (2022). Reproductive and developmental
toxicity of perfluorooctane sulfonate (PFOS) in the white-footed mouse (Peromyscus leucopus). Reproductive
Toxicology 113: 120-127. http : //dx. doi. org/1.0.1.01.6/j. repro tox.2022.08.01.1.
NASEM. (2021). Review of U.S. EPA's ORD staff handbook for developing IRIS assessments: 2020 version. Washington, DC:
National Academies Press, http://dx.doi.org/1.0.1.7226/26289
NCHS. (2019). Health, United States - Data Finder. 2019: Table 23. Hyattsville, MD.
fattps: //www.cdc. go v/nchs/hus/contents201.9 .htm
Needham, LL; Grandjean, P; Heinzow, B; Jergensen, PJ; Nielsen, F; Patterson, DG, et al. (2011). Partition of environmental
chemicals between maternal and fetal blood and tissues. Environmental Science and Technology 45: 1121-1126.
http://dx.doi.org/10.1021/esl019614
Negri, E; Metruccio, F; Guercio, V; Tosti, L; Benfenati, E; Bonzi, R, et al. (2017). Exposure to PFOA and PFOS and fetal
growth: a critical merging of toxicological and epidemiological data [Review], Critical Reviews in Toxicology 47: 482-
508. http://dx.doi.org/10.1.080/1.0408444.201.6.1.271.972
Nelson, JW; Hatch, EE; Webster, TF. (2010). Exposure to Polyfluoroalkyl Chemicals and Cholesterol, Body Weight, and Insulin
Resistance in the General US Population. Environmental Health Perspectives 118:197-202.
http://dx.doi.org/10.1289/ehp.09011.65
Ngo, HT; Hetland, RB; Sabaredzovic, A; Haug, LS; Steffensen, IL. (2014). In utero exposure to perfluorooctanoate (PFOA) or
perfluorooctane sulfonate (PFOS) did not increase body weight or intestinal tumorigenesis in multiple intestinal
neoplasia (Min/+) mice. Environmental Research 132: 251-263. http://dx.doi.Org/10.1016/i.envres.2014.03.033
Nian, M; Huo, X; Zhang, J; Mao, Y; Jin, F; Shi, Y; Zhang, J. (2022). Association of emerging and legacy per- and
polyfluoroalkyl substances with unexplained recurrent spontaneous abortion. Ecotoxicology and Environmental Safety
239: 113691. http://dx.doi.Org/10.1016/i.ecoenv.2022.l 1.3691.
Nian, M; Li, QQ; Bloom, M; Qian, ZM; Syberg, KM; Vaughn, MG, et al. (2019). Liver function biomarkers disorder is
associated with exposure to perfluoroalkyl acids in adults: Isomers of C8 Health Project in China. Environmental
Research 172: 81-88.
Nielsen, C; Andersson Hall, U; Lindh, C; Ekstrom, U; Xu, Y; Li, Y, et al. (2020). Pregnancy-induced changes in serum
concentrations of perfluoroalkyl substances and the influence of kidney function. Environmental Health: A Global
Access Science Source 19: 80.
Nilsson, S; Smurthwaite, K; Aylward, LL; Kay, M; Toms, LM; King, L, et al. (2022). Associations between serum perfluoroalkyl
acid (PFAA) concentrations and health related biomarkers in firefighters. Environmental Research 215: 114370.
http://dx.doi.Org/10.1016/i.envres.2022.l 1.4370
Niu, J; Liang, H; Tian, Y; Yuan, W; Xiao, H; Hu, H, et al. (2019). Prenatal plasma concentrations of Perfluoroalkyl and
polyfluoroalkyl substances and neuropsychological development in children at four years of age. Environmental
Health: A Global Access Science Source 18: 53.
NJDWQI. (2018). Health-based maximum contaminant level support document: Perfluorooctane sulfonate (PFOS): Appendix A.
https://www.state.nj.us/dep/watersupply/pdf/pfos-recommendation-appendix-a.pdf
NLM. (2022). ChemlDplus: Perfluorooctane sulfonic acid. Available online at https://chem.nlm.nih.gov/chemidplus/name/pfos
NOAA. (2017). NOAA's National Status and Trends Data. Silver Spring, MD: U.S. Department of Commerce, National Oceanic
and Atmospheric Administration, National Centers for Coastal Ocean Science. Retrieved from
https://products.coastalscience.noaa.gov/nsandt_data/data.aspx
NTP. (2019). NTP technical report on the toxicity studies of perfluoroalkyl sulfonates (perfluorobutane sulfonic acid,
perfluorohexane sulfonate potassium salt, and perfluorooctane sulfonic acid) administered by gavage to Sprague
Dawley (Hsd:Sprague Dawley SD) rats. (Toxicity Report 96). Research Triangle Park, NC.
Ode, A; Kallen, K; Gustafsson, P; Rylander, L; Jonsson, BA; Olofsson, P, et al. (2014). Fetal exposure to perfluorinated
compounds and attention deficit hyperactivity disorder in childhood. PLoS ONE 9: e95891.
R-24
-------
APRIL 2024
OECD. (2001). Test no. 416: Two-generation reproduction toxicity. In OECD guidelines for the testing of chemicals, Section 4:
Health effects. Paris, France, http://dx.doi.org/10.1787/9789264070868-en
OECD. (2018). Toward a new comprehensive global database of per- and polyfluoroalkyl substances (PFASs): Summary report
on updating the OECD 2007 list of per- and polyfluoroalkyl substances (PFASs). (ENV/JM/MONO(2018)7).
http://www.oecd.org/chemicalsafety/poi1al-perfluoriiiated-chemicals/
Okada, E; Sasaki, S; Kashino, I; Matsuura, H; Miyashita, C; Kobayashi, S, et al. (2014). Prenatal exposure to perfluoroalkyl acids
and allergic diseases in early childhood. Environment International 65: 127-134.
http: //dx. doi. org/1.0.1.01.6/j. eiivint .201.4,01.. 007
Okada, E; Sasaki, S; Saijo, Y; Washino, N; Miyashita, C; Kobayashi, S, et al. (2012). Prenatal exposure to perfluorinated
chemicals and relationship with allergies and infectious diseases in infants. Environmental Research 112: 118-125.
http://dx.doi.Org/10.1016/i.envres.2011.10.003
Olsen, G, .W., B.,urlew, M.,.M.; Burris, J, .M.; Mandel, J, .H. (2001). A Longitudinal Analysis of Serum Perfluorooctane
Sulfonate (PFOS) and Perfluorooctanoate (PFOA) Levels in Relation to Lipid and Hepatic Clinical Chemistry Test
Results from Male Employee Participants of the 1994/95, 1997 and 2000 Fluorochemical Medical Surveillance
Program. Final Report. (Epidemiology, 220-3W-05). St. Paul, MN: 3M Company.
https://www.ag.state.mn.us/OflIce/Cases/3M/docs/PTX/PTX 1799.pdf
Olsen, GW; Burris, JM; Burlew, MM; Mandel, JH. (2003a). Epidemiologic assessment of worker serum
perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA) concentrations and medical surveillance
examinations. Journal of Occupational and Environmental Medicine 45: 260-270.
http://dx.doi.org/10.1097/01.iom.0000Q52958.59271.10
Olsen, GW; Burris, JM; Ehresman, DJ; Froehlich, JW; Seacat, AM; Butenhoff, JL; Zobel, LR. (2007). Half-life of serum
elimination of perfluorooctanesulfonate,perfluorohexanesulfonate, and perfluorooctanoate in retired fluorochemical
production workers. Environmental Health Perspectives 115: 1298-1305.
https://heronet.epa.gov/heronet/iiidex.cfm/reference/dowiiload/reference_id/1429952
Olsen, GW; Ehresman, DJ; Buehrer, BD; Gibson, BA; Butenhoff, JL; Zobel, LR. (2012a). Longitudinal assessment of lipid and
hepatic clinical parameters in workers involved with the demolition of perfluoroalkyl manufacturing facilities. Journal
of Occupational and Environmental Medicine 54: 974-983.
Olsen, GW; Hansen, KJ; Stevenson, LA; Burris, JM; Mandel, JH. (2003b). Human donor liver and serum concentrations of
perfluorooctanesulfonate and other perfluorochemicals. Environmental Science and Technology 37: 888-891.
http: // dx. doi. org/1.0.1.021. /es020955c
Olsen, GW; Lange, CC; Ellefson, ME; Mair, DC; Church, TR; Goldberg, CL, et al. (2012b). Temporal trends of perfluoroalkyl
concentrations in American Red Cross adult blood donors, 2000-2010. Environmental Science & Technology 46: 6330-
6338. http://dx.doi.org/1.0.1.021./es300604p
Omoike, OE; Pack, RP; Mamudu, HM; Liu, Y; Strasser, S; Zheng, S, et al. (2020). Association between per and polyfluoroalkyl
substances and markers of inflammation and oxidative stress. Environmental Research 196: 110361.
http: //dx. doi. org/1.0.1.01.6/j. envres.2020.1.1.0361.
Omoike, OE; Pack, RP; Mamudu, HM; Liu, Y; Wang, L. (2021). A cross-sectional study of the association between
perfluorinated chemical exposure and cancers related to deregulation of estrogen receptors. Environmental Research
196: 110329. http://dx.doi.Org/10.1016/i.envres.2020.1.1.0329
Osorio-Yanez, C; Sanchez-Guerra, M; Cardenas, A; Lin, PID; Hauser, R; Gold, DR, et al. (2021). Per- and polyfluoroalkyl
substances and calcifications of the coronary and aortic arteries in adults with prediabetes: Results from the diabetes
prevention program outcomes study. Environment International 151: 106446.
http: //dx. doi. org/1.0.1.01.6/j. envint .2021.. 1.06446
Ou, Y; Zeng, X; Lin, S; Bloom, MS; Han, F; Xiao, X, et al. (2021). Gestational exposure to perfluoroalkyl substances and
congenital heart defects: A nested case-control pilot study. Environment International 154: 106567.
http://dx.doi.org/10.1.01.6/j envint.2021.. 1.06567
Oulhote, Y; Coull, B; Bind, MA; Debes, F; Nielsen, F; Tamayo, I, et al. (2019). Joint and independent neurotoxic effects of early
life exposures to a chemical mixture: A multi-pollutant approach combining ensemble learning and g-computation.
Environmental Epidemiology 3: e063.
Oulhote, Y; Steuerwald, U; Debes, F; Weihe, P; Grandjean, P. (2016). Behavioral difficulties in 7-year old children in relation to
developmental exposure to perfluorinated alkyl substances [Review], Environment International 97: 237-245.
Padula, AM; Ning, X; Bakre, S; Barrett, ES; Bastain, T; Bennett, DH, et al. (2023). Birth outcomes in relation to prenatal
exposure to per-and polyfluoroalkyl substances and stress in the environmental influences on child health outcomes
(echo) program [Supplemental Data], Environmental Health Perspectives 131: (037006)037001-037011.
http://dx.doi.org/10.1289/EHP10723
Pan, K; Xu, J; Long, X; Yang, L; Huang, Z; Yu, J. (2023). The relationship between perfluoroalkyl substances and hypertension:
A systematic review and meta-analysis [Review], Environmental Research 232: 116362.
http: //dx. doi. org/1.0.1.01.6/j. envres.202 3.1.1.6362
Pan, Y; Cui, Q; Wang, J; Sheng, N; Jing, J; Yao, B; Dai, J. (2019). Profiles of Emerging and Legacy Per-/Polyfluoroalkyl
Substances in Matched Serum and Semen Samples: New Implications for Human Semen Quality. Environmental
Health Perspectives 127: 127005.
R-25
-------
APRIL 2024
Pan, Y; Zhu, Y; Zheng, T; Cui, Q; Buka, SL; Zhang, B, et al. (2017). Novel Chlorinated Polyfluorinated Ether Sulfonates and
Legacy Per-/Polyfluoroalkyl Substances: Placental Transfer and Relationship with Serum Albumin and Glomerular
Filtration Rate. Environmental Science and Technology 51: 634-644.
Papadopoulou, E; Nicolescu, A; Haug, LS; Husey, T; Deleanu, C; Dirven, H; Lindeman, B. (2022). Lipoprotein profiles
associated with exposure to poly- and perfluoroalkyl substances (PFASs) in the EuroMix human biomonitoring study.
Environmental Pollution 308: 119664. http://dx.doi.org/1.0.1.01.6/j.en.vpol.2022.1.1.9664
Papadopoulou, E; Poothong, S; Koekkoek, J; Lucattini, L; Padilla-Sanchez, JA; Haugen, M, et al. (2017). Estimating human
exposure to perfluoroalkyl acids via solid food and drinks: Implementation and comparison of different dietary
assessment methods. Environmental Research 158: 269-276. fattp://dx.doi.org/1.0.1.01.6/i.envres.201.7.06,01.1.
Papadopoulou, E; Stratakis, N; Basagana, X; Brantsaster, AL; Casas, M; Fossati, S, et al. (2021). Prenatal and postnatal exposure
to PFAS and cardiometabolic factors and inflammation status in children from six European cohorts. Environment
International 157: 106853. fattp://dx.doi.org/1.0.1.01.6/i.envint.2021.1.06853
Patel, JC; Mehta, BC. (1999). Tetanus: Study of 8,697 cases. Indian Journal of Medical Sciences 53: 393-401.
Peden-Adams, MM; Keller, JM; Eudaly, JG; Berger, J; Gilkeson, GS; Keil, DE. (2008). Suppression of humoral immunity in
mice following exposure to perfluorooctane sulfonate. Toxicological Sciences 104: 144-154.
fattp: //dx. doi. org/1.0.1.093/tox sci/kfnO 5 9
Penland, TN; Cope, WG; Kwak, TJ; Strynar, MJ; Grieshaber, CA; Heise, RJ; Sessions, FW. (2020). Trophodynamics of Per- and
Polyfluoroalkyl Substances in the Food Web of a Large Atlantic Slope River. Environmental Science and Technology
54: 6800-6811. http://dx.doi.org/10.1021/acs.est.9b05007
Pennati, G; Corno, C; Costantino, ML; Bellotti, M. (2003). Umbilical flow distribution to the liver and the ductus venosus in
human fetuses during gestation: an anatomy-based mathematical modeling. Medical Engineering & Physics 25: 229-
238. http://dx.doi.org/10.1016/sl350-4533(02)001.92-3
Pereiro, N; Moyano, R; Blanco, A; Lafuente, A. (2014). Regulation of corticosterone secretion is modified by PFOS exposure at
different levels of the hypothalamic-pituitary-adrenal axis in adult male rats. Toxicology Letters 230: 252-262.
Perez, F; Nadal, M; Navarro-Ortega, A; Fabrega, F; Domingo, JL; Barcelo, D; Farre, M. (2013). Accumulation of perfluoroalkyl
substances in human tissues. Environment International 59: 354-362.
Petersen, MS; Hailing, J; Jergensen, N; Nielsen, F; Grandjean, P; Jensen, TK; Weihe, P. (2018). Reproductive function in a
population of young Faroese men with elevated exposure to polychlorinated biphenyls (pcbs) and perfluorinated
alkylate substances (pfas). International Journal of Environmental Research and Public Health 15: n/a.
Peterson, AK; Eckel, SP; Habre, R; Yang, T; Faham, D; Amin, M, et al. (2022). Detected prenatal perfluorooctanoic acid
(PFOA) exposure is associated with decreased fetal head biometric parameters in participants experiencing higher
perceived stress during pregnancy in the MADRES cohort. 9. http://dx.doi.Org/10.1016/j.envadv.2022.100286
Petro, EM; D'Hollander, W; Covaci, A; Bervoets, L; Fransen, E; De Neubourg, D, et al. (2014). Perfluoroalkyl acid
contamination of follicular fluid and its consequence for in vitro oocyte developmental competence. Science of the
Total Environment 496: 282-288.
Pilkerton, CS; Hobbs, GR; Lilly, C; Knox, SS. (2018). Rubella immunity and serum perfluoroalkyl substances: Sex and analytic
strategy. PLoS ONE 13: e0203330.
Pitter, G; Zare Jeddi, M; Barbieri, G; Gion, M; Fabricio, ASC; Dapra, F, et al. (2020). Perfluoroalkyl substances are associated
with elevated blood pressure and hypertension in highly exposed young adults. Environmental Health: A Global Access
Science Source 19: 102. http://dx.doi.Org/10.l 1.86/s1.2940~020~00656~0
Poothong, S; Padilla-Sanchez, JA; Papadopoulou, E; Giovanoulis, G; Thomsen, C; Haug, LS. (2019). Hand Wipes: A Useful
Tool for Assessing Human Exposure to Poly- and Perfluoroalkyl Substances (PFASs) through Hand-to-Mouth and
Dermal Contacts. Environmental Science and Technology 53: 1985-1993. http://dx.doi.org/10.1.02 l/acs.est.8M)S303
Poothong, S; Papadopoulou, E; Padilla-Sanchez, JA; Thomsen, C; Haug, LS. (2020). Multiple pathways of human exposure to
poly- and perfluoroalkyl substances (PFASs): From external exposure to human blood. Environment International 134:
105244. littp://dx.doi.org/1.0.1.01.6/i.envint.201.9.1.05244
Poothong, S; Thomsen, C; Padilla-Sanchez, JA; Papadopoulou, E; Haug, LS. (2017). Distribution of novel and well-known poly-
and perfluoroalkyl substances (PFASs) in human serum, plasma, and whole blood. Environmental Science and
Technology 51: 13388-13396.
Porpora, MG; Lucchini, R; Abballe, A; Ingelido, AM; Valentini, S; Fuggetta, E, et al. (2013). Placental transfer of persistent
organic pollutants: a preliminary study on mother-newborn pairs. International Journal of Environmental Research and
Public Health 10: 699-711.
Porter, AK; Kleinschmidt, SE; Andres, KL; Reusch, CN; Krisko, RM; Taiwo, OA, et al. (2022). Antibody response to COVID-
19 vaccines among workers with a wide range of exposure to per- and polyfluoroalkyl substances. Environment
International 169: 107537. http://dx.doi.Org/10.1016/i.enviiit.2022.107537
Powley, CR; George, SW; Russell, MH; Hoke, RA; Buck, RC. (2008). Polyfluorinated chemicals in a spatially and temporally
integrated food web in the Western Arctic. Chemosphere 70: 664-672.
http://dx.doi.Org/10.1016/i.chemosphere.2007.06.067
Predieri, B; Iughetti, L; Guerranti, C; Bruzzi, P; Perra, G; Focardi, SE. (2015). High Levels of Perfluorooctane Sulfonate in
Children at the Onset of Diabetes. International Journal of Endocrinology 2015: 234358.
R-26
-------
APRIL 2024
Preston, EV; Rifas-Shiman, SL; Hivert, MF; Zota, AR; Sagiv, SK; Calafat, AM, et al. (2020). Associations of per- and
polyfluoroalkyl substances (PFAS) with glucose tolerance during pregnancy in project viva. Journal of Clinical
Endocrinology and Metabolism 105: E2864-E2876.
Preston, EV; Webster, TF; Oken, E; Claus Elenn, B; Mcclean, MD; Rifas-Shiman, SL, et al. (2018). Maternal plasma per- and
polyfluoroalkyl substance concentrations in early pregnancy and maternal and neonatal thyroid function in a
prospective birth cohort: Project Viva (USA). Environmental Elealth Perspectives 126: 027013.
Pritchard, JA. (1965). Changes in the blood volume during pregnancy and delivery [Review], Anesthesiology 26: 393-399.
http://dx.doi.Org/1.0.1.097/00000542-1.96507000-00004
Pumarega, J; Gasull, M; Koponen, J; Campi, L; Rantakokko, P; Elenriquez-Hernandez, LA, et al. (2023). Prepandemic personal
concentrations of per- and polyfluoroalkyl substances (PFAS) and other pollutants: Specific and combined effects on
the incidence of COVID-19 disease and SARS-CoV-2 infection. Environmental Research 237: 116965.
http: //dx. doi. org/1.0.1.01.6/j .envies.202 3.116965
Purdue, MP; Rhee, J; Denic-Roberts, H; McGlynn, KA; Byrne, C; Sampson, J, et al. (2023). A Nested Case-Control Study of
Serum Per- and Polyfluoroalkyl Substances and Testicular Germ Cell Tumors among U.S. Air Force Servicemen.
Environmental Elealth Perspectives 131: 77007. http: //dx. doi. org/1.0.1.2 89/EHP1.2603
Puttige Ramesh, N; Arora, M; Braun, JM. (2019). Cross-sectional study of the association between serum perfluorinated alkyl
acid concentrations and dental caries among US adolescents (NHANES 1999-2012). British Medical Journal Open 9:
e024189.
Qazi, MR; Xia, Z; Bogdanska, J; Chang, SC; Ehresman, DJ; Butenhoff, JL, et al. (2009). The atrophy and changes in the cellular
compositions of the thymus and spleen observed in mice subjected to short-term exposure to perfluorooctanesulfonate
are high-dose phenomena mediated in part by peroxisome proliferator-activated receptor-alpha (PPARalpha).
Toxicology 260: 68-76. http://dx.doi.Org/10.1016/i.tox.2009.03.009
Qin, XD; Qian, Z; Vaughn, MG; Eluang, J; Ward, P; Zeng, XW, et al. (2016). Positive associations of serum perfluoroalkyl
substances with uric acid and hyperuricemia in children from Taiwan. Environmental Pollution 212: 519-524.
Qin, XD; Qian, ZM; Dharmage, SC; Perret, J; Geiger, SD; Rigdon, SE, et al. (2017). Association of perfluoroalkyl substances
exposure with impaired lung function in children. Environmental Research 155: 15-21.
Qiu, L; Qian, Y; Liu, Z; Wang, C; Qu, J; Wang, X; Wang, S. (2016). Perfluorooctane sulfonate (PFOS) disrupts blood-testis
barrier by down-regulating junction proteins via p38 MAPK/ATF2/MMP9 signaling pathway. Toxicology 373: 1-12.
Qiu, L; Wang, H; Dong, T; Eluang, J; Li, T; Ren, H, et al. (2020). Perfluorooctane sulfonate (PFOS) disrupts testosterone
biosynthesis via CREB/CRTC2/StAR signaling pathway in Leydig cells. Toxicology 449: 152663.
http: //dx. doi. org/1.0.1.01.6/j. tox.2020.1.52663
Qiu, L; Zhang, X; Zhang, X; Zhang, Y; Gu, J; Chen, M, et al. (2013). Sertoli cell is a potential target for perfluorooctane
sulfonate-induced reproductive dysfunction in male mice. Toxicological Sciences 135: 229-240.
Qu, A; Cao, T; Li, Z; Wang, W; Liu, R; Wang, X, et al. (2021). The association between maternal perfluoroalkyl substances
exposure and early attention deficit hyperactivity disorder in children: a systematic review and meta-analysis.
Environmental Science and Pollution Research 28: 67066-67081. http://dx.doi.org/10.1007/sl 1.356-021.-1.51.36-2
Qu, J; Zhao, Y; Zhang, L; Elu, S; Liao, K; Zhao, M, et al. (2022). Evaluated serum perfluoroalkyl acids and their relationships
with the incidence of rheumatoid arthritis in the general population in Elangzhou, China. Environmental Pollution 307:
119505. http://dx.doi.Org/10.101.6/i.eiiYpol.2022.1.1.9505
Qu, JH; Lu, CC; Xu, C; Chen, G; Qiu, LL; Jiang, JK, et al. (2016). Perfluorooctane sulfonate-induced testicular toxicity and
differential testicular expression of estrogen receptor in male mice. Environmental Toxicology and Pharmacology 45:
150-157.
Quaak, I; de Cock, M; de Boer, M; Lamoree, M; Leonards, P; van de Bor, M. (2016). Prenatal Exposure to Perfluoroalkyl
Substances and Behavioral Development in Children. International Journal of Environmental Research and Public
Health 13.
Radke, EG; Glenn, B; Galizia, A; Persad, A; Nachman, R; Bateson, T, et al. (2019). Development of outcome-specific criteria for
study evaluation in systematic reviews of epidemiology studies. Environment International 130: 104884.
http: //dx. doi. org/1.0.1.01.6/j. eiivint .201.9.05.078
Rahman, ML; Zhang, C; Smarr, MM; Lee, S; Honda, M; Kannan, K, et al. (2019). Persistent organic pollutants and gestational
diabetes: A multi-center prospective cohort study of healthy US women. Environment International 124: 249-258.
Rantakokko, P; Mannisto, V; Airaksinen, R; Koponen, J; Viluksela, M; Kiviranta, H; Pihlajamaki, J. (2015). Persistent organic
pollutants and non-alcoholic fatty liver disease in morbidly obese patients: A cohort study. Environmental Health: A
Global Access Science Source 14: 79.
Reardon, AJF; Khodayari Moez, E; Dinu, I; Goruk, S; Field, CJ; Kinniburgh, DW, et al. (2019). Longitudinal analysis reveals
early-pregnancy associations between perfluoroalkyl sulfonates and thyroid hormone status in a Canadian prospective
birth cohort. Environment International 129: 389-399.
Remucal, CK. (2019). Spatial and temporal variability of perfluoroalkyl substances in the Laurentian Great Lakes [Review].
Environmental Science: Processes & Impacts 21: 1816-1834. http://dx.doi.org/1.0.1.039/c9em00265k
Ren, Y; Jin, L; Yang, F; Liang, H; Zhang, Z; Du, J, et al. (2020). Concentrations of perfluoroalkyl and polyfluoroalkyl substances
and blood glucose in pregnant women. Environmental Health: A Global Access Science Source 19: 88.
Reyes, L; Manalich, R. (2005). Long-term consequences of low birth weight [Review], Kidney International Supplement 68:
S107-S111. http://dx.doi.org/10.1.1.1.1/i 1523-1755.2005.09718.x
R-27
-------
APRIL 2024
Rhee, J; Chang, VC; Cheng, I; Calafat, AM; Botelho, JC; Shearer, JJ, et al. (2023). Serum concentrations of per- and
polyfluoroalkyl substances and risk of renal cell carcinoma in the Multiethnic Cohort Study. Environment International
180: 108197. http://dx.doi.Org/10.1016/i.enviiit.2023.1.081.97
Robledo, CA; Yeung, E; Mendola, P; Sundaram, R; Maisog, J; Sweeney, AM, et al. (2015). Preconception maternal and paternal
exposure to persistent organic pollutants and birth size: the LIFE study. Environmental Health Perspectives 123: 88-94.
http://dx.doi.org/10.1289/ehp. 1.30801.6
Rokoff, LB; Rifas-Shiman, SL; Coull, BA; Cardenas, A; Calafat, AM; Ye, X, et al. (2018). Cumulative exposure to
environmental pollutants during early pregnancy and reduced fetal growth: the Project Viva cohort. Environmental
Health: A Global Access Science Source 17: 19.
Romano, ME; Heggeseth, BC; Gallagher, LG; Botelho, JC; Calafat, AM; Gilbert-Diamond, D; Karagas, MR. (2022). Gestational
per- and polyfluoroalkyl substances exposure and infant body mass index trajectory in the New Hampshire Birth
Cohort Study. Environmental Research 215: 114418. littp://dx.doi.org/1.0.1.01.6/j.envres.2022.1.1.441.8
Romano, ME; Xu, Y; Calafat, AM; Yolton, K; Chen, A; Webster, GM, et al. (2016). Maternal serum perfluoroalkyl substances
during pregnancy and duration of breastfeeding. Environmental Research 149: 239-246.
Rosen, EM; Kotlarz, N; Knappe, DRU; Lea, CS; Collier, DN; Richardson, DB; Hoppin, JA. (2022). Drinking water-associated
PFAS and fluoroethers and lipid outcomes in the GenX exposure study. Environmental Health Perspectives 130:
97002. http://dx.doi.org/10.1289/EHP11033
Rosner, B. (2015). Fundamentals of biostatistics (8th ed.). Boston, MA: Brooks/Cole, Cengage Learning.
Rotander, A; Toms, LM; Aylward, L; Kay, M; Mueller, JF. (2015). Elevated levels of PFOS and PFHxS in firefighters exposed
to aqueous film forming foam (AFFF). Environment International 82: 28-34.
Ruark, CD; Song, G; Yoon, M; Verner, MA; Andersen, ME; Clewell, HJ; Longnecker, MP. (2017). Quantitative bias analysis for
epidemiological associations of perfluoroalkyl substance serum concentrations and early onset of menopause.
Environment International 99: 245-254. http://dx.doi.Org/10.1016/i.envint.2016.l 1.030
Ruffle, B; Vedagiri, U; Bogdan, D; Maier, M; Schwach, C; Murphy-Hagan, C. (2020). Perfluoroalkyl Substances in U.S. market
basket fish and shellfish. Environmental Research 190: 9932-9932. http://dx.doi.Org/10.1016/i.envres.2020.109932
Rusch, GM. (1979). An Acute Inhalation Toxicity Study of T-2307 CoC in the Rat. (3M-EPA-0029829). Rusch, GM.
Rylander, L; Lindh, CH; Hansson, S. R.; Broberg, K; Kallen, K. (2020). Per- and polyfluoroalkyl substances in early pregnancy
and risk for preeclampsia: a case-control study in Southern Sweden. Toxics 8: 43.
Sagiv, SK; Rifas-Shiman, SL; Fleisch, AF; Webster, TF; Calafat, AM; Ye, X, et al. (2018). Early Pregnancy Perfluoroalkyl
Substance Plasma Concentrations and Birth Outcomes in Project Viva: Confounded by Pregnancy Hemodynamics?
American Journal of Epidemiology 187: 793-802.
Salgado-Freiria, R; Lopez-Doval, S; Lafuente, A. (2018). Perfluorooctane sulfonate (PFOS) can alter the hypothalamic-pituitary-
adrenal (HPA) axis activity by modifying CRF1 and glucocorticoid receptors. Toxicology Letters 295: 1-9.
Salgado, R; Lopez-Doval, S; Pereiro, N; Lafuente, A. (2016). Perfluorooctane sulfonate (PFOS) exposure could modify the
dopaminergic system in several limbic brain regions. Toxicology Letters 240: 226-235.
Salgado, R; Pereiro, N; Lopez-Doval, S; Lafuente, A. (2015). Initial study on the possible mechanisms involved in the effects of
high doses of perfluorooctane sulfonate (PFOS) on prolactin secretion. Food and Chemical Toxicology 83: 10-16.
Salihovic, S; Stubleski, J; Karrman, A; Larsson, A; Fall, T; Lind, L; Lind, PM. (2018). Changes in markers of liver function in
relation to changes in perfluoroalkyl substances - A longitudinal study. Environment International 117: 196-203.
Salvalaglio, M; Muscionico, I; Cavallotti, C. (2010). Determination of energies and sites of binding of PFOA and PFOS to
human serum albumin. Journal of Physical Chemistry B 114: 14860-14874. http://dx.doi.org/10.1.021/ipl 06584b
Sanchez Garcia, D; Sjodin, M; Hellstrandh, M; Norinder, U; Nikiforova, V; Lindberg, J, et al. (2018). Cellular accumulation and
lipid binding of perfluorinated alkylated substances (PFASs) - A comparison with lysosomotropic drugs. Chemico-
Biological Interactions 281: 1-10.
Sasaki, K; Harada, K; Saito, N; Tsutsui, T; Nakanishi, S; Tsuzuki, H; Koizumi, A. (2003). Impact of airborne perfluorooctane
sulfonate on the human body burden and the ecological system. Bulletin of Environmental Contamination and
Toxicology 71: 408-413. http://dx.doi.org/10.1007/s00128-003-Q179-x
Schaider, LA; Balan, SA; Blum, A; Andrews, DQ; Strynar, MJ; Dickinson, ME, et al. (2017). Fluorinated compounds in US fast
food packaging. Environmental Science & Technology Letters 4: 105-111.
littp: //dx. doi. org/1.0.1.021. /ac s. estlet t. 6b0043 5
Schecter, A; Colacino, J; Haffner, D; Patel, K; Opel, M; Papke, O; Birnbaum, L. (2010). Perfluorinated compounds,
poly chlorinated biphenyls, and organochlorine pesticide contamination in composite food samples from Dallas, Texas,
USA. Environmental Health Perspectives 118: 796-802. http://dx.doi.org/10.1289/ehp.09Q1347
Schillemans, T; Donat-Vargas, C; Lindh, CH; de Faire, U; Wolk, A; Leander, K; Akesson, A. (2022). Per- and Polyfluoroalkyl
Substances and Risk of Myocardial Infarction and Stroke: A Nested Case-Control Study in Sweden. Environmental
Health Perspectives 130: 37007. http://dx.doi.org/10.1289/EHP9791
Schreder, E; Dickman, J. (2018). Take Out Toxics: PFAS Chemicals in Food Packaging. Schreder, Erika Schreder; Dickman,
Jennifer. https://48h57c2131ua3c3fmqlne58b-wpengine.netdna-ssl.eom/wp-content/uploads/2019/05/Take-Out-Toxics-
Full-Report.pdf
Schumann, G; Bonora, R; Ceriotti, F; Ferard, G; Ferrero, CA; Franck, PFH, et al. (2002). IFCC Primary Reference Procedures
for the Measurement of Catalytic Activity Concentrations of Enzymes at 37°C. Part 4. Reference Procedure for the
R-28
-------
APRIL 2024
Measurement of Catalytic Concentration of Alanine Aminotransferase. Clinical Chemistry and Laboratory Medicine
40: 718-724. https://doi.org/10.1515/CCLM.20Q2.124
Scinicariello, F; Buser, MC; Abadin, HG; Attanasio, R. (2020a). Perfluoroalkyl substances and anthropomorphic measures in
children (ages 3-11 years), NHANES 2013-2014. Environmental Research 186: 109518.
Scinicariello, F; Buser, MC; Balluz, L; Gehle, K; Murray, HE; Abadin, HG; Attanasio, R. (2020b). Perfluoroalkyl acids,
hyperuricemia and gout in adults: Analyses of NHANES 2009-2014. Chemosphere 259: 127446.
Scorer, CG. (1964). The descent of the testis. Archives of Disease in Childhood 39: 605-609.
Seacat, AM; Thomford, PJ; Hansen, KJ; Clemen, LA; Eldridge, SR; Elcombe, CR; Butenhoff, JL. (2003). Sub-chronic dietary
toxicity of potassium perfluorooctanesulfonate in rats [Erratum], Toxicology 183: 117-131.
Seacat, AM; Thomford, PJ; Hansen, KJ; Olsen, GW; Case, MT; Butenhoff, JL. (2002). Subchronic toxicity studies on
perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicological Sciences 68: 249-264.
Selgrade, MK. (2007). Immunotoxicity: The risk is real [Review], Toxicological Sciences 100: 328-332.
http://dx.doi.org/lQ.lQ93/toxsci/kfm244
Seo, SH; Son, MH; Choi, SD; Lee, DH; Chang, YS. (2018). Influence of exposure to perfluoroalkyl substances (PFASs) on the
Korean general population: 10-year trend and health effects. Environment International 113: 149-161.
Sevelsted, A; Giirdeniz, G; Rago, D; Pedersen, CT; Lasky-Su, JA; Checa, A, et al. (2022). Effect of perfluoroalkyl exposure in
pregnancy and infancy on intrauterine and childhood growth and anthropometry. Sub study from COPSAC2010 birth
cohort. EBioMedicine 83: 104236. http://dx.doi.org/1.0.1.01.6/j.ebioiii.2022.1.04236
Shah-Kulkarni, S; Kim, BM; Hong, YC; Kim, HS; Kwon, EJ; Park, H, et al. (2016). Prenatal exposure to perfluorinated
compounds affects thyroid hormone levels in newborn girls. Environment International 94: 607-613.
Shankar, A; Xiao, J; Ducatman, A. (2011). Perfluoroalkyl chemicals and chronic kidney disease in US adults. American Journal
of Epidemiology 174: 893-900. http://dx.doi.org/10.1093/aje/kwrl71
Shapiro, GD; Dodds, L; Arbuckle, TE; Ashley-Martin, J; Ettinger, AS; Fisher, M, et al. (2016). Exposure to organophosphorus
and organochlorine pesticides, perfluoroalkyl substances, and polychlorinated biphenyls in pregnancy and the
association with impaired glucose tolerance and gestational diabetes mellitus: The MIREC Study. Environmental
Research 147: 71-81.
Shearer, JJ; Callahan, CL; Calafat, AM; Huang, WY; Jones, RR; Sabbisetti, VS, et al. (2021). Serum concentrations of per- and
polyfluoroalkyl substances and risk of renal cell carcinoma. Journal of the National Cancer Institute 113: 580-587.
http://dx.doi.org/lQ.lQ93/inci/diaal43
Shen, C; Ding, J; Xu, C; Zhang, L; Liu, S; Tian, Y. (2022). Perfluoroalkyl mixture exposure in relation to fetal growth: potential
roles of maternal characteristics and associations with birth outcomes. Toxics 10: 650.
http: //dx. doi. org/1.0.33 90/toxic s 1.01.1.0650
Sheng, N; Cui, R; Wang, J; Guo, Y; Wang, J; Dai, J. (2018). Cytotoxicity of novel fluorinated alternatives to long-chain
perfluoroalkyl substances to human liver cell line and their binding capacity to human liver fatty acid binding protein.
Archives of Toxicology 92: 359-369. http://dx.doi.org/1.0.1.007/s00204-01.7-2055-1
Sheng, N; Wang, J; Guo, Y; Wang, J; Dai, J. (2020). Interactions of perfluorooctanesulfonate and 6:2 chlorinated polyfluorinated
ether sulfonate with human serum albumin: A comparative study. Chemical Research in Toxicology 33: 1478-1486.
Shi, F; Almerick T Boncan, D; Wan, HT; Chan, TF; Zhang, EL; Lai, KP; Wong, CK. (2022). Hepatic metabolism gene
expression and gut microbes in offspring, subjected to in-utero PFOS exposure and postnatal diet challenges.
Chemosphere 308: 136196. http://dx.doi.org/10.1.01.6/i.ch.eniospliere.2022.1.361.96
Shih, YH; Blomberg, AJ; Bind, MA; Holm, D; Nielsen, F; Heilmann, C, et al. (2021). Serum vaccine antibody concentrations in
adults exposed to per- and polyfluoroalkyl substances: A birth cohort in the Faroe Islands. Journal of
Immunotoxicology 18: 85-92. http://dx.doi.org/10.1.080/1.547691.X.2021.. 1.922957
Shin, HM; Bennett, DH; Calafat, AM; Tancredi, D; Hertz-Picciotto, I. (2020). Modeled prenatal exposure to per- and
polyfluoroalkyl substances in association with child autism spectrum disorder: A case-control study. Environmental
Research 186: 109514.
Shoaff, J; Papandonatos, GD; Calafat, AM; Chen, A; Lanphear, BP; Ehrlich, S, et al. (2018). Prenatal exposure to perfluoroalkyl
substances: Infant birth weight and early life growth. Environmental Epidemiology 2: eOlO.
Shoeib, M; Harner, T; M Webster, G; Lee, SC. (2011). Indoor sources of poly- and perfluorinated compounds (PFCS) in
Vancouver, Canada: implications for human exposure. Environmental Science and Technology 45: 7999-8005.
http: //dx. doi. org/1.0.1.021. /es 1.03 562v
Shrestha, S; Bloom, MS; Yucel, R; Seegal, RF; Rej, R; Mccaffrey, RJ, et al. (2017). Perfluoroalkyl substances, thyroid
hormones, and neuropsychological status in older adults. International Journal of Hygiene and Environmental Health
220: 679-685.
Singer, AB; Whitworth, KW; Haug, LS; Sabaredzovic, A; Impinen, A; Papadopoulou, E; Longnecker, MP. (2018). Menstrual
cycle characteristics as determinants of plasma concentrations of perfluoroalkyl substances (PFASs) in the Norwegian
Mother and Child Cohort (MoBa study). Environmental Research 166: 78-85.
http: //dx. doi. org/1.0.1.01.6/j. envres.201.8.05.01.9
Sinisalu, L; Sen, P; Salihovic, S; Virtanen, SM; Hyoty, H; Ilonen, J, et al. (2020). Early-life exposure to perfluorinated alkyl
substances modulates lipid metabolism in progression to celiac disease. Environmental Research 188: 109864.
http://dx.doi.Org/10.1016/i.envres.2020.109864
R-29
-------
APRIL 2024
Siwakoti, RC; Cathey, A; Ferguson, KK; Hao, W; Cantonwine, DE; Mukheijee, B, et al. (2023). Prenatal per- and
polyfluoroalkyl substances (PFAS) exposure in relation to preterm birth subtypes and size-for-gestational age in the
LIFECODES cohort 2006-2008. Environmental Research 237: 116967. http://dx.doi.Org/10.1016/i.envies.2023.l 1.6967
Skendzel, LP. (1996). Rubella immunity. Defining the level of protective antibody [Review], American Journal of Clinical
Pathology 106: 170-174. http://dx.doi.Org/10.1093/aicp/106.2.170
Skogheim, TS; Villanger, GD; Weyde, KVF; Engel, SM; Suren, P; 0ie, MG, et al. (2019). Prenatal exposure to perfluoroalkyl
substances and associations with symptoms of attention-deficit/hyperactivity disorder and cognitive functions in
preschool children. International Journal of Elygiene and Environmental Elealth 223: 80-92.
Skuladottir, M; Ramel, A; Rytter, D; Haug, LS; Sabaredzovic, A; Bech, BH, et al. (2015). Examining confounding by diet in the
association between perfluoroalkyl acids and serum cholesterol in pregnancy. Environmental Research 143: 33-38.
Smit, LA; Lenters, V; Efoyer, BB; Lindh, CH; Pedersen, HS; Liermontova, I, et al. (2015). Prenatal exposure to environmental
chemical contaminants and asthma and eczema in school-age children. Allergy 70: 653-660.
Song, X; Tang, S; Zhu, H; Chen, Z; Zang, Z; Zhang, Y, et al. (2018). Biomonitoring PFAAs in blood and semen samples:
Investigation of a potential link between PFAAs exposure and semen mobility in China. Environment International
113: 50-54.
Spliethoff, HM; Tao, L; Shaver, SM; Aldous, KM; Pass, KA; Kannan, K; Eadon, GA. (2008). Use of newborn screening program
blood spots for exposure assessment: declining levels of perluorinated compounds in New York State infants.
Environmental Science and Technology 42: 5361-5367. http://dx.doi.org/1.0.1.021./es8006244
Spratlen, MJ; Perera, FP; Lederman, SA; Rauh, VA; Robinson, M; Kannan, K, et al. (2020a). The association between prenatal
exposure to perfluoroalkyl substances and childhood neurodevelopment. Environmental Pollution 263: 114444.
Spratlen, MJ; Perera, FP; Lederman, SA; Robinson, M; Kannan, K; Herbstman, J; Trasande, L. (2020b). The association between
perfluoroalkyl substances and lipids in cord blood. Journal of Clinical Endocrinology and Metabolism 105: 43-54.
Starling, AP; Adgate, JL; Hamman, RF; Kechris, K; Calafat, AM; Dabelea, D. (2019). Prenatal exposure to per- and
polyfluoroalkyl substances and infant growth and adiposity: The healthy start study. Environment International 131:
104983.
Starling, AP; Adgate, JL; Hamman, RF; Kechris, K; Calafat, AM; Ye, X; Dabelea, D. (2017). Perfluoroalkyl substances during
pregnancy and offspring weight and adiposity at birth: Examining mediation by maternal fasting glucose in the healthy
start study. Environmental Health Perspectives 125: 067016.
Starling, AP; Engel, SM; Richardson, DB; Baird, DD; Haug, LS; Stuebe, AM, et al. (2014a). Perfluoroalkyl Substances During
Pregnancy and Validated Preeclampsia Among Nulliparous Women in the Norwegian Mother and Child Cohort Study.
American Journal of Epidemiology 179: 824-833.
Starling, AP; Engel, SM; Whitworth, KW; Richardson, DB; Stuebe, AM; Daniels, JL, et al. (2014b). Perfluoroalkyl substances
and lipid concentrations in plasma during pregnancy among women in the Norwegian Mother and Child Cohort Study.
Environment International 62: 104-112. http://dx.doi.Org/10.1016/i.envint.2013.10.004
StataCorp. (2021). Stata Statistical Software: Release 17 [Computer Program], College Station, TX: StataCorp LLC.
Steenland, K; Kugathasan, S; Barr, DB. (2018). PFOA and ulcerative colitis. Environmental Research 165: 317-321.
Steenland, K; Tinker, S; Frisbee, S; Ducatman, A; Vaccarino, V. (2009). Association of Perfluorooctanoic Acid and
Perfluorooctane Sulfonate With Serum Lipids Among Adults Living Near a Chemical Plant. American Journal of
Epidemiology 170: 1268-1278. fattp://dx.doi.org/1.0.1.093/aje/kwp279
Steenland, K; Tinker, S; Shankar, A; Ducatman, A. (2010). Association of perfluorooctanoic acid (PFOA) and perfluorooctane
sulfonate (PFOS) with uric acid among adults with elevated community exposure to PFOA. Environmental Health
Perspectives 118: 229-233. http://dx.doi.org/10.1289/ehp.0900940
Stein, CR; Ge, Y; Wolff, MS; Ye, X; Calafat, AM; Kraus, T; Moran, TM. (2016a). Perfluoroalkyl substance serum
concentrations and immune response to FluMist vaccination among healthy adults. Environmental Research 149: 171-
178.
Stein, CR; Mcgovern, KJ; Pajak, AM; Maglione, PJ; Wolff, M. (2016b). Perfluoroalkyl and polyfluoroalkyl substances and
indicators of immune function in children aged 12-19 y: National Health and Nutrition Examination Survey. Pediatric
Research 79: 348-357. http://dx.doi.org/10.1038/pr.2015.213
Stein, CR; Savitz, DA; Dougan, M. (2009). Serum levels of perfluorooctanoic acid and perfluorooctane sulfonate and pregnancy
outcome. American Journal of Epidemiology 170: 837-846. http://dx.doi.org/1.0.1.093/aie/kwp212
Stein, CR; Wolff, MS; Calafat, AM; Kato, K; Engel, SM. (2012). Comparison of polyfluoroalkyl compound concentrations in
maternal serum and amniotic fluid: a pilot study. Reproductive Toxicology 34: 312-316.
fattp: //dx. doi. org/1.0.1.01.6/j .reprotox.201.2.05.039
Stock, NL; Furdui, VI; Muir, DC; Mabury, SA. (2007). Perfluoroalkyl contaminants in the Canadian Arctic: evidence of
atmospheric transport and local contamination. Environmental Science and Technology 41: 3529-3536.
fattp: //dx. doi. org/1.0.1.021. / es062709x
Stelevik, S, .B. (2012) Immunotoxic effects of dietary toxicants: Focus on prenatal exposure to acrylamide, polychlorinated
biphenyls and dioxins. (Doctoral Dissertation). University of Oslo, Oslo. Retrieved from
https://www.duo.uio.no/bitstream/handl e/10852/28057/dravhandling-stolevik.pdf?sequence=3&isAllowed=y
Stratakis, N; Rock, S; La Merrill, MA; Saez, M; Robinson, O; Fecht, D, et al. (2022). Prenatal exposure to persistent organic
pollutants and childhood obesity: A systematic review and meta-analysis of human studies. Obesity Reviews 23(S1):
el3383. http://dx.doi.org/10.1.1.1.1/obr. 1.3383
R-30
-------
APRIL 2024
Stram, M; Hansen, S; Olsen, SF; Haug, LS; Rantakokko, P; Kiviranta, H; Halldorsson, TI. (2014). Persistent organic pollutants
measured in maternal serum and offspring neurodevelopmental outcomes—a prospective study with long-term follow-
up. Environment International 68: 41-48.
Su, TC; Kuo, CC; Hwang, JJ; Lien, GW; Chen, MF; Chen, PC. (2016). Serum perfluorinated chemicals, glucose homeostasis and
the risk of diabetes in working-aged Taiwanese adults. Environment International 88: 15-22.
Sun, Q; Zong, G; Valvi, D; Nielsen, F; Coull, B; Grandjean, P. (2018). Plasma concentrations of perfluoroalkyl substances and
risk of Type 2 diabetes: A prospective investigation among U.S. women. Environmental Health Perspectives 126:
037001.
Tan, Y; Taibl, KR; Dunlop, AL; Barr, DB; Panuwet, P; Yakimavets, V, et al. (2023). Association between a Mixture of Per- and
Polyfluoroalkyl Substances (PFAS) and Inflammatory Biomarkers in the Atlanta African American Maternal-Child
Cohort. Environmental Science & Technology, http://dx.doi.org/1.0.1.021./acs.est. 3c04688
Tao, L; Kannan, K; Wong, CM; Arcaro, KF; Butenhoff, JL. (2008). Perfluorinated compounds in human milk from
Massachusetts, USA. Environmental Science and Technology 42: 3096-3101. http://dx.doi.org/10.1.02l/es702789k
Taylor, KW; Hoffman, K; Thayer, KA; Daniels, JL. (2014). Polyfluoroalkyl chemicals and menopause among women 20-65
years of age (NHANES). Environmental Health Perspectives 122: 145-150. http://dx.doi.org/10.1289/ehp.1306707
Thayer, KA; Angrish, M; Arzuaga, X; Carlson, LM; Davis, A; Dishaw, L, et al. (2022). Systematic evidence map (SEM)
template: report format and methods used for the US EPA integrated risk information system (iris) program,
provisional peer reviewed toxicity value (PPRTV) program, and other "fit for purpose" literature-based human health
analyses (manuscript-in-progress) (pp. 1-69). Thayer, KA; Angrish, M; Arzuaga, X; Carlson, LM; Davis, A; Dishaw,
L; Druwe, I; Gibbons, C; Glenn, B; Jones, R; Kaiser, JP; Keshava, C; Keshava, N; Kraft, A; Lizarraga, L; Persad, A;
Radke, EG; Rice, G; Schulz, B; Shaffer, R; Shannon. T; Shapiro, A; Thacker, S; Vulimiri, S; Williams, AJ; Woodall,
G; Yost, E; Blain, R; Duke, K; Goldstone, AE; Hartman, P; Hobbie, K; Ingle, B; Lemeris, C; Lin, C; Lindahl, A;
McKinley, K; Soleymani, P; Vetter, N.
Thomford, PJ. (2002). 104-week dietary chronic toxicity and carcinogenicity study with perfluorooctane sulfonic acid potassium
salt (PFOS; T-6295) in rats (pp. 002148-002363). (Study No. 6329-183). Madison, WI: Covance Laboratories.
Thompson, J; Lorber, M; Toms, LM; Kato, K; Calafat, AM; Mueller, JF. (2010). Use of simple pharmacokinetic modeling to
characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environment
International 36: 390-397. http://dx.doi.Org/10.1016/i.enviiit.2010.02.008
Thomsen, C; Haug, LS; Stigum, H; Frashaug, M; Broadwell, SL; Becher, G. (2010). Changes in concentrations of perfluorinated
compounds, polybrominated diphenyl ethers, and polychlorinated biphenyls in Norwegian breast-milk during twelve
months of lactation. Environmental Science and Technology 44: 9550-9556. http://dx.doi.org/10.1.021/esl021.922
Tian, M; Reichetzeder, C; Li, J; Hocher, B. (2019a). Low birth weight, a risk factor for diseases in later life, is a surrogate of
insulin resistance at birth. Journal of Hypertension 37: 2123-2134. http://dx.doi.org/10.1097/HJH.00000000000Q2156
Tian, Y; Liang, H; Miao, M; Yang, F; Ji, H; Cao, W, et al. (2019b). Maternal plasma concentrations of perfluoroalkyl and
polyfluoroalkyl substances during pregnancy and anogenital distance in male infants. Human Reproduction 34: 1356-
1368.
Tian, Y; Miao, M; Ji, H; Zhang, X; Chen, A; Wang, Z, et al. (2020). Prenatal exposure to perfluoroalkyl substances and cord
plasma lipid concentrations. Environmental Pollution 268: 115426. http://dx.doi.Org/10.1016/i.envpol.2020.115426
Tian, Y; Zhou, Q; Zhang, L; Li, W; Yin, S; Li, F; Xu, C. (2023). In utero exposure to per-/polyfluoroalkyl substances (PFASs):
Preeclampsia in pregnancy and low birth weight for neonates. Chemosphere 313: 137490.
http://dx.doi.Org/10.1016/i.chemosphere.2022.137490
Tian, YP; Zeng, XW; Bloom, MS; Lin, S; Wang, SQ; Yim, SHL, et al. (2019c). Isomers of perfluoroalkyl substances and
overweight status among Chinese by sex status: Isomers of C8 Health Project in China. Environment International 124:
130-138.
Timmermann, CA; Budtz-Jergensen, E; Jensen, TK; Osuna, CE; Petersen, MS; Steuerwald, U, et al. (2017a). Association
between perfluoroalkyl substance exposure and asthma and allergic disease in children as modified by MMR
vaccination. Journal of Immunotoxicology 14: 39-49.
Timmermann, CA; Budtz-Jergensen, E; Petersen, MS; Weihe, P; Steuerwald, U; Nielsen, F, et al. (2017b). Shorter duration of
breastfeeding at elevated exposures to perfluoroalkyl substances. Reproductive Toxicology 68: 164-170.
Timmermann, CA; Rossing, LI; Grantved, A; Ried-Larsen, M; Dalgard, C; Andersen, LB, et al. (2014). Adiposity and glycemic
control in children exposed to perfluorinated compounds. Journal of Clinical Endocrinology and Metabolism 99: E608-
E614. http://dx.doi.org/10.1210/ic.2013-3460
Timmermann, CAG; Jensen, KJ; Nielsen, F; Budtz-Jergensen, E; van Der Klis, F; Benn, CS, et al. (2020). Serum Perfluoroalkyl
Substances, Vaccine Responses, and Morbidity in a Cohort of Guinea-Bissau Children. Environmental Health
Perspectives 128: 87002.
Timmermann, CAG; Pedersen, HS; Weihe, P; Bjerregaard, P; Nielsen, F; Heilmann, C; Grandjean, P. (2021). Concentrations of
tetanus and diphtheria antibodies in vaccinated Greenlandic children aged 7-12 years exposed to marine pollutants, a
cross sectional study. Environmental Research 203: 111712. http://dx.doi.org/1.0.1.01.6/j.envres.2021.. 1.1.1712
Toft, G; Jonsson, BA; Bonde, JP; Nargaard-Pedersen, B; Hougaard, DM; Cohen, A, et al. (2016). Perfluorooctane Sulfonate
Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in
Danish Boys (1980-1996). Environmental Health Perspectives 124: 151-156.
R-31
-------
APRIL 2024
Toft, G; Jonsson, BAG; Lindh, CH; Giwercman, A; Spano, M; Heederik, D, et al. (2012). Exposure to perfluorinated compounds
and human semen quality in arctic and European populations. Human Reproduction 27: 2532-2540.
littp: //dx. doi. org/1.0.1.093/humrep/des 1.8 5
Tomy, GT; Budakowski, W; Halldorson, T; Helm, PA; Stern, GA; Friesen, K, et al. (2004). Fluorinated organic compounds in an
eastern Arctic marine food web. Environmental Science and Technology 38: 6475-6481.
littp: //dx. doi. org/1.0.1.021. / es049620g
Tosh, PK; Kennedy, RB; Vierkant, RA; Jacobson, RM; Poland, GA. (2009). Correlation between rubella antibody levels and
cytokine measures of cell-mediated immunity. Viral Immunology 22: 451-456.
http://dx.doi.org/10.1089/vim.2009.0068
Trudel, D; Horowitz, L; Wormuth, M; Scheringer, M; Cousins, IT; Hungerbuhler, K. (2008). Estimating consumer exposure to
PFOS and PFOA. [erratum appears in Risk Anal. 2008 Jun;28(3):807], Risk Analysis 28: 251-269.
http://dx.doi.Org/1.0.l 1.1.1/i. 1.539-6924.2008.01.01.7.x
Tsai, MS; Chang, SH; Kuo, WH; Kuo, CH; Li, SY; Wang, MY, et al. (2020). A case-control study of perfluoroalkyl substances
and the risk of breast cancer in Taiwanese women. Environment International 142: 105850.
Tsai, MS; Lin, CC; Chen, MH; Hsieh, WS; Chen, PC. (2017). Perfluoroalkyl substances and thyroid hormones in cord blood.
Environmental Pollution 222: 543-548.
Tsai, MS; Lin, CY; Lin, CC; Chen, MH; Hsu, SH; Chien, KL, et al. (2015). Association between perfluoroalkyl substances and
reproductive hormones in adolescents and young adults. International Journal of Hygiene and Environmental Health
218: 437-443.
U.S. EPA. (1991). Guidelines for developmental toxicity risk assessment. Federal Register 56: 63798-63826.
U.S. EPA. (1996). Guidelines for reproductive toxicity risk assessment (pp. 1-143). (EPA/630/R-96/009). Washington, DC: U.S.
Environmental Protection Agency, Risk Assessment Forum, https://www.epa.gov/sites/production/files/2014-
1.1/docuiiients/guidelines _repro toxicitv.pdf
U.S. EPA. (1998). Health effects test guidelines OPPTS 870.3800 reproduction and fertility effects [EPA Report], (EPA 712-C-
98-208). Washington D.C.: U.S. Environmental Protection Agency, Office of Prevention, Pesticides and Toxic
Substances.
U.S. EPA. (2000). Methodology for deriving ambient water quality criteria for the protection of human health (2000).
(EPA/822/B-00/004). Washington, DC: U.S. Environmental Protection Agency, Office of Water.
http://www.epa.gov/waterscience/criteria/hunianhealth/method/complete.pdf
U.S. EPA. (2002). A review of the reference dose and reference concentration processes. (EPA630P02002F). Washington, DC.
https: //www.epa. gov/sites/production/f'ites/2014-12/documents/rfd-final.pdf
U.S. EPA. (2005a). Guidelines for carcinogen risk assessment [EPA Report], (EPA630P03001F). Washington, DC.
https://www.epa.gov/sites/production/files/2013-09/documents/cancer guidelines J'inal 3-25-05.pdf
U.S. EPA. (2005b). Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens [EPA Report],
(EPA/630/R-03/003F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
https://www.epa.gov/risk/supplemental-guidance-assessing-susceptibility-early-life-exposure-carcinogens
U.S. EPA. (2010). 2008-2009 National Rivers and Streams Assessment Fish Tissue Study. Washington, DC.
https://www.epa.gov/fish-tech/2008-2009-national-rivers-and-streams-assessment-fish-tissue-studv
U.S. EPA. (201 la). 2010 Great Lakes Human Health Fish Tissue Study. Washington, DC: U.S. Environmental Protection
Agency, National Coastal Condition Assessment, https://www.epa.gov/fish-tech/2010-great-lakes-human-health-fish-
tissue-study
U.S. EPA. (2011b). Exposure factors handbook: 2011 edition [EPA Report], (EPA/600/R-090/052F). Washington, DC: U.S.
Environmental Protection Agency, Office of Research and Development, National Center for Environmental
Assessment, http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=236252
U.S. EPA. (2012). Benchmark dose technical guidance. (EPA/100/R-12/001). Washington, DC: U.S. Environmental Protection
Agency, Risk Assessment Forum, https://www.epa.gov/risk/benchmark-dose-technical-guidance
U.S. EPA. (2015). 2013-2014 National Rivers and Streams Assessment Fish Tissue Study. Washington, DC.
https://www.epa.gov/fish-tech/2013-2014-national-rivers-and-streams-assessment-fish-tissue-study
U.S. EPA. (2016a). Drinking water health advisory for perfluorooctane sulfonate (PFOS) [EPA Report], (EPA 822-R-16-004).
Washington, DC: U.S. Environmental Protection Agency, Office of Water.
https://www.epa.gov/sites/production/files/2016-05/documents/pfos health_advisory_fuial 508.pdf
U.S. EPA. (2016b). Drinking water health advisory for perfluorooctanoic acid (PFOA) [EPA Report], (EPA 822-R-16-005).
Washington, DC: U.S. Environmental Protection Agency, Office of Water.
https://www.epa.gov/sites/production/files/2016-05/documents/pfoa health advisory final 508.pdf
U.S. EPA. (2016c). Health effects support document for perfluorooctane sulfonate (PFOS) [EPA Report], (EPA 822-R-16-002).
Washington, DC: U.S. Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division.
https://www.epa.gov/sites/production/files/2016-05/documents/pfos hesd J'inal 508.pdf
U.S. EPA. (2016d). Health effects support document for perfluorooctanoic acid (PFOA) [EPA Report], (EPA 822-R-16-003).
Washington, DC: U.S. Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division.
https://www.epa.gov/sites/production/files/2016-05/documents/pfoa hesd fiiial-plain.pdf
U.S. EPA. (2016e). National Coastal Condition Assessment: 2015 Results. Washington, DC. https://www.epa.gov/national-
aquatic-resource-surveys/national-coastal-condition-assessment-2015-results
R-32
-------
APRIL 2024
U.S. EPA. (2017). Occurrence Data for the Unregulated Contaminant Monitoring Rule: UCMR 3 (2013-2015). Washington,
D.C.: U.S. Environmental Protection Agency, Office of Water. Retrieved from https:/Avww. epa. go v/moni toring-
miregulated-chliik.iiig-water-coiitaiiiiiiaiits/occurrence-data-uiiregulated-coiitaiiiiiiaiit#3
U.S. EPA. (2020a). ORD staff handbook for developing IRIS assessments (public comment draft) [EPA Report], (EPA/600/R-
20/137). Washington, DC: U.S. Environmental Protection Agency, Office of Research and Development, Center for
Public Health and Environmental Assessment, https://cfpub.epa.gov/ncea/iris drafts/recordisplay .cfm?deid=350086
U.S. EPA. (2020b). Systematic review protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA (anionic and acid forms) IRIS
assessments: Supplemental information appendix A [EPA Report], (EPA/635/R-20/131). Washington, DC: US EPA,
ORD, CPHEA, Integrated Risk Information System, littps://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=P 1.010Y0L.txt
U.S. EPA. Announcement of final regulatory determinations for contaminants on the Fourth Drinking Water Contaminant
Candidate List, 86 FR 12272 (2021a). https://www.govinfo.gov/content/pkg/FR-2021-03-03/pdf/2021-04184.pdf
U.S. EPA. (2021b). Emerging issues in food waste management: Persistent chemical contaminants [EPA Report], (EPA/600/R-
21/115). U.S. Environmental Protection Agency, Office of Research and Development.
https://www.epa.gov/system/files/docunients/2021-08/emerging-issues-in-food-waste-management-persistent-
clieniicat-contaminaiits.pdf"
U.S. EPA. (2021c). EXTERNAL PEER RE VIEW DRAFT: Proposed Approaches to the Derivation of a Draft Maximum
Contaminant Level Goal for Perfluorooctane Sulfonic Acid (PFOS) (CASRN 1763-23-1) in Drinking Water [EPA
Report], Washington, DC: U.S. Environmental Protection Agency (EPA).
https://sab.epa.gov/ords/sab/f?p= 1.00:1.8:1.6490947993: ::RP J. 8:P1.8 II):2601.
U.S. EPA. (202Id). Final Regulatory Determination 4 Support Document [EPA Report], (EPA 815R21001). U.S. Environmental
Protection Agency (EPA). https://www.regulations.gov/document/EPA-HQ-OW-2019-0583-0284
U.S. EPA (U.S. Environmental Protection Agency). (2022a). IRIS toxicological review of perfluorobutanoic acid (PFBA,
CASRN 375-22-4) and related salts [EPA Report], (EPA/635/R-22/277Fa). Washington, D.C.: U.S. Environmental
Protection Agency, https://iris.epa.gov/ChemicalLanding/&substance_nmbr=701
U.S. EPA. (2022b). ORD staff handbook for developing IRIS assessments [EPA Report], (EPA 600/R-22/268). Washington, DC:
U.S. Environmental Protection Agency, Office of Research and Development, Center for Public Health and
Environmental Assessment, https://cfpub.epa.gov/ncea/iris draft s/recordi splay.cfm?deid=356370
U.S. EPA. (2022c). Review of EPA's Analysis to Support EPA's National Primary Drinking Water Rulemaking for PFAS.
(EPA-SAB-22-008). U.S. Environmental Protection Agency, Science Advisory Board.
U.S. EPA. (2023a). Technical support document - per- and polyfluoroalkyl substances (PFAS) occurrence & contaminant
background [EPA Report]. (EPA-822-P-22-007).
U.S. EPA (U.S. Environmental Protection Agency). (2023b). Toxicological Review of Perfluorohexanoic Acid (PFHxA) and
Related Salts (Final Report, 2023) [EPA Report], (EPA/635/R-23/027F).
https://cfpub.epa.gov/iicea/iris draft s/recordi splay. cfm?deid=35731.4
U.S. EPA. (2024). Human health toxicity assessment for perfluorooctane sulfonic acid (PFOS) [EPA Report], Washington, DC:
U.S. Environmental Protection Agency, Office of Water.
Uhl, SA; James-Todd, T; Bell, ML. (2013). Association of Osteoarthritis with Perfluorooctanoate and Perfluorooctane Sulfonate
inNHANES 2003-2008. Environmental Health Perspectives 121: 447-452,452e441-443.
Valenti, L; Pelusi, S; Bianco, C; Ceriotti, F; Berzuini, A; Iogna Prat, L, et al. (2021). Definition of Healthy Ranges for Alanine
Aminotransferase Levels: A 2021 Update. Hepatology Communications 5: 1824-1832.
http://dx.doi.Org/https://doi.org/10.1002/hep4.1794
Valvi, D; Oulhote, Y; Weihe, P; Dalgard, C; Bjerve, KS; Steuerwald, U; Grandjean, P. (2017). Gestational diabetes and offspring
birth size at elevated environmental pollutant exposures. Environment International 107: 205-215.
http://dx.doi.Org/10.1016/i.envint.2017.07.016
van den Dungen, MW; Murk, AJ; Kampman, E; Steegenga, WT; Kok, DE. (2017). Association between DNA methylation
profiles in leukocytes and serum levels of persistent organic pollutants in Dutch men. Environmental Epigenetics 3:
dvxOOl.
van der Veen, I; Harming, AC; Stare, A; Leonards, PEG; de Boer, J; Weiss, JM. (2020). The effect of weathering on per- and
polyfluoroalkyl substances (PFASs) from durable water repellent (DWR) clothing. Chemosphere 249: 126100.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere .2020.1261.00
van Gerwen, M; Colicino, E; Guan, H; Dolios, G; Nadkarni, GN; Vermeulen, RCH, et al. (2023). Per- and polyfluoroalkyl
substances (PFAS) exposure and thyroid cancer risk. EBioMedicine 97: 104831.
littp: //dx. doi. org/1.0.1.01.6/j. ebiom.202 3.1.04831.
Velarde, MC; Chan, AFO; Sajo, MEJ, V; Zakharevich, I; Melamed, J; Uy, GLB, et al. (2022). Elevated levels of perfluoroalkyl
substances in breast cancer patients within the Greater Manila Area. Chemosphere 286 Pt 1: 131545.
http://dx.doi.org/10.1.01.6/j.clieniospliere.2021.131.545
Velez, MP; Arbuckle, TE; Fraser, WD. (2015). Maternal exposure to perfluorinated chemicals and reduced fecundity: the
MIREC study. Human Reproduction 30: 701-709. http://dx.doi.org/10.1093/humrep/deu350
Verner, MA; Loccisano, AE; Morken, NH; Yoon, M; Wu, H; Mcdougall, R, et al. (2015). Associations of Perfluoroalkyl
Substances (PFAS) with Lower Birth Weight: An Evaluation of Potential Confounding by Glomerular Filtration Rate
Using a Physiologically Based Pharmacokinetic Model (PBPK). Environmental Health Perspectives 123: 1317-1324.
http://dx.doi.org/10.1.289/ehp. 1.408837
R-33
-------
APRIL 2024
Verner, MA; Longnecker, MP. (2015). Comment on "enhanced elimination of perfluorooctanesulfonic acid by menstruating
women: evidence from population-based pharmacokinetic modeling" [Letter], Environmental Science & Technology
49: 5836-5837. http://dx.doi.org/10.1.021/acs.est.SbO01.87
Verner, MA; Ngueta, G; Jensen, ET; Fromme, H; Voelkel, W; Nygaard, UC, et al. (2016). A simple pharmacokinetic model of
prenatal and postnatal exposure to perfluoroalkyl substances (PFASs). Environmental Science and Technology 50: 978-
986. http://dx.doi.org/10.1.02l/acs.est.SM)4399
Vestergren, R; Cousins, IT. (2009). Tracking the pathways of human exposure to perfluorocarboxylates [Review], Environmental
Science and Technology 43: 5565-5575. http://dx.doi.org/10.1.02l/es90Q228k
Vesterholm Jensen, D; Christensen, J; Virtanen, HE; Skakkebask, NE; Main, KM; Toppari, J, et al. (2014). No association
between exposure to perfluorinated compounds and congenital cryptorchidism: a nested case-control study among 215
boys from Denmark and Finland. Reproduction 147: 411-417.
Volkel, W; Genzel-Boroviczeny, O; Demmelmair, El; Gebauer, C; Koletzko, B; Twardella, D, et al. (2008). Perfluorooctane
sulphonate (PFOS) and perfluorooctanoic acid (PFOA) in human breast milk: results of a pilot study. International
Journal of Hygiene and Environmental Health 211: 440-446. http://dx.doi.Org/10.1016/i.iiheh.2007.07.024
von Ehrenstein, OS; Fenton, SE; Kato, K; Kuklenyik, Z; Calafat, AM; Hines, EP. (2009). Polyfluoroalkyl chemicals in the serum
and milk of breastfeeding women. Reproductive Toxicology 27: 239-245.
http://dx.doi.Org/10.1016/i.reprotox.2009.03.001
Vuong, A; Yolton, K; Webster, GM; Sjodin, A; Calafat, AM; Braun, JM, et al. (2016). Prenatal polybrominated diphenyl ether
and perfluoroalkyl substance exposures and executive function in school-age children. Environmental Research 147:
556-564.
Vuong, AM; Braun, JM; Yolton, K; Wang, Z; Xie, C; Webster, GM, et al. (2018a). Prenatal and childhood exposure to
perfluoroalkyl substances (PFAS) and measures of attention, impulse control, and visual spatial abilities. Environment
International 119: 413-420.
Vuong, AM; Xie, C; Jandarov, R; Dietrich, KN; Zhang, H; Sjodin, A, et al. (2020a). Prenatal exposure to a mixture of persistent
organic pollutants (POPs) and child reading skills at school age. International Journal of Hygiene and Environmental
Health 228: 113527.
Vuong, AM; Yolton, K; Braun, JM; Sjodin, A; Calafat, AM; Xu, Y, et al. (2020b). Polybrominated diphenyl ether (PBDE) and
poly- and perfluoroalkyl substance (PFAS) exposures during pregnancy and maternal depression. Environment
International 139: 105694.
Vuong, AM; Yolton, K; Wang, Z; Xie, C; Webster, GM; Ye, X, et al. (2018b). Childhood perfluoroalkyl substance exposure and
executive function in children at 8 years. Environment International 119: 212-219.
Vuong, AM; Yolton, K; Xie, C; Dietrich, KN; Braun, JM; Webster, GM, et al. (2019). Prenatal and childhood exposure to poly-
and perfluoroalkyl substances (PFAS) and cognitive development in children at age 8 years. Environmental Research
172: 242-248.
Wambaugh, JF; Setzer, RW; Pitruzzello, AM; Liu, J; Reif, DM; Kleinstreuer, NC, et al. (2013). Dosimetric anchoring of in vivo
and in vitro studies for perfluorooctanoate and perfluorooctanesulfonate. Toxicological Sciences 136: 308-327.
http://dx.doi.org/10.1093/toxsci/kfl204
Wan, C; Han, R; Liu, L; Zhang, F; Li, F; Xiang, M; Ding, W. (2016). Role of miR-155 in fluorooctane sulfonate-induced
oxidative hepatic damage via the Nrf2-dependent pathway. Toxicology and Applied Pharmacology 295: 85-93.
Wan, HT; Zhao, YG; Leung, PY; Wong, CK. (2014). Perinatal exposure to perfluorooctane sulfonate affects glucose metabolism
in adult offspring. PLoS ONE 9: e87137. http://dx.doi.org/10.1371/joumal.pone.0087137
Wang, B; Zhang, R; Jin, F; Lou, H; Mao, Y; Zhu, W, et al. (2017). Perfluoroalkyl substances and endometriosis-related infertility
in Chinese women. Environment International 102: 207-212.
Wang, H; Du, H; Yang, J; Jiang, H; O, K; Xu, L, et al. (2019a). PFOS, PFOA, estrogen homeostasis, and birth size in Chinese
infants. Chemosphere 221: 349-355.
Wang, H; Li, W; Yang, J; Wang, Y; Du, H; Han, M, et al. (2023a). Gestational exposure to perfluoroalkyl substances is
associated with placental DNA methylation and birth size. Science of the Total Environment 858: 159747.
http://dx.doi.Org/1.0.1.01.6/i.scitoteiiY.2022.1.59747
Wang, H; Wei, K; Wu, Z; Liu, F; Wang, D; Peng, X, et al. (2022a). Association between per- and polyfluoroalkyl substances and
semen quality. Environmental Science and Pollution Research 30: 27884-27894. http://dx.doi.org/10.1007/sl 1356-022-
241.82-3
Wang, H; Yang, J; Du, H; Xu, L; Liu, S; Yi, J, et al. (2018a). Perfluoroalkyl substances, glucose homeostasis, and gestational
diabetes mellitus in Chinese pregnant women: A repeat measurement-based prospective study. Environment
International 114: 12-20.
Wang, IJ; Hsieh, WS; Chen, CY; Fletcher, T; Lien, GW; Chiang, HL, et al. (2011). The effect of prenatal perfluorinated
chemicals exposures on pediatric atopy. Environmental Research 111: 785-791.
littp: //dx. doi. org/1.0.1.01.6/j. envres.201.1..04,006
Wang, J; Pan, Y; Cui, Q; Yao, B; Wang, J; Dai, J. (2018b). Penetration of PFASs across the blood cerebrospinal fluid barrier and
its determinants in humans. Environmental Science and Technology 52: 13553-13561.
Wang, J; Zeng, XW; Bloom, MS; Qian, Z; Hinyard, LJ; Belue, R, et al. (2019b). Renal function and isomers of
perfluorooctanoate (PFOA) and perfluorooctanesulfonate (PFOS): Isomers of C8 Health Project in China.
Chemosphere 218: 1042-1049.
R-34
-------
APRIL 2024
Wang, Y; Han, W; Wang, C; Zhou, Y; Shi, R; Bonefeld-Jergensen, EC, et al. (2019c). Efficiency of maternal-fetal transfer of
perfluoroalkyl and polyfluoroalkyl substances. Environmental Science and Pollution Research 26: 2691-2698.
Wang, Y; Liu, W; Zhang, Q; Zhao, H; Quan, X. (2015a). Effects of developmental perfluorooctane sulfonate exposure on spatial
learning and memory ability of rats and mechanism associated with synaptic plasticity. Food and Chemical Toxicology
76: 70-76. http://dx.doi.Org/10.1016/i.fct.2014.12.008
Wang, Y; Rogan, WJ; Chen, HY; Chen, PC; Su, PH; Chen, HY; Wang, SL. (2015b). Prenatal exposure to perfluroalkyl
substances and children's IQ: The Taiwan maternal and infant cohort study. International Journal of Hygiene and
Environmental Health 218: 639-644.
Wang, Y; Rogan, WJ; Chen, PC; Lien, GW; Chen, HY; Tseng, YC, et al. (2014). Association between maternal serum
perfluoroalkyl substances during pregnancy and maternal and cord thyroid hormones: Taiwan maternal and infant
cohort study. Environmental Health Perspectives 122: 529-534.
Wang, Y; Starling, AP; Haug, LS; Eggesbo, M; Becher, G; Thomsen, C, et al. (2013). Association between perfluoroalkyl
substances and thyroid stimulating hormone among pregnant women: a cross-sectional study. Environmental Health: A
Global Access Science Source 12: 76. http://dx.doi.Org/10.l 1.86/1.476-069X-1.2-76
Wang, Y, u; Zhao, H; Zhang, Q; Liu, W, ei; Quan, X, ie. (2015c). Perfluorooctane sulfonate induces apoptosis of hippocampal
neurons in rat offspring associated with calcium overload. Toxicology Research 4: 931-938.
Wang, Y; Zhang, L; Teng, Y; Zhang, J; Yang, L; Li, J, et al. (2018c). Association of serum levels of perfluoroalkyl substances
with gestational diabetes mellitus and postpartum blood glucose. Journal of Environmental Sciences 69: 5-11.
Wang, Z; Luo, J; Zhang, Y; Li, J; Zhang, J; Tian, Y; Gao, Y. (2023b). High maternal glucose exacerbates the association between
prenatal per- and polyfluoroalkyl substance exposure and reduced birth weight. Science of the Total Environment 858:
160130. http://dx.doi.Org/10.1016/i.scitotenv.2022.160130
Wang, Z; Shi, R; Ding, G; Yao, Q; Pan, C; Gao, Y; Tian, Y. (2022b). Association between maternal serum concentration of
perfluoroalkyl substances (PFASs) at delivery and acute infectious diseases in infancy. Chemosphere 289: 133235.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere .2021.133235
Warembourg, C; Maitre, L, ea; Tamayo-Uria, I; Fossati, S; Roumeliotaki, T; Aasvang, GM, et al. (2019). Early-life
environmental exposures and blood pressure in children. Journal of the American College of Cardiology 74: 1317-
1328.
Washino, N; Saijo, Y; Sasaki, S; Kato, S; Ban, S; Konishi, K, et al. (2009). Correlations between prenatal exposure to
perfluorinated chemicals and reduced fetal growth. Environmental Health Perspectives 117: 660-667.
littp: // dx. doi. org/1.0.1.2 89/ehp. 1.1.681.
Watkins, DJ; Josson, J; Elston, B; Bartell, SM; Shin, HM; Vieira, VM, et al. (2013). Exposure to perfluoroalkyl acids and
markers of kidney function among children and adolescents living near a chemical plant. Environmental Health
Perspectives 121: 625-630. http://dx.doi.org/1.0.1289/ehp. 1.205838
Webster, GM; Venners, SA; Mattman, A; Martin, JW. (2014). Associations between perfluoroalkyl acids (PFASs) and maternal
thyroid hormones in early pregnancy: a population-based cohort study. Environmental Research 133: 338-347.
littp: //dx. doi. org/1.0.1.01.6/j. envres.201.4.06,01.2
Wen, HJ; Wang, SL; Chen, PC; Guo, YL. (2019a). Prenatal perfluorooctanoic acid exposure and glutathione s-transferase Tl/Ml
genotypes and their association with atopic dermatitis at 2 years of age. PLoS ONE 14: e0210708.
Wen, HJ; Wang, SL; Chuang, YC; Chen, PC; Guo, YL. (2019b). Prenatal perfluorooctanoic acid exposure is associated with
early onset atopic dermatitis in 5-year-old children. Chemosphere 231: 25-31.
Wen, LL; Lin, LY; Su, TC; Chen, PC; Lin, CY. (2013). Association between serum perfluorinated chemicals and thyroid
function in U.S. adults: the National Health and Nutrition Examination Survey 2007-2010. Journal of Clinical
Endocrinology and Metabolism 98: E1456-E1464. littp://dx.doi.org/1.0.1.21.0/jc.201.3-1.282
Wen, X; Wang, M; Xu, X; Li, T. (2022). Exposure to Per- and Polyfluoroalkyl Substances and Mortality in U.S. Adults: A
Population-Based Cohort Study. Environmental Health Perspectives 130: 67007. littp://dx.doi.org/1.0.1.289/EHP1.0393
Weng, J; Hong, C; Tasi, J; Shen, CY, u; Su, P; Wang, S. (2020). The association between prenatal endocrine-disrupting chemical
exposure and altered resting-state brain fMRI in teenagers. Brain Structure and Function 225: 1669-1684.
Whitworth, KW; Haug, LS; Baird, DD; Becher, G; Hoppin, JA; Skjaerven, R, et al. (2012a). Perfluorinated compounds in
relation to birth weight in the Norwegian Mother and Child Cohort Study. American Journal of Epidemiology 175:
1209-1216. littp://dx.doi.org/1.0.1.093/aje/kwrf59
Whitworth, KW; Haug, LS; Baird, DD; Becher, G; Hoppin, JA; Skjaerven, R, et al. (2012b). Perfluorinated compounds and
subfecundity in pregnant women. Epidemiology 23: 257-263. http://dx.doi.org/10.1097/EDE.0b013e31823b5031
WHO. (2018). The immunological basis for immunization series. Module 3: Tetanus. Geneva.
https://apps.who.int/iris/bitstream/handle/10665/275340/9789241.51.3616-eng.pdf?sequence=l&isAllowed=y
Wiener, RC; Waters, C. (2019). Perfluoroalkyls/polyfluoroalkyl substances and dental caries experience in children, ages 3-11
years, National Health and Nutrition Examination Survey, 2013-2014. Journal of Public Health Dentistry 79: 307-319.
Wikstrom, S; Hussein, G; Lingroth Karlsson, A; Lindh, CH; Bornehag, CG. (2021). Exposure to perfluoroalkyl substances in
early pregnancy and risk of sporadic first trimester miscarriage. Scientific Reports 11: 3568.
http://dx.doi.org/10.1038/s41598-021-82748-6
Wikstrom, S; Lin, PI; Lindh, CH; Shu, H; Bornehag, CG. (2020). Maternal serum levels of perfluoroalkyl substances in early
pregnancy and offspring birth weight. Pediatric Research 87: 1093-1099.
R-35
-------
APRIL 2024
Wikstrom, S; Lindh, CH; Shu, H; Bornehag, CG. (2019). Early pregnancy serum levels of perfluoroalkyl substances and risk of
preeclampsia in Swedish women. Scientific Reports 9: 9179.
Wimsatt, J; Villers, M; Thomas, L; Kamarec, S; Montgomery, C; Yeung, LW, et al. (2016). Oral perlluorooctane sulfonate
(PFOS) lessens tumor development in the APC(min) mouse model of spontaneous familial adenomatous polyposis.
BMC Cancer 16: 942.
Winquist, A; Hodge, JM; Diver, WR; Rodriguez, JL; Troeschel, AN; Daniel, J; Teras, LR. (2023). Case-Cohort Study of the
Association between PFAS and Selected Cancers among Participants in the American Cancer Society's Cancer
Prevention Study II LifeLink Cohort. Environmental Health Perspectives 131: 127007.
http://dx.doi.org/10.1289/EHP13174
Wong, F; Macleod, M; Mueller, JF; Cousins, IT. (2014). Enhanced elimination of perlluorooctane sulfonic acid by menstruating
women: evidence from population-based pharmacokinetic modeling. Environmental Science and Technology 48: 8807-
8814. fattp: //dx. dot ore/1.0.1.021. /es500796v
Woodcraft, MW; Ellis, DA; Rafferty, SP; Burns, DC; March, RE; Stock, NL, et al. (2010). Experimental characterization of the
mechanism of perfluorocarboxylic acids' liver protein bioaccumulation: the key role of the neutral species.
Environmental Toxicology and Chemistry 29: 1669-1677. http://dx.doi.org/10.1.002/etc. 1.99
Woods, MM; Lanphear, BP; Braun, JM; McCandless, LC. (2017). Gestational exposure to endocrine disrupting chemicals in
relation to infant birth weight: A Bayesian analysis of the HOME Study. Environmental Health: A Global Access
Science Source 16: 115. http://dx.doi.ore/10.llB6/sl294Q-Q17-Q332-3
Workman, CE; Becker, AB; Azad, MB; Moraes, TJ; Mandhane, PJ; Turvey, SE, et al. (2019). Associations between
concentrations of perfluoroalkyl substances in human plasma and maternal, infant, and home characteristics in
Winnipeg, Canada. Environmental Pollution 249: 758-766.
Worley, RR; Moore, SM; Tierney, BC; Ye, X; Calafat, AM; Campbell, S, et al. (2017). Per- and polyfluoroalkyi substances in
human serum and urine samples from a residentially exposed community. Environment International 106: 135-143.
Wu, XM; Bennett, DH; Calafat, AM; Kato, K; Strynar, M; Andersen, E, et al. (2014). Serum concentrations of perfluorinated
compounds (PFC) among selected populations of children and Adults in California. Environmental Research 136C:
264-273. littp: //dx.doi. org/10.1016/j.envres.201.4,09.026
Xiao, C; Grandjean, P; Valvi, D; Nielsen, F; Jensen, TK; Weihe, P; Oulhote, Y. (2019). Associations of exposure to
perfluoroalkyl substances with thyroid hormone concentrations and birth size. Journal of Clinical Endocrinology and
Metabolism 105: 735-745.
Xing, J; Wang, G; Zhao, J; Wang, E; Yin, B; Fang, D, et al. (2016). Toxicity assessment of perlluorooctane sulfonate using acute
and subchronic male C57BL/6J mouse models. Environmental Pollution 210: 388-396.
Xu, H; Mao, Y; Hu, Y; Xu, B. (2020a). Association between exposure to polyfluoroalkyi chemicals and increased fractional
exhaled nitric oxide in adults. Environmental Research In Press: 110450.
littp: //dx. doi. org/10.1016/j. envres.2020.1.1.0450
Xu, H; Zhou, Q; Zhang, J; Chen, X; Zhao, H; Lu, H, et al. (2020b). Exposure to elevated per- and polyfluoroalkyi substances in
early pregnancy is related to increased risk of gestational diabetes mellitus: A nested case-control study in Shanghai,
China. Environment International 143: 105952.
Xu, J; Guo, CS; Zhang, Y; Meng, W. (2014). Bioaccumulation and trophic transfer of perfluorinated compounds in a eutrophic
freshwater food web. Environmental Pollution 184: 254-261. http://dx.doi.org/10.1.01.6/i .eiivpol.201.3.09.01.1.
Xu, Y; Fletcher, T; Pineda, D; Lindh, CH; Nilsson, C; Glynn, A, et al. (2020c). Serum Half-Lives for Short- and Long-Chain
Perfluoroalkyl Acids after Ceasing Exposure from Drinking Water Contaminated by Firefighting Foam. Environmental
Health Perspectives 128: 77004.
Xu, Y; Li, Y; Scott, K; Lindh, CH; Jakobsson, K; Fletcher, T, et al. (2020d). Inflammatory bowel disease and biomarkers of gut
inflammation and permeability in a community with high exposure to perfluoroalkyl substances through drinking
water. Environmental Research 181: 108923.
Xu, Z; Du, B; Wang, H; Li, Z; Wu, Y; Wang, Q, et al. (2023). Perfluoroalkyl substances in umbilical cord blood and blood
pressure in offspring: a prospective cohort study. Environmental Health 22: 72. http://dx.doi.org/10.1.1.86/s 1.2940-023-
01.023-5
Yahia, D; Tsukuba, C; Yoshida, M; Sato, I; Tsuda, S. (2008). Neonatal death of mice treated with perlluorooctane sulfonate.
Journal of Toxicological Sciences 33: 219-226. http://dx.doi.org/10.21.31/its.33.219
Yamaguchi, M; Arisawa, K; Uemura, H; Katsuura-Kamano, S; Takami, H; Sawachika, F, et al. (2013). Consumption of seafood,
serum liver enzymes, and blood levels of PFOS and PFOA in the Japanese population. Journal of Occupational Health
55: 184-194. http://dx.doi.org/10.1539/ioh. 12-0264-QA
Yang, BY; Wu, J; Niu, X; He, C; Bloom, MS; Abudoukade, M, et al. (2022). (In Press) Low-level environmental per- and
polyfluoroalkyi substances and preterm birth: A nested case-control study among a Uyghur population in northwestern
China. Exposure and Health. http://dx.doi.org/10.1007/sl2403-021-0Q454-0
Yang, C; Lee, HK; Zhang, Y; Jiang, LL; Chen, ZF; Chung, ACK; Cai, Z. (2019). In Situ Detection and Imaging of PFOS in
Mouse Kidney by Matrix-Assisted Laser Desorption/Ionization Imaging Mass Spectrometry. Analytical Chemistry 91:
8783-8788. http://dx.doi.org/10.1021/acs.analchem.9b00711.
Yang, D; Han, J; Hall, DR; Sun, J; Fu, J; Kutarna, S, et al. (2020a). Nontarget Screening of Per- and Polyfluoroalkyi Substances
Binding to Human Liver Fatty Acid Binding Protein. Environmental Science and Technology 54: 5676-5686.
R-36
-------
APRIL 2024
Yang, J; Wang, H; Du, H; Fang, H; Han, M; Xu, L, et al. (2020b). Serum perfluoroalkyl substances in relation to lipid
metabolism in Chinese pregnant women. Chemosphere 273: 128566.
http: //dx. doi. org/1.0.1.01.6/j. chemosphere.2020.1.28566
Yang, L; Ji, H; Liang, H; Yuan, W; Song, X; Li, X, et al. (2022a). Associations of perfluoroalkyl and polyfluoroalkyl substances
with gestational hypertension and blood pressure during pregnancy: A cohort study. Environmental Research 215:
114284. http://dx.doi.org/1.0.1.01.6/jenvres.2022.1.14284
Yang, L; Li, J; Lai, J; Luan, H; Cai, Z; Wang, Y, et al. (2016a). Placental transfer of perfluoroalkyl substances and associations
with thyroid hormones: Beijing prenatal exposure study. Scientific Reports 6: 21699.
Yang, L; Wang, Z; Shi, Y; Li, J; Wang, Y; Zhao, Y, et al. (2016b). Human placental transfer of perfluoroalkyl acid precursors:
Levels and profiles in paired maternal and cord serum. Chemosphere 144: 1631-1638.
Yang, M; Li, LY; Qin, XD; Ye, XY; Yu, S; Bao, Q, et al. (2021). Perfluorooctanesulfonate and perfluorooctanoate exacerbate
airway inflammation in asthmatic mice and in vitro. Science of the Total Environment 766: 142365.
http://dx.doi.org/10.1016/iscitoteiiY.2020.142365
Yang, Q; Guo, X; Sun, P; Chen, Y; Zhang, W; Gao, A. (2018). Association of serum levels of perfluoroalkyl substances (PFASs)
with the metabolic syndrome (MetS) in Chinese male adults: A cross-sectional study. Science of the Total Environment
621: 1542-1549.
Yang, Z; Liu, HY; Yang, QY; Chen, X; Li, W; Leng, J; Tang, NJ. (2022b). Associations between exposure to perfluoroalkyl
substances and birth outcomes: A meta-analysis. Chemosphere 291:132909.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere .2021.132 909
Yao, Q; Gao, Y; Zhang, Y; Qin, K; Liew, Z; Tian, Y. (2021). Associations of paternal and maternal per- and polyfluoroalkyl
substances exposure with cord serum reproductive hormones, placental steroidogenic enzyme and birth weight.
Chemosphere 285: 131521. h ttp://dx.doi. org/10.1.01.6/j .chemosphere.2021.131.521.
Yao, Q; Shi, R; Wang, C; Han, W; Gao, Y; Zhang, Y, et al. (2019). Cord blood per- and polyfluoroalkyl substances, placental
steroidogenic enzyme, and cord blood reproductive hormone. Environment International 129: 573-582.
Ye, L; Zhao, B; Yuan, K; Chu, Y; Li, C; Zhao, C, et al. (2012). Gene expression profiling in fetal rat lung during gestational
perfluorooctane sulfonate exposure. Toxicology Letters 209: 270-276. http://dx.doi.org/10.1.01.6/j.toxtet.201.1.. 1.2.01.3
Yeung, EH; Bell, EM; Sundaram, R; Ghassabian, A; Ma, W; Kannan, K; Louis, GM. (2019). Examining endocrine disruptors
measured in newborn dried blood spots and early childhood growth in a prospective cohort. Obesity 27: 145-151.
Yeung, LWY; Robinson, SJ; Koschorreck, J; Mabury, SA. (2013). Part II. A Temporal Study of PFOS and Its Precursors in
Human Plasma from Two German Cities in 1982-2009. Environmental Science and Technology 47: 3875-3882.
http://dx.doi.org/10.1021/es4004153
Young, W; Wiggins, S; Limm, W; Fisher, CM; Dejager, L; Genualdi, S. (2022). Analysis of Per- and Poly(fluoroalkyl)
Substances (PFASs) in Highly Consumed Seafood Products from U.S. Markets. Journal of Agricultural and Food
Chemistry 70: 13545-13553. http://dx.doi.org/10.1021/acs.iafc.2c04673
Yu, WG; Liu, W; Jin, YH. (2009a). Effects of perfluorooctane sulfonate on rat thyroid hormone biosynthesis and metabolism.
Environmental Toxicology and Chemistry 28: 990-996. http://dx.doi.Org/10.1897/08-345.l
Yu, WG; Liu, W; Jin, YH; Liu, XH; Wang, FQ; Liu, L; Nakayama, SF. (2009b). Prenatal and postnatal impact of
perfluorooctane sulfonate (PFOS) on rat development: a cross-foster study on chemical burden and thyroid hormone
system. Environmental Science and Technology 43: 8416-8422. http://dx.doi.org/10.1021/es901602d
Yu, WG; Liu, W; Liu, L; Jin, YH. (2011). Perfluorooctane sulfonate increased hepatic expression of OAPT2 and MRP2 in rats.
Archives of Toxicology 85: 613-621. http: //dx. doi. org/1.0.1.007/s00204-01.0-061.3-x
Yu, Y; Qin, XD; Bloom, MS; Chu, C; Dai, X; Li, QQ, et al. (2022). Associations of prenatal exposure to perfluoroalkyl
substances with preterm birth: A family-based birth cohort study. Environmental Research 214: 113803.
http://dx.doi.Org/10.1016/i.envres.2022.l 1.3803
Yu, Y; Quan, X; Wang, H; Zhang, B; Hou, Y; Su, C. (2023). Assessing the health risk of hyperuricemia in participants with
persistent organic pollutants exposure - A systematic review and meta-analysis [Review], Ecotoxicology and
Environmental Safety 251: 114525. http://dx.doi.Org/10.1016/i.ecoenv.2023.l 1.4525
Yu, Y; Wang, C; Zhang, X; Zhu, J; Wang, L; Ji, M, et al. (2019). Perfluorooctane sulfonate disrupts the blood brain barrier
through the crosstalk between endothelial cells and astrocytes in mice. Environmental Pollution 256: 113429.
Zabaleta, I; Blanco-Zubiaguirre, L; Baharli, EN; Olivares, M; Prieto, A; Zuloaga, O; Elizalde, MP. (2020). Occurrence of per-
and polyfluorinated compounds in paper and board packaging materials and migration to food simulants and foodstuffs.
Food Chemistry 321: 126746. http://dx.doi.Org/10.1016/i.foodchem.2020.126746
Zabaleta, I; Negreira, N; Bizkarguenaga, E; Prieto, A; Covaci, A; Zuloaga, O. (2017). Screening and identification of per- and
polyfluoroalkyl substances in microwave popcorn bags. Food Chemistry 230: 497-506.
http: //dx. doi. org/1.0.1.01.6/j. foodchem.201.7.03.074
Zafeiraki, E; Gebbink, WA; Hoogenboom, R; Kotterman, M; Kwadijk, C; Dassenakis, E; van Leeuwen, SPJ. (2019). Occurrence
of perfluoroalkyl substances (PFASs) in a large number of wild and farmed aquatic animals collected in the
Netherlands. Chemosphere 232: 415-423. http://dx.doi.Org/10.1016/i.chemosphere.2019.05.200
Zare Jeddi, M; Dalla Zuanna, T; Barbieri, G; Fabricio, ASC; Dapra, F; Fletcher, T, et al. (2021a). Associations of Perfluoroalkyl
Substances with Prevalence of Metabolic Syndrome in Highly Exposed Young Adult Community Residents-A Cross-
Sectional Study in Veneto Region, Italy. International Journal of Environmental Research and Public Health 18: 1194.
http: //dx. doi. org/1.0.33 90/ij erph 1.8031.1.94
R-37
-------
APRIL 2024
Zare Jeddi, M; Soltanmohammadi, R; Barbieri, G; Fabricio, ASC; Pitter, G; Dalla Zuanna, T; Canova, C. (2021b). To which
extent are per-and poly-fluorinated substances associated to metabolic syndrome? [Review]. Reviews on
Environmental Health, http://dx.doi.org/10.1.51.5/reveh-2020-0144
Zareitalabad, P; Siemens, J; Hamer, M; Amelung, W. (2013). Perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid
(PFOS) in surface waters, sediments, soils and wastewater - A review on concentrations and distribution coefficients
[Review], Chemosphere 91: 725-732. http://dx.doi.org/1.0.1.01.6/j .chemosphere. 201.3.02.024
Zeeshan, M; Yang, Y; Zhou, Y; Huang, W; Wang, Z; Zeng, XY, et al. (2020). Incidence of ocular conditions associated with
perfluoroalkyl substances exposure: Isomers of C8 Health Project in China. Environment International 137: 105555.
Zell-Baran, LM; Dabelea, D; Norris, JM; Glueck, DH; Adgate, JL; Brown, JM, et al. (2023). Prenatal Exposure to Poly- and
Perfluoroalkyl Substances (2009-2014) and Vaccine Antibody Titers of Measles, Mumps, Rubella, and Varicella in
Children Four to Eight Years Old from the Healthy Start Cohort. Environmental Health Perspectives 131: 127018.
http://dx.doi.org/10.1289/EHP12863
Zeng, HC; Zhang, L; Li, YY; Wang, YJ; Xia, W; Lin, Y, et al. (2011). Inflammation-like glial response in rat brain induced by
prenatal PFOS exposure. NeuroToxicology 32: 130-139. http://dx.doi.org/1.0.1.01.6/i.iieuro.201.0.1.0.001.
Zeng, X; Chen, Q; Zhang, X; Li, H; Liu, Q; Li, C, et al. (2019a). Association between prenatal exposure to perfluoroalkyl
substances and asthma-related diseases in preschool children. Environmental Science and Pollution Research 26:
29639-29648.
Zeng, X; Chen, T; Cui, Y; Zhao, J; Chen, Q; Yu, Z, et al. (2023). In utero exposure to perfluoroalkyl substances and early
childhood BMI trajectories: A mediation analysis with neonatal metabolic profiles. Science of the Total Environment
867: 161504. http://dx.doi.Org/10.1016/i.scitotenv.2023.161504
Zeng, XW; Bloom, MS; Dharmage, SC; Lodge, CJ; Chen, D; Li, S, et al. (2019b). Prenatal exposure to perfluoroalkyl substances
is associated with lower hand, foot and mouth disease viruses antibody response in infancy: Findings from the
Guangzhou Birth Cohort Study. Science of the Total Environment 663: 60-67.
Zeng, XW; Li, QQ; Chu, C; Ye, WL; Yu, S; Ma, H, et al. (2020). Alternatives of perfluoroalkyl acids and hepatitis B virus
surface antibody in adults: Isomers of C8 Health Project in China. Environmental Pollution 259: 113857.
Zeng, XW; Lodge, CJ; Dharmage, SC; Bloom, MS; Yu, Y; Yang, M, et al. (2019c). Isomers of per- and polyfluoroalkyl
substances and uric acid in adults: Isomers of C8 Health Project in China. Environment International 133: 105160.
Zeng, XW; Qian, Z; Emo, B; Vaughn, M; Bao, J; Qin, XD, et al. (2015). Association of polyfluoroalkyl chemical exposure with
serum lipids in children. Science of the Total Environment 512-513: 364-370.
Zhang, B; Wei, Z; Gu, C; Yao, Y; Xue, J; Zhu, H, et al. (2022a). First evidence of prenatal exposure to emerging poly- and
perfluoroalkyl substances associated with e-waste dismantling: Chemical structure-based placental transfer and health
risks. Environmental Science & Technology 56: 17108-17118. http://dx.doi.org/10.1.021./acs.est.2c0S92S
Zhang, C; Sundaram, R; Maisog, J; Calafat, AM; Barr, DB; Buck Louis, GM. (2015a). A prospective study of prepregnancy
serum concentrations of perfluorochemicals and the risk of gestational diabetes. Fertility and Sterility 103: 184-189.
http: //dx. doi. org/1.0.1.01.6/i. fertns tert .2014.10.001.
Zhang, H; Lu, H; Yu, L; Yuan, J; Qin, S; Li, C, et al. (2021). Effects of gestational exposure to perfluorooctane sulfonate on the
lung development of offspring rats. Environmental Pollution 272: 115535.
http: //dx. doi. org/1.0.1.01.6/i. envpol.2020.1.1.5 53 5
Zhang, H; Yolton, K; Webster, GM; Ye, X; Calafat, AM; Dietrich, KN, et al. (2018a). Prenatal and childhood perfluoroalkyl
substances exposures and children's reading skills at ages 5 and 8 years. Environment International 111: 224-231.
Zhang, L; Liang, J; Gao, A. (2023a). Contact to perfluoroalkyl substances and thyroid health effects: A meta-analysis directing
on pregnancy [Review], Chemosphere 315: 137748. http://dx.doi.org/1.0.1.01.6/j.chemosphere.2023.1.37748
Zhang, L; Ren, XM; Guo, LH. (2013a). Structure-based investigation on the interaction of perfluorinated compounds with human
liver fatty acid binding protein. Environmental Science and Technology 47: 11293-11301.
http://dx.doi.org/10.1021/es4Q26722
Zhang, Q; Liu, W; Zhao, H; Zhang, Z; Qin, H; Luo, F; Niu, Q. (2019). Developmental perfluorooctane sulfonate exposure
inhibits long-term potentiation by affecting AMPA receptor trafficking. Toxicology 412: 55-62.
Zhang, S; Tan, R; Pan, R; Xiong, J; Tian, Y; Wu, J; Chen, L. (2018b). Association of perfluoroalkyl and polyfluoroalkyl
substances with premature ovarian insufficiency in Chinese women. Journal of Clinical Endocrinology and Metabolism
103: 2543-2551.
Zhang, T; Fu, S; Yu, K; Albanes, D; Moore, SC; Purdue, MP; Stolzenberg-Solomon, RZ. (2023b). Nested Case-Control Studies
Investigating Serum Perfluorooctanoate and Perfluorooctane Sulfonate Levels and Pancreatic Ductal Adenocarcinoma
in Two Cohorts. Environmental Health Perspectives 131: 107702. http://dx.doi.org/1.0.1.289/EHP1.3208
Zhang, T; Qin, X. (2014). Assessment of fetal exposure and maternal elimination of perfluoroalkyl substances. Environmental
Science: Processes & Impacts 16: 1878-1881.
Zhang, T; Sun, H; Lin, Y; Qin, X; Zhang, Y; Geng, X; Kannan, K. (2013b). Distribution of poly- and perfluoroalkyl substances
in matched samples from pregnant women and carbon chain length related maternal transfer. Environmental Science
and Technology 47: 7974-7981.
Zhang, T; Sun, H; Qin, X; Gan, Z; Kannan, K. (2015b). PFOS and PFOA in paired urine and blood from general adults and
pregnant women: assessment of urinary elimination. Environmental Science and Pollution Research 22: 5572-5579.
Zhang, X; Chen, L; Fei, XC; Ma, YS; Gao, HW. (2009). Binding of PFOS to serum albumin and DNA: Insight into the
molecular toxicity of perfluorochemicals. BMC Molecular Biology 10: 16. http://dx.doi.org/1.0.1.1.86/1471 -21.99-1.0-1.6
R-38
-------
APRIL 2024
Zhang, X; Xue, L; Deji, Z; Wang, X; Liu, P; Lu, J, et al. (2022b). Effects of exposure to per- and polyfluoroalkyl substances on
vaccine antibodies: A systematic review and meta-analysis based on epidemiological studies [Review], Environmental
Pollution 306: 119442. littp://dx.doi.org/1.0.1.01.6/j.envpol.2022.1.1.9442
Zhang, Y; Beesoon, S; Zhu, L; Martin, JW. (2013c). Biomonitoring of perfluoroalkyl acids in human urine and estimates of
biological half-life. Environmental Science and Technology 47: 10619-10627.
Zhang, Y; Beesoon, S; Zhu, L; Martin, JW. (2013d). Isomers of perfluorooctanesulfonate and perfluorooctanoate and total
perfluoroalkyl acids in human serum from two cities in North China. Environment International 53: 9-17.
littp: //dx. doi. org/1.0.1.01.6/j. eiivint .201.2.1.2.007
Zhang, Y; Mustieles, V; Sun, Y; Oulhote, Y; Wang, YX; Messerlian, C. (2022c). Association between serum per- and
polyfluoroalkyl substances concentrations and common cold among children and adolescents in the United States.
Environment International 164: 107239. http://dx.doi.Org/10.1016/i.enviiit.2022.107239
Zhang, Y; Mustieles, V; Wang, YX; Sun, Q; Coull, B; Sun, Y, et al. (2023c). Red blood cell folate modifies the association
between serum per- and polyfluoroalkyl substances and antibody concentrations in U.S. adolescents. Environmental
Science & Technology 57: 2445-2456. http://dx.doi.org/10.1021/acs.est.2c07152
Zhang, Y; Pan, C; Ren, Y; Wang, Z; Luo, J; Ding, G, et al. (2022d). Association of maternal exposure to perfluoroalkyl and
polyfluroalkyl substances with infant growth from birth to 12 months: A prospective cohort study. Science of the Total
Environment 806: 151303. http://dx.doi.org/1.0.1.01.6/j.scitotenv.2021.151.303
Zhang, Z; Wang, F; Zhang, Y; Yao, J; Bi, J; He, J, et al. (2022e). Associations of serum PFOA and PFOS levels with incident
hypertension risk and change of blood pressure levels. Environmental Research 212: 113293.
littp: //dx. doi. org/1.0.1.01.6/j. envres.2022.1.1.32 93
Zhao, L; Zhang, Y; Zhu, L; Ma, X; Wang, Y; Sun, H; Luo, Y. (2017a). Isomer-Specific Transplacental Efficiencies of
Perfluoroalkyl Substances in Human Whole Blood. Environmental Science & Technology Letters 4: 391-398.
Zhao, W; Zitzow, JD; Ehresman, DJ; Chang, SC; Butenhoff, JL; Forster, J; Hagenbuch, B. (2015). Na+/taurocholate
cotransporting polypeptide and apical sodium-dependent bile acid transporter are involved in the disposition of
perfluoroalkyl sulfonates in humans and rats. Toxicological Sciences 146: 363-373.
Zhao, W; Zitzow, JD; Weaver, Y; Ehresman, DJ; Chang, SC; Butenhoff, JL; Hagenbuch, B. (2017b). Organic anion transporting
polypeptides contribute to the disposition of perfluoroalkyl acids in humans and rats. Toxicological Sciences 156: 84-
95.
Zhao, Y; Jin, H; Qu, J; Zhang, S; Hu, S; Xue, J; Zhao, M. (2022a). The influences of perfluoroalkyl substances on the rheumatoid
arthritis clinic. BMC Immunology 23: 10. http://dx.doi.org/10.1186/sl2865-022-00483-7
Zhao, Y; Liu, W; Qu, J; Hu, S; Zhang, L; Zhao, M, et al. (2022b). Per-/polyfluoroalkyl substance concentrations in human serum
and their associations with immune markers of rheumatoid arthritis. Chemosphere 298: 134338.
littp: //dx. doi. org/1.0.1.01.6/j. chemosphere.2022.1.343 38
Zheng, T; Kelsey, K; Zhu, C; Pennell, KD; Yao, Q; Manz, KE, et al. (2023). Adverse birth outcomes related to concentrations of
per- and polyfluoroalkyl substances (PFAS) in maternal blood collected from pregnant women in 1960-1966.
Environmental Researchl 17010. littp://dx.doi.org/1.0.1.01.6/j.envres.2023.11701.0
Zhong, SQ; Chen, ZX; Kong, ML; Xie, YQ; Zhou, Y; Qin, XD, et al. (2016). Testosterone-Mediated Endocrine Function and
TH1/TH2 Cytokine Balance after Prenatal Exposure to Perfluorooctane Sulfonate: By Sex Status. International Journal
of Molecular Sciences 17.
Zhou, W; Zhang, L; Tong, C; Fang, F; Zhao, S; Tian, Y, et al. (2017a). Plasma perfluoroalkyl and polyfluoroalkyl substances
concentration and menstrual cycle characteristics in preconception women. Environmental Health Perspectives 125:
067012.
Zhou, Y; Bao, WW; Qian, ZM; Dee Geiger, S; Parrish, KL; Yang, BY, et al. (2017b). Perfluoroalkyl substance exposure and
urine CC16 levels among asthmatics: A case-control study of children. Environmental Research 159: 158-163.
Zhou, Y; Hu, LW; Qian, ZM; Chang, JJ; King, C; Paul, G, et al. (2016). Association of perfluoroalkyl substances exposure with
reproductive hormone levels in adolescents: By sex status. Environment International 94: 189-195.
Zhou, Y; Hu, LW; Qian, ZM; Geiger, SD; Parrish, KL; Dharmage, SC, et al. (2017c). Interaction effects of polyfluoroalkyl
substances and sex steroid hormones on asthma among children. Scientific Reports 7: 899.
Zhu, Y; Qin, XD; Zeng, XW; Paul, G; Morawska, L; Su, MW, et al. (2016). Associations of serum perfluoroalkyl acid levels
with T-helper cell-specific cytokines in children: By gender and asthma status. Science of the Total Environment 559:
166-173.
Zong, G; Grandjean, P; Wang, X; Sun, Q. (2016). Lactation history, serum concentrations of persistent organic pollutants, and
maternal risk of diabetes. Environmental Research 150: 282-288.
R-39
------- |