A EPA
External Review Draft
EPA/635/R-22/039a
www.epa.gov/iris
Toxicological Review of Formaldehyde—Inhalation
[CASRN 50-00-0]
April 2022
Integrated Risk Information System
Center for Public Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Toxicological Review of Formaldehyde—Inhalation
DISCLAIMER
This document is a public comment draft for review purposes only. This information is
distributed solely for the purpose of public comment It has not been formally disseminated by EPA.
It does not represent and should not be construed to represent any Agency determination or policy.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
This document is a draft for review purposes only and does not constitute Agency policy.
ii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
CONTENTS
AUTHORS | CONTRIBUTORS | REVIEWERS xvii
PREFACE ON ASSESSMENT METHODS AND ORGANIZATION xix
EXECUTIVE SUMMARY li
ES.l Overall Summary li
ES.2 Hazard Assessment Summary liv
ES.3 Dose-response Assessment Summary Ivi
ES.4 Susceptible Populations and Lifestages lix
1. HAZARD IDENTIFICATION 1-1
1.1. SUMMARY OF USES, HUMAN EXPOSURE, AND TOXICOKINETICS 1-1
1.1.1. Chemical Properties and Uses of Formaldehyde 1-1
1.1.2. Exposure to Formaldehyde 1-2
1.1.3. Toxicokinetics of Formaldehyde 1-2
1.2. SYNTHESIS OF EVIDENCE FOR EFFECTS ON THE RESPIRATORY SYSTEM 1-10
1.2.1. Sensory Irritation 1-11
1.2.2. Pulmonary Function 1-34
1.2.3. Immune-mediated Conditions, Focusing on Allergies and Asthma 1-77
1.2.4. Respiratory Tract Pathology 1-149
1.2.5. Respiratory Tract Cancers 1-196
1.3. SYNTHESIS OF EVIDENCE FOR NONRESPIRATORY EFFECTS 1-341
1.3.1. Nervous System Effects 1-341
1.3.2. Developmental and Reproductive Toxicity 1-382
1.3.3. Lymphohematopoietic Cancers 1-434
1.4. SUMMARY AND EVALUATION 1-544
1.4.1. Susceptible Populations and Lifestages 1-544
1.4.2. Summary of Evidence Integration Conclusions for Effects Other Than Cancer 1-555
1.4.3. Summary of Evidence Integration Conclusions for Carcinogenicity 1-558
2. DOSE-RESPONSE ANALYSIS 2-1
2.1. INHALATION REFERENCE CONCENTRATION FOR EFFECTS OTHER THAN CANCER 2-1
This document is a draft for review purposes only and does not constitute Agency policy.
iii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
2.1.1. Choice of Studies and Endpoints and Calculation of PODs 2-2
2.1.2. Derivation of Candidate Reference Concentrations 2-25
2.1.3. Selection of Organ- or System-specific Reference Concentrations 2-32
2.1.4. Summary of Organ- or System-specific RfCs and RfC Selection 2-36
2.1.5. Previous IRIS Assessment: Reference Value 2-42
2.2. INHALATION UNIT RISK ESTIMATE FOR CANCER 2-42
2.2.1. Unit Risk Estimates for Nasal Cancer 2-45
2.2.2. Derivation of a Myeloid Leukemia Unit Risk Estimate Based on Human Data 2-82
2.2.3. Summary of Unit Risk Estimates and the Preferred Estimate for Inhalation Unit
Risk 2-93
2.2.4. Adjustment of Human-based Unit Risk Estimates for Potential Increased Early-life
Susceptibility 2-96
2.2.5. Cancer Risk Based on Background Cancer Incidence and Internal Dose of
Endogenous and Exogenous Formaldehyde 2-99
2.2.6. Preferred Inhalation Unit Risk Estimate 2-101
2.2.7. Previous IRIS Assessment: Inhalation Unit Risk 2-109
REFERENCES R-l
This document is a draft for review purposes only and does not constitute Agency policy.
iv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
TABLES
Table I. General approach to literature search strategies xxiv
Table II. Criteria and presentation of strength of the evidence for each mechanistic event and for
potential associations between events relating to potential respiratory health
effects xxxii
Table III. Information most relevant to describing primary considerations informing causality
during evidence syntheses xxxv
Table IV. Primary considerations for assessing the strength of evidence for the health effects
studies in humans and, separately, animals xxxix
Table V. Examples of the interpretation and application of mechanistic evidence xli
Table VI. Framework for strength of evidence judgments (human evidence) xlii
Table VII. Framework for strength of evidence judgments (animal evidence) xliv
Table VIII. Overall evidence integration judgments for characterizing potential human health
hazards (noncancer health effects and cancer outcomes) in the evidence
integration narrative xlvi
Table IX. Criteria for applying cancer descriptors to overall confidence conclusions for cancer
types xlvii
Table X. Considerations for study selection for quantification of dose-response and derivation of
toxicity values xlix
Table ES-1. Evidence integration judgments for noncancer health effects and the reference
concentration (RfC) lii
Table ES-2. Cancer evidence integration judgments, carcinogenicity descriptor, and inhalation
unit risk (IUR) for cancer incidence liii
Table 1-1. Summary of controlled human exposure studies of formaldehyde and human sensory
irritation 1-16
Table 1-2. Summary of epidemiological studies of residential exposures to formaldehyde and
human sensory irritation 1-25
Table 1-3. Mechanistic evidence most informative to the occurrence of sensory irritation after
formaldehyde inhalation 1-31
Table 1-4. Evidence integration summary for effects on sensory irritation 1-33
Table 1-5. Common measures of pulmonary function reported in studies of formaldehyde
inhalation 1-37
Table 1-6. Formaldehyde effects on pulmonary function in laboratory settings (changes over <1
year) 1-41
Table 1-7. Formaldehyde effects on pulmonary function in occupational settings (long-term
effects) 1-48
Table 1-8. Formaldehyde effects on pulmonary function among adults in residential settings 1-58
Table 1-9. Formaldehyde effects on pulmonary function among children in residential or school
settings 1-63
Table 1-10. Mechanistic evidence most informative to the occurrence of decreased pulmonary
function after formaldehyde inhalation 1-68
Table 1-11. Evidence integration summary for effects on pulmonary function 1-76
Table 1-12. History of allergy-related conditions in relation to formaldehyde exposure, by age
group 1-85
Table 1-13. Skin prick tests in relation to formaldehyde exposure, by age group 1-92
Table 1-14. Allergy symptoms or skin prick tests in relation to formaldehyde exposure in workers 1-94
This document is a draft for review purposes only and does not constitute Agency policy.
v DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-15. Prevalence of asthma in children or adults in relation to residential or school
formaldehyde exposure in studies with relatively low exposures (<0.05 mg/m3) 1-99
Table 1-16. Prevalence of asthma in children or adults in relation to residential formaldehyde
exposure in studies with relatively high exposures (>0.05 mg/m3) 1-104
Table 1-17. Prevalence of asthma in relation to occupational formaldehyde exposure 1-109
Table 1-18. Exacerbation of asthma symptoms in relation to residential formaldehyde exposure 1-113
Table 1-19. Controlled acute exposure chamber studies of pulmonary function with
formaldehyde exposure among people with asthma 1-115
Table 1-20. Respiratory conditions in infants and young children in relation to residential
formaldehyde exposure 1-118
Table 1-21. Effect modification by environmental tobacco smoke: results from studies in
children and adults 1-125
Table 1-22. Mechanistic evidence most informative to the development of immune-mediated
conditions after formaldehyde inhalation 1-134
Table 1-23. Summary of changes in cell counts and soluble immunological factors in the blood
following formaldehyde exposure 1-140
Table 1-24. Evidence integration summary for effects on immune-mediated conditions, including
allergies and asthma 1-147
Table 1-25. Formaldehyde effects on respiratory pathology in occupational settings 1-154
Table 1-26. Chronic respiratory pathology studies in animals 1-171
Table 1-27. Subchronic respiratory pathology studies in animals 1-178
Table 1-28. Selected short-term respiratory pathology studies in animals (see Appendix A.5.5 for
others) 1-187
Table 1-29. Mechanistic evidence most informative to the development of respiratory tract
pathology after formaldehyde inhalation 1-193
Table 1-30. Evidence integration summary for effects of formaldehyde inhalation on respiratory
pathology 1-195
Table 1-31. Age-standardized (world) incidence rates of nasopharyngeal cancer per 100,000 per
year 1-202
Table 1-32. Epidemiological studies of formaldehyde exposure and risk of nasopharyngeal
cancers 1-212
Table 1-33. Epidemiological studies of formaldehyde exposure and risk of sinonasal cancers 1-239
Table 1-34. Studies of formaldehyde exposure and risk of cancer of oropharynx/hypopharynx 1-259
Table 1-35. Epidemiological studies of formaldehyde exposure and risk of laryngeal cancer 1-273
Table 1-36. Squamous cell carcinoma (SCC) incidence in rats exposed to formaldehyde for
>2 years 1-290
Table 1-37. Respiratory tract cancer—chronic and subchronic (with long-term follow up)
exposure in rats, mice, and hamsters 1-296
Table 1-38. Concordance of temporal and dose-response relationships among formaldehyde
effects induced in F344 rat nasal epithelium in vivo 1-306
Table 1-39. Genotoxicity and mutagenicity in the upper respiratory tract 1-307
Table 1-40. Direct measurements of DNA synthesis in the upper respiratory tract 1-309
Table 1-41. Epithelial pathology, cytotoxicity, and regenerative proliferation in the upper
respiratory tract 1-311
Table 1-42. Summary considerations for upper respiratory tract (URT) carcinogenesis 1-333
Table 1-43. Evidence integration summary for effects of formaldehyde inhalation on URT
cancers 1-338
This document is a draft for review purposes only and does not constitute Agency policy.
vi DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-44. Summary of alterations in neurobehavioral tests in relation to formaldehyde
exposure in observational epidemiology and controlled exposure studies 1-347
Table 1-45. Summary of human studies of nervous system disease risk in relation to
formaldehyde exposure 1-351
Table 1-46. Developmental neuropathology in experimental animal studies 1-360
Table 1-47. Neural sensitization in experimental animal studies 1-363
Table l-48.Tests of motor-related behaviors in experimental animal studies 1-367
Table 1-49. Tests of learning and memory in experimental animal studies 1-371
Table 1-50. Evidence integration summary for nervous system effects after formaldehyde
inhalation 1-379
Table 1-51. Epidemiology studies describing effects on time to pregnancy in relation to
formaldehyde exposure 1-386
Table 1-52. Epidemiology studies describing effects on spontaneous abortion in relation to
formaldehyde exposure 1-391
Table 1-53. Epidemiology studies describing effects on prenatal growth and births outcomes in
relation to formaldehyde exposure 1-396
Table 1-54. Epidemiology studies describing male reproductive toxicity in relation to
formaldehyde exposure 1-399
Table 1-55. Summary of developmental effects observed in animal studies following inhalation
exposure to formaldehyde 1-408
Table 1-56. Summary of female reproductive effects observed in animal studies following
inhalation exposure to formaldehyde 1-413
Table 1-57. Summary of male reproductive effects observed in animal studies following
inhalation exposure to formaldehyde 1-421
Table 1-58. Evidence integration summary for effects of formaldehyde inhalation on
reproduction and development 1-433
Table 1-59. Summary high confidence studies of reported exposure-response trends describing
the effect estimates of association between formaldehyde exposure and risk of
myeloid leukemia 1-448
Table 1-60. Epidemiological studies of formaldehyde exposure and risk of myeloid leukemia 1-455
Table 1-61. Epidemiological studies of formaldehyde exposure and risk of lymphatic leukemia 1-472
Table 1-62. Epidemiological studies of formaldehyde exposure and risk of multiple myeloma 1-487
Table 1-63. Epidemiological studies of formaldehyde exposure and risk of Hodgkin lymphoma 1-502
Table 1-64. Cumulative incidence of hematopoietic cancers in B6C3F1 mice and F344 rats 1-513
Table 1-65. Summary of animal evidence of lymphohematopoietic cancers and bone marrow
histopathology following inhalation exposure to formaldehyde 1-514
Table 1-66. Summary conclusions regarding plausible mechanistic events associated with
formaldehyde induction of lymphohematopoietic cancers 1-538
Table 1-67. Evidence integration summary for effects of formaldehyde inhalation on LHP cancers ... 1-542
Table 2-1. Eligible studies for POD derivation and rationale for decisions to not select specific
studies 2-4
Table 2-2. Summary of derivation of PODs for sensory irritation 2-9
Table 2-3. Summary of derivation of PODs for pulmonary function 2-11
Table 2-4. Summary of derivation of PODs for allergies and current asthma based on
observational epidemiology studies 2-15
Table 2-5. Summary of derivation of POD for squamous metaplasia based on observations in
F344 rats (Kerns et al., 1983) 2-19
This document is a draft for review purposes only and does not constitute Agency policy.
vii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-6. Summary of derivation of PODs for squamous metaplasia based on studies in F344
and Wistar rats (Woutersen et al., 1989; Kerns et al., 1983) 2-20
Table 2-7. Adjusted time-weighted average formaldehyde exposures for Taskinen et al. (1999) 2-22
Table 2-8. Summary of derivation of PODs for reproductive toxicity in females 2-23
Table 2-9. Summary of derivation of PODs for reproductive toxicity in males 2-25
Table 2-10. Health effects and corresponding derivation of candidate RfCs 2-30
Table 2-11. Organ- or system-specific RfCs for formaldehyde inhalation 2-36
Table 2-12. Proposed overall RfC for formaldehyde inhalation 2-41
Table 2-13. Demographic details about the NCI industrial workers cohort 2-46
Table 2-14. Exposure details about the NCI industrial workers cohort 2-46
Table-2-15. Relative risk estimates for mortality from nasopharyngeal malignancies (ICD-8 code
147) by level of formaldehyde exposure for different exposure metrics 2-49
Table-2-16. Regression coefficients from NCI log-linear trend test models for NPC mortality from
cumulative exposure to formaldehyde 2-49
Table 2-17. Extra risk estimates for nasopharyngeal cancer mortality from various levels of
continuous exposure to formaldehyde 2-50
Table 2-18. ECooos, LECooos, and inhalation unit risk estimates for nasopharyngeal cancer mortality
from formaldehyde exposure based on the Beane Freeman et al. (2013) log-
linear trend analyses for cumulative exposure 2-51
Table 2-19. ECooos, LECooos, and inhalation unit risk estimates for nasopharyngeal cancer
incidence from formaldehyde exposure based on the Beane Freeman et al.
(2013) log-linear trend analyses for cumulative exposure 2-52
Table -2-20. F344 rat nasal cancer data 2-55
Table-2-21. Dosimetric and mechanistic information supporting dose-response assessment
based on rat nasal tumors 2-57
Table 2-22. Benchmark concentrations and human equivalents using formaldehyde flux and DPX
as dose metrics 2-63
Table 2-23. BBDR model estimated extra risk of SCC in human respiratory tract compared with
EPA's modeling of extra risk of NPC from the human occupational epidemiology
data 2-68
Table 2-24. Evaluation of BBDR modeling issues 2-70
Table 2-25. Sensitivity of BBDR modeled human SCC risk at 0.15 ppm to small variations in
normal (aN) and initiated (ai) cell replication rates 2-75
Table 2-26. Unit risk estimates derived from benchmark estimates 2-80
Table 2-27. Comparison and basis of unit risk estimates for nasopharyngeal cancer in humans
and nasal squamous cell carcinomas in rats 2-81
Table 2-28. Strengths and uncertainties in the cancer type-specific unit risk estimate for
nasopharyngeal cancer 2-82
Table 2-29. Relative risk estimates for mortality from multiple myeloma (ICD-8 code 203),
leukemia (ICD-8 codes 204-207), myeloid leukemia (ICD-8 code 205), and
other/unspecified leukemia (ICD-8 code 207) by level of formaldehyde exposure
for different exposure metrics 2-84
Table 2-30. Regression coefficients for leukemia, myeloid leukemia, and myeloid plus
other/unspecified leukemias mortality from NCI trend test models of cumulative
exposure 2-85
Table 2-31. Extra risk estimates for myeloid plus other/unspecified leukemia mortality from
various levels of continuous lifetime exposure to formaldehyde 2-88
This document is a draft for review purposes only and does not constitute Agency policy.
viii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-32. ECoos, LECoos, and inhalation unit risk estimates for myeloid plus other/unspecified
leukemia mortality from formaldehyde exposure based on log-linear trend
analyses of cumulative exposure data from the Beane Freeman et al. (2009)
study 2-89
Table 2-33. ECoos, LECoos, and inhalation unit risk estimates for myeloid plus other/unspecified
leukemia incidence from formaldehyde exposure based on Beane Freeman et al.
(2009) log-linear trend analyses for cumulative exposure 2-90
Table 2-34. ECoos and LECoos estimates for mortality and incidence and incidence unit risk
estimates for all leukemia and for myeloid leukemia using alternate approaches
(all person-years) 2-90
Table 2-35. Exposure-response modeling (all person-years) and (incidence) unit risk estimate
derivation results for different leukemia groupings 2-91
Table 2-36. Strengths and uncertainties in the cancer type-specific unit risk estimate for myeloid
leukemia 2-92
Table 2-37. Inhalation unit risk estimates by cancer type based on human data 2-93
Table 2-38. Adult-based unit risk estimates for nasopharyngeal cancer for use in ADAF
calculations and risk estimate calculations involving less-than-lifetime exposure
scenarios 2-98
Table 2-39. NPC incidence risk from exposure to constant formaldehyde exposure level of 1
Hg/m3 from ages 0 to 70 years 2-99
Table 2-40. Inhalation unit risk 2-102
Table 2-41. Summary of BMR/EC estimates 2-102
FIGURES
Figure I. Overview of assessment methods for hazard identification xxiii
Figure II. Summary depictions of evaluation of epidemiology studies xxviii
Figure III. Process for evidence integration xxxvii
Figure 1-1. Schematic of the rat upper respiratory tract depicting the gradient of formaldehyde
concentration formed following inhalation exposure, both from anterior to
posterior locations, as well as across the tissue depth 1-5
Figure 1-2. Prevalence of eye irritation in controlled human exposure studies of formaldehyde 1-14
Figure 1-3. Prevalence of eye irritation among study groups exposed to formaldehyde in
residential settings and controlled human exposure studies 1-24
Figure 1-4. Possible mechanistic associations between formaldehyde exposure and sensory
irritation 1-29
Figure 1-5. Forest plots depicting mean difference in pulmonary function (percentage predicted)
between exposed and comparison groups for FEVi, FVC, and FEF 1-46
Figure 1-6. Association of PEFR measured at bedtime and in the morning with household mean
formaldehyde concentration among children less than 15 years of age
(Krzyzanowski et al., 1990) 1-62
Figure 1-7. Possible mechanistic associations between formaldehyde exposure and decreased
pulmonary function 1-65
Figure 1-8. Relative risk estimates for prevalence of allergy-related conditions in children and
adults in relation to formaldehyde in residential and school settings 1-84
This document is a draft for review purposes only and does not constitute Agency policy.
ix DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Figure 1-9. Relative risk estimates for prevalence of asthma in children and adults in relation to
formaldehyde by exposure level in general population and occupational studies 1-99
Figure 1-10. Relative risk of persistent wheeze and of increased frequency of symptoms among
children with wheeze in relation to residential formaldehyde exposure 1-113
Figure 1-11. Examination of effect modification by family or personal history of atopy 1-124
Figure 1-12. Possible mechanistic associations between formaldehyde exposure and immune-
mediated conditions, including allergies and asthma 1-128
Figure 1-13. Inflammatory and immune cells involved in asthma 1-129
Figure 1-14. Example cross-section levels in rat nasal passages used for histopathological
evaluations from Kerns et al. (1983) 1-153
Figure 1-15. Squamous metaplasia in medium and high confidence chronic and subchronic
respiratory pathology studies of inhaled formaldehyde 1-158
Figure 1-16. Squamous metaplasia incidence in high and medium confidence rat studies of
chronic and subchronic formaldehyde exposure duration 1-159
Figure 1-17. The four epithelial cell populations that line the nasal lateral wall in monkeys and
rats are portrayed in this image 1-161
Figure 1-18. Possible mechanistic associations between formaldehyde exposure and respiratory
tract pathology 1-191
Figure 1-19. Schematic diagram of the human upper respiratory tract (i.e., nose, nasal cavity,
paranasal sinuses, pharynx, larynx), as well as neighboring structures (from
Vokes et al. (1993)) 1-197
Figure 1-20. Epidemiological studies reporting nasopharyngeal cancer risk estimates 1-211
Figure 1-21. Highest (medium) confidence epidemiological studies reporting sinonasal cancer
risk estimates 1-237
Figure 1-22. Epidemiological studies reporting oropharyngeal or hypopharyngeal cancer risk
estimates 1-258
Figure 1-23. Epidemiological studies reporting laryngeal cancer risk estimates 1-272
Figure 1-24. Nasal SCCs in rats exposed to formaldehyde for at least 2 years 1-291
Figure 1-25. An integrated cancer mode-of-action (MOA) network for the URT 1-305
Figure 1-26. Mechanistic relationships relevant to URT carcinogenesis 1-314
Figure 1-27. Network of adverse outcome pathways relevant to URT carcinogenesis 1-315
Figure 1-28. Human studies of medium or high confidence examining the potential for
formaldehyde exposure to cause ALS 1-345
Figure 1-29. Nervous system effects in animal studies 1-356
Figure 1-30. Medium confidence animal studies of nervous system effects 1-357
Figure 1-31. Risk of spontaneous abortion associated with maternal occupational formaldehyde
exposure 1-390
Figure 1-32. Animal studies evaluating the effects of formaldehyde inhalation exposure on
developmental toxicity 1-404
Figure 1-33. Animal studies evaluating female reproductive toxicity 1-412
Figure 1-34. Animal studies evaluating male reproductive toxicity 1-416
Figure 1-35. Medium and high confidence animal studies evaluating male reproductive toxicity 1-417
Figure 1-36. The hematopoietic pathway and likely sites of neoplastic transformation for LHPs 1-438
Figure 1-37. All epidemiological studies reporting myeloid leukemia risk estimates 1-453
Figure 1-38. High and medium confidence epidemiological studies reporting myeloid leukemia
risk estimates 1-454
Figure 1-39. Epidemiological studies reporting acute myeloid leukemia risk estimates 1-455
Figure 1-40. Epidemiological studies reporting lymphatic leukemia risk estimates 1-471
This document is a draft for review purposes only and does not constitute Agency policy.
x DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Figure 1-41. Epidemiological studies reporting multiple myeloma risk estimates 1-486
Figure 1-42. Epidemiological studies reporting multiple Hodgkin lymphoma estimates 1-502
Figure 1-43. Simplified hematopoiesis 1-519
Figure 2-1. Candidate RfCs with corresponding POD and composite UF 2-32
Figure 2-2. Organ- or system-specific RfC scatterplot 2-38
Figure 2-3. Illustration of noncancer toxicity value estimations 2-40
Figure 2-4. Fit to the rat tumor incidence data using the model and assumptions in Conolly et al.
(2003) 2-60
Figure 2-5. Cellular proliferation measured by DNA labeling in studies >12 weeks 2-66
Figure 2-6. Dose-response for normal cell division rate, aN, versus formaldehyde flux to tissue for
the F344 rat nasal epithelium 2-72
Figure 2-7. Small variations to ai(flux) for flux <475 pmol/mm2-h carried out for sensitivity
analysis 2-74
This document is a draft for review purposes only and does not constitute Agency policy.
xi DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
ABBREVIATIONS
ACA
activated carbon aerosol
ACTH
adrenocorticotropic hormone
ADAF
age-dependent adjustment factor
ADH
alcohol dehydrogenase
ADME
absorption, distribution, metabolism, excretion
ALDH2
aldehyde dehydrogenase 2
ALI
air-liquid interface
ALL
acute lymphoblastic leukemia
ALM
anterior lateral meatus
ALS
amyotrophic lateral sclerosis
AML
acute myeloid leukemia
ANOVA
analysis of variance
AOP
adverse outcome pathways
AON
adverse outcome network
ATS
American Thoracic Society
ATSDR
Agency for Toxic Substances and Disease Registry
BAL
bronchoalveolar lavage
BALF
bronchoalveolar lavage fluid
BBDR
biologically based dose-response
BER
base excision repair
BM-MSC
bone marrow mesenchymal stem cell
BMC
benchmark concentration
BMCL
benchmark concentration, lower confidence bound
BMD
benchmark dose
BMDL
benchmark dose lower limit
BMI
body mass index
BMR
benchmark response
BrdU
5-bromodeoxyuridine
BTPS
body temperature and ambient pressure saturated (with water vapor)
BW
body weight
CA
chromosomal aberration
CASN
Chemical Abstracts Service Number
CASRN
Chemical Abstracts Service Registry Number
CDC
Centers for Disease Control and Prevention
CE
cumulative exposure
CFD
computational fluid dynamicfs]
CFU
colony-forming unit
CFU-GM
colony-forming unit-granulocytes and macrophages
CGRP
calcitonin gene related protein
CI
confidence interval
CUT
Chemical Industry Institute of Toxicology
CLL
chronic lymphatic leukemia
CML
chronic myeloid leukemia
CNS
central nervous system
COPD
chronic obstructive pulmonary disease
COSMIC
Catalogue of Somatic Mutations in Cancer
cRfC
candidate reference concentration
CRH
corticotropin-releasing hormone
CS
conditioned stimulus
CTL
cytotoxic T lymphocytes
CV
coefficient of variation
This document is a draft for review purposes only and does not constitute Agency policy.
xii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
DDC
DNA crosslink
DG
dentate gyrus
DNA
deoxyribonucleic acid
DPX
DNA-protein crosslink
DSB
double strand break
EA
ethyl acetate
EBV
Epstein-Barr virus
EC
effective concentration
ECHRS
European Community Respiratory Health Survey
EEG
electroencephalogram
EI
exposure index
EMG
electromyelogram
EPA
Environmental Protection Agency
ER
endoplasmic reticulum
ETS
environmental tobacco smoke
FANC
Fanconi anemia family
FDR
fecundability density ratio
FEF
forced expiratory flow
FEV
forced expiratory volume
FSH
follicle-stimulating hormone
FVC
forced vital capacity
GD
gestational day
GLP
good laboratory practice
GM
granulocyte, monocyte
GSNO
S-nitrosoglutathione
GSNOR
S-nitrosoglutathione reductase
GSD
geometric standard deviation
GSH
glutathione
HCHO
formaldehyde
HCL
hairy cell leukemia
HDM
house dust mite
HEC
human equivalent concentrations
HERO
Health and Environmental Research Online
HHRA
Human Health Risk Assessment
HIC
highest ineffective concentrations
HL
Hodgkin lymphoma
hmDNA
hypermethylated DNA
HPA
hyp othalamic-pituitary-adr enal
HPG
hyp othalamic-pituitary-gonadal
HPO
hyp othalamic-pituitary-o varian
HR
hazard ratio
HSC
hematopoietic stem cells
HSPC
hematopoietic stem and progenitor cells
HUVEC
and umbilical vein endothelial cells
IARC
International Agency for Research on Cancer
IB
information bias
ICD
International Classification of Disease
I FN
interferon
i.p.
intraperitoneal
IQR
interquartile range
IRIS
Integrated Risk Information System
ISAAC
International Study of Arthritis and Allergies in Children
IUR
inhalation unit risk
JEM
job-exposure matrix
JNK
Jun N-terminal protein kinase
This document is a draft for review purposes only and does not constitute Agency policy.
xiii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
LEC
lowest effective concentration
LI
labeling index
LH
luteinizing hormone
LHP
lymphohematopoietic
LM
lateral meatus
LOAEL
lowest-observed-adverse-effect level
LOD
limit of detection
LOQ
limit of quantification
LRT
lower respiratory tract
MldG
malondialdehyde-deoxyguanosine
MAP
mitogen activated protein
MEF
mid-expiratory flow
MDA
malondialdehyde
MDS
myelodysplastic syndrome
mRNA
messenger RNA
miRNA
microRNA
ML
myeloid leukemia
MLE
maximum likelihood estimate
MM
multiple myeloma
MMEF
maximum mid-expiratory flow
MN
micronuclei
MOA
mode of action
MOR
mortality odds ratio
MS
mass spectrometry
MSC
Mesenchymal stem cell
NADP
Nicotinamide adenine dinucleotide phosphate [NADPH], reduced form
NALT
nasal-associated lymphoid tissues
NAS
National Academy of Sciences
NASA
National Aeronautics and Space Administration
NATA
National-Scale Air Toxics Assessment
NCEA
National Center for Environmental Assessment
NCHS
National Center for Health Statistics
NCI
National Cancer Institute
NER
nucleotide excision repair
NHL
non-Hodgkin lymphomas
NIOSH
National Institute for Occupational Safety and Health
NK
natural killer
NLMS
National Longitudinal Mortality Study
NMDA
iV-methyl-D-aspartate receptor
NO
nitric oxide
NOAEL
no-observed-adverse-effect level
NOS
Nitric oxide synthase
NOx
nitrogen oxides
NP
nonprotein
NPC
nasopharyngeal cancer
NRC
National Research Council
NREMS
nonrapid eye movement sleep
NTP
National Toxicology Program
OB
olfactory bulb
OE
olfactory epithelium
OHPC
oropharyngeal/hypopharyngeal cancer
OR
odds ratio
ORD
Office of Research and Development
OSB
oriented strand board
osRfC
organ/system reference concentration
This document is a draft for review purposes only and does not constitute Agency policy.
xiv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
OVA
ovalbumin
PA
polypoid adenoma
PBL
peripheral blood lymphocytes
PBPK
physiologically based pharmacokinetic
PECO
Populations, Exposures, Comparisons, Outcomes
PEF
peak expiratory flow
PEFR
peak expiratory flow rate
PG
periglomerular
PMR
proportionate mortality ratio
PND
postnatal day
POD
point of departure
PODadj
point of departure, adjusted
PODhec
point of departure, human equivalent concentration
POE
portal of entry
PTM
posttranslational modification
PPL
prolymphocyte leukemia
ppm
parts per million
RANTES
regulated on activation, normal T-cell expressed and secreted
RE
respiratory epithelium
REMS
rapid eye movement sleep
RC
room control
RfC
reference concentration
RfD
oral reference dose
RIL
recommended indoor limit
ROS
reactive oxygen species
RR
relative risk
SA
spontaneous abortion
SB
selection bias
see
squamous cell carcinoma
SCE
sister chromatid exchange
SCF
stem-cell factor
SD
standard deviation
SE
standard error
SEER
Surveillance, Epidemiology, and End Results
SEM
standard error of the mean
SES
socioeconomic status
SIR
standardized incidence ratio
SMR
standardized mortality ratio
SNC
sinonasal cancer
SNP
single nucleotide polymorphism
SOD
superoxide dismutase
SPES
symptom questionnaire (German translation)
SPIR
Standardized Proportional Incidence Ratio
SPT
skin prick tests
SRR
summary relative risk
SSB
strand breaks
TCL
T cell lymphoma
TE
transitional epithelium
TH
tyrosine hydroxylase
THF
tetrahydrofolate
TLV
threshold limit value
TNF
tumor necrosis factor
TPA
12 0 tetradecanoylphorbol-13-acetate
TRP
transient receptor potential (channel)
TSCE
two-stage clonal expansion
This document is a draft for review purposes only and does not constitute Agency policy.
xv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
TSFE
time since first exposure
TSLP
thymic stromal lymphopoietin
TTP
time-to-pregnancy
TWA
time-weighted average
UCL
upper confidence limit
UCOD
underlying cause of death
UDS
unscheduled DNA synthesis
UF
uncertainty factor
UFa
uncertainty factor, interspecies
UFc
uncertainty factor, composite
UFd
uncertainty factor, database
UFFI
urea foam insulation
UFh
uncertainty factor, intraspecies
UFl
uncertainty factor, LOAEL-to-NOAEL
UFs
uncertainty factor, subchronic-to-chronic
UFFI
urea formaldehyde foam insulation
ULLI
unit length labeling index
URT
upper respiratory tract
U.S.
United States of America
UV
ultraviolet
VAS
visual analogue scale
VC
vital capacity
VOC
volatile organic compound
WBC
white blood cell
XRCC
X-ray repair cross-complementing
This document is a draft for review purposes only and does not constitute Agency policy.
xvi DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
AUTHORS | CONTRIBUTORS | REVIEWERS
Assessment Team
Barbara Glenn (Chemical Manager)
Andrew Kraft (Chemical Manager)
Thomas F. Bateson
Susan Makris
Deborah Segal
Ravi Subramaniam
Suryanarayana Vulimiri
Glinda Cooper
Jason Fritz
Jennifer Jinot
John Whalan
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
Separated from U.S. EPA
U.S. EPA/Region 8
Retired from U.S. EPA
Retired from U.S. EPA
Other Contributors
James Avery
Jane Caldwell
Brian Pachkowski
Keith Salazar
Lynn Adams
Jacqueline Moya
Amanda Persad
Linda Phillips
Lynn Flowers
Vincent Cogliano
David Bussard
Karen Hogan
Alan Sasso
Bob Sonawane
Emma Lavoie
Retired from U.S. EPA
Retired from U.S. EPA
N.J. Department of Environmental Protection
U.S. EPA/OCSPP/OPPT
Separated from U.S. EPA
Retired from U.S. EPA
U.S. EPA/ORD/CPHEA
Retired from U.S. EPA
U.S. EPA/ORD/OSAPE
Separated from U.S. EPA
Retired from U.S. EPA
Retired from U.S. EPA
U.S. EPA/ORD/CPHEA
Retired from U.S. EPA
U.S. EPA/ORD/OSAPE
Production Team
Maureen Johnson
Ryan Jones
Dahnish Shams
Vicki Soto
Samuel Thacker
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
U.S. EPA/ORD/CPHEA
Executive Direction
Wayne E. Cascio
Kay Holt
Samantha Jones
Kristina Thayer
ORD/CPHEA Center Director
ORD/CPHEA Deputy Center Director
ORD/CPHEA Associate Director
ORD/CPHEA/CPAD Division Director
This document is a draft for review purposes only and does not constitute Agency policy.
xvii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Internal Reviewers
This assessment was provided for review to scientists in EPA's Program and Region Offices.
Comments were submitted by:
Office of the Administrator/Office of Children's Health
Protection
Office of Air and Radiation/Office of Air Quality Planning and
Standards
Office of Air and Radiation/Office of Transportation and Air
Quality
Office of Chemical Safety and Pollution Prevention/Office of
Pesticide Programs
Office of Chemical Safety and Pollution Prevention/Office of
Pollution Prevention and Toxics
Office of Land and Emergency Management
Region 2, New York
Region 4, Atlanta
External Reviewers
1 This assessment was provided for review to other federal agencies and Executive Offices of the
2 President Comments were submitted by:
Department of Defense
Department of Health and Human Services/Agency for Toxic Substances and Disease Registry
Department of Health and Human Services/National Institute of Environmental Health Sciences
Department of Health and Human Services/National Institute for Occupational Safety and Health
Executive Office of the President/Office of Management and Budget
Small Business Administration Office of Advocacy
3 This assessment was released for public comment on [month] [day], [year] and comments were due
4 on [month] [day], [year]. The public comments are available on the IRIS website. A summary and
5 EPA's disposition of the comments from the public is included in Appendix [X] and is also available
6 on the IRIS website. Comments were received from the following entities:
7 This assessment was peer reviewed by independent expert scientists external to EPA (specify NAS
8 panel) and a peer-review meeting was held on [month] [day], [year]. The external peer-review
9 comments are available on the IRIS website. A summary and EPA's disposition of the comments
10 received from the independent external peer reviewers and from the public is included in
11 Appendix [X] and is also available on the IRIS website.
COMPANY NAME
COMPANY NAME
Location
Location
NAME
NAME
NAME
NAME
Affiliation, Location
Affiliation, Location
Affiliation, Location
Affiliation, Location
This document is a draft for review purposes only and does not constitute Agency policy.
xvm
DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
PREFACE ON ASSESSMENT METHODS AND
ORGANIZATION
This Preface presents information to orient the reader to the assessment, including
background information on the development of the Toxicological Review of Formaldehyde—
Inhalation and a description of the focus and underlying framework for this assessment. The
evaluation of formaldehyde's toxicity was informed by what is known about the toxicokinetics of
inhaled formaldehyde (see Section 1.1.3 and Appendix A.2), and this knowledge is reflected in the
organization of the Hazard Identification section. This Preface summarizes the approaches and
methods for the identification of the literature and evaluation of study methods, syntheses of
results for specific health hazard categories within streams of evidence, and integration of the
evidence across human, experimental animal, and mechanistic studies. Finally, the approach used
to select studies and their data for deriving quantitative (dose-response) values is described.
BACKGROUND INFORMATION
The Toxicological Review was prepared under the auspices of the U.S. Environmental
Protection Agency's (EPA's) Integrated Risk Information System (IRIS) program. Assessment
development was based on EPA guidelines as well as standard IRIS procedures (U.S. EPA. 2020)
that were reviewed by the National Academy of Sciences, Engineering and Medicine (NASEM1)
fNASEM. 20211. In 1990 and 1991, an oral reference dose (RfD) (reference value for ingested
formaldehyde) and an inhalation unit risk (IUR) value for cancer, respectively, were developed for
formaldehyde. A previous draft of the inhalation assessment was developed between 1998 and
2010. That document was reviewed by an external peer-review panel convened by the National
Research Council (NRC) between June 2010 and April 2011 (NRC. 2011). The newly developed,
current assessment addresses the comments from the NRC panel on that prior draft (see
Appendix D).
For additional information about this assessment or for general questions regarding IRIS,
please contact EPA's IRIS Hotline at 202-566-1676 (phone), 202-566-1749 (fax), or
hotline.iris@epa.gov.
1 This Toxicological Review and related assessment documents refer to NASEM as well as the National Academy
of Sciences (NAS) and National Research Council (NRC). These names apply to the same organization during
different timeframes (generally with different panel members on the different review panels).
This document is a draft for review purposes only and does not constitute Agency policy.
xix DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
GENERAL ASSESSMENT ORGANIZATION
The Toxicological Review critically reviews the publicly available studies relevant to human
health hazards that may result from formaldehyde inhalation and describes the level of confidence
in the supporting evidence. When there was sufficient confidence in the evidence supporting a
hazard and appropriate studies and data were available, toxicity values were derived using either
analyses of dose-response or selected no-observed-adverse-effect or lowest-observed-adverse-
effect levels (NOAELs or LOAELs). Although this review focused on exposure through inhalation,
general population exposure to formaldehyde can occur via inhalation, ingestion, and dermal
contact
The Toxicological Review is organized into the following sections: Introduction (consisting
of a Preface and an Executive Summary); Hazard Identification; and Dose-Response Analysis.
Supplemental Information to the Toxicological Review is provided in a separate document,
Supplemental Information to the Toxicological Review of Formaldehyde—Inhalation, containing
appendices that support hazard identification and dose-response evaluation. The appendices
include a description of the chemical properties and uses of formaldehyde; information specifically
addressing exposure, toxicokinetics, and genotoxicity; supporting information for health hazard
conclusions in the Toxicological Review (i.e., literature search strategies and results for each health
hazard; conclusions of the evaluation of study methodology; additional analyses); dose-response
modeling; a list of previous legislation and assessments by other agencies; and responses to
external peer-review comments on a prior draft IRIS assessment In addition, an abridged version
of the main assessment conclusions and underlying analyses is provided in a third document, the
Assessment Overview. Additional documents produced during assessment development are
available on the IRIS website fhttp: //www.epa.gov/iris].
The NRC recommendations on the 2010 draft IRIS assessment fNRC. 20111 were
substantive and prompted development of a new (from scratch) assessment using a framework for
evidence identification, evaluation, and integration to provide a more systematic and transparent
process. As a result, different decisions were made, some as a response to the comments received
and others as part of the systematic approach to evaluating the available evidence.
For the purposes of this assessment, potential human health hazards from formaldehyde
exposure were identified and evaluated. These include sensory irritation; decreased pulmonary
function; respiratory tract pathology; immune-mediated conditions, focusing on allergies and
asthma; nervous system effects; reproductive and developmental toxicity; and carcinogenicity.
These health outcomes were identified based on previous reviews of formaldehyde toxicity and
health assessments by other agencies, including the International Agency for Research on Cancer
(IARC), Agency for Toxic Substances and Disease Registry (ATSDR), and the National Toxicology
Program (NTP) (NTP. 2014: IARC. 2012: ATSDR. 2010.1999). For each health hazard, the literature
regarding specific health effects was synthesized within each of the human, animal, and mechanistic
streams of evidence and then integrated across the streams of evidence. The evidence integration
This document is a draft for review purposes only and does not constitute Agency policy.
xx DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
includes a narrative summary of the key evidence and a corresponding level of evidence judgment
(i.e., evidence demonstrates, evidence indicates [likely], evidence suggests, or evidence
inadequate2) as to whether formaldehyde inhalation exposure may pose a human hazard for
specific types of cancer or individual noncancer health effects, given appropriate exposure
circumstances. The assessment provides evidence integration judgments for each unit of analysis
that can be reasonably supported by the available health effect-specific evidence base, so that a
given health hazard may have a single judgment or multiple judgments at more granular outcome
groupings. The evidence integration for cancer concludes with a descriptor summarizing the
weight of evidence for cancer according to EPA's cancer guidelines fU.S. EPA. 2005al For each
credible hazard identified (in this assessment, judgments of evidence demonstrates or evidence
indicates [likely]), the "appropriate exposure circumstances" alluded to during hazard
identification in Section 1 are more fully evaluated and defined through dose-response analysis in
Section 2 (including, depending on the evidence available, the derivation of toxicity values).
Based on the current understanding of the toxicokinetics of formaldehyde inhalation
exposure (see Appendix A.2), several practical working assumptions were applied to this
assessment. Although some uncertainties remain, the organization and analyses in the assessment
assume that inhaled formaldehyde is not distributed to an appreciable extent beyond the
respiratory tract to distal tissues; thus, it is assumed that inhaled formaldehyde acts via a pathway
different from a direct interaction with tissues distal to the portal of entry (POE) to elicit observed
systemic effects. Similarly, it is assumed that formaldehyde does not cause appreciable changes in
normal metabolic processes associated with formaldehyde in distal tissues. Thus, studies
examining potential associations between levels of formaldehyde or formaldehyde byproducts in
tissues distal to the POE (e.g., formate in blood or urine, brain formaldehyde levels) and health
outcomes are not considered relevant here to interpreting the human health hazards of inhaled
formaldehyde.
The Toxicological Review includes an inhalation reference concentration (RfC) value for
lifetime exposure. The inhalation RfC (expressed in units of [ig of substance/m3 air) is defined as an
estimate (with uncertainty spanning perhaps an order of magnitude) of a continuous daily
exposure of formaldehyde to the human population (including sensitive subgroups) that is likely to
be without an appreciable risk of deleterious effects during a lifetime. A carcinogenicity assessment
was also performed, including derivation of an inhalation unit risk value (IUR), which is an upper-
bound excess lifetime cancer risk estimated to result from continuous exposure to an agent at a
concentration of 1 |ig/m:i in air. In addition, organ/system-specific RfCs (osRfCs) were derived for
the various noncancer health endpoints, when supported by the available evidence. These may be
useful when considering cumulative risk scenarios. Multiple candidate RfCs (cRfCs) were
2 These level of evidence judgments and their implications are described in detail in the IRIS Handbook fU.S.
EPA, 20201. Note that none of the health effects evaluated in this assessment approached the level of evidence
needed to support a judgment of strong evidence supports no effect, so this level is not discussed.
This document is a draft for review purposes only and does not constitute Agency policy.
xxi DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Toxicological Review of Formaldehyde—Inhalation
sometimes compared before choosing a representative osRfC. An osRfC was typically selected from
cRfCs based on use of higher confidence studies, and higher confidence in the cRfC derivation
(including point-of-departure [POD] selection). Where relevant, mechanistic understanding
regarding the development of specific health effects (e.g., temporal progression, potential
thresholds in dose-response), as well as knowledge of susceptibility, was used to inform
approaches to derive points of departure (PODs), uncertainty factors, or confidence levels for the
quantitative estimates (e.g., osRfCs, RfC, IUR). Where possible, the assessment attempts to describe
the level of response observed across different exposure levels within the range of the data, and to
discuss transparently the uncertainties and assumptions when deriving toxicity value estimates
(e.g., cRfCs, IUR). In addition, as the temporal window of exposure relevant to particular outcomes
may vary, the window of exposure expected to be most relevant to each toxicity value is discussed
in Section 2, Dose-Response Analysis, when applicable.
A confidence level of high, medium, or low was assigned to each osRfC and the overall RfC
based on the reliability of the associated POD and cRfC calculation(s). Confidence in the POD and
cRfC calculation(s) included considerations of the quality, timing, and variability of the exposure
estimates in an epidemiological study or the exposure protocols in an animal study. Moreover,
higher confidence was placed in the osRfC when the POD was identified close to the range of the
observed data. Finally, confidence in the coverage and quality of the database of studies that
informed the hazard conclusion for that organ/system was assigned. The evidence base for
different health outcomes varies in size, coverage of critical endpoints, and quality of the studies;
this confidence level reflects database completeness for each of the organ systems.
SUMMARY OF ASSESSMENT METHODS AND APPROACHES
The approaches implemented throughout different stages of this assessment can be
grouped into those used to (1) identify and evaluate individual studies; (2) synthesize and integrate
the evidence, including interpreting the support for particular human health effects across different
streams of evidence (i.e., human, animal, and mechanistic studies) and developing summary
conclusions; and (3) select and analyze studies and data to derive quantitative (dose-response)
values. The process for hazard identification, which involves hazard-specific literature searches,
outcome/endpoint-specific evaluation of study methods, synthesis of information within each
streams of evidence, and integration across streams of evidence, is displayed in Figure I. The
process involves a successive focusing on the more informative outcomes/endpoints within each
hazard domain and the most methodologically sound studies.
This document is a draft for review purposes only and does not constitute Agency policy.
xxii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Literature Searches (Hazard-specific)
1
Reference retrieval
Reference lists
Inclusion/ exclusion criteria
(based on PECO)
H
Reference screening by hazard domain
o Excluded references grouped by PECO category
o Included references grouped by lines of
evidence (human, animal, mechanistic)
o Literature search diagrams by hazard domain
~
Evaluation of study methods (Outcome/ endpoint specific)
Outcome-specific
evaluation criteria
for health effects
studies in humans
and animals;
informed by ADME
research
*
Study evaluation tables
o By outcome and study
o Study confidence by outcome
Medium
Syntheses of results
Low
Unlnformative
Interpretation of
results from health
effect studies in
humans and animals
(consistency,
magnitude of effect,
dose-response,
coherence, etc.)
Evaluation and
interpretation of
mechanistic evidence
Judgments on health effects
separately for human or animal
studies
based on health effects and biological
plausibility from mechanistic studies
Overall evidence judgments
regarding potential of chemical to
cause health effects in humans, using
judgments on human and animal
evidence and mechanistic inference3
Evidence demonstrates
Evidence indicates (likely)
Evidence suggests
Evidence inadequate
Figure I. Overview of assessment methods for hazard identification.
This figure illustrates the flow of evidence through the assessment, sequentially focusing on the most
useful information, as well as the decision-making processes for arriving at evidence judgments regarding
the potential for noncancer health effects and for developing specific types of cancer. aMechanistic
inference considered during evidence integration included biological plausibility or relevance of animal
study results to humans and identification of susceptible groups. Notes: For this assessment, "compelling
evidence of no effect" was not reached for any of the human or animal evidence evaluations; as such,
criteria for evidence integration when compelling evidence of no effect was present are not discussed in
this assessment. Importantly, hazard identification for carcinogenicity includes an additional step of
assigning a descriptor regarding the potential for formaldehyde to cause cancer (this step is not shown
but is discussed in this section below (see Table IX). Abbreviations: HERO = Health and Environmental
Research Online; PECO = Populations, Exposures, Comparisons, Outcomes; ADME = absorption,
distribution, metabolism, excretion; MOA = mode of action.
1 Literature Search Strategy
2 The literature search strategy used to identify primary research pertaining to formaldehyde
3 inhalation was conducted using the databases and approaches listed in Table I. A separate search
4 strategy was developed for each health hazard considered in the assessment These strategies are
5 described in detail in Appendix A.5, with PECO criteria, and literature flow diagrams depicting the
6 systematic search and sorting process. Generally, health outcomes and search terms were selected
This document is a draft for review purposes only and does not constitute Agency policy.
xxiii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
after reviewing the draft Toxicological Review for Formaldehyde (2010) and other relevant health
assessments or reviews of formaldehyde toxicity. A series of comprehensive literature searches
was conducted beginning in 2012 and updated annually through 2016, after which the completed
2017 Step 1 draft IRIS formaldehyde-inhalation assessment was suspended at the request of senior
EPA management. When the IRIS assessment was unsuspended in March 2021
(http: //www.epa. gov/sites/production/files/2 0 21 -
03/documents/iris_program_outlook_mar2021.pdf), systematic evidence mapping (SEM) methods
f Keshava et al.. 2020: Wolffe etal.. 20201 were employed to survey the newer literature and
expedite updating the unsuspended draft (see Appendix F for the methods and results of the
formaldehyde SEM update). In these searches, electronic database queries were supplemented
using various approaches to identify additional papers, including review of reference lists in
identified publications and national-level health assessments. Several meta-analyses of
formaldehyde effects, with different conclusions, have been published for a few health outcomes.
Reviews and meta-analyses were reviewed to identify relevant publications and background
information.
Table I. General approach to literature search strategies
Databases3
Health hazard searches'1
Web of Science
ToxNet
PubMed
TSCATS2
(formaldehyde, formalin, paraformaldehyde, OR CASN 50-00-0) AND:
Sensory Irritation0
Pulmonary Function0
Immune-Mediated Conditions, focusing on Allergies and Asthma
Respiratory Tract Pathology
Developmental and Reproductive Toxicity
Nervous System Effects
Cancer
Inflammation and Immune Effects (mechanistic information^
aPubMed: http://www.ncbi.nlm.nih.gov/pubmed/, Web of Science:
http://apps.webofknowledge.com/WOS GeneralSearch input.do?product=WOS&search mode=. ToxNet:
toxicology information previously contained in ToxNet were integrated into other NLM products (see
https://www.nlm.nih.gov/toxnet/index.html for where to access).
Specific parameters and keywords for each hazard-specific database search strategy are included in Appendix A.5.
CA systematic search strategy was not applied to the database of animal studies on this health outcome. Sensory
irritation in animals is a well-described phenomenon. For pulmonary function, there was an extensive set of
research studies on humans, and therefore, the few studies on this endpoint in animals were not reviewed.
dThis separate, systematic literature search was performed to augment the analyses of mechanisms relevant to
other health effect-specific searches.
The citations returned from these literature searches were screened using health outcome-
specific PECO criteria (see Appendix A.5). In general, although studies of other routes of exposure
might inform the mechanistic understanding of potential health hazards, the formaldehyde
literature database is extensive and the toxicokinetics following inhalation exposure is expected to
This document is a draft for review purposes only and does not constitute Agency policy.
xxiv DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
differ significantly from those observed after exposure via other routes; thus, the evaluations of
potential health effects and mechanistic information focused on inhalation exposure studies (with
the exception of genotoxicity). Publications were typically excluded if they contained no
information about formaldehyde exposure or were descriptions of analytic methods using
formaldehyde. Ambient levels of formaldehyde in outdoor air are significantly lower than those
measured in the indoor air of workplaces or residences, and the exposure range was narrow in
many epidemiological studies of ambient exposure (<0.005 mg/m3), limiting their sensitivity to find
any associations with health outcomes even if they existed. In addition, the potential for exposure
misclassification for estimates of individual exposure using mean formaldehyde concentrations
from central outside monitors is greater than from indoor formaldehyde measurements. Therefore,
the few studies examining health effects in relation to outdoor formaldehyde concentrations were
excluded. Other exclusions were based on specific criteria relating to each health hazard, which are
summarized in each of the respective health hazard sections in Appendix A.5.
In addition to the health effects listed in Table I, relevant literature on additional topics
(e.g., formaldehyde exposure, toxicokinetics, mechanisms of carcinogenesis) was identified. While a
thorough effort was made to identify all relevant studies for each of these topic areas (see
Appendix A for details), these discussions do not include specific tracking of the selection of
individual studies (e.g., based on PECO criteria). The references identified and selected through the
literature search process, including bibliographic information and abstracts, can be found on the
formaldehyde page of the Health and Environmental Research Online (HERO) website3:
http://hero.epa.gov/hero/index.cfm/project/page/project id/4051.
For the literature update from 2016-2021 using SEM approaches (overlapping with the
searches used for the 2017 draft), while the aforementioned description of the search and
screening process was largely identical (see Appendix F) a few differences are important to note.
Most notably, after screening the studies for PECO relevance, only those studies meeting the PECO
criteria and judged as likely to have a potential impact on the conclusions or toxicity values
described in the suspended 2017 draft are synthesized in this assessment Studies meeting PECO
criteria that were judged to have no impact on those conclusions or toxicity values are summarized
in Appendix F, along with explanations for these decisions. These latter studies are not further
discussed or synthesized in the assessment
3HER0 (Health and Environmental Research On-line) is a database of scientific studies and other references
used to develop EPA's assessments that support critical decision-making and is aimed at understanding the
health and environmental effects of pollutants and chemicals. It is developed and managed in EPA's Office of
Research and Development (0RD) by the Center for Public Health and Environmental Assessment. The
database includes more than 3,000,000 scientific articles and associated data from the peer-reviewed
literature. New studies are added continuously to HERO.
This document is a draft for review purposes only and does not constitute Agency policy.
xxv DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Study Evaluations
All human and experimental animal health effect studies identified in the search and
screening processes described above, without regard to study results, were considered for use in
assessing the evidence for health effects associated with inhalation exposure to formaldehyde. This
full body of evidence is discussed and evaluated in Section 1, Hazard Identification.
Study methods were evaluated to assign a level of confidence in the results of the study with
respect to the hazard question under consideration. The study confidence levels were high,
medium, and low confidence, and not informative, and are presented as italicized text in the body of
the assessment These evaluations were performed on a health outcome-specific basis, rather than
a study-specific basis; thus, a single study was sometimes evaluated multiple times for different
endpoints, sometimes involving slightly different considerations. High confidence studies generally
had no notable methodological limitations for an outcome, while medium confidence studies were
considered well conducted but had specific issues that might introduce a minor amount of
uncertainty about attribution of the results solely to formaldehyde exposure on the health outcome
in question. Methodological limitations of low confidence studies are considered to be significant,
but the outcome-specific results might still be of limited use (e.g., as support for observations from
other studies; to identify potential data gaps). The evaluations for studies identified as not
informative were documented (see Appendix A.5), but these data are not discussed in the
Toxicological Review. In general, if a study or individual analysis (e.g., when multiple health
outcomes or cohorts were assessed) was judged to have multiple severe limitations, or if reporting
deficiencies precluded the ability to conduct an evaluation, the experiment was concluded to be not
informative. When potential limitations were identified, the evaluations considered the anticipated
direction (i.e., bias toward or away from the null) and magnitude of the impact of the limitation(s)
on the study results (when possible). Emphasis was placed on discerning limitations that would be
expected to produce a substantive change in the results.
The evaluations focused on potential sources of bias or other limitations (including reduced
sensitivity) that can affect the validity or interpretation of a study's results. Thus, the confidence
conclusions for individual studies reflect an interpretation of the reliability of the study results for
answering each particular hazard question. The general procedure involved evaluating specific
methodological features (see below), although the categories differed somewhat between
observational epidemiological, animal toxicological, and human-controlled exposure studies. The
appendices contain summary evaluation tables developed for studies in each health hazard domain,
which provide both relevant study characteristics and the conclusions of the evaluations.
Evaluation conclusions also are included in the tables summarizing the evidence for each health
effect in the Toxicological Review. In addition to the evaluations of the individual health effect
studies, systematic evaluations of individual mechanistic studies were conducted in relation to
several important health domains when this information could contribute to judgments about the
human and animal evidence or hazard conclusions, specifically: biomarkers of genotoxicity in
This document is a draft for review purposes only and does not constitute Agency policy.
xxvi DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
exposed humans, mechanistic data related to potential respiratory health effects, and mechanistic
data related to potential nervous system effects (see Appendix A.4.6, A.5.6, and A.5.7, respectively).
Individual study evaluations for literature on exposure, toxicokinetics and other mechanistic data
were not systematically conducted and documented.
In some situations, in which key study details or results were not presented, the study
author(s) were contacted to obtain this information. Any additional study details obtained from the
authors are noted in the evaluation summary tables and evidence tables.
Evaluation of Observational Epidemiology Studies
Classification scheme
For each type of health outcome examined, the epidemiological studies were evaluated for
each of the categories of information relevant to internal validity (bias) that could lead to an under-
or overestimate of risk and to other features that could affect the interpretation of the results or
limit the ability to detect a true association (e.g., narrow exposure range). The categories used for
the epidemiological studies included population selection, exposure (measurement and
levels/range), outcome ascertainment, consideration of confounding, and analytic approach. The
potential for selection bias, information bias (relating to exposure and to outcome), and
confounding was evaluated. A pictorial summary of the conclusions from the outcome-specific
evaluation process was created (see Figure II). Studies that evaluated more than one outcome
might be categorized differently for each outcome. The classification of a study could also vary
among different analytical groups within a study (e.g., studies of children and adults, with separate
analyses for each group), depending on the information presented for the different analyses.
This document is a draft for review purposes only and does not constitute Agency policy.
xxvii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
SE IE Cf Oth
Overall
Confidence
High
SB
l& Cf Oth
Overall
Confidence
Medium
4
I-
SB
IB
Cf
Oth
Overall
Confidence
LOW
1
¦
¦
¦
1
SE IE Cf
Oth
Overall
Confidence
1 w
Low
Figure II. Summary depictions of evaluation of epidemiology studies.
The extent of column shading reflects the degree of limitation. Different colors are intended to visually
distinguish the columns and have no other meaning. The direction of anticipated bias is indicated by
arrows: "4," for overall confidence indicates anticipated impact would be likely to be toward the null
(i.e., attenuated effect estimate); '"Y" for overall confidence indicates anticipated impact would be likely
to be away from the null (i.e., spurious or inflated effect estimate). Panel A: High confidence study; Panel
B: Medium confidence study with likely attenuated effect estimate; Panel C: Two possible examples for a
low confidence study. Color blocks that straddle the midline indicate that the direction of bias is unknown
or not predictable. The depiction on the right indicates that selection bias (SB) likely resulted in an
overestimate of the effect estimate, indicated by the colored block above the midline. Abbreviations:
SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or analysis.
1 The synthesis of evidence (see next section) focuses on the medium and high confidence
2 studies, if available, taking into account differences in populations and settings (e.g., children and
3 adults; occupational, residential, or in schools), exposure levels, and other aspects of the studies.
4 Formaldehyde exposure considerations specific to observational epidemiological studies
5 All residential or school-based studies with measures of formaldehyde exposure were
6 included in the hazard identification evaluation; because the database of studies with direct
7 measurements is relatively large, residential studies with indirect measures of formaldehyde
8 exposure (e.g., based on age of building or presence of plywood) were not included. Most of the
9 included studies attempted to estimate average formaldehyde levels using area samples placed in
This document is a draft for review purposes only and does not constitute Agency policy.
xxviii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
one or more locations; measurement periods ranged from 30 minutes to 2 weeks. A few studies
included more than one sampling period (e.g., sampling on multiple days in different seasons over
the course of a year). Studies in adults and in children indicate that area-based (e.g., residential or
school) samples are highly correlated with personal samples fLazenbv et al.. 2012: Gustafson etal..
20051: therefore, the use of measures based on residential (e.g., bedroom) samples rather than
personal samples was not considered to be a limitation when evaluating a study.
There was also variation in the exposure measurements used within occupational settings.
For hazard identification, an accurate characterization of "high" versus "low" exposure or "exposed"
versus "nonexposed" may be able to provide a sufficient contrast to examine associations, even if
there is considerable heterogeneity within the high exposure group. Exposure assessments in
occupational studies involved one or more area samples in specific task areas, personal samples, or
a combination of both. Sampling periods ranged from less than 1 hour to an entire work shift over
1 or more days. Concentrations were reported as an average of all samples for a particular location
or as a time-weighted average (TWA) over the sampling period. Generally, a TWA concentration
from a full-shift measurement using personal sampling was preferred as a more precise estimate of
average exposure. Other occupational studies that used a formaldehyde-specific exposure
definition or semiquantitative measure (e.g., duration, number of embalmings) also were included,
although they were concluded to be limited to some extent by exposure misclassification. Studies
of certain occupational groups with considerable exposure to formaldehyde (e.g., embalmers,
pathologists, wood or garment workers) were included as proxies for formaldehyde exposure
because formaldehyde is the predominant chemical exposure in these jobs and a large contrast is
expected between exposed and unexposed groups.
Evaluation of controlled exposure studies in humans
A process incorporating aspects of the evaluation approaches used for epidemiological
studies and experimental animal studies (see below) was used to evaluate controlled exposure
studies in humans. The evaluation categories included exposure generation, outcome classification,
consideration of possible bias (i.e., randomization and blinding), consideration of confounding
(i.e., adequacy of randomization), and details of analysis and presentation of results. A study was
judged to be low confidence if the exposure generation method resulted in exposure to substances
other than formaldehyde (e.g., emissions from pressed wood products), allocation to the order of
exposure categories was not random, or subjects were not blinded to their exposure order.
Evaluation of experimental studies
Classification scheme
Toxicological studies differ systematically from observational epidemiological studies
because the former seek to control both the exposure and nonexposure conditions of an
experiment This leads to some differences in approach and interpretation. In general, however,
This document is a draft for review purposes only and does not constitute Agency policy.
xxix DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
toxicological study evaluations considered similar categories to the epidemiological studies. The
categories were based on the design of a toxicological study, including test animals, experimental
design (e.g., duration of exposure, timing of endpoint evaluations, allocation procedures), exposure
conduct, endpoint evaluation procedures, and data presentation and analysis. The specifics of the
considerations applied were different for each type of health outcome examined (see
Appendix A.5).
As the expectation is that experimental studies should attempt to control all variables, any
study limitation interpreted as capable of influencing the data was considered to have negatively
affected the quality (e.g., validity, accuracy) of the results. Thus, these "confounding factors" differ
substantially from what would be deemed a potential "confounder" in epidemiological studies.
Formaldehyde exposure considerations specific to controlled exposure fanimal or human! studies
Typical human exposures to formaldehyde can be complex and difficult to translate to
experimental systems. Experimental exposure to formaldehyde by inhalation is typically achieved
through volatilization of formalin or depolymerization of paraformaldehyde. Methanol, present in
aqueous formaldehyde solutions to inhibit polymerization, is a potential confounder of associations
between observed health outcomes and formaldehyde exposure via formalin. As experimental
studies, including controlled exposure studies in either humans or animals should aim to control all
variables other than the exposure or manipulations of interest; coexposure to methanol in these
studies introduces uncertainty that the effects were caused by formaldehyde alone. Inhaled
methanol could affect health endpoints or introduce quantitative uncertainty. An example of the
former would be if methanol were distributed to different locations than inhaled formaldehyde,
where it could either directly cause effects or, theoretically, be metabolized to formaldehyde and
cause effects. An example of the latter would be that, because methanol is metabolized to
formaldehyde in vivo, substantial coexposure to methanol could result in differences in tissue-
specific formaldehyde levels at identical external formaldehyde exposure levels when different test
articles are used. This limitation typically introduces a bias toward an effect and is of particular
concern in studies observing systemic effects after exposure. Thus, the test article used to generate
the formaldehyde atmosphere in experimental studies was critically evaluated (see Appendix A.5
for details), including consideration of whether a methanol-only control group was used.4 Although
4While one study used a sprayer in a heated vessel to generate formaldehyde from a formalin solution
containing a known concentration of methanol (Kamata et al.. 19971. presumably resulting in the release of
formaldehyde and methanol in proportions that would be conserved from liquid to gas (i.e., allowing air
methanol levels to be relatively accurately estimated based on air formaldehyde levels), the remaining
formalin studies generally evaporated formalin from solution. Notably, the liquid: air partitioning of methanol
and formaldehyde is influenced by the proportions of these agents in aqueous solutions (Albert et al.. 2000).
Thus, as chamber methanol levels were not analytically measured in the other identified studies, a methanol
control group may not eliminate uncertainty. Unfortunately, a calculation for estimating methanol levels
released (e.g., by evaporation) from formalin solutions at different levels of inhaled formaldehyde was not
identified.
This document is a draft for review purposes only and does not constitute Agency policy.
xxx DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
this evaluation was applied to all experimental systems, conclusions about the level of uncertainty
introduced by this coexposure varied by health outcome, with a far greater level of concern for
potential impacts on nonrespiratory health effects (see Section 1.3, Nervous System Effects,
developmental and reproductive system effects, and lymphohematopoietic (LHP) cancers), as
compared to respiratory health effects (see Section 1.2). This disproportionate level of concern is
primarily based on two factors: (1) as compared to formaldehyde, which does not appear to be
distributed to distal sites in appreciable amounts, inhaled methanol would be readily transported
beyond the portal of entry (POE) and could elicit direct effects at distal target tissues, and
(2) certain systemic effects evaluated in this assessment (i.e., reproductive and developmental
toxicity, nervous system effects) are health outcomes known to be a target of methanol toxicity,
while other health outcomes, although generally less well studied, have not been clearly associated
with methanol exposure fU.S. EPA. 20131. These issues are discussed further in each major
endpoint discussion in Sections 1.2 and 1.3.
For certain health outcomes, the irritant and odorant nature of formaldehyde gas and the
inescapable nature of these exposures (animals cannot terminate exposure at irritating levels), can
complicate interpretations of causality. In addition, reflex bradypnea is an irritant response that
exists in rodents, typically at formaldehyde concentrations exceeding 1 mg/m3 (see Section 1.1.3),
but not humans and can cause large variations between the administered and internal exposures.
Although the understanding of irritation-related responses, including reflex bradypnea in rodents,
is incomplete (e.g., responses following repeated and prolonged exposure are not well studied;
see Appendix A.3), it is generally assumed that irritation- and odorant-specific changes are either
short lived or markedly reduced shortly after formaldehyde exposure is removed. In light of these
considerations, care was taken to consider in detail the specifics of the study protocols related to
formaldehyde exposure (e.g., determining whether a sufficient duration was allotted between
exposure and testing, evaluating whether the exposure levels tested were capable of introducing
variables such as reflex bradypnea) for certain health outcomes.
Overall, as in observational studies in humans, considerations related to the quality of the
exposure paradigms used in experimental studies typically had the strongest influence on study
confidence determinations.
Evaluation of mechanistic studies
For the datasets described previously, evaluations of individual mechanistic studies
involving formaldehyde inhalation in experimental animals or in vitro models of gaseous
formaldehyde exposure considered the same general features evaluated for more apical measures
of toxicity (i.e., evaluations of exposure quality and study design were emphasized). The specific
criteria were simplified, however, to accommodate the increased heterogeneity of the available
mechanistic studies, as compared to the data available for apical measures of toxicity. Similarly,
study evaluations of individual mechanistic studies involving exposed humans emphasized
consideration of exposure assessment, study design, outcome ascertainment, and comparison
This document is a draft for review purposes only and does not constitute Agency policy.
xxxi DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
groups for potential sources of bias and their potential impact For the mechanistic studies related
to potential noncancer respiratory effects, given the large number of studies identified, individual
experiments were characterized as high or medium confidence, low confidence, or not informative.
These evaluations emphasized exposure-related considerations and were designed to identify the
mechanistic data most likely to be associated with constant, chronic inhalation exposure to
formaldehyde (see Appendix A.5.6 for additional details). As these individual study evaluations
were less endpoint specific than the evaluations of the individual health effect-specific studies,
these evaluations were generally less rigorous. Subsequently, groupings of studies or related
endpoints were evaluated to assess the strength of the evidence for different "mechanistic events"
as robust, moderate, slight, or indeterminate. Likewise, potential associations between mechanistic
events were judged based on the tissue(s)/region(s) assessed and known biological roles within
those tissues for the identified mechanistic events. The criteria and presentation of decisions for
the strength of the mechanistic evidence relating to potential respiratory health effects are
illustrated in Table II. For studies of ge no toxicity biomarkers in exposed humans, conclusions
about bias and sensitivity were drawn using the same approach as for other epidemiological
studies.
Table II. Criteria and presentation of strength of the evidence for each
mechanistic event and for potential associations between events relating to
potential respiratory health effects
Evidence
judgment3
Mechanistic events
Associations between mechanistic
events
Criteria for conclusions
Presentation15
Criteria for conclusions
Presentation15
Strongest
Robust
Direct evidence supporting an
effect in multiple, consistent
high or medium confidence
studies'5
o
Emphasized in
Text
Formaldehyde-specific data
demonstrate a linkage
(i.e., inhibition of
mechanistic event "A"
prevents or reduces the
occurrence of event "B";
events "A" and "B" are
linked by concentration,
location, or temporality)
Moderate
Direct or indirect (e.g., genetic
changes) evidence supporting an
effect in at least one high or
medium confidence study, with
supporting evidence
(e.g., consistent changes
suggesting an effect in low
confidence studies)15
O
Emphasized in
Text
• An association between
events "A" and "B" is
known based on
established (basic) biology
• An association has been
demonstrated for similar
chemicals or effects
->
Slight
• Evidence supporting an effect
in one hypothesis-generating
fN|
v ~
An association is justifiable,
or even expected, based on
¦¦¦»
This document is a draft for review purposes only and does not constitute Agency policy.
xxxii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Evidence
judgment3
Mechanistic events
Associations between mechanistic
events
Criteria for conclusions
Presentation15
Criteria for conclusions
Presentation15
high or medium confidence
study
• Evidence suggesting an effect
in multiple, reasonably
consistent low confidence
studies
Minimal
Discussion in
Text
underlying biology, but it
has not been well
established (note: events for
which a biological
association appears unlikely
are not linked)
Weakest
Indetermin
-ate
• Evidence suggesting an effect
in one low confidence study
• A set of low confidence studies
with inconsistent results
Not included
in figures; may
be noted in
text
N/A
N/A
• Evidence cannot be
interpreted (no data; no
pattern in results within or
across studies)
• Data suggest no change
Not included
in figures or
synthesis text
N/A
N/A
aFor consistency, the words used to describe the judgments for apical health effect endpoints in human or animal
studies were applied (see subsequent section, Evidence Integration and Confidence Conclusions for Noncancer
and Cancer Health Outcomes), although the criteria herein are less rigorous (i.e., when evaluating sets of studies),
unlike the conclusions for apical health effects.
Supporting evidence and documentation for these decisions is provided in Appendix A.5.6, with only the evidence
on mechanistic changes (irrespective of the results) most informative to the health effect-specific discussions
presented in Sections 1.2.1-1.2.4.
cThe presence of a comparable or stronger set of studies with directly conflicting evidence results in the
identification of the next weaker evidence descriptor (e.g., robust evidence with conflicting data would be
moderate); note that the purpose of this evaluation was not to identify mechanistic events for which there was
robust evidence of no change; however, the plausibility of the pathways (considering evidence for a lack of
changes in expected events) is discussed in later sections.
1 Synthesis of the Available Evidence for Each Health Outcome
2 Sections 1.2 and 1.3 include syntheses of the entire body of evidence for the following
3 health hazard categories: sensory irritation; reduced pulmonary function, respiratory tract
4 pathology, immune-mediated conditions, focusing on allergies and asthma; cancer (respiratory
5 tract cancers, lymphohematopoietic cancers); nervous system effects (motor neuron disease, tests
6 of general motor-related behaviors, neural sensitization, learning or memory, neuropathology);
7 developmental and female reproductive toxicity; and male reproductive toxicity. Health hazard
8 categories were chosen based on prior reviews, as well as the specifics of the available literature.
9 The units of analysis within an overall hazard category for which a hazard conclusion was
10 developed were determined based on biologic considerations (i.e., specific to an organ system and
This document is a draft for review purposes only and does not constitute Agency policy.
xxxiii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
considering the degree to which endpoints are related) and the number of studies that evaluated a
particular outcome. Thus, hazard conclusions were developed for consolidated sets of related
health endpoints within an overall hazard category in some instances (e.g., male reproductive
toxicity).
For each unit of analysis (hazard category, or hazard subgrouping), and depending on the
data available, separate syntheses were developed for each of the three streams of evidence:
namely, human and animal health effect studies and mechanistic studies. These evidence
syntheses, which incorporate the evaluations of the strengths and limitations of the available
studies as well as considerations related to the toxicokinetics of inhaled formaldehyde, provide a
discussion of the information provided by each stream of evidence regarding the potential for
exposure to formaldehyde via inhalation to result in specific health effects. All informative studies
(see above), regardless of the magnitude or direction of results (i.e., whether yielding positive or
null results) were considered in assessing the evidence; however, the focus of the synthesis was on
the high and medium confidence studies, when available. Descriptive information about study
methods and detailed results are generally presented in tabular or graphical displays, with
supportive text The narrative summaries discuss the nature and breadth of the available
literature, highlighting details that contribute to the analysis of the strength of evidence regarding
causality in the next section.
The syntheses of the separate streams of evidence—human health effect studies, animal
health effect studies, and mechanistic studies—involved related considerations that differed due to
the nature of the study designs and applicability of the data (see Table III). Consistency, magnitude
of effects, and dose-response gradients were emphasized in the synthesis of results of
epidemiological and controlled human exposure studies. While the precision of effect estimates
could add to the strength of evidence for a health effect, all of the results were summarized.
Consistency between studies was examined by comparing study results by confidence level, specific
methodological features that contributed to potential bias, exposure setting, and level of exposure.
The primary considerations for synthesizing the results of animal studies were consistency
(e.g., across species and across research groups, with consideration of study confidence), magnitude
and severity of the effects, dose-response, and coherence of findings for related effects. The
information from mechanistic studies in humans or animals relevant to each apical outcome was
synthesized, highlighting information that could inform either biological plausibility, coherence,
susceptibility, relevance to humans or an improved understanding of dose-response. Given the
exposure-related issues specific to formaldehyde and the abundance of data available, the
mechanistic evaluations in this assessment focused almost exclusively on in vivo studies of
inhalation exposures, with rare exception (e.g., evaluation of in vitro genotoxicity studies).
This document is a draft for review purposes only and does not constitute Agency policy.
xxxiv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table III. Information most relevant to describing primary considerations
informing causality during evidence syntheses
Consideration
Description and synthesis methods
Consistency
• Examines the similarity of results (e.g., direction; magnitude) across studies.
When inconsistencies exist, the synthesis considers whether results were "conflicting"
(i.e., unexplained positive and negative results in similarly exposed human populations or
in similar animal models) or "differing" (i.e., mixed results explained by differences
between human populations, animal models, exposure conditions, or study methods)
(U.S. EPA, 2005a) based on analyses of potentially important explanatory factors such as:
• Confidence in studies' results, including study sensitivity (e.g., some study results that
appear to be inconsistent may be explained by potential biases or other attributes that
affect sensitivity, resulting in variations in the degree of confidence accorded to the
study results)
• Exposure, including route (if applicable), levels, duration, etc.
• Populations or species, including consideration of potential susceptible groups or
differences across lifestages at exposure or endpoint assessment
• Toxicokinetic information as an explanation for any observed differences in responses
across route of exposure, other aspects of exposure, species, or lifestages
The interpretation of the consistency of the evidence and the magnitude of the reported
effects will emphasize biological significance as more relevant to the assessment than
statistical significance. Statistical significance (as reported by p-values, etc.) provides no
evidence about effect size or biological significance, and a lack of statistical significance
will not be automatically interpreted as evidence of no effect.
Strength (effect
magnitude) and
precision
• Examines the effect magnitude or relative risk, based on what is known about the
assessed endpoint(s), and considers the precision of the reported results based on
analyses of variability (e.g., confidence intervals; standard error). In some cases, this
may include consideration of the rarity or severity of the findings (in the context of the
health effect being examined).
Syntheses will analyze results both within and across studies and may consider the utility
of combined analyses (e.g., meta-analysis). While larger effect magnitudes and precision
(e.g., p < 0.05) help reduce concerns about chance, bias, or other factors as explanatory,
syntheses should also consider the biological or population-level significance of small
effect sizes. Thus, a lack of statistical significance should not be automatically
interpreted as evidence of no effect.
Biological
gradient/dose-
response
• Examines whether the results (e.g., response magnitude, incidence, severity) change in
a manner consistent with changes in exposure (e.g., level, duration), including
consideration of changes in response after cessation of exposure.
Syntheses will consider relationships both within and across studies, acknowledging that
the dose-response (e.g., shape) can vary depending on other aspects of the experiment,
including the outcome and the toxicokinetics of the chemical. Thus, when dose-response
is lacking or unclear, the synthesis will also consider the potential influence of such
factors on the response pattern.
This document is a draft for review purposes only and does not constitute Agency policy.
xxxv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Consideration
Description and synthesis methods
Coherence
• Examines the extent to which findings are cohesive across different endpoints that are
known/expected to be related to, or dependent on, one another (e.g., based on known
biology of the organ system or disease, or mechanistic understanding such as
toxicokinetic/dynamic understanding of the chemical or related chemicals). In some
instances, additional analyses of mechanistic evidence from research on the chemical
under review or related chemicals that evaluate linkages between endpoints or organ-
specific effects may be needed to interpret the evidence. These analyses may require
additional literature search strategies.
Syntheses will consider potentially related findings, both within and across studies,
particularly when relationships are observed within a cohort or within a narrowly defined
category (e.g., occupation, strain or sex, lifestage of exposure). Syntheses will emphasize
evidence indicative of a progression of effects, such as temporal- or dose-dependent
increases in the severity of the type of endpoint observed.
Mechanistic evidence
related to biological
plausibility
• There are multiple uses for mechanistic information (see 9.2), and this consideration
overlaps with "coherence." This examines the biological support (or lack thereof) for
findings from the human and animal health effect studies and becomes more impactful
on the hazard conclusions when notable uncertainties in the strength of those sets of
studies exist. These analyses can also improve understanding of dose- or duration-
related development of the health effect. In the absence of human or animal evidence
of apical health endpoints, the synthesis of mechanistic information will drive evidence
integration conclusions (when such information is available).
Syntheses can evaluate evidence on precursors, biomarkers, or other molecular or
cellular changes related to the health effect(s) of interest to describe the likelihood that
the observed effects result from exposure. This will be an analysis of existing evidence,
and not simply whether a theoretical pathway can be postulated. This analysis may not
be limited to evidence relevant to the PECO but may also include evaluations of
biological pathways (e.g., for the health effect; established for other, possibly related,
chemicals). The synthesis will consider the sensitivity of the mechanistic changes and the
potential contribution of alternative or previously unidentified mechanisms of toxicity.
Natural experiments
• Specific to epidemiological studies and rarely available, these examine effects in
populations that have experienced well-described, pronounced changes in exposure to
the chemical of interest (e.g., blood lead levels before and after banning lead in
gasoline).
1 Evidence Integration and Integration Judgments for Noncancer and Cancer Health Outcomes
2 For transparency in the sequential decision steps taken to draw overall evidence
3 integration judgments, a two-step, sequential process was used (Figure III). First, judgments
4 regarding the strength of the evidence from the available human and animal studies were made.
5 These judgments incorporated mechanistic evidence (or MOA understanding) in exposed humans
6 and animals, respectively, that informed the biological plausibility and coherence of the available
7 human or animal health effect studies. Second, an overall conclusion(s) was drawn by integrating
8 the animal and human evidence judgments and incorporating inferences regarding the human
This document is a draft for review purposes only and does not constitute Agency policy.
xxxvi DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
relevance of the animal evidence (i.e., based on default assumptions or empirical evidence),
coherence across the human and animal evidence, and susceptibility.
STEP 1: INTEGRATION OF HEALTH EFFECT STEP 2: OVERALL INTEGRATION OF
AND MECHANISTIC EVIDENCE IN EVIDENCE FOR HAZARD ID
HUMANS OR ANIMALS
HUMAN EVIDENCE JUDGMENT
The synthesis of evidence about health
effects and mechanisms from human
studies is combined (integrated) to make a
judgment about health effects in human
studies.
ANIMAL EVIDENCE JUDGMENT
The synthesis of evidence about health
effects and mechanisms from animal
studies is combined (integrated) to make a
judgment about health effects in animal
studies.
Figure III. Process for evidence integration.
Human and animal evidence judgments from Step 1 and the overall evidence integration
conclusion from Step 2 were reached using decision frameworks (see Tables IV, V, and VI) adapted
from considerations originally described by Austin Bradford Hill (Hill. 19651. In the first step, the
strength of the human and, separately, the animal evidence (with consideration of mechanistic
information in humans and animals, respectively, including in vitro or other relevant models) for
each noncancer health effect (or groups of related effects) and specific cancer type (or groups of
related cancer types) was summarized using the following terms: robust, moderate, slight, and
indeterminate. The strength of the human and animal evidence was determined starting from the
evidence syntheses that summarized the evidence from the available human and animal health
effects studies, respectively, and then considering coherence of effects and biological plausibility
based on mechanistic evidence, which could add to or detract from the strength of evidence.
Syntheses of mechanistic data that might inform potential respiratory health effects (Section 1.2.1-
1.2.4), which involved an integrated and systematic review process (see Appendix A.5.6),
emphasize the sequence(s) of mechanistic events interpreted to have the most reliable evidence
(e.g., mechanistic events and associations with robust evidence are preferred). Based on the known
or presumed linkages, these events are organized from a "plausible initial effect of exposure" (e.g., a
potential direct interaction between inhaled formaldehyde and biological materials) to each apical
EVIDENCE INTEGRATION CONCLUSION
The judgments regarding the human and
animal evidence are integrated in light of
evidence on the human relevance of the
findings in animals, susceptibility, and the
coherence of the findings across evidence
streams to draw a conclusion about the
evidence for health effects in humans.
This document is a draft for review purposes only and does not constitute Agency policy.
xxxvii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 toxicity endpoint in a linear fashion, regardless of tissue region. Additional details and other
2 mechanistic changes that might contribute to the observed health effects are discussed in
3 Appendix A.5.6. Note, however, that the lack of mechanistic data explaining an association did not
4 discount results from human or animal health effect studies. To draw these judgments, a modified
5 set of considerations was applied to evidence from studies in humans and animals (Table III).
6 Examples of ways that mechanistic evidence was used in causal analyses and derivation of toxicity
7 values are described in Table IV.
This document is a draft for review purposes only and does not constitute Agency policy.
xxxviii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table IV. Primary considerations for assessing the strength of evidence for the health effects studies in humans
and, separately, animals3
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
The structured categories and criteria in Tables VI and VII will guide the application of strength-of-evidence judgments for an outcome or health effect. Evidence
synthesis scenarios that do not warrant an increase or decrease in evidence strength will be considered "neutral."
Risk of bias;
sensitivity (across
studies)
• An evidence base of high or medium confidence studies increases
strength.
• An evidence base of mostly low confidence studies decreases strength. An exception to
this is when the primary issues resulting in low confidence are related to insensitivity.
This may increase evidence strength in cases where an association is identified because
the expected impact of study insensitivity is toward the null.
• Decisions to increase strength for other considerations in this table should generally not
be made if there are serious concerns for risk of bias.
Consistency
• Similarity of findings for a given outcome (e.g., of a similar magnitude,
direction) across independent studies or experiments increases
strength, particularly when consistency is observed across populations
(e.g., location) or exposure scenarios in human studies, and across
laboratories, populations (e.g., species), or exposure scenarios
(e.g., duration, route, timing) in animal studies.
• Unexplained inconsistency (conflicting evidence) decreases strength. Generally,
strength should not be decreased if discrepant findings can be reasonably explained by
study confidence conclusions, variation in population or species, sex, or lifestage,
exposure patterns (e.g., intermittent or continuous), levels (low or high), duration or
intensity. However, any decisions about decreased strength will be determined by the
extent to which residual questions about the evidence may persist.
Strength (effect
magnitude) and
precision
• Evidence of a large magnitude effect (considered within or across
studies), can increase strength. Effects of a concerning rarity or
severity can also increase strength, even if they are small magnitude.
• Precise results from individual studies or across the set of studies
increases strength, noting that biological significance is prioritized over
statistical significance.
• The presence of small effects is not typically used to decrease confidence in a body of
studies. However, if effect sizes that are small in magnitude are concluded not to be
biologically significant, or if there are only a few studies with imprecise results, then
strength is decreased.
• In animal studies, an example of evidence that can decrease strength involves an effect
for which there is a lesser level of concern under some conditions (e.g., rapid
reversibility after removal of exposure). Note that many reversible effects are of high
concern. Such a decision is informed by factors such as the toxicokinetics of the chemical
and the conditions of exposure (see U.S. EPA (1998)), judgments regarding the
potential for delayed or secondary effects, as well as the exposure context focus of the
assessment (e.g., addressing intermittent or short-term exposures).
Biological
gradient/dose-
response
• Evidence of dose-response increases strength. Dose-response may be
demonstrated across studies or within studies and it can be dose or
duration dependent. It may also not be a monotonic dose-response
• A lack of dose-response when expected based on biological understanding and having a
wide range of doses/exposures evaluated in the evidence base can decrease strength.
This document is a draft for review purposes only and does not constitute Agency policy.
xxxix DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
(monotonicity should not necessarily be expected), and the analysis
will consider the extent to which this might be explained by the
available evidence (e.g., different outcomes may be expected at low
versus high doses due to activation of different mechanistic pathways
or induction of systemic toxicity at very high doses).
• Decreases in a response after cessation of exposure (e.g., symptoms of
current asthma) also may increase strength by increasing certainty in a
relationship between exposure and outcome (this is applicable to
human observational studies, but not experimental studies).
• In rare cases, and typically only in toxicology studies, the duration of exposure might
reveal an inverse association with effect magnitude (e.g., due to tolerance or
acclimation). Similar to the discussion of reversibility above, a decision about whether
this decreases strength depends on the exposure context focus of the assessment and
other factors.
• If the data are not adequate to evaluate a dose-response pattern, then strength is
neither increased nor decreased.
Coherence
• Biologically related findings within an organ system, or across
populations (e.g., sex) increase strength, particularly when a temporal-
or dose-dependent progression of related effects is observed within or
across studies, or when related findings of increasing severity are
observed with increasing exposure.
• An observed lack of expected coherent changes (e.g., well-established biological
relationships), particularly when observed for multiple related endpoints, will typically
decrease evidence strength. The decision to decrease depends on the strength of the
expected relationship(s), and considers factors (e.g., dose and duration of exposure)
across studies of related changes.
Mechanistic
evidence related
to biological
plausibility
• Mechanistic evidence of precursors or health effect biomarkers in well-
conducted studies of exposed humans or animals, in appropriately
exposed human or animal cells, or other relevant human or animal
models (for the human or animal evidence, respectively) increases
strength, particularly when this evidence is observed in the same
cohort/population exhibiting the health outcome.
• Evidence of changes in biological pathways or providing support for a
proposed MOA in models also increases strength, particularly when
support is provided for rate-limiting or key events, or changes are
conserved across multiple components of the pathway or MOA.
• Mechanistic understanding is not a prerequisite forjudging the evidence, and thus
absence of knowledge should not be used a basis for decreasing strength NTP (2015):
NRC (2014a).The human relevance of animal findings is assumed unless there is
sufficient evidence to the contrary [see IARC (2006): U.S. EPA (2005a)l.
• Mechanistic evidence in well-conducted studies that demonstrates that the health
effect(s) are unlikely to occur, or only likely to occur under certain scenarios (e.g., above
certain exposure levels), can decrease evidence strength. A decision to decrease
depends on an evaluation of the strength of the mechanistic evidence supporting vs.
opposing biological plausibility, as well as the strength of the health effect-specific
findings (e.g., stronger health effect data require more certainty in mechanistic evidence
opposing plausibility).
aThese ideas build upon the discussion for assessing causality of disease in Hill (1965). although the use or interpretation of some of the terms differs.
bWhile humans are "exposed" and not "dosed," and nor are animals "dosed" via inhalation, "dose-response" is used for convention throughout the assessment, although it is
acknowledged that 'exposure-response' may be more appropriate in many contexts.
cThere is a clear overlap in the use of mechanistic evidence to interpret coherence (e.g., informing the relatedness or comparability of potentially coherent health findings) and
biological plausibility. The available mechanistic information is also considered during the subsequent step of evidence integration across streams of evidence (see Table VIII).
dAlthough it is not separately listed, Hill's consideration of 'analogy' (information for a similar but different association that supports causation) is indirectly encompassed by the
evaluation of coherence during the review of environmental health studies; however, this use of analogous chemicals or exposure scenarios is less common.
This document is a draft for review purposes only and does not constitute Agency policy.
xl DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table V. Examples of the interpretation and application of mechanistic
evidence
Mechanistic inferences considered
Potential specific applications within the assessment
Biological plausibility: as applied herein,
this applies to information that either
strengthens or weakens an interpretation
of the likelihood of an association
between exposure and the health effect.
Thus, in some instances, differing levels of
biological plausibility (or certainty) might
be drawn. It is important to note that the
lack of mechanistic data explaining an
association is not used to discount
observations from human or animal
studies. The interpretation of biological
plausibility considers the existing
knowledge for how the health effect
develops and can involve analyses of
information at different levels of biological
organization (e.g., molecular, tissue).
Evidence Integration (Animal or Human Health Effects)
• Observations of important mechanistic changes in exposed humans
or animals that are plausibly associated with the health outcome in
question can strengthen the confidence in the health effect
findings for either the human or animal evidence base, particularly
when the changes are observed in the same exposed population
presenting the health effect.
• The absence of expected mechanistic changes in an exposed
population might diminish the plausibility of an association. This
considers the sensitivity of the changes and the potential
contribution of alternative or unidentified toxicity mechanisms.
• Inconsistent evidence (i.e., heterogeneous results) across different
animal species or human populations might be explained by
evidence that mechanisms differ or are not/less operant in the
different populations (e.g., evidence demonstrating that certain
populations cannot metabolize a chemical to its reactive
metabolite; evidence that gene expression variability correlates
with response variability).
Human relevance of findings in animals: in
the absence of sufficiently justifiable
mode of action (MOA) information, effects
in animal models are assumed to be
relevant to humans (U.S. EPA, 2005a). In
this assessment, for potential health
hazards where the evidence from animal
models is likely to influence the overall
hazard conclusion, the available
mechanistic evidence was considered in
light of human relevance.
Evidence Integration (Overall Hazard Description)
• Evidence establishing that the mechanisms underlying the animal
response do not operate in humans, or that animal models do not
suitably inform a specific human health outcome can support the
view that the animal response is irrelevant to humans. In these
cases, the animal response provides neither an argument for nor
an argument against an overall hazard judgment.
• Observations of mechanistic changes in exposed humans that are
similar or coherent with mechanistic or toxicological changes in
experimental animals (and which are interpreted to be associated
with the health outcome under evaluation) strengthen the human
relevance of the animal findings.
Potential susceptibilities: When a
mechanistic understanding of how a
health outcome develops, or MOA, is
known or hypothesized, knowledge about
the presence and sensitivity (e.g., across
lifestages), or modifying factors
(e.g., genetics) of important events in that
MOA can help identify vulnerable groups.
Susceptibility, Dose-Response Analysis, and Uncertainty
• Identification of lifestages or groups potentially at greatest risk can
add clarity to hazard descriptions and inform uncertainties on
whether the most vulnerable populations have been adequately
tested.
• Knowledge of potential or expected vulnerabilities can inform
selection of studies for quantitative analysis (e.g., prioritizing
studies including such populations).
This document is a draft for review purposes only and does not constitute Agency policy.
xli DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Mechanistic inferences considered
Potential specific applications within the assessment
Biological understanding, including the
identification of precursor events: When
mechanistic data can reasonably describe
how effects develop, this information may
inform the situations or scenarios
expected to result in these effects.
Further, well-studied MOAs can
sometimes identify mechanistic precursor
events that can be qualitatively or
quantitatively linked to the apical health
effect in question with reasonable
confidence.
Dose-Response Analysis
• Understanding how effects develop might support the use of, for
example, particular models (e.g., models assuming effects do not
occur below certain levels; biologically based models; models
integrating data across several closely related outcomes) or
measures of exposure (e.g., different external or internal metrics).
• Uncertainty in the dose-dependence of responses in animals or
humans can be influenced by the occurrence of precursor events,
which can add to or subtract from the plausibility of the findings for
use in dose-response analyses. Relatedly, in rare instances,
well-established precursor events might be used as surrogates in
dose-responses analyses when the health effect-specific data are
less certain.
1 Decision frameworks, with criteria described in Tables VI and VII were used to develop the
2 judgments concerning the strength of evidence for a health effect within each of the human and
3 animal evidence bases, weighing the strengths and weaknesses of both positive and null studies.
4 These frameworks, which add clarity, consistency, and transparency to the evidence evaluations
5 and conclusions, are consistent with generally accepted principles in epidemiology and toxicology
6 and are meant to convey a distribution of confidence in each body of evidence pertaining to a
7 hazard, a process that relies on expert judgment
Table VI. Framework for strength of evidence judgments (human evidence)
Strength of
evidence
judgment
Description
Robust
... evidence in
human studies
(strong signal
of effect with
little residual
uncertainty)
A set of high or medium confidence independent studies reporting an association between the
exposure and the health outcome, with reasonable confidence that alternative explanations,
including chance, bias, and confounding, can be ruled out across studies. The set of studies is
primarily consistent, with reasonable explanations when results differ; an exposure-response
gradient is demonstrated; and the set of studies includes varied populations. Additional
supporting evidence, such as associations with biologically related endpoints in human studies
(coherence) or large estimates of risk or severity of the response, may increase confidence but
are not required.
In exceptional circumstances, a finding in one study may be considered to be robust, even when
other studies are not available (e.g., analogous to the finding of angiosarcoma, an exceedingly
rare liver cancer, in the vinyl chloride industry).
Mechanistic evidence from exposed humans or human cells, if available, may add support
informing considerations such as exposure-response, temporality, coherence, and MOA, thus
raising the level of certainty to robust for a set of studies that otherwise would be described as
moderate.
Moderate
A smaller number of studies (at least one high or medium confidence study with supporting
evidence), or with some heterogeneous results, that do not reach the degree of confidence
This document is a draft for review purposes only and does not constitute Agency policy.
xlii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Strength of
evidence
judgment
Description
... evidence in
human studies
(signal of effect
with some
uncertainty)
required for robust. For multiple studies, there is primarily consistent evidence of an
association, but there may be lingering uncertainty due to potential chance, bias or
confounding.
For a single study, there is a large magnitude or severity of the effect, or a dose-response
gradient, or other supporting evidence, and there are not serious residual methodological
uncertainties. Supporting evidence could include associations with related endpoints, including
mechanistic evidence from exposed humans or human cells, if available, based on
considerations such as exposure-response, temporality, coherence, and MOA, thus raising the
level of certainty to moderate for a set of studies that otherwise would be described as slight.
Slight
... evidence in
human studies
(signal of effect
with large
amount of
uncertainty)
One or more studies reporting an association between exposure and the health outcome,
where considerable uncertainty exists. In general, only low confidence studies may be
available, or considerable heterogeneity across studies may exist. Supporting coherent
evidence is sparse. Strong biological support from mechanistic evidence in exposed humans or
human cells may also be independently interpreted as slight. This also includes scenarios where
there are serious residual uncertainties across studies (these uncertainties typically relate to
exposure characterization or outcome ascertainment, including temporality) in a set of largely
consistent medium or high confidence studies. This category serves primarily to encourage
additional study where evidence does exist that might provide some support for an association,
but for which the evidence does not reach the degree of confidence required for moderate.
Indeterminate
... evidence in
human studies
(signal cannot
be determined
for or against
an effect)
No studies available in humans or situations when the evidence is inconsistent or primarily of
low confidence
Compelling
evidence of no
effect
... in human
studies
(strong signal
for lack of an
effect with little
uncertainty)
Several high confidence studies showing null results (for example, an odds ratio of 1.0), ruling
out alternative explanations including chance, bias, and confounding with reasonable
confidence. Each of the studies should have used an optimal outcome and exposure
assessment and adequate sample size (specifically for higher exposure groups and for
susceptible populations). The 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.
This document is a draft for review purposes only and does not constitute Agency policy.
xliii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table VII. Framework for strength of evidence judgments (animal evidence)
Strength of
evidence
judgment
Description
Robust
... animal
evidence
The set of high or medium confidence experiments includes 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
can reasonably rule out the potential for nonspecific effects (e.g., resulting from toxicity) to
have resulted in the findings. Any inconsistent evidence (evidence that cannot be reasonably
explained by the respective study design or differences in animal model) is from a set of
experiments of lower confidence. At least two of the following additional factors in the set of
experiments increases certainty in the evidence for the health outcome(s): coherent effects
across multiple related endpoints (may include mechanistic evidence); 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. Alternatively, mechanistic data in
animals or animal cells that address the above considerations or that provide experimental
support for a MOA that supports causality with reasonable confidence may raise the level of
certainty to robust for evidence that otherwise would be described as moderate or,
exceptionally, slight, or indeterminate.
Moderate
... animal
evidence
A set of evidence that does not reach the degree of certainty required for robust, but which
includes at least one high or medium confidence study and information strengthening the
certainty in the evidence for the health outcome(s). Although the results are largely consistent,
notable uncertainties remain. However, while inconsistent evidence or evidence indicating
nonspecific effects (e.g., toxicity) may exist, it is not sufficient to reduce or discount the level of
concern regarding the positive findings from the supportive experiments or it is from a set of
experiments of lower confidence. The set of experiments supporting the effect provide
additional information supporting causality, such as consistent effects across laboratories or
species; coherent effects across multiple related endpoints (may include mechanistic evidence);
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 data in animals or animal cells that address the
above considerations or that provide information supporting causality with reasonable
confidence may raise the level of certainty to moderate for evidence that otherwise would be
described as slight.
Slight
... animal
evidence
Scenarios in which there is a signal of a possible effect, but the evidence is conflicting or weak.
Most commonly, this includes situations where only low confidence experiments are available
and supporting coherent evidence is sparse. It also applies when one medium or high
confidence experiment is available without additional information increasing the certainty in
the evidence (e.g., corroboration within the same study or from other studies). Lastly, this
includes scenarios in which there is evidence that would typically be characterized as moderate,
but inconsistent evidence (evidence that cannot be reasonably explained by the respective
study design or differences in animal model) from a set of experiments of higher confidence
(may include mechanistic evidence) exists. Strong biological support from mechanistic studies
in exposed animals or animal cells may also be independently interpreted as slight. Notably, to
encourage additional research, it is important to describe situations where evidence exists that
might provide some support for an association but is insufficient for a conclusion of moderate.
This document is a draft for review purposes only and does not constitute Agency policy.
xliv DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Toxicological Review of Formaldehyde—Inhalation
Strength of
evidence
judgment
Description
Indeterminate
...animal
evidence
No animal studies were available, or a set of low confidence animal studies exist that are not
reasonably consistent or are not informative to the hazard question under evaluation.
Compelling
evidence of no
effect
... in animal
studies
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, postexposure latency, and endpoint evaluation
procedures, and to address potentially susceptible populations and lifestages.
In the next step (i.e., after judging the strength of the human and animal evidence
separately), the entire body of evidence was integrated across the human and animal evidence,
considering mechanistic information on the human relevance of the animal evidence and coherence
of the findings across streams of evidence, to arrive at an overall evidence integration judgment
regarding the evidence for causation (Table VIII). This evidence integration framework interprets
the instructions and examples provided in the cancer guidelines fU.S. EPA. 2005al to allow clarity
and consistency in the evaluation of each potential human hazard. The evidence integration
framework is consistent with the cancer guidelines in that evidence in humans generally has
greater weight than evidence in animals. In the absence of sufficiently justifiable MOA information,
effects in animal models are assumed to be relevant to humans. In this assessment, for potential
health hazards where the evidence from animal models influenced the overall evidence integration
judgment, the available mechanistic evidence was considered to inform human relevance.
For each potential health effect evaluated, a narrative evidence integration summary and
judgment was developed. The overall evidence integration judgments of evidence demonstrates,
evidence indicates [likely], evidence suggests and evidence inadequate (to judge hazard) are
defined in Table VIII and presented as bolded text throughout the assessment, accompanied by a
description of the conditions of expression (e.g., exposure levels, exposure patterns) in the studies
that served as the basis for the judgment Importantly, for the purposes of this assessment, the
same evidence integration approach was used to draw evidence integration judgments for both
noncancer health effects and specific cancer types. This approach uses the methods and
considerations and described in the EPA cancer guidelines fU.S. EPA. 2005al Consistent with these
guidelines, for carcinogenicity, a final step of categorizing the totality of the evidence using a
"descriptor" was performed, as described in Table IX.
This document is a draft for review purposes only and does not constitute Agency policy.
xlv DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table VIII. Overall evidence integration judgments for characterizing
potential human health hazards (noncancer health effects and cancer
outcomes) in the evidence integration narrative
Overall evidence
integration judgment
in narrative
Explanation and example scenarios
Evidence demonstrates
This signifies a very high level of certainty that formaldehyde exposure causes the
health effect in humans.
• This category was3 used if there was robust human evidence supporting an effect.
• This category could also be used with moderate human evidence and robust animal
evidence if there was strong mechanistic evidence that MOAs and key precursors
identified in animals were anticipated to occur and progress in humans.
Evidence indicates
(likely)b
This reflects a reasonable certainty that the relationship between formaldehyde
exposure and the health outcome is causal, although there may be some outstanding
questions that remain.
• This category was used if there is robust animal evidence supporting an effect and
slight-to-indeterminate human evidence, or with moderate human evidence when
strong mechanistic evidence was lacking.
• This category 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 were not sufficient to reduce or discount
the level of concern, or mechanistic evidence in the slight or indeterminate evidence
base (e.g., precursors) existed to increase confidence in the moderate evidence.
Evidence suggests (but
is not sufficient to
infer)c
This conveys some concern that formaldehyde may cause a particular health effect in
humans, but there were very few studies that contributed to the evaluation, the
evidence was very weak or conflicting, or the methodological conduct of the studies
was poor. Given the substantial degree of uncertainty, additional research would
provide valuable information for future evaluations.
• This category was used if there was slight human evidence and slight-to-
indeterminate animal evidence.
• This conclusion level was also used with slight animal evidence and slight-to-
indeterminate human evidence.
• This category could also be used with moderate human evidence and sliaht or
indeterminate animal evidence, or with moderate animal evidence and slight or
indeterminate human evidence. In these scenarios, there were outstanding issues
regarding the moderate evidence that reduced the level of concern or confidence in
the reliability of the findings, or mechanistic evidence in the slight or indeterminate
evidence base (e.g., null results in well-conducted evaluations of precursors) existed
to decrease confidence in the moderate evidence.
• Exceptionally, when there is general scientific understanding of mechanistic events
that result in a hazard, this category could also be used if there was strong
mechanistic evidence that was sufficient to identify a cause for concern—in the
This document is a draft for review purposes only and does not constitute Agency policy.
xlvi DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Overall evidence
integration judgment
in narrative
Explanation and example scenarios
absence of adequate conventional studies in humans or in animals (i.e., indeterminate
evidence in both).
Evidence inadequated
This conveys either a lack of information or an inability to interpret the available
evidence.
• This category was used if there was indeterminate human and animal evidence.
• This category could also be used with sliaht-to-robust animal evidence and
indeterminate human evidence if strong mechanistic information indicated that the
animal evidence was unlikely to be relevant to humans.
A conclusion of inadequate is not a determination that the agent does not cause
adverse health outcomes or is safe. It generally indicates that further research is
needed.
Note: This table does not supersede or alter direction provided in EPA guidelines. It is meant only to provide added
transparency for conclusions drawn regarding the level of evidence from human, animal, and mechanistic studies,
terminology of "was" refers to the default option; terminology of "could also be" refers to alternative options.
bFor some applications, such as benefit-cost analysis, to better differentiate the categories of evidence demonstrates and
evidence indicates (likely), the latter category should be interpreted as evidence that supports an exposure-effect linkage that
is likely to be causal.
cHealth effects characterized as having evidence demonstrates and evidence indicates (likely) (and, in some cases, evidence
suggests) are evaluated for use in dose-response assessment. When the database includes at least one well-conducted study
and a judgment of evidence suggests is drawn, quantitative analyses may still be useful for some purposes (e.g., providing a
sense of the magnitude and uncertainty of estimates for health effects of potential concern, ranking potential hazards, or
setting research priorities), but not for others [see related discussions in U.S. EPA (2005b)]. It is critical to transparently
convey the extreme uncertainty in any such estimates.
dSpecific narratives for each of the health effects with an evidence integration judgment of evidence inadequate may be
deemed unnecessary.
1 For carcinogenesis only, the weight of evidence as to whether formaldehyde inhalation
2 exposure is carcinogenic to humans was summarized using descriptors, consistent with EPA
3 guidelines (U.S. EPA. 2005a) (Table IX). For this assessment, the descriptors build upon the overall
4 evidence integration judgments for individual cancer types, as described in Table VIII; however,
5 this does not alter or supersede direction provided in EPA guidelines. These descriptors are bolded
6 and italicized.
Table IX. Criteria for applying cancer descriptors to overall confidence
conclusions for cancer types
Cancer descriptor
Criteria
Carcinogenic to humans
This descriptor was used if the evidence demonstrates that, for at least one
cancer type, formaldehyde inhalation exposure caused the increase in cancer
incidence or mortality.
This document is a draft for review purposes only and does not constitute Agency policy.
xlvii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Cancer descriptor
Criteria
This descriptor could also be used in rare instances if the evidence indicates that
formaldehyde inhalation exposure likely causes different cancer types across
evidence bases (e.g., when one type of cancer is based on human evidence and
tumors at another site is supported by animal evidence), consistent with EPA
guidelines (U.S. EPA, 2005a) that site concordance is not reauired. Such a decision
would depend on mechanistic understanding (i.e., in this example, the decision
would consider differences in tumor types or ADME across species).
Likely to be carcinogenic to
humans
This descriptor was used if the evidence indicates that, for at least one cancer
type, formaldehyde inhalation exposure likely caused the increase in cancer
incidence or mortality.
Similar to the rationale provided above, this descriptor could also be used in rare
instances when the evidence suggests formaldehyde inhalation exposure may
cause multiple tumor types, depending on mechanistic inference.
Suggestive evidence of
carcinogenic potential
This descriptor was used if, for the evidence relating to carcinogenicity, the
evidence was only suggestive that formaldehyde inhalation exposure may cause
any of the observed increases in cancer incidence or mortality for any cancer type.
This would reflect a substantial degree of uncertainty in any potential causal
inference.
Inadequate evidence to
assess carcinogenic potential
This descriptor was used if the evidence was inadequate to draw a conclusion
regarding cancers of any type with any confidence. This might reflect a lack of
information or highly conflicting information.
Not Likely to be carcinogenic
to humans
This descriptor conveys a high degree of certainty that there is negligible concern
for carcinogenic effects. A substantial amount of evidence would be required to
support this descriptor (see (U.S. EPA, 2005a).
Quantitative Dose-Response Assessment
This formaldehyde assessment includes development of organ/system-specific RfCs (osRfC)
and an overall RfC for noncancer effects, as well as an IUR for carcinogenic effects, presented in
units of ng/m3.5 From among the body of evidence used for the hazard identification assessment,
selection of the studies for dose-response assessment used information from the study confidence
evaluations, with particular emphasis on conclusions regarding the characteristics of the study
population (considering potential susceptible groups) and the accuracy of formaldehyde exposure,
the severity of the observed effects, and the exposure levels analyzed (see Appendix B). Based on
the data available in this assessment, the subset of studies used to develop RfCs and unit risk
estimates were from those noncancer health outcomes and specific cancer types with an overall
evidence demonstrates or evidence indicates [likely] judgment regarding the potential for
formaldehyde inhalation to cause those effects (see Section 2).
For each health effect for which a value was derived, one or more studies were determined
to be suitable for use in quantitative exposure-response assessment, and these are discussed in
Section 2.1 for effects other than cancer and in Section 2.2 for specific cancer types. A POD was
5 Throughout this assessment, a conversion of 1 ppm = 1.23 mg/m3 formaldehyde is used.
This document is a draft for review purposes only and does not constitute Agency policy.
xlviii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
determined for several health effects, including sensory irritation, pulmonary function, respiratory
tract pathology, prevalence of current asthma, allergic conditions, developmental and female
reproductive toxicity, male reproductive toxicity, respiratory tract cancers (i.e., nasopharyngeal
cancer), and lymphohematopoietic cancers (i.e., myeloid leukemia). In some cases, estimates
considered information from mechanistic studies (see Table ES-2, footnote c for examples of how
these data were considered quantitatively). Specifically, for some outcomes (i.e., nasal cancers;
noncancer respiratory tract pathology), analyses included efforts to apply dosimetry models
estimating the uptake of inhaled formaldehyde, including an evaluation of modeling efforts to
account for the potential contribution of endogenous formaldehyde on uptake (see Section 2.2).
Candidate osRfCs or cancer unit risk values were estimated for each of these noncancer or cancer
health outcomes, respectively, and the associated uncertainties were discussed. In addition to the
overall evidence integration judgment for concluding that formaldehyde inhalation results in
specific health effects (which incorporates the individual study confidence), a confidence level of
high, medium, or low was assigned to each osRfC regarding the reliability of the associated POD
calculation(s). Confidence in the completeness of the database for each osRfC was also assigned.
These judgments were used to select the RfC, draw an overall level of confidence in the RfC, and
determine the completeness of the formaldehyde literature database. For noncancer health
hazards, multiple graphical depictions were developed to display PODs, uncertainty factors, and
candidate osRfCs across outcomes and studies, as well as the context of these estimates (e.g., in
relation to the study-specific results, in relation to known human exposures to formaldehyde).
Organ/system-specific RfCs, a single, overall RfC, and unit risk were selected; the specific rationale
is described in Section 2, Dose-Response Analysis. For the derivation of the cancer inhalation unit
risk (IUR) estimate, exposure-response analyses for nasopharyngeal cancer (NPC) from an
occupational cohort study and cancers of the nose across two bioassays in rats, and for
lymphohematopoietic malignancies from an occupational cohort study, were considered. The IUR
was based on the preferred unit risk estimate for NPC and application of age-dependent adjustment
factors (see Section 2.2.6). An overall level of confidence was assigned to the IUR. For one
mechanism that contributes to cancer risk, cytotoxicity-induced regenerative proliferation, a
contributing mechanism which appears to involve a threshold, cRfCs were derived using different
data sets from rat bioassays.
Table X. Considerations for study selection for quantification of dose-response
and derivation of toxicity values
Factor
Considerations
Overall
Confidence
Conclusion
For this assessment, if the data were amenable, a toxicity value was estimated for health effects
with evidence integration judgments of evidence demonstrates or evidence indicates [likely].
Although it may sometimes be possible to develop toxicity values for judgments of evidence
suggests, given the particulars of the available data in this assessment, toxicity values were not
estimated.
This document is a draft for review purposes only and does not constitute Agency policy.
xlix DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Factor
Considerations
Study
Confidence
Studies with appropriate study designs (e.g., long-term bioassays were preferred for animal
studies of most health effects), reasonably complete reporting of results, and with no identified
sources of selection bias, information bias, or confounding that would substantially alter
interpretation of study results.
Population
Human studies were preferred over animal studies. Dose-response information for the most
susceptible subgroups was evaluated, if appropriate.
Exposure
information
Studies with risk estimates for multiple exposure levels or regression coefficients per unit of
formaldehyde concentration were generally preferred over LOAELs or NOAELs because they
provided information about the shape of the concentration-response curve and allowed for
benchmark dose modeling.
The role of endogenously generated formaldehyde in human diseases is largely unknown.
This includes endogenous formaldehyde generated during normal cellular metabolic processes, as
well as formaldehyde produced endogenously within cells (e.g., in the liver) as a breakdown
product of external exposures to other chemicals, including ingestion of caffeine fSummers etal..
2012: Hohnloser et al.. 1980) and methanol-rich foods or beverages, such as fruit-based liquors
(Riess etal.. 2010). The mode of action by which toxicity at distal sites, such as bone marrow or
reproductive tissues, may occur in response to inhalation of formaldehyde over long periods, also is
not known. Once formaldehyde is inhaled and interacts with extracellular aqueous matrices such
as mucus in nasal passages and is hydrated, the biochemical reactivity of inhaled formaldehyde and
endogenous formaldehyde are likely to be very similar, given that there are no differences in
chemical structure. However, no specific data are available to inform whether there may be
differences in interactions with specific extracellular or intracellular macromolecular targets in
vivo. While the rate of cellular detoxification of exogenous formaldehyde remains unknown, the
production and subsequent detoxification of endogenous formaldehyde appears to be kept under
strict control and has been well described (Burgos-Barragan etal.. 2017bl.
The focus of the assessment is to estimate the risk over background that results from only
the exogenous exposure, and the assessment assumes that background incidence of cancer or other
health hazard that may potentially be attributed to endogenous formaldehyde is already accounted
for in the background. Endogenous formaldehyde might be responsible for some portion of
background risks for some health outcomes, particularly when normal detoxification pathways are
deficient (e.g.. Pontel etal.. 2015): but that possibility is not the purpose of this review. This
assessment does consider and discuss the potential impact of normal levels of endogenous
formaldehyde on the penetration and distribution of inhaled formaldehyde, based on recent
dosimetric models ffCampbell Tr et al.. 2020: Schroeter etal.. 20141: see Section 2.2). In addition,
efforts to incorporate the unknown contribution of endogenous formaldehyde to background
cancer incidence in an attempt to bound low-dose human cancer risks from formaldehyde exposure
have been published using a measure of internal dose for inhaled formaldehyde. These papers are
discussed in Section 2.2 and Appendix B.2.3.
This document is a draft for review purposes only and does not constitute Agency policy.
1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
EXECUTIVE SUMMARY
ES.l OVERALL SUMMARY
This IRIS health assessment presents a systematic evaluation of the publicly available
studies relevant to inhalation exposure to formaldehyde and potential adverse health outcomes.
The purpose of the review was to identify hazards that may result from formaldehyde inhalation
and to describe the level of confidence in each conclusion. When there was sufficient confidence in
a hazard and the studies and data available, toxicity values were derived using either analyses of
dose-response or selected no-observed-adverse-effect or lowest-observed-adverse-effect levels
(NOAEL or LOAEL). The conclusions of the assessment are summarized in Tables ES-1 and ES-2.
The evidence identification, evaluation, and integration framework depicted in Figure I was
used to conduct the assessment Potential health hazards were evaluated, including sensory
irritation; reduced pulmonary function; immune system effects, focusing on allergic conditions and
asthma; respiratory tract pathology; nervous system effects; reproductive and developmental
toxicity; and cancer. Several extensively studied cancer sites were specifically evaluated, including
cancers of the upper respiratory tract (i.e., nasopharyngeal cancer, sinonasal cancer, cancers of the
oropharynx/hypopharynx, and laryngeal cancer) and of the lymphohematopoietic system
(i.e., Hodgkin lymphoma, multiple myeloma, myeloid leukemia, and lymphatic leukemia).
This document is a draft for review purposes only and does not constitute Agency policy.
li DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table ES-1. Evidence integration judgments for noncancer health effects and
the reference concentration (RfC)
Noncancer health effect
Confidence in
health effect
POD basis
Confidence in
POD
UFC
osRfC
(mg/m3)
Decreased pulmonary function
evidence indicates
[likely]'
Human
high
3
0.007
Allergic conditions
evidence indicates
[likely]
Human
high
3
0.008
Current asthma symptoms or degree of
asthma control
evidence indicates
[likely]
Human
medium
10d
0.006d
Sensory irritation
evidence
demonstrates
Human
medium
10
0.009
Female reproductive or developmental
toxicity
evidence indicates
[likely]
Human
low
10
0.01
Respiratory tract pathology
evidence
demonstrates
Rat
medium
30d
0.003d
Male reproductive toxicity
evidence indicates
[likely]
Rat
low
3000
0.001
Nervous system effects3
evidence suggests
Not Derived
-
-
-
Confidence in
health effects
PODs basis
Confidence in PODs
UFC
Confidence in
database
RfC
(mg/m3)
Overall
confidence
RfCb:
Medium or High
Human
Medium or High
3 or 10d
High
0.007
High
Abbreviations and definitions: RfC = reference concentration: An estimate (with uncertainty spanning perhaps an order of
magnitude) of a continuous inhalation exposure of a chemical to the human population (including sensitive subpopulations),
that is likely to be without risk of deleterious noncancer effects during a lifetime. osRfC = organ- or system-specific RfC: an RfC
based on the evidence for effects on that particular organ or system. UFC = composite (total) uncertainty factor; POD = point
of departure.
aThree separate judgments were drawn for nervous system effects, all evidence suggests; specifically, the evidence suggests,
but is not sufficient to infer, that formaldehyde inhalation might cause increases in amyotrophic lateral sclerosis incidence or
mortality, developmental neurotoxicity, or behavioral toxicity.
bBasis for RfC—sensory irritation, decreased pulmonary function, current asthma symptoms or degree of asthma control, and
allergic conditions. The corresponding osRfCs (i.e., based on human studies with medium or high confidence in the health
effects and PODs) are highlighted in gray, which also have the lowest UFC values.
cFor decreased pulmonary function, the judgment evidence indicates [likely] was drawn for long-term exposure durations. For
acute or intermediate exposure durations (hrs to wks), the evidence is inadequate to draw judgments.
dThese two osRFCs and the RfC are based on multiple studies and candidate values, sometimes with different UFCs applied.
The UFC values shown in this table and Figure 2-2 reflect the candidate values selected to represent each osRfC [i.e., the UFC
applied to the POD from Krzyzanowski et al. (1990) for asthma and from Woutersen et al. (1989) for respiratory pathology].
This document is a draft for review purposes only and does not constitute Agency policy.
lii DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table ES-2. Cancer evidence integration judgments, carcinogenicity
descriptor, and inhalation unit risk (IUR) for cancer incidence
Cancer type investigated
Evidence
integration
judgment for
cancer type
risk
Unit risk
estimate
basis
Unit risk
estimate (per
Hg/m3)
ADAF-adjusted
unit risk
estimate (per
Hg/m3)a
Confidence
in the unit
risk estimate
Nasopharyngeal cancer
(or nasal cancer in animals)
evidence
demonstrates*1
Human
6.4 x 10"6
1.1 X 10"5
medium
Animalc
8.9 x 10"6
to 1.8 x 10"5
NAd
medium
Myeloid leukemia
evidence
demonstrates8
Human
3.4 x 10"5
NAf
low
Sinonasal cancer
evidence
demonstrates8
No usable data
-
-
Oropharyngeal/Hypo-
pharyngeal cancer
evidence
suggests
Not derived
-
-
Multiple myeloma
evidence
suggests
Not derived
-
-
Hodgkin lymphoma
evidence
suggests
Not derived
-
-
Laryngeal cancer
evidence
inadequate
Not derived
-
-
Lymphatic leukemia
evidence
inadequate
Not derived
-
-
Carcinogenicity Descriptor:
Carcinogenic to Humans
Total cancer risk (IUR)1':
1.1 x 10 '' per |Jg/mConfidence in the IUR is Medium
Abbreviations and definitions: IUR = inhalation unit risk: the upper-bound excess lifetime cancer risk estimated to result from
continuous exposure to an agent at a concentration of 1 pg/m3 in air; ADAF = age-dependent adjustment factor.
aADAF adjustments are recommended for cancers for which there is sufficient evidence that formaldehyde has, at least in part,
a mutagenic MOA (see Section 2.2.4).
bThe judgment of evidence demonstrates for NPC cancer is based on robust human evidence of increased risk in groups
exposed to occupational formaldehyde levels, and robust animal evidence of nasal cancers in rats and mice that exhibits
steeply increasing incidence at high formaldehyde levels. Strong mechanistic support is provided across species (primarily
rats, but also mice, monkeys, and humans), including genotoxicity, epithelial damage or remodeling, and cellular proliferation
that are consistent with neoplastic development in a regional, temporal, and dose-related fashion.
cWhile the preferred unit risk estimate for NPC is based on a cancer mortality study in humans, several estimates in general
agreement with each other were also derived based on animal nasal tumor incidence. These estimates used multiple
mechanistic and statistical models, including biologically based dose-response (BBDR) modeling (see Section 2.2.2). In
addition, an RfC for one mechanism contributing to nasal cancer development, specifically cytotoxicity-induced regenerative
cell proliferation, was estimated to be between 0.006 and 0.018 mg/m3 based on calculations using animal data. Specifically,
this narrow RfC range was estimated based on cRfCs from a pathology study of hyperplasia, labeling studies of proliferating
cells, and BBDR modeling results (see Section 2.2.2).
dNA = not applicable; an ADAF-adjusted value was not calculated for the unit risk estimates based on the animal data on nasal
cancer, as the human unit risk estimate for NPC was the preferred estimate.
0 The judgment of evidence demonstrates for myeloid leukemia is based on robust human evidence of increased risk in groups
exposed to occupational formaldehyde levels. Supporting mechanistic evidence consistent with leukemia development is
provided across numerous studies of peripheral blood isolated from exposed workers, including evidence of mutagenicity and
This document is a draft for review purposes only and does not constitute Agency policy.
liii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
other genotoxic damage in lymphocytes and myeloid progenitors, and perturbations to immune cell populations. The animal
evidence is inadequate and the findings to date suggest that there may be a lack of concordance across species for leukemia,
as leukemia was not increased in two well-conducted chronic bioassays of rats or mice, and the available animal data provide
weak mechanistic support for LHP cancers. No MOA has been established to explain how formaldehyde inhalation can cause
myeloid leukemia without systemic distribution (inhaled formaldehyde does not appear to be distributed to an appreciable
extent beyond the respiratory tract to distal tissues).
fNA = not applicable; no ADAF adjustment is recommended for myeloid leukemia because the MOA is unknown (see
Section 1.3.3).
gThe judgment of evidence demonstrates for sinonasal cancer is based primarily on robust human evidence of increased risk in
groups exposed to occupational formaldehyde levels. The strong animal and mechanistic evidence for nasal cancers across
species is interpreted to provide moderate evidence supportive of sinonasal cancer (a judgment of moderate rather than
robust reflects some uncertainty in interpreting the nasal cavity findings in animals as fully applicable to the specific human
disease of sinonasal cancer; see Section 1.2.5).
hThe full lifetime (ADAF-adjusted) IUR estimate is based on the ADAF-adjusted estimate for nasopharyngeal cancer (which
includes a mutagenic MOA; see Section 1.2.5). Less-than-lifetime exposure scenarios with a very large fraction of exposure
during adulthood may not warrant ADAF adjustment, and one may choose to use the unadjusted unit risk estimate of 6.4 x 10"
6 per ng/m3. Otherwise, see Table 2-39 for an illustration of how to apply the ADAFs to obtain total cancer risk estimates for
less-than-lifetime exposure scenarios (see Section 2.2.4).
ES.2 HAZARD ASSESSMENT SUMMARY
ES.2.1. Noncancer Effects
Overall, the integrated evidence demonstrates that inhalation of formaldehyde causes
increased sensory irritation and respiratory tract pathology in humans, given appropriate exposure
circumstances. Well-conducted studies in humans and animals support these hazard conclusions,
and strong mechanistic evidence in animals provides plausible modes of action (MOAs) for the
identified endpoints.
The available evidence indicates that formaldehyde inhalation likely causes decreased
pulmonary function, an increased frequency of current asthma symptoms or difficulty controlling
asthma, and increased allergic responses in humans, given appropriate exposure circumstances.
These conclusions were supported primarily by evidence in exposed humans, with supportive
mechanistic evidence indicating that formaldehyde inhalation results in biological changes related
to these outcomes in exposed animals. In addition, the evidence indicates that inhalation of
formaldehyde likely causes female reproductive or developmental toxicity and reproductive
toxicity in men, given appropriate exposure circumstances. The conclusion for female reproductive
or developmental toxicity is supported by evidence in humans, specifically increases in time-to-
pregnancy (TTP) and spontaneous abortion risk; mechanistic evidence explaining such effects
without systemic distribution of formaldehyde is lacking. The conclusion for male reproductive
toxicity is supported primarily by coherent evidence of several alterations to the male reproductive
system in animals exposed to very high levels of formaldehyde (>6 mg/m3), with some
corroborative changes in an occupational epidemiological study; although no MOA is available,
some relevant mechanistic changes have been observed in well-conducted studies of the male
reproductive organs of exposed rodents.
Lastly, while a number of studies reported evidence of potential neurotoxic effects,
including developmental neurotoxicity, behavioral toxicity, and an increased incidence of, or
This document is a draft for review purposes only and does not constitute Agency policy.
liv DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Toxicological Review of Formaldehyde—Inhalation
mortality from, the motor neuron disease amyotrophic lateral sclerosis (ALS), due to limitations in
the database (e.g., poor methodology, lack of consistency), the integration of the evidence for each
of these manifestations of potential neurotoxicity ultimately resulted in the determination that the
evidence suggests, but is not sufficient to infer, that formaldehyde inhalation may pose a human
health hazard, and additional study is warranted. The available data on potential nervous system
effects were considered insufficient for developing quantitative toxicity estimates.
cc o 2 Cancer
VjallCCl
Formaldehyde is Carcinogenic to Humans by the Inhalation Route of Exposure. This
conclusion is independently supported by three evidence integration judgments:
• The evidence demonstrates that formaldehyde inhalation causes nasopharyngeal cancer
(NPC) in humans. This is based primarily on observations of increased risk of NPC in groups
exposed to occupational formaldehyde levels and nasal cancers in mice and several strains
of rats, with strong, reliable, and consistent mechanistic evidence in both animals and
humans (i.e., robust evidence for both the human and animal evidence, and strong
mechanistic support for the human relevance of the animal data). The nasopharynx,
although not typically specified in animal studies, is the region adjacent to the nasal cavity,
where the animal evidence was predominantly observed (thus, the animal evidence is
judged as robust). In addition, the evidence is sufficient to conclude that a mutagenic MOA
of formaldehyde is operative in formaldehyde-induced nasopharyngeal carcinogenicity.
• The evidence demonstrates that formaldehyde inhalation causes sinonasal cancer (SNC)
in humans. This is based primarily on observations of increased risk of SNC in groups
exposed to occupational formaldehyde levels (i.e., robust human evidence) and supported
by apical and mechanistic evidence for nasal cancers across multiple animal species. Some
uncertainties remain in the interpretation of the animal nasal cavity data as wholly
applicable to interpreting human sinonasal cancer (thus, the animal evidence is judged as
moderate). In addition, while uncertainties remain, the evidence is sufficient to conclude
that a mutagenic MOA of formaldehyde is operative in formaldehyde-induced sinonasal
carcinogenicity.
• The evidence demonstrates that formaldehyde inhalation causes myeloid leukemia in
humans. This is based primarily on observations of increased risk in groups exposed to
occupational formaldehyde levels. This evidence integration judgment is further supported
by other studies of human occupational exposure that provide strong and coherent
mechanistic evidence identifying clear associations with additional endpoints relevant to
LHP cancers, including an increased prevalence of multiple markers of mutagenicity and
other genotoxicity in peripheral blood cells of exposed workers, other perturbations to
immune cell populations in blood (primarily from human studies), and evidence of other
systemic effects (i.e., developmental or reproductive toxicity). Generally, evidence
supporting the development of LHP cancers after formaldehyde inhalation has not been
observed in experimental animals (i.e., rodents), including a well-conducted, chronic cancer
bioassay in two species, a similar lack of increased leukemias in a second rat bioassay, and
multiple mechanistic evaluations of relevant biological changes, including genotoxicity
(i.e., inadequate evidence). The exact mechanism(s) leading to cancer formation outside of
the respiratory tract are unknown.
This document is a draft for review purposes only and does not constitute Agency policy.
lv DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
The hazard conclusion for cancer is consistent with those drawn by other expert review
panels. Formaldehyde was classified as a known carcinogen by the NTP f20111 and a Group 1
carcinogen by IARC f2012. 20061, both based on evidence for nasal cancers in humans and animals
and myeloid leukemia in humans, with supporting data on mechanisms of carcinogenesis. In
addition, an expert committee convened by the NAS confirmed the conclusions of the NTP 12th
Report on Carcinogens (RoC) and conducted an independent review of the literature through 2013,
concluding that formaldehyde is a known carcinogen. The European Union and Health Canada
concluded that formaldehyde is a genotoxic carcinogen with a cytotoxic MOA (SCOEL. 2017: ECHA.
2012: Health Canada. 2006. 20011.
ES.3 DOSE-RESPONSE ASSESSMENT SUMMARY
ES.3.3. Inhalation Reference Concentration (RfC) for Noncancer Effects:
The reference concentration (the RfC) of 0.007 mg/m3 is the concentration one can breathe
every day for a lifetime that is not anticipated to cause any harmful noncancer health effects.
Organ- or system-specific reference concentrations fosRfCsl
In this assessment, the RfC is based on several osRfCs, which are themselves based on
candidate reference concentrations (cRfCs). The cRfCs are estimates for a specific endpoint based
on a single, specific study within an organ- or system-specific hazard domain. The osRfCs differ
from the associated cRfCs only when there are multiple cRfCs for the same organ system.
The osRfCs that were used to calculate the overall RfC in this assessment were all based on
epidemiological studies and were interpreted with either high- or medium-confidence based on
(1) the study results (i.e., confidence in the individual studies used to derive the osRfC), (2) the
point of departure (POD) and the cRfC derivation, and (3) the hazard determination (the strongest,
highest confidence judgment of evidence demonstrates was preferred) (see Table ES-1). In
general, the studies preferred as the basis for the derivation of the RfC were those human studies
that best represented the general population, including sensitive subgroups. An osRfC was typically
selected from those cRfCs that had a greater degree of certainty with regard to both reliability of
study results and cRfC derivation (including POD selection). In addition, candidate RfCs with lower
composite uncertainty factors (UFcs) were preferred.
The overall RfC is within the narrow range (0.006-0.009 mg/m3) of the group of respiratory
system-related osRfCs (sensory irritation, pulmonary function, allergy-related conditions, and
current asthma prevalence or degree of control). The health effects generally were observed in the
range of indoor formaldehyde concentrations in population studies (effects were observed in
studies at approximately 35-40 ng/m3), and these were used to arrive at the osRfCs associated with
the lowest UFcs. Thus, the selected RfC is at the upper end of the range of outdoor formaldehyde
levels recorded in some locations (average or median levels of formaldehyde in outdoor air
typically range from 0.4 to 10 ng/m3), and it would be expected that levels in indoor air would
This document is a draft for review purposes only and does not constitute Agency policy.
lvi DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
exceed this concentration in many situations. However, as the RfC is interpreted to be without
appreciable risk, even in sensitive subgroups, it is important to note that the potential for health
effects in individuals at concentrations between the RfC (0.007 mg/m3) and levels at which health
effects have been observed in the available population studies (~35-40 |ig/m:i] is unknown.
Although the RfC is designed to apply to exposures over a lifetime, the relevant window of
exposure for some of the effects observed in the contributing studies may be less than a lifetime.
Sensory irritation is an immediate response to reactive compounds like formaldehyde. The
relevant window of exposure for effects on asthma outcomes is also less than lifetime, although the
time frame for the control of asthma symptoms (i.e., a few weeks) is different than that for the
prevalence of current asthma symptoms or a decrease in pulmonary function (i.e., the past
12 months). In addition, the relevant window of exposure for female reproductive or
developmental outcomes is from conception to the end of the pregnancy.
Overall confidence in the RfC is high, based on high confidence in the composite set of
studies used to derive the RfC, high confidence in the completeness of the literature database
supporting the judgment that formaldehyde causes the adverse effects identified (although
uncertainties remain for other potential health effects), and medium-to-high confidence in the
derivation of the candidate RfC numerical values.
ES.3.4. Quantitative Estimate of Carcinogenic Risk from Inhalation Exposure:
The inhalation unit risk (IUR) is 1.1 x 10"5 per |J.g/m3, which is an upper-bound estimate of
the increased lifetime risk of cancer from inhaling 1 ng/m3 of formaldehyde for 70 years. The
estimate is based on an estimate of increased risk for NPC, for which evidence demonstrates that
formaldehyde inhalation causes this type of cancer in humans. The IUR does not incorporate a unit
risk estimate for myeloid leukemia (also for which the evidence demonstrates that formaldehyde
inhalation causes this type of cancer in humans) presented in Section 2 of this assessment because
the interpretation of the published exposure-response modelling results was deemed too
uncertain.6 This estimate also does not incorporate risk from sinonasal cancer for which the
evidence demonstrates that formaldehyde inhalation exposure causes this type of cancer in
humans given appropriate exposure circumstances, as amenable data were unavailable. The IUR is
based on the modeling results of the association of cumulative formaldehyde exposure with NPC
mortality in an occupational cohort followed by the National Cancer Institute (Beane Freeman et al..
20131. The regression coefficient from the dose-response model (log-linear models) was applied to
age-specific cancer incidence rates from the National Cancer Institute's (NCI) Surveillance,
Epidemiology, and End Results (SEER) database using life-table methods to estimate the upper
6 A charge question is provided for external peer review asking for advice regarding the development of a unit
risk estimate for myeloid leukemia and how, if at all, the unit risk estimate might inform the quantification of
risk for cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
lvii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
bound on the extra risk7 expected at a formaldehyde concentration of 0.1 ppm. The IUR is
expressed as the upper-bound number of extra cancer cases estimated for a lifetime inhalation
exposure to 1 |ig/m:i. This estimate, based on a human study, was found to be within the range of
estimates derived using experimental animal data, including estimates that incorporate BBDR
modeling approaches using available mechanistic evidence (see Section 2.2). The estimated IUR for
total cancer prior to any age adjustments is 6.4 x 10"6 per ng/m3 (see Table ES-2). EPA guidelines
recommend that ADAFs be used when estimating the risk of NPC from childhood inhalation
exposures to formaldehyde because the NPCs are judged to be due, at least in part, to a mutagenic
MOA. In the absence of information to support a chemical-specific age adjustment factor, EPA's
default ADAFs should be applied. Thus, the unit risk estimate was adjusted using age-dependent
adjustment factors (ADAFs) to address expected increased susceptibility from early-life exposures
(see Table ES-1).
Overall confidence in the IUR is medium. The availability of suitable human data from
which to derive unit risk estimates eliminates one of the major sources of uncertainty inherent in
most unit risk estimates—the uncertainty associated with interspecies extrapolation. The NCI
longitudinal cohort study used as the basis for the preferred unit risk estimate is a well-conducted
study for the purposes of deriving unit risk estimates and there is high confidence in the study's
results. However, it was the only independent study with adequate exposure estimates for the
derivation of unit risk estimates.
There are some uncertainties that could result in an underestimation of the IUR. An
important uncertainty is the inability to derive unit risk estimates for all cancer sites with
conclusions of evidence demonstrates that formaldehyde inhalation exposure causes these cancer
types given appropriate exposure circumstances, resulting in an underestimate of the IUR, Since
industrial workers are healthier than the general population overall, the unit risk estimates derived
from the NCI worker cohort data could underestimate the cancer risk for the general population to
an unknown, but likely small, extent Given the high survival rates for NPC, cancer incidence risk
estimates were calculated using the dose-response relationships from the NCI mortality study to
reduce the potential to underestimate the unit risk. However, the calculation required certain
assumptions, thus, the estimates may under- or overpredictthe true risk by an amount expected to
be relatively small.
Because a mutagenic MOA was established for NPC, the IUR was calculated using linear low-
dose extrapolation from the 95% lower bound on the exposure level associated with the extra risk
level serving as the benchmark response, which is considered to be a plausible upper bound on the
risk at lower exposure levels. The low dose extrapolation is a source of uncertainty potentially
7 Extra risk is defined as (Rx - Ro)/(l - Ro), where Rx is the lifetime risk in the exposed population and Ro is
the lifetime risk in an unexposed population; it is the added risk applied to the portion of the population that
did not show background tumors.
This document is a draft for review purposes only and does not constitute Agency policy.
lviii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
resulting in overestimation of the IUR, possibly by a substantial (e.g., over an order of magnitude)
extent
ES.4 SUSCEPTIBLE POPULATIONS AND LIFESTAGES
Overall, the most extensive research on the health effects of inhaled formaldehyde and
susceptible groups indicates a greater susceptibility among children to respiratory disease,
manifested as reduced pulmonary function, increased prevalence of current asthma, and greater
asthma severity (reduced asthma control). More research is needed to investigate the role of sex,
race, nutrition, exercise, and coexposures that may modulate susceptibility to formaldehyde
toxicity. Increased early-life susceptibility for cancer is assumed because of the mutagenic MOA for
NPC carcinogenicity. Health status and disease, particularly related to the respiratory system, are
likely to be modifying factors of formaldehyde toxicity. Studies suggest that asthmatics are more
susceptible than nonasthmatics to declines in respiratory function following formaldehyde
exposure. Based on multiple mechanistic studies of respiratory hypersensitivity, it also appears
likely that persons with preexisting respiratory allergies would be more sensitive to the respiratory
health effects of formaldehyde exposure, although the data informing potential associations
between more generalized atopy and respiratory effects in the available human studies were
inconsistent Experimental animal studies and occupational studies indicate that nasal lesions are
more severe among individuals with prior nasal damage which could result in heightened
susceptibility to the development of nasal cancer following formaldehyde exposure.
In addition, epidemiological and toxicological studies identify female reproductive or
developmental toxicity as a hazard of formaldehyde exposure. At this time, it is not clear whether
increased time to pregnancy and spontaneous abortion rates seen in occupationally exposed
women are due to reproductive system toxicity or to toxicity to the developing fetus. Finally,
reproductive toxicity in males has been associated with formaldehyde inhalation, although this
association has only been tested in well-conducted studies of rodents at very high formaldehyde
concentrations.
This document is a draft for review purposes only and does not constitute Agency policy.
lix DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
1. HAZARD IDENTIFICATION
Potential health hazards from the inhalation of formaldehyde were evaluated across
multiple health domains, including sensory irritation; reduced pulmonary function; immune system
effects, focusing on allergies and asthma; respiratory tract pathology; nervous system effects;
reproductive and developmental toxicity; and cancer. Research results for several cancer sites
were evaluated, specifically cancers of the upper respiratory tract ([URT]; i.e., nasopharyngeal
cancer, sinonasal cancer, cancers of the oropharynx and hypopharynx, laryngeal cancer) and of the
lymphohematopoietic system (i.e., Hodgkin lymphoma, multiple myeloma, myeloid leukemia,
lymphatic leukemia). The evidence regarding the potential for formaldehyde exposure to cause
other cancer types (i.e., lung non-Hodgkin lymphoma, brain, bladder, colon, pancreas, prostate,
skin) were not systematically evaluated because only a few studies reported analyses for these
cancer sites (see Appendix A.5.9 for detail). Multiple health endpoints were evaluated within each
of these hazard domains using primary research studies in human populations and experimental
animals and in supporting mechanistic studies. The mechanistic studies informing all potential
respiratory effects were considered and analyzed together due to the potential interdependencies
of the mechanisms involved (see Appendix A.5.6). The majority of studies evaluating the potential
toxicity of formaldehyde inhalation exposure have focused on effects at the portal of entry (POE),
primarily the URT, with less research available to inform potential systemic, or nonrespiratory,
effects. Thus, the synthesis of the evidence for each identified health endpoint is provided in
Section 1.2 for potential respiratory system-related effects (including cancer and noncancer
endpoints) and in Section 1.3 for potential nonrespiratory health effects.
1.1. SUMMARY OF USES, HUMAN EXPOSURE, AND TOXICOKINETICS
1.1.1. Chemical Properties and Uses of Formaldehyde
Formaldehyde (CASRN 50-00-0) is an aliphatic aldehyde noted for its reactivity and
versatility as a chemical intermediate. At room temperature, pure formaldehyde is a colorless gas
with a strong pungent, and irritating odor. Formaldehyde is readily soluble in water, alcohols,
ether, and other polar solvents. Due to its chemical properties (see Appendix A.1 for additional
details), formaldehyde is widely used in both commercial and industrial settings. Based on EPA's
Chemical Data Reporting, the national production volume for formaldehyde was 3.9 billion lb/yr in
2011 and between 1 and 5 billion lbs/yr for 2012 through 2015
(https: //chemview. epa. gov/chemview/#).
This document is a draft for review purposes only and does not constitute Agency policy.
1-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
Products containing formaldehyde are widespread in industry and in the home.
Approximately 55% of the consumption of formaldehyde is in the production of industrial resins
fNTP. 20101. Formaldehyde is used in plywood adhesives, surface coatings, molding compounds,
laminates, phenolic thermosetting, resin curing agents, and other products fWHO. 19891.
Formaldehyde is used in smaller quantities for the preservation and embalming of biological
specimens. It is also used as a germicide, an insecticide, and a fungicide in some products. Some
industries with the greatest potential for exposure to the workforce include health services,
business services, printing and publishing, chemical manufacturing, garment production, beauty
salons, and furniture manufacturing flARC. 19951.
1.1.2. Exposure to Formaldehyde
Generally, formaldehyde levels are higher in the indoor environment than in ambient air.
Indoor sources of formaldehyde in air include building materials and household products
(e.g., volatilization from pressed wood products, carpets, fabrics, insulation, permanent-press
clothing, latex paint), as well as household sources of combustion (e.g., gas burners, kerosene
heaters, cigarettes) fWHO. 20101. Median indoor air concentrations in some European countries in
both commercial and residential buildings ranged from 10 to 50 |J.g/m3 (Sarigiannis etal.. 2011:
Salthammer etal.. 20101. Indoor average formaldehyde concentrations reported since 2000 in U.S.
and Canadian conventional homes ranged from 12 to 39 |J.g/m3 (see Appendix A.l.2). For example,
a fairly large study of 398 homes in Los Angeles, CA, Houston, TX, and Elizabeth, NJ, between 1999
and 2001 reported formaldehyde levels of 22 ± 7.1 |ig/m3 (Weisel et al.. 20051. Higher levels are
found in mobile homes and trailers. In 2018, annual site averages of formaldehyde concentrations
outdoors ranged from 0.25 - 11.06 |J.g/m3 (0.20 - 9.01 ppb), with an overall annual site average
concentration of 2.97 |J.g/m3 (2.42 ppb) (EPA's Ambient Monitoring Archive for HAPs, which
includes data from the Air Quality System database and other data sources at
https://www.epa.gov/amtic/amtic-air-toxics-data-ambient-monitoring-archive). A full summary of
the information on formaldehyde exposure is included in Appendix A.1.2. Under the National-Scale
Air Toxics Assessment (NATA) program, EPA has conducted an emissions inventory for a variety of
hazardous air pollutants (HAPs), including formaldehyde. NATA uses the emissions inventory data
to model nationwide air concentrations/exposures (https://www.epa.gov/national-air-toxics-
assessment). The most recent NATA data are for 2014. The results of the 2014 ambient air
concentration modeling for formaldehyde suggest that county mean air levels range from 0.1 to
2.78 |J.g/m3 with a national mean of 1.3 |ig/m3 [personal communication to EPA (Palma. 2018)].
1.1.3. Toxicokinetics of Formaldehyde
Formaldehyde is a respiratory irritant for which the human body has developed several
detoxification and removal processes, especially at the site(s) of first contact (i.e., nasal passages for
inhalation). Thus, this discussion of the toxicokinetics of inhaled formaldehyde at the POE is
organized according to the most likely sites of first contact between inhaled formaldehyde and
This document is a draft for review purposes only and does not constitute Agency policy.
1-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
biological materials, in the context of the known anatomy and potential elimination processes of the
respiratory tract tissues. A more comprehensive summary of what is known about the absorption,
distribution, metabolism, and excretion of inhaled formaldehyde is provided in Appendix A.2. This
section also includes a discussion of published analyses of the potential impact of endogenous
levels of formaldehyde produced during normal cellular metabolism on the toxicokinetics of
inhaled formaldehyde.
Distribution of Inhaled Formaldehyde
Much of what is known about the uptake and distribution of formaldehyde is based on
experimental animal studies, primarily in monkeys and rats. Several of the key considerations for
evaluating the toxicokinetics of inhaled formaldehyde at the POE in the rat nose are represented
schematically in Figure 1-1. Species differences in the structure of the airways and breathing
patterns, as well as the composition of the surface epithelium at various nasal locations, are
important considerations when interpreting results in experimental animals and extrapolating
observations to humans. While the nasal passages in humans are generally similar to those in other
mammalian species, one key difference is that humans and nonhuman primates have nasal
passages adapted for both oral and nasal (oronasal) breathing, as opposed to obligate nasal
breathing in rodents. A second key difference regards the shape and complexity of the nasal
turbinates, with relatively simple shapes in humans, and complex, folded patterns in rodents. In
general, these differences provide better protection of the rodent lower respiratory tract against
inhaled toxicants than is provided to the human lower respiratory tract (Harkema et al.. 20061.
Uptake of formaldehyde (defined as retention within the respiratory tract tissue), based on
rough estimates determined from the amount of formaldehyde removed from the air, indicates that
the vast majority of formaldehyde is removed from inhaled air by the upper respiratory tract (URT)
in monkeys fCasanova etal.. 1991: Monticello etal.. 19891. dogs fEgle. 19721 and rats fKimbell et
al.. 2001b: Chang etal.. 1983: Heck etal.. 1983: Kerns etal.. 1983). Further, dosimetric modeling
studies in humans have shown close agreement with observations of exposed rodents, namely, that
90-95% of inhaled formaldehyde is deposited in the URT (Yang etal.. 2020: Kimbell etal.. 2001b:
Overton etal.. 2001: Subramaniam et al.. 1998). Most recently, Yang et al. (2020) conducted
inhalation studies in 120 (70 female and 50 male) healthy human volunteers and measured their
absorption of formaldehyde and selected volatile organic compounds. The absorbed formaldehyde
Cinh - Cexh was seen to be linearly related to Cmh- The slope of this straight line, which expresses a
mean deposition rate for the range of concentrations from 2 ppb to 18 ppb was determined to be
0.97, indicating that most of the inhaled formaldehyde is absorbed, on average, at these low
concentrations. This is consistent with prior understanding regarding the extent of formaldehyde
absorbed. A detailed description of dosimetry modeling efforts in humans, monkeys, and rats is
provided in Appendix B.2.2. As demonstrated in monkeys and rats, and as modeled in humans, a
concentration gradient of inhaled formaldehyde follows an anterior-to-posterior distribution, with
high concentrations of formaldehyde distributed to squamous, transitional, and respiratory
This document is a draft for review purposes only and does not constitute Agency policy.
1-3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
epithelium, and less uptake by olfactory epithelium. Except under exercise conditions or with
exposure to high formaldehyde concentrations, very little formaldehyde reaches more distal sites
such as the lung. The possibility that more extensive distribution to the LRT may occur when
people are regularly breathing through the mouth or when they have an upper respiratory tract
infection has not been directly investigated (see Sections 1.2.2 and 1.2.3 for discussions of the
available, indirect evidence). Likewise, no specific toxicokinetic studies focusing on the possibility
of inhaled formaldehyde distributing to the developing fetus were identified; however, based on
current understanding of its reactivity and distribution, it is unlikely that inhaled formaldehyde
would reach the developing fetus.
Asgharian et al. (2012.) developed a pharmacokinetic model for transport of formaldehyde
and other gases in the human lung, across the air-tissue interface towards arterial blood, that
explicitly incorporates information on partition coefficient, metabolism and tissue reactivities
(considered as saturable and first-order clearance pathways). This was a substantial improvement
over the approach in Overton et al. (2001) that was used for providing formaldehyde dose to the
lung in the Conolly et al. (2004) model for extrapolating cancer risk to the human; Overton et al.
(2001) did not model the tissue kinetics [and hence the systemic dose] but assumed a constant
mass transfer coefficient. There are several noteworthy results from this paper:
• Surface flux rates of formaldehyde appeared to be predictive of local tissue concentrations.
• 97% of the inhaled formaldehyde was absorbed.
• Formaldehyde did not penetrate beyond 60 [im of tissue depth in any breathing scenario,
thus predicting that systemic penetration is not likely to take place.
• This model predicted a 25% higher tracheal mass flux of formaldehyde, and
correspondingly lesser flux to the deep lung, than Overton et al. (2001). It is important to
note that this quantitative result is not relevant to the dose-response modeling in this
assessment (see Sections 2.1.1 and 2.2.1). While the extrapolation model by Conolly et al.
(2004) uses formaldehyde dose to the human lung as input, this model is not used in this
assessment and lung cancer is not identified as a hazard (see Section 1.2.5).
This document is a draft for review purposes only and does not constitute Agency policy.
1-4 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Rat nose [formaldehyde
(in red), squamous
epithelium (se),
respiratory epithelium
[re), nasal turbinates with
olfactory epithelium (NT-
oe) or with re (NT-re),
cribiform plate (CP), and
olfactory bulb (OB)]
Inspired air and
formaldehyde (red)
Mucus
Cilia
Epithelium
[epithelial cells
(EC) and goblet
cells (GC)]
Basement membrane
Lamina propria
[systemic circulation
(SC) and lymphoid
tissue (NALT)]
Gradient
formaldehyde
concentration
Nasopharynx (to lower
respiratory tract)
•fcNALT*
Figure 1-1. Schematic of the rat upper respiratory tract depicting the gradient
of formaldehyde concentration formed following inhalation exposure, both
from anterior to posterior locations, as well as across the tissue depth.
Modeling based on observations in rodents predicts a similar pattern of distribution in humans. Drawing
is based in part on images by NRC (2011) and Harkema et al. (2006). Note: Other components (e.g., naris,
transitional epithelium) have been omitted for clarity.
1 Corley et al. (2015.) developed integrated air and tissue transport models for predicting
2 airway region-specific tissue dose of tobacco smoke in the rat and human, upper and lower,
3 respiratory tracts. Their approach coupled CFD models for gas transport in the airways with airway
4 region-specific PBPK models for tissue transport, and included realistic, transient breathing
5 patterns. Although the paper was aimed at tobacco smoke, results were separately provided for the
6 acrolein, formaldehyde and acetaldehyde constituents. Metabolic interactions and reactions were
7 described by clearance through a saturable enzymatic pathway, a first order pathway representing
8 intrinsic tissue reactivity, and a first order binding to DNA to form DPX. Details on regional
9 distribution of metabolic enzymes and local blood perfusion rates were incorporated and the
10 simulations were carried out until breath-by-breath, steady-state kinetics was achieved in all
This document is a draft for review purposes only and does not constitute Agency policy,
1-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Toxicological Review of Formaldehyde—Inhalation
tissues. These calculations of regional tissue concentrations as a function of tissue depth are a
substantial improvement over other dosimetry models that could model only airway wall flux rates
of formaldehyde. The primary results relevant to this assessment were as follows:
• Formaldehyde does not penetrate deep into epithelial or subepithelial tissue even in the
olfactory region where the penetration was greatest, and therefore does not transport
directly to the systemic blood circulation at moderate exposure concentrations.
• As with prior formaldehyde rat dosimetry models, their model predicted greatest initial
uptake rates of the gas in the anterior respiratory nasal region. However, the uptake was
greater in the anterior dorsal olfactory epithelium when area under the curve (AUC)
concentrations were calculated by integrating the concentration profile over time of
exposure as well as depth normal to the air-tissue interface under more realistic transient
breathing profiles.
• The simulation covered only oral inhalation in the human because the purpose of the
research was to investigate uptake from cigarette smoke. In the human, oral and laryngeal
tissues received the greatest local tissue dose. Overall formaldehyde absorbed was 97% at 2
and 6 ppm and about 94% at 15 ppm exposure concentrations.
• Formaldehyde surface fluxes did not correlate well with local time dependent tissue
concentration AUCs for all nasal tissues in the rat; the AUCs were significantly higher in the
olfactory region than would be predicted by surface flux alone. This finding was counter to
the conclusion in Asgharian et al as detailed above.
The modeling approach in Corley et al. (2015) could potentially make a tangible difference
in extrapolated dose over that computed by solely surface flux-based models in the case of reactive
gases that result in adverse effects in the rat olfactory region. Because the findings of formaldehyde
induced cancer or non-cancer effects in the URT of the rat are not observed in the olfactory region
(see Section 1.2), this modeling approach by Corley et al. (2015.) was not applied.
As inhaled formaldehyde enters the URT, it interacts with the mucociliary apparatus, the
first line of defense against inhaled materials in the nose. In nasal mucus, most of the formaldehyde
is rapidly converted to methanediol (~99.9%) and a minor fraction remains as free formaldehyde
(~0.1%) (Bogdanffv et al.. 1986). Inhaled formaldehyde induces mucostasis and ciliastasis in the
rat that extends from anterior to posterior regions of the nasal cavity depending on the
concentration and duration of exposure (Morgan et al.. 1986a). Thus, inhalation of higher
concentrations can potentially slow clearance mechanisms and increase the proportion of
formaldehyde that is available to react with cellular components or that is distributed to epithelium
and systemic circulation. Whether mucostasis or ciliastasis is induced with longer exposure
duration to low levels of formaldehyde is not known. Methanediol is assumed to be better able to
penetrate the tissues while free formaldehyde reacts with macromolecules. It is assumed that the
equilibrium is rapid, hence that the methanediol:free formaldehyde equilibrium ratio is maintained
(Fox. 1985). Formaldehyde levels are reduced through interactions with components of the mucus
and through mucociliary clearance, through reactions with cellular materials at the plasma
This document is a draft for review purposes only and does not constitute Agency policy.
1-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
membrane of the respiratory epithelium, via interactions with glutathione (GSH) and other
macromolecules in the intracellular and extracellular space, through localized metabolism and
conjugation reactions, and through reversible interactions with intracellular materials. These
processes result in the formation of a gradient of formaldehyde across the tissue space, with the
greatest formaldehyde concentration at the apical surface of the mucosa, and the lowest levels of
formaldehyde at deeper components of the tissue, such as the nasal-associated lymphoid tissues
(NALT) and blood vessels.
Several uncertainties exist regarding the transition of inhaled formaldehyde from the
mucociliary layer to the underlying epithelium. Although direct experimental evidence is lacking,
the biochemical properties of formaldehyde make it likely that inhaled formaldehyde (in the
hydrated or anhydrated form) undergoes passive transport, via simple diffusion, across biological
membranes. As a result, higher extracellular formaldehyde levels would be expected to result in
increased diffusion into the cell owing to the concentration gradient formed. However, this
concentration gradient may be affected by endogenous formaldehyde levels, since in humans, as in
other animals, formaldehyde is an essential metabolic intermediate in all cells (Thompson etal..
20091.
Two groups of researchers, Schroeter et al. (2014) and Campbell Jr et al. (2020) developed
toxicokinetic models of formaldehyde uptake that incorporate the production of endogenous
formaldehyde in nasal tissue. Schroeter et al. (2014) revised the fluid dynamic modeling by Kimbell
et al. (2001a: 2001b) to explicitly include tissue pharmacokinetics. The Campbell Jr et al. (2020)
model simulates observed data for formaldehyde-induced DNA mono-adducts (N2-hydroxymethyl-
dG) using exogenous and endogenous formaldehyde adduct data published after 2010. This model
was based on a modification of Andersen et al. (2010) which simulated formaldehyde-induced
DNA-protein cross-links (DPX). Both models, Schroeter et al. (2014) and Campbell Jr et al. (2020),
predicted the endogenous formaldehyde to reduce uptake of inhaled formaldehyde from the air
phase to the tissue compartment
In the first model, net desorption of the gas was predicted at exposure concentrations below
lppb in humans. In the second model developed only for the rat, the model was calibrated with the
restriction that formaldehyde absorption in the nose occurs only at exposure concentrations above
0.3 ppm based upon the available experimental DNA adduct data, and the model predicted that the
inhalation rate must exceed the tissue clearance rate for formaldehyde to be absorbed by the tissue.
The results from both these pioneering projects add to our characterization of uncertainties related
to formaldehyde dose-response at low exposures; at sufficiently low levels of exogenous
formaldehyde, the contribution of endogenous formaldehyde could become significant.
Additionally, when including endogenous formaldehyde in an analysis it is important to incorporate
considerations of the large variability in these levels. [The impact of this variability was apparent,
for example, from the individual animal data on DNA adducts formed by formaldehyde in Swenberg
et al. (2013.), kindly made available to EPA by the authors. A number of animals in these data had
This document is a draft for review purposes only and does not constitute Agency policy.
1-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
very high endogenous levels of these adducts; in these animals, the total (endogenous plus
exogenous) internal dose even at a low inhaled exposure concentration of 2 ppm, as measured by
the level of DNA adducts, was comparable to the mean total internal dose measured in the group of
animals exposed at 10 ppm. At this dose, considerable carcinogenicity was observed in animal
bioassays in other studies.] There are also crucial uncertainties in the measurements of free
endogenous formaldehyde levels as highlighted by Campbell Jr etal. (2020) and discussed further
in Appendix A.2.
EPA evaluated the Schroeter et al. (2014) model and determined that the model predicts
any external exposure to cause some, albeit very small, increase in formaldehyde tissue
concentration over background levels. EPA's evaluation, as detailed in Appendix A.2, pointed to
critical uncertainties in model assumptions; therefore, this model was not directly used in EPA
calculations. However, it was seen that EPA benchmark concentrations based on formaldehyde as a
dose metric in Sections 2.1.1 and 2.2.1 do not change appreciably when results from Schroeter et al.
(2014) are used.
Extrapolation of results in Campbell Jr etal. (2020) to humans is not possible because the
data and the model are specific to rats. These models and a discussion of studies of formaldehyde
distribution in the URT are discussed further in context of the toxicokinetics of inhaled
formaldehyde in Appendix A. 2.
Metabolism, Binding, and Removal of Inhaled Formaldehyde
In the URT, formaldehyde is predominantly metabolized by glutathione-dependent class III
alcohol dehydrogenase (ADH3) and by a minor pathway involving aldehyde dehydrogenase 2
(ALDH2) to formate. Formate can either enter the one-carbon pool leading to protein and nucleic
acid synthesis or is further metabolized to CO2 and eliminated in expired air or excreted in urine
unchanged. ADH3 and ALDH2 show region-specific differences in distribution in the respiratory
and olfactory mucosa, and higher levels of ADH3 activity have been reported in the cytoplasm of the
respiratory and olfactory epithelial cells of rats and in the nuclei of olfactory sensory cells, as
compared to other regions of the nasal mucosa fKeller etal.. 19901. The presence of areas of high
enzyme activity highlights a significant barrier to the penetration of inhaled formaldehyde beyond
the respiratory epithelium.
Formaldehyde can interact with macromolecules either by noncovalently binding to
glutathione (GSH), tetrahydrofolate (THF), or albumin in nasal mucus or by covalently forming
DNA-protein crosslinks (DPXs), DNA-DNA crosslinks (DDCs), hydroxymethyl-DNA (hm-DNA)
adducts (see Appendix A.2), or protein adducts, such as N6-formyllysine (Edrissi etal.. 2013b:
Edrissi etal.. 2013al. In rats and monkeys, a concentration-dependent increase in DPX formation is
observed in nasal passages. Metabolic incorporation studies with 14C-formaldehyde have shown
both covalent binding and metabolic incorporation in nasal tissues (Casanova and Heck. 1987:
Casanova-Schmitz etal.. 1984b). Inhaled formaldehyde induces a concentration-dependent
This document is a draft for review purposes only and does not constitute Agency policy.
1-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
increase in N2-hydroxymethyl deoxyguanosine (N2-hm-dG) adducts, another form of formaldehyde-
induced covalent DNA modification, in the nasal passages of monkeys and rats. Recently, analytical
methods have been developed that can distinguish between N2-hm-dG adducts from exogenous
(inhaled) formaldehyde and N2-hm-dG adducts from endogenous formaldehyde fLu etal.. 2012: Lu
etal.. 2011: Moeller etal.. 2011: Lu etal.. 2010al. For example, an increase in exogenous
formaldehyde adducts has been observed in rat nasal tissue at 0.7-15 ppm (0.86-18.45 mg/m3)
formaldehyde without any significant increases in endogenous adducts following a single 6-hour
exposure (Lu etal.. 20111 or at 10 ppm (12.3 mg/m3) after exposure to formaldehyde for 1 or
5 days (6 hrs/day) fLu et al.. 2010al. However, in a more recent study with a lower detection limit
for adducts and testing lower formaldehyde exposure levels, Leng et al. (2019) did not observe an
increase in exogenous hmDNA adducts or DPXs, including in nasal and respiratory tissues as well as
at systemic sites (e.g., bone marrow), at formaldehyde levels of 0,1, 30, or 300 ppb (up to
0.37 mg/m3) after exposure for 28 days. The lack of detectable exogenous adducts in the URT at
0.3 ppm (0.37 mg/m3) helps to inform the evolving understanding of formaldehyde-induced DPX at
lower concentrations, which would benefit from additional study. DNA monoadducts (Yu etal..
2015a: Lu etal.. 2011: Moeller etal.. 2011: Lu etal.. 2010al and DPXs fLai etal.. 20161 derived from
exogenous formaldehyde were detectable in nasal tissues, but not in distal tissues (including the
bone marrow), of experimental animals exposed by inhalation, supporting that exogenous
formaldehyde is not systemically distributed. Also, toxicokinetic studies showed that labeled
carbon from inhaled formaldehyde measured in bone marrow of rats was the result of metabolic
incorporation from the 1-Carbon (1C) pool, not covalent binding, further supporting the lack of
transport of formaldehyde or metabolites of formaldehyde to the distal tissues (Casanova-Schmitz
etal.. 1984bl. Finally, inhalation exposure to formaldehyde does not appear to alter blood
formaldehyde levels (approximately 0.1 mM across different species), suggesting that inhaled
formaldehyde is not significantly absorbed into blood fKleinniienhuis etal.. 2013: Casanova etal..
1988: Heck etal.. 19851.
The toxicokinetics of formaldehyde may be influenced by certain formaldehyde-related
effects, such as mucociliary clearance (Morgan etal.. 19831. reflex bradypnea (rodents only) and
corresponding reductions in minute volume (Chang and Barrow. 1984: Chang etal.. 19811. and
dynamic changes in tissue structure fKamata etal.. 19971. all of which have the potential to
modulate formaldehyde uptake and clearance. For example, during repeated inhalation exposure
to formaldehyde, mice but not rats lower their minute volume thereby restricting the intake of the
gas (Chang and Barrow. 1984: Chang etal.. 1981). which may impact dosimetric adjustment if the
dose-response results from these studies are extrapolated to humans. Exposure to formaldehyde
can also cause a perturbation of ADH3-dependent pathways involved in cell proliferation (Nilsson
etal.. 2004: Hedbergetal.. 20001. protein modification and cell signaling (Que etal.. 20051. S-
nitrosoglutathione (GSNO) metabolism, and deregulation of nitric oxide-dependent pathways
fThompson etal.. 20101. In rats exposed by inhalation to high concentrations of formaldehyde, a
This document is a draft for review purposes only and does not constitute Agency policy.
1-9 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
rapid GSH depletion can occur, which may result in more free formaldehyde available for covalent
binding and a decrease in metabolic incorporation fCasanova and Heck. 19871.
Assumptions based on what is known about the distribution and metabolism of
formaldehyde and its detoxification products allow inferences to be made about how inhaled
formaldehyde is eliminated as CO2 in expired air or in various forms in urine. Approximately
one-third of inhaled formaldehyde is estimated to be removed in the URT mucus (Schlosser. 1999).
It is expected that the majority of this formaldehyde would be removed from the URT via
esophageal clearance and excreted in urine in various forms. A large amount of inhaled
formaldehyde penetrating the mucociliary layer of the URT is metabolized in the nasal cavity, giving
rise to formate, which can be excreted in urine. Part of this formate may also be further oxidized
and eliminated in the exhaled breath as CO2. Some formaldehyde is incorporated into the 1C pool
and repurposed for protein and nucleic acid synthesis.
1.2. SYNTHESIS OF EVIDENCE FOR EFFECTS ON THE RESPIRATORY
SYSTEM
Research on several noncancer respiratory health effects was synthesized for the following
health domains: sensory irritation (see Section 1.2.1), pulmonary function (see Section 1.2.2),
immune system effects focusing on allergies and asthma (see Section 1.2.3), and respiratory tract
pathology (see Section 1.2.4). Synthesis of the evidence relevant to potential carcinogenicity at
respiratory sites focused on cancers in the upper respiratory tract ([URT]; see Section 1.2.5), as less
has been reported concerning cancer associations at other respiratory sites (see Appendix A.5.9 for
details).
As previously described, inhaled formaldehyde is highly reactive at the portal of entry
(POE), that is, nose and upper airways, which results in alterations to the local tissues that could
give rise to respiratory system health effects. The potential noncancer effects, in particular, involve
many of the same biological processes; thus, a high degree of overlap across the mechanistic
changes underlying these responses is expected. Similarly, because the potential respiratory health
effects are interrelated, effects on one outcome may affect others. Accordingly, an overarching
evaluation of the mechanistic information pertinent to any or all potential noncancer respiratory
system health effects (some of which is relevant to carcinogenicity) was performed
(see Appendix A.5.6). The primary mechanistic conclusions drawn from this overarching
evaluation are summarized in the MOA analyses in Sections 1.2.1-1.2.4. Section 1.2.3 includes a
discussion expanded to include mechanistic changes in nonrespiratory tissues that might relate to
respiratory system health effects, although these findings are also relevant to the nonrespiratory
(systemic) health effects reviewed in Section 1.3.
Finally, an essential component of the analysis of potential carcinogenicity at respiratory
sites involves evaluating whether inhaled formaldehyde causes genotoxicity or mutagenicity.
This document is a draft for review purposes only and does not constitute Agency policy.
1-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Because abundant information exists on this topic, the data are comprehensively described in
Appendix A.4, with the primary conclusions summarized in Section 1.2.5. Some of the conclusions
from the genotoxicity evidence analyzed in Appendix A.4 are also relevant to interpretations
regarding potential cancers at nonrespiratory (distal) sites in Section 1.3.3).
1.2.1. Sensory Irritation
This section describes research on formaldehyde inhalation and sensory irritation in
experimental and observational studies in humans. Although not formally evaluated for this
review, formaldehyde inhalation-induced sensory irritation in animals is a well-established
phenomenon (Nielsen et al.. 1999: Barrow etal.. 1983: Chang etal.. 1981: Kane and Alarie. 1977).
Formaldehyde has been found to be a sensory irritant of the eyes and respiratory tract in
several epidemiological studies causing mild to severe symptoms, including itching, stinging, and
watering eyes; sneezing and rhinitis; sore throat; coughing; and bronchial constriction. Symptoms
of eye irritation were reported at lower concentrations than symptoms of the nose or throat. Many
epidemiology studies evaluated symptoms of irritation among residents exposed to formaldehyde
in their homes, workers involved in the production or use of formaldehyde products, and anatomy
students participating in the dissection of formaldehyde-preserved cadavers. In addition, data from
several controlled human exposure studies are available that evaluated acute responses among
healthy or asthmatic volunteers during rest or exercise (see Table 1-1). The controlled human
exposure studies showed that the irritant response to formaldehyde is an immediate phenomenon
apparent at concentrations of 0.1 mg/m3, the lowest concentration evaluated, and higher. The
irritation resolves when exposure is removed (Krakowiaketal.. 1998: Sauder etal.. 1986: Andersen
and Molhave. 1983: Andersen. 19791. Concentration was related to both prevalence and severity of
symptoms. In addition, a large variability in sensitivity to the irritant properties of formaldehyde at
specific concentrations was observed f Mueller et al.. 2 013: Berglund etal.. 20121. Because of the
wide variability in responses, it has been difficult for experimental studies to characterize the
exposure-response relationship in the lower range of concentrations experienced by the general
population. Sensory irritation is understood to occur as a result of direct interactions of
formaldehyde with cellular macromolecules in the nasal mucosa leading directly or indirectly to
stimulation of trigeminal nerve endings located in the respiratory epithelium.
Studies in humans provide robust evidence of sensory irritation based on the controlled
human exposure studies and observational epidemiology studies, and this effect also is well
described and accepted across a range of experimental animal species [robust). Further, there is an
established MOA for this well-studied health effect, based primarily on mechanistic evidence in
experimental animals, and this MOA is interpreted to be operant in humans. Overall, a judgment
was drawn that the evidence demonstrates that inhalation of formaldehyde causes sensory
irritation in humans, given the appropriate exposure circumstances. The primary support for this
conclusion is based on residential studies with mean formaldehyde concentrations >0.05 mg/m3
This document is a draft for review purposes only and does not constitute Agency policy.
1-11 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
(range 0.01 to approximately 1.0 mg/m3) and controlled human exposure studies testing responses
to concentrations 0.1 mg/m3 and above.
Literature Search and Screening Strategy
The identification of relevant epidemiology studies (i.e., both observational and controlled
exposure studies) on sensory irritation included systematic literature searches in PubMed and Web
of Science through September 2016 (see Appendix A.5.2 for search details), and a systematic
evidence map updating the literature through 2021 (see Appendix F). Based on the extensive
database of research studies on relevant apical endpoints in humans after formaldehyde exposure,
systematic searches for studies of sensory irritation in experimental animals were not conducted.
However, mechanistic data informing this health effect were identified and evaluated as part of the
overarching review of mechanistic data relevant to potential respiratory health effects (see
Appendix A.5.6 for details). Epidemiological studies describing reports of sensory irritation based
on questionnaire responses or objective measures, such as eye blink frequency or conjunctival
redness, were included. Articles reporting on case reports, illness investigations, and surveillance
studies were not included because the studies were not designed to derive an effect estimate of the
association between measures of irritation and formaldehyde exposure. The bibliographic
databases, search terms, and specific strategies used to search them are provided in Appendix A.5.2
and A.5.6, as are the specific PECO criteria. Literature flow diagrams summarize the results of the
sorting process using these criteria and indicate the number of studies that were selected for
consideration in the assessment through 2016 (see Appendix F for the identification of newer
studies through 2021). The relevant health effect studies in humans, as well as the mechanistic
data informative to sensory irritation, were evaluated to ascertain the level of confidence in the
study results for hazard identification (see Appendix A.5.2 and A.5.6).
Methodological issues considered in evaluation of studies
This review focused on the results of controlled human exposure studies and observational
studies of exposure to residential populations. The relevant period for the assessment of irritant
responses was considered to be concurrent with the time period of the exposure assessment
because the symptoms associated with irritation occur immediately (Krakowiak et al.. 1998:
Andersen and Molhave. 1983: Andersen. 19791. The controlled human exposure studies were able
to evaluate symptoms in a controlled environment; therefore, the exposure-response relationship
was more precise, and potential confounders were of less concern. However, the study groups
were selected for age (younger adults) and were healthy enough to conform to study protocols.
These studies evaluated formaldehyde concentrations above 0.1 mg/m3, while exposure levels in
the residential studies ranged from 0.01 (the limit of detection [LOD] in the available studies) to
approximately 1 mg/m3, with a large proportion of residences having levels less than 0.1 mg/m3.
The studies of residential formaldehyde exposure included a wider range of ages (adults and
children) and potentially susceptible individuals, some of whom had existing respiratory issues and
This document is a draft for review purposes only and does not constitute Agency policy.
1-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
other health conditions. Evaluations of individual mechanistic studies emphasized consideration of
issues related to exposure conduct, as previously described (see Preface and Appendix A.5.6).
Sensory Irritation Studies in Humans
The following discussion is organized by exposure setting, starting first with evidence from
controlled human exposure studies, followed by studies of residential exposure, and then
laboratory and occupational studies. Evidence tables for each exposure setting (see Tables 1-1
and 1-2) are organized by level of confidence in the study's results and then by publication year.
Controlled human exposure studies fshort-term exposures!
Controlled human exposure studies testing exposures from less than 1 hour to 5 hours
reported slight-to-moderate irritation of the eyes, nose, and throat detected by subjects at
formaldehyde concentrations beginning at around 0.3-0.4 mg/m3 (see Table 1-1), although the
data do not clearly identify the concentration at which symptoms of irritation begin. Eye irritation
was reported at lower concentrations than nasal or throat irritation, and symptoms increased in
frequency and severity with exposure level.
Both prevalence and severity of symptoms were associated with increasing concentration
between 0.12 and 2.5 mg/m3 fMueller et al.. 2013: Berglund etal.. 2012: Lang etal.. 2008: Kulle et
al.. 1987: Andersen and Molhave. 1983: Bender etal.. 19831. Overall, the prevalence of eye
irritation increased from <10 to >80% across several studies with formaldehyde concentrations of
0-4 mg/m3 (see Figure 1-2). The prevalence of mild-to-moderate irritation varied among
individuals at specific concentration levels. For example, at concentrations above 2 mg/m3,
prevalence ranged from 53 to 100% (Kulle etal.. 1987: Schachter etal.. 1987: Witek etal.. 1987:
Schachter et al.. 1986a: Andersen and Molhave. 19831. Possible reasons for the variation may
include differences in exposure duration or differences in the characteristics of the volunteers
(e.g., inter individual variation due to smoking status, prior exposure history, or respiratory health).
Participants in all of the studies were 18 to 35 years old. Two studies by one research group
reported a much lower symptom prevalence (27%) among healthy and asthmatic subjects exposed
to 3.7 mg/m3 formaldehyde for 60 minutes (Green etal.. 19871. This response is not directly
comparable to the other studies, however, because the authors only presented irritation prevalence
for more severe symptoms (moderate severity or greater).
Only a few studies evaluated whether symptom prevalence or severity changed over the
course of the exposure period. One research group recruited university volunteers and compared
their responses to controlled formaldehyde exposure against responses in hospital laboratory
workers with routine exposure to formaldehyde; responses were similar between the two groups
during the 40-minute period at 2 ppm (Schachter et al.. 1987: Schachter etal.. 1986a). The study of
the laboratory workers was concluded to have medium confidence because some study aspects may
have reduced the study's sensitivity, including that the previous formaldehyde exposure was not
characterized, and other characteristics, such as being a smoker, were not controlled. The
This document is a draft for review purposes only and does not constitute Agency policy.
1-13 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
university volunteers reported the highest symptom scores when subjects first entered the
exposure chamber with declines over the 40-minute exposure period. Andersen and Molhave
(1983) also found that eye irritation was experienced earlier in the exposure period among subjects
exposed to higher concentrations (1 and 2 mg/m3) and that symptom severity increased and then
plateaued or decreased after 3 hours. However, the initiation of symptoms was delayed at lower
concentrations (0.3 and 0.5 mg/m3), and symptom severity continued to increase over the rest of
the exposure period. Other studies involving exposures from a few minutes to 1 hour also reported
irritation responses that slightly decreased or plateaued (Green etal.. 1987: Bender etal.. 1983).
Note that Bender et al. (1983) used a protocol involving exposure to the eyes only, which may
involve a different type of response compared to inhalation. Therefore, these few studies suggest
that some acclimatization may occur over a few hours at higher concentrations; however, this
phenomenon may not be apparent when concentrations are lower (<1 mg/m3). Further, based on
the few studies available, individuals with long-term occupational exposure to formaldehyde do not
appear to respond differently than individuals with no previous known exposure.
1 .OOT
0.80"
E o
o =
0)
E S
^ 0)
(/)
0.60-
Q. 0.40-
O A nd erson 1 983
¦ S chachter 1 986
~
A Kulle 1987
V S chachter 1 987
~ G reen 1987
m
• Green 1 987A
V
* Witek 1 987A
• Bender1983
0.20"
m o d e rate seve rity
o.oo-
-i 1 1 1 r
1 2
¦ ¦ ¦ I ¦
3
Formaldehyde (mg/m )
Figure 1-2. Prevalence of eye irritation in controlled human exposure studies
of formaldehyde.
Studies that randomly assigned the order of exposure levels are graphed in relation to formaldehyde
concentration. Three studies limited by reporting deficiencies regarding randomization (Bender et al.,
1983) or blinding (Kulle et al., 1987; Andersen and Molhave, 1983) are graphed with open symbols. The
results from Schachter et al. (1987) also are graphed in open symbols because subjects were also exposed
to formaldehyde through their occupations or cigarette smoke. Two studies reported increases in
symptom intensity or scores for eye irritation but did not report prevalence and are not graphed (Mueller
This document is a draft for review purposes only and does not constitute Agency policy.
1-14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
et al., 2013; Lang et al., 2008; Yang et al., 2001). Note that the figure does not convey differences in
severity scores, which also increased with formaldehyde level.
In addition to subjective reports, some investigators evaluated objective measures,
including eye blink frequency, conjunctival redness, and nasal flow and resistance fMueller etal..
2013: Lang etal.. 2008: Andersen and Molhave. 1983: Andersen. 19791. Eye blink frequency was
increased at exposure levels above those where subjective symptoms were reported. For example,
two studies evaluated responses to a combination of concentration peaks superimposed on a
constant formaldehyde exposure (Mueller etal.. 2013: Lang etal.. 20081. Lang et al. (20081 found
that increased eye blink frequency and conjunctival redness occurred at 0.62-1.2 mg/m3 among
subjects who also reported symptoms of eye irritation at 0.37 mg/m3. Mueller et al. (2013.) found
no exposure-related effect on blinking frequency and conjunctival redness, although total symptom
scores increased beginning at 0.37 mg/m3 with peaks of 0.7 mg/m3 in a group with nasal
hypersensitivity. Studies using objective measures of nasal irritation reported variable results
including no change in nasal flow and resistance between 0.19 and 0.62 mg/m3 (Lang etal.. 2008). a
decrease in nasal mucus flow at a concentration of 0.37 mg/m3 and higher (Andersen and Molhave.
1983). and an increase in nasal flow rate among hypersensitive participants at 0.86 mg/m3 (Mueller
etal.. 20131. Subjects exhibited a large degree of individual variability in sensitivity for both
objective and subjective responses fMueller etal.. 2013: Berglund etal.. 2012: Lang etal.. 20081.
This document is a draft for review purposes only and does not constitute Agency policy.
1-15 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-1. Summary of controlled human exposure studies of formaldehyde
and human sensory irritation
Study and design
Results
Mueller et al. (2013)
Design: N = 41, age 32 years, nonsmoking, healthy male
volunteers; categorized into hyposensitive and hypersensitive
based on C02 sensitivity measurements in nasal mucosa
(cutpoint median 80.3 mm on visual analogue scale [VAS]).
Exposure order randomly assigned; repeated measures cross-
over design; blinding not described. Five 4-hour exposure
conditions, 1 per day, over 5 days. Four 15-minute cycle
exercise segments during exposure period.
Outcome: Irritation assessed by conjunctival redness (digital
photographs), blinking frequency (blinks counted in 60-
second segments from 5-minute video, two counters blind to
concentration), tear film break-up time (time to first close of
eyelid while staring at mark on wall), nasal flow and
resistance (rhinomanometry), and validated symptom
questionnaire (SPES German translation) measured before
and 15 minutes before end of exposure. Severity rated using
VAS with 100-mm scale.
Exposure: 4 hours in groups of 2. Clean air, 0.3 + 4 peaks of
0.6 ppm, 0.4 + 4 peaks of 0.8 ppm, 0.5 ppm and 0.7 ppm (0.0,
0.37 + 0.74, 0.49 + 0.98, 0.62, and 0.86 mg/m3).a
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Confidence: High
Results presented in graphs of difference between pre- and end
of test values. Large variability in scores between subjects for
all measures. Blinking frequency and conjunctival redness—no
exposure-related effect, tear film break-up time—increased in
0.4/0.8 ppm and 0.5 ppm (p < 0.05), nasal flow rate increased in
hypersensitive 0.7 ppm (p < 0.01); total symptom score
increased in hypersensitive at 0.3/0.6 ppm (p < 0.001) and
0.4/0.8 ppm (p < 0.01), perception of impure air increased in
hypersensitive at all exposure levels (including clean air,
0.01 ppm). Control for "negative affectivity" did not alter
associations.
Combined eye symptom score reported to be increased with
higher scores among hypersensitives at all exposures except
0.7 ppm (0.86 mg/m3). Changes in scores were not statistically
significant and no exposure-response was observed (results in
online supplemental resource 10 in Mueller et al. (Mueller et
al., 2013)). Severity measured using VAS ranged between
-0.2 and 2.1 mm).
SPES Symptom Score (SD)—Eye Irritation
mg/m3 Hypo- Hyper-
Average/peak sensitive3 sensitive3
0 -0.17(2.02) 1.96(7.59)
0.37/0.74 0.23 (2.65) 2.13 (4.71)
0.49/0.98 0.62 (5.71) 1.43 (5.31)
0.62 -0.09(2.14) 1.24(2.84)
0.86 0.94(4.56) 0.52(4.14)
3Sensitivity categorized as above or below
median for nasal sensitivity to C02 irritation.
This document is a draft for review purposes only and does not constitute Agency policy.
1-16 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Berglund et al. (2012)
Design: N = 31 healthy volunteers, 52% male, age 24.5 years,
nonsmokers. Exposure concentrations randomly presented;
blinding not described.
Outcome: Participants evaluated detection of odor and nasal
irritation for each "sniff" with forced-choice responses
(yes-yes, yes-no, no-yes and no-no). Goal was to identify the
concentration at which a participant detected nasal irritation
in all (100%) of the 12 presentations.
Exposure: Series of 18 concentrations; 6.36-1,000 ppb
(0.0078-1.23 mg/m3).a
12 presentations at each concentration plus 72 blanks; 1 sniff
in exposure hood (<3 seconds) followed by clean air, 3 sniffs
per minute; 36 exposures per each of eight 12-minute
sessions over 4 hours.
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Results
None of the 31 participants detected nasal irritation in 100% of
12 presentations at any formaldehyde concentration. 13% false
alarms (reports of detection of odor or irritation for blanks).
Large variation in individual distributions of percentage
detections for nasal irritation vs. log concentration. Authors
could not calculate threshold distributions for irritation. See
pooled data below (see Figure 5 in paper).
Formaldehyde
Confidence: High
This document is a draft for review purposes only and does not constitute Agency policy.
1-17 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Lang et al. (2008)
Design: N = 21, age 19-39 years, nonsmoking, healthy
volunteers. Exposure order randomly assigned; double
blinded. Ten 4-hour exposure conditions, 1 per day, over
10 days. Three 15-minute cycle exercise segments during
exposure period.
Outcome: Irritation assessed by conjunctival redness (digital
slit lamp photographs, two scorers), blinking frequency
(90-second count from 6-minute video), nasal flow and
resistance (rhinomanometry), and symptom questionnaire
(SPES German translation) measured before, three times
during, and after exposure, and after last exposure day.
Rated on 5 levels (0-5).
Exposure: 4 hours in groups of 4. Clean air, 0.15, 0.3, and
0.5 ppm (0.0, 0.19, 0.37, and 0.62 mg/m3); additional 0.3 and
0.5 ppm with peaks up to 1.0 ppm (1.23 mg/m3).a
Additional 0.0, 0.3, and 0.5 ppm with ethyl acetate (EA)
introduced as a "mask" for formaldehyde odor.
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations not reported.
Confidence: High
Blinking frequency, conjunctival redness significantly increased
at 0.5 ppm with peaks of 1.0 ppm.
Symptoms: Maximum scores at 195 minutes; eye and olfactory
symptom scores were elevated at 0.3 ppm (p < 0.05). With
control for "negative affectivity," eye irritation symptoms
significantly associated with 0.5 ppm with EA or 0.5 ppm with
peaks. Severity: Average severity scores were less than 2
("somewhat").
Nasal irritation: no significant increase in objective measures;
symptoms significantly increased at 0.5 ppm and 0.3 ppm with
coexposure to EA (also an irritant; p < 0.05).
Green et al. (1989)
Design: N = 24,10 male, mean age 24 ± 0.7 yr, nonsmoking,
no history of allergies or hay fever. Random assignment to
order of exposure; double blinded. Four 15-min exercise
segments in the 2-hr exposure period.
Outcome: Symptoms questionnaire (presence and severity,
scored none = 0 to severe = 5) before, and four times during
exposure. Testing pre- and during exposure period
(approximate 15-min intervals).
Exposure: 2 h, four exposures over 4 weeks, clean air, 3 ppm
(3.69 mg/m3)a, 0.5 mg/m3 activated carbon aerosol (ACA),
HCHO + ACA.
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Confidence: High
Symptom scores presented graphically for 80-min time point.
Formaldehyde treatment elevated symptom scores (p < 0.05) at
all time points for eye, nasal and throat irritation, odor, chest
discomfort. No effect modification by ACA exposure. Average
eye irritation scores <1.5 at 80 minutes; similar response at all
measurements (20, 50, 80, and 110 minutes).
No separate effect on cough by formaldehyde, but combined
formaldehyde and ACA exposure resulted in elevated score for
cough at 20 minutes (p < 0.02) and 80 minutes (p < 0.05).
This document is a draft for review purposes only and does not constitute Agency policy.
1-18 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Green et al. (1987)
Design: n = 22, mean age 26.9 ± 3.6 years, nonsmoking, no
history of allergies or hay fever. Random assignment to order
of exposure; single blinded. Two 15-min exercise segments in
the 60-min exposure period.
Outcome: Symptoms questionnaire (presence and severity,
scored none = 0 to severe = 5) before, and four times during
exposure. Testing pre- and during exposure period
(approximate 15-min intervals).
Exposure: 60 minute, clean air and 3 ppm (3.69 mg/m3).a
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Confidence: High
Kulle (1993); Kulle etal.(1987)
Design: Group 1 (N = 10), Group 2 (N = 9), nonsmoking
healthy, age 26.3 ± 4.7 years, 53% male. Exposure order
randomly assigned; Blinding not reported. 3-hour exposures
each week, at same time on five occasions. 8-minute exercise
segment every half hour during 2-ppm exposure.
Outcome: Symptom questionnaires before and after each
exposure, and 24-hours postexposure. Severity was scored
none, mild, moderate, severe (0-5).
Exposure: 3 hour, Group 1: 0.0, 0.5,1.0, or 2.0 ppm (0.0, 0.62,
1.23, 2.46 mg/m3)a at rest, and an additional 2.0 ppm with
exercise; Group 2: 0.0,1.0, or 3.0 ppm (0.0,1.23, or
3.69 mg/m3) at rest, and an additional 2.0 ppm with exercise.
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Confidence: Medium
Results
Mean symptom scores associated with 3-ppm exposure at all
time points, difference from clean air statistically significant for
odor, nose or throat irritation, and eye irritation. Individual
severity scores ranged from none to severe.
Prevalence of scores > moderate
severity at 3 ppm (p < 0.01)
Healthy
Asthmatic
(%)
(%)
Odor
23
31
Nose/throat
32
31
Eye
27
19
Mean difference in scores before and after exposure period:
Linear dose-response (N = 19) for odor and eye irritation, 0,1,
and 2 ppm (p < 0.0001); and nose/throat (Group 2, p = 0.054).
Log-linear dose-response for odor and eye irritation, 0, 0.5,1.0
and 2.0 ppm (Group 1, p < 0.05). Test for nonlinearity not
significant. Data presented graphically, prevalence reported in
Kulle et al. (1993), Table 3 in the paper.
Prevalence
Concentration # (mild/moderate)
0
19
0.05
0.62
10
0
1.23
19
0.26
2.46
19
0.53
3.69
9
1.0
Deficiencies in reporting detail regarding blinding and
quantitative results
This document is a draft for review purposes only and does not constitute Agency policy.
1-19 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Witek et al. (1987): Witek et al. (1986)
Symptoms during exercise not different from rest
Design: n = 15 with asthma, ages 18-35 years, nonsmoking.
Random assignment to order of exposure; double blinded.
Two protocols (at rest and during exercise).
Prevalence (%) and severity scores during rest
0 ppm 2 ppm
# (%) Sa # (%)
Sa
Outcome: Symptoms questionnaire, severity scores (0-4).
Testing at beginning and at 30 min during and 4- to 8-hr and
24-hr postexposure.
Odor
Eye
5(33.3) 7
1(7) 2
15(100)
11(73.3)
30
16
Exposure: 40 minutes, 0 and 2 ppm (2.46 mg/m3).a
Formaldehyde generation via thermal depolymerization of
paraformaldehyde over boiling 2-propanol, dynamic chamber,
Nose
Throat
3 (20) 4
4(26.7) 4
7 (46.7)
5(33.3)
10
6
analytical concentrations reported.
aTotal severity score across al
subjects
Confidence: Medium
Symptoms reported to have disappeared postexposure.
Schachter et al. (1986a); Witek et al. (1986)
Design: N = 15 healthy, age 18-35 yr, nonsmokers. Random
assignment to order of exposure, double blinded. Two
protocols (at rest and during exercise), separated by 4 days.
Symptoms during exercise not different from rest; highest
symptom scores at beginning of exposure with decrease by 30
minutes.
Prevalence (%) and severity scores during rest
0 ppm 2 ppm
Outcome: Symptoms questionnaire at beginning and at 30
min during exposure and at 8 and 24 hr after exposure,
severity scores (0-4).
# (%) Sa
# (%)
Sa
Odor
7 (46.7) 7
12 (80.0)
18
Exposure: 40 min; clean air and 2 ppm (2.46 mg/m3).a
Eye
0 0
8(53.3)
12
Formaldehyde generation via thermal depolymerization of
paraformaldehyde over boiling 2-propanol, dynamic chamber,
analytical concentrations reported.
Nose
Throat
4(26.7) 4
2(13.3) 2
6 (40.0)
4 (26.7)
7
4
Confidence: Medium
aTotal severity score across al
subjects
Co-exposure to 2-propanol
Eye Irritation Severity by Exposure, n (%)
0 ppm
2 ppm
Mild
Moderate
0
0
5(33.3)
2(13.3)
Severe
0
1(7)
This document is a draft for review purposes only and does not constitute Agency policy.
1-20 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Andersen and Molhave (1983); Andersen
(1979)
Design: N = 16 healthy students, age 30-33, 68.8 % male,
31.2% smokers, groups of four over 4 days. Exposure order
determined by Latin square design, blinding not described.
Testing before (during 2-hour clean air) and two times during
exposure.
Outcome: Subjects used a pointer to express the degree of
airway irritation (scale 1 to 100) while being exposed.
Exposure: 5 hours; 0.3, 0.5,1.0 and 2.0 mg/m3 (0.24, 0.40,
0.81 and 1.61 ppm respectively).
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations reported.
Confidence: Medium
Variation exposure concentrations, reporting deficiencies
regarding blinding, potential confounding by smoking
Schachter et al. (1987)
Design: N = 15 healthy hospital laboratory workers routinely
exposed to formaldehyde as part of their job, age 32 ± 11.3
years, 33.3% male, N = 2 smokers. Random assignment to
order of exposure, double blinded. Two dose levels, four
exposure conditions, 2 days at rest and 2 days with exercise.
One 10-minute exercise segment at 5 minutes in the
40-minute exposure period.
Outcome: Symptoms diary, scores 0-4, at t = 0, t = 30
minutes, and 4-8 hours and 24 hours postexposure.
Exposure: 40 minutes; clean air and 2.0 ppm (2.46 mg/m3).a
Formaldehyde generation via thermal depolymerization of
paraformaldehyde over boiling 2-propanol, dynamic chamber,
analytical concentrations reported.
Confidence: Medium
Co-exposure to 2-propanol, potential confounding by smoking
Results
Irritation prevalence with clean air was not reported. At end of
exposure to 0.3, 0.5,1.0, and 2.0 mg/m3 of formaldehyde; 3, 5,
15 and 15 subjects respectively of the 16 who participated
reported conjunctival irritation, dryness in the nose and throat.
Smokers were found to be less sensitive than nonsmokers.
Severity: Maximum individual scores ranged from 30 (slight
discomfort) at 0.3 mg/m3 to 50 (discomfort) at 3 mg/m3. After
the first 2 hours, discomfort increased during the exposure
period at 0.3 and 0.5 mg/m3. In two highest concentrations,
discomfort reported during first hour, increased to hour 3, then
plateaued or decreased.
Eye blinking increased at 2.0 mg/m3 (1.70 ppm).
Subjects reported no symptoms the next morning.
Symptoms during exercise not different from rest.
Prevalence and scores during rest
Concentration (ppm)
0 2
# (%)
Sa
# (%)
Sa
Odor
7 (46.7)
10
12 (80.0)
22
Eye
0
0
7 (46.7)
9
Nose
1 (0.07)
2
0
0
Throat
1 (0.07)
2
0
0
aTotal Score Across all Subjects
Eye Irritation Severity by Exposure, # (%)
0 ppm 2 ppm
Mild 0 5(33.3)
Moderate 0 2 (13.3)
Severe 0 0
This document is a draft for review purposes only and does not constitute Agency policy.
1-21 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Bender et al. (1983)
Design: Panels of seven volunteers from Battelle Memorial
Institute (age, health status, smoking status, and gender not
reported) exposed to clean air and formaldehyde. Individuals
who responded to 1.3 and 2.2 ppm formaldehyde were
tested.
Order of exposure assignment not reported, blinding not
described. Eye-only exposures for 6 minutes.
Outcome: Response time (seconds); proportion of subjects
with shorter response time to formaldehyde than to clean air.
Subjective score (0-3) when first detected and after
6 minutes.
Exposure: 6 minutes, eye only, 0, 0.35, 0.56, 0.7, 0.9 and
1.0 ppm (0.0, 0.43, 0.69, 0.86,1.11, and 1.23 mg/m3).a
Formaldehyde generation via thermal depolymerization of
paraformaldehyde, dynamic chamber, analytical
concentrations not reported.
Confidence: Low
Reporting deficiencies regarding analytical concentrations,
random allocation and blinding. Sample size <10.
Median time to first irritant response decreased with increasing
concentration (Cochran'sx2 test for trend). Severity index
increased with increasing concentration.
Proportion with shorter response to
formaldehyde compared to clean air
Respondents
PPM
Total
#
%
0
28
-
-
0.35
12
5
41.7
0.56
26
14
53.8
0.7
7
4
57.1
0.9
5
3
60.0
1.0
27
20
74.1*
p < 0.05, compared to control
Abbreviations: ACA = activated carbon aerosol; ATS = American Thoracic Society; EA = ethyl acetate; HCHO = formaldehyde;
NASA = National Aeronautics and Space Administration; S = Symptom score; SPES = symptom questionnaire; UFFI = urea foam
insulation.
Concentrations reported by authors as ppm or ppb converted to mg/m3.
Studies in residential settings
Two studies investigated the prevalence of irritation symptoms in relation to residential
formaldehyde exposure during the 1980s fLiu etal.. 1991: Sexton etal.. 1986: Hanrahan et al..
19841. These studies met the criteria for a high confidence study but did not describe or provide a
reference for the questionnaire used to assess symptoms. Two studies of occupational exposure in
mobile trailers (Main and Hogan. 1983: Olsen and Dossing. 19821 are included with this group
because the exposure settings (mobile homes with particle board paneling) are similar.
Formaldehyde exposure was associated with an increasing prevalence of eye irritation in all of
these studies (see Table 1-2 and Figure 1-3). One study, Olsen etal. (1982), assessed the severity of
symptoms as well as their presence within the previous month using a linear analogue scale.
Among those reporting symptoms of eye irritation, a severity at approximately the midpoint of the
scale was reported, which is consistent with the mild or moderate severity reported by the
controlled human exposure studies. Two studies in residential populations analyzed exposure-
response relationships and observed a statistically significant relationship between increasing
formaldehyde concentration (from approximately 0.01 to >0.60 mg/m3) and symptoms of irritation
using logistic regression models with adjustment for age, gender, smoking behavior and other
potential confounders fLiu etal.. 1991: Sexton etal.. 1986: Hanrahan et al.. 19841. Data were
This document is a draft for review purposes only and does not constitute Agency policy.
1-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
collected on current symptoms occurring after participants had moved into their homes (Hanrahan
etal.. 19841 or those that occurred during the 2 weeks prior to the end of the one-week
formaldehyde sampling period fLiu etal.. 19911. Although the sampling period used by Hanrahan
et al. (1984) was shorter (1 hour), the presence of smokers or gas appliances in the home, sources
that might contribute to variability in concentrations, were not associated with indoor
formaldehyde concentrations. Therefore, the formaldehyde concentrations measured by both
studies were considered to be relevant to the time frame of the symptom reports. Other emissions
released from the same sources as formaldehyde that also can contribute to eye irritation, such as
phenols from resins in floor or wall coverings or pinene and terpenes from wood products, were
not analyzed. However, a strong exposure-response relationship with formaldehyde, as a
cumulative measure (ppm-hr) or a 1-hour concentration, was reported by two medium confidence
studies, which is unlikely to be explained to a great extent by unmeasured confounding. Although
limited by low participation rates, participants were randomly selected for recruitment, and the
investigators noted that the characteristics of the respondents and nonrespondents, such as age of
housing stock, demographics, and formaldehyde concentrations, were comparable.
Figure 1-3 graphs prevalence of eye irritation (or burning eyes) by formaldehyde
concentration reported by controlled human exposure studies and residential studies that
evaluated concentrations below 1 mg/m3. These results are complementary for the most part and
indicate a consistent pattern in response to formaldehyde concentrations between 0 and 1 mg/m3.
As seen in Figures 1-2 and 1-3, the concentration-response curve for eye irritation in the Kulle et al.
(19871 study was shifted to the right compared to other studies that evaluated multiple
concentration levels. The study by Bender et al. (19831 used a protocol that involved exposure to
the eyes only, although the concentration-response pattern was similar to the studies that
evaluated exposure via inhalation. Two controlled human exposure studies that also evaluated
concentrations below 1 mg/m3 used a different metric to measure symptoms, a subjective symptom
score using a validated questionnaire (Mueller etal.. 2013: Lang etal.. 2008). The results of the two
studies differed; Lang et al. (2008) reported an increase in symptom scores for eye irritation at
0.3 mg/m3, although with control for responses to questions that assessed "negative affectivity," the
association was not observed until 0.5 mg/m3, and Mueller et al. (20131 reported no effect related
to formaldehyde exposure.
Other URT symptoms were reported by these studies as well, including irritation of the nose
and throat A recent study of formaldehyde levels in redecorated homes in China and respiratory
symptoms among residents exposed from 1 month to 3 years, reported a higher prevalence of nasal
irritation, and throat irritation among adults and children at concentrations above 0.08 mg/m3
(Zhai etal.. 2013). The association was independent of other factors including age, gender, smoking
in the family, occupation, education, presence of domestic animals, family history of allergy, and
ventilation frequency.
This document is a draft for review purposes only and does not constitute Agency policy.
1-23 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
It
•
Anderson 1983
¦
Ku lie 1987
~
Bender 1983
T
Main & Hogan 1983
~
Olsen & Dossing 1982
o
Hanrahan 1984
~
Liu 1991
Formaldehyde {mg/m
Figure 1-3. Prevalence of eye irritation among study groups exposed to
formaldehyde in residential settings and controlled human exposure studies.
Different symbols are used for each study. Olsen and Dossing (1982) and Main and Hogan (1983) are
occupational studies with exposure in mobile trailer offices and are presented with the residential mobile
home studies. Prevalence at formaldehyde concentrations measured among comparison groups is
graphed if reported (Holnessand Nethercott, 1989; Holmstrom and Wilhelmsson, 1988; Horvath et al.,
1988; Olsen and Dossing, 1982). Error bars are standard error (SE) calculated by EPA. Average weekly
concentrations in three categories for Liu et al. (1991) were estimated from the midpoint of each category
of reported weekly cumulative exposure (ppm-hour) and an assumption that individuals spent 60% of a
24-hour period at home.
This document is a draft for review purposes only and does not constitute Agency policy.
1-24 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-2. Summary of epidemiological studies of residential exposures to
formaldehyde and human sensory irritation
Study and design
Results
Zhaietal. (2013)
Jan 2008-Dec 2009 (China) (prevalence)
Population: 186 homes in Shenyang surveyed, homes were decorated
in past 4 years and occupied within the past 3 years; randomly selected
one adult from each house, plus 82 children (assisted by parents);
characteristics of participants were not described.
Outcome: Reported symptoms and disorders via questionnaire Ferris
(1978).
Exposure: Cited code for indoor environmental pollution control of civil
building engineering (GB50325-2001); sampling period not reported.
Samplers in breathing zone in bedroom, living room, and kitchen;
N = 558 in 186 homes; exposure groups "polluted" homes:
>0.08 mg/m3, mean 0.09-0.13 mg/m3, range 0.01-0.55 mg/m3, in three
rooms; nonpolluted <0.08 mg/m3, mean 0.04-0.047 mg/m3.
Analysis: Compared symptom prevalence for children and adults by
exposure category (reported p-values); multivariate logistic regression
of respiratory system symptoms (all) in children and adults, adjusting
for age, gender, smoking in family, occupation, education, ventilation
frequency, domestic pets, house facing, family history of allergy, height,
weight.
Evaluation:3
For analysis of combined symptoms:
Respiratory system symptoms and disorders
by exposure group (N = 186 adults, 82
children)
Symptom
>0.08
mg/m3 (%)
<0.08
mg/m3 (%)
Cough, adults
Cough, children
Phlegm, adults
Phlegm, children
Wheeze, adults
Wheeze, children
Nasal irritation,
adults
Odor disorder,
adults
Throat irritation,
adults
16.0*
25
6.7
15
5.0
10
52.1*
31.9*
4.5
8.1
3.0
6.7
3.0
6.6
16.4
3.0
13.4
*p < 0.05, **p < 0.01
Association of formaldehyde exposure with
respiratory system symptoms in adults and
children (N = 186 adults, 82 children)
Overall
SK
in
ft
Oth
Confidence
Medium
¦
Odds Ratio
95% CI
Adults3
2.6
1.8, 3.8
Children15
4.3
2.1, 8.8
Combined analysis does not distinguish URT irritation symptoms from
asthma-related symptoms; sampling period not reported.
aOther statistically significant covariates were
ventilation frequency (OR = 1.6) and domestic
pets (OR = 1.5)
bOther statistically significant covariates were
ventilation frequency (OR = 1.8) and family
history of allergy (OR = 1.9)
Liu et al. (1991); Sexton et al. (1986) (California)
Prevalence survey, 1984-1985.
2,203 randomly selected mobile home occupants recruited, 44%
response (836 of 1,895 contacted). 1,394 residents in 663 mobile
homes in summer and 1,096 residents in 523 mobile homes in winter.
20-64 years of age.
Outcome: Symptoms (occurrence during 1 week prior to end of
sampling period) from mailed questionnaire, questionnaire not
described.
Exposure: Formaldehyde sampling using passive monitors mailed to
participants, 7-day samples, two rooms.
Average concentration: 0.091 (SD 0.069, range <0.01 (LOD)-0.464) ppm
in summer and 0.091 (SD 0.052, range 0.017-0.314) in winter. (0.11 (SD
0.095), range <0.012-0.57 mg/m3)
Cumulative formaldehyde: formaldehyde concentration x hours spent
in the residence (ppm-hr).
Significant associations with burning/tearing eyes,
stinging/burning skin in summer, and
burning/tearing eyes, chest pain, sore throat in
winter (effect estimates from logistic regression
model were not presented).
Prevalence Burning/Tearing Eyes
ppm-hr
Summer
(%)
Winter (%)
<7.0
13.3
10.8
7.0-12
17.1
14.7
>12.0
21.4
20.6
Burning/tearing eyes higher among females in
regression models.
This document is a draft for review purposes only and does not constitute Agency policy.
1-25 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Analysis: Logistic regression adjusting for age, gender, smoking status,
time spent at home, and chronic respiratory/allergy status.
Evaluation:3
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Reporting deficiencies regarding type of questionnaire and quantitative
results
Hanrahan et al. (1984) (Wisconsin)
Prevalence survey, 1979
61 teenage and adult occupants from 65 of 208 randomly selected
mobile homes. Mean age 48 yrs, 61% female. Participants blinded to
exposure status.
Outcome: Current symptoms with occurrence since moving into home
from self-administered questionnaire, questionnaire not described.
Exposure: Formaldehyde measurements: 1-hour samples, average of
measurements in two rooms.
Median: 0.16 ppm. Range: <0.1 ppm to 0.80 ppm. Outdoor mean
(SD) = 0.04 (0.03) ppm. Windows closed, smoking banned, gas
appliances turned off for 30 minutes prior to measurements.
Analysis: Logistic regression adjusting for age, gender, and smoking.
Evaluation:3
A statistically significant concentration-response
relationship was reported individually for burning
eyes and eye irritation; no regression coefficients
provided.
Burning Eyes
Concentration
(ppm)
Prevalence (%)a
0.1
<5
0.2
17.5
0.5
65
0.8
80
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Predicted response estimated by EPA from
graphical presentation of logistic regression results
normalized to mean age.
Formaldehyde concentration not associated with
presence of smoker in home or gas appliances.
Regression model showed higher prevalence of eye
irritation in younger persons.
Limited sampling period for formaldehyde measurements;
Questionnaire not described
Olsen and Dossing (1982) (Denmark)
Prevalence survey, 1979.
Exposed: 66 of 70 employees of seven mobile day care centers (average
of 6 months old) paneled indoors with urea formaldehyde glued particle
board; mean age 29 years, 10/90 percentiles 19/40 years. Referent: 26
of 34 employees randomly selected from three control (nonmobile
home) centers with no materials containing formaldehyde. Mean age
32 years, 10/90 percentiles 25/38 years. All worked in day care centers
for >3 months.
Outcome: Prevalence (yes/no), Severity of symptoms experienced
within 1 month measured in centimeters on scale from 0 to 10, "linear"
analogue self-assessment method."
Exposure: Formaldehyde measurements taken after questionnaire
study: 2-hour samples in 2-4 locations in the homes. Mean mobile
units = 0.43 mg/m3 (range 0.24-0.55 mg/m3).
Mean referent = 0.08 mg/m3 (range 0.05-0.11 mg/m3).
The average frequency of mucous membrane
irritation of eyes, nose, and throat was 3x higher
among staff of mobile units vs. stationary
institutions (p < 0.01). Symptoms disappeared after
end of work.
Percentage with affirmative answer3
Exposed
Referent
(%)
(%)
Eye
56
14.6
Nose/throat
74
25
Estimated by EPA from bar chart in Figure 1 in the
paper.
This document is a draft for review purposes only and does not constitute Agency policy.
1-26 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Analysis: Prevalence and average impact scores compared.
Evaluation:3
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Some uncertainties regarding temporal concordance of exposure and
symptom assessments
Main and Hogan (1983)
Prevalence survey
21 exposed individuals working in two mobile trailers for 34 months
(mean [SD] age 38 [9] years, 76% male)
18 referent staff members who did not work in the trailers (mean [SD]
age 30 [6] years, 50% male)
Outcome: Modified ATS questionnaire
Exposure: Three 1-hour area samples taken on four occasions (August,
September, December, April) always on a Monday. At least one sample
was taken from each office in both trailers.
Concentration range 0.12-1.6 ppm (0.15-1.97 mg/m3)a
Analysis: Group comparisons, x2 statistic
Evaluation:3
Symptom Prevalence While at Work
Ex-
Ref-
posed
erent
X2
Symptom
1
CM
II
00
1
II
(p-value)
Eye
0.71
0.0
20.9
irritation
(<0.001)
Nasal
0.33
0.0
7.3 (0.01)
symptoms
Throat
0.48
0.0
11.5
irritation
(0.001)
SB IB Cf Oth
Overall
Confidence
Low
~
~
1 1
~
Potential dissimilarity between comparison groups; more exposure to
ETS among referent; small sample size
LOD = limit of detection; RD50 = concentration resulting in a 50% reduction in the respiratory rate; RIL = recommended indoor
limit; VOC = volatile organic compound.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.2). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
1 Laboratory and occupational exposure
2 The studies of anatomy students and formaldehyde-exposed workers provide further
3 evidence that formaldehyde exposure is associated with symptoms of eye, nose, and throat
4 irritation. These studies are summarized in tables in the appendix for sensory irritation
5 (Appendix A.5.2). Exposure levels experienced during anatomy laboratory courses and in
6 occupational settings were high and variable. Formaldehyde levels during anatomy courses
7 generally averaged 0.9 mg/m3 and above during the lab, with short-term peaks above 5 mg/m3
8 (Takahashi et al.. 2007: Kriebel etal.. 2001: Wantke etal.. 2000: Kriebel etal.. 1993: Uba et al..
This document is a draft for review purposes only and does not constitute Agency policy.
1-27 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
19891. These exposures were episodic, one to two sessions per week, for 1-4 hours. Study designs
that analyzed reported symptoms and formaldehyde levels measured in close temporal proximity
were considered less subject to information bias. The intensity of symptoms fKriebel etal.. 20011
and prevalence or frequency of occurrence fTakigawa etal.. 2005: Wantke etal.. 20001 of
symptoms was related to exposure during the lab. Over time, the magnitude of the increase in
symptoms during a laboratory session was reported to decline over the succeeding weeks of the
course fKriebel etal.. 2001: Kriebel etal.. 19931. Kriebel et al. (20011 modeled average
formaldehyde concentration during each lab session in relation to irritation symptoms (separate
models for eye, nose, and throat irritation) and reported that intensity of eye irritation symptoms
increased by 1.22% per unit increase in ppm, and the magnitude of the increase in intensity
declined with each successive week during the course.
Formaldehyde concentrations in the workplace varied by industry. Examples of industrial
formaldehyde levels include mean levels of 0.26 mg/m3 in a formaldehyde-producing plant in
Sweden (Holmstrom and Wilhelmsson. 19881. 0.96 mg/m3 in a melamine-formaldehyde resin-
producing plant (Neghab etal.. 20111 in Iran, and 1.04 mg/m3 in a particleboard plant (Horvath et
al.. 19881. Excursions above 2 mg/m3 were measured in some industries. Most of the studies
compared responses in exposed groups to those in a referent group, and symptoms of URT and eye
irritation were associated with exposure status in these studies. One study also reported a strong
exposure-related trend for burning nose, stuffy nose, burning eyes, itchy nose, sore throat, and itchy
eyes in multiple regression models, although quantitative results were not reported (Horvath etal..
19881.
Evidence on Mode of Action for Sensory Irritation
Sensory irritation is understood to occur as a result of direct interactions of formaldehyde
with cellular macromolecules in the nasal mucosa leading directly or indirectly to stimulation of
trigeminal nerve endings located in the respiratory epithelium. While other mechanistic changes
(e.g., oxidative stress; airway inflammation; damage or dysfunction of the respiratory epithelium)
and biological differences (e.g., nasal morphology; underlying allergy, infection, or other respiratory
conditions) are expected to be strong modifiers of this sequence of events, this pathway is
interpreted as likely to be the dominant mechanism by which formaldehyde exposure causes
sensory irritation. The primary evidence for this conclusion includes mechanistic changes in the
URT, which are supported by robust or moderate formaldehyde-specific data (see summary
interpretations in Figure 1-4 and Table 1-3; Appendix A.5.6 includes additional details and evidence
supporting other relevant mechanistic changes, some of which are discussed briefly below), and the
relationships described are largely well understood biological phenomena, or they have been
demonstrated following formaldehyde exposure. This mechanistic understanding provides strong
support for the biological plausibility of this effect. Although the primary support for an MOA
reliant on stimulation of receptors on nasal trigeminal nerve endings is from studies in
This document is a draft for review purposes only and does not constitute Agency policy.
1-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
experimental animal models, the mechanistic events presumed to be driving sensory irritation after
formaldehyde exposure are expected to be conserved in humans.
Possible Initial Alterations Secondary Alterations
Effector-Level Changes
Key Hazard Feature
# <*>
f1 oxidative
stress in URT
URTTRPA1
binding
Trigeminal nerve
stimulation in URT
Centrally mediated
sensory irritation
Legend
EVIDENCE
RELATIONSHIP
^ Plausibly an initial
effect of exposure
Key feature of sensory
irritation
0 Robust
( ) Moderate
O Slight
—> Robust
--> Moderate
Slight
Figure 1-4. Possible mechanistic associations between formaldehyde
exposure and sensory irritation.
An evaluation of the formaldehyde exposure-specific mechanistic evidence informing the potential for
formaldehyde exposure to cause respiratory health effects (see Appendix A.5.6 for clarifying details)
identified this sequence of mechanistic events as likely to be the dominant mechanism by which
formaldehyde inhalation could cause sensory irritation.
As illustrated in Figure 1-4, formaldehyde exposure appears to result in activation of
chemosensory afferents, likely C fibers, in the URT, presumably in the anterior third of the nasal
cavity, based on the pattern of chemosensory activation and consistent with the distribution of
inhaled formaldehyde (see Appendix A.5.6], This activation initiates central signals that result in
the burning sensation characteristic of sensory irritation. The rapid detection of these sensations in
exposed individuals, as well as insights from other irritants, suggest a receptor-mediated event that
is dependent on formaldehyde penetration to the nerve endings, which may not have an exposure
duration threshold. In vitro and ex vivo studies suggest that activation of the trigeminal nerve by
formaldehyde is mediated, at least in large part, through cation channels, primarily the Transient
Receptor Potential A1 channel (TRPA1). Alongside the centrally mediated physiological response,
the initial activation of the trigeminal nerve is also known to cause a localized release of
neuropeptides, such as substance P, from nerve terminals (not shown in Figure 1-4], which can
affect local inflammatory and immune responses. Observations of these local neuropeptide
changes have been reported at slightly higher formaldehyde levels than those shown to activate the
trigeminal nerve, generally at >1 mg/m3, although the data suggest that they too may be dependent
on TRPA1 activation. All of these direct and indirect interactions could act independently or
together in a concentration- and duration-dependent manner.
This document is a draft for review purposes only and does not constitute Agency policy,
1-29 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
While the response to some irritant chemicals exhibits desensitization or fading of the
irritant response over time (e.g., through receptor downregulation) fNielsen. 19911. it is not clear
this is the case with formaldehyde. As previously discussed, results from acute, controlled human
exposure studies indicate that some acclimatization may occur over exposures of a few hours at
higher concentrations; however, this reduction in symptoms is less apparent (or may be absent)
when concentrations are lower (<1 mg/m3), and changes to this response pattern in humans over
time, particularly with exposure longer than 1 day, remain poorly tested. Studies of reflex
bradypnea in rodents (see Appendix A.3), a phenomenon dependent on the activation of the
trigeminal nerve, show that repeated exposure for up to a month elicits a similar level of activation
of this pathway. However, uncertainties with the rodent data include a nonconstant exposure
(i.e., there is at least partial recovery from the reflex effects in rodents with continued exposure in
acute studies of minutes to hours, while the available short-term studies employed work hour-like
exposure periodicity) and testing only at reflex bradypnea-inducing levels (e.g., >1 mg/m3). It is
unclear whether the results based on acute or episodic exposures apply to long-term responses to
constant oronasal exposure in humans (who do not exhibit reflex bradypnea) at lower
formaldehyde levels.
Sensitivity (i.e., activation of this pathway) is expected to vary between individuals due to
differences in TRPA1 channel sensitivity or access of formaldehyde to TRPA1 channels, as might
occur due to differences in airway structure, mucus production, or TRPA1 channel density. Thus,
enhanced irritation could plausibly occur directly as a result of sensitization of the receptors to
formaldehyde with prolonged exposure or due to the accumulation of other factors that could
reduce the threshold for TRPA1 activation by formaldehyde, or indirectly by increased access of
formaldehyde to trigeminal nerve endings following damage to juxtaposed epithelial cells or
reduced mucociliary function. Airway inflammation has been shown to reduce the threshold for
activation of afferent fibers, through an unknown mechanism fCarr and Undem. 20011. and lipid
peroxidation byproducts can independently stimulate sensory nerve activation. These latter
possibilities are of particular relevance, as exposure to formaldehyde (possibly even at lower levels,
e.g., <1 mg/m3) appears to result in airway inflammation and increased oxidative stress.
Conversely, other modifications to the respiratory epithelium following formaldehyde exposure
(e.g., at levels causing effects such as squamous metaplasia, which is generally observed in animals
at >2.5 mg/m3; see Section 1.2.4) could plausibly result in a decreased access of formaldehyde to
trigeminal nerve receptors. However, while the structure and function of the URT across species is
similar, interpretation of compensatory or adaptive changes within the human URT following long-
term exposure based on findings in experimental animals is difficult to infer, and modification of
sensory nerve signaling in the context of these important scenarios has, for the most part, not been
directly tested. In addition, studies of related chemicals suggest that human sensitivity may also be
dependent on demographic factors such as age, gender (women are generally more sensitive), and
allergy status fShusterman. 2007: Hummel et al.. 20031. complicating an understanding of changes
This document is a draft for review purposes only and does not constitute Agency policy.
1-30 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 in sensitivity. While additional studies clarifying modifications to the sensitivity of this pathway
2 with longer-term exposure or under different exposure scenarios would be useful, it is likely that
3 rodents acutely exposed to ~0.2 mg/m3 formaldehyde under normal conditions would exhibit this
4 effect, and exposed humans are expected to be more sensitive.
Table 1-3. Mechanistic evidence most informative to the occurrence of sensory
irritation after formaldehyde inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
"t URT
Oxidative
Stress
High or Medium
Human: Increased nasal epithelial MldG adducts (oxidative
stress and lipid peroxidation marker) (Bono et al.,
2016): unknown duration (but likely years) at >0.066 mg/m3
Direct and indirect evidence of
elevated reactive oxygen species
(ROS), possibly at low
concentrations (e.g., at
>0.066 mg/m3; maximum of
0.444 mg/m3) with prolonged
human exposure
Moderate
Animal: mRNA changes indicating increased stress-response
proteins (Andersen et al., 2008): short-term exposure
at >2.46 mg/m3
£
o
Human: Increased nasal lavage nitrites (Priha et al.,
2004): acute (8-hr shift) exposure at 0.19 mg/m3
Data suggest elevated oxidative
stress at very low formaldehyde
concentrations with acute and
short-term exposure.
Animal: Increased glutathione peroxidase and/or nonprotein
sulfhydryl groups (Cassee et al., 1996; Cassee and
Feron, 1994): short-term (3 d) duration at 3.94 and
4.43 mg/m3, respectively
Trigeminal
Nerve
Stimulation
High or Medium
Human: None
Increased activity of trigeminal
nerve afferents at <0.5 mg/m3
following acute exposure in
anesthetized rats
Robust
(data are
primarily from
acute
exposure)
Animal: Increased afferent nerve activity (Tsubone and
Kawata, 1991): acute duration exposure resulted in ~20%
at 0.62 mg/m3 and ~50% at 2.21 mg/m3; (Kulle and
Cooper, 1975): acute exposure (threshold detection at 25
seconds) at 0.31 mg/m3
£
o
Human: None
Supportive indirect evidence from
ex vivo and in vitro experiments
Animal: Indirect evidence: with acute exposure, dose-
dependent increase in nerve currents and CI—release in
intact rat trachea (LllO et al., 2013), and stimulation
using in vitro neuronal preparations (Kunkler et al.,
2011; Mcnamara et al., 2007)
TRPA1
Stimulation
High or Medium
Human: None
Indirect data identify TRPA1 as a
molecular target for
formaldehyde exposure-induced
sensory effects
Moderate
(data are
primarily from
acute or
short-term
exposure)
Animal: Formaldehyde and related chemicals such as acrolein
activate the trigeminal system in wild-type mice, but not
TRPA1 knockout mice following acute exposure, at least at
high exposure levels (YonemitSU et al., 2013); taken
together with the established role for TRPA1 in acrolein-
induced sensory effects (e.g., (Bautista et al., 2006)),
these data indirectly support a role for TRPA1 in sensory
nerve-related changes following formaldehyde exposure
o
Human: None
Indirect data identify TRPA1 as a
molecular target of formaldehyde
exposure with acute or short-
term exposure; inhibitor studies
demonstrate that downstream
Animal: Formaldehyde activates TRPA1 in in vitro and ex vivo
models relevant to acute inhalation exposure of the URT and
upperlrt(Luo et al., 2013; Mcnamara et al..
This document is a draft for review purposes only and does not constitute Agency policy.
1-31 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
2007), and is well established in in vivo models using
formalin as a pain stimulus (not a focus of this review);
inhibition of TRPA1 channels localized to sensory nerve
endings reduce formaldehyde exposure-induced nerve
currents in rat trachea (LllO et al.. 2013) and
immune-related responses in mice (Wll et al.. 2013; Lu
et al.. 2005) with short-term (2- or 4-wk) exposure at 1 or
3 mg/m3
effects of sensory nerve
stimulation depend on TRPA1
stimulation
Integrated Summary of Evidence on Sensory Irritation
Symptoms of sensory irritation were consistently reported by studies of formaldehyde
exposure in multiple settings, and both prevalence and severity of symptoms increased with the
level of exposure. Sensory irritation is an acute phenomenon, and symptoms resolve when
exposure is removed (Sauder etal.. 1986: Andersen and Molhave. 1983: Andersen. 19791. The
irritant effects of formaldehyde on the eyes and URT were reported by several controlled human
exposure studies that evaluated responses among healthy or asthmatic volunteers using relatively
high formaldehyde concentrations (0.12 and 3.7 mg/m3) during rest or exercise. In addition to
subjective reports, some investigators evaluated objective measures, including eye blink frequency,
conjunctival redness, and nasal flow and resistance (Mueller et al.. 2 013: Lang etal.. 2008:
Andersen and Molhave. 1983: Andersen. 19791. Eye blink frequency was increased at exposure
levels above those where subjective symptoms were reported. Symptoms of sensory irritation also
were documented in the epidemiological literature among residential and occupational
populations, and students exposed in anatomy classes. Exposed groups described eye, nose, and
throat symptoms with formaldehyde exposure, including itching, stinging, and watering eyes;
sneezing and rhinitis; sore or dry throat; and coughing. Average formaldehyde concentrations for
exposed populations were 0.9 mg/m3 (median) among anatomy students (Kriebel etal.. 1993) and
0.2 mg/m3 and lower among residential populations (Zhai etal.. 2013: Liu etal.. 1991: Hanrahan et
al.. 1984). A statistical exposure-response relationship for the prevalence of eye irritation or
burning eyes was described using regression models in some studies fKriebel etal.. 2001: Kriebel et
al.. 1993: Liu etal.. 1991: Horvath etal.. 1988: Kulle etal.. 1987: Hanrahan et al.. 19841. Alternative
explanations for these symptoms can be ruled out since there is strong evidence from controlled
human exposure studies and residential studies, with exposure-response trends that were adjusted
for potential confounders, including age, gender, and smoking. Coexposures in homes, such as that
from terpenes, phenol, and acetaldehyde, which are emitted from wood products, carpets and wall
coverings, and combustion, were present at lower levels compared to formaldehyde. Sensory
irritation also was reported among groups in exposure settings without those coexposures
(e.g., controlled human exposure studies, anatomy labs). NO2, which is emitted from gas stoves, has
not been correlated with formaldehyde levels in homes fMullen et al.. 20151.
This document is a draft for review purposes only and does not constitute Agency policy.
1-32 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
The magnitude or severity of symptoms does not appear to worsen over periods of
prolonged exposure, and some studies have observed decreases over observation periods lasting a
few weeks. However, change in responses over time has been examined in only a few studies.
Notably, controlled human exposure studies involving occupationally exposed individuals did not
observe responses that were less sensitive than those among subjects with no occupational
exposure, suggesting that the response persists even with prolonged exposure. Controlled human
exposure studies that examined change in response during exposures at relatively high levels
(>1 mg/m3) reported higher symptom scores initially with subsequent declines suggestive of
acclimation during exposure fGreen etal.. 1987: Schachter etal.. 1986a: Andersen and Molhave.
19831. However, at lower concentrations (0.3 and 0.5 mg/m3), the initiation of symptoms was
delayed, and symptom severity continued to increase during the exposure period f Andersen and
Molhave. 19831. Overall, these few studies suggest that some acclimatization may occur over a few
hours at higher concentrations; however, this phenomenon may not be apparent when
concentrations are lower (0.05 mg/m3 (range 0.01 to approximately 1.0 mg/m3) and
controlled human exposure studies testing responses to concentrations 0.1 mg/m3 and above
(Table 1-4).
Table 1-4. Evidence integration summary for effects on sensory irritation
Evidence
Evidence judgment
Hazard determination
Human
Robust, based on:
Human health effect studies:
• Four high and medium confidence studies of symptom prevalence (eye,
nose, throat) among adults and children in residential settings (mean
>0.05 mg/m3 formaldehyde, range 0.01 to approximately 1.0 mg/m3)
• Numerous high and medium confidence studies involving acute exposure
(controlled human exposure studies)
• Numerous high and medium confidence studies with longitudinal designs
(occupational, panel studies of medical school pathology/ anatomy lab
courses)
The evidence demonstrates that
formaldehyde inhalation causes
sensory irritation in humans
given appropriate exposure
circumstances3
Primarily based on well-
conducted residential studies
with mean formaldehyde
concentrations >0.05 mg/m3
This document is a draft for review purposes only and does not constitute Agency policy.
1-33 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
• Consistent observations of irritation symptoms in all studies; clear
exposure-response trends
Biological Plausibility. No directly relevant human mechanistic studies were
found
and controlled human exposure
studies testing >0.1 mg/m3
Potential susceptibilities:
Potentially large variations in
sensitivity are expected,
depending primarily on
differences in nasal health
(including allergy or
inflammatory status) and
physiology
Animal
Robust, based on:
Animal health effect studies: Although animal studies were not formally
evaluated, formaldehyde inhalation-induced sensory irritation in rodents is a
well-documented phenomenon (e.g., reflex bradypnea in mice and rats; see
Appendix A.3).
Biological Plausibility. Robust and moderate evidence for mechanistic events
from animal studies identifies stimulation of the trigeminal nerve as the
dominant MOA
Other
inferences
• Relevance to humans: Assumed, based on similarities in systems
mediating the identified MOA across species
• MOA: Trigeminal nerve stimulation is likely to be the dominant
mechanism
• Other: This effect does not appear to worsen with longer exposure
durations, although uncertainties remain
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2.
1.2.2. Pulmonary Function
This section describes research on formaldehyde inhalation and pulmonary function effects
in experimental and observational studies in humans. The systematic review process assigned
controlled human exposure studies of acute exposure involving healthy individuals to the review of
pulmonary function and the studies involving asthmatic volunteers to the review of effects on
immune-mediated conditions and their results are summarized there (see section 1.2.3). However,
since all of these studies involved measurements of pulmonary function, the results of the studies
involving participants with asthma have been integrated with the evidence from studies of acute
exposure in healthy individuals in this section. Animal studies of analogous endpoints were not
included in the hazard evaluation because there were few directly relevant studies in the peer-
reviewed literature and the extensive literature on these endpoints in humans was considered
adequate to draw a hazard conclusion.
While studies involving acute exposures (<24 hours) reported either no change or
inconsistent responses, more consistent effects were available from studies of occupational
populations exposed over long periods and children exposed in residential settings. The acute,
controlled human exposure studies involving healthy or asthmatic volunteers consistently did not
observe changes, even at high concentrations, although two studies by one research team observed
small decrements (<5%) when longer exercise components (15 minutes) were included. Studies
using shorter exercise components (8-10 minutes) reported no changes. Two studies of asthmatic
volunteers included an allergen challenge (dust mites, pollen), which resulted in a hyperreactive
bronchial response at a lower challenge dose associated with formaldehyde exposure compared to
clean air in one study that imposed mouth breathing (nose clips). Many of the studies of
This document is a draft for review purposes only and does not constitute Agency policy.
1-34 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
occupational groups or dissection labs observed pulmonary function declines over the course of the
workday or lab; however, most did not account for diurnal changes, limiting the interpretation of
these results.
The review of the epidemiological literature provides evidence that long-term
formaldehyde exposure is associated with declines in pulmonary function, including forced
expiratory volume (FEVi), forced vital capacity (FVC), FEVi/FVC, and expiratory flow rates.
Pulmonary function was lower in highly exposed occupational groups employed at exposed jobs for
long durations compared to their nonexposed or lesser-exposed comparison groups. The few
longitudinal studies found some evidence of declines in some measures in excess of that expected
from aging, although the duration of follow-up and individual variation combined with small group
sizes may have resulted in lack of associations with other measures. There are few studies of
residential exposure; however, a clear exposure-response relationship in children was reported by
a well-conducted residential study with most household concentrations <0.045 mg/m3
(Krzvzanowski et al.. 19901.
There is mechanistic support, primarily from studies in animals, for the biological
plausibility of formaldehyde exposure-induced effects on decreased pulmonary function, although a
definitive MOA(s) has not been fully defined. Overall, the most relevant mechanistic evidence
(predominantly evidence interpreted as moderate or robust) included inflammatory structural
alterations and eosinophil increases in the lower airways that appear to be at least partially related
to indirect activation of sensory nerve endings. However, the initial cellular or tissue modifications
that ultimately lead to these later events are not understood, and given the limitations of the
available studies, it is unclear whether and to what extent certain events would be triggered with
chronic, low-level exposure. Although there is an expectation that other important mechanistic
events would be identified with additional study, the available data were interpreted to provide
reasonable support for the biological plausibility of the observed associations and to identify what
is likely to be an incomplete mechanism by which formaldehyde inhalation could cause decreased
pulmonary function.
Spirometric measures are used along with other diagnostic criteria in the evaluation of
asthma and chronic obstructive pulmonary disease in individuals. While a group mean decrement
in any pulmonary function measure does not indicate that the prevalence of these respiratory
diseases has increased, EPA considered a decrease in mean values to suggest a shift toward a
decline in the respiratory health status of the population. Poor pulmonary function, as well as a
decrease in pulmonary function, is an important health endpoint associated with the development
of chronic respiratory disease, coronary heart disease, and mortality (Clayton et al.. 2014: Menezes
etal.. 2014: Young etal.. 2007: Sin etal.. 2005: Schroeder etal.. 2003: Schunemann etal.. 2000:
Sorlie etal.. 19891. The American Thoracic Society evaluated the clinical significance of small
average declines in pulmonary function observed in a population in response to air pollutants and
concluded that although the magnitude of the observed declines may not be clinically relevant to an
This document is a draft for review purposes only and does not constitute Agency policy.
1-35 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
individual, a shift in the population distribution toward lower pulmonary function, assuming the
association is causal, may have a large impact on public health fATS. 20001.
Overall, based on moderate human evidence from observational epidemiology studies, with
corresponding slight evidence for an effect in animals based on mechanistic studies supporting
biological plausibility, the evidence indicates that long-term inhalation of formaldehyde likely
causes decreased pulmonary function in humans given the appropriate exposure circumstances.
The primary support for this conclusion includes a study of children and adults in a residential
setting (mean, 0.03 mg/m3, maximum 0.17 mg/m3) and numerous studies of workers with long-
term exposure to >0.2 mg/m3. The evidence is inadequate to interpret whether acute or
intermediate-term (hour-weeks) formaldehyde exposure might cause this effect.
Literature Search and Screening Strategy
The identification of human health effect studies of formaldehyde exposure and effects on
pulmonary function involved literature searches in PubMed and Web of Science through September
2016 (see Appendix A.5.3 for details), and a systematic evidence map updating the literature
through 2021 (see Appendix F). Studies were included if the exposure to formaldehyde was
quantified and if analyses compared outcomes in relation to exposure for one or more of a standard
set of pulmonary function measures (see Table 1-5). Studies that evaluated both short-term as well
as long-term exposure to formaldehyde were reviewed. Observational studies of human
populations evaluated exposures in residential communities, school classrooms and university lab
courses, and industrial and other workplace settings. Controlled human exposure studies, which
exposed subjects for minutes or hours, also were included. The mechanistic evidence informing
this health effect was identified and evaluated as part of the overarching review of mechanistic data
relevant to potential respiratory health effects (see Appendix A.5.6 for details). The bibliographic
databases, search terms, and specific strategies used to search them are provided in Appendix A.5.3
and A.5.6, as are the specific PECO criteria. Literature flow diagrams summarize the results of the
sorting process through 2016 using these criteria and indicate the number of studies that were
selected for consideration in the assessment (see Appendix F for the identification of newer studies
through 2021). The relevant health effect studies in humans, and the mechanistic data informative
to changes in pulmonary function, were evaluated to ascertain the level of confidence in the study
results for hazard identification (see Appendices A.5.3 and A.5.6).
Methodological issues considered in evaluation of studies
Pulmonary function is assessed using spirometry, which measures the volume and speed of
air that is exhaled or inhaled. Several parameters can be measured during spirometric testing to
characterize an individual's respiratory health. Some common measures evaluated in the studies of
formaldehyde exposure are defined in Table 1-5. It was preferred if the measurement of
pulmonary function outcomes used by the studies followed the guidelines published by the
American Thoracic Society fTepper etal.. 2012: Miller etal.. 2005a: Miller etal.. 2005b: Pellegrino
This document is a draft for review purposes only and does not constitute Agency policy.
1-36 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
etal.. 20051 or provided a description of the protocols and reference equations that were used. In
addition to the use of conventional spirometric equipment, peak expiratory flow has been
measured in research settings using portable flow meters operated by study participants trained in
their use. Although it requires careful training and monitoring, this method has the advantage that
it can be used in large epidemiological studies and multiple measurements can be obtained over
time (Tepper etal.. 2012). Studies of residential exposure to formaldehyde were conducted in this
way (Kriebel etal.. 2001: Krzvzanowski etal.. 19901.
Table 1-5. Common measures of pulmonary function reported in studies of
formaldehyde inhalation
Measure
Definition
Vital Capacity (VC)
(Liters at BTPS)
The volume of air between a full inspiration and maximal expiration (an
unforced maneuver)
Forced Vital Capacity (FVC)
(Liters at BTPS)
The maximum volume of air forcibly exhaled after a maximal inspiration
Forced Expiratory Volume, 1 second (FEVi)
(Liters at BTPS)
The volume of air that is exhaled with maximal force in the first second
Forced Expiratory Flow
25-75% (FEF25-75) (L/sec)
The mean forced expiratory flow in the 25th and 75th percentiles of FVC (also
called maximum mid-expiratory flow [MMEF, MEF])
Ratio of FEVi to FVC (FEVi/FVC)
Proportion of vital capacity exhaled in the first second of forced expiration
Peak Expiratory Flow Rate (PEF or PEFR)
(L/sec at BTPS or L/min)
The maximum flow obtained from a person's maximum forced expiration
starting from the point of a maximal lung inflation
BTPS: Body temperature and ambient pressure saturated with water vapor.
Source: Miller et al. (2005a).
Pulmonary function varies by race or ethnic origin, gender, age, and height and is best
compared when normalized to expected pulmonary function based on these variables (Tepper et
al.. 2012: Pellegrino etal.. 2005: Hankinson et al.. 19991. Studies that did not adjust or otherwise
account for these variables when comparing results between exposure groups were not considered.
Pulmonary function also is associated with smoking status (Becklake and White. 1993). which was
considered in the evaluation of potential confounding. FEVi and PEFR exhibit diurnal variation and
this complicates the interpretation of changes across a work shift or during a laboratory session if
no comparisons were made with an unexposed group fChan-Yeung. 2000: Lebowitz etal.. 19971.
Pulmonary Function Studies in Humans
The synthesis of pulmonary function first discusses responses to acute exposures including
experimental study designs (controlled human exposure studies) or analyses of changes across a
work shift or lab session in occupational groups or medical school anatomy students. Controlled
human exposure studies of pulmonary function change among asthmatic volunteers are
summarized in Section 1.2.3 (Immune-mediated Conditions, Focusing on Allergies and Asthma), but
This document is a draft for review purposes only and does not constitute Agency policy.
1-37 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
their results are most informative to the pulmonary function outcome and are included in the
integration of evidence in this section. Then, panel studies of students in anatomy labs with
intermediate-duration exposure over a period of weeks or months are discussed. Subsequently,
studies of long-term exposures are synthesized involving occupational groups or residential
populations of adults and children. Evidence tables for each exposure setting (see Tables 1-6
through 1-10) are organized by level of confidence in the study's results and then descending
publication year. The table summarizing the studies of occupational exposure are organized first
by study design (cross-sectional, longitudinal), then by confidence in study results and descending
publication year.
Generally, in the included studies of formaldehyde exposure and effects on pulmonary
function, groups exposed to formaldehyde during the course of their jobs experienced TWA
concentrations above 0.2 mg/m3 with intermittent peaks above 1 mg/m3. Students meeting once or
twice a week in anatomy labs experienced fluctuating concentrations during dissections averaging
between 0.1 and >1.0 mg/m3. Formaldehyde concentrations in residential or primary school
settings are much lower and less variable (<0.1 mg/m3). EPA included both the higher exposure
and the lower exposure studies in its evaluation of pulmonary function effects.
Acute and intermediate-duration formaldehyde exposure
Controlled human exposure studies
Formaldehyde exposures (0.62-3.7 mg/m3), lasting from minutes to up to 5 hours, have not
induced pulmonary function deficits in healthy, nonexercising volunteers in controlled human
exposure studies (see Appendix A.5.3 for study summaries). The studies exposed small numbers
(<25) of diverse individuals, often including males and females of varying age, and two included
current smokers [31% of participants in the study described in Andersen (1979) and Andersen and
Molhave (1983), and 13% of the participants in Schachter et al. (1987)]. In some studies, the
variation around the mean change in lung function was quite large suggesting that the response to
exposure was large in some individuals, and in others, the response was small (Schachter etal..
1987: Schachter etal.. 1986b: Witek etal.. 1986).
In contrast to the studies of exposure without exercise, small but statistically significant
deficits in pulmonary function (e.g., decreased FEVi, FVC, FEV3, FEF25-75, specific airways
conductance) during formaldehyde exposures of 2.5 or 3.7 mg/m3 were reported in two studies
from one research group that included two or more 15-minute exercise regimens within the study
protocol (Green etal.. 1989: Green etal.. 1987). These effects were not seen, however, in studies
with shorter exercise segments [8-10 minutes; (Kulle etal.. 1987: Schachter etal.. 1987: Schachter
etal.. 1986b)]. Although the average change in lung function was generally small, some individuals
exhibited clinically significant deficits, even after only 2 hours of exposure, suggesting that
individual susceptibility may be an important consideration (Green etal.. 1987).
This document is a draft for review purposes only and does not constitute Agency policy.
1-38 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Changes in pulmonary function across a work shift or anatomy course lab session
Daily changes in pulmonary function measures (e.g., across a work shift or during a lab
session lasting a few hours) were assessed in studies among workers employed for several years in
exposed jobs or among students enrolled in an anatomy lab. Most of the studies measured changes
only among exposed individuals; measurements in a comparison group would have allowed
adjustment for diurnal effects. One study using repeated peak expiratory flow measures taken by
students trained in the procedure at multiple points during dissection lab sessions found that PEF
declined over the course of a lab and these daily declines became attenuated over successive weeks
fKriebel et al.. 20011. Kriebel et al. (2001) also measured overall changes after a few weeks'
duration (see next section, Exposure durations <1 year).
Several studies reported daily cross-shift change in pulmonary function, although the same
measures were not evaluated by all of the studies (see Appendix A.5.3).The interpretation of
responses in the occupational groups is complicated because workers had significant previous
exposure to formaldehyde (>0.2 mg/m3) and few studies included an unexposed comparison group.
Occupational studies in the wood products or chemical industries reported declines across a shift in
one or more of FEVi, FVC and FEVi/FVC fNeghab etal.. 2011: Herbert etal.. 1994: Alexandersson
and Hedenstierna. 1989: Horvath etal.. 1988: Alexandersson etal.. 19821. However, declines in
these measures were not observed among other cohorts of plywood workers fMalaka and Kodama.
1990). industrial workers (Lofstedtetal.. 2009). workers using acid hardening lacquers
(Alexandersson and Hedenstierna. 1988). nor among funeral workers during an embalming session
(Holness and Nethercott. 1989). The magnitude and direction of changes also were varied among
anatomy students who were assumed to have no prior significant exposure to formaldehyde
fBinawaraetal.. 2010: KhaliqandTripathi. 2009: Akbar-Khanzadeh and Mlvnek. 1997: Akbar-
Khanzadeh etal.. 1994: Chia etal.. 1992: Uba etal.. 19891. The heterogeneity in results cannot be
explained by the study evaluation conclusions indicating confidence in a study's results (high,
medium, low). Studies of exposure in dissection labs that evaluated an unexposed referent group or
measured change in pulmonary function prior to the first lab generally reported that referent
groups also experienced a change (either an increase or decrease) in pulmonary function, further
complicating interpretations.
Daily declines in FEF25-75 were more consistently reported by the occupational studies of
wood products employees fNeghab etal.. 2011: Malaka and Kodama. 1990: Horvath etal.. 1988:
Alexandersson etal.. 1982). and exposed groups had larger decrements compared to the referent
groups among the two studies that reported cross-shift changes in both groups (Malaka and
Kodama. 1990: Horvath etal.. 1988). Further, although FEF25-75 was not reported by Holness et al.
(1985). 2.3 and 8.5% decreases in FEF50 and FEF75, respectively, were observed during embalming
sessions among 22 embalmers, in contrast to a 1.2 and 1.9% increase, respectively, among 13
referent individuals assessed over a 2- to 3-hour period.
This document is a draft for review purposes only and does not constitute Agency policy.
1-39 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
Exposure durations <1 year—changes among anatomy/pathology students
Three panel studies examined pulmonary function changes over the course of 10 weeks,
12 weeks, and 7 months among anatomy students exposed to formaldehyde, with average
concentrations ranging from 0.12 to 6.2 mg/m3 intermittently (once or twice a week: Kriebel etal..
2001: Kriebel etal.. 1993: Uba etal.. 1989): see Table 1-6]. The primary source of formaldehyde
exposure in the laboratory air was formalin, a preservative composed of a mixture of formaldehyde
(37%) and methanol (14%). Methanol is not expected to be associated with pulmonary function
deficits and would not be a strong confounder in these studies (U.S. EPA. 2013). One study that
measured pulmonary function using spirometry did not observe statistically significant declines
over 7 months fFVC. FEV1. FEV1/FVC. FEF25-751. Uba et al.. 19891. Two studies by the same
research group using repeated peak expiratory flow measures taken by students trained in the
procedure at multiple points during the lab sessions suggested an average decline in PEF over 2 to
several weeks related to concentration averaged over the entire duration, as well as reductions
during dissections that decreased in magnitude over time (Kriebel etal.. 2001: Kriebel etal.. 1993).
Cumulative exposure (ppm-minutes) summed over all previous weeks was not a significant
predictor of changes in pulmonary function. The measurement of multiple measures of PEF per
student in the studies by Kriebel et al. (2001; 19931 increased the precision of the mean value and,
consequently, the statistical power to detect a significant change. Interpretation of the analyses by
both Kriebel et al. and Uba et al. is complicated by the consideration that class attendance as well as
formaldehyde concentrations decreased over the semester in the studies (Kriebel etal.. 2001: Uba
etal.. 19891.
This document is a draft for review purposes only and does not constitute Agency policy.
1-40 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-6. Formaldehyde effects on pulmonary function in laboratory settings
(changes over <1 year)
Study and design
Results
Pulmonary function by test day (mean ± SD)
(/V = 96)
Before
exposure
FVC (L)
5.246 ± 1.025
(Day 1)
FEVi(L)
4.379 ±0.846
FEF25-75 (L/sec)
4.492 ± 1.216
FEVi/FVC
0.835
2 Weeks
FVC (L)
5.277 ± 1.027
FEVi(L)
4.409 ± 0.824
FEF25-75 (L/sec)
4.484 ± 1.151
FEVi/FVC
0.836
7 months
FVC (L)
5.308 ± 1.027
FEVi(L)
4.399 ±0.823
FEF25-75 (L/sec)
4.392 ± 1.198
FEVi/FVC
0.829
Reference: Uba et al. (1989)
Panel study, California
Population: 96 medical students (72.5% participation) during a
7-month anatomy class meeting twice a week for 4 hours. Mean
age: 24.3 years, 88% white, 73.8% male, nonsmokers,
12 asthmatics.
Exposure: Personal sampling monitors (impingers) in the breathing
zone, 32 samples during different class periods in 7-month period.
Short-term samples (N = 16) for peak concentrations during
dissection.
Range of TWA formaldehyde: below LOD (0.05 ppm) to 0.93 ppm
(0.06 to 1.14 mg/m3).a
Monthly averages in September, October, and May: 0.6, 0.8, and
0.1 ppm (0.74, 0.98, and 0.12 mg/m3),a respectively.
Peak concentrations: During dissection: mean 1.9 ppm (2.3 mg/m3)'
range 0.1 to 5.0 ppm (0.12 to 6.1 mg/m3),a observing dissection:
mean 1.2 ppm (1.5 mg/m3)a range 0.2 to 2.0 ppm (0.25 to
2.5 mg/m3).a
Methods: Pre- (noon) and postlab spirometric measures (ATS
methods) taken before the class began, after the first 2 weeks, and
after 7 months.
Analyzed using repeated measures ANOVA, adjusted for sex.
Evaluation:3
SB IB Cf Oth
Overall
Confidence
High
Reference: Kriebel et al. (2001) Panel study, USA
Population: 51 gross anatomy students (out of 54 total) during a 12-
week class meeting once per week for 2.5 hours. Mean age:
24.9 years, 23.7% male, two current smokers, four with history of
asthma.
Exposure: Continuous monitoring in six homogenous sampling
zones (LOD = 0.05 ppm). 12-minute work-zone concentrations
calculated per student using sampling data and recorded work
locations.
Geometric mean concentration: 0.7 ppm (0.9 mg/m3)a (GSD:
2.13 ppm). Peak 12-min concentration: 10.91 ppm (13.4 mg/m3).a
Average concentration: 1.1 ppm (1.35 mg/m3)a (SD = 0.56 ppm).
Concentrations decreased over 12-week semester.
Methods: Spirometry (FEVi, FVC) using ATS criteria before 1st
exposure and during 10th week. Pre- and postlab PEF
measurements obtained for at least 1 week for 38 students. PEF as
fraction of value before 1st lab session; individual pre-lab and cross-
lab change data analyzed together in relation to recent, average,
and cumulative formaldehyde in single generalized estimating
equations model. Generalized estimating equations regression
Exposure metrics: Recent exposure = mean
concentration during 2.5-hour lab; cumulative
exposure = ppm-minutes for all previous
weeks;
past average exposure: Cumulative exposure
divided by total number of minutes of
exposure.
PEF as fraction of baseline (before 1st lab)
(L/s per ppm)
(5 (SE) p-value
Recent exposure -1.05 (0.33) 0.002
Recent exposure 0.69 (0.24) 0.004
*ln(wk)
Past average -o.52 (0.30) 0-08
exposure
Cold on lab day -1.67 (0.41) 0.001
No association with cumulative exposure.
Pulmonary function among asthmatics not
This document is a draft for review purposes only and does not constitute Agency policy.
1-41 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
adjusted for cold on lab day.
Evaluation:3
different.
SB
IB
Cf
Oth
Overall
Confidence
1 1
Medium
Attrition and declining concentration over course—bias to healthy
individuals and toward null
Reference: Kriebel et al. (1993) Panel study, USA
Population: 24 clinical anatomy students (out of 25 total) during a
10-week anatomy class meeting once a week for 3 hours. Mean age
PEF (L/min) during course (mean ± SD)
[n = 20)
26,42% male, 1 current smoker, five reported history of asthma.
Exposure: Personal samples in the breathing zone, 1-1.5 hours
sampling periods.
Formaldehyde concentration geometric mean: 0.73 ppm
(0.9 mg/m3),a GSD 1.22; range: 0.49-0.93 ppm
Weeks 1-2 PEF (L/min) 538.9 (86.9)
Weeks PEF (L/min) 529.4 (88.4)
9-10a
Weeks PEF (L/min) 536.6 (86.2)
24-25
(0.6-1.14 mg/m3); 8 samples. No trend in concentrations over
semester.
Pentachlorophenol: ND (LOD = 83 ng/m3.
Methods: PEF measured by trained students pre- and postlab and
1-3 times during lab using Mini-Wright peak flowmeters. Mean
absolute value (SD) pre- and cross-lab change in pulmonary function
analyzed in separate models using multivariate linear models,
including asthma, asthma x week, eye and nose or throat
symptoms.
Evaluation:3
aEnd of course.
Decrement over 10-week course,
P = -2.7 ± 1.1 L/min per week; p = 0.01,
Model included asthma, asthma x week,
eye symptoms, nose symptoms.
SB
IB
Cf
Oth
Overall
Confidence
n
~
Medium
Limited sample size
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.3). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
1 Long-term formaldehyde exposure
2 Occupational exposure
3 Overall, the set of occupational studies indicates that inhalation of formaldehyde over long
4 periods at work is associated with declines in measures of pulmonary function. With only a few
5 exceptions, average values for FEVi, FVC, and FEF measured before a work shift at the beginning of
6 the work week were lower among exposed workers than average values in their referent groups
7 (see Table 1-7). However, the differences were relatively small and some were imprecise. The
8 occupational groups under study were exposed to high average formaldehyde concentrations
This document is a draft for review purposes only and does not constitute Agency policy.
1-42 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
(>0.2 mg/m3) in a variety of industries, including funeral homes (embalming), wood products
(plywood, cabinetry), chemical products (formaldehyde resins), and manufacturing. Employees
had worked at these jobs for at least 5 years, and in a few studies, for more than 10 years. While a
few studies conducted longitudinal analyses, most of the occupational studies were cross-sectional
in design, recruiting only current employees, and likely were limited by lead time bias, a selection
bias that results in attenuated effect estimates. In general, when only current employees are
recruited for a cross-sectional study of an exposure that causes symptoms, there is a possibility that
former workers may have left their jobs to reduce their exposure (lead time bias, healthy worker
survival effect). Further, for studies that recruited from among those present on the day of the
study, if employees were not present because of symptoms related to their formaldehyde exposure,
attenuated effect estimates may have been observed fAlexandersson and Hedenstierna. 1988:
Alexanderssonetal.. 19821.
The healthy worker effect and survivor (lead time) bias raised a concern for selection bias
for several cross-sectional occupational studies, some of which had no other notable limitations
(Lofstedt et al.. 2011 a: Neghab etal.. 2011: Lofstedt etal.. 2009: Milton etal.. 1996: Malaka and
Kodama. 1990: Nunn et al.. 1990: Alexandersson and Hedenstierna. 1989.1988: Holmstrom and
Wilhelmsson. 1988: Alexandersson etal.. 1982: Schoenberg and Mitchell. 19751. In addition, one
study compared pulmonary function values in individuals exposed occupationally to individuals in
a community population, raising a concern about the healthy worker effect and a possible bias
toward a null association (Holness and Nethercott. 1989). Community populations include
employed individuals, as well as people who are unemployed, ill or disabled, or retired, with a
spectrum of health conditions. Among the prospective studies, loss to follow-up of exposed
participants with symptoms because they moved to jobs with less or no exposure, also was evident
f Lofstedt et al.. 2 011 a: Nunn etal.. 1990: Alexandersson and Hedenstierna. 19891. This type of
selection bias also could result in an attenuated effect estimate. For other studies, exposure to
other substances that affect pulmonary function, such as dust or environmental tobacco smoke,
appeared to be more prevalent in the referent group, and was not adjusted for in the analysis, also
resulting in a potential bias toward the null (Herbert etal.. 1994: Main and Hogan. 19831. Despite
the bias toward the null, most studies observed associations of measures of pulmonary function
with formaldehyde exposure, which increased EPA's confidence in their findings.
Figure 1-5 presents forest plots of the difference in mean FEVi, FVC, and FEF25-75 between
exposed and referent groups for 10 study results. Overall, while no difference in means was found
by a few of the 10 studies for one or more of the measures, most of the comparisons indicate that
exposed groups had lower mean values compared to their respective referent group. Studies that
reported only the absolute values or used a different analysis could not be plotted. One study of
laboratory technicians found differences, even though the referent group had relatively high
average formaldehyde exposure [0.125 mg/m3; fKhamgaonkar and Fulare. 19911], Another study
This document is a draft for review purposes only and does not constitute Agency policy.
1-43 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
among workers in the wood products industry reported a decrease in FEVi/FVC but not other
measures fHerbertetal.. 19941.
This document is a draft for review purposes only and does not constitute Agency policy.
1-44 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Author, Year
HCHO
Ref
Wood Products
Malaka (high), 1990
56
93
Alexandersson, 1988
38
18
Horvath, 1988
109
254
Alexandersson, 1982
47
20
Chemical Manufacture
Neghab, 2011
70
24
Lofstedt, 2009
64
134
Schoenberg, 1975
15
15
Embalming
Holness, 1989
84
38
"T
1"
Mean Difference [95% CI]
-3.6 [ -8.9, 1.7]
-8.3 [-15 8,-0 8]
-2.0 [ -4 9 , 0 9 ]
-9.5 [-16.5 ,-2.5]
-12.2 [-18 9,-5.5]
-1.9 [ -5.4, 1.6]
-1.7 [-12.8, 9.4]
-1.5 [ -6 4, 3.4]
-200 -10.0 00 10.0
Mean Difference. FEV 1 (%)
200
Author. Year HCHO
Ref
Wood Products
Malaka (high), 1990
56
93
Holmstrom (Grp 2), 1988
98
36
Alexandersson, 1988
38
18
Horvath, 1988
109
254
Alexandersson, 1982
47
20
Chemical Manufacture
Neghab, 2011
70
24
Lofstedt, 2009
64
134
Holmstrom (Grp 1), 1988
70
36
Schoenberg, 1975
15
15
Embalming
Holness, 1989
84
38
-200
~T
Mean Difference [95% CI]
-0.3 [ -3.6 , 3.0 ]
-8.1 [-12.6,-3.6]
-4.8 [-11.5 , 1.9]
-2.0 [ -4 8 , 0 8 ]
-5.6 [-11 8, 0.6]
-13.9 [-20.6, -7.2]
-0.6 [ -4.1 , 2.9]
-6 6 [-114,-18]
-h 7.9 [ -1.5,17.3]
1
-0.4 [ -4.9, 4.1 ]
-10.0 00 10.0 20.0
Mean Difference, FVC (%)
This document is a draft for review purposes only and does not constitute Agency policy,
1-45 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Toxicological Review of Formaldehyde—Inhalation
Author, Year
HCHO Ref
Mean Difference [95% CI]
Wood Products
Malaka (high), 1990
56 93
-10.4 [-17.1 ,-3.7]
Alexandersson, 1988 38 18 ^
-9.8 [ -22.9 , 3.3 ]
Horvath,1988
109 254
-2.0 [ -7.1 , 3.1 ]
Alexandersson, 1982
47 20
-5.6 [-17 8, 6.6]
-20.0 -10.0 0.0 10.0 20.0
Mean Difference, FEF (%)
Figure 1-5. Forest plots depicting mean difference in pulmonary function
(percentage predicted) between exposed and comparison groups for FEVi,
FVC, and FEF.
The plots include results from eight studies that reported the percentage of predicted normal function
accounting for age, gender, and height, and three studies that reported mean absolute values and mean
reference values for exposed and referent groups from which the percentage of the reference group
could be calculated. The forest plot compares the mean difference between all exposed and referent
groups when available, although one study reported appropriate data only for subgroups [e.g., low and
high exposure categories; (Malaka and Kodama, 1990)1. The average of the standard deviations for a
spirometric parameter specific to an exposure group, weighted by the size of the referent group, was
used when no statistics from the individual study were available (Alexandersson and Hedenstierna, 1988;
Holmstrom and Wilhelmsson, 1988; Alexandersson et al., 1982).
In addition to accounting for age, gender, and height, most of the studies adjusted for
smoking in their statistical analyses or otherwise addressed potential confounding by smoking.
The studies evaluated three types of occupational settings—wood products industries,
chemical production, and mortuaries—and employees in these industries were exposed to other
chemical and physical agents that may co-occur with formaldehyde. Other common exposures in
the wood products industry can include phenols and other solvents contained in resins and glues,
terpenes, and dust, while embalming fluids include methanol. Phenol and terpenes are not
expected to have strong effects on pulmonary function, particularly at the concentrations reported
by the studies. However, occupational exposure to high concentrations of wood dust (>2 mg/m3)
has been associated with reductions in pulmonary function fMandrvk etal.. 20001. Many of the
studies of wood products workers reported measurements for dust, terpenes, and phenol, stating
that levels were a fraction of occupational exposure limits. Studies that either adjusted for dust
This document is a draft for review purposes only and does not constitute Agency policy.
1-46 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
levels or compared effects in formaldehyde-exposed groups with and without dust exposure did
not find an independent effect by dust fMalaka and Kodama. 1990: Holmstrom and Wilhelmsson.
19881. The chemical industries included manufacture of formaldehyde products such as
formaldehyde-phenol or formaldehyde-melamine resins and may involve exposures to phenols,
other alcohols, VOCs, and other compounds, some of which may affect pulmonary function.
However, since a pattern of reduction in pulmonary function was observed across several different
exposure settings, all involving high formaldehyde exposure, confounding by a coexposure becomes
less likely to be an alternative explanation for the observed associations. Three studies conducted
longitudinal analyses of small groups of workers with continued exposure over 4-6 years (Lofstedt
etal.. 2011a: Nunn etal.. 1990: Alexandersson and Hedenstierna. 19891. All three longitudinal
studies measured FEVi and reported no change in the cohorts over the study period. However, one
study of workers at a formaldehyde-urea resin manufacturing factory reported that among exposed
nonsmokers, the annual decline was -45 mL/year (95% CI -28, -62 mL/year), which is 50%
greater than the expected rate of age-related decline in FEVi in nonsmokers (29 mL/year Redlich et
al.. 2014: Lee and Fry. 20101. The annual decline among unexposed nonsmokers in this study was
-29 mL/year, consistent with the expected rate of decline with age. In addition, Alexandersson and
Hedenstierna (1989) reported a decline in FEF25-75 at a TWA concentration of 0.42-0.5 mg/m3.
FEF25 -75 percentage among the carpentry workers declined by -168 ± 46 mL/second (10.1
L/minute) for each year of exposure over a 5-year period (p < 0.001). There was a larger decrease
among nonsmokers compared to smokers, which might not be surprising since decreased
pulmonary function is associated with smoking (-212 mL/sec/yr and -60 mL/sec/yr,
respectively). The annual decrease was corrected for normal aging and reference pulmonary
function spirometry values. The number of years that participants were followed by the three
studies, 4-6 years, is the minimum length of time considered adequate to observe changes with
time f Redlich etal.. 20141. and the size of the exposure groups was quite small. Given the large
amount of within-person variability in these measures when assessed over time, these studies
would have had limited sensitivity to detect a small longitudinal change. Further, the studies were
limited by potential differential loss to follow-up of exposed individuals who may have changed
jobs or left the industry because of the irritation effect of formaldehyde. Despite the low sensitivity
of these studies, some declines in FEVi and FEF25-75 were reported.
Duration of work in an exposed job was associated with decreased pulmonary function
values in two studies fNeghab etal.. 2011: Schoenberg and Mitchell. 19751. but not others
(Holmstrom and Wilhelmsson. 1988: Horvath etal.. 1988: Alexandersson etal.. 1982). These
analyses controlled for age, height, gender, and cigarette smoking. One study examined
associations with cumulative exposure (ppm-years) and observed reductions in pulmonary
function measures (FEVi, FEVi/FVC, and FEF25-75) among male employees at a plywood company
who had worked an average of 6-7 years fMalaka and Kodama. 19901. In addition to other relevant
covariates, this analysis controlled for cigarette smoking and dust levels in the regression model.
This document is a draft for review purposes only and does not constitute Agency policy.
1-47 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 Another study among wood products employees reported no association with a cumulative
2 exposure measure, but did not present the results quantitatively fHolmstrom and Wilhelmsson.
3 1988).
Table 1-7. Formaldehyde effects on pulmonary function in occupational
settings (long-term effects)
Study and design
Results
Prevalence stu
dies
Reference: Horvath et al. (1988)
Cross-sectional study, Wisconsin.
Population: 109 exposed (workers at a particleboard and
molded products operation, 68.6% of all exposed), average age
37.4 ± 11.7 years, 57% males; average work duration in
exposed: 10.3 years (1-20 years). 254 unexposed (workers
from nearby food processing facilities; average age
34.2 ± 12.1 years, 44% male).
53% current and former smokers.
Exposure: 8-hour TWA measured using personal passive
monitors on the day of the exam (LOD 0.15 mg/m3). Area levels
measured with an active sampling train (impingers).
TWA 0.69 ppm, range 0.17-2.93 ppm (0.85 mg/m3, range
0.21-3.60 mg/m3),a and 0.05 ppm, range 0.03-0.12 ppm
(0.062 mg/m3, range 0.037-0.15 mg/m3)b in the exposed and
unexposed industries, respectively.
Other exposures in exposed:
Respirable particulates (PEL 5 mg/m3): median 0.11 mg/m3;
phenol (PEL 5 ppm): mean 0.15 ppm; carbon monoxide (PEL
50 ppm): mean 7.35 ppm; sodium hydroxide (PEL 2 mg/m3):
0.4-0.21 mg/m3; nitrogen dioxide: ND; acrolein: ND.
Methods: Spirometry (volumetric) before and after the work
shift. Pulmonary function (ATS methods) as percentage of
predicted normal compared between exposed and unexposed
(unpaired t-test); multiple linear regression of baseline absolute
values by exposure group, adjusting for age, height, sex, and
smoking.
Evaluation:3
SB IB Cf Oth
Overall
Confidence
High
Comparison of mean preshift pulmonary
function (percentage predicted (SD))
Exposed Referent
FEVi(L)
103 (13)
105 (13)
FVC (L)
105 (12)
107 (13)
FEVi/FVC
96 (8)
95 (8)
PEFR (L/sec)
100 (23)
103 (22)
FEF25-75 (L/sec)
83 (22)
85 (25)
FEF25 (L/sec)
6.91 (2.12)
6.73 (1.98)
FEF50 (L/sec)
4.5 (1.46)
4.38(1.43)
FEF75 (L/sec)
1.63 (0.8)
1.66 (0.77)
p > 0.05
Exposure group was not associated with baseline
absolute values in multiple linear regression
models. Work duration was not associated with
preshift pulmonary function.
This document is a draft for review purposes only and does not constitute Agency policy.
1-48 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Reference: Neghab et al. (2011)
Cross-sectional study, Iran.
Population: 70 male workers at a local melamine-formaldehyde
resin-producing factory with current exposure to formaldehyde
and >2 years work history (mean age 38.2 ± 8.4 years, work
duration 13.2 ± 7.8 years, 24.3% smokers).
24 healthy males from the same industry and comparable
socioeconomic and demographic status, and no present or
former formaldehyde or other exposure to respiratory irritants.
100% participation (mean age 40.0 ± 8.2 years, work duration
14.5 ± 8.1 years, 25% smokers).
Exposure: Area samples (N = 7) in seven workshops with
exposure and one area sample in office area (sampling in
different time points and shifts). Sampling time 40 minutes.
Exposed mean formaldehyde: 0.78 ± 0.4 ppm
(0.96 ± 0.49 mg/m3)b; referent: not detected.
Methods: Pulmonary function tests (Vitalograph COMPACT),
ATS methods) before and at the end of the work shift on the
first working day of week, percentage predicted.
Group comparisons and cross-shift difference among exposed,
and multiple linear regression analysis of pulmonary function
comparing exposed and referent adjusting for smoking, age,
weight, height.
Evaluation:3
Percentage predicted pulmonary function
(mean (SD))
Exposed Referent
Preshift (N = 70) (N = 24)
VC
FVC
FEVi
FEVi/FVC
PEF
77.9 (12.0)a
86.6 (14.5)a
86.6 (14.4)a
100.2 (8.8)
90.9 (15.9)
99.3 (21.0)
100.5 (14.5)
98.8 (14.6)
98.8 (5.3)
89.8 (31.2)
SB IB
Cf
Oth
Overall
Confidence
u
Medium
Healthy worker survivor bias
difference between exposed and referent,
p< 0.025
Difference in pulmonary function between
exposure groups
Regression coefficients (percentage difference; SD
provided by author; p-value):
VC -21.43 (3.48) (p = 0.001)
FVC -13.88 (3.44) (p = 0.001)
FEVi -12.23 (3.42) (p = 0.001)
Change in pulmonary function per year work
duration
Regression coefficients (unit change/year):
VC-0.1 (p = 0.315)
FVC -0.43 (p = 0.02)
FEVi -0.375 (p = 0.035)
FEVi/FVC-0.1 (p = 0.225)
PEF-0.28 (p = 0.2)
Reference: (Herbert et al.. 1994)
Cross-sectional study, Canada.
Population: 99 oriented strand board workers (exposed, 98%
participation), mean age 35.4 years, 51.5% smokers; work
duration 5.1 years; 165 oil/gas field plant workers (not exposed
to formaldehyde or oil and gas vapors) from same geographic
area (82% participation), mean age 34.9 years, 27.9% smokers,
work duration 10 years. Excluded 14 workers in referent with
hydrogen sulfide exposure.
Exposure: TWA formaldehyde and dust concentrations at OSB
plant based on 21 hours of continuous sampling in the
breathing zone at five work sites on 2 separate days.
Formaldehyde range: 0.07-0.27 ppm (0.09-0.33 mg/m3),b dust
mean: 0.27 mg/m3, 2.5 nm diameter.
Methods: Spirometric testing (volumetric, best of five
satisfactory maneuvers) at start of work shift and after 6 hours
(ATS guidelines).
Analysis ANCOVA controlling for age, height, and smoking.
Preshift pulmonary function (mean) by
exposure group
OSB Oilfield
FEVi (mL)
FVC(mL)
FEVi/FVC (%)
4.203
5.364
78.6a
4.223
5.257
80.3a
ap = 0.028
Risk of airway obstruction (FEVi/FVC< 75%) by
smoking category (N = number below criteria)
Odds
Ratio 95% CI
Nonsmokers (17) 1.68 0.54,5.25
Exsmokers (15) 1.08 0.32,3.64
Current (25) 2.98 1.10,8.07
This document is a draft for review purposes only and does not constitute Agency policy.
1-49 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Evaluation:3
SB
IB Cf Oth
Overall
Confidence
Medium
N
Healthy worker survivor bias; possible irritant exposure in
referent, coexposure to dust.
Reference: Khamgaonkar and Fulare (1991)
Cross-sectional study, India.
Population: 74 individuals working in anatomy and
histopathology departments at three colleges and exposed to
formaldehyde. Selected every 2nd person from occupational
list. Comparison group matched by age and sex (N = 74)
(individuals not working in laboratories with formaldehyde).
Comparable for mean height and weight. Excluded persons
with a history of pulmonary disease before their present
occupation.
Exposure: Multiple 30-minute area samples collected in the
breathing zone in both the exposed (N = 43) and unexposed
(N = 18) areas.
Mean (SD) exposed 1.00 ppm (0.556), range 0.036-2.27 ppm
(1.23 mg/m3 (0.68), range 0.044-2.79 mg/m3).b
Referent 0.102 ppm (0.115), range 0-0.52 ppm (0.125 mg/m3
(0.141) range ND-0.64 mg/m3).b
Methods: Pulmonary function tests on a subset of 37 exposed
and 37 comparison individuals on a Monday morning after days
of no exposure.
Evaluation:3
Mean pulmonary function values by
exposure group
Exposed Referent
(A/= 37) (A/= 37)
FVC (L)
MMEFR (L/sec)
FEVi (%)
2.18
1.55
60.68
2.63a
2.71b
78.74a
ap < 0.01, bp < 0.05
SB IB Cf Oth
M
Overall
Confidence
Medium
Possible exposures in referent that affect pulmonary function;
exposure to formaldehyde in referent labs.
Reference: Malaka and Kodama (1990)
Cross-sectional study, Indonesia.
Population: Male employees at plywood company. Exposed
workers (N = 93) randomly selected with stratification by
smoking status and work duration (<5 and >5 years; 93%
participation), mean age 26.6 years, work duration
6.2 ± 2.4 years; unexposed group (N = 93) matched for age,
ethnicity, and smoking status (53%), mean age 28.8 years,
similar in height, work duration 6.7 ± 2.3 years, worked in areas
where formaldehyde was not used, and had no previous or
current exposure to formaldehyde based on occupational
histories; 93% participation rate.
Exposure: Area sampling and personal monitoring. Average
exposed 0.9 ppm (1.1 mg/m3),b range 0.22-3.48 ppm
Mean baseline spirometric values (adjusted
for dust) (SD)
Exposed
Referent
FEVi/FVC (%)
FEVi (L)
FVC (L)
FEF25-75%
(L/sec)
84.7 (6.5)
2.78 (0.41)
3.28 (0.44)
3.04 (0.76)
86.9 (4.9)a
2.82 (0.30)a
3.37 (0.36)
3.44 (0.78)a
ap < 0.005
This document is a draft for review purposes only and does not constitute Agency policy.
1-50 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
(0.27-4.28 mg/m3)b; calculated by EPA from weighted average
of area specific averages in Table 2 in the paper; referent
0.003-0.07 ppm (0.0037-0.09 mg/m3).b
Cumulative exposure measure developed using area
concentrations and duration in current job (mean
6.29 ppm-year, SD 2.72). Categorized into none (N = 93), low
(<5 ppm-yr) (N = 37), and high (>5 ppm-yr) (N = 56).
Other exposures: average total dust 1.35 mg/m3, average
respirable dust 0.6 mg/m3.
Methods: Baseline (Monday) and cross-shift spirometric
measurements (volumetric) followed ATS methods.
Pulmonary function (percentage of expected function) by
category of cumulative exposure analyzed using analysis of
covariance. Stepwise regression of pulmonary function on
cumulative formaldehyde (continuous) adjusted for age, height,
weight, cigarettes/day, and dust.
Evaluation:3
Multiple regression model of pulmonary
function3
P(per ppm-yr
FA)
FEVi/FVC (%)
FEVi (L)
FVC (L)
FEF25-75 (L/sec)
-0.347b
-0.015b
NS
-0.043b
Adjusted for age, height, weight,
cigarettes/day, and dust.
bp < 0.05
Mean pulmonary function (percentage predicted)
(SD) by Categories of Cumulative Exposure
None
Low
High
SB IB
Cf
Oth
Overall
Confidence
u
Medium
Healthy worker survivor bias
FEVi 94.4 (20.0) 87.4 (10.2) 90.8 (12.7)
FVC 92.0(9.2) 87.1(8.4) 91.7(10.4)
FEVi/FVC 86.9 (4.9) 85.3 (6.4) 84.4 (6.5)
FEF25-75 90.4(20.0) 79.5(18.2) 80.0(20.1)
Dust was not associated with any pulmonary
function measures.
Reference: Holness and Nethercott (1989)
Cross-sectional study of funeral workers, Canada.
Population: 67 currently active embalmers and 17 formerly
active, recruited through a list of funeral homes from a district
funeral directors association (86.6% participation). Average
work duration 10 years. Unexposed group (N = 38) recruited
from large service organization and paid student volunteers.
Exposure: Average concentration from two area samples
(impingers), measured during embalming procedures lasting
from 30 to 180 minutes, 0.36 ± 0.19 ppm, range 0.08-0.81 ppm
(0.44 ± 0.23 mg/m3, range 0.10-1.0 mg/m3).b
Unexposed average concentration: 0.02 ppm (0.025 mg/m3).b
Methods: Information on symptoms, past and family medical
history, and work practices by questionnaire.
Spirometry (volumetric) tests on 22 embalmers before and after
embalming procedure and on 13 referents 2-3 hours after first
test.
Pulmonary function (percentage predicted) compared using
multiple regression, correcting for age, height, and pack-years
smoked.
Evaluation:3
Comparisons of baseline pulmonary function
(percentage predicted) (SD)
Exposed
(N = 84)
Unexposed
(N = 38)
FVC
FEVi
FEVi/FVC
FEF50
FEF75
100.5 (12.3)
99.2 (12.9)
98.4(7.9)
104.8 (29.7)
76.2 (32.9)
100.9 (11.5)
100.7 (12.9)
99.4 (8.7)
110.3 (34.5)
86.6 (36.0)
Active [N = 67 Inactive [N = 17)
FVC
FEVi
FEVi/FVC
FEF50
FEF75
100.7 (12.2)
100.8(12.19)
98.9 (7.8)
107.5 (28.7)
80.8(33.1)
95.8 (12.0)a
93.1 (14. l)b
96.6 (8.0)
94.1 (32.3)
57.1 (24.7)
ap = 0.0385, bp = 0.0652
SB IB
Cf
Oth
Overall
Confidence
Medium
-
This document is a draft for review purposes only and does not constitute Agency policy.
1-51 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Comparison groups selected from different source populations.
Reference: Alexandersson and Hedenstierna (1988)
Cross-sectional study, carpentry shop, Sweden.
Population: 38 exposed employees working with
acid-hardening lacquers for the previous 12 months [mean age
(SD): 34 (10) years, mean duration employment 7.8 years] and
at work on the study day. 18 referent employees at the same
company (mean age [SD] 37 [9] years). Asthmatics excluded.
Exposure: Personal exposure monitored during three to four
15-minute periods during the workday. No formaldehyde
measurements reported for referent group.
TWA 0.40 mg/m3, range: 0.12-1.32 mg/m3. Peak concentration
(15 minute) 0.70 mg/m3, range 0.14-2.6 mg/m3.
Additional measurements of solvents and dust (4 hr)—
considered very low compared to Swedish threshold limit
values.
Methods: Spirometry (volumetric) on Monday after 2 days
unexposed and again at end of shift on second day. Half of
referent employees tested before and half tested after shift.
Compared difference from sex, age, and height matched
reference values.
Evaluation:3
Pulmonary function before work on Monday
(Mean difference from reference values)
Exposed
(N = 38)
Referent
(N = 18)
Difference (SD) Difference (SD)
FVC (L)
FEVi(L)
FEV%
FEV25-75
(L/sec)
-0.24 (0.64)*
-0.21 (0.51)**
-0.7 (6.7)
-0.10 (0.98)
0.03 (0.65)
0.15 (0.42)
1.8 (5.3)
0.31 (0.76)
*p < 0.05; **p < 0.01
Difference from reference values greater among
nonsmokers than smokers.
SB IB Cf Oth
Overall
Confidence
Medium
Healthy worker survivor bias; small samples.
Reference: Holmstrom and Wilhelmsson (1988)
Cross-sectional study, Sweden.
Population: 3 study groups: 70 individuals (87% male) in
formaldehyde products group (mean age 36.9 years); 100
furniture workers exposed to formaldehyde and wood dust
(93% males, mean age 40.5 years). Comparison group, 36
persons (56% male, mean age 39.9 years), primarily local
government clerks. 100% participation. Mean duration of
employment 10.4 years for exposed and 11.4 years for referent
group.
Exposure: Mean formaldehyde in 1985.
Group 1: mean 0.26 ± 0.17 mg/m3, range 0.05-0.5 mg/m3, Dust
<1 mg/m3.
Group 2: mean 0.25 ± 0.05 mg/m3, range 0.2-0.3 mg/m3, dust
1.65 ± 1.06 mg/m3.
Referent: mean 0.09 mg/m3.
Data on formaldehyde concentrations available 1979-1984 and
from 1 to 2 hours personal sampling in breathing zone at
different workstations in 1985.
Mean annual exposure estimated for each participant from
start of employment; dose-years.
Pulmonary function values compared to
expected by exposure group
FA FA-dust
exposed exposed Referent
(N = 70) (N = 98) (N = 36)
FVC
Observed
Expected
FEV%
Observed
Expected
4.979a
5.556
80.8
80.6
4.929a
5.593
78.3
79.5
4.539
4.718
81.4
80.7
'paired t-test comparing observed to
expected, p < 0.001.
No correlation of FVC with cumulative
formaldehyde dose or years of service >5 years.
This document is a draft for review purposes only and does not constitute Agency policy.
1-52 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Other exposures (phenol, ammonia, epichlorhydrin, methanol,
ethanol) <1% of occupational exposure limit.
Methods: Spirometric measures analyzed as percentage of
expected normal based on age, sex, smoking, height, and
weight.
Evaluation:3
SB IB Cf Oth
KM
Overall
Confidence
Medium
Healthy workers; comparison groups selected from different
source populations.
Change in pulmonary function per unit exposure
rank (/V = 90)
Reference: Levine et al. (1984b)
Cross-sectional study, USA, 1978.
Population: 105 white, male morticians attending postgraduate
course (94% participation).
Exposure: # embalmings.
Exposure index: rank ordering of the total # embalmings;
divided into categories of low and high exposure based on
# bodies embalmed, matched on age (within 3 years).
Methods: Completed self-reported respiratory disease
questionnaire (ATS) and detailed occupational history;
pulmonary function testing (volumetric spirometer) (N = 99),
analysis of 90 with complete data after excluding pipe and cigar
smokers.
Evaluation:3
SB IB Cf Oth
Overall
Confidence
Medium
Uncertainty regarding assignment of exposure rank.
Variable
Exposure Rank
FVC (L) +0.0003
FEVi (L) -0.0001
FEVi/FVC +0.0019
FEF25-75 (L/s) -0.0016
FEF25-75/FVC -0.0002
Rank FVC/predicted -0.0547
Rank FEVi/predicted +0.0229
FEF25-75/predicted -0.0676
Coefficients were not statistically significant
(p > 0.05).
Multiple regression equations adjusted for age,
height, number of pack-years, and exposure
index.
Comparison of pulmonary function by exposure
group (low, high) in nonsmokers (N = 24); mean
(SE)
Measure
Low
High
FVC (L)
4.69 (0.22)
4.56 (0.32)
FVC %
100.5 (3.1)
98.9 (3.4)
predicted
FEVi (L)
3.80 (0.22)
3.64 (0.27)
FEVi %
108.9 (3.3)
105.5 (4.1)
predicted
FEVi/FVC
0.807 (0.02)
0.797 (0.02)
FEF25-75 (L/sec)
4.28 (0.48)
3.88 (0.49)
FEF25-75 %
117.9 (8.8)
110.5 (11.7)
predicted
Groups matched on age, similar in height
Group comparisons, p > 0.05
This document is a draft for review purposes only and does not constitute Agency policy.
1-53 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Reference: Alexandersson et al. (1982)
Cross-sectional study, Sweden.
Population: 47 exposed carpentry workers employed at the
plant for >1 year and at work on study day (mean age 35 years,
mean duration 5.9 years) and 20 unexposed employees. No
asthmatics were included.
Exposure: TWA concentration, measured using personal
sampling over a working day, 0.47 mg/m3 (range
0.05-1.62 mg/m3).
Other exposures: Terpenes: range ND-9 mg/m3; dust (all
particle sizes) mean 0.5 mg/m3 (range 0.3-0.7 mg/m3).
Methods: Spirometric measurements (volumetric, ATS
methods) Monday morning preshift and after work for exposed.
Pulmonary function was measured in the unexposed in the
morning or the afternoon. Statistical analysis of preshift values
and cross-shift change, two-tailed Student's t-test. Linear
regression of association with duration of employment.
Evaluation:3
Comparisons of pre-shift mean pulmonary
function (SD)
Exposed
Referent
(A/= 47)
(N = 20)
FVC (L)
5.73 (0.14)
6.0 (0.2)
FEVi(L)
4.52 (0.12)a
4.86 (0.15)
FEV%
79.2 (1.0)
80.7 (1.32)
MMF
4.94 (0.2)
5.08 (0.31)
(L/sec)
CV%
16.7 (1.07)
17.1 (1.5)
difference from reference value, p = 0.08
No association with duration of employment
(quantitative results not presented).
SB
IB Cf Oth
Overall
Confidence
Medium
N
Healthy worker survivor bias.
Reference: Schoenberg and Mitchell (1975)
Cross-sectional study, USA.
Population: Employees using formaldehyde-phenol resin in the
filter acrylic wool filter department of a filter manufacturing
plant.
Exposed production line workers and supervisors, N = 63 (94%
of recruited); younger age and cigarette smoking (packs/yr) less
among present line group compared to never on-line.
Exposure: Measurements taken by insurance company during
same month; 0.5-1 mg/m3.
3 breathing zone samples, 10.6-16.3 mg/m3.
Exposure groups
Present line, N = 40
Previous line, N = 8
Never-on-line, N = 15
Some in never-on-line had some exposure.
Other exposures:
Phenol, four breathing zone samples, 7-10 mg/m3.
Methods: Standardized questionnaire, pulmonary function
measured before and after shift on Monday and Friday
(pneumotachometer); 5 maneuvers, average of best two used
to calculate values; compared to predicted based on age,
height, and gender.
Monday preshift pulmonary function by
exposure duration (mean, SEM)
1-4
Never
<1 year
years
>5 years
(N = 15)
(N = 15)
(N = 10)
(,N = 15)
FVCa
104.3
103.7
108.8
112.2
(2.9)
(2.9)
(2.7)
(3.8)
FEVia
98.9
100.7
99.6
97.2
(3.6)
(3.1)
(3.5)
(4.4)
FEVi/FV
79.4
79.9
74.1
71.2
C, %b
(1.3)
(1.4)
(2.2)
(2.6)c
MEF5o%/
90.3
87.1
73.6
64.0
FVC, %b
(4.0)
(6.1)
(8.4)
(6.2)d
percentage predicted
Standardized to cigarette consumption of 15
pack-years
cDifferent from never-on-line group (p < 0.05)
dDifferent from never-on-line group (p < 0.005)
This document is a draft for review purposes only and does not constitute Agency policy.
1-54 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Evaluation:3
SB IB Cf Oth
B
Overall
Confidence
Medium
Healthy survival effect. Multiple exposures: formaldehyde,
phenol. Phenol is an irritant but is not expected to be
associated with pulmonary function at these levels.
Reference: Main and Hogan (1983)
Cross-sectional study, USA.
Population: 21 exposed individuals working in two mobile
trailers for 34 months (mean age 38 ± 9 years, 76% male, 19%
nonsmokers).
18 referent individuals who did not work in the trailers (mean
age 30 ± 6 years, 50% male, 22% nonsmokers).
Exposure: Three 1-hour area samples using impingers taken on
four occasions (August, September, December, April) always on
a Monday. At least one sample from each office in both trailers.
Concentration range 0.12 to 1.6 ppm (0.15-1.97 mg/m3).b
Methods: Volumetric spirometer, percentage predicted FEVi
and FVC stratified by smoking status (unadjusted group means
compared using t-tests).
Evaluation:3
Mean pulmonary function (percentage
predicted)
Exposed
(N = 14)
Unexposed
(N = 17)
FEVi
FVC
FEFso
FEF75
%A FEF50
98
94
93
69
55
99
97
90
70
43
SB
IB
Cf
Oth
Overall
Confidence
Low
¦
¦
Comparison groups selected from different sources (possible
unmeasured confounding), ETS in referent; small sample size
(low sensitivity).
Longitudinal studies
Reference: Nunn et al. (1990)
Prospective study at chemical factory manufacturing urea
formaldehyde resin, Duxford, England.
Population: Exposed: 164 workers, aged 25 or older, exposed to
free formaldehyde in 1980; 29% <35 years, 46% current
smokers, 22% employed >22 years; referent: 129 workers from
bonded structures division at same factory in 1980; 39%
<35 years, 45% current smokers, 4% employed >22 years.
Followed over 6 years (1980-1985).
Exposure: Area samples (1-6 hours) periodically, 1979 and
1985, and personal sampling for representative exposed
workers, 1985 to 1987. Exposure prior to 1976 based on
subjective determinations and knowledge of process changes
and industrial hygiene measures. Pre-1979 levels estimated as
low, medium, and high, corresponding to an 8-hour day.
Decline in FEVi with age by smoking history
(mean slope, mL/year (95% CI)
Smoking
status
Exposed
N
Unexposed
N
Never
-45
26
-29
13
(-28, -62)
(-7, -51)
Ex-
-33
34
-40
31
smoker
(-20, -46)
(-26, -54)
Current
-46
57
-46
36
(-33, -59)
(-32, -61)
Total
-42
117
-41
80
(-34, -51)
(-32, -50)
Among those lost to follow-up, FEVi was less than
predicted among 75% of 12 exposed and 33% of
27 referent compared to 36% of 117 exposed and
This document is a draft for review purposes only and does not constitute Agency policy.
1-55 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
TWA of 0.1-0.5 ppm (0.12-0.62 mg/m3),b 0.6-2.0 ppm
(0.74-2.46 mg/m3),b and >2 ppm, respectively.
Other exposures: Records examined for random sample of 20
per group; more exposure to asbestos, carbon and glass fibers,
siliceous fillers, aliphatic amines in referent group; both groups
exposed to phenol and urea formaldehyde resin (not free
formaldehyde).
Methods: Data on FEVi and FVC (volumetric spirometer)
highest of two readings within 5% of each other) obtained from
routine annual health screenings conducted by the same nurse
throughout the study period. Follow-up complete for 76% of
exposed and 74% of unexposed. FEVi values adjusted for
height (FEVi/height3), regressed on time of screening visit for
each worker, adjusted for age in 1980, smoking status in 1980,
and at final assessment, maximum and mean exposure,
assessment level, and total duration of exposure.
Evaluation:3
45% of 80 referent followed.
Concern for selection bias: loss to follow-up higher among
exposed with low pulmonary function compared to referent;
referent exposed to other potential irritants.
Reference: Alexandersson and Hedenstierna (1989)
Prospective occupational study, follow-up of Alexandersson et
al. (1982), Sweden.
Population: 47 exposed cabinetry workers and 20 unexposed
workers examined in 1980, 34 exposed and 18 unexposed were
examined again in 1984. Of the 34 originally exposed, 13 had
been reassigned to other unexposed jobs. Average exposure
duration among exposed and transferred workers: 11 years.
Exposure: Personal monitoring during 3 or 4 15-minute periods
during workday.
TWA 0.42 ± 0.27 mg/m3 in 1980 and 0.50 ± 0.12 mg/m3 in 1984.
Other exposures: terpenes ND; respirable dust: mean
0.1 ± 0.2 mg/m3.
Methods: Spirometric measures (volumetric, ATS methods)
compared with reference values for sex, age, height, and
weight. 5-year change corrected for age-dependent change.
Results presented by smoking status.
Evaluation:3
Annual change(1980-1984) in exposed, mean
(SD)
Smokers Nonsmokers
All
(N = 10) (N = 11)
(N = 21)
FVC (mL/yr)
-15 (24) -10 (26)
-12 (16)
FEVi(mL/yr)
-15 (21) -31 (20)
-24 (20)
FEVi/FVC
-0.1(0.4) -0.4 (0.2)a
-0.3 (0.3)
(%/yr)
FE F25-75
-60 (69) -212 (66)a
-168 (46)
(mL/s/yr)
CV% (%/yr)
-0.6 (0.3) 0.2 (0.4)
-0.2 (0.3)
SB IB Cf Oth
S
Overall
Confidence
Medium
ap < 0.001, compared to predicted normal
Pulmonary function was unchanged among
referent group.
Pulmonary function was correlated with
formaldehyde concentration in unadjusted
regression analysis.
Pulmonary function improved after a 4-week
holiday.
Healthy worker survivor bias; small sample.
This document is a draft for review purposes only and does not constitute Agency policy.
1-56 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Reference: Lofstedt et al. (2011b)
Prospective study; follow-up of Lofstedt et al. (2011a),
Sweden.
Population: One of four foundries opted out of follow-up, plus
39 were lost to follow-up. 25 of 64 workers from 2009 study
involved with Hot Box method; 55 of 134 referents from 2009
study working outside core-production and die-casting halls;
not exposed to chemicals. Prevalence of childhood allergy
lower in exposed than in referent in 2005 (4 vs. 31%, p < 0.05);
higher prevalence of nasal symptoms among referent in 2005.
Exposure: Formaldehyde, isocyanic acid, and methyl isocyanate
measurements on same day as spirometry.
Monoisocyanates: Mean of 4 to 5 15-minute samples
Formaldehyde: sampling over entire shift
Individual exposure estimated for 2001 and 2005 (mg/m3);
levels 50% lower in 2005 (mean, range).
2001 0.098 (0.094)
2005 0.045 (0.043)
0.014-0.44
0.01-0.19
SB IB Cf Oth
Overall
Confidence
Low
Limited sample size to detect small changes between 2001 and
2005; concern for survivor bias; coexposure to methyl
isocyanate and isocyanic acid in exposed—unable to
differentiate for comparisons of change from 2001 to 2005.
Decreased across shift pulmonary function
reported in 2001 was correlated with decreased
preshift pulmonary function in 2005.
VC r = 0.51, FEV r = 0.57, p < 0.05
Preshift value and change in pulmonary
function (percentage predicted), 2001-2005
2001 2001-2005
Mean (SD)
Mean
(SD)
Range
VC
Exposed 93.3(12.1) -0.8(4.2) -11.2-6.5
Referent 93.9(10.8) -0.4(3.8) -11.0-5.9
FEVi
Exposed 94.4 (11.6) -1.3 (5.5) -14.0-8.8
Referent 96.3 (11.6) 0.3 (5.3) -13.8-10.3
Correlation low between formaldehyde and either methyl
isocyanate (r = -0.20) or isocyanic acid (r = 0.09); 61% of
exposed were coremakers where formaldehyde levels were
highest and isocyanate levels were lower.
Methods: Pulmonary function by spirometry (volumetric) using
ATS guidelines. Pre- and postshift after 2 days with no
exposure. Percentage predicted using Swedish reference.
Regression analysis of formaldehyde adjusted for MIC, smoking,
and childhood allergy.
Evaluation:3
Across shift change was not different between
exposure groups (data not provided).
No association of formaldehyde with change in
pulmonary function at follow-up in regression
analysis (data not provided).
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.3). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
Concentrations reported by authors as ppm or ppb converted to mg/m3.
This document is a draft for review purposes only and does not constitute Agency policy.
1-57 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Exposure in residences or school
Adults
Results among four studies of residential exposure among adults are difficult to compare
because different methods were used to assess pulmonary function and two of the studies did not
report results quantitatively (Norback etal.. 1995: Broder et al.. 1988c) (see Table 1-8). A cross-
sectional study of residential formaldehyde exposure in a large, representative sample in Arizona
observed an association with declines in PEFR among adult smokers at formaldehyde
concentrations between 0.049 and 0.172 mg/m3, but not among the group as a whole
fKrzvzanowski etal.. 19901. Another study among elderly nursing home residents observed an
elevated risk of low pulmonary function (defined as values falling in the lower 20% of the
distribution) in association with formaldehyde concentrations above the median level measured in
each nursing home (Bentaveb etal.. 2015). The overall median and range of formaldehyde
concentrations was 0.007 mg/m3 and 0.001-0.021 mg/m3, respectively, but the concentrations
associated with elevated risks varied according to the median in each nursing home. Two
additional studies that assessed effects of formaldehyde exposure on pulmonary function in
primarily adult residential populations exposed to concentrations between 0.009 and 0.279 mg/m3
reported no associations, although the outcomes evaluated by each study were not directly
comparable (Norback etal.. 1995: Broder et al.. 1988c).
The study by Krzyzanowski et al. (1990). which used the most thorough exposure-
assessment protocol and included repeated measurements of PEF (thus enhancing the ability to
detect an association at the lower concentrations found in the homes) was interpreted with high
confidence. Of the residential studies, only Krzyzanowski et al. (1990) examined effect modification
by smoking status. Confidence in the regression results by Norback et al. (1995) is low because
most of the measured formaldehyde concentrations were less than the LOD and the sensitivity of
the study was low. Overall, results from the small set of studies suggest that adults in general did
not experience declines in pulmonary function at average formaldehyde levels less than
0.05 mg/m3; however, declines may be experienced at lower concentrations among susceptible
subsets (e.g., elderly, smokers).
Table 1-8. Formaldehyde effects on pulmonary function among adults in
residential settings
Study and Design
Results
Reference: Krzyzanowski et al. (1990);Quackenboss et al.
(1989c)
Cross-sectional study, Arizona, USA.
Population: A stratified random sample of 202 households of municipal
employees, selected based on information about potential exposure
(age of housing) and potential susceptibility obtained from an initial
screening questionnaire. Households with children aged 5-15 years
(613 adults and 298 children) were eligible for inclusion.
Change in PEFR (L/min) in relation to
indoor formaldehyde, ages >15 yrs.
(N = 526; 8,463 observations)
Formaldehyde (household 0.09 (0.27)
mean)
Morning formaldehyde (vs. -5.9 (1.1)a
bedtime)
Bedroom formaldehyde -0.07 (0.04)b
This document is a draft for review purposes only and does not constitute Agency policy.
1-58 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and Design
Results
Mean age: >15: 37 years, percentage male: >15: 43.4%, percentage
white: >15: 70.4%, 24.4% current smokers.
Asthma prevalence: >15: 12.9%.
Exposure: Sampling: two one-week samples from each individual's
kitchen, living area, and bedroom using passive sampling tubes
(sensitivity 12 ng/m3 for 1 week, 15% accuracy).
Average formaldehyde concentration, 26 ppb [0.032 mg/m3],b
maximum 140 ppb, [0.172 mg/m3].b
The majority of subjects (83%) lived in homes with 2-week average
concentrations below 40 ppb [0.049 mg/m3].b
Methods: Trained subjects measured peak expiratory flow rates
(PEFRs) using Mini-Wright peak flow meters four times daily, in the
morning, at noon, in the early evening, and before bed, for 2 weeks.
The largest of three test results was recorded for each test period.
Analysis of PEFR in relation to indoor formaldehyde concentration,
random effects model adjusting for asthma status, smoking status, SES,
N02 levels, episodes of acute respiratory illness, and time of day.
Analysis performed separately for ages younger and older than
15 years.
Evaluation:3
x morning
Morning x smoking -7.4 (2.6)a
Bedroom 0.59 (0.13)a
formaldehyde x morning x s
moking
Bedroom -0.007
formaldehyde2 x morning x (0.001)a
smoking
Constant 491.7 (8.5)
ap < 0.05, b0.05 < p < 0.10
In adults, only the morning PEFR values
were affected by formaldehyde
concentrations. Smoking status was shown
to affect the relationship between PEFR
and formaldehyde exposure.
SB IB Cf Oth
Overall
Confidence
High
Reference: Bentaveb et al. (2015)
Cross-sectional study, 2009-2011; 7 European countries.
Population: 600 elderly residents (20 randomly selected per home)
permanently living in randomly selected nursing homes (8 per city) in
selected city in seven countries. Exclusion criteria stated (neurological
or psychiatric disorders), 71.8% female, 62.8% >80 years old, 35%
active smokers, 13.8% passive smoking.
Exposure: Measurements in common room; 1-week samples; also
measured particulates, N02, ozone, temperature, humidity and C02;
range of 1 week averages 0.001-0.021 mg/m3, median 0.006 mg/m3;
categorical (low and high) based on median concentration in each
nursing home.
Methods: Assessed by same team in all countries; medical visit and
standardized questionnaire (European Community Respiratory Health
Survey); lifetime COPD (ever told by doctor; spirometry (ATS/European
Respiratory Society guidelines), percentage predicted. General
estimating equations analysis, accounting for correlations within
nursing homes; adjusted OR (95% CI) for risk of values <20% of
distribution; stratification by presence of ventilation.
Evaluation:3
Association of formaldehyde (cutpoint
median in the nursing home) with
pulmonary function among elderly nursing
home residents
aORa 95% CI
FEVi
FVC
FEVi/FVC < 70%
1.12
1.16
0.46
0.97-1.28
1.06-1.28
0.12-1.66
aaOR: adjusted OR
Stratification by poor (n = 436) or adequate
(n = 105) ventilation.
FEVi aOR (95% CI), 2.65 (1.29, 5.45).
SB IB Cf Oth
Overall
Confidence
Medium
This document is a draft for review purposes only and does not constitute Agency policy.
1-59 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and Design
Results
Confounding by coexposures was not assessed; range of average
concentrations within low and high exposure categories associated
with overall effects is not known.
Reference: Broder et al. (1988b, 1988c); Broder et al. (1988a)
Cross-sectional study, February 1983-March 1984, Toronto, Canada.
Population: 1,726 occupants from 517 households with urea
formaldehyde foam insulation (UFFI) identified from registry
maintained by Urea Formaldehyde Foam Insulation Information and
Coordination Centre, Consumer and Corporate Affairs, Canada (50%
male, mean age 40 years, 80% over 16 years, 18% current smokers).
231 referent households (n = 720) selected at random from streets
adjacent to UFFI households (49% male, mean age 35 years, 20%
current smokers). Interviewers and respondents were not blinded with
respect to the focus of the study or the presence of UFFI insulation.
Exposure: Formaldehyde sampling 5 hours on 2 successive days in
central hallway, all bedrooms and in yard.
Inside: referent 0.035 ppm, range 0.006-0.112 ppm [0.043 mg/m3,
range 0.007-0.138 mg/m3].b 90% 0.061; UFFI 0.043 ppm, range
0.007-0.227 [0.053 mg/m3, range 0.009-0.279 mg/m3],b 90%
0.073 ppm.
Outside: referent 0.005 ppm, UFFI 0.005 ppm.
Carbon dioxide sampled in central hallway and in yard (as indication of
ventilation).
Methods: Questionnaire on symptoms and household characteristics,
spirometry (minimum of three satisfactory tests, recorded largest
value). Testing on ages 10 years and older.
Statistical comparisons by group and within group (multiple linear
regression), adjusted for date of examination, gender, age, race,
height, smoking, total hours spent in house per week.
Evaluation:3
Formaldehyde concentration within group
was not associated with pulmonary
function in multiple regression models
adjusting for covariates listed in column,
"Study and Design," (results not
presented).
Between-group comparisons were not
informative for formaldehyde associations
because formaldehyde concentrations
were comparable.
SB IB Cf Oth
Overall
Confidence
Medium
For within-group analyses,
formaldehyde.
Results not presented quantitatively for
Reference: Norback et al. (1995)
Cross-sectional study, Uppsala, Sweden.
Population: 88 men and women (47 with asthma symptoms and 41
without) who agreed to participate (57%) from a group of 154 eligible
randomly selected from 488 preliminary subjects from general
population of Uppsala in 1990, aged 20-44 years. Mean duration in
homes 6 years (range 0.5-31 years).
Exposure: Field measurements: October 1991—April 1992.
Formaldehyde (one 2-hour sample) and guanine (house dust mites) in
the bedroom at pillow height. Room temperature, air humidity, VOCs,
respirable dust, and C02 in living room and bedroom.
Formaldehyde mean (range):
29 (<5-110 ng/m3) in homes of those with nocturnal breathlessness.
FEVimean percentage predicted (SD):
106% (13%).
PEF mean variability (range): 5% (1-18%).
FEVi percentage predicted and PEF
variability (during the day) were not
associated with log-transformed
formaldehyde concentration using
Kendall's rank correlation test (data not
presented).
This document is a draft for review purposes only and does not constitute Agency policy.
1-60 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
Study and Design
Results
17 (<5-60 ng/m3) in homes without symptoms.
Formaldehyde and VOCs concentrations were correlated and could not
be evaluated in same regression model (no data presented).
Methods: Structured interview, spirometry (N = 82), blinded to
exposure.
FEVi spirometry, percentage predicted; multiple regression model,
Kendall's rank correlation test.
Evaluation:3
SB
IB
Cf
Oth
Overall
Confidence
Low
Exposure: Low sensitivity, most exposed to concentration
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
HCHO (ppb}
Figure 1-6. Association of PEFR measured at bedtime and in the morning with
household mean formaldehyde concentration among children less than
15 years of age (Krzvzanowski et al.. 1990).
Reproduced with permission.
Two other studies among children evaluated exposure to formaldehyde at home (Franklin
etal.. 20001 and at school fWallner etal.. 20121. The range of formaldehyde concentrations was
similar to those in the homes evaluated by Krzyzanowski et al. (1990). While no associations were
reported for FVC or FEVi by either of the two studies that evaluated these measures fWallner etal..
2012: Franklin et al.. 2000). Wallner et al. (2012) also measured maximal expiratory flow at 50 or
75% of FVC (MEF50 and MEF75) and observed an approximate 3% decrease per standard deviation
increase in formaldehyde concentration measured in elementary school classrooms. Several
pollutants were evaluated by this study and a few also were associated with MEF75. These
pollutants, benzylbutylphthalate and polybrominated diphenylether congeners, both measured in
dust, would be expected to originate from different sources than formaldehyde, and therefore,
would not be expected to be highly correlated with formaldehyde in air. The exposure contrast in
the homes evaluated by Franklin et al. (2000) was relatively small, limiting the ability of the study
to detect an association with formaldehyde. The interquartile range was 0.011-0.035 mg/m3, and
concentrations between 0.062 and 0.107 mg/m3, which was the range in the higher exposure
group, were found only in 10 homes.
The studies of formaldehyde exposure in homes and schools are limited in their ability to
detect a small reduction in pulmonary function associated with formaldehyde exposure at
concentrations below 0.1 mg/m3 (see Table 1-9). However, a methodologically robust study
reported an association with reductions in peak expiratory flow rate (PEFR) in this concentration
range (Krzyzanowski et al.. 1990). These findings are supported by declines in MEF50 and MEF75
(but not other measures) in a second, more limited study (Wallner etal.. 2012).
This document is a draft for review purposes only and does not constitute Agency policy.
1-62 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-9. Formaldehyde effects on pulmonary function among children in
residential or school settings
Study and design
Results
Reference: Krzvzanowski et al. (1990); Quackenboss et al. (1987)
Cross-sectional study, Arizona.
Population: A stratified random sample of 202 households of municipal
employees, selected based on information about potential exposure (age
of housing) and potential susceptibility obtained from an initial screening
questionnaire. Households with children aged 5-15 years (613 adults and
298 children) were eligible for inclusion.
Mean age: <15: 9.3 years, percentage male: <15: 50.2%, percentage white:
<15: 67.3%, Asthma prevalence: <15:15.8%.
Exposure: Sampling: two 1-week samples from each individual's kitchen,
living area, and bedroom using passive sampling tubes (sensitivity 12 pg/m3
for 1 week, 15% accuracy).
Average concentration, 26 ppb [0.032 mg/m3],ba maximum 140 ppb,
(0.172 mg/m3).b
The majority of subjects (83%) lived in homes with 2-week average
concentrations below 40 ppb (0.049 mg/m3).b
Methods: Trained subjects measured peak expiratory flow rates (PEFRs)
using Mini-Wright peak flow meters four times daily, in the morning, at
noon, in the early evening, and before bed, for 2 weeks. The largest of
three test results was recorded for each daily test period (e.g., morning,
bedtime).
Analysis of PEFR in relation to indoor formaldehyde concentration, random
effects longitudinal model including morning and bedtime formaldehyde
concentration, adjusting for asthma status, smoking status, SES, N02 levels,
episodes of acute respiratory illness, and time of day. Analysis performed
separately for ages younger and older than 15 years.
Evaluation:3
Change in PEFR (L/min) in relation to indoor
formaldehyde, random effects longitudinal
model, ages <15 (N = 208; 3,021 observations)
Factor
ft(SE)
Formaldehyde (household
mean, ppb)
Morning formaldehyde (vs.
bedtime)
Bedroom formaldehyde
*morning
Bedroom formaldehyde
squared *morning
Morning*asthma
Bedroom
formaldehyde*morning*
asthma
Bedroom formaldehyde
squared *morning*asthma
Constant
-1.28 (0.46)a
-6.1 (3.0)a
0.09 (0.15)
0.0031 (0.0015)a
4.59(9.60)
-1.45 (0.53)a
0.031 (0.006)a
349.6(13.2)
ap < 0.05, b0.05 < p < 0.10
PEFR decreased in children as formaldehyde
concentrations increased with a difference noted
between the measurements taken in the morning
vs. bedtime. The morning PEFR was further
decreased in children with asthma.
SB IB Cf Oth
Overall
Confidence
High
Reference: Wallner et al. (2012)
Cross-sectional study; Austria.
Population: 433 children (aged 6-10 years) with spirometry of 596 eligible
(72.7%) in two classrooms each at 9 of 19 schools that volunteered to
participate in study (50% male). 53% of the children were exposed to
environmental tobacco smoke at home.
Exposure: Pollutant measurements for 252 agents: 2 samples in each
classroom, 1 per season (autumn, spring).
Formaldehyde: 24-hour sampling period.
34 chemicals selected for statistical analysis were those with substantial
variation across schools based on an arbitrarily selected criterion (ratio of
between-school variance to the pooled within-school variance >4).
Methods: Questionnaire completed by parents, spirometry assessed at
school between 8:30 am and 12:30 pm by trained technician, ATS protocol
except 6-second minimum exhalation time (not feasible in children).
Values expressed as percentage of reference based on age, gender, height,
and weight. Regression of log-transformed values on mean concentration
of chemical adjusted for education and occupation of parents, urban/rural
Percentage change in pulmonary function (95%
CI) per 1 SD change in formaldehyde
concentration
% Change
95% CI
FVCa
-0.94
-3.29, 1.35
FEVi3
-2.16
-4.80, 0.41
MEF75b
-3.31
-6.6, -0.08
MEFso
-2.60
-4.31, -0.91
Associations with ethylbenzene, m-, p-xylene,
and o-xylene in air, tris (1,3-dichlor-2-propyl)-
phosphate in particulate matter, and
benzylbutylphthalate (FEVi only) and
polybrominated diphenylether congeners in
dust were statistically significant.
bAssociations with benzylbutylphthalate and
polybrominated diphenylether congeners in
dust also were statistically significant.
This document is a draft for review purposes only and does not constitute Agency policy.
1-63 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
residence, and # smokers at home. No adjustment of statistical significance
criterion for multiple comparisons (exploratory).
Evaluation:3
SB IB
Cf
Oth
Overall
Confidence
Medium
¦
No adjustment for coexposures in classroom that were also associated with
pulmonary function, but correlation not anticipated.
Reference: Franklin et al. (2000)
Cross-sectional study, Australia.
Population: 224 healthy children (116 girls, 108 boys) with no current or
history of upper or lower respiratory tract disease based on responses to
respiratory health questionnaire and household inventory distributed
through local primary schools.
Age provided by author: <50 ppb, 9.5 years (SD 1.6); >50 ppb, 9.2 years
(SD 1.9).
Exposure: 3 to 4-day passive samples collected in the child's bedroom and
the main living area of the house, average of both rooms; 214 homes.
TWA categorized into two groups: <50 ppb (0.062 mg/m3)b and >50 ppb
(10 homes).
Additional information from author:
Mean (SD): 20.1 ppb (15.6) (0.025 mg/m3)a; range ND-86.6 ppb
(ND-0.107 mg/m3)b.
Median (IQR): 15.6 ppb (0.019 mg/m3)a (range 9.2-28.1)
(0.011-0.035 mg/m3).b
Methods: Clinical respiratory measures obtained at children's hospital.
Measured spirometry (ATS guidelines), exhaled nitric oxide, and skin prick
tests for seven common allergens.
Evaluation:3
SB IB Cf Oth
Overall
Confidence
Medium
Mean pulmonary function (SD) by exposure
group3
<50 ppb
>50 ppb
FVC (L)
2.21(0.55)
2.18 (0.46)
Percentage
99.1(10.2)
101.4 (7.3)
predicted
FEVi
1.89 (0.46)
1.83 (0.24)
Percentage
96.3(11.1)
97.2 (5.4)
predicted
FEV/FVC (%)
89.1(9.2)
93.1 (11.3)
aNot reported; data provided to EPA by author;
percentage predicted based on age, sex, and
height.
eNO levels by exposure category
HCHO(ppb)
eNO (ppb)
Range
>50
15.5
10.5-22.9
<50
8.7a
7.9-9.6
ap = 0.002, linear regression adjusted for age,
atopic status.
Limited exposure contrast; few subjects in high exposure group.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.3).3). SB = selection bias;
IB = information bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of
limitation. Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be
likely to be toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be
likely to be away from the null (i.e., spurious or inflated effect estimate).
bConcentrations reported by authors as ppm or ppb converted to mg/m3.
1 Evidence on Mode of Action for Decrements in Pulmonary Function
2 Although an MOA for formaldehyde-related effects on pulmonary function remains
3 incompletely defined, it is considered likely that these associations involve the indirect activation of
4 sensory nerve endings in the lower respiratory tract (LRT) or increases in airway eosinophils, or
5 both (see Figure 1-7). Moderate evidence exists for the mechanistic changes that could be directly
6 related to decrements in pulmonary function (e.g., inflammatory changes in airway structure), and
This document is a draft for review purposes only and does not constitute Agency policy.
1-64 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
moderate or robust evidence supports the linkages between events in this pathway. However, the
initial cellular or tissue modifications that ultimately lead to these later events are not understood,
and given the limitations of the available studies, it is unclear whether certain events would be
triggered at low-exposure levels. It is also possible that structural and functional changes in the
upper respiratory tract (URT) might contribute to decreased pulmonary function, for example,
through narrowing of the upper airways or an altered release of cytokines or other soluble
mediators in the URT; however, these possibilities are considered unlikely to be significant drivers
of these effects (see additional discussion below]. Overall, the airway inflammatory changes, which
may be at least partially related to indirect activation of sensory nerve endings, is judged as likely to
be an incomplete mechanism by which formaldehyde inhalation could cause decreased pulmonary
function. As the mechanistic event(s) critical to understanding the observed relationship remain
unknown, including how sensory nerve endings in the LRT might be stimulated without
distribution of inhaled formaldehyde to the LRT, it is expected that important insights would be
gained with additional study, particularly studies testing longer exposure durations. Although
much of the mechanistic support is from studies in experimen tal animals, it is expected that related
mechanisms are operant in exposed humans and could contribute to the consistent decrements in
pulmonary function observed in the available epidemiology studies. Variation in sensitivity is likely
to be affected by underlying respiratory health status and the exposure history of the individuals,
including exposure to known allergens.
O -<
I* oxidative Sensory nerve 't' LRT rieuro- LRT micro- 1s Eosinophils
stress in LRT stimulation in LRT peptides vascular leakage in LRT*
> < * ~ -o
airway edema/ Decreased pulmonary
inflammatory function
structural change*
URT protein/ DNA ¦f1 oxidative ¦f' URT URT mucociliary URT epithelial
modification stress in URT neuropeptides dysfunction damage
Mucus membrane Decreased pulmonary
change in URT function
Legend
Plausibly an initial
effect of exposure
Key feature of decreased
pulmonary function
EVIDENCE
Robust
( ) Moderate
( : Slight
RELATIONSHIP
Robust
--> Moderate
Slight
*effects are amplified
with allergen exposure
Figure 1-7. Possible mechanistic associations between formaldehyde
exposure and decreased pulmonary function.
An evaluation of the formaldehyde exposure-specific mechanistic evidence informing the potential for
formaldehyde exposure to cause respiratory health effects (see Table 1-10 and Appendix A.5.6) identified
these sequences of mechanistic events as those most directly relevant to interpreting effects on
pulmonary function. Evidence of airway inflammatory changes, including eosinophil recruitment to both
This document is a draft for review purposes only and does not constitute Agency policy.
1-65 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
the upper and lower respiratory tract (URT and LRT; upper pathway), is considered as likely to represent
an incomplete mechanism by which formaldehyde inhalation could cause decreased pulmonary function,
although whether certain events occur at lower exposure levels is unclear, and other unexplored
mechanistic events are expected to contribute. URT modifications, primarily structural changes (bottom
pathway), may also contribute; however, this is not interpreted as likely to be a significant contributing
mechanism.
The most plausible support for a mechanism(s) that explains the observed decreases in
pulmonary function includes evidence of increased airway eosinophils and other immunogenic
changes that could be attributed to sensory nerve activation in the LRT (presumably, the vagus
nerve) of exposed rodents, although the potential involvement of LRT sensory nerve stimulation is
poorly studied (i.e., slight evidence). It is expected that LRT sensory nerve activation would be
reliant on a secondary response to TRP channel-activating stimuli increased in the LRT via indirect
mechanisms, such as increased LRT oxidative stress or inflammatory mediators, or both, released
from activated immune cells. This response is unlikely to result from direct stimulation of the
nerve by inhaled formaldehyde or in response to cellular damage, as inhaled formaldehyde is
unlikely to reach the LRT in appreciable amounts and overt epithelial damage in the LRT is not
supported by the available evidence (see Appendix A.5.6). While it might also be explained by a
central trigeminal-to-vagal neural reflex response to irritation of the URT (i.e., a "nasobronchial"
reflex8), the existence of this reflex in humans is debated and a clear scientific consensus does not
exist fGiavina-Bianchi etal.. 2016: Sahin-Yilmaz and Naclerio. 2011: Togias. 2004.19991.
Stimulation of sensory nerve endings can cause a localized release of neuropeptides.
Accordingly, moderate evidence indicates that formaldehyde exposure results in increased LRT
neuropeptides, including substance P, typically at formaldehyde concentrations >2.5 mg/m3, with
coherent moderate evidence for rapid activation of the primary receptor for substance P, the
neurokinin (NK1) receptor, after acute exposure to higher formaldehyde levels. Further, the
activation of the substance P pathway has been experimentally linked to formaldehyde-induced
leakage of the LRT microvasculature. Airway edema and related inflammatory structural changes
(i.e., in airway bronchi), which have been reported in experimental animals following short-term
formaldehyde exposures ranging from >0.3 to >3 mg/m3 and which appear to be exacerbated by
prior allergen exposure, may represent consequences of increased microvascular leakage and
inflammation (see below). To date, potential experimental linkages between these structural
changes and sensory nerve stimulation or substance P signaling have not been studied after
formaldehyde exposure. Similarly, while these changes could lead to an increased permeability to
bronchoconstrictors such as histamine, and while substance P itself can increase the
responsiveness of airway smooth muscle, these endpoints were generally unexamined in the
available studies. Any or all of these immunogenic changes could plausibly contribute to airway
8Note: neural reflexes involving afferent and efferent activity of the vagus nerve (e.g., across different LRT
regions), some of which may involve C fibers and TRP channels, are better established fMazzone and Undem.
20161
This document is a draft for review purposes only and does not constitute Agency policy.
1-66 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
narrowing or obstruction and affect pulmonary function, although airway obstruction would
generally be expected to require much higher exposure levels or effects that cumulate over an
extended period of time. Importantly, however, the majority of the evidence available to inform
these immunogenic changes is from studies of short-term exposure.
Substance P and NK1R signaling has been implicated in establishing the successful
recruitment and adhesion of eosinophils to inflamed airways, and it can promote immune cell
survival and activation through the release of cytokines and chemokines (Mashaghi etal.. 20161.
Moderate evidence for an association between formaldehyde exposure and increases in LRT
eosinophils was identified, including amplification of the response of these cells in rodents
previously exposed to allergens. Considering the evidence across the URT and LRT, a generalized
increase in airway eosinophils after formaldehyde exposure is supported by robust evidence.
Increased airway eosinophils have been reported following exposure of laboratory rodents for
several weeks at effective concentrations above 0.5 mg/m3, with increases generally not being
observed following acute exposure. Recruitment of eosinophils to the airways might be related to
the moderate evidence for LRT markers of oxidative stress, as eosinophils can release toxic
mediators, including lipid-active factors and reactive oxygen species (again noting that it is
considered more likely that any oxidative stress increases would result from changes in
inflammatory factors and immune cells in the LRT, rather than LRT epithelial damage). However,
the activation characteristics of the recruited airway eosinophils, including factors released, have
not been defined, preventing a more complete understanding of whether and how these cells might
decrease pulmonary function in these contexts.
As noted above, modifications to the URT respiratory epithelium could also result in
changes that might indirectly affect pulmonary function. Such modifications include potential
effects on immunological functions, such as an altered release of secreted factors from damaged
epithelial cells, or effects on structural functions (e.g., modified clearance or barrier processes due
to dysfunction of the mucociliary apparatus or cell type transitions, or narrowing of upper airways
due to inflammation or proliferation). If increased URT cytokines or other soluble mediators were
to reach the LRT, they could contribute to decreased pulmonary function through airway
hyperreactivity or hypersensitivity to challenges such as allergen exposure (Hulsmann and
Deiongste. 19961. However, it is expected that most immune factors released from URT respiratory
epithelial cells are tightly controlled and locally acting, and that modest increases would be unlikely
to have significant effects on the lower airways and lungs. Similarly, it is reasonable to presume
that physical modifications to the URT would need to be severe to cause a noticeable change in
function, which would not be expected with typical exposure scenarios. Direct, formaldehyde-
specific examinations of any such associations between the robust evidence for structural URT
changes and LRT effects were not identified, further limiting the interpretation of this potential
association.
This document is a draft for review purposes only and does not constitute Agency policy.
1-67 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
While evidence for some events at low formaldehyde levels (e.g., <1 mg/m3) exists, some of
the more convincing associations have only been tested at high formaldehyde concentrations.
Additionally, the supporting mechanistic evidence is generally from studies of short-term (i.e., days
to weeks) exposure. Therefore, the relevance and sensitivity of the proposed mechanistic pathways
to chronic, low-level exposure scenarios is uncertain. It is also presumed that several important
mechanistic events are currently unidentified. In particular, the initial effects of formaldehyde
exposure that lead to the LRT changes remain undefined, although speculative, untested scenarios
explaining the associations can be hypothesized based on the data available. Similarly, no
explanation exists for the observed exaggerated effects on some mechanistic events following prior
allergen exposure. Overall, however, although a definitive MOA has not been fully identified,
several contributing mechanistic events interpreted with moderate or robust evidence appear to
impact pulmonary function and, taken together, these data provide support for the biological
plausibility of formaldehyde exposure-induced decreases in pulmonary function (See Table 1-10).
Table 1-10. Mechanistic evidence most informative to the occurrence of
decreased pulmonary function after formaldehyde inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
Modifications in the nose and upper airways
Modification
of biological
macro-
molecules
(see
Appendix A.2
and A.4 on
ADMEand
Genotoxicity
for additional
detail)
High or Medium
Human: No direct evidence [note: binding of formaldehyde
to albumin and other soluble proteins in human mucus has
been demonstrated in vitro; e.g., (Bogdanffv et al.,
1987)1; hemoglobin adducts are observable after months-
to-vears exposure at ~0.2 mg/m3(Bono et al., 2012).
Consistent with its known
chemistry, formaldehyde
can modify cellular
macromolecules, including
DNA, and interact with
soluble factors such as
albumin and glutathione,
after exposure to low levels
(e.g., <0.5 mg/m3) across a
wide range of exposure
durations.
Robust
Animal: Multiple animal studies testing various exposure
durations demonstrate that inhaled formaldehyde can bind
and modify biological macromolecules, which is consistent
with the known biological reactivity of formaldehyde;
evidence includes increased DNA-protein crosslinks (DPXs),
hydroxymethyl (hm) DNA adducts, and reactions with
glutathione [e.g., increased DPXs are observed at
>0.37 mg/m3 (Casanova etal., 1989)1; and hmDNA
adducts and protein adducts are observed at >0.86 mg/m3
(Edrissi et al., 2013b; Lu et al., 2011; Lu et al.,
2010a).
o
—1
N/A: Sufficient information for 'robust' from high or medium confidence studies.
Impaired
mucociliary
function
(see
Appendix A.5.6
for additional
detail and
discussion)
High or Medium
Human: Decreased mucus flow at >0.3 mg/m3 after acute
exposure and pathological changes in mucociliary clearance
in workers at mean exposed levels of 0.25-0.26 mg/m3 after
chronic exposure (Holmstrom and Wilhelmsson,
1988; Andersen and Molhave, 1983).
Decreased mucus flow and
ciliary beat, and impaired
clearance, in humans and
rats at >0.25 and
>2.5 mg/m3, respectively
(observed across exposure
durations), eventually
leading to cilia loss.
Robust
Animal: Mucociliary function was generally unaffected at
<0.57 mg/m3 after short-term exposure, with minor changes
noted at the next exposure level, around 2.5 mg/m3; robust
changes were observed at the next highest concentrations
This document is a draft for review purposes only and does not constitute Agency policy.
1-68 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
tested, >7.27 mg/m3 after acute or short-term exposure;
there was a general lack of recovery with longer exposure
duration (e.g., (Monticello et al., 1989; Morgan et
al., 1986a; Morgan et al., 1986c); see
Appendix A.5.6).
O
-J
Human: Increases in ciliary activity at 1.23 mg/m3 in
dissociated human nasal epithelial cells (Wang et al.,
2014), with decreased ciliary beating frequency in human
epithelial cells at >3.46 mg/m3 (Wang et al., 2014;
Schafer et al., 1999): in vitro, acute exposure.
Suggestive of decreased
ciliary beat and ciliastasis at
>5 mg/m3 in humans and
animals withacute
exposure, and ciliary
damage at >0.5 mg/m3 with
short-term exposure;
usually preceded by initial
effects including slight
increases in activity.
Animal: Ciliastasis and mucostasis after acute exposure in
vitro (Morgan et al., 1984): frog palates at
>5.36 mg/m3 (with early activity increases, even at
1.69 mg/m3); structural cilia changes were also observed
(Monteiro-Riviere and Popp, 1986): short-term
exposure at >0.5 mg/m3; and (Abreu et al., 2016):
acute exposure at 0.25, but not 1.2-3.7 mg/m3.
Structural
change in URT
mucus
membrane or
nasal
obstruction
High or
Human: Membrane hypertrophy, atrophy, rhinitis
(Lvapina et al., 2004): chronic (yrs) exposure at
0.87 mg/m3.
Mucus membrane damage
and swelling in humans at
0.87 mg/m3 with chronic
exposure
Moderate
(particularly
in persons
with nasal
damage)
Animal: None
£
o
Human: Data suggest increased mucosal swelling, nasal
obstruction or rhinitis in workers by (Holmstrom and
Wilhelmsson, 1988): chronic exposure at0.26 mg/m3,
and (Norback et al., 2000): short-term exposure at
<0.016 mg/m3, which did not increase in severity with longer
exposure; increased mucosal swelling was also noted in
symptomatic nasal distress patients, but not healthy controls
(Falk et al., 1994): acute (2-hr) exposure at
>0.073 mg/m3.
Observations at
<0.26 mg/m3 in humans or
at >3.5 mg/m3 in rats
support data from the
chronic duration study and
suggest increased acute
vulnerability of people with
a prior nasal condition.
Animal: Rhinitis and necrosis in rats after acute or short-term
exposure, generally at >3.5 mg/m3 (see Appendix A.5.5).
URT epithelial
damage or
dysfunction
(see
Section 1.2.4
for additional
detail)
High or Medium
Human: Indirect data indicating epithelial damage, including
loss of ciliated cells, in occupational studies at 0.1 to
>2 mg/m3 (Ballarin et al., 1992; Holmstrom et
al., 1989c; Edling et al., 1988; Holmstrom and
Wilhelmsson, 1988; Edling et al., 1987a), with
some equivocal findings (Bovsen et al., 1990);
however, these histopathological symptom scores included
hyperplasia and metaplasia, which complicate
interpretation.
Duration-dependent
epithelial damage, typically
at >2.5 mg/m3 in
subchronic or chronic rat
studies, and with
supportive indirect findings
from human studies at
0.1-0.2 mg/m3, generally
correlates with inhibited
mucociliary activity.
Robust
Animal: Increased epithelial damage and related nasal
lesions [e.g., (Andersen et al., 2010)1: duration
dependent, typically >2.46 mg/m3 in subchronic and chronic
studies, with general correlation with inhibited mucociliary
activity; goblet cell loss noted in monkeys (Monticello et
al., 1989): short-term (1 wk) exposure at 7.38 mg/m3;
This document is a draft for review purposes only and does not constitute Agency policy.
1-69 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
indirect evidence mRNA or miRNA changes associated with
apoptosis (Rager et al., 2014; Rager et al., 2013):
short-term (2-d in macques or 28-d in rats) exposure at
>2.46 mg/m3.
o
Human: None
Studies suggest that nasal
epithelial damage is
increased, even in
short-term studies, at
>2.5 mg/m3.
Animal: Goblet cell damage and decreased junctional
proteins between epithelial cells in rats (Arican et al.,
2009): subchronic (12-wk) exposure at 18.5 mg/m3; mRNA
changes in DNA repair in rats (Andersen et al., 2010):
short-term (1-wk) exposure, but not longer (4- to 13-wk)
durations at >12.3 mg/m3; rhinitis and necrosis in rats after
acute or short-term (1- to 3-d) exposure at >3.94 or
4.43 mg/m3.
¦f URT
oxidative
stress
See Section 1.2.1, Evidence on mode of action..., for a description of the direct and indirect
evidence of elevated reactive oxygen species (ROS), possibly at very low concentrations (e.g., at
>0.066 mg/m3) with prolonged exposure.
Moderate
This document is a draft for review purposes only and does not constitute Agency policy.
1-70 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
'T* Neuro-
peptide
release
High or Medium
Human: None
Indirect evidence after
subchronic exposure in a
mouse study at
2.46 mg/m3; Indirect
evidence for acute
activation of the receptor
for substance P in rats at
>18 mg/m3.
Moderate
(for 'T* neuro-
peptides)
Moderate
(for NK1R
stimulation)
(note:
relevant to
both URT and
LRT)
Animal: Increased substance P in plasma in mice (Flliimaki
et al., 2004b): subchronic exposure at 2.46 mg/m3;
microvascular leakage in rats (ItO et al., 1996): acute
exposure to 18.45 mg/m3; this was inhibited by NK1 receptor
antagonists (note: substance P binds NK1R).
Low
Human: Substance P in nasal lavage (in URT) is increased in
human volunteers with ocular exposure (He et al.,
2005): 4-d (5-min/d) exposure at 3 mg/m3, not 1 mg/m3.
Data suggest formaldehyde
activates TRP channels on
sensory neurons, leading to
release of CGRP and
substance P, with acute or
short-term exposure at
>1 mg/m3. An inhibitor
study in isolated rat LRT
tissue provides evidence of
NK1R involvement,
although the relevant
inhalation exposure levels
are unknown.
Animal: In URT models, formaldehyde stimulates release of
calcitonin gene-related protein (CGRP) in in vitro models
relevant to inhalation exposure of the URT (Klinkler et
al., 2011); experiments usingthe related chemical,
acrolein, suggest this is TRPAl-mediated (Klinkler et al.,
2011).
In LRT models, inhibition of substance P receptor (NK1R)
inhibited formaldehyde-induced currents in isolated rat
trachea (LUO et al., 2013); increased substance P and
CGRP in mouse BAL, both amplified with ovalbumin (OVA)
sensitization, and both involved TRP activation (Wu et al.,
2013): short-term exposure at 3 mg/m3.
Nasal cellular
inflammatory
response
High or Medium
Human: None
Cellular infiltration
observed by histology,
primarily neutrophils, but
indirectly supporting other
immune cell infiltration, in
short-term animal studies
at 7.38 mg/m3. Indirect
evidence of increases in
granulocytes (and possibly
lymphocytes) at
2.46 mg/m3 with short-
term exposure.
Moderate
(^ granulo-
cytes:
neutrophils
and
eosinophils)
(Note: data
on lympho-
cytes were
indeterm-
inate)
Animal: Increased inflammatory response, mostly
neutrophils but also mention of lymphocytes and other
inflammatory cells (e.g., assumed monocytes, basophils and
eosinophils) (Monticello et al., 1989): short-term (1-
or 6-wk) exposure at 7.38 mg/m3; "inflammatory cell"
infiltration (Andersen et al., 2008): acute or short-
term (1-d to 3-wk) exposure at 7.38 mg/m3; miRNA changes
associated with inflammation in rats and nonhuman
primates (Rager et al., 2014; Rager et al., 2013):
short-term (1- or 4- wk, with some miRNA changes reversible
with 1-week recovery) exposure at 2.46 mg/m3; in rats, 35
formaldehyde-responsive transcripts in the nose known to
be related to immune cells indirectly indicated increases in
granulocytes (i.e., eosinophil and neutrophil markers) and
lymphocyte changes (Andersen et al., 2010): short-
term (1-wk, but not >4-wk) exposure at >12.3 mg/m3.
Low
Human: N/C in nasal lavage cell counts, but increased total
protein (Priha et al., 2004): occupational^ exposed (8-
hr shift) 0.19 mg/m3; allergy-independent increased
eosinophils, permeability (albumin index) and total protein
in lavage (Pazdrak et al., 1993): acute (2-hr) exposure
at 0.5 mg/m3; increased eosinophils, leukocytes, and
permeability (albumin index) in lavage (Krakowiak et
al., 1998): acute (2-hr) exposure at 0.5 mg/m3(reversible);
indirect evidence of eosinophil infiltration (increased
Suggestive of cellular
inflammation, particularly
eosinophils, at 0.5 mg/m3
and indirect markers of
eosinophil recruitment at
lower levels in humans,
following acute exposure;
neutrophil inflammation
observed at >6 mg/m3 in
This document is a draft for review purposes only and does not constitute Agency policy.
1-71 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
Modifications
in th
markers: lysozyme and eosinophil cationic protein), but not
neutrophils, at very low levels (Norback et al., 2000):
<0.02 mg/m3 for unknown duration (likely >months) in
schools.
rats with short-term
exposure.
Animal-. Neutrophil inflammation (Monteiro-Riviere
and Popp, 1986): short-term exposure at >6 mg/m3.
e lower airways
'T* Lower
respiratory
tract (LRT)
microvascular
leakage
High or Medium
Human: None
Demonstrated increased
leakage from acute
exposure >6.15 mg/m3 in
1 study, which appears to
be mediated by
substance P.
Moderate
(only
examined in
acute
studies)
Animal: Increased in rats(ltO et al., 1996): acute
exposure at >6.15 mg/m3; note: inhibited at 18.45 mg/m3 by
NK1 receptor antagonist (note: substance P binds NK1R), but
not histamine or bradykinin antagonists.
£
o
Human: None
One study suggests acute
exposure as low as
1.23 mg/m3 induces
microvascular leakage,
although continued
exposure appeared (at least
in the near-term) to result
in less leakage.
Animal'. Transiently increased in rats (Kimura et al.,
2010): acute exposure at >1.23 mg/m3 (duration-
independent); note: leakage blocked by inhibiting mast cells,
but not blocking cyclooxygenases; indirect mechanistic data
following injection of formalin into the trachea, causing
leakage that appeared to be dependent on substance P
release after stimulation of C-fiber afferents (Lundberg
and Saria, 1983).
'T* Airway
edema or
other
inflammatory
structural
changes
High or Medium
Human: None
Bronchial edema in one
short-term study at
0.31 mg/m3.
Moderate
(may require
high
exposure
levels or
allergen
sensitization
to elicit
pronounced
changes)
Animal: Increased edema in lung bronchi, but not alveoli,
without signs of inflammation in lower airways in guinea pigs
(Riedel et al., 1996): 5 d at 0.31 mg/m3, not at
0.16 mg/m3.
£
o
Human: None
Airway structural changes
with allergen sensitization
in two species (and, to a
lesser extent, without
sensitization) with short-
term exposure at
>3 mg/m3.
Animal'. Airway structural changes consistent with
inflammation (e.g., wall thickening; cell infiltration) in mice
(Jung et al., 2007), some evidence for which was slight
(Wu et al., 2013; Liu et al., 2011a), and in mice and
rats sensitized with OVA (Wu et al., 2013; LiU et al.,
2011a; Qiao et al., 2009), but not in nonsensitized
rats (Qiao et al., 2009): all 2- to 3-wk exposure at
>3 mg/m3 [Note: most studied bronchial airways].
LRT sensory
nerve
activation
High or
Human: None
No evidence to evaluate
Slight
(levels
required for
potential
activation are
unknown;
may involve
TRPA1
binding)
Animal: None
o
Human: None
A single acute rat study and
indirect evidence from
potentially related
exposures suggest that
lower airway sensory nerve
afferents may be activated,
but the inhaled
formaldehyde levels
required for such potential
activation have not been
Animal'. With acute exposure, dose-dependent increase in
nerve currents and CI" release in intact rat trachea (LUO et
al., 2013), with supporting evidence of substance P and
NK receptor involvement. Indirectly, increased substance P
and CGRP were observed in mouse lung tissue, both were
amplified with OVA, and both were dependent on TRP
activation (Wu et al., 2013): short-term exposure at
3 mg/m3. Note: the potential involvement of
This document is a draft for review purposes only and does not constitute Agency policy.
1-72 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
tracheobronchial reflexes, as is shown with direct LRT
stimulation by irritants including cigarette smoke
constituents and capsaicin (e.g., (Widdicombe, 1998)),
may provide indirect support.
experimentally
demonstrated.
This document is a draft for review purposes only and does not constitute Agency policy.
1-73 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
¦f LRT
oxidative
stress
High or Medium
Human: Increased exhaled nitric oxide, a noninvasive marker
of lower airway inflammation and oxidative stress, in healthy
or asthmatic children (Flamant-Hulin et al., 2010;
Franklin et al., 2000): unknown duration (likely
months to years: classrooms or homes) at 0.04-0.06 mg/m3,
but not in elderly nursing home patients at lower levels
(Bentaveb et al., 2015) for unknown duration (likely
months to years) at 0.005-0.01 mg/m3.
Increased biomarkers
(indirect evidence) of
oxidative stress in children
at >0.04 mg/m3, but not in
elderly individuals at
<0.01 mg/m3 with
prolonged (months-years)
exposure, with indirect
support from a subchronic
rat study at >6 mg/m3.
Moderate
(observed in
children at
low levels:
~0.04 mg/m3)
Animal: Increased iron and zinc, indirect markers of potential
oxidative stress, in lungs of male rats: 13 weeks at
>6.15 mg/m3(Ozen et al., 2003).
£
o
-J
Human: None
Multiple studies in two
species suggest elevated
oxidative stress at
>1 mg/m3 with short-term
exposure.
Animal: In mice: NO and NOS activity increased with 3 d
exposure at 3 mg/m3 (Yan et al., 2005), GSH levels
decreased with 3-wk exposure at >0.5 mg/m3 (Ye et al.,
2013b), and increased ROS or lipid peroxidation markers
were observed with 3-wk exposure at >1 mg/m3 (Ye et al.,
2013b) or 2-wk exposure at >6.15 mg/m3 (Jung et al.,
2007), but decreased with acute exposure in one study
(Matsuoka et al., 20 10): 24-h exposure at
0.12 mg/m3.
In rats: short-term studies at >12.3 mg/m3 demonstrated
increased total oxidant levels and decreased total
antioxidant level (Avdin et al., 2014), increased lipid
peroxidation markers and protein oxidation markers (Sul et
al., 2007), and decreased gamma-glutamvl transpeptidase
(indirect evidence) (Dinsdale et al., 1993).
¦f LRT
eosinophils'1
(see Appendix
A.5.6for
discussion of
LRT evidence
on other cell
types and
soluble factors)
High or
Human: None
Increased after subchronic
exposure to 2.5 mg/m3 in
mice coexposed to antigen.
Moderate
(with short-
term
exposure at
>0.5 mg/m3;
note:
moderate
evidence for
increases in
total BAL cells
or total white
blood cells,
under similar
conditions;
see Appendix
A.5.6)
Animal: 'T* in rats at 2.5 mg/m3 with coexposure to the
antigen, ovalbumin (ova) (Fujimaki et al., 2004b).
o
—j
Human: Two studies did not observe increases following
acute exposure at 0.1 mg/m3 ((CaSSet et al., 2007);
note: trend toward ^) and 0.5 mg/m3 (Ezrattv et al.,
2007) with allergen coexposure (i.e., dust mite antigen;
pollen).
Evidence of increases with
short-term exposure (in
general, at >0.5 mg/m3) in
both rats and mice; the
data suggest that changes
may not occur after acute
exposure.
Animal: 1s in four short-term studies of mice in the absence
of antigen [12.3 mg/m3; (Jung et al., 2007)1, with
antigen (>~12.3 mg/m3 with house dust mite antigen;
(Sadakane et al., 2002)a), or both with and without
antigen
(at 0.5-3 mg/m3 ± OVA (LiU et al., 2011a), and at
3 mg/m3 ± OVA (Wll et al., 2013)); "T* in one short-term
rat study at 0.5-3.1 mg/m3 with OVA antigen (Qiao et al.,
2009)
One acute rat study did not observe effects at 6.2 mg/m3
without antigen (Kimura et al., 2010).
This document is a draft for review purposes only and does not constitute Agency policy.
1-74 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
aReported as 0.5% formaldehyde solution; concentration assumed to be >12.3 mg/m3 (Sadakane et al., 2002).
bThere was also slight evidence for increases in eosinophil attractant and adhesion factors (see Appendix A.5.6).
Integrated Summary of Evidence for Pulmonary Function
Duration of exposure appears to play an important role in epidemiological associations for
pulmonary function. Declines in pulmonary function measures have not been observed by
controlled human exposure studies of short-term formaldehyde exposure among healthy
volunteers, although one research group reported that longer exercise periods (15 minutes)
resulted in small changes. Controlled studies of pulmonary function responses to formaldehyde
inhalation among volunteers with asthma also did not observe changes in this potentially sensitive
group (see Section 1.2.3, Table 1-19). One exception was a heightened response to a dust mite
challenge in the formaldehyde inhalation arm compared to the clean air exposure in one study that
used nose clips, although a different study did not observe an increased response in a study with a
similar design but using a pollen challenge and no nose clip. Studies of change across the work shift
or during pathology labs reported mixed results, which are difficult to interpret because most
studies did not evaluate changes in an appropriate referent group.
Associations with long-term formaldehyde exposure were observed more consistently;
measures of FEVi, FVC, FEVi/FVC, and expiratory flow rates were generally lower in highly exposed
occupational groups compared to their nonexposed or lesser-exposed comparison groups. While
the direction of the associations was generally consistent, some effect estimates were imprecise.
The differences may be a result of individual variability, lower sensitivity in some studies to detect
small mean differences or changes, or random variation. Another source of variation may be
incomplete control for confounders (e.g., smoking, dust, other pollutant exposure), although some
studies did adjust for these factors and still observed an independent association with
formaldehyde, and associations were found among groups with different exposure settings.
Smoking, health status, and lifestage may increase sensitivity to inhaled formaldehyde. The
limited number of population-based studies evaluating lower exposure levels indicates that while,
in general, no associations were observed among adults, declines were reported for smokers and
the elderly living in nursing homes. The study with the strongest design and methods found an
association with declines in PEFR among adult smokers and increasing average formaldehyde
concentration between 0.049 and 0.172 mg/m3 fKrzvzanowski et al.. 19901. In this large,
population-based sample, the investigators also observed a linear relationship between increased
formaldehyde exposure and decreased peak expiratory flow rate (PEFR) among children exposed
to average concentrations of 0.032 mg/m3 (26 ppb), and a stronger response was observed among
children with asthma. This finding is supported by declines in some of the pulmonary function
measures in a more limited study in schools (Wallner etal.. 2012).
While there were very few studies in humans that inform potential biological mechanisms
(i.e., several studies indirectly support inflammatory changes in the LRT), experimental evidence
This document is a draft for review purposes only and does not constitute Agency policy.
1-75 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
primarily from animal studies provides robust or moderate evidence of mechanistic changes that
can be plausibly associated with effects on pulmonary function, including increases in airway
eosinophils and other inflammatory airway changes that appear to be at least partially dependent
on indirect activation of sensory nerve endings in the LRT. Taken together, the data provide what is
likely to be an incomplete mechanism explaining how formaldehyde exposure might result in
decreased pulmonary function. Uncertainties remain regarding the initial cellular or tissue
modifications that ultimately lead to the observed mechanistic changes in the lower airways, and it
is unclear whether certain events would be triggered with chronic, low-level exposure.
Overall, based on moderate human evidence from observational epidemiology studies, with
corresponding slight evidence for an effect in animals based on mechanistic studies supporting
biological plausibility, the evidence indicates that long-term inhalation of formaldehyde likely
causes decreased pulmonary function in humans given the appropriate exposure circumstances.
The primary support for this conclusion includes a study of children and adults in a residential
setting (mean, 0.03 mg/m3, maximum 0.17 mg/m3) and several studies of workers with long-term
exposure to >0.2 mg/m3 (see Table 1-11). The evidence is inadequate to interpret whether acute
or intermediate-term (hour to weeks) formaldehyde exposure might cause this effect (see
Table 1-11).
Table 1-11. Evidence integration summary for effects on pulmonary function
Evidence
Evidence judgment
Hazard determination
Long-term Exposure (years)
Human
Moderate for Lona-Term Exposure (years), based on:
Human health effect studies:
• 1 high and two medium confidence studies in residential and school
populations indicating that susceptible individuals may experience
reduced pulmonary function at lower average concentrations
(<0.05 mg/m3).
• Numerous high and medium confidence studies showing a pattern of
reduced mean pulmonary function in formaldehyde-exposed occupational
groups across a variety of exposure settings and countries. However,
some inconsistencies were noted for specific measures; possible
explanations may be random variation and low study sensitivity.
• Concentration-related associations from four high and medium
confidence adjusted analyses indicate an independent association for
formaldehyde exposure suggesting confounding is not an alternative
explanation.
• Longitudinal declines were reported for one occupational population and
a panel study of medical students, but null or equivocal associations were
identified from other studies, all with possible differential loss to follow-
up and low sensitivity.
Biological Plausibility: Some indirectly supportive mechanistic information
from well-conducted human studies exists related to increased lower airway
oxidative stress following exposures likely to span months to years.
The evidence indicates that
long-term inhalation of
formaldehyde likely causes
decreased pulmonary function
in humans given appropriate
exposure circumstances3
Primarily based on a study of
children and adults in a
residential setting (mean,
0.03 mg/m3, maximum
0.17mg/m3) and several studies
of workers with long-term
exposure to >0.2 mg/m3
Potential Susceptibilities:
Variation in sensitivity is
anticipated to depend on age
and respiratory health, with the
potential for children to be
more sensitive.
Animal
Slight, based on:
This document is a draft for review purposes only and does not constitute Agency policy.
1-76 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Biological Plausibility: Robust and moderate evidence for several mechanistic
events, primarily from experimental animal studies, provides support for
inflammatory changes in the lower airways, including eosinophil increases,
which appear to be at least partially dependent on indirect stimulation of
sensory nerve endings. While evidence exists for some changes in the range
of 0.3-0.5 mg/m3 with exposure for several weeks, some potential
associations in the identified, incomplete MOA pathway have only been
tested at higher (i.e., >1 mg/m3) levels and with shorter-term exposures.
Animal health effect studies: Not formally evaluated.
Other
inferences
Relevance to humans: The observed mechanistic changes are expected to
occur in humans, given similarities across species in the systems that appear
to be involved, and some support is based on studies in both humans and
animals (e.g., lower airway oxidative stress).
MOA: Not established, but likely to involve airway eosinophil increases and
stimulation of airway sensory nerve endings.
Acute or Intermediate-Term Exposure (hours to weeks)
Humans
Indeterminate for Acute or Intermediate-Term Exposure (hours to weeks),
based on:
Human health effect studies:
Small reductions in two controlled human exposure studies of healthy
volunteers (1 lab) with longer exercise periods (15 min), but no associations
with other exposure protocols (including those with <10 min exercise
periods) in studies involving healthy subjects or asthmatics (see discussion
above and Section 1.2.3 for pulmonary function results in asthmatics);
inconsistent results among studies of medical school dissection labs and
cross-shift measurements in occupational studies.
Biological Plausibility: Increases in lower airway eosinophils were not
observed in the few low confidence acute studies in humans available.
The evidence is inadequate to
draw judgments regarding acute
or intermediate-term exposure
(hours to weeks)
Animals
Indeterminate, based on:
Biological Plausibility: Although some mechanistic changes relevant to
pulmonary function, including most of the immunogenic effects, were
altered after short-term exposure in animals, given the dependence of some
of the key mechanistic events (e.g., URT damage or dysfunction; LRT
oxidataive stress) on exposure duration, it remains unclear whether the
potential mechanistic pathways would be relevant to interpreting acute or
short-term exposure scenarios.
Animal health effect studies: Not formally evaluated.
1
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2
1.2.3. Immune-mediated Conditions, Focusing on Allergies and Asthma
2 This section examines the evidence pertaining to the effect of formaldehyde exposure on
3 immune-mediated responses, primarily in the respiratory system, focusing on allergy-related
4 conditions (e.g., rhinitis, rhinoconjunctivitis) and asthma; sensitization related to dermal exposure
5 is not a focus of this review. Experimental animal studies were ultimately concluded to be
6 unsuitable models (i.e., indeterminate) for evaluating allergy-related conditions and asthma as
7 apical outcomes (see discussion in Immune-mediated Conditions, Focusing on Allergies and Asthma,
8 in Animal Studies). Additionally, a few studies that indirectly suggested that respiratory immune
This document is a draft for review purposes only and does not constitute Agency policy.
1-77 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
function (i.e., the ability to respond to infection) could be affected by formaldehyde exposure are
introduced. However, in the context of the health effects data available, it was determined that
these particular findings were better suited to discuss within the wider context of potential
mechanistic changes that might explain respiratory health hazards (see Appendix A.5.6 and
discussion below in Evidence on MOAfor Immune-mediated Conditions), rather than as an
independent health hazard to be evaluated. The mechanistic studies considered most relevant to
these health outcomes provided biological support for the immune-mediated conditions observed
in humans, although complete and definitive MOAs could not be established and several changes
thought to be important to the development or progression of asthma, in particular, were not
identified. The few available studies on developmental immunotoxicity in animals
(hypersensitivity studies) were indeterminate in regard to the information necessary to draw
conclusions.
The general population studies in children and adults provided evidence of an association
between formaldehyde exposure and prevalence of rhinitis or rhinoconjunctivitis, with a relative
risk of approximately 1.2 for formaldehyde exposures of around 0.04-0.06 mg/m3. Although the
effect size was small, these are relatively common conditions and could result in a large impact in
the population. A stronger association (two-fold risk) was seen in the only study of eczema.
Eczema, while not indicative of an allergic respiratory response, is often associated with other
allergic disorders, including those affecting the respiratory system [e.g., allergic rhinitis; (Weidinger
and Novak. 2016a. b), and it appears that some inhaled allergens may have the potential to
exacerbate this condition (Mendell etal.. 2011: Morren et al.. 1994). The available general
population studies also provided evidence of an association between formaldehyde exposure and
the prevalence of current asthma, as determined by symptoms or medication use in the past
12 months in studies with some participants exposed above 0.05 mg/m3, but associations were not
seen in settings with an exposure range less than 0.05 mg/m3. The two studies examining asthma
control or severity among children with asthma suggest associations may be seen at lower
exposures (e.g., 0.04 mg/m3) in this potentially susceptible population. Relatively strong
associations were seen in studies examining prevalence of current asthma in relation to
formaldehyde exposure in occupational settings (exposures above 0.10 mg/m3). The mechanistic
evidence indicates that formaldehyde exposure can induce bronchoconstriction and lead to the
development of hyperresponsive airways,9 particularly with allergen sensitization. These
heightened responses may be due to a combination of potentially progressive changes, including
neurogenic increases in tachykinins and eosinophil recruitment and activation in the lung. The
mechanistic studies also provided consistent evidence that formaldehyde may stimulate a number
9Hyperresponsive airways (or hyperresponsiveness) represents a mechanistic event (supported by robust
evidence) and a potential key feature of respiratory health hazards that is defined to encompass any of a
range of relevant airway features, including hyperreactivity (exaggerated response) and hypersensitivity
(lower dose to elicit response). See Appendix A.5.
This document is a draft for review purposes only and does not constitute Agency policy.
1-78 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
of immunological and neurological processes related to asthmatic responses; however, a molecular
understanding of how formaldehyde exposure favors asthmatic T-helper 2 (Th2) responses has not
been experimentally established.
Overall, based primarily on a moderate level of human evidence supporting an association
from the available epidemiological studies, with corresponding slight evidence for an effect in
animals based on mechanistic studies in animals supporting biological plausibility, the evidence
indicates that inhalation of formaldehyde likely causes an increased risk of prevalent allergic
conditions and prevalent asthma symptoms, as well as decreased control of asthma symptoms,
given the appropriate exposure circumstances. The primary basis for this conclusion involves
studies of occupational settings (>0.1 mg/m3) and population studies where formaldehyde
concentrations measured in schools and homes averaged between 0.03 and <0.1 mg/m3.
Literature Search Strategy
The primary databases used for the literature search were PubMed, Web of Science, and
Toxline, with the last update of the search completed in September 2016 (see Appendix A.5.4 and
A.5.6), and a systematic evidence map updating the literature through 2021 (see Appendix F). The
focus of this review was on studies with a direct measure of formaldehyde exposure in relation to
measures of allergic conditions or asthma, reflecting the question of whether formaldehyde
exposure influences the sensitization response to respiratory allergens. This included the
identification of studies of specific health outcomes and particular exposure scenarios in studies of
exposed humans (Appendix A.5.4), studies of hypersensitivity in animals (Appendix A.5.4 and
A.5.6), and relevant mechanistic data identified and evaluated as part of the overarching review of
mechanistic data relevant to potential respiratory health effects (Appendix A.5.6). For the human
health effect studies, several exposure settings and scenarios were included that encompassed
different exposure durations and time windows. These included controlled human exposure
studies among asthmatics, residential and school settings, as well as occupational studies.
Controlled human exposure studies of pulmonary function change among asthmatic volunteers,
including two studies that assessed whether formaldehyde exposure changed the response to an
allergen challenge, are summarized in this section, but their results are most informative to the
pulmonary function outcome and are included in the integration of evidence in that section (see
Section 1.2.2). Specific types of outcome measures within the category of allergic conditions include
questionnaire-based ascertainment of history of rhinitis, rhinoconjunctivitis, hay fever, pet allergy,
eczema, or dermatitis; physician documentation of a specific diagnosis (e.g., atopic dermatitis); and
allergic sensitization based on skin prick tests. Allergic conditions were grouped by site (nose and
eyes, skin). Eczema is not a contact allergy but can be triggered by reactions to respiratory and
other types of allergens (as well as by other factors). Food allergies were not included in the
literature search. Measures of asthma include questionnaire-based ascertainment of prevalence of
current asthma (e.g., within past 12 months), incidence of asthma, and measures of asthma control
(based on symptom frequency and medication use in the past 2-4 weeks).
This document is a draft for review purposes only and does not constitute Agency policy.
1-79 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
While not a particular focus of this review, the search also encompassed several studies of
lower respiratory infection. Given the frequency and general transiency of upper respiratory
infections such as the common cold in human populations (which may complicate epidemiological
evaluations), as well as their generally benign nature, this endpoint is not discussed in detail in this
section, although they were identified and evaluated in the wider context of potential mechanisms
for respiratory health hazards (see Appendix A.5.6).
One potential mechanism for inducing hypersensitivity is the potential to elicit a
formaldehyde-specific antibody response, specifically IgE. The presence of formaldehyde-specific
IgE in workers occupationally exposed to formaldehyde was described in case reports fVandenplas
etal.. 2004: Kim etal.. 20011. but larger studies in exposed populations or in asthma patients
indicate this is a relatively uncommon occurrence, seen in no or only a few individuals (Hisamitsu
etal.. 2011: Doietal.. 2003: Krakowiaketal.. 1998: Wantke etal.. 1996b: Grammer etal.. 1990:
Thrasher et al.. 1990). Formaldehyde-specific IgE was not included as an outcome for analysis in
this section. However, a broader consideration of antibody responses following formaldehyde
exposure is considered in the mechanistic evaluation of potential respiratory effects (see
Appendix A.5.6).
The bibliographic databases, search terms, and specific strategies used to search them are
provided in Appendix A.5.4 and A.5.6, as are the specific PECO criteria. Literature flow diagrams
summarize the results of the sorting process using these criteria and indicate the number of studies
that were selected for consideration in the assessment through 2016 (see Appendix F for the
identification of newer studies through 2021). The relevant human health effect studies
(i.e., meeting the requirements outlined above), studies of hypersensitivity in animals, and
mechanistic data informative to immune-related conditions and asthma were evaluated to
ascertain the level of confidence in the study results for hazard identification (see Appendix A.5.4
and A.5.6).
Methodological issues considered in evaluation of studies
The evaluation criteria were developed after discussions with two groups of clinical and
epidemiology experts in allergy10 and asthma11 regarding sensitivity, specificity, and interpretation
of various types of outcome measures used in the identified observational epidemiological studies.
These discussions were conducted without regard to the magnitude or direction of results
10Dr. Hasan Arshad, University of Southampton, Southhampton, U.K.; Dr. Peter Gergen, National Institute of
Allergy and Infectious Diseases, Bethesda, Maryland; Dr. Elizabeth Matsui, Johns Hopkins University,
Baltimore, Maryland; Dr. Dan Norback, Uppsala University, Uppsala, Sweden; Dr. Matthew Perzanowski,
Columbia University, New York, New York.
"Asthma: Dr. Lara Akinbami, Centers for Disease Control and Prevention, Atlanta, Georgia; Dr. Peter Gergen,
National Institute of Allergy and Infectious Diseases, Bethesda, Maryland; Dr. Christine Joseph, University of
Michigan, Ann Arbor, Michigan; Dr. Felicia Rabito, Tulane University, New Orleans, Louisiana; Dr. Carl-Gustaf
Bornehag, Karlstad University, Karlstad, Sweden.
This document is a draft for review purposes only and does not constitute Agency policy.
1-80 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
pertaining to formaldehyde or other exposures. Three studies were reclassified from asthma to
lower respiratory symptoms in infants and toddlers; see discussion in Appendix A.5.4.
EPA also evaluated the exposure measurement protocol used in the epidemiological studies,
considering the length of the exposure period, consideration of temperature, relative humidity, and
LOD and percentage 0.1 to >0.5 mg/m3. The remaining were general
population studies of adults and children, with exposure measured in homes or schools or with a
personal monitor. In the general population settings, most exposures were <0.050 mg/m3, with
relatively few results for exposures from >0.05 to approximately 0.1 mg/m3. EPA used 0.05 mg/m3
as a cutpoint to examine results in lower exposure groups compared to higher general population
exposures.
The study evaluation conclusions are indicated with the summaries of study results. Within
each subsection of a table (e.g., sections of studies of children or studies of adults), studies are
further grouped by confidence level (i.e., high, medium, and low categories). Results from low
confidence studies are shaded in gray. The corresponding synthesis of evidence focuses on the
medium and high confidence studies, taking into account differences in populations (i.e., children,
adults) and exposure levels.
One study was difficult to classify fSmedie and Norback. 20011. This is the only study that
examined incidence of allergies or asthma; the prospective design is a considerable strength of the
study. However, the exposure assessment (conducted in classrooms in the baseline year and in
Year 3 of the 4-year follow-up) was limited by a high prevalence of values below the detection limit:
(54% of 1993 samples and 24% of 1997 samples were below 0.005 mg/m3; geometric mean 0.004
and mean 0.008 mg/m3). The analysis was conducted using formaldehyde as a continuous variable,
without discussing the influence of the values below the detection limit Thus, EPA classified this as
a low confidence study because of uncertainties regarding the analysis. However, given that this
was the only study that evaluated incidence of allergies or asthma using prospective study design,
this section also considers the potential impact of this study fSmedie and Norback. 20011 on overall
conclusions if it had been characterized as a medium confidence study.
In this section, where feasible (based on similar type of measures, referent groups, and
analysis), EPA conducted a meta-analysis to calculate a summary effect estimate for related results.
These analyses used random effects models with a restricted maximum likelihood estimator,
weighing the studies based on variance.
This document is a draft for review purposes only and does not constitute Agency policy.
1-81 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Immune-mediated Conditions, Focusing on Allergies and Asthma, in Human Studies
In the following sections, the evidence regarding allergic conditions (symptoms, skin prick
tests) from general population studies is discussed by age category (i.e., children, adults). For
asthma, general population studies of asthma incidence and prevalence and degree of control
among children and adults are discussed by exposure setting (general population, occupational). In
addition, responses among asthmatics to acute exposure are described (controlled human exposure
studies), followed by other respiratory conditions in infants and toddlers, and a discussion of
factors that may increase susceptibility. The studies are summarized in tables for these outcomes
(see Tables 1-12 through 1-21) that are ordered by age group, confidence in study results, and
publication year. The three tables of asthma prevalence (see Tables 1-15 through 1-17) group
studies of populations with exposure to relatively low levels or relatively high levels of
formaldehyde in residential or school settings and occupational groups exposed to higher levels.
Allergic conditions
The high and medium confidence general population studies provided evidence that
formaldehyde exposure is associated with an increased prevalence of rhinitis or rhinoconjunctivitis
(see Figure 1-8A, Table 1-12). These studies were conducted in school children in France fAnnesi-
Maesano etal.. 20121. Romania fNeamtiu etal.. 20191. and Korea fYon etal.. 20191. and in adults in
France (Billionnet etal.. 2011) and Japan (Matsunaga etal.. 2008). The exposure range was similar
in these studies and estimated RRs were comparable for rhinitis endoints ranging from 1.14 to 1.21
for comparisons of the higher exposed to the referent groups. One study of school children in
Malaysia measured very low formaldehyde concentrations in classrooms (mean 4.2 ug/m3, max
18.0 ug/m3), and did not observe an association with rhinitis prevalence fNorback et al.. 20171. The
classification of rhinoconjunctivitis by Annesi-Maesano et al. (2012) was the most sensitive and
specific of the measures, and the narrower confidence intervals in this study reflected the larger
sample size. No other pollutants (e.g., NOx, PM2.5, acetaldehyde, acrolein, ETS) analyzed by this
study were associated with rhinoconjunctivitis. For eczema, only one study was available, with a
two-fold risk seen at exposures of approximately 0.06 mg/m3 (Matsunaga etal.. 20081. Neamtiu et
al. (2019) studied "allergy-like symptoms" in school children occurring in the past week using a
translated ISAAC questionnaire. The definition for allergy-like symptoms included a combination of
symptoms involving the eyes, rhinitis symptoms, and skin conditions. Students exposed to
formaldehyde concentrations in classrooms >0.035 mg/m3 (median 0.045 mg/m3) had a 3-fold
odds of experiencing allergy-like symptoms within the past week compared to students exposed to
<0.035 mg/m3 (OR 3.23, 95% CI 1.31, 8.00). Two studies examined more than two exposure groups
(Annesi-Maesano etal.. 2012: Matsunaga etal.. 20081 and observed the highest relative risk in the
highest exposure group compared to the referent group, with weaker or no associations seen in the
lower exposure categories (see Figure 1-8B). Further, an analysis by categories of rhinitis severity
in children observed a statistically significant increasing trend in risk fYon etal.. 20191. The
This document is a draft for review purposes only and does not constitute Agency policy.
1-82 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
inclusion of the study by Smedje and Norback (20011 as a medium confidence study did not change
the interpretation of the evidence.
A relative risk of 1.4 for formaldehyde exposures above approximately 0.035 mg/m3 and
atopy based on skin prick tests was also seen in a study in children fGarrettetal.. 19991. but not in
the study by Palczynski et al. (1999) (see Table 1-13). Both of these were classified as medium
confidence with respect to the results in children. The exposure range examined in Garrett et al.
(19991 is wider than that in Palczynski et al. (19991. and the exposure measurement protocol (four
1-day samples in different seasons) was an additional strength of the study by Garrett et al. (19991.
This study also reported associations between formaldehyde exposure and both wheal size and the
number of positive tests (from a mean of approximately 1.5 in the lowest to 4.0 in the highest
category of exposure). A limitation of the skin prick test studies was the uncertainty regarding the
congruence between the exposure measure and the exposure during the relevant time window
with respect to development of sensitization. In particular, all of the residences in the study by
Palczynski et al. (19991 had been built 10 years prior to enrollment in the study, and sensitization
may have occurred years before the exposure assessment, possibly when exposure levels were
higher. A similar concern was raised for Garrett et al. (1999), as the authors did not report the age
of the housing stock for participants and 74% of the children had lived in their homes at least
5 years.
Results from the two occupational studies were mixed (see Table 1-14). Both are
considered low confidence based primarily on limitations of the outcome ascertainment used in
these studies.
Because of the limitations noted above with respect to interpretation of skin prick tests,
EPA has higher confidence in the studies of allergy-related conditions. Consistent results were
observed across this set of studies in children and adults comprising diverse populations. The
pattern of exposure-response seen in the studies with sufficient sample size and range of exposure
to examine these patterns suggests that formaldehyde exposure at levels seen in the general
population studies can enhance the immune hypersensitivity response to allergens. The studies of
allergy-related conditions are summarized in Figure 1-8.
This document is a draft for review purposes only and does not constitute Agency policy.
1-83 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
A. Highest Exposure Group per Study
Children
Norback et al. 2017-
(rhinitis)-
IMeamtiu et al. 2019-
(eye, nose and skin symptoms) -
Yon et al. 2019-
(rhinitis) -
Annesi-Maesano et al. 2012-
(rhinoconjunctivitis)-
Adults
Billionnet et al. 2001-
(rhinitis) -
Matsunaga et al. 2008-
(rhinitis)-
(eczema)-
No quantitative results reported;
Reported no association
Formaldehyde Levels (mg/m3)
Total Approximate Referent RR
N Midpoint
462
280
246
6,683
916
998
998
0.004
0.045
0.027
0.044
0.06
0.07
0.07
Per unit mg/m3
<0.035 3.23
Per 0.01 1.21
mg/m3
<0.019 1.19
<0.028 1.14
<0.058 1.22
<0.058 2.25
0.4
4 5
10
B. All Exposure Groups
Children
Norback et al. 2017
(rhinitis)
Neamtiu et al. 2019
(eye, nose and skin symptoms)
Yon et al. 2019
(rhinitis)
Annesi-Maesano et al. 2012
(rhinoconjunctivitis)
Adults
Billionnet et al. 2001
(rhinitis)
Matsunaga et al. 2008
(rhinitis)
(eczema)-
No quantitative results reported;
Reported no association
~
Formaldehyde Levels (mg/m3)
Total Approximate Referent RR
N Midpoint
462
280
246
6,683
916
998
998
0.004
0.045
0.027
0.025
0.044
0.06
0.028
0.045
0.07
0.028
0.046
0.07
Per unit mg/m3
<0.035
Per 0.01
mg/m3
<0.019
<0.019
<0.028
<0.022
<0.022
3.23
1.21
1.11
1.19
1.14
1.16
0.85
1.17
1.03
1.11
2.36
10
Figure 1-8. Relative risk estimates for prevalence of allergy-related conditions
in children and adults in relation to formaldehyde in residential and school
settings.
Results are depicted for rhinitis (diamond), eczema (circle) and symptom combinations (square). Study
details are described in Table 1-12. High and medium confidence studies are included in figure. Open
symbols are for studies in children; closed symbols are for studies in adults. Panel A depicts the results
from the highest exposure group in each study; Panel B depicts the results from all exposure groups in
each study.
This document is a draft for review purposes only and does not constitute Agency policy.
1-84 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-12. History of allergy-related conditions in relation to formaldehyde
exposure, by age group
Study and design3
Results
Nasal and ocular
Dermatologic
Children
Annesi-Maesano et al.
(2012) (France)
Prevalence survey, n = 6,683, ages
9-10 years, participation rate 69%.
Sampling from 108 schools, all
classes of specified grade level per
school.
Exposure: 5-day samples in
classrooms. Median (75th
percentile) 0.027 (0.034) mg/m3
(estimated from Figure 1 in paper).
Outcome: Parent report, sneezing
and runny nose, with itchy eyes,
without a cold, in past 12 months.
Evaluation3:
Rhinoconjunctivitis prevalence 11.8%,
OR (95% CI) (adjusted)
<0.0191 mg/m3 1.0 (referent)
>0.0191-0.0284 1.11 (0.94, 1.37)
>0.0284-~0.055 1.19 (1.03, 1.39)
(Confidence intervals estimated from Figure
3 in paper)
Adjusted for age, gender, passive smoking,
maternal and paternal history of asthma
and allergic diseases.
Not examined
SB IB Cf Oth
Overall
Confidence
High
Yon etal. (2019)
(Seongnam City, Korea)
Prevalence study, n = 427 school
children recruited from 22
randomly selected classrooms at
11 elementary schools; 68.9%
participation rate, ages 10-14
years.
Exposure: Formaldehyde sampling
in each classroom using monitors
with pumps during the 1st and 2nd
half of the school year.
Mean 0.027 ± 0.077 mg/m3; as
high as 0.06 mg/m3 in some
classrooms.
Duration and sampling methods
were not described.
Outcome: rhinitis definition:
presence of characteristic
symptoms and /or signs during the
previous 12 months using ISAAC
questionnaire, Self report. Rhinitis
severity: low, medium, high.
Rhinitis prevalence: 57.6%, n = 246
OR (95% CI) per 1 Mg/m3
1.019 (1.002,1.037) adjusted for age, sex,
environmental tobacco smoke exposure,
and physician-diagnosed allergic rhinitis in
parents.
Rhinitis severity
OR (95% CI) per
n 1 ng/m3
Control 181 Reference
Mild 44 1.019(0.991,1.048)
Moderate/ 202 1.025 (1.007,1.044)
Severe
P trend = 0.014
This document is a draft for review purposes only and does not constitute Agency policy.
1-85 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
Evaluation:
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Letter to the editor providing
minimal details on formaldehyde
distribution and demographic
characteristics.
Neamtiu et al. (2019)
(Romania)
Prevalence survey; n = 139 males
and 141 females, 89.7%
participation rate.
Sampling from five primary schools
in one county, 3 classrooms per
school.
Exposure: 5-day samples in each
classroom.
Median (75th percentile)
0.035 (0.045) mg/m3,
maximum = 0.066 mg/m3.
Outcome: Allergy-like symptoms in
the past week based on ISAAC
questionnaire, as skin conditions
(e.g., rash, itch, eczema), eye
disorders (e.g., red, dry, swollen,
itching, or burning eyes, or
sensation of "sand in the eyes,"
and rhinitis symptoms (e.g., itching
nose, sneezes, and/or stuffy or
blocked nose.
Evaluation3
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Selection of schools was part of a
larger European framework.
Appropriate methods for exposure
assessment and outcome
ascertainment instruments appear
to have been used.
Outcome definition for allergy-like
symptoms using ISAAC
questionnaire included combined
symptoms of rhinitis (nose), eye
and skin conditions.
Allergy-like symptoms (eyes, nose and
skin)
OR (95% CI), above compared to below
median (0.035 mg/m3):
3.23 (1.31, 8.00).
Logistic regression model adjusted for age,
gender, N02, CO, C02, temperature,
relative humidity, ventilation rate, and
tobacco smoke exposure for the past week.
Norback et al. (2017)
(Malaysia)
Prevalence survey, n = 462
randomly selected children
Rhinitis, weekly symptoms during previous
3 months.
Prevalence 18.8%.
This document is a draft for review purposes only and does not constitute Agency policy.
1-86 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
recruited from 8 randomly
selected schools (15 students in
each of 4 randomly selected
classes per school). 96%
participation rate. Mean age
14 years (range 14-16 years), 48%
male.
Exposure: Formaldehyde sampled
continuously over 7 days in each
classroom using diffusion
samplers. Samplers placed 2
meters above floor, methods
described.
Mean concentrations
formaldehyde indoor 4.2 Mg/m3,
max 18.0 ug/m3,100% samples
above the detection limit.
Outside 5.5 ug/m3, max 6.0 Mg/m3,
100% samples above the detection
limit.
Outcome: Rhinitis defined by two
questions combined regarding
nasal catarrh or nasal congestion in
standardized questionnaire. Cases
defined by reporting symptoms
weekly over a 3-month period.
Evaluation3:
No association with formaldehyde in initial
model; quantitative results were not
reported.
Initial stepwise multiple logistic regression
model including indoor exposures (C02,
N02, formaldehyde and VOC), personal
factors (sex, race, current smoking, atopy,
parental asthma/allergy) and home
environment factors (ETS, dampness/mold,
recent indoor painting).
SB IB Cf Oth
Overall
Confidence
Medium
Quantitative results were not
reported. Very low indoor
formaldehyde concentrations.
Isa et al. (2020) (Malaysia)
Prevalence survey; n = 182 males
and 288 females, participation not
reported.
8 randomly selected schools
(4 urban, 4 suburban), randomly
selected students from 4 classes
(Form two, aged 14 years) during
August-November 2018 &
February 2019.
Exposure: One-hour samples in
four classes during class session.
Median (IQR) Urban: 13.2
(9.3) Mg/m3, Suburban: 3.1
(5.2) Mg/m3 (reported as mg/m3
but likely Mg/m3).
Outcome: Allergy information and
symptoms within defined period
Rhinitis in last 12 months 55.5%
OR (95% CI) per 10 units formaldehyde
(reported as mg/m3 but likely Mg/m3).
3.32 (1.69, 6.51)
Adjusted for atopy, sex, doctor's diagnosed
asthma, parental asthma/ allergy and
urban/suburban location.
Association observed for N02
OR (95% CI) per Mg/m3
2.07 (1.10, 3.89)
Skin allergy in last 12 months 14.5%
OR (95% CI) per 10 units formaldehyde
(reported as mg/m3 but likely Mg/m3).
2.41 (0.96, 6.07)
Adjusted for atopy, sex, doctor's diagnosed
asthma, parental asthma/ allergy and
urban/suburban location.
Association observed for N02
OR (95% CI) per Mg/m3
3.68(1.07, 12.69)
This document is a draft for review purposes only and does not constitute Agency policy.
1-87 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
using ECRHS and ISAAC
questionnaires. Allergic symptoms
in last 12 months: rhinitis, skin
allergy.
Evaluation:
SB
IB
Cf
Oth
Overall
Confidence
¦
Low
Uncertainty in exposure
concentrations and distribution
given short sampling duration, very
low concentrations in half the
schools with unclear proportion of
samples less than the LOD, and
analysis using concentration as a
continuous variable. Participation
details not reported. Unknown
impact of potential confounding by
N02 on formaldehyde associations.
Huang et al. (2017)
(Shanghai, China)
Case-control study, n = 409
children, aged 5-10 years, who
were participants in a previous
cross-sectional study (2011-2012)
selected from 88 kindergartens
located in 6 Shanghai districts.
Eligible children lived in homes not
renovated in prior two years and
agreed to home inspection during
March 2013-December 2014.
Exposure: Formaldehyde sampling
in child's bedroom, 24 hours, in
breathing zone (detection range:
0.012-0.08 mg/m3). Average
concentration (pg/m3), 24-hr
21.5 ±13; 6-hr 22.2 ±17.9.
Range 6.0-60.0 |ig/m3, 2 homes
above.
Outcome: History of airway
diseases using translated ISAAC
questionnaire; Current rhinitis: In
the past 12 months, has your child
had a problem with sneezing, or a
runny, or a blocked nose when
he/she did not have a cold or the
flu?
Evaluation3:
Current rhinitis 41.4%
OR (95% CI) per IQR (15.2 |Jg/m3)
0.72 (0.47, 1.10).
Logistic regression adjusted for age, sex,
family history of atopy, family annual
income, household (ETS), early and current
household dampness-related exposures,
early antibiotics exposure, early home
decoration, and the inspection season.
SB IB Cf Oth
Overall
Confidence
Low
This document is a draft for review purposes only and does not constitute Agency policy.
1-88 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
Concern for selection bias
(eligibility based on home
renovation and asthma status),
difference in ventilation methods
by case status suggests
uncontrolled confounding, low
formaldehyde concentrations
Hsu et al. (2012) (Taiwan)
Case-control study, n = 48 allergic
rhinitis cases, 36 eczema cases 42
controls, recruited through
kindergartens and day care
centers, ages 3-9 years at
enrollment. Participation rate
(clinic exam and home measures)
approximately 5% of potential
cases and controls (but differential
at various steps).
Exposure: 2-hour household
sample (probably bedroom;
converted from ppb)
Median (25th, 75th percentile):
Controls 0.017 (0.005,
0.030) mg/m3
Outcome: Initial screening through
parent report of history (ages 2-6)
with confirmation (1-3 years later)
by clinical examination.
Evaluation3:
Allergic rhinitis
Formaldehyde concentrations lower in
cases than in controls:
(n) Median (25th, 75th percentile) mg/m3
Controls (42) 0.017 (0.005, 0.030)
Allergic rhinitis (48) 0.005 (0.005, 0.020)
(p = 0.02)
Mann-Whitney nonparametric test
Eczema
Formaldehyde concentrations lower in
cases than in controls:
(n) Median (25th, 75th percentile) mg/m3
Controls (42) 0.017 (0.005, 0.030)
Eczema (36) 0.006 (0.005, 0.018)
(p = 0.07)
Mann-Whitney nonparametric test
SB
IB
Cf
Oth
Overall
Confidence
Low
1
¦
¦
¦
1
Low and differential (at various
steps) participation rate. Short
exposure sampling period and no
information on protocol. Limited
analysis. Uncertainty regarding
distribution (percentage
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
Location: 44 and 21% near road for
cases and controls, respectively.
Exposure: Household sample
(sampling period not reported, but
closed windows and use of
duplicates).
Geometric mean, 25th, and 75th
percentiles in controls: 0.043
(0.024, 0.115) mg/m3.
Outcome: Atopic dermatitis based
on medical history, skin prick test
and IgE (criteria not provided).
Evaluation3:
SB IB Cf Oth
Overall
Confidence
i i
Low
llll
Selection and recruitment process
not reported; sampling period not
reported and specific criteria for
case definition not reported;
potential confounders not
addressed (age and type of
housing and location differed
between cases and controls, as
measure of socioeconomic status).
Limited analysis.
Smedje and Norback
(2001) (Sweden) Prospective
(incidence) study, children, 1,258
without asthma at baseline, 88
incident cases of pollen allergy and
50 incident cases of pet allergy in
4-year follow-up; 78% participation
in follow-up, mean age 10.3 years
at baseline. School-based sample;
1st, 4th, and 7th grades.
Exposure: Two 4-hour samples in
2-5 classrooms per school;
measured in 1993 (n = 98) and
1995 (n = 101).
mean 0.008 mg/m3, geometric
mean 0.004 mg/m3(min, max)
(<0.005, 0.072) mg/m3,54% of
1993 samples and 24% of 1995
samples below detection limit
(0.005 mg/m3); median among
those above detection
limit = 0.010 mg/m3. Individual
student values based on average of
1993 and 1995 classrooms (<0.005
to 0.042 mg/m3).
Allergies (incidence)
RR (95% CI) per 0.010 mg/m3,
Pollen allergy: 1.3 (0.95,1.7)
Pet allergy: 1.1(0.7,1.7)
Adjusted for sex, age, history of atopy,
smoking.
Not examined
This document is a draft for review purposes only and does not constitute Agency policy.
1-90 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
Outcome: Parent report, hay
fever/pollen allergy or pet dander
allergy.
Evaluation3:
SB IB Cf Oth
ES
Overall
Confidence
Low
Exposure measures in only 2 of the
4 years; uncertainty about
distribution; relatively high
percentage
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nasal and ocular
Dermatologic
pregnancy, median age ~30.
Recruited through obstetric clinics
and public health nurses.
Exposure: 24-hour personal
sample (converted from ppb).
Median 0.030, maximum
0.161 mg/m3.
Outpoints based on 30th, 60th, and
90th percentiles (<0.022,
0.022-0.033, 0.034-0.57, and
>0.058 mg/m3).
Outcome: Self-report, treatment
for allergic rhinitis or atopic
eczema in past 12 months.
Evaluation3:
SB IB a
Oth
Overall
Confidence
HH
Medium
0.033
0.034- 301 0.85 (0.51, 1.40)
0.057
0.058- 100
0.131
(trend p-value)
0.058-0.161 vs.
<0.058
1.17 (0.60,2.28)
1.22
(0.91)
(0.68, 2.20)
Low participation rate but
potential for differential
participation (by formaldehyde
exposure and disease status)
unlikely. Lack of data pertaining to
sensitivity and specificity of the
ascertainment method for these
conditions.
0.034-0.057 301
0.058-0.131 100
(trend p-value)
0.058-0.161 vs.
<0.058
per 12.3 mg/m3
1.11 (0.50,2.42)
2.36 (0.92,6.09)
(0.08)
2.25 (1.01,5.01)
1.16 (0.99, 1.35)
Adjusted for age, gestation, parity, family
history (of asthma, atopic eczema, allergic
rhinitis), smoking status, current passive
smoking at home and work, mold in
kitchen, indoor domestic pets, dust mite
antigen level, family income, education,
and season.
(Midpoint of highest quartile estimated as
0.0.07 mg/m3 based on personal
communication (Matsunaga. 2012))
Adjusted for same factors as allergic rhinitis
analysis.
Additional analyses examined effect
modification by family history of asthma,
atopic eczema, or allergic rhinitis, see
Figure 1-11 in this report.
(Midpoint of highest quartile estimated as
0.0.07 mg/m3 based on personal
communication (Matsunaga. 2012)).
Evaluation of sources of bias or study limitations (see details in Appendices A.5.1 and A.5.4). SB = selection bias;
IB = information bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of
limitation. Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be
likely to be toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be
likely to be away from the null (i.e., spurious or inflated effect estimate).
Table 1-13. Skin prick tests in relation to formaldehyde exposure, by age
group
Study and design
Results
Children
Garrett et al. (1999) (Australia)
Atopy prevalence: 88/145 = 0.61
Prevalence survey, n = 148 (53 asthma cases, 95 controls;
combined for this analysis; some cases and controls from
same household; three excluded for total n = 145), ages 7-14
(mean 10.2) years.
Exposure: 4-day (one per season) measures in home
(bedroom, living room, kitchens, outdoors). 74% of the
children had lived in the house for at least 5 years; 34% for
entire life.
Median (maximum) 0.0158 (0.139) mg/m3.
Outcome: Atopy based on skin prick tests to 12 allergens (cat,
dog, grass mix #7, Bermuda grass, house dust, two dust mite,
five fungi).
Exposure (mg/m3) N Proportion with atopy
<0.020 30 0.33
0.020-0.050 75 0.64
0.050-0.139 40 0.75
(trend p-value) (<0.001)
per 0.020 mg/m3 increase OR 1.42 (0.99, 2.04)
Odds ratio, adjusted for parental asthma history, sex; other
factors examined (passive smoke, pets, indoor N02, fungal
spores, house dust mite allergens). (Similar trend seen based on
bedroom measure: prevalence 0.50, 0.59, 0.74, trend p = 0.06.)
Exposure Number of
(mg/m3) N allergens3 Wheal size3
This document is a draft for review purposes only and does not constitute Agency policy.
1-92 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Evaluation3:
SB IB Cf Oth
ej
Overall
Confidence
Medium
<0.020
0.020-0.050
0.050-0.139
(trend p-value)
30
75
40
1.3
3.4
3.9
(0.004)
0.5
1.0
1.3
(0.002)
Estimated from Figure i (Garrett et al.. 1999)
Uncertainty about effect of recruitment process and about
time window of exposure measurement with respect to skin
prick test results.
Children and adults (stratified)
Palczynski et al. (1999) (Poland)
Prevalence survey, n = 278 adults ages 16-65 years; n = 186
children ages 5-16 years from 120 households with children
(random selection from 10-year-old apartment houses).
Participation rate not reported.
Exposure: 24-hour household sample (area not specified)
Mean (±SD) (minimum, maximum) 0.026 (±0.011) (0.002,
0.067) mg/m3; 2% >0.050.
Outcome: Allergy based on skin prick tests (SPT) to five
allergens (dust, dust mites, feathers, grasses)
Evaluation3
Children:
SB IB Cf Oth
en
Overall
Confidence
Medium
Adults:
SB IB Cf Oth
Overall
Confidence
Low
Uncertainty about time window of exposure measurement for
skin prick test results (greater uncertainty in adults than in
children). Not informative above 0.050 mg/m3 because of
sample size (<5).
Positive Skin
(n) Prick Test (%)
IgE
(>100 kU/L) (%)
Children
<0.025 mg/m3
(101)
34.7
37.6
0.025-0.050
(82)
28.0
32.9
0.051-0.067
(4)
25.0
25.0
Adults
<0.025 mg/m3
(142)
29.6
26.1
0.025-0.050
(131)
28.2
25.6
0.051-0.067
(5)
60.3
40.0
Additional analyses demonstrated effect modification by
environmental tobacco smoke, see Table 1-21 in this report.
Abbreviations: SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or analysis.
Evaluation of sources of bias or study limitations (see details in Appendix 3.5.3.2 and Table C.5.3.2-2). Extent of column
shading reflects degree of limitation. Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates
anticipated impact would be likely to be toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates
anticipated impact would be likely to be away from the null (i.e., spurious or inflated effect estimate).
This document is a draft for review purposes only and does not constitute Agency policy.
1-93 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-14. Allergy symptoms or skin prick tests in relation to formaldehyde
exposure in workers
Study and design
Results
Allergy symptoms
Fransman et al. (2003) (NewZealand)
Prevalence survey. Plywood mill workers, n = 112. Participation rate 66%. Mean age
34.5 years, 71% men, mean duration 4.7 years.
Exposure: Personal samples (15-minute samples) in jobs held by 49 workers: (n),
geometric mean (igeometric standard deviation) (mg/m3).
all (22) 0.080 (3.0)
dryers (14) 0.070 (3.2) (one outlier)
pressing (5) 0.160 (2.7)
other areas 0.030-0.040 mg/m3 (at or near detection limit)
Total inhalable dust (full-shift personal samples): geometric mean 0.7 mg/m3.
Outcome: Self-report, allergy symptoms based on sensitivity to house dust, food,
animals or grasses/plants.
Evaluation3:
Overall
IB
Cf
oth
Confidence
Low
¦
1
Uncertain impact of outcome classification (includes food allergies). Selection out of
the exposed work force of "affecteds" possible in this type of prevalence study. "Low"
exposure group exposed to levels of formaldehyde up to 0.080 mg/m3. Either
limitation would result in reduced (attenuated) effect estimate.
Allergy symptoms prevalence
Low (<0.080 mg/m3, n = 38) 31.6%
High (>0.080 mg/m3; n = 11) 45.5%
OR (95% CI) (>0.080 vs. <0080 mg/m3):
2.4(0.5, 11.8)
Adjusted for age, sex, ethnicity,
smoking. Internal comparison by
exposure category limited to the 49
workers with same job titles as those
with the 22 air sample measurements.
Dust not related to high formaldehyde
exposure. Not clear if these specific
symptoms were or were not related to
other exposures (e.g., endotoxin).
Skin prick tests
Herbert et al. (1994) (Canada)
Prevalence survey. Oriented strand board manufacturing (n = 99). Comparison group
(n = 165) oil field workers, not exposed to gas or vapors. Participation rate 98% in
workers, 82% in comparison group. Mean age ~35 years in both groups.
Exposure: 21 hours continuous area sampling, 2 consecutive days
Saw line, debarking: 0.090-0.160 mg/m3
Postheat, press conveyor, packaging, storage: 0.200-0.290 mg/m3
Preheat conveyor: 0. 330 mg/m3
Total dust: mean 0.27 mg/m3, median aerodynamic equivalent diameter = 2.5 [am.
Outcome: Atopy based on SPT to six allergens (wheat, rye, Alternaria, cat, house dust,
birch; four of these are common allergens in this area).
Evaluation3:
S6 IB Cf Oth
Overall
Confidence
Low
¦
Uncertainty about time window of exposure measurement with respect to skin prick
test results; some uncertainty about referent group.
Atopy prevalence not reported
OR (95% CI) 0.75 (0.40, 1.35)
Dust exposure considered low; not
included in analysis.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.4). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
This document is a draft for review purposes only and does not constitute Agency policy.
1-94 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Asthma
Asthma affects approximately 5-10% of the U.S. population, and results in a significant
individual and societal burden in terms of morbidity, health care costs, and indirect costs [e.g., due
to absences from work fShenolikar etal.. 2011: Bahadori et al.. 20091], The potential for
formaldehyde to induce or exacerbate asthma symptoms has been described in occupational
settings in reports spanning several decades (see for example, Nordman etal.. 1985: Popa etal..
19691. Characterization of this risk on a population level requires more extensive evaluation.
Epidemiological studies have investigated potential associations between formaldehyde and
asthma in children and adults using formaldehyde measurements conducted in occupational,
residential, and school-based settings. The outcomes studied include the incidence of asthma
(i.e., the number of people newly diagnosed with asthma in a period of time), prevalence of current
asthma (typically ascertained through a set of questions pertaining to symptoms or medication use
over a period of time, e.g., past 12 months), and asthma control (typically ascertained through a
larger set of symptoms, medication, and medical care use over a shorter period of time,
e.g., 2-4 weeks). Asthma control pertains to the extent to which symptoms can be reduced or
eliminated with medication. The prevalence of current asthma includes newly diagnosed patients,
as well as previously diagnosed patients who are experiencing the expression (and thus the costs
and burden) of this condition. EPA considered "ever had asthma" to be of limited use in this review,
as the formaldehyde measures available do not reflect cumulative exposures that could be related
to cumulative risk, and thus EPA did not include results using the definition, "ever had asthma."
However, there were a small number of studies where asthma was not defined clearly but study
details appeared to indicate that the definition was not "ever had asthma"; these were included but
the limitation was noted. Altered lung function in people with asthma, examined in acute
controlled exposure studies, is also discussed in this section, although these acute, high exposure
scenarios are of less direct relevance to the question of risks of chronic exposures.
Asthma prevalence and incidence studies
The collection of studies evaluated associations between formaldehyde exposure and
prevalence of current asthma, as determined by symptoms or medication use in the past
12 months. The six medium or high confidence studies in homes or schools with relatively low
exposures (<0.05 mg/m3, most from approximately 0.02 to 0.04 mg/m3) reported relative risks
around 1.0 (see Table 1-15, Figure 1-9A). This set of studies included a variety of designs and
populations; the school-based studies are large (from 1,014 to 6,683 total participants). The case
definition of wheezing during the past year used by Venn et al. (20031 is interpreted to be relevant
to a definition of current asthma as used in this assessment since 88% of the cases also reported
using a reliever inhaler in the past year. The results of Smedje and Norback (2001) are consistent
with these studies, and so the inclusion of this as a medium confidence study would not change the
interpretation of the evidence. A study that assessed a definition of "asthma-like" symptoms among
This document is a draft for review purposes only and does not constitute Agency policy.
1-95 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
school children indicates that asthma symptoms, a less specific outcome compared to "current
asthma" may occur at lower formaldehyde concentrations fNeamtiu etal.. 20191. This study in
Romania observed an OR of 2.7 (95% CI: 1.04, 6.97) with the prevalence of asthma-like symptoms
in the past week comparing children exposed to formaldehyde concentrations above and below the
median (0.035 mg/m3, maximum 0.066 mg/m3).
Six medium confidence general population studies in children or adults where a proportion
of the study sample had exposures of 0.05-0.1 mg/m3 were available (see Table 1-16; Figure 1-9B).
Two of these included both children and adults (Zhai etal.. 2013: Krzvzanowski etal.. 19901. and
each provides evidence of a greater susceptibility in children. Both studies compared effects in
groups exposed to levels approximately 0.08 mg/m3 or above to lower exposed groups; a limitation
of the Krzyzanowski et al. (1990) analysis is the relatively small number in the highest exposure
group (n = 21). The sRR in children for these two studies was 4.5 (95% CI: 0.76, 27). One other
study of children (mean age 10 years) was a hospital-based case-control study that diagnosed
prevalent asthma using the ISAAC questionnaire over 3 or more months, and an FEVi increase of
15% in response to (3-agonist inhalation (Liu etal.. 20181. The authors reported an association with
formaldehyde levels based on a regression analysis using quartiles of formaldehyde concentration
(OR = 2.736, 95% CI: 1.098, 5.516). Exposure levels in the highest quartile ranged from 0.05 to 0.14
mg/m3. Of note, a Canadian intervention study of impacts on symptom exacerbation among
asthmatic children from increasing ventilation rates in homes reported that a 50% reduction in
formaldehyde concentrations in the bedroom was associated with a 14 to 20% decrease in the
annual change in some symptoms or medical care in the intervention group (Laioie etal.. 20141.
Geometric mean concentrations of 0.037 mg/m3 were measured in the intervention group at
baseline. However, other coexposures were reduced by the intervention resulting in uncertainty in
the independent effect of formaldehyde, although the reductions were to a lessor extent and
separate effects of the other factors were not analyzed. Two other medium confidence studies with
exposures above 0.05 mg/m3 were conducted only in adults (Billionnet etal.. 2011: Matsunaga et
al.. 2008). Billionnet et al. (2011) compared the asthma outcome for subjects exposed to exposures
greater than the 75th percentile of 0.028 mg/m3 to those exposed to less than the 75th percentile.
While most of the study population was exposed to lower concentrations, a portion were exposed
to concentrations as high as 0.09 mg/m3, which likely influenced the observed RR of 1.4. EPA has
lower confidence in the results of Matsunaga et al. (2008) because of the lower sensitivity and
specificity of the asthma ascertainment The pattern of results in this exposure range of 0.05-
0.1 mg/m3 was indicative of an elevated risk, as none of the point estimates were below 1.0;
however, the confidence intervals around each of the estimates indicated some variability in the
data (see Figure 1-9).
Epidemiological studies in occupational settings examining the incidence of asthma in a
cohort of individuals after they initially enter a workplace have not been conducted. The available
studies generally did not attempt to examine the timing of symptoms in relation to when the
This document is a draft for review purposes only and does not constitute Agency policy.
1-96 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Toxicological Review of Formaldehyde—Inhalation
subjects are present in the workplace (i.e., over the course of a workday or comparison between
workdays and weekend days) and so would not have the level of detail that would be included in a
clinical workup of occupational asthma; rather, these studies can be thought of as studies of the
prevalence of current asthma among workers exposed to formaldehyde. The occupational
exposure literature included three medium confidence studies of plywood and other layered wood
manufacturing workers in Canada (Herbert etal.. 1994). New Zealand (Fransman et al.. 2003). and
Indonesia (Malaka and Kodama. 1990): each of these studies included between 93 and 112 exposed
workers (see Table 1-17). Exposure levels varied by work area, but generally ranged from 0.10 to
>0.50 mg/m3. A greater than three-fold increased risk of asthma was seen in each of these studies;
the sRR for these three studies was 3.79 (95% CI 1.98, 7.28). One of the wood worker studies
addressed potential confounding by dust exposure by the inclusion of this variable in the analysis
fMalaka and Kodama. 19901. and another study specifically noted that the measured dust levels
were not related to high formaldehyde exposure and that the asthma symptoms were not strongly
related to other exposures including endotoxin measures (Fransman et al.. 2003). The results from
these studies may represent underestimates of risk; two factors contribute to this concern. All of
the studies were prevalence surveys of workers who have remained in a workplace for some time
(e.g., 2 or more years), which could be biased by the loss of affected individuals from the workforce
(e.g., because of the "healthy worker effect" inherent in this type of study design). In addition, in
two of the studies, the comparison group included workers who may have also been exposed to
formaldehyde or other respiratory irritants (Fransman etal.. 2003: Herbert etal.. 1994). Inclusion
of this type of exposure in the comparison group reduces the possibility that the observed
associations were influenced by differential reporting of asthma among the exposed but raises the
possibility that the relative risk estimated against this comparison group underestimates the risk
that would be represented by a comparison with a population that does not have these other
exposures. Another limitation to note is that the sensitivity and specificity of the symptom-based
questionnaire measures may be lower in occupational settings than in general population studies;
EPA did not find validation data specific to these types of wood manufacturing settings. However,
given the strength of the relative risks, the consistency of the associations seen in the three
different workplaces and populations, and the likelihood that the observed associations were
underestimates of the true associations, these studies collectively support a strong association
between formaldehyde concentrations above approximately 0.100 mg/m3 in occupational settings
and increased prevalence of current asthma (see Figure 1-9).
This document is a draft for review purposes only and does not constitute Agency policy.
1-97 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
A. General Population, Low Exposures (< 0.050 mg/m )
Children
Palczynski etal., 1999
Venn et al., 2003
Annesi-Maesano et al., 2012
Mi et al., 2006
Kim et al., 2011
Adults
Palczynski et al., 1999
Matsunaga et al., 2008
B. General Population, High Exposures (> 0.050 mg/m3)
Children
Liu et al., 2018
Zhai et al., 2013
Krzyzanowski et al., 1990
Adults
Zhai et al., 2013
Krzyzanowski et al., 1990
Billionnet et al. 2001
Matsunaga et al., 20081
Not calculated (0% in referent)
Reported as "not significantly related" but
rate of wheeze "somewhat higher" with
higher exposure
Formaldehyde
Levels (mg/m3)
Total
Approximate
N
Midpoint
Referent
RR
187
0.037
<0.025
0.98
190
0.019
<0.016
1.14
0.027
1.08
0.041
1.04
6,683
0.025
<0.019
1.10
0.044
0.90
1,414
0.010
per 0.010 mg/m3
1.30
1,028
0.030
per 0.010 mg/m3
1.04
278
0.037
<0.025
0.72
998
0.028
<0.022
0.80
0.046
0.72
Forma Idehyd
e Levels (mg/m3)
Total
Approximate
N
Midpoint
Referent
RR
360
0.05
per quartile
2.74
82
0.115
<0.08
12
298
0.09
<0.049
2
186
0.115
<0.08
t
613
0.09
<0.049
916
0.046
<0.028
1.43
998
0.07
<0.058
2.65
C. Occupational Settings (> 0.01 mg/m )
Herbert et al., 1994-
Fransman et al. 2003 (all) -
(duration > 6.5 years)-
(> 0.080 mg/m3)-
Malaka and Kodama, 1990-
-A-
0.5 1 2 5 10
Relative Risk
50
Total
N
Formaldehyde Levels (mg/m3)
Approximate
Midpoint Referent
RR
99
0.20
<0.08
12
112
0.08
<0.049
1.5
0.08
<0.08
3.1
0.16
<0.049
4.3
93
0.12
<0.058
2.65
This document is a draft for review purposes only and does not constitute Agency policy.
1-98 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Figure 1-9. Relative risk estimates for prevalence of asthma in children and
adults in relation to formaldehyde by exposure level in general population
and occupational studies.
Study details are described in Tables 1-15 (Panel A), 1-16 (Panel B), and 1-17 (Panel C). High and medium
confidence studies included in figures. Lajoie et al. (2014) was not included in the figure because the study
assessed percent change in current asthma symptoms over 12 months, not relative effect. Levels for most
of the participants in the study groups in Panel A, low exposure, were < 0.05 mg/m3. The exposure value
for Liu et al. (2018) is the 75% percentile concentration, which resulted in classifying the study as high
exposure. Exposure levels in Billionnet et al. (2011) ranged to a maximum of 0.09 mg/m3, which resulted
in classifying the study as high exposure. Effect estimates are RR or OR.
Table 1-15. Prevalence of asthma in children or adults in relation to
residential or school formaldehyde exposure in studies with relatively low
exposures (SO.05 mg/m3)
Study and design3
Results
Studies in children and adults (stratified)
Palczynski et al. (1999) (Poland)
Prevalence survey; n = 278, ages 16-65 years and n = 187, ages 5-15 years
from 120 households with children (random selection, 10-year old
apartments). Participation rate not reported.
Exposure: 24-hour household sample (area not specified).
Mean (±SD) (minimum, maximum) 0.026 (±0.011) (0.002, 0.067) mg/m3
2% >0.050 mg/m3
Outcome: Bronchial asthma diagnosed using American Thoracic Society
criteria.
Evaluation3:
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
¦' 1
Uncertainty regarding asthma definition. Not informative above 0.050 mg/m3
because of sample size (n = 4).
Children results: Asthma prevalence 4.8%
Exposure category (n) prevalence
All children <0.025 mg/m (101) 5.0%
0.025-0.050 (82) 4.9%
0.0501-0.067 (4) 0.0%
Adult results: Asthma prevalence: 5.8%
Exposure category (n) prevalence
All adults <0.025 mg/m3
0.025-0.050
0.0501-0.067
(142) 6.3%
(131) 4.6%
(5) 20.0%
Studies in children
Annesi-Maesano et al. (2012) (France)
Prevalence survey; n = 6,683, ages 9-10 years, participation rate 69%.
Sampling from 108 schools, all classes of specified grade level per school.
Exposure: 5-day samples in classrooms.
Median (75th percentile) 0.027 (0.034) mg/m3 (estimated from Figure 1 in
paper).
Outcome: Asthma based on International Study of Asthma and Allergies in
Childhood questionnaire.
Evaluation3:
SB IB Cf Oth
Overall
Confidence
High
Prevalence 6.9%, OR (95% CI)
<0.0191 mg/m3 1.0 (referent)
>0.0191-0.0284 1.10 (0.85,1.39)
>0.0284-~0.055 0.90 (0.78,1.07)
(Confidence intervals estimated from Figure 4
in paper.)
Adjusted for age, gender, passive smoking,
and paternal or maternal history of asthma
or allergic disease.
Additional analyses examined effect
modification by atopy status, see Figure 1-11
in this report.
Kim et al. (2011) (Korea)
Prevalence of asthma: 6.9%
OR (95% CI), per 0.010 mg/m3:
Asthma, current 1.04 (0.78,1.40).
This document is a draft for review purposes only and does not constitute Agency policy.
1-99 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Prevalence survey; n = 1,028, mean age 10 years, participation rate 96%.
Sampling from 12 schools, 2-3 classes per school.
Exposure: 7-day samples in classrooms (n = 34) and one outdoor area per
school (n = 12) (all samples collected in same season).
Mean (±SD), (minimum, maximum) Indoor 0.028 (±0.0083) 0.016,
0.047 mg/m3.
Outcome: Asthma based on current use of asthma medication or asthma
attack in past 12 months.
Evaluation3:
Adjusted for age, sex, self-reported pet or
pollen allergy, environmental tobacco smoke
at home, other home environment (indoor
dampness, remodeling, changing floor, age of
home).
SB IB Cf Oth
Overall
Confidence
High
Neamtiu et al. (2019) (Romania)
Prevalence survey; n = 139 males and 141 females, 89.7% participation rate
Sampling from five primary schools in one county, 3 classrooms per school.
Exposure: 5-day samples in each classroom.
Median (75th percentile) 0.035 (0.045) mg/m3, maximum = 0.066 mg/m3.
Outcome: Asthma-like symptoms based on International Study of Asthma and
Allergies in Childhood questionnaire, asthma-like symptoms defined as difficult
breathing, dry cough and wheezing in the past week (any symptom).
Evaluation3
Asthma-like symptoms
OR (95% CI), above compared to below
median (0.035 mg/m3):
2.7 (1.04, 6.97)
Logistic regression model adjusted for age,
gender, N02, CO, C02, temperature, relative
humidity, ventilation rate, and tobacco
smoke exposure for the past week.
SB
IB
Cf Oth
Overall
Confidence
Medium
¦
Medium
Appropriate methods for exposure assessment and outcome ascertainment
instruments appear to have been used although outcome definition (asthma-
like symptoms) is not specific for current asthma.
Mi et al. (2006) (Shanghai, China)
Prevalence survey; n = 1,414, ages 12-17 (mean 13) years, percentage with
environmental tobacco smoke not reported, participation rate 99%. Sampling
from 10 schools, 3 7th-grade classes per school.
Exposure: 4-hour samples in 30 classrooms.
Mean (±SD), (minimum, maximum) 0.009 (±0.0089) (0.003, 0.020) mg/m3. No
information on LOD or percentage
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Nested case-control; n = 190 persistent wheeze cases, 214 controls, ages
9-11 years. Participation rate 79% among cases, 59% among controls.
Exposure: 3-day samples in bedroom; median ~0.022 mg/m3;
median in top quartile 0.041 mg/m3.
Outcome: Parent report, wheeze in past year (reported for both of two
periods, 1995-1996 and 1998), validated by medical records for 115 cases and
164 controls.
Evaluation3:
(49) 1.0
(46) 1.14
(51) 1.08
(44) 1.04
(referent)
(0.65, 2.00)
(0.62, 1.86)
(0.59, 1.82)
SB
IB
Cf Oth
Overall
Confidence
Medium
n
¦
!'
<0.016 mg/m3
0.020-0.022
0.022-0.032
0.032-0.123
(trend p = 0.93)
Adjusted for age, sex, socioeconomic status.
Similar results in group with validation of
case status from prescription asthma
medication records.
(Median in top quartile provided in email
from Dr. Venn, March 29, 2012.)
Uncertainty about time window of exposure measure.
Branco et al. (2020) (Portugal)
Prevalence survey: School children, n=648 preschoolers (3-5 years) and n=882
primary school children (6-10 years) randomly recruited from urban and rural
nursery (n=17) and primary schools (n=8), participation rate 39%.
Exposure: Daily exposure based on time-averaged air concentration and
reported time in specific school locations. Continuous monitoring in each room
(24 h to 9 days). Mean formaldehyde concentration (SD) 35.3 (43.1) Mg/m3;
Table in article also stated that these values were the median (IQR).
Outcome: Asthma diagnosis by study physicians based on either reported
symptoms using ISAAC questionnaire or a report of ever having 1 or more
symtoms plus spirometry before and after bronchodilator (ERS/ATS and Global
Initiative for Asthma guidelines).
Evaluation3:
SB
IB
Cf Oth
Overall
Confidence
Low
¦
OR (95% CI) per IQR increase in exposure
0.66 (0.37, 1.21).
OR (95% CI) above compared to below the
median
1.19 (0.60, 2.39).
Logistic regression models adjusted for site
(urban, rural), study phase, sex, age group,
BMI and parental history of asthma. Also
controlled for surrogates of home indoor
exposure including mother's education, living
with smoker. Other covariates for contact
with farm animals during 1st year of life, pets
at home in previous year &/or 1st year of life.
Concern regarding potential for selection bias (low participation and missing
values) and decreased specificity of asthma diagnosis by including very young
children (<5 years), 42% of sample.
Yon et al. (2019) (Seongnam City, Korea)
Prevalence study, n = 427 school children recruited from 22 randomly selected
classrooms at 11 elementary schools; 68.9% participation rate, ages
10-14 years.
Exposure: Formaldehyde sampling in each classroom using monitors with
pumps during the 1st and 2nd half of the school year.
Mean 0.027 ± 0.077 mg/m3; as high as 0.06 mg/m3 in some classrooms.
Duration and sampling methods were not described.
Outcome: current asthma definition: presence of characteristic symptoms and
/or signs during the previous 12 months using ISAAC questionnaire, Self report.
Evaluation:
Current asthma prevalence n = 10
OR (95% CI) per 1 Mg/m3
1.023 (0.96,1.089) adjusted age, sex,
environmental tobacco smoke exposure,
keeping a pet at home, and physician-
diagnosed asthma and allergic dermatitis in
parents.
SB
IB
Cf Oth
Overall
Confidence
Low
¦
Few children with asthma contributed to analyses
Letter to the editor providing minimal details on formaldehyde distribution
and demographic characteristics
This document is a draft for review purposes only and does not constitute Agency policy.
1-101 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Madureira et al. (2016) (Porto, Portugal)
Children, case-control, October 2012—April 2013, random recruitment of 38
residences among asthmatic children and 30 residences among nonasthmatic
children previously identified in a cross-sectional study (Madureira et al.,
2015). n=1099 children (aged 8-10 years, 69% of recruited). Excluded
respondents with a recent renovation or who had moved since responding.
Exposure: Continuous passive sampling in bedroom over 7 days.
Formaldehyde concentrations all above the detection limit.
Outcome: For asthma cases, parents responded yes to both of 2 questions in
ISAAC questionnaire: 1) Has your child ever had asthma diagnosed
by a doctor? and 2) In the past 12 months, has your child had wheezing or
whistling in the chest? Parents of controls responded no to both questions.
Evaluation:
SB
IB
Cf
Oth
Overall
Confidence
Low
m ¦
Small sample size, potential for selection bias, no adjustment for confounding
and some differences noted between cases and controls.
Formaldehyde concentration in bedroom,
mg/m3
Cases Controls
38 30
0.015(0.010) 0.017(0.095)
0.011 0.015
0.007-0.018 0.009-0.022
0.004-0.051 0.005-0.043
N
Mean (SD)
Median (SD)
IQR
Min-max
p value = 0.199
Hsu et al. (2012) (Taiwan)
Case-control study; n = 9 cases, 42 controls, recruited through kindergartens
and day care centers, ages 3-9 years at enrollment. Participation rate (clinic
exam and home measures) approximately 5% of potential cases and controls).
Exposure: 2-hour household sample (probably bedroom; converted from ppb)
Median (25th, 75th percentile): Controls 0.017 (0.005, 0.030) mg/m3.
Outcome: Initial screening through parent report of history (ages 2-6 years)
with confirmation by clinical examination.
Evaluation3:
Formaldehyde concentrations lower in cases
than in controls:
(n) Median (25th, 75th percentile) mg/m3
Controls (42) 0.017 (0.005, 0.030)
Asthma cases (9) 0.005 (0.004, 0.012)
(p = 0.03)
Nonparametric (Mann-Whitney) comparison
of formaldehyde by group.
SB
ie
a
Oth
Overall
Confidence
Low
1
¦
¦
1—
Low and differential (at various steps) participation rate. Short exposure
sampling period and no information on protocol. Limited analysis.
Uncertainty regarding distribution (percentage
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
SB IB Cf Oth
en
Overall
Confidence
Low
Asthma definition includes current asthma and ever asthma. Uncertainty
regarding selection processes (high prevalence of family history of asthma in
cases [86%] and controls [96%]); uncertainty about analysis and distribution.
Hulin et al. (2010) (France)
Case-control; (n = 32 urban cases, 31 urban controls; n = 24 rural cases, 24
rural controls), mean age 12.5 years. Drawn from previous school-based
surveys. Participation rates 22 and 13% in urban cases and controls, 52 and
75% in rural cases and controls, respectively.
Exposure: 7-day sample in living room; median (minimum, maximum)
Total (n = 112) 0.019 (0.004, 0.075) mg/m3
Urban, winter (n = 36) 0.020 (0.007, 0.075) mg/m3
Urban, summer (n = 11) 0.021 (0.004, 0.060) mg/m3
Rural, summer (n = 49) 0.016 (0.005, 0.063) mg/m3
Outcome: Parent report of child's history of asthma, use of asthma
medications, or wheezing in past 12 months.
Evaluation3:
OR (95% CI) for above vs. below median)
Total sample: 1.7 (0.7, 4.4)
urban OR = 0.24 (0.04,1.5)
rural OR = 9.0 (1.0, 98)
(interaction p < 0.05)
(Confidence intervals estimated from figure
in the paper.)
Adjusted for age, sex, family history of
allergy, passive smoke exposure during
childhood, and allergic rhinitis.
Levels of other pollutants that are risk factors
for asthma were higher in urban areas.
SB IB Cf Otti
Overall
Confidence
Low
Small sample size and uncertain interpretation of the stratified analyses (and
unspecified n in analysis of current asthma).
Smedje and Norback (2001) (Sweden).
Prospective (incidence) study. 1,258 without asthma at baseline, 56 incident
cases of asthma in 4-year follow-up (incidence rate 1.1% per year); 78%
participation in follow-up, mean age 10.3 years at baseline. School-based
sample; 1st, 4th, and 7th grades.
Exposure: Two 4-hour samples in 2-5 classrooms per school; measured in
1993 (n = 98) and 1995 (n = 101).
Mean 0.008 mg/m3, geometric mean 0.004 mg/m3, (min, max) (<0.005,
0.072) mg/m3, 54% of 1993 samples and 24% of 1997 samples below detection
limit (0.005 mg/m3); median among those above detection
limit = 0.010 mg/m3. Individual student values based on average of 1993 and
1997 classrooms (<0.005 to 0.042 mg/m3).c
Outcome: Parent report of physician diagnosis of asthma and six lower
respiratory symptom questions; previous validation study (73% sensitivity,
99% specificity).
Analysis: Odds ratio, adjusted for sex, age, history of atopy, smoking.
Evaluation3:
OR (95% CI) per 0.010 mg/m3:
total sample: 1.2 (0.8,1.7)
with history of atopy: 0.6 (0.3,1.3)
no history of atopy: 1.7(1.1,2.6)
(Atopy defined at baseline based on positive
response to questions on childhood eczema,
allergy to pollen, or allergy to pet dander.)
Additional analyses examined effect
modification by atopy status, see Figure 1-11
in this report.
Overall
SB IB
CI
Orh
Confidence
Low
¦
¦
•
Exposure measures in only 2 of the 4 years; uncertainty about distribution;
relatively high percentage
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Alternative Evaluation: Medium (based on strengths of prospective study of
incidence).
Related References: Smedie et al. (1997).
Studies in adults
Norback et al. (1995) (Sweden)
Nested case-control within random population sample; n = 47 cases, n = 41
controls, ages 20-44 (mean 32) years. Participation rate 64 and 57%,
respectively, among selected cases and controls.
Exposure: 2-hour sample measured in bedroom.
Mean (Min, Max) 0.029 (<0.005, 0.110) mg/m3.
Strongly correlated with total volatile organic compounds (correlation
coefficient not shown).
Mean duration in home = 6 years (minimum 0.5, maximum 31).
Outcome: Cases defined by positive response to: asthma attack in past
2 months, nocturnal breathlessness in past 12 months, or current use of
asthma medication. Controls responded "no" to all three questions.
Analysis: Odds ratio, adjusted for age, sex, current smoking, wall-to-wall
carpets, and house dust mites.
Evaluation3:
Uncertainty about exposure (most values 0.05 mg/m3)
Study and design3
Results
Studies of children and adults (stratified)
Zhai et al. (2013) (china)
Household survey with random selection of participants within household;
186 homes
186 adults, 82 children.
Exposure: Samples in three rooms per house (bedroom, living room,
kitchen); sampling time not specified.
64% of the 186 houses, and 24% of the 82 houses with children were
>0.08 mg/m3 ("polluted").
Outcome: Ferris (1978) questionnaire
Evaluation3:
Prevalence by exposure category
Children n (%)
<0.08 mg/m3 62 3.22
0.08-0.15 mg/m3 20 40.0
RR 12.4 (2.9, 53.7) [calculated by EPA]
Adults
<0.08 mg/m3 66 0.0
0.08-0.15 mg/m3 120 1.6
RR not calculated
This document is a draft for review purposes only and does not constitute Agency policy.
1-104 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
SB IB a Oth
EH
Overall
Confidence
Medium
Uncertainty regarding exposure measurement period and validation of case
ascertainment in this population. Although potential confounders were not
considered in asthma-only analysis, given the magnitude of the results, the
formaldehyde association is unlikely to be explained only by confounders.
For adults, small number of positive responses.
Krzyzanowski et al. (1990) (United States, Arizona)
Prevalence survey. Adults (n = 613 ages >15 years, mean 37) and children
(n = 298 ages 5-15 years, mean 9.3) from 202 households (stratified sample
from municipal employees). Participation rate not reported. 67% white.
Exposure: Two 1-week samples (opposite seasons) in kitchen, living area,
and bedroom (converted from ppb)
Household: mean 0.032 mg/m3
<0.049 mg/m3 83.7%
0.049-0.074 10.0%
0.074-0.172 6.3%
Only a few values above 0.111 mg/m3
Outcome: Ferris (1978) questionnaire (physician diagnosed).
Evaluation3:
Children and Adults
Children:
Prevalence
asthma, current (physician diagnosed)
15.8%
(n), asthma prevalence by exposure
category,
(248) 11.7%
(24) 4.2%
(21) 23.8%
SB IB Cf Oth
Overall
Confidence
Medium
For children, relatively small n in higher exposure categories; for adults,
incomplete reporting
Related references: Quackenboss et al. (1989a); Quackenboss
et al. (1989b).
<0.049 mg/m3
0.049-0.074
0.074-0.172
(trend p < 0.03)
Log-linear models, stratified by
environmental tobacco smoke, adjusted for
socioeconomic status, ethnicity.
Highest vs. lowest group: RR (95% CI) 2.0
(0.88, 4.8) (EPA calculation, unadjusted)
Additional analyses demonstrated effect
modification by environmental tobacco
smoke, see Table 1-21 in this report.
Adults:
Prevalence of asthma 12.9%
wheeze without a cold
21.5%
shortness of breath with wheezing
14.0%
Reported as "not significantly related" but
rate of wheeze was "somewhat higher" with
higher exposure.
Studies of children
Liu et al. (2018) (China)
Hospital based case-control study, n = 180 cases, 180 controls, mean age 10
years, sex and age comparable. Participation rate not reported.
Exposure: Two-month samples in living room and bedroom. N02 and PM
also measured.
Household: median (range), 75th pet
Cases 0.0384 (0.012-0.142), 0.057 mg/m3
Control 0.0251 (0.012-0.094), 0.046 mg/m3
Outcome: Asthma diagnosis via ISAAC questionnaire (2 or more incidents of
cough, wheezing, and dyspnea for 3 or more consecutive days). Plus FEVi
increased by >15% after (3-agonist inhalation and persistent asthma was
stable for 3 or more months prior to study.
Evaluation3:
Current asthma
OR (95% CI), formaldehyde by quartile
2.736(1.098, 5.516)
Regression models adjusted for history of
allergy, breastfeeding, ETS and PM2.5
Association of lower magnitude (OR =
also was reported for PM2.5
2.029)
Note: the units for the odds ratio were not
provided, but authors stated that quartiles
of concentration were included in the
model.
This document is a draft for review purposes only and does not constitute Agency policy.
1-105 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
SB IB Cf Oth
Overall
Confidence
Medium
While reporting details were brief, citations were
given and appropriate methods for exposure and outcome ascertainment
appear to have been used and the sampling period for formaldehyde was
adequate. Coexposures to PM and N02 were simultaneously controlled. Lack
of clarity for exposure units in regression results.
Laioie et al. (2014) (Quebec, Canada)
Intervention study October 2008-June 2011, n = 43 intervention group,
n = 40 control group; Asthmatic children with exacerbation requiring medical
care in the past year referred by physicians at tertiary care center, 3-
12 years old, (n=83, 71.5% of those meeting inclusion criteria) in homes with
low ventilation rates (<0.30 ACH). Randomly assigned to intervention to
increase ventilation rates by 0.15 ACH.
Exposure: Passive air sampling for formaldehyde in bedroom, 6-8 days,
during winter and summer seasons; intervention group pre- and post-
intervention, Fall/winter measurements: Pre- geometric mean 0.037 (0.032-
0.043) mg/m3; 30.1% homes > 0.050 mg/m3; post- geometric mean 0.024
(0.021-0.028) mg/m3; 0% homes > 0.050 mg/m3;
Control group, Pre- geometric mean 0.037 (0.031-0.043) mg/m3; 25.5%
homes > 0.050 mg/m3; post- geometric 0.035 (0.030-0.041) mg/m3; 22.9%
homes > 0.050 mg/m3;
Outcome: Symptom prevalence or medical care over last 12 months, ISAAC
questionnaire administered to parents;
Evaluation:
Current asthma
Change from year 1 to year 2 in prevalence
of asthma symptoms and medical care in the
past year associated with a 50% reduction in
formaldehyde concentration. Analyses in
intervention group, n = 43:
Outcome
> 1 episode
Wheezing
Night cough
> 1 emergency
Room visit
% Change (95% CI) p value
-14.8 (-28.6,-0.9) 0.037
-20.4 (-35.7,-5.0) 0.010
-16.0 (-30.5,-1.5) 0.031
SB IB
Cf
Oth
Overall
Confidence
¦
Medium
Analyses used mixed linear models with
repeated measures, adjusted for age and
eczema.
Other outcomes analyzed with no
statistically significant decrease were
disturbed sleep, severe wheezing, > 4
episodes wheezing, effort wheezing, rhinitis,
> 1 hospitalization
Small sample size
Other coexposures that have been associated in literature with asthma
symptoms also declined in intervention group (toluene, ethylbenzene,
styrene, limonene, alpha-pinene, airborne mold spores), although
formaldehyde reduction was greatest.
Tavernier et al. (2006) (United Kingdom)
Case-control study, n = 105 cases, 95 controls (from two primary care
practices, age- and sex-matched), ages 4-16 years, lower socioeconomic
status. Participation rate 50%.
Exposure: 5-day sample in living room and bedroom.
Outcome: Asthma based on validated screening questionnaire (84% positive
predictive value; but included questions on respiratory infection).
Analysis: Odds ratio, conditional logistic regression, adjusted for measured
exposures (e.g., endotoxin, Der p 1, particulate matter) and other risk
factors.
Evaluation3:
OR (95% CI), by exposure tertile (exposure
levels not reported; median in Gee et al.
(2005) reported as 0.037 and 0.049 mg/m3
in living room and bedroom, respectively)
Lowest
Middle
3.40)
Highest
2.52)
Living room
1.0 (referent)
Bedroom
1.0 (referent)
0.82 (0.33, 2.05) 1.26 (0.47,
1.22(0.49,3.07) 0.99(0.39,
SB IB Cf
Oth
Overall
Confidence
| |
Low
Uncertainty regarding selection process and loss of almost half of the cases.
Outcome classification includes questions that are not specific to asthma.
This document is a draft for review purposes only and does not constitute Agency policy.
1-106 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Uncertainty as to exposure range, particularly upper tertile (no response
from email to corresponding author).
Related Reference: Gee et al. (2005)
Garrett et al. (1999) (Australia)
Case-control study. 53 cases (physician diagnosis), 88 controls (no asthma
diagnosis) from 80 households (some cases and controls from same
household), ages 7-14 (mean 10.2) years.
Exposure: 4-day (1 per season) measures in home (bedroom, living room,
kitchen), and outdoors.
Median (maximum) Indoor 0.0158 (0.139) mg/m3
Outcome: Parent report, doctor-diagnosed asthma, and respiratory
symptom questionnaire.
Evaluation3:
Overall
SR
ip
ct
Oth
Confidence
L
¦
Low
Uncertainty about asthma definition (current asthma or ever asthma?).
Uncertainty about effect of recruitment process and ability to fully address
household correlation of cases and controls; could result in attenuated
effect estimate. Incomplete reporting of results (adjusted results reported
as "not statistically significant").
Incomplete reporting of results
(n), proportion with asthma (overall
proportion 53/148 = 0.36):
<0.020 mg/m3 (31) 0.16
0.020-0.050 (76) 0.39
0.050-0.139 (41) 0.44
(trend = 0.02)
Adjusted for parental asthma history, sex.
Adjusted results reported as "not
statistically significant" (numeric results not
reported).
This document is a draft for review purposes only and does not constitute Agency policy.
1-107 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Studies of adults
Billionnet et al. (2011) (France)
Prevalence survey, n = 905 adults from 490 dwellings (drawn from nationally
representative sample; 13.6% participation rate), median age 44
(15-89) years; 48% men.
Exposure: One-week sample in bedroom
Median, 75th percentile (minimum, maximum) 0.0194, 0.028 (0.0013,
0.0863) mg/m3
Outcome: Asthma based on self-report, asthma attack, woken by shortness
of breath, or using asthma medication in past 12 months
Evaluation3:
SB
IB a Oth
Overall
Confidence
Medium
h
Low participation rate but potential for differential participation (by
formaldehyde exposure and disease status) unlikely.
Prevalence of asthma: 8.6%
OR (95% CI), adjusted for multiple risk
factors, above vs. below 75th percentile
(0.028-0.0863 vs. <0.028 mg/m3):
1.43 (0.8, 2.4)
(Confidence intervals estimated from graph)
Adjusted for age, gender, smoking status,
relative humidity, mold, pets, outdoor
sources of pollution within 500-meter
radius, highest education level in household,
time of data collection.
Matsunaga et al. (2008) (Japan)
Prevalence survey. Adults, n = 998 women, mean 17th week of pregnancy,
median age ~30 years. Recruited through obstetric clinics and public health
nurses. Osaka prefecture, Japan. Participation rate 17% of pregnant women
in the area.
Exposure: 24-hour personal sample (converted from ppb)
Median 0.030, maximum 0.161 mg/m3
Outpoints based on 30th, 60th, and 90th percentiles (<0.022, 0.022-0.033,
0.034-0.57, and >0.058 mg/m3)
Outcome: Self-report, treatment for asthma in past 12 months
Evaluation3:
Asthma (2.1% prevalence)
mg/m3
n
OR
(95% CI)
<0.022
298
1.0
(referent)
0.022-0.033
299
0.80
(0.23, 2.84)
0.034-0.057
301
0.72
(0.19, 2.77)
0.058-0.161
100
2.15
(0.41, 11.3)
(trend p-value)
(0.47)
0.058 to 0.161 vs.
2.65
(0.63, 11.1)
<0.058
SB IB Cf Oth
en
Overs!!
Confidence
Medium
Low participation rate but potential for differential participation (by
formaldehyde exposure and disease status) unlikely. Potential low
sensitivity of outcome measure; uncertainty regarding specificity but COPD
unlikely to be common in this population.
Adjusted for age, gestation, parity, family
history (asthma, atopic eczema, allergic
rhinitis), smoking, passive smoking, mold in
kitchen, indoor domestic pets, dust mite
antigen level, family income, education,
season of data collection.
(Midpoint of highest quartile estimated as
0.07 mg/m3 based on personal
communication (Matsunaga. 2012)
Studies of children and adults (combined analysis)
YeattS et al. (2012) (United Arab Emirates)
Prevalence survey; n = 1,590 (1,007 ages 19-50 years, 583 ages 6-18 years
from 628 nationally representative sample of household (75% household
participation).
Outcome: Asthma, wheeze symptoms based on several standardized
questionnaires.
Analysis: Odds ratio, adjusted for sex, urban/rural area, age group,
household tobacco smoke; children and adults combined in analysis.
Exposure: 7-day sample (living room)
71% 1 per week
Wheezing in
past
12 months
Wheezing in
past 4 weeks
Difficulty
breathing or
chest tightness
in past
12 months
Difficulty
breathing or
chest tightness
Prevalence
(%)
9.2
6.1
12.0
7.0
OR
(95% CI)
0.64
(0.71, 2.42)
3.5
(0.81,14.9)
1.43
(0.83, 2.46)
6.5
(1.9, 22.3)
This document is a draft for review purposes only and does not constitute Agency policy.
1-108 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Evaluation3:
SB 12
Cf
Oth
Overall
Confidence
Low
T
Difficult to disentangle possible effects of sulfur dioxide from those of
formaldehyde (similar effect sizes; moderate-strong correlation; could result
in inflated effect estimate. Does not separate analysis of children and
adults; only 29% above LOD—analyzed as above vs. below LOD
once or more
times a month
Similar results seen with sulfur dioxide.
Formaldehyde levels (mg/m3):
Geometric
75th
mean
percentile
Cases
0.054
0.108
Controls
0.043
0.115
p-value not reported (>0.05)
Choi et al. (2009) (Korea)
Case-control study, n = 36 allergic asthma cases, 28 controls, recruited
through university outpatient clinic; recruitment procedures not described.
Mean age cases 15.4 years (SD = 3.4; controls 16.2 years (SD = 4.1). Housing
age and type: cases 58% <3 years old and 72% apartments; controls 29%
<3 years old and 50% apartments. Location: 44 and 21% near road for cases
and controls, respectively.
Exposure: Household sample (sampling period and area not reported, but
closed windows and use of duplicates).
Geometric mean, 25th, and 75th percentiles in controls: 0.043 (0.024,
0.115) mg/m3
Outcome: "Allergic asthma" based on medical history, skin prick test, and IgE
(criteria not provided).
Evaluation3:
SB IB Cf Oth
Overall
Confidence
1 1 1 [
Low
Selection and recruitment process not reported; sampling period not
reported and specific criteria for case definition not reported; potential
confounders (age and type of housing and location differed between cases
and controls, as measure of socioeconomic status) not addressed. Limited
analysis.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.4). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
Table 1-17. Prevalence of asthma in relation to occupational formaldehyde
exposure
Study and design3
Results
Fransman et al. (2003) (NewZealand)
Prevalence survey. Plywood mill workers, n = 112. Participation rate 66%. Mean
age 34.5 years, 71% men, mean duration 4.7 years. Internal comparison by
exposure level and external comparison group (n = 415) from general population
(random sample) surveys in the study area.
Exposure: Personal samples (15-minute samples) in jobs held by 49 workers: (n),
geometric mean (igeometric standard deviation) (mg/m3)
Prevalence of asthma in exposed workers,
external comparison group 20.5%, 12.5%
(n) OR (95% CI):
All workers (112) 1.5 (0.9, 2.8)
By duration:
<2 years (34) 0.5 (0.2,1.7)
2-6.5 years (39) 1.0(0.3,2.7)
>6.5 years (39) 3.1(1.3,7.2)
This document is a draft for review purposes only and does not constitute Agency policy.
1-109 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
all (22)0.080(3.0)
dryers (14) 0.070 (3.2) (one outlier)
pressing (5) 0.160 (2.7)
other areas 0.030-0.040 mg/m3 (at or near detection limit)
Total inhalable dust (full-shift personal samples): geometric mean 0.7 mg/m3.
Dust levels highest among composers; formaldehyde levels in this group
were 0.080 mg/m3) (11) 4.3 (0.7, 27.7)
Weaker association with terpenes (OR 2.0
for high vs. low exposure); no association
with other exposures (e.g., dust,
endotoxin) examined in this study.
Adjusted for age, sex, ethnicity, smoking.
Internal comparison by exposure category
based on job title (limited to workers with
same job titles as those with the 22 air
sample measurements).
Selection out of the exposed work force of "affecteds" possible in this type of
prevalence study. "Low" exposure group exposed to levels of formaldehyde up
to 0.080 mg/m3. Either limitation would result in reduced (attenuated) effect
estimate.
Herbert et al. (1994) (Canada)
Prevalence survey. Oriented strand board manufacturing (n = 99). Comparison
group (n = 165) oil field workers, not exposed to gas or vapors. Participation rate
98% in workers, 82% in comparison group. Mean age ~35 years.
Exposure: 21 hours continuous area sampling, two consecutive days
Saw line, debarking: 0.090-0.160 mg/m3
Postheat, press conveyor, packaging, storage 0.200-0.290 mg/m3
Preheat conveyor 0. 330 mg/m3
Total dust: mean 0.27 mg/m3, median aerodynamic equivalent diameter = 2.5 [am
Outcome: International Union Against Tuberculosis and Lung Disease (1986)
questionnaire (symptoms past 12 months).
Evaluation3:
SB
IB a Oth
Overall
Confidence
Medium
h
Prevalence in exposed workers,
comparison group
Asthma 13.3%, 3.0%
Wheeze attacks 25.3%, 9.7%
Woken by shortness of breath
8.1%, 1.2%
OR (95% CI)
Asthma 5.48(1.85,16.2)
Wheeze attacks 3.34(1.66,6.73)
Woken by shortness of breath
6.78(1.40, 32.7)
Adjusted for age, smoking. Dust exposure
considered low, not included in analysis.
Selection out of the exposed work force of "affecteds" possible in this type of
prevalence study, and some uncertainty about referent group.
Malaka and Kodama (1990) (Indonesia)
Prevalence survey. Plywood workers, n = 93 exposed (93% participation rate), 93
unexposed from same plant, matched by age, ethnicity, smoking history (all men).
Mean age ~27 years, mean duration 6 years.
Exposure: Personal and area samples (duration not reported)
Mean by area (converted from ppm)
Exposed—Plywood: 0.78 mg/m3; Particle board: 2.9; Block board: 0.62 mg/m3
Other ("unexposed"): <0.086 mg/m3
Outcome: Ferris (1978) questionnaire. Asthma based on "ever had attack of
wheezing that made you feel short of breath?" or ever diagnosed with asthma
and experienced currently; occupational asthma not defined.
Evaluation3:
Prevalence in exposed workers,
comparison group
Occupational asthma 14%, 8%
Asthma 30%, 8%
OR (95% CI):
Occupational asthma
2.84 (not reported) (p = 0.02)
Asthma
6.31 (not reported) (p < 0.01)
Adjusted for age, smoking, dust
This document is a draft for review purposes only and does not constitute Agency policy.
1-110 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
SB IS
Cf
Oth
Overall
Confidence
1 1
Medium
4-
i i
Selection out of the exposed work force of "affecteds" possible in this type of
prevalence study. "Unexposed" exposure group exposed to levels of
formaldehyde up to 0.086 mg/m3. Either limitation would result in reduced
(attenuated) effect estimate, "occupational asthma" not defined, and lack of
clarity in asthma definition pertaining to current prevalence.
Neghab et al. (2011) (Iran)
Prevalence survey, melamine-formaldehyde resin plant, n = 70 exposed, 24
unexposed (office workers from same plant, no present or past exposure to
formaldehyde or other respiratory irritant chemicals; all men). Similar
demographics, smoking history. Participation rate 100%. Duration >2 years.
Exposure: Area samples (40 minutes) in seven workshops and one area sample in
office area (converted from ppm)
Exposed (mean ±SD) 0.96 (±0.49) mg/m3; unexposed nondetectable
Outcome: Ferris (1978) questionnaire, wheezing symptoms (period not
specified).
Evaluation3:
Prevalence in exposed workers,
comparison group:
Wheezing symptoms 48.6%, 8.3%;
OR (95% CI not reported) OR 10.4
(p = 0.001)
Overall
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Toxicological Review of Formaldehyde—Inhalation
Asthma control studies
The previous discussion focused on the association between formaldehyde and prevalence
of current asthma (i.e., symptoms or use of medications in the past 12 months). A different
question concerns the association between formaldehyde and asthma control among people with
asthma. This population could represent a group with greater susceptibility or vulnerability than
the general population. EPA identified two studies that examined symptom frequency and
medication use in the past 4 weeks (see Table 1-18). In the United Kingdom, Venn et al. (2003)
examined symptoms recorded in daily diaries over the course of 1 month in relation to
formaldehyde levels measured in the child's home (3-day samples from bedrooms). No association
was seen with the prevalence of wheezing during the past year in the case-control analysis (as
discussed in the previous section), but among the 193 cases, a two- to three-fold increased risk of
frequent symptoms (defined as symptoms recorded on >10 consecutive days) was seen in the
highest quartile of exposure (>0.032 mg/m3) compared with <0.016 mg/m3, with some evidence of
an increased risk at even lower exposures (see Figure 1-10; p-value for trend = 0.05). For nighttime
symptoms, which may be most relevant with respect to measurements taken in the bedroom, the
relative risk estimate was 3.33 (95% CI 1.23, 9.02; p-value for trend = 0.02). The case definition of
wheezing during the past year is interpreted as relevant to the definition of current asthma as used
in this assessment, since 88% of the cases also reported using a reliever inhaler in the past year.
These results were not impacted by inclusion of measures of room dampness in the models and
were stronger when limited to patients with atopy (based on positive skin prick test results). In a
smaller study of 37 low-income children in Boston, Dannemiller et al. (2013) observed higher
formaldehyde levels in homes of children with poor asthma control compared to those with better
asthma control (geometric mean 0.066 and 0.042 mg/m3, p = 0.078; see Table 1-18).
This document is a draft for review purposes only and does not constitute Agency policy.
1-112 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
10
0,1
Prevalence of
Persistent Wheeze
I reqiiency of Symptoms
Among Children with Persistent Wheeze:
Daytime Symptoms • Nighttime Symptoms
I ,,4-H
a:
II"
0.021 0.027 0.040
0.021 0.027 0.040
0.021 0.027 0.040
Formaldehyde Exposure (mg/m1)
Figure 1-10. Relative risk of persistent wheeze and of increased frequency of
symptoms among children with wheeze in relation to residential
formaldehyde exposure.
Effect modification by disease status: comparison of formaldehyde associations with prevalence of
current asthma (persistent wheeze) and with increased frequency of symptoms only among cases. Data
from Venn et al. (2003): study details in Table 1-18.
Table 1-18. Exacerbation of asthma symptoms in relation to residential
formaldehyde exposure
Study and design3
Results
Venn et al. (2003) (United Kingdom)
Symptom control among persistent wheeze cases
(symptoms during past year) (n = 193), ages 9-11
years. Participation rate 79%.
Exposure: 3-day samples in bedroom during home
visit.
Median ~0.022 mg/m3
Median in top quartile 0.039 mg/m3
(Maximum and median in top quartile provided in
email from Dr. Venn to Glinda Cooper, March 29,
2012.)
Outcome: 1-month daily diaries recording
symptoms: daytime and nighttime wheezing, chest
tightness, breathlessness, and cough, each
measured on 0-to-5 scale. "Frequent" symptoms
defined as recorded on >10 days.
(n cases, percentage with frequent symptoms), OR (95% CI), adjusted
for age, sex, socioeconomic status (Carstairs deprivation index):
Frequent nighttime symptoms
<0.016 mg/m3 (39,41%) 1.0 (referent)
0.020-0.022 (35, 49%) 1.40 (0.54, 3.62)
0.022-0.032 (36, 53%) 1.61 (0.62, 4.19)
0.032-0.083 (33, 67%) 3.33 (1.23, 9.01)
(trend p = 0.02)
OR per quartile increase:
full sample 1.45 (1.06,1.98)
limited to atopic cases 2.06 (1.37,3.09)
Frequent daytime symptoms
<0.016 mg/m3 (37,62%) 1.0 (referent)
0.020-0.022 (34, 47%) 0.47 (0.17, 1.25)
0.022-0.032 (37,73%) 2.00 (0.71,5.65)
0.032-0.083 (32,73%) 2.08 (0.71,6.11)
This document is a draft for review purposes only and does not constitute Agency policy.
1-113 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Analysis: Odds ratio, adjusted for age, sex, and
Carstairs deprivation index
Evaluation:
SB IB Cf Oth
Overall
Confidence
High
(trend p = 0.05)
OR per quartile increase:
full sample 1.40 (1.00,1.94)
limited to atopic cases 1.68 (1.10, 2.57)
Additional adjustment for dampness or other exposures including visible
mold, total VOCs, or N02, did not affect formaldehyde results.
Similar results in group with validation of case status from prescription
asthma medication records.
(Median in top quartile provided in email from Dr. Venn, March 29,
2012.)
Dannemiller et al. (2013) (United states)
Symptom control among 37 asthma cases, mean
age 10.5 years. Participation rate 79% (37 out of
47)
Exposure: 30-minute pumped sample in kitchen
(converted from ppb)
Median 0.044 mg/m3
Range 0.006-0.162 mg/m3
31% >0.060 mg/m3
Outcome: Five-question survey about symptom
control in past 4 weeks at same time as
environmental sampling.
Analysis: Examined season, temperature, and
relative humidity
Evaluation3:
Asthma Control Question
Geometric mean formaldehyde (mg/m3)
Frequency A/(%)with Most
during past most severe severe All other
4 weeks... rating group groups p-value
Asthma interfered 5 (14%)
with activities
Shortness of
SB IB Cf Oth
en
Overall
Confidence
Medium
breath
Nighttime
symptoms
Used rescue
inhaler or
nebulizer
medication
Asthma control
rating
Score <12 (very
poor control)
3 (8%)
4 (11%)
4 (11%)
3 (8%)
6 (16%)
0.070
0.079
0.065
0.055
0.074
0.066
0.042 0.066
0.043 0.086
0.043 0.184
0.044 0.409
0.043 0.128
0.042 0.078
Recruitment is not from a well-defined population.
Limited exposure measurement period (but quality
control details provided).
Related reference: Sandel et al. (2014)
Similar results adjusted for season.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.4). SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
1 Acute exposure—controlled chamber studies—people with asthma
2 Most of the acute formaldehyde exposure studies among adults with asthma provide little
3 or no evidence of an immediate effect on pulmonary function in response to formaldehyde
4 inhalation (see Table 1-19); however, no controlled exposure studies have been conducted in
5 children with asthma. The exposure duration in these studies ranges from 10 minutes to 3 hours,
6 and so does not represent a chronic exposure scenario. The studies are fairly small (ranging from 7
7 to 19 participants) and use various measures of pulmonary function (e.g., FEVi, FVC) and airway
8 reactivity. Only two of these studies included an assessment of the response to an allergen
9 challenge: dust mite in Casset et al. (2006) and grass pollen in Ezratty et al. (2007). One of these
This document is a draft for review purposes only and does not constitute Agency policy.
1-114 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
Toxicological Review of Formaldehyde—Inhalation
studies demonstrated a reduction in the average dose of mite allergen required for a 20% decrease
in FEVi from baseline (PD20 FEVi) after a 30-minute exposure via mouth breathing only to 92.2
|ig/m3 of formaldehyde compared to ambient air controls (32 |J.g/m3 formaldehyde) [54.7 ng versus
73.2 ng, respectively; fCasset et al.. 20061], Formaldehyde exposure also increased the late-phase
response, expressed as the maximum fall in FEVi from baseline observed during the 6-hour follow-
up, by 15% in FEVi in the exposed individuals compared to an 11% reduction among controls.
However, these effects were not observed in the study by Ezratty et al. (20071. One difference in
these studies is that the Casset et al. (20061 protocol used a nose clip, thus resulting in inhalation
solely by mouth. In addition, for all of these studies, the severity of asthma among the volunteers in
these experiments is not known; thus, the results may not be generalizable to all people with
asthma.
Table 1-19. Controlled acute exposure chamber studies of pulmonary function
with formaldehyde exposure among people with asthma
Results
Study and design
Exposure
measures
Pulmonary function
Bronchial
challenge—airway
reactivity
Studies with allergen challenge
Ezrattv et al. (2007)
n = 12, ages 18-44, nonsmoking,
positive history of pollen allergy.
Design: Random assignment to order
of exposure (2 weeks apart); double
blinded. Testing pre-and every hour
up to 8 hours postexposure. Grass
pollen (5 allergens) challenge (protocol
described).
Evaluation: High confidence
Randomized, double blinded, detailed
data presentation
60 minutes, 0
and
0.500 mg/m3
No difference in FVC or FEVi before
or immediately after (data not
shown)
Early phase response—PD15
FEVi grass allergen: compared
with placebo, higher in five
and unchanged in seven after
exposure
Median (range) index of
reactivity:
Placebo 0.25 (0.10-2.0)
Exposed 0.80 (0.15-2.0)
(p = 0.06)
Late-phase response (8 hours
postexposure and allergen
challenge)
PD15 FEVi
Placebo 0.17(0.03-4.0)
Exposed 0.23
(0.01-3.6) (p = 0.42)
This document is a draft for review purposes only and does not constitute Agency policy.
1-115 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Exposure
measures
Results
Pulmonary function
Bronchial
challenge—airway
reactivity
Casset et al. (2006)
n = 19, ages 19-35 years, nonsmoking,
positive IgE to dust mites.
Design: Random assignment to order
of exposure (3 weeks apart); double
blinded. Mean formaldehyde
exposure at home 0.037 ±
0.004 mg/m3 (24-hour sample).
Testing pre- and every hour up to
6 hours postexposure. House dust
mite challenge (Der p 111.08 Mg/mL,
11.12 |am) (protocol described).
Evaluation: High confidence
Randomized, double blinded, detailed
data presentation; applies to mouth
breathing.
30 minutes,
0.032
(background)
and
0.092 mg/m3
Nose clip
(breathing by
mouth)
No difference in at-pretreatment or
early-posttreatment assessment;
Late-phase response-
Mean ± SE reduction FEVi:
Placebo 11 ± 1.6
Exposed 15 ± 1.6 (p = 0.046)
Early phase response—PD2o
FEVi Der pi
Mean ± SE; median (ng):
Placebo 73.2 ± 17.3; 39.7
Exposed 54.7 ± 12.6; 28.1
(p = 0.05)
Studies without allergen challenge
Harving et al. (1990)
n = 15, ages 15-36, nonsmoking.
Design: Random assignment to
exposure order (one per week);
double blinded. Testing pre- and near
end of exposure period.
Evaluation: High confidence
Randomized, double blinded, detailed
analysis.
Related Reference: Harving et al.
(1986)
90 minutes,
filtered air
(8), 0.120
and
0.850 mg/m3
No difference in: FEVi Raw SGaw
0.008 mg/m3 100.9 2.21 10.67
0.12 mg/m3 99.4 2.23 10.63
0.85 mg/m3 105.0 2.29 11.17
No difference in challenge
test:
PC20PEF
0.008 mg/m3 0.29
0.12 mg/m3 0.36
0.85 mg/m3 0.26
Green et al. (1987)
n = 16, ages 19-35 years, nonsmoking.
Design: Two 15-minute exercise
segments in 60-minute exposure
period. Random assignment to order
of exposure; single blinded. Testing
pre- and during exposure period, ~15
minute intervals.
Evaluation: Medium confidence
Randomized, single blinded
60 minute,
clean air and
3,000 ppb
[0,
3.69 mg/m3]
No difference in FVC, FEVi,SGaw, or
other lung function measures
At 55 minutes FVC FEVi SGaw
Control 4.62 3.54 0.114
3 ppm 4.56 3.46 0.111
No difference in challenge
test:
PDssSGaw
Control 3.69
3 ppm 3.86
Sauder et al. (1987)
n = 9, ages 29-40, nonsmoking.
Design: Clean air followed by
formaldehyde (1 week apart); blinding
of participant not specified. Testing
during and at end of exposure.
Evaluation: Low confidence
Not randomized, blinding not specified
3 hours,
clean air and
3,000 ppb
[0,
3.69 mg/m3]
No difference in FVC, FEVi,SGaw, or
other lung function measures.
At 180 minutes FVC FEVi SGaw
Control 4.11 3.02 0.101
3 ppm 4.16 3.07 0.106
No difference in challenge
test:
PD35 SGaw
Control 0.93
3 ppm 0.96
This document is a draft for review purposes only and does not constitute Agency policy.
1-116 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Exposure
measures
Results
Pulmonary function
Bronchial
challenge—airway
reactivity
Witek et al. (1987); Witek et
al. (1986)b
n = 15, ages 18-35 years, nonsmoking
Design: Two protocols (at rest and
during exercise). Random assignment
to order of exposure; double blinded.
Testing during and at 10 and 30
minutes postexposure; PEFR assessed
from 1 to 24 hours postexposure.
Evaluation: High confidence
Randomized, double blinded;
nonparametric analysis could be
preferred but individual data provided
40 minutes,
0 and 2,000
ppb [0,
2.46 mg/m3]
Few difference in FVC, FEVi, Raw, or
other lung function measures
At 30 min postexposure, resting
protocol
FVC FEVi Raw
Control 0.82 -0.31 -6.64
2 ppm -2.78 0.60 -3.05
Similar patterns in exercise protocol.
No decline in PEFR over 24 hours in
either group.
PD2o FEVi mean ± SD; median
Pre-exposure:
24.0 ± 15.7; 27.4
Postexposure:
13.6 ±20.5; 3.1
(p = 0.12)
Krakowiak et al. (1998)
n = 10, ages 23-52 years, some
smokers, with occupational
formaldehyde exposure
Design: Single blinded. Testing
2 hours pre- and up to 24 hours after
exposure.
Evaluation: Low confidence
Not randomized, single blinding, SE or
SD not reported
2 hours,
0.500 mg/m3
No difference in FEVi or PEF (mean
values shown on graph; no indication
of variability in measures)
No difference in challenge
test (PD2o FEVi) (mean values
shown on graph; no indication
of variability in measures)
Sheppard et al. (1984)
n = 7, ages 18-37, nonsmoking
Design: Two protocols (at rest and
during exercise). >1 day apart;
blinding of participant not specified.
Testing before and 2 minutes after
exposure.
Evaluation: Low confidence
Not randomized, blinding not specified
10 minutes,
0,1,000, and
3,000 ppb
[0, 1.23,
3.69 mg/m3]
formalin
No difference between pre- and post
SGawc in either protocol:
Resting Exercise
Control -1.0 1.8
1 ppm 0.2 2.2
3 ppm NC 2.9
NC= not conducted
Not assessed
Abbreviations: Double blinded = investigator and participants unaware of which exposure; single blinded = participants were
unaware of exposure. Late phase: between 4 and 6 hours after end of house dust mite bronchial challenge. PDX = dose
required to induce an x% reduction in the specified pulmonary function measure (i.e., PD15 FEVi = dose required to induce a
15% reduction in FEVi); Raw = airway resistance; SGaw = specific airway conductance (corrected for lung volume); PEFR = peak
expiratory flow rate.
bWitek et al. (1987) includes the same subjects as the Witek et al. (1986) paper, but with additional results presented in 1987.
cPostminus preexposure SGaw (liters x cm H20/liter); negative value indicates lower SGaw postexposure.
1 Other respiratory conditions in infants and toddlers
2 Five studies examined other respiratory conditions in infants and toddlers (see Table 1-20).
3 Three of these were considered medium confidence studies and are discussed below. Roda et al.
4 (20H) was a follow-up of 2,940 infants in a birth cohort, with questionnaires regarding respiratory
5 symptoms including lower respiratory infections and wheeze, completed by parents at 1, 3, 6, 9 and
This document is a draft for review purposes only and does not constitute Agency policy.
1-117 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
12 months. Formaldehyde exposure was modeled based on housing characteristic data and the
mean of four 1-week samples taken in homes at 1, 6, 9, and 12 months in a randomly selected
subset of 196 homes. The sensitivity and specificity of the modeling was estimated as 72.4 and
73.6% respectively for categorization based on the median and 57.4 and 82.1% for categorization
based on tertiles. EPA noted in its evaluation, however, that the modeling was not tested on a
separate sample, and thus these model characteristic estimates may be high. Rumchev et al. (2002)
is a study of emergency room visits for what was characterized as asthma (based on discharge
diagnosis); information on the recruitment and selection process was not presented. The relatively
young age of the children (mean 24 months, range 6 to 36 months) does not reflect the phenotypic
expression of asthma, and thus this study likely represents various respiratory tract infections and
wheezing episodes. Two 8-hour measures, in different seasons, of formaldehyde were taken in case
and control homes; the length of time between the hospital visit and the study was not specified.
Both of these studies reported associations between the examined outcome and residential
formaldehyde levels, with effects seen above 0.020 mg/m3 in Roda et al. (2011) and above
0.060 mg/m3 (possibly above 0.050 mg/m3) in Rumchev et al. (2002). Although the conditions
included in these studies do not fit within the usual classification of asthma, these respiratory
conditions may have implications for subsequent disease risk, and in the case of Rumchev et al.
(2002) (emergency room visits), also reflects an outcome with accompanying health care costs.
The association of formaldehyde exposure with symptoms consistent with increased lower
respiratory infections also may be indicative of immune suppression in the children, although this
was not directly tested in the available studies, and mechanistic findings that may support these
observations were similarly indirect and inconclusive (see Evidence on Mode-of-Action
Section below). Although the congruence between the outcomes examined within these two
studies is not clear, the results of these studies indicate that the relationship between indoor
formaldehyde exposure and respiratory conditions in infants and toddlers is an area requiring
additional research.
Table 1-20. Respiratory conditions in infants and young children in relation to
residential formaldehyde exposure
Study and design3
Results
Roda et al. (2011) (France)
Birth cohort, infants (singleton, >2,500 g)
followed through age 1 year; n = 2,940 with
12-month questionnaire and formaldehyde
measures (70% of 4,177 initial enrollees; 76%
of those completing at least one
questionnaire).
Exposure: Questionnaire on home
characteristics at baseline and updated at 3,
6, 9 and 12 months. N = 196 randomly
selected for predictive modeling analysis; 4 1-
week measures at 1, 6, 9 and 12 months.
OR (95% CI)
Lower respiratory tract infection (Prevalence through age 1 year 45.8%)
Per interquartile range increase:
1.32 (1.11,1.55)
Above vs. below median (0.02 mg/m3):
1.20(1.03,1.41)
Top tertile vs. lowest tertile:
1.31 (1.10,1.57)
Lower respiratory tract infection with wheeze
(Prevalence through age 1 year 22.3%)
Per interquartile range increase:
1.41 (1.14,1.74)
This document is a draft for review purposes only and does not constitute Agency policy.
1-118 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Predictive model used to assign subjects to
categorical levels.
LOD 0.008 mg/m3. Median 0.020 mg/m3; IQR
0.014, 0.027 mg/m3.
Exposure prediction model for high vs. low
(based on median):
Sensitivity 72.4%; Specificity 73.6%
Exposure prediction model by fertile:
Sensitivity 57.4%; Specificity 82.1%.
Outcome: Parent questionnaire at 1, 3, 6, 9,
and 12 months used to define lower
respiratory infections with and without
wheeze
Evaluation3:
Above vs. below median (0.02 mg/m3)
1.31 (1.07,1.59)
Top fertile vs. lowest fertile:
1.43 (1.14,1.79)
Adjusted for sex, prenatal and postnatal environmental tobacco smoke
exposure, breastfeeding history, number of older siblings, day care
attendance, furry pets in the home, humidity, parental history of asthma,
and socioeconomic status.
S3
IB
a oth
Overall
Confidence
Medium
!
¦
L!
Did not test predictive model on separate
sample (may overestimate sensitivity and
specificity)
Rumchev et al. (2002) (Australia)
Case-control, n = 88 cases, n = 104 controls
(health department); ages 6 months to 3 years
(mean 25 months for cases, 20 months for
controls). Participation rates not reported.
Exposure: Two 8-hour measures (winter,
summer) in home (living room, bedroom)
mean (max) (mg/m3)
living room: 0.028 (0.244); bedroom: 0.030
(0.189)
Outcome: Emergency room discharge
diagnosis of asthma
Evaluation3:
OR (95% CI) by exposure category15:
0.010-0.029 mg/m3 0.95 (0.8, 1.1)
0.030-0.049 0.95 (0.8, 1.2)
0.050-0.059 1.2 (0.9, 1.6)
>0.060 1.39 (1.1, 1.7)
Per 0.010 mg/m3: 1.003 (1.002,1.004)
(OR and 95% CI for all categories except >0.060 mg/m3 estimated from figure
in the paper; numbers in each exposure were not reported)
Adjusted for age, sex, allergic sensitization to common allergens, family
history of asthma, relative humidity, indoor temperature, socioeconomic
status, pets, air conditioning, gas appliances, smoking inside, house dust mite
levels
Overall
Confidence
SB IS Cf
Oth
HH"
Medium
Recruitment process not described;
uncertainty as to what is included within this
case definition and length of time between
emergency room visit and subsequent
exposure measure.
Related References: Rumchev et al.
(2004)
Li et al. (2019) (Hong Kong)
Birth cohort (2013-2014), Infants aged
<4 months (>2.5 kg, gestation >36 weeks)
followed to 18 months;
n = 963 (67% of recruited) with outcome and
exposure data.
New onset wheeze
Prevalence 12.5% at mean age of 13.4 months.
HR (95% CI) per 10 |ig/m3
1.002 (1.001,1.003)
This document is a draft for review purposes only and does not constitute Agency policy.
1-119 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Exposure: Air sampling (N02, formaldehyde),
72 hour samples at 6 months of age
(concentrations not reported), ISAAC
questionnaire included questions on
environmental conditions in residence.
Outcome: Parent questionnaire (ISAAC) prior
to 4 months, weekly respiratory health diary
and monthly telephone survey to 18 months.
New onset wheeze (time to event) measured
from 6 to 18 months of age.
Evaluation:
SB IB Cf Oth
Overall
Confidence
m
Low
Concern for
Cox proportional hazard models adjusted for N02 (ng/m3), sex, neo-natal
respiratory illness, sibling, keeping pets, cooking fuel, and family history of
non-asthma allergy or asthma.
selection bias. Participation rate was very low
(29% of eligible agreed) and of those selected
there was notable data loss, data was
complete for 67%. No comparisons of
participants and nonparticipants and no
descriptive statistics provided for study
sample. No control for smoking or ETS.
This document is a draft for review purposes only and does not constitute Agency policy.
1-120 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Yu et al. (2017) (Hong Kong)
Birth cohort (2009-2011), Infants aged
<4 months, followed to 18 months;
n = 535 (76.2% of recruited) with outcome and
exposure data.
Exposure: Air sampling at 6 months of age in
bedroom (N02, formaldehyde), sampling
period not reported, ISAAC questionnaire
included questions on environmental
conditions in residence.
Mean (SD) concentrations
N02 42.4 (30.97) Mg/m3; formaldehyde 51.09
(74.94) Mg/m3;
Outcome: Parent questionnaire (ISAAC) prior
to 4 months, weekly respiratory health diary
and monthly telephone survey to 18 months.
New onset wheeze (time to event) measured
from 6 to 18 months of age.
Evaluation:
New onset wheeze
Prevalence 11% at mean age of 11.4 months.
HR (95% CI) per 10 Mg/m3
1.004 (1.001,1.007)
Cox proportional hazard models adjusted for N02 (Mg/m3), sex, neo-natal
respiratory illness, sibling, keeping pets, cooking fuel, living area (ft2) and
family history of non-asthma allergy or asthma.
Overall
SB IB Cf Oth
Confidence
Low
No details provided
for exposure measurements. Concern for
selection bias. Participation rate was very low
(29% of eligible agreed) and of those selected
there was notable data loss, data was
complete for 76%. No comparisons of
participants and nonparticipants. No control
for ETS
Raaschou-Nielsen et al. (2010)
(Denmark)
Birth cohort, n = 343, infants of mothers with
asthma (83% of 411 enrollees, 90% of 378 who
participated through 18 months).
Exposure: 10-week samples in bedrooms, 1 to
3 sampling periods (at 6,12, and 18 months of
age). Analysis of variance: 31% between and
69% within person.
mean 0.020 mg/m3
median 0.018 mg/m3
95th percentile 0.037 mg/m3
Outcome: Daily symptom diaries kept from
birth to 18 months (reviewed at clinic visit
every 6 months), recording of wheezing
symptoms affecting activity or sleep.b
Evaluation3:
(n), OR (95% CI) by exposure quintiles. Outcome = any symptom day:
<0.012 mg/m3
0.012-0.016
0.016-0.020
0.020-0.026
>0.026
(67) 1.0 (referent)
(69) 1.11 (0.47, 2.63)
(68) 1.21 (0.51. 2.92)
(71) 1.40 (0.57, 3.47)
(68) 0.67 (0.29,1.54) (trend p = 0.49)
Adjusted for sex, area of residence, education of mother and
log-transformed baseline lung function
SB IB Cf Oth
Overall
Confidence
Low
This document is a draft for review purposes only and does not constitute Agency policy.
1-121 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Analysis does not take into account important
features of the data (e.g., temporal variations
in symptoms and in formaldehyde); could have
masked an association.
Evaluation of sources of bias or study limitations (see Appendix A.5.1 and A.5.4). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Susceptibility: modifying factors affecting prevalence of asthma or allergic sensitization
Asthma and atopic sensitization are hypothesized to be affected by a combination of genetic
and environmental factors. Several sensitization and asthma studies included analyses pertaining
to effect modification by factors that may help elucidate pathogenesis and susceptibility, such as
atopy (see Figure 1-11). In the study of adult women by Matsunaga et al. (2008), the association
between use of medication for atopic eczema and formaldehyde exposure was stronger among
women with no family history of allergy (OR 2.96, 95% CI 0.87,10.12) than among women with a
family history of allergy (OR 1.63, 95% CI 0.58, 4.57) at exposures of 0.058 to 0.161 mg/m3
compared with <0.058 mg/m3. The pattern across exposure levels also revealed an increase in risk
of atopic eczema at lower exposures in the negative family history group (OR 1.37,1.88, and 4.21)
compared with the positive family history group (OR 0.80, 0.92, and 1.45) (see Figure 1-11A). In
the study of asthma incidence in relation to formaldehyde measures in school by Smedje and
Norback (2001), the association between formaldehyde and asthma in the full sample was
relatively weak (OR 1.2, 95% CI 0.8,1.7), but there was some divergence in estimated effects in
analyses stratified by history of atopy: OR 1.7 (95% CI 1.1, 2.6) among children without a positive
history, and OR 0.6 (95% CI 0.3,1.3) among children with a positive history (see Figure 1-11B).
The pattern is difficult to interpret in the study by Annesi-Maesano et al. (2012.) (see Figure 1-11C),
as an indication of effect modification at lower exposures was not seen at higher exposures. Note
that the direction of effect modification seen in Matsunaga et al. (2008) and in Smedje and Norback
(2001) differ from that described in the preceding section (i.e., the stronger association between
formaldehyde and asthma control among children with atopy compared to nonatopics in Venn et al.
(2003). Examination of the presence of interactions and the factors contributing to them requires
large studies designed to test specific hypotheses defined a priori; thus, additional research is
needed to address the question of potential effect modification of atopic eczema or asthma
symptom prevalence by atopy status.
Tobacco smoke represents an environmental factor that may increase the incidence of
hypersensitivity responses in formaldehyde-exposed individuals. Two studies included IgE or
asthma analyses stratified by environmental tobacco smoke exposure among children and adults
(nonsmokers) (Palczvnski etal.. 1999: Krzvzanowski et al.. 1990). There was some evidence of
This document is a draft for review purposes only and does not constitute Agency policy.
1-122 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
effect modification by environmental tobacco smoke (i.e., stronger associations, or associations
seen at lower formaldehyde exposures, seen with this coexposure). In the Palczynski et al. (1999)
study, there was no association between formaldehyde and either IgE levels or asthma prevalence
in the full sample of children or of adults. Analyses stratified by the presence of environmental
tobacco smoke exposure in the home, however, indicated associations between formaldehyde (at
levels of 0.025-0.050 mg/m3) and (1) elevated IgE in children (but not adults), and (2) asthma in
adults (but not in children). In the study by Krzyzanowski et al. (1990), an association between
formaldehyde and asthma was seen in children exposed to environmental tobacco smoke, but
evidence of this type of effect modification was not seen in adults (see Table 1-21). Additional
studies are needed to establish if this interaction is seen only in children, only in adults, in adults
and children, or in neither group.
One other source of effect modification was examined in the study by Hulin et al. (2010), a
case-control study conducted in an urban and a rural area in France, with 32 and 24 cases,
respectively, in each area. The formaldehyde levels were similar in the two areas, but a strong
effect modification by area was seen, with an elevated risk seen in the rural area (OR 9.0) and a
decreased risk seen in the urban area (OR 0.24). Both estimates have wide confidence intervals.
These findings could be due to chance or could reflect interactions with other exposures or other
differences between the areas. The uncertainty in interpreting these stratified results contributed
to the low confidence rating for this study. Additional studies examining modifying factors would
be informative.
This document is a draft for review purposes only and does not constitute Agency policy.
1-123 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
A. Effect Modification of Atopic Eczema Risk in Adults
by Family History of Allergy
~ Family history - positive
A Family history - negative
0.02 0.04 0.06
Formaldehyde Exposure (mg/m3)
Data from Matsunaga et al., 2008
B. Effect Modification of Risk of Asthma Incidence in Children
by Atopy Status
o: ¦-
I i
~ History of atopy
A No history of atopy
1
0.02 0.04 0.06
Formaldehyde Exposure (mg/m3)
Data from Smedge and Norback, 2001
C. Effect Modification of Prevalence of Current Asthma in Children
by Atopy Status
~ Atopy - positive
A Atopy - negative
i
0.02 0.04 0.06
Formaldehyde Exposure (jig/m3)
Data from Annesi-Maesano et al., 2012
Figure 1-11. Examination of effect modification by family or personal history of atopy.
(A) Relative risk of prevalence of atopic eczema in adults (Matsunaga et al., 2008); study details in
Table 1-12. Family history defined as parent or sibling with doctor-diagnosed asthma, atopic eczema, or
allergic rhinitis. (B) Relative risk of incidence of asthma in children (Smedje and Norback, 2001); study
details in Table 1-15. Atopy defined at baseline as a positive response to questions on childhood eczema,
allergy to pollens, and allergy to pet dander. (C) Relative risk of prevalence of asthma in children (Annesi-
Maesano et al., 2012); study details in Table 1-15. Atopy based on positive skin prick test (10 allergens).
This document is a draft for review purposes only and does not constitute Agency policy.
1-124 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-21. Effect modification by environmental tobacco smoke: results from
studies in children and adults
Study and design3
Results
Palczvnski et al. (1999) (Poland)
Prevalence survey, n = 278, ages 16-65 and n = 187,
ages 5-15 years from 120 households with children
(random selection, 10-year old apartments).
Participation rate not reported.
Exposure: 24-hour household sample (area not
specified)
Mean (±SD) (minimum, maximum) 0.026 (±0.011)
(0.002, 0.067) mg/m3
2% >0.050 mg/m3
Outcome: Bronchial asthma diagnosed using American
Thoracic Society criteria.
Evaluation3:
SB
IB
Cf Oth
Overall
Confidence
Medium
1
Uncertainty regarding asthma definition. Not
informative above 0.050 mg/m3 because of sample size
(n = 4).
N per group (Percentage with
Current Asthma)
Environmental Tobacco Smoke
Exposure (mg/m3)
Positive
Negative
Children, IgE >100 kU/L
<0.025
39(38.5)
55 (29.1)
0.025-0.050
44(52.3)
46 (23.9)
0.051-0.067
2 (0.0)
1 (100.0)
(Fisher's exact test
p-value, children)
(0.005)
Adults, IgE >100 kU/L
<0.025
34(23.5)
67 (29.9)
0.025-0.050
36(22.2)
57 (26.3)
0.051-0.067
2 (0.0)
2 (0.0)
Children, Asthma
<0.025
39 (6.9)
55 (5.4)
0.025-0.050
44 (2.3)
46(6.5)
0.051-0.067
2 (0.0)
1 (0.0)
Adults, Asthma
<0.025
34 (5.9)
67 (4.4)
0.025-0.050
36(13.9)
57(1.8)
0.051-0.067
2 (0.0)
2 (0.0)
(Fisher's exact test
p-value, adults)
(0.03)
Krzyzanowski et al. (1990) (United states,
Arizona)
Prevalence survey, adults (n = 613 ages >15, mean 37)
and children (n = 298 ages 5-15, mean 9.3) from 202
households (stratified sample from municipal
employees). Participation rate not reported. 67%
whites
Exposure: Two one-week samples (opposite seasons) in
kitchen, living area, and bedroom (converted from ppb).
Household: mean 0.032 mg/m3
<0.049 mg/m3 83.7%
0.049-0.074 10.0%
0.074-0.172 6.3%
Only a few values above 0.111 mg/m3.
Outcome: Asthma and symptoms based on Ferris
(1978) (physician diagnosed)
Evaluation3:
N per group (Percentage with Current
Asthma)
Children
Environmental Tobacco Smoke
Exposure
(mg/m3)
Positive
Negative
<0.049
106(15.1)
142 (8.5)
0.049-0.074
12 (0.0)
12 (8.3)
0.074-0.172
11(45.5)
10 (0.0)
(trend p-value)
(<0.05)
(>0.50)
Log-linear models, stratified by environmental tobacco smoke,
adjusted for socioeconomic status, ethnicity.
Adults: Results reported as "not significantly related" but rate of
wheeze was "somewhat higher" with higher exposure; analyses
stratified by environmental tobacco smoke exposure not reported.
SB IB a Oth
Overall
Confidence
Medium
For children, relatively small n in higher exposure
categories; for adults, incomplete reporting.
This document is a draft for review purposes only and does not constitute Agency policy.
1-125 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design3
Results
Related references: Quackenboss et al. 1989a);
1989c)
1 Immune-mediated Conditions, Focusing on Allergies and Asthma, in Animal Studies
2 The animal studies most relevant to evaluating potential effects on allergy-related
3 conditions and asthma, as well as a single study suggesting a potential increased vulnerability to
4 respiratory infections, are discussed in the sections below.
5 Allergy-related conditions and asthma
6 There are currently no universally accepted animal models applicable to humans for
7 determining dose-response relationships or the potency of low molecular weight chemicals to
8 induce allergic symptoms via the inhalation route (IPCS. 2012). The majority of the experimental
9 animal formaldehyde studies that are most relevant to interpreting these respiratory
10 immune-mediated conditions used the ovalbumin (OVA) murine model, the best studied animal
11 model of asthma. However, the OVA mouse model has several limitations relative to human data
12 for hazard characterization. They include the following:
13 • Key features of human asthma are absent or minimal in the OVA model, including a lack of
14 airway remodeling fShin etal.. 20091 and minimal airway hyperreactivity and eosinophilic
15 inflammation fMullane and Williams. 20141
16 • OVA challenge models a small subset of endpoints and genes compared with those in
17 humans fMullane and Williams. 20141
18 • The OVA model elicits an acute disease in contrast to the chronic condition in humans (Shin
19 etal.. 20091. and the antigen ovalbumin has questionable relevance and poor translatability
20 for human asthma (Mullane and Williams. 2014: Bates etal.. 2009)
21 • A standardized method for OVA administration is lacking; this precludes comparing results
22 between laboratories and evaluating study protocols (Bates etal.. 2009)
23 • There is uncertainty regarding the biological significance of airway hyperreactivity in mice
24 (Bates etal.. 2009)
25 In light of these limitations, EPA concluded for this assessment that the OVA model was
26 more appropriate for examining mechanistic questions in support of hazard identification, based in
27 part on the reasonably large number of well-conducted human studies on these endpoints. As such,
28 the experimental animal studies were considered to be less informative than human studies for
29 drawing interpretations regarding the potential for formaldehyde inhalation exposure to induce or
30 exacerbate allergy-related conditions or asthma, and these studies are discussed below as
31 mechanistic information that may add insight to the apical effects observed in exposed humans.
This document is a draft for review purposes only and does not constitute Agency policy.
1-126 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Other respiratory conditions
One experimental animal study of medium or high confidence evaluated endpoints related
to the potential for formaldehyde exposure to cause other immune-mediated respiratory
conditions, and reported a decrease in pulmonary antibacterial activity in mice exposed to
1.23 mg/m3 formaldehyde for less than 1 day (lakab. 1992). While such a finding could indirectly
suggest that formaldehyde exposure might predispose animals to developing lower respiratory
infections, this hypothesis was not specifically tested and other notable uncertainties with the study
design exist (see Appendix A.5.6). Animal studies of long-term duration that are specifically
designed to examine the functional capacity of the respiratory immune response would be
informative.
Evidence on Mode of Action for Immune-mediated Conditions, Including Allergies and Asthma
An integrated evaluation of the abundant mechanistic information that might be relevant to
the potential development of immune-mediated conditions following formaldehyde inhalation
exposure is described in Appendix A.5.6, including evaluations of the individual mechanistic
studies. The evaluation includes the somewhat heterogeneous data related (either directly or
indirectly) to possible increases in respiratory infections after exposure, although those data are
not discussed in detail in this section. Thus, this discussion focuses on mechanistic information that
may inform the potential for formaldehyde to affect allergic conditions or asthma. This includes
animal models using the allergen, OVA, which, although they do not fully capture the phenotype of
human asthma or allergy-related conditions, can provide insight into some of the mechanistic
changes that are relevant to these human conditions.
As shown in Figure 1-12, the integrated analysis identified several pathways describing
potential associations between the most relevant mechanistic data available, with several of the
initial or early events in these hypothesized pathways (e.g., oxidative stress and inflammatory
changes) generally observed to occur at lower formaldehyde levels than other downstream changes
(see Table 1-22). Overall, the mechanistic support for airway hyperresponsiveness was stronger
(i.e., based primarily on moderate evidence of mechanistic events and their relationships).
Although a definitive MOA(s) could not be defined, and it is unclear whether some important events
would occur with chronic low-level formaldehyde exposure, the data were interpreted to identify
an incomplete mechanism(s) by which formaldehyde exposure could cause this effect (see
Figure 1-12), providing biological plausibility for inflammatory airway changes that could
contribute to respiratory immune-mediated conditions. The mechanistic support for allergic
sensitization was less clear (i.e., based on some potentially relevant events interpreted with
moderate evidence and, in general, slight evidence for the relationships between events) because
reliable data identifying mechanistic changes typically thought to be essential for sensitization,
including changes in IgE, were lacking. However, moderate evidence for several mechanistic
changes relevant to these responses was identified, providing some biological support.
This document is a draft for review purposes only and does not constitute Agency policy.
1-127 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 Importantly, while many individual mechanistic events observed in animals are considered to be
2 relevant to interpreting changes that may occur in the human airways, including potential amplified
3 responses to inhaled materials, it is unclear how translatable these pathways are to interpreting
4 complex human diseases like asthma, and notable key events have not been observed. Some of the
5 data most informative to drawing conclusions for these health endpoints are described in greater
6 detail below (see Tables 1-22 and 1-23).
Initial Alterations
Secondary Alterations
Effector-Level Changes
O
-->»
oxidative Sensory nerve LRT neuro f" LRT micro- ^Eosinophils •f airway edema/
stress in LRT stimulation in LRT peptides vascular leakage in LRT* inflammatory
structural change*
Key Hazard Feature
Airway hyper-
responsiveness*
• o o o
"t1 stress 'f oxidative 4, CDS*! cells ^ IL-4, 4^ IFNy Altered ES
hormone stressin blood in blood in blood cells
o o^o o
o
Altered antibody Allergic Airway hyper-
responses* sensitization responsiveness*
o
URT epithelial "T1 URT ¦f'airway 'T'CD8^ T cells & ^Eosinophils Sustained airway Allergic Airway hyper-
damage inflammatory neuropeptides Th2-related jn LRT* inflammation* sensitization responsiveness*
cells & factors cytokines in LRT
Legend
^ Plausibly an initial
effect of exposure
Key feature of
}V§ respiratory immune-
mediated conditions
EVIDENCE
Q11 Robust
( ) Moderate
( ) Slight
RELATIONSHIP
—> Robust
--> Moderate
Slight
*effects are amplified
with allergen exposure
Figure 1-12. Possible mechanistic associations between formaldehyde
exposure and immune-mediated conditions, including allergies and asthma.
An evaluation of the formaldehyde exposure-specific mechanistic evidence informing the potential for
formaldehyde exposure to cause respiratory health effects (see Tables 1-22 and 1-23, and Appendix A.5.6)
identified these mechanistic pathways as most relevant to interpreting effects on respiratory immune-
related conditions such as asthma and allergic responses. Similar to effects on pulmonary function,
events related to indirect stimulation of lower respiratory tract (LRT) sensory nerve endings (top pathway)
were considered as likely to represent an incomplete mechanism by which formaldehyde inhalation could
cause airway hyperresponsiveness, although whether certain events occur with chronic, low-level
exposure remains unclear. While the observed alterations to circulating antibodies (i.e., primarily related
to IgG and not IgE) following formaldehyde exposure might contribute to the development of both allergic
sensitization and airway hyperresponsiveness (middle pathway), in the absence of additional clarifying
data, this couid not be identified as a likely mechanism for these effects. Likewise, the slight evidence of
altered T cell-related airway responses and, secondarily, inflammatory eosinophil responses might be
useful for explaining allergic sensitization (bottom pathway) if additional data were available to better
explain the pattern and strength of these associations. Conversely, sustained airway inflammation, at
least in animals previously sensitized to an allergen, was interpreted as likely to be an incomplete
explanatory mechanism for airway hyperresponsiveness, although the sequence of events leading to
This document is a draft for review purposes only and does not constitute Agency policy,
1-128' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
inflammation remain unclear. Interdependencies between the top and bottom pathways are likely to
exist for airway hyperresponsiveness.
It is informative to consider the formaldehyde-specific mechanistic information in the
context of the known pathogenesis of human asthma and related conditions. Asthmatic airways are
characterized by an infiltration of eosinophils, plasma B cells, activated mast cells, and T cells that
contribute to thickening of the airway wall, mucous secretion, airway remodeling and airway
hyperresponsiveness. Initiation and perpetuation of asthma are believed to be the result of Th2
activity fCohn etal.. 20041. Specifically, Th2 cells accumulate in the airway and secrete cytokines
IL-4 and IL-13, which stimulate B cells to produce IgE (Barnes. 20081 (see Figure 1-13). Mast cells
bind IgE and display this immunoglobulin as an allergen-specific receptor on their surfaces. When
an allergen binds to this IgE, the mast cell is activated, triggering its release of several
bronchoconstrictors (e.g., histamine, leukotrienes), which drive the disease state. Th2 cells also
release IL-5 that activates eosinophils following their migration into the airways. The precise role
of eosinophils in asthma is unknown, but they are thought to contribute to inflammation (Barnes.
20081. Immune function and inflammatory responses do not fully explain the pathogenesis of
asthma, particularly with respect to the varying phenotypes seen at a clinical level (Anderson.
20081. The interaction between nerve cells and the immune system also includes evidence that
neuropeptide release may contribute to neurogenic inflammation and heightened airway
responsiveness (Veres etal.. 2009).
Inhaled allergens
Figure 1-13. Inflammatory and immune cells involved in asthma
This document is a draft for review purposes only and does not constitute Agency policy.
1-129 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
Inhaled allergens activate sensitized mast cells by crosslinking surface-bound IgE molecules to release
prostaglandin D2. Epithelial cells release stem-cell factor (SCF), which is important for dendritic cells,
which are conditioned by thymic stromal lymphopoietin (TSLP) secreted by epithelial cells and mast cells
to release the chemokines CC-chemokine ligand 17 (CCL17) and CCL22, which act on CC-chemokine
receptor 4 (CCR4) to attract T-helper 2 (TH2) cells. TH2 cells have a central role in orchestrating the
inflammatory response in allergy through the release of interleukin-4 (IL-4) and IL-13 (which stimulate B
cells to synthesize IgE), IL-5 (which is necessary for eosinophilic inflammation), and 11-9 (which stimulates
mast-cell proliferation). Epithelial cells release CCL11, which recruits eosinophils via CCR3. Patients with
asthma may have a defect in regulatory T (Treg) cells, which may favor further TH2-cell proliferation.
Reprinted from Barnes (2008) with permission from Nature Publishing Group.
The mechanistic evidence that provides the most direct information regarding the potential
role of formaldehyde in respiratory hypersensitivity responses consists of three high or medium
confidence studies (Larsen etal.. 2013: Fujimaki et al.. 2004b: Ito etal.. 1996: Riedel etal.. 1996:
Swiecichowski etal.. 1993).12 These studies all differed in the conditions under which
formaldehyde affected asthma-relevant endpoints, specifically increased bronchoconstriction and
airway hyperresponsiveness, using short-term and acute exposures in sensitized and nonsensitized
animals. Formaldehyde exposure of 0.369 to 36.9 mg/m3 increased bronchoconstriction in guinea
pigs exposed for 2 to 8 hours (Swiecichowski etal.. 19931. Both the in vivo and ex vivo data from
this study indicate that smooth muscle airways are a (presumably indirect) target for
formaldehyde. A 5-day formaldehyde exposure of 0.31 mg/m3 prior to OVA sensitization increased
OVA-induced bronchoconstriction in guinea pigs, indicating that formaldehyde exposure enhances
reactivity to OVA sensitization (Riedel etal.. 1996). Finally, a single 60-minute formaldehyde
exposure of 7.0 mg/m3 induced bronchoconstriction in OVA-sensitized mice housed only in humid,
but not dry, environments, indicating that the bronchoconstrictive effects of formaldehyde may be
impacted by humidity fLarsen etal.. 20131. Taken together with supportive findings from a
number of low confidence human and animal studies (see Appendix A.5.6), results across multiple
species indicate that formaldehyde exposure is sufficient to trigger bronchoconstriction in both
sensitized and nonsensitized animals, and that exposure appears to result in the development of
hyperresponsive airways,13 particularly in sensitized animals. This finding is consistent with the
evidence supporting increases in microvascular leakage, edema, and other inflammatory airway
changes with formaldehyde exposure after allergen sensitization (see Section 1.2.2 and
Appendix A.5.6). Overall, the data do not indicate that formaldehyde is itself immunogenic, but
instead suggest formaldehyde may augment immune responses to other allergens.
Other findings that may be relevant to asthma or allergic conditions with at least a
moderate level of evidence include increases in airway eosinophils, increases in protein mediators
12Note: Swiecichowski et al. Q9931 and Leikauf (1992) are interpreted to use the same cohort of animals.
13As the challenge stimuli used in the formaldehyde studies included allergens as well as nonimmunological
stimuli, and because most experiments did not attempt to delineate the specifics of the functional changes,
"airway hyperresponsiveness" or "hyperresponsive airways" encompasses any of a range of possible airway
features: hyperreactivity (exaggerated response), hypersensitivity (lower dose to elicit response), altered
ventilatory parameters (e.g., maximal response, resistance), recovery (longevity of response), or others.
This document is a draft for review purposes only and does not constitute Agency policy.
1-130 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
of bronchoconstriction such as tachykinins, and changes in antibody titers (see Section 1.2.2 and
Table 1-22). Although a precise role for eosinophils in asthma is unknown (i.e., eosinophilia is not
necessary for the development of asthma), eosinophilic airway inflammation (presumably
mediated by Th2 lymphocytes) is a hallmark of asthma f George and Brightling. 20161: the
formaldehyde-specific evidence indicates that eosinophils are increased in both the upper and
lower airways following formaldehyde exposure, particularly with allergen sensitization (see
Section 1.2.2). As activation of eosinophils can induce airway hyperresponsiveness and perpetuate
further recruitment of inflammatory mediators into the airway (Cohn etal.. 2004). these changes
provide coherent biological support for the more apical immune-mediated conditions. In addition,
as previously discussed (see Section 1.2.2), it appears that formaldehyde exposure mediates (at
least in part) lung inflammation via tachykinins in rats and mice. For example, high or medium
confidence studies show that substance P, a tachykinin and NK1 ligand, is dose-dependently
increased in mice exposed for 12 weeks to 0.1 to 2.5 mg/m3 formaldehyde (Fujimaki etal.. 2004b).
and that an antagonist of the NK1 receptor can completely abrogate formaldehyde-induced airway
inflammation, at least following a 10-minute formaldehyde exposure at 18 mg/m3 (Ito etal.. 1996).
Somewhat surprisingly, however, the formaldehyde-induced increases in substance P observed by
Fujimaki et al. (2004b) were not observed in animals sensitized to OVA, despite the observation
that airway eosinophils were increased at 2.5 mg/m3 formaldehyde only in animals that were
sensitized. Thus, some uncertainties remain. The results related to antibody production, although
providing moderate evidence of an effect, were difficult to interpret in the context of their relevance
to asthma. Specifically, while evidence from human and animal studies suggests that formaldehyde
exposure modifies antibody responses, the most consistently observed responses were associated
with changes in IgG, not IgE (see Table 1-22). The relevance of IgG-related responses to asthma or
allergies is unclear.
Several other airway changes relevant to asthma or allergic conditions were not supported
by moderate or robust evidence in the available studies. For example, slight evidence suggests
changes in CD8+ T cells or asthma-relevant Th2 cytokines, including IL-4 [and, to a lesser extent, IL-
5 and RANTES (regulated on activation, normal T cell expressed and secreted)], in the lungs after
exposure to 0.5-12 mg/m3 formaldehyde in both sensitized and nonsensitized rodents; however,
no changes in IL-13 or histamine have been reported. At the cellular level, while slight evidence
suggests that CD8+ T cells might be increased in naive rodents exposed to >7 mg/m3 formaldehyde,
mast cells or other T cell populations did not appear to be changed in the few studies that examined
them, and none of the identified studies investigated other cells of interest (e.g., dendritic cells,
smooth muscle cells).
Immune-related changes in the blood may also be relevant to interpreting the development
of allergic conditions, and possibly asthma, albeit indirectly. A number of studies, across different
human and animal populations, spanning an array of formaldehyde exposure scenarios, have
reported changes in blood cell counts and secreted factors. Although some of the specific changes
This document is a draft for review purposes only and does not constitute Agency policy.
1-131 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
vary across studies, taken together, the data provide robust evidence of an association between
formaldehyde exposure and hematological effects. Interestingly, some changes noted in the blood
of individuals exposed to formaldehyde are contrary to the cellular changes noted in the
respiratory tract (e.g., CD8+ T cells appear to be increased in the respiratory tract and decreased in
the blood) (see additional discussion in Appendix A.5.6). Potential explanations could include
recruitment of subsets of immunoresponsive cells from the circulation to the irritated and inflamed
respiratory tract (e.g., due to a gradient of chemoattractants or other factors across tissue
compartments, potentially resulting from sustained airway inflammation), or species differences in
responses (i.e., LRT data are mostly from animal studies, while the data in blood are primarily from
humans); however, none of the identified human studies report data across tissue compartments,
and the animal data do not address such hypotheses. Overall, similar to the cellular changes in the
LRT, no explanation exists for how formaldehyde exposure could affect blood immune cell counts.
One of the most consistent blood cell changes observed across studies was a decrease in the
total number of white blood cells (WBCs), including moderate evidence of CD8+ T cell decreases
following formaldehyde exposure and a corresponding increase in the ratio of CD4+/CD8+ T cells
(see Table 1-23). Depending on the specific stimuli, stimulated CD8+ T cells can produce interferon-
y (IFN-y) and inhibit production of IL-4 and immunoglobulin (i.e., IgE) responses fHolmes etal..
19971. or their phenotype can be driven toward production of excess IL-4, a situation hypothesized
to be associated with atopic asthma (Lourenco et al.. 2016). IL-4 can stimulate T cell receptors on
CD4+ and CD8+ T cells (Serre etal.. 2010). and can both drive CD4+ T cells toward a Th2 response
(Kopfetal.. 1993) and influence the activation and development of antigen-specific CD8+ T cell
immunity by shifting the phenotype of these cells from IFN-y production to IL-4 production (Erb
and Le Gros. 19961. Moderate evidence provides support for increases in blood IL-4 (slight
evidence suggests similar increases in the LRT) and decreases in IFN-y after formaldehyde
exposure. Interestingly, several lines of evidence suggest a pattern of immune cell effects related to
formaldehyde concentration, with potential stimulation at lower formaldehyde exposure levels and
decreases at higher levels. This included slight evidence of changes in total T cells, NK cells, and IL-
10. A complex relationship exists between IL-10, NK cells, and subsets of CD4+ T cells (e.g., TH1 and
TH2 cells), which can affect antibody responses (Moore etal.. 2001). However, the potential effects
of formaldehyde exposure on the specific phenotype of CD4+ or CD8+ T cells, or on the relationship
between changes in lymphocyte populations or secreted factors and respiratory hypersensitivity,
have not been well studied and remain to be elucidated.
Several other changes in the blood are of interest to the development of immune-mediated
conditions (see Appendix A.5.6 for additional discussion). Moderate evidence indicates that
formaldehyde exposure alters the percentage of B cells in the circulation. These cells produce
antibodies upon stimulation with antigen (e.g., allergens) and can contribute to airway
hyperresponsiveness fHamelmann et al.. 19971. While this finding, along with slight evidence of
increased antigenic markers, suggests the potential for alteration of the adaptive immune response
This document is a draft for review purposes only and does not constitute Agency policy.
1-132 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
after formaldehyde exposure, this observation alone is insufficient to indicate functional changes
such as exposure-induced differences in clonal expansion and differentiation to antibody-producing
cells, evidence of which would support a more convincing biological relationship. In addition, red
blood cell counts were decreased in both human and animal studies (moderate evidence), generally
at formaldehyde concentrations above 0.5 mg/m3, although the relevance of these changes to
respiratory system health effects is unknown. It is plausible that sustained increases in oxidative
stress (markers for which are consistently elevated in blood and respiratory tissues after
formaldehyde exposure), or other soluble factors that could segue from airway inflammation, might
affect the viability of circulating erythrocytes and immune cells, or the circulating precursors for
these cells; however, no evidence exists to substantiate this hypothesis. An increased level of the
circulating stress hormone, corticosterone (the major animal glucocorticoid; in humans, it is
Cortisol), with short-term, but not acute, formaldehyde exposure is also suggested. Persistent
increases in circulating glucocorticoids can also negatively impact the function and health of
circulating immune cells, causing immunosuppression of most cell types (O'Connor et al.. 2000).
However, these potential linkages have also not been examined.
Overall, although additional studies clarifying inconsistencies across the studies would be
informative, the available data support a conclusion that formaldehyde exposure can modify
immune system function in the blood across a range of concentrations and exposure durations.
Many of these observations would benefit from more specific studies on WBCs focused on
understanding the phenotype of the modified cells, and the profile of secreted factors in the blood,
particularly after formaldehyde exposures of varying duration and concentration. Taken together,
the available mechanistic studies provide consistent evidence that formaldehyde may stimulate a
number of immunological and neurological processes related to allergic or asthmatic responses;
however, a molecular understanding of how formaldehyde exposure might favor asthmatic Th2
responses has not been experimentally established and additional experimental support is
necessary to interpret the translatability of these pathways to complex human airway diseases such
as asthma. Importantly, the evidence supports that formaldehyde exposure induces
bronchoconstriction with and without allergen sensitization, providing strong biological support
for the development of hyperresponsive airways that could contribute to at least some of the
observed respiratory immune-related symptoms. This heightened bronchoconstriction response
may be due to a combination of neurogenic mechanisms through reduction of anti-inflammatory
molecules or increased tachykinins, increased Th2 cytokines and antibodies, and eosinophil
recruitment and activation in the lung. Immune- and inflammatory-related changes in the blood
provide additional support for exposure-induced alterations relevant to the development of these
immune-mediated conditions. Additional studies are necessary to clarify the incomplete
mechanisms that describe the association between formaldehyde exposure and these effects, as
well as the exposure concentration and duration dependence of some of the more influential
This document is a draft for review purposes only and does not constitute Agency policy.
1-133 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 findings from the current studies. Collectively, the available studies provide mechanistic support
2 for the biological plausibility of the formaldehyde exposure-induced changes observed in humans.
Table 1-22. Mechanistic evidence most informative to the development of
immune-mediated conditions after formaldehyde inhalation3
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
Modifications in the upper or lower respiratory tract (URT or LRT)
Some of these mechanistic changes have been discussed in previous sections.
See Section 1.2.2, Evidence on mode of action..., for presentation of the evidence for:
LRT oxidative stress (moderate); LRT sensory nerve activation (slight); LRT neuropeptides (moderate); LRT
microvascular leakage (moderate); LRT eosinophils (moderate); airway edema or other inflammatory structural change;
and URT epithelial damage (robust)
Upper
airway
indicators
of altered
immune
function
(inferred
from URT
infections)
High or Medium
Human: Increased frequency and duration of URT infections
in symptomatic workers; increased chronic URT inflammation
(and decreased function of blood neutrophils, but N/C in
leukocyte counts) in exposed workers (Lvapina et al.,
2004): chronic (yrs) exposure at 0.87 mg/m3(Note: recent
URT infection was often an exclusion criterion in
observational studies focusing on pulmonary function)
Indirect evidence of
decreased immune
capacity in a human study
of long-term exposure at
0.87 mg/m3 (note: mRNA
changes were not
necessarily indicative of a
decreased immune
response)
Slight
(indirect
evidence of
^URT
infection)
Animal: mRNA changes suggestive of altered immune
response (Andersen et al., 2010): short-term (>1 wk)
exposure at >12.3 mg/m3
Low
Human: None
No evidence to evaluate
Animal'. None
Lower
airway
indicators
of altered
immune
function
(inferred
from LRT
infections)
High or Medium
Human: Increased LRT infections in infants (Roda et al.,
2011): 32-41% increase in incidence per 0.0124 mg/m3
increase in formaldehyde (LOD: 0.008 mg/m3); ~l-year
exposure at 0.020 mg/m3 (median)
Indirect evidence in a single
study of infants exposed to
a median of 0.020 mg/m3
observing an association
between exposure and
increased infections. One
acute mouse study also
provided indirect support
for an increased likelihood
of respiratory infections.
Moderate
(indirect
support for an
increased
propensity for
LRT infections,
particularly
during
development)
Animal: Decreased antibacterial activity in mice (Jakab,
1992): acute exposure at 1.23 mg/m3, noting that this finding
appeared to be particularly sensitive to the pattern of
formaldehyde exposure
Low
Human: Increased emergency room visits for episodes
including LRT infections (RumcheV et al., 2002): children
aged 6-36 months at mean levels 0.028-0.030 mg/m3
(maximum 0.12-0.22)
Direct and indirect evidence
of impaired LRT immune
function in children and in a
short-term rat study,
respectively
Animal: Decreased expression of immune-related genes in rat
lung (Sul et al,, 2007), specifically HSP701a (involved in
antigen presentation), complement four binding protein (binds
necrotic or apoptotic cells for cleanup), and Fc portion of IgGiii
(involved in leukocyte activation): 2 wk exposure at
>6.15 mg/m3
Changes in
pulmonary
function
with
High or
Medium
Human: None
Acute and short-term
studies in two animal
species demonstrate that
formaldehyde increases
Robust
(^ Hyper-
responsive
airways'5)
Animal: [allergen challenge]: With ovalbumin [OVA]
sensitization, increased airway obstruction in guinea pigs
This document is a draft for review purposes only and does not constitute Agency policy.
1-134 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
challenge
(e.g., with
broncho-
constrictor
allergen)
(Note: un-
provoked
responses
are not
included)
(Riedel etal., 1996): short-term exposure at 0.31 mg/m3
and increased reactivity in mice (Larsen et al., 2013):
acute exposure at ~5-7 mg/m3 in humid or dry environments;
[acetylcholine challenge]: Increased airway resistance and
reactivity in guinea pigs (Swiecichowski et al., 1993;
Leikauf, 1992): acute exposure at 1.23 mg/m3
responsiveness to allergens
and bronchoconstrictors,
particularly with prior
sensitization, at levels as
low as 0.31 mg/m3
O
Human: [histamine challenge]: Hyperreactive airways with
prolonged exposure (Gorski and Krakowiak, 1991):
>1 year exposure at <0.5 mg/m3, but N/C after acute exposure
(Krakowiak et al., 1998): at 0.5 mg/m3; [allergen
challenge]: hypersensitivity with acute exposure when
exposure was restricted to mouth breathing in allergic
asthmatics with a large allergen (mite) (CaSSet et al.,
2006): acute exposure at 0.1 mg/m3; N/C after oronasal
exposure in allergic asthmatics using a different allergen
(pollen), including a methacholine (MCh) responsiveness test
after allergen exposure (EzrattV et al., 2007): acute
exposure at 0.5 mg/m3
Suggestive evidence of
increases with prolonged
exposure, and possibly
acute mouth-breathing
exposure when challenged
with specific allergens, but
not acute exposure alone, to
<0.5 mg/m3 in human
adults; also, increased at >3
mg/m3 in short-term or
acute studies across three
species, particularly with
prior sensitization
Animal: [MCh challenge]: Hyperresponsive airways (increased
reactivity and sensitivity) with exposure in mice and rats (Wu
et al., 2013; Liu et al., 2011a; Qiao et al., 2009):
short-term exposure at >3 mg/m3, and in monkeys (Biagini
et al., 1989): acute exposure at 3.1 mg/m3; in mice and
rats, this response was amplified with OVA sensitization; TRP
antagonists reduced the hyperresponsiveness in mice (Wu et
al.. 2013)
Sustained
Inflam-
mation
High or Medium
Human: Increased exhaled nitric oxide, a noninvasive and
indirect marker of lower airway inflammation and oxidative
stress, in healthy or asthmatic children (Flamant-Hulin et
al., 2010; Franklin et al., 2000): unknown exposure
duration (likely months to years; in classrooms or homes) at
0.04-0.06 mg/m3
Immune cell counts are
continually elevated in a
subchronic mouse study
with allergen stimulation at
2.46 mg/m3; increased
biomarkers (indirect
evidence) of lower airway
inflammation are observed
in children with prolonged
exposure.
Moderate
(may require
allergen
sensitization)
Animal: Eosinophils and monocyte counts remain elevated
with continued exposure for subchronic duration with allergen
(ova) sensitization (Fuiimaki et al., 2004b): 12 wk
exposure at 2.46 mg/m3
£
o
¦-J
Human: None
BAL cell counts and
histologic evidence suggest
that inflammation persists
for several weeks with
short-term exposure, and
these effects are amplified
by allergen
Animal: Immune cell counts were increased with short-term
exposure in several studies at >0.5 mg/m3 (see Table 1-23;
histological evidence of inflammation without epithelial
damage was noted in short-term exposure studies, typically at
higher concentrations, which were amplified by allergen
(e.g., >3 mg/m3; (Wu et al., 2013; Kimura et al.,
2010)
t CD8+T
cells in LRT
High or
Medium
Human: none
No evidence to evaluate
Slight
(at >7 mg/m3,
but allergen
Animal: none
This document is a draft for review purposes only and does not constitute Agency policy.
1-135 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
O
—1
Human: none
A study in rats and another
in mice suggest that CD8+ T
cells in the BAL might be
increased after short-term
exposure to high (>7
mg/m3) levels, although a
second mouse study
reported no changes
stimulus
unstudied)
(note: mixed,
indeterminate
evidence for B
cells, and CD4+
cells; Appendix
A.5.6)
Animal: Increased in short-term exposure studies in rats [at 7.4
mg/m3; (Sandikci et al., 2007b)l and mice [at 12.3
mg/m3; (Jung et al., 2007)1; no change with short-term
exposure in a mouse study at >6.2-12.3 mg/m3 (Kim et al.,
2013a)
t Th2-
related
(primarily)
cytokines in
LRT
High or
Medium
Human: none
No evidence to evaluate
Slight
(T* IL-4 at >0.5
mg/m3 and IL-
5 at >6 mg/m3)
(note: mixed,
indeterminate
evidence for
IL-10, IL-6, IL-
13, and for Thl
cytokines; see
Appendix
A.5.6)
Animal: none
o
—J
Human: No change in IL-4 or IL-5 at 0.5 mg/m3 after acute
exposure and pollen coexposure (EzrattV et al., 2007)
IL-4 was increased in short-
term studies of rats and
mice at levels as low as
0.5 mg/m3, with amplified
increases with antigen; IL-5
was increased in 2 of 3
studies in mice only testing
higher (>6mg/m3) levels
Animal: 'T* IL-4 in 4 studies in mice and one study in rats (all
short-term exposure) testing exposures of 0.5-12.3 mg/m3 and
observing larger increases with antigen (OVA) administration
(Wu et al., 2013; Liu et al., 2011a; Qiao et al.,
2009; Jung et al., 2007; Lu et al., 2005)
'T* IL-5 in 2 short-term exposure studies in mice at >6.2 mg/m3
(Jung et al., 2007; Sadakane et al., 2002)
No change in IL-4 in a short-term exposure study in mice at
>12.3 mg/m3 with co-administered house dust mite antigen
(Sadakane et al., 2002)c
Modifications in the blood
[[See Table 1-23 for cellular and cytokine responses in the blood]]
Total IgE
High or
Medium
Human: None
Slight (at > 3 mg/m3)
Based on no changes in a
high or medium confidence
subchronic mouse study at
<2.46 mg/m3 and evidence
of increased IgE in two
short-term low confidence
formalin studies in mice at
>3 mg/m3, but no evidence
for changes in low
confidence studies in mice
or humans at <2 mg/m3
Moderate
for IgG
Slight
for IgE
(only with
specific
exposure
scenarios)
Indeterminate
for IgM or IgA
(i.e., very little
evidence; data
not shown: see
Appendix
A.5.6)
Animal: No evidence suggesting changes (Fuiimaki et al.,
2004b): subchronic exposure at <2.46 mg/m3
o
—1
Human: No evidence suggesting changes (Ohmichi et al.,
2006; Erdei et al., 2003; Wantke et al., 2000;
Palczvnski et al., 1999; Wantke et al., 1996b):
short-term exposure at <1.8 mg/m3 (duration in Erdei et al.
unknown)
Animal: Evidence of increases in mice, which were increased
further by OVA sensitization (Wu et al., 2013; Jung et
al., 2007): short-term exposure at >3 mg/m3; evidence of no
changes in mice by FA alone (Kim et al., 2013b; Gu et
al., 2008), although FA exacerbated house dust mite-
induced IgE (Kim et al., 2013b): short-term exposure at
0.12-1.2 mg/m3
Formal-
dehyde
(FA)-
Specific IgE
High or Medium
Human: Elevated in one study of children (Wantke et al.,
1996a): years of exposure (assumed) at ~0.06 compared to
~0.03 mg/m3 (note: elevations were unrelated to symptoms);
N/C in adults (Kim et al., 1999): 4 years at 3.74 mg/m3
Slight (in children)
Based on increases in a high
or medium confidence long-
term study of children at
<0.1 mg/m3; although, no
changes were observed in a
Animal: None
This document is a draft for review purposes only and does not constitute Agency policy.
1-136 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
O
Human: No evidence of changes across multiple studies in
adults(Ohmichi et al., 2006;Zhou et al., 2005;
Wantke et al., 1996b; Gorski and Krakowiak,
high or medium confidence
long-term study of adults at
3.74 mg/m3 and there was
no clear evidence of
changes across multiple low
confidence short-term and
long-term studies in adults
at <1.81 mg/m3
1991; Thrasher et al., 1987): short-term (weeks) or
long-term (years) exposure at ~0.1-1.81 mg/m3; however,
findings were unclear in two adult studies of long-term
exposure in which a small proportion of subjects did have FA-
igE (Dvkewicz et al., 1991; Thrasher et al., 1990),
and one study noted slight increases with longer exposure
(Wantke et al., 2000): 10 wk, not 5 wk, at 0.265 mg/m3
Animal'. No change in guinea pigs with acute challenge (Lee
et al., 1984) at 2.5 or 4.9 mg/m3 after short-term exposure
to 7.4 or 12.3 mg/m3 (note: no measures without
formaldehyde and isotype was unspecified)
Antigen-
Specific IgE
(does not
include FA-
specific Ig)
High or
Medium
Human: None
Slight
Based on no changes in a
high or medium confidence
subchronic study with i.p.
antigen sensitization and
evidence in low confidence
short-term studies in mice
exposed to >1 mg/m3 that
appears to be highly
situational (e.g., dependent
on duration and periodicity
of formaldehyde exposure,
and antigen type and
administration route)
Animal-, n/c in ovA-igE (Fuiimaki et al., 2004b): 12 wk
exposure at 0.1-2.46 mg/m3 (OVA i.p.)
£
o
Human: None
Animal: Increased OVA-specific IgE in mice in two short-term
exposure studies (Gu et al., 2008; Tarkowski and
Gorski, 1995): 10 d at 2 mg/m3 (but not 1 d/wkfor 7 wk, or
when OVA sensitization i.p.) and 5 wk at 0.98 mg/m3 with i.p.
OVA (but not <4 wk), respectively; however, N/C in mice in
three short-term (all 4-wk) exposure studies: (Wu et al.,
2013) at 3 mg/m3 with s.c. OVA sensitization, (Kim et al.,
2013b) at 0.2-1.23 mg/m3 with dermal house dust mite
(hdm) sensitization, and (Sadakane et al., 2002) at
>12.3 mg/m3 with i.p. HDM sensitization b
Total IgG
High or Medium
Human: Decreased in a single study of exposed workers
(Avdin et al., 2013): 7 vr exposure at 0.264 mg/m3
Moderate
Based on decreased total
IgG in a high or medium
confidence long-term study
in adult workers exposed to
0.264 mg/m3, and a high or
medium confidence short-
term study in rats exposed
to >6.15 mg/m3. IgG
isoforms were affected in 2
of 3 low confidence short-
term mouse studies, but not
a low confidence study of
children at low levels
Animal: Decreased total IgG in rats (Sapmaz et al.,
2015): short-term exposure at >6.15 mg/m3
£
o
-J
Human: N/C in children at ~0.007-0.07 mg/m3 (Erdei et al.,
2003): unknown exposure duration (likely months-years)
Animal: IgGl (N/C in lgG2a) increased by FA alone, whereas FA
exacerbated lgG2a (N/C in IgGl) in atopic-prone mice (Kim
et al., 2013b): short-term exposure at 0.25, but not 1.2,
mg/m3; increased IgGl and lgG3, but decreased lgG2a and 2b,
in C57 mice (Jung et al., 2007): short-term exposure at
>6.15 mg/m3;
N/C in IgG Balb/c mice (Gil et al., 2008): short-term
exposure at <1 mg/m3
This document is a draft for review purposes only and does not constitute Agency policy.
1-137 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
FA-Specific
IgG
High or
Medium
Human: Slight (i.e., <10%) increase in a single study of adults
(Kim et al., 1999): years of exposure at 3.74 mg/m3
Moderate
Based on slight increases in
a high or medium
confidence long-term study
of adults at 3.74 mg/m3 and
increases in low confidence
studies of adults with long-
term exposure at <1 mg/m3,
but not with short-term
exposure at higher levels;
studies in children were not
identified
Animal: None
£
o
Human: Increased in two studies (Thrasher et al,, 1990;
Thrasher et al., 1987) and unclear in one study in which
5/55 subjects did have FA-IgG (DvkewicZ et al., 1991):
all three studies examined years of exposure at <0.1-<1.0
mg/m3; N/C in one study (Wantke et al., 2000): short-
term exposure at 0.265 mg/m3
Animal-. No change in guinea pigs with acute challenge (Lee
et al., 1984) at 2.5 or 4.9 mg/m3 after short-term exposure
to 7.4 or 12.3 mg/m3 (note: the study did not present
measures without formaldehyde exposure, and isotype was
unspecified)
Antigen-
Specific IgG
(does not
include FA-
specific Ig)
High or Medium
Human: None
Moderate (with inhaled
antigen)
Based on increased OVA-
lgGl in a high or medium
confidence short-term study
in guinea pigs at 0.31 mg/m3
with inhaled allergen, but
not a longer high or medium
confidence mouse study at
similar levels using injected
allergen. Similarly, a long-
term low confidence study
observed increased IgG
sensitization to airway
antigens in children,
whereas several low
confidence studies in mice
and rats suggest that IgG
sensitization does not occur
when antigen is injected.
Animal: Increased OVA-specific IgGl in guinea pigs (Riedel
et al., 1996): 5 d at 0.31 mg/m3 with inhaled OVA;
questionable decrease (i.e., effects were observed at 0.49, but
not 2.46, mg/m3) in OVA-lgGl and OVA-lgG3 in mice
(Fuiimaki et al., 2004b): 12 wks exposure with i.p. OVA
sensitization (N/C in OVA-lgG2)
£
o
-J
Human: Increased IgG against 2 bacterial pathogens by linear
regression in 3rd grade children with respiratory complaints
(Erdei et al., 2003): <0.1 mg/m3, unknown exposure
duration (likely years, home measures)
Animal: N/C in OVA-IgG or Der f-lgGl in mice (Wll et al.,
2013; Gu et al., 2008; Sadakane et al., 2002): up
to 5 wk exposures at 0.123-3 mg/m3 or >12.3 mg/m3 b; N/C in
IgG specific to vaccine antigens in rats (Holmstrom et al.,
1989a): 22 months exposure at 15.5 mg/m3. In all cases, s.c.
or i.p. exposure was used for sensitization
Circulating
Stress
Hormones
High or
Medium
Human: None
Increased at 3 mg/m3
formaldehyde in a study in
rats with short-term, but
not acute, exposure
Slight
Animal: Increased corticosterone in rats with short-term, but
not acute, exposure (Sorg et al., 2001a): at ~3 mg/m3
£
o
-J
Human: None
No evidence to evaluate
Animal: None
Modifications in other non-Respiratory Tissues
Oxidative
stress
in
nonrespira-
tory tissues
High or Medium
Human: Increased marker of lipid peroxidation in adult serum
lymphocytes (Bono et al., 2010): likely months-to-years
exposure (assumed) at >0.066 mg/m3; Increased F2-
Isoprostanes (suggested as the best in vivo biomarker of lipid
peroxidation) in urine (Romanazzi et al., 2013): 0.21
mg/m3 chronic occupational exposure (indirect for effects in
blood), although smoking and formaldehyde were not
additive, both were independently associated with
Two studies in adults
indicate elevated oxidative
stress markers at >0.066-
0.21 mg/m3 with long-term
exposure. Given the
uncertainty regarding use of
urine to reflect associations
in blood, one study
Moderate
This document is a draft for review purposes only and does not constitute Agency policy.
1-138 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
ROS—Note: serum and urine IsoP measures are often
correlated [e.g., (RodrigO et al., 2007)1, suggesting that
urine levels may reflect similar serum changes
contributes as indirect
evidence
Animal: None
O
Human: Increased oxidative stress biomarkers (F2-
Isoprostanes; malondialdehyde) in urine (Bellisario et al.,
2016): work-shift exposure at ~0.034 mg/m3 (indirect for
effects in blood; responses likely reflect short-term exposure)
Several studies in three
species suggest increases in
markers of oxidative stress
with acute or short-term
exposure, even at
formaldehyde levels <1
mg/m3; it is not clear
whether and to what extent
this persists with long-term
exposure
Animal: Increased oxidative stress markers in mice (Ye et
al., 2013b; Matsuoka et al., 2010): acute or short-
term exposures at as low as 0.12 mg/m3; increased oxidative
stress markers and protein indicators in rats (Aydin et al.,
2014; Im et al., 2006): short-term exposure at 6.48-12.3
mg/m3, although one study with a longer exposure (10 wk)
observed a decrease in MDA in rats (Katsnelson et al.,
2013): at 12.8 mg/m3; other indicators in rodents included
decreased gsh (Katsnelson et al., 2013; Ye et al.,
2013b) and increased NO and SOD (Matsuoka et al.,
2010): short-term exposure at >1 mg/m3
Cell counts
in immune
tissues (not
including
bone
marrow)
High or Medium
Human: None
Suppression of CD8+ T cells
in immune tissues
(e.g., spleen) is indicated in
one 8-wk mouse study, with
indirect support from a
second short-term mouse
study, at around 2 mg/m3;
effects on CD4+/CD8+ ratio
were mixed across 2
subchronic mouse studies
Moderate (for
n!/ CD8+ T cell
response in
spleen and
thymus)
Slight
NK cells (^ at
low level; \|/ at
high level)
Indeterminate
for other cell
counts
Animal: Decreased CD8+ T cells and increased CD4+/CD8+ ratio
in both thymus (immature immune cells) and spleen (mature
immune cells) in male mice (Ma et al., 2020): Eight weeks
of exposure at 2 mg/m3; No change in splenic CD4+/CD8+ ratio
in female mice (Fuiimaki et al., 2004b): 12 wk at up to
2.46 mg/m3; Increased splenic regulatory T cells (subset of
CD4+) and indirect markers for suppression of effector T cell
(CD8+) activity in female mice (Park et al., 2020): short-
term exposure at >1.38 mg/m3
£
o
Human: None
Multiple short-term mouse
studies suggest that overall
splenic cell T and B cells are
unchanged; however, 2
studies suggest that NK cells
may be affected (1 study
showed NK cells were
stimulated at low
formaldehyde levels, and
another that high levels are
inhibitory/toxic)
Animal: N/C in tissue weight, total cellularity or T or B cell
counts in mice (Kim et al., 2013a; Gu et al., 2008;
Dean et al., 1984); altered NK cell number and function
was noted in mice, with one study showing decreases (Kim et
al., 2013a): 2-3 wk at 12.3 mg/m3, and another showing
increases (Gu et al., 2008): 5 wk at up to 0.12 mg/m3, and
a third showing N/C in lymphocyte proliferation, functional
parameters, IgM production, or NK cytotoxicity (Dean et al.,
1984): 3 wk at 18.5 mg/m3
Systemic
indicators
of altered
High or
Medium
Human: None
No evidence to evaluate
Indeterminate
Animal: None
This document is a draft for review purposes only and does not constitute Agency policy.
1-139 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Endpoint-specific findings and confidence
Summary of evidence
Conclusion
immune
function
O
Human: Increased autoantibodies in adults (Thrasher et
al., 1990): long-term exposure at 0.06-0.95 mg/m3
1 study in adults suggests
that autoantibodies are
elevated with low-level,
long-term exposure;
somewhat in contrast, one
mouse study suggests
short-term high-level
exposure improves host
response to bacteria
Animal: Improved cell-mediated immune response to bacteria
challenge, but N/C against tumor challenge or delayed-type
hypersensitivity response in mice (Dean et al., 1984): 3
wk exposure at 18.5 mg/m3 (Note: N/C in vitro measures of
immune cell function in the same study)
aSeveral studies examining the lineage and maturity of immune and non-immune cells in the bone marrow and other systemic
tissues (e.g., blood; spleen) are not discussed in this section. Although it is possible that differences in the maturation
phenotype of cells could indirectly contribute to the immune changes of interest to this section, such alterations would be
expected to cause functional or other detectable changes in more apical mechanistic events relevant to immune responses in
the respiratory system. Thus, this discussion focuses on those mechanistic events considered more directly relevant to these
POE outcomes. Please see Section 1.3.3 for a discussion of these cell lineage and maturation markers in the context of
lymphohematopoietic cancer MOA.
bAs the challenge stimuli used in the formaldehyde studies included allergens as well as nonimmunological stimuli, and because
most experiments did not attempt to delineate the specifics of the functional changes, "airway hyperresponsiveness" or
"hyperresponsive airways" encompasses any of a range of possible airway features: hyperreactivity (exaggerated response),
hypersensitivity (lower dose to elicit response), altered ventilatory parameters (e.g., maximal response, resistance), recovery
(longevity of response), or others.
cReported as 0.5% formaldehyde solution; concentration assumed to be >12.3 mg/m3 (Sadakane et al., 2002).
Table 1-23. Summary of changes in cell counts and soluble immunological
factors in the blood following formaldehyde exposure
Endpoint
No changes observed
(above dashed line= human studies;
below dashed line= animal studies;
high or medium confidence = *and bold)
Significant3 increases (1^) or decreases (^)
(above dashed line= human studies; below
dashed line= animal studies;
high or medium confidence = *and bold)
Conclusion
(notes)
mg/m3
Length13
References (details)
mg/m3
Length13
References (details)
| White blood cells (WBCs) |
Total
WBCs
0.87
0.25
0.018
Years
Years
Yearsc
(Lvapina et al.,
2004)*
(Avdin et al., 2013)*
(Erdei et al., 2003)
(asthmatic children)
1.6
si N/Ae
(<1)
\|/<0.29
Years
(same
cohort)
Yr vs.
Mo
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
2010)*d
(Bassig et al., 2016)*
(Thrasher et al.,
1990)
(Kuo et al., 1997)
Moderate si in
WBCs e
>9.23
8 wk
(Morgan et al.,
2017) (mice)*
>2.46f
(indirect)
si 0.5-3
Short
Short
(Rager et al., 2014)*
(rats)
(Zhang et al., 2013b)
(mice)
Granulocytes
All
si 1.6
Years
(same
cohort)
(Hosgood et al.,
2013)*; (Zhang et al.,
2010)*d
(Bassig et al., 2016)*
Slight si in
granulocytes
(appears to
reflect
potential
This document is a draft for review purposes only and does not constitute Agency policy.
1-140 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
No changes observed
(above dashed line= human studies;
below dashed line= animal studies;
high or medium confidence = *and bold)
Significant3 increases (1^) or decreases (^)
(above dashed line= human studies; below
dashed line= animal studies;
high or medium confidence = *and bold)
Conclusion
(notes)
changes in
neutrophils at
higher
concentrations
with short-
term or longer
exposure)
mg/m3
Length13
References (details)
mg/m3
Length13
References (details)
18.5
Short
(Dean et al., 1984)
(mice)h
Neutr
ophils
0.25
<0.29
0.018
Years
Years
Yearsc
(Avdin et al., 2013)*
(Kuo et al., 1997)
(Erdei et al., 2003)
(asthmatic children)
>1/0.87
Years
(Lvapina et al.,
2004)* (i.e., function, in
workers with URT
dysfunction)
>9.23
0.5-3
8 wk
Short
(Morgan et al.,
2017) (mice) (mice)
(Zhang et al., 2013b)
(mice)
viz 13
Short
(Katsnelson et al.,
2013) (rats)
Eosino
phils
<0.29
0.018
Years
Yearsc
(Kuo et al., 1997)
(Erdei et al., 2003)
(asthmatic children)
>9.23
8 wk
(Morgan et al.,
2017) (mice) (mice)
Baso
phils
<0.29
Years
(Kuo et al., 1997)
No animal studies identified
Lymphocytes
All
0.2 & 0.8
N/Ae
(<1) 0.51
<0.29
0.018
Months
Yr vs.
Mo
Weeks
Years
Yearsc
(Jia et al., 2014)*
(Thrasher et al.,
•4, 1.6
t 0.25
Years
(same
cohort)
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
Indeterminate
(multiple
changes noted,
but pattern is
indiscernible)
1990)(Ying et al.,
2010)*d
1999)
(Kuo et al., 1997)
(Erdei et al., 2003)
(asthmatic children)
(Bassig et al., 2016)*
(Avdin et al., 2013)*
18.5
>9.23
Short
8 wk
(Dean et al., 1984)
(mice)h
(Morgan et al.,
2017)* (mice)
¦f 13
viz 0.5-3
Short
Short
(Katsnelson et al.,
2013) (rats)
(Zhang et al., 2013b)
(mice)
This document is a draft for review purposes only and does not constitute Agency policy.
1-141 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
No changes observed
(above dashed line= human studies;
below dashed line= animal studies;
high or medium confidence = *and bold)
Significant3 increases (1^) or decreases (4^)
(above dashed line= human studies; below
dashed line= animal studies;
high or medium confidence = *and bold)
Conclusion
Endpoint
mg/m3
Length13
References (details)
mg/m3
Length13
References (details)
(notes)
1.6
Years
(same
cohort)
Years
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
t 0.99
t0.2&
0.8
¦f N/Ae
(<1)
/T* 0.51
>1/0.47
>1/0.36
Months
Months
Yr vs.
Mo
Weeks
Years
Years
(Ye et al., 2005)* (peak
levels up to 1.69 mg/m3)
Moderate for
altered number
of B cells
(direction of
change may
differ by
exposure levels
or duration)
B Cells
0.25
0.09-0.7
2010)*
(Bassig et al., 2016)*
(Avdin et al., 2013)*
(Thrasher et al.,
1987)
(Jia et al., 2014)*
(Thrasher et al.,
1990)(Ying et al.,
1999)
(Costa et al., 2019)*
(peak levels to 3.94
mg/m3)
(Costa et al., 2013)*
(peak levels to 0.69
mg/m3)
No animal studies identified
T Cells
(Total)
0.2-0.8
N/Ae
(<1)
Months
Yr vs.
Mo
(Jia et al., 2014)*
(Thrasher et al.,
1990)
•4, 1.6
>1/0.99
t 0.36
t 0.25
>1/0.9
>1,0.51
>1/ >0.09
Years
(same
cohort)
Months
Years
Years
Years
Weeks
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
2010)*d
(Bassig et al., 2016)*
(Ye et al., 2005)*
(peak levels to 1.69
mg/m3)
(Costa et al., 2013)*
(peak levels to 0.69
mg/m3)
(Avdin et al., 2013)*
(Jakab et al., 2010)
(Ying et al., 1999)
(Thrasher et al.,
1987) (levels up to 0.68
mg/m3)
Slight for
altered total T
cells
(mixed results
suggest dose-
dependence,
with \1/ at
higher levels;
possible 'T* at
low levels, with
longer
duration)
¦f 7.4
Short
(Sandikci et al.,
2007a, b) (rats)
This document is a draft for review purposes only and does not constitute Agency policy.
1-142 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint
No changes observed
(above dashed line= human studies;
below dashed line= animal studies;
high or medium confidence = *and bold)
mg/m3
Length13
1.6
Years
(same
cohort)
0.99
Months
0.47
Years
0.25
Years
0.2-0.8
Months
References (details)
Significant3 increases (1^) or decreases (4^)
(above dashed line= human studies; below
dashed line= animal studies;
high or medium confidence = *and bold)
mg/m3
Length1
References (details)
Conclusion
(notes)
(Hosgood et al..
2013)* (note: ¦i, Treg
cells)
(Zhang et al.. 2010)*
t 0.36
n|,0.51
Years
Weeks
(Costa et al.. 2013)*
(peak levels to 0.69
mg/m3)
(Ying et al.. 1999)
T Cells
(CD4+)
(Bassig et al.. 2016)*
(Ye et al.. 2005)*
(peak levels up to 1.69
mg/m3)
(Costa et al.. 2019)*
(peak levels to 3.94
mg/m3)
(Aydin et al.. 2013)*
(Jia et al.. 2014)*
Indeterminate
(mostly N/C,
but variable
and,
considering
also studies of
spleen (above),
suggests
effects might
exist for certain
subsets of CD4
cells)
No animal studies identified
0.36
0.25
0.2-0.8
T Cells
(CD8+)
Years
Years
Months
(Costa et al.. 2013)*
(peak levels to 0.69
mg/m3)
(Aydin et al.. 2013)*
(Jia et al.. 2014)*
N/C CD4/CD8 ratio in these 3 studies (or in
(Thrasher et al.. 1990)comparing
durations)
•i, 1.6
Years
(same
cohort)
Months
>1/0.99
Weeks
n|,0.51
Years
t 0.47
(Hosgood et al..
2013)*; (Zhang et al..
2010)*d
(Bassig et al.. 2016)*
(particularly memory cells)
(Ye et al.. 2005)*
(peak levels to 1.69
mg/m3)
(Ying et al.. 1999)
(Costa et al.. 2019)*
Moderate
\1/ CD8 and 'T*
CD4/CD8 ratio
(likely dose-
dependence,
as consistent
observations
are at higher
levels)
(peak levels to 3.94
mg/m3)
CD4/CD8 ratio in all but one of these studies
No animal studies identified
NK
Cells
•i, 1.6
Years
(same
cohort)
>1/0.36
Years
t 0.25
Years
t0.2
Months
N/C at 0.8
(Hosgood et al..
2013)*; (Zhang et al..
2010)*d; (Bassig et
al.. 2016)*
(Costa et al.. 2013)*
(peak levels to 0.69
mg/m3)
(Aydin et al.. 2013)*
(Jia et al.. 2014)*
Slight for
altered number
of NK cells
(mixed results
suggest dose-
dependence
like total T
cells)
No animal studies identified
This document is a draft for review purposes only and does not constitute Agency policy.
1-143 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
No changes observed
(above dashed line= human studies;
below dashed line= animal studies;
high or medium confidence = *and bold)
Significant3 increases (1^) or decreases (^)
(above dashed line= human studies; below
dashed line= animal studies;
high or medium confidence = *and bold)
Conclusion
(notes)
Endpoint
mg/m3
Length13
References (details)
mg/m3
Length13
References (details)
Mono
cytes
1.6
0.25
Years
(same
cohort)
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
¦f 0.018
Yearsc
(Erdei et al., 2003)
(asthmatic children)
Indeterminate
(data suggest
N/C, at least in
human adults)
2010)*d
(Bassig et al., 2016)*
(Aydin et al., 2013)*
>9.23
8 wk
(Morgan et al.,
2017) (mice)
n!/ 18.5
\1/ 0.5, not
3
Short
Short
(Dean et al., 1984)
(mice)
(Zhang et al., 2013b)
(mice)
Red Blood
Cells
0.25
<0.29
0.018
Years
Years
Yearsc
(Avdin et al., 2013)*
(Kuo et al., 1997)
(Erdei et al., 2003)
(asthmatic children)
1.6
>1/0.87
Years
Years
(Hosgood et al.,
2013)*; (Zhang et al.,
2010)*d
(Lvapina et al.,
2004)* (association with
duration)
Moderate in
RBCs'
(suggests dose-
and duration-
dependence)
>9.23
8 wk
(Morgan et al.,
2017) (mice)
n!/ 0.5-3
Short
(Zhang et al., 2013b)
(mice)
Platelets
0.87
<0.29
0.018
Years
Years
Yearsc
(Lvapina et al.,
2004)* (Kuo et al.,
1.6
Years
(same
cohort)
(Hosgood et al.,
2013)*; (Zhang et al.,
Slight \1/ in
platelets'
(possible dose-
dependence as
noted above)
1997)
(Erdei et al., 2003)
(asthmatic children)
2010)*d
(Bassig et al., 2016)*
>9.23
8 wk
(Morgan et al.,
2017) (mice)
¦f 0.5-3
Short
(Zhang et al., 2013b)
(mice)
| Secreted factors and immune markers |
Primarily Thl-related
TNF-a
1.8
0.2-0.8
Years
Months
(Seow et al., 2015)*
(peak levels to 6.9 mg/m3)
(Jia et al., 2014)*
t 0.25
Years
(Avdin et al., 2013)*
Slight 'T*
TNF-a and C3
No animal studies identified
Compl
ement
0.25
Years
(Avdin et al., 2013)*
(i.e., C3, C4)
t 6.15
Short
(Sapmaz et al.,
2015)* (rats; i.e., C3)
IFN-v
4, 0.8
Months
(Jia et al., 2014)*
Moderate
-------
Toxicological Review of Formaldehyde—Inhalation
No changes observed
Significant3 increases (1^) or decreases (4^)
(above dashed line= human studies;
(above dashed line= human studies; below
below dashed line= animal studies;
dashed line= animal studies;
high or medium confidence = *and bold)
high or medium confidence = *and bold)
Conclusion
Endpoint
mg/m3
Length13
References (details)
mg/m3
Length13
References (details)
(notes)
IL-10
1.8
"t 0.2-0.8
Years
Months
(Seow et al., 2015)*d
(i.e., using less strict 20%
FDR)
(Jia et al., 2014)*
Slight
IL-10
(suggests dose-
dependence
like total T
No animal studies identified
cells)
No human studies identified
Indeterminate
IL-6
0.12
Acute
(Matsuoka et al.,
2010) (mice)
IL-6
V)
+¦•
C
m
¦*->
o
re
CXCL1
1 and
CCL17
1.8
Years
(Seow et al., 2015)*
(i.e., using stringent 10%
FDR)
Slight \1/
(chemo-
attractants
for neutrophils:
IL-8, and
+->
m
No animal studies identified
E
(V
-C
IL-8
•i, 0.2-0.8
Months
(Jia et al., 2014)*
lymphocytes:
Cxclll, Ccll7)
u
No animal studies identified
Tal
and
IL-2R
-T N/Ae
(<1)
Yr vs.
Mo
(Thrasher et al.,
1990) (antigen reactivity
Indeterminate
(data suggest
N/C in B cell
activation
markers)
a>
.c
4-»
o
No animal studies identified
markers)
CD27
and
1.6
Years
(Bassig et al., 2016)*
(B cell activation markers)
CD30
No animal studies identified
Abbreviations and definitions: Derf = Dermatophagoides farina (house dust mite) and OVA= ovalbumin (major
protein of chicken egg whites): both immunogenic materials used to stimulate an allergy-like response;
FDR = false discovery rate; N/C = no change; Treg = T regulatory cells, a subset of helper T cells; short = short-term.
Notes: Formaldehyde concentrations typically reflect average or median levels in human studies (e.g., when effects
were not observed); Gray shading = no data meeting the inclusion criteria were available (see Appendix A.5.6);
one study observing increased substance P and related changes in the serum (Fuiimaki et al., 2004b) is presented
in the context of changes in the respiratory system (see Section 1.2.2).
Primarily, this reflects reporting of a statistically significant change; in rare instances where a p-value was not
given, changes are indicated if the authors discussed the change as a significant effect.
bHuman study exposure durations are indicated as "years," "months," "weeks," "acute," or "Yr vs. Mo" (see
footnote d) and defined based on the anticipated exposure duration for the majority of the exposed
population(s); these durations are interpreted to approximate animal study exposure durations of chronic
(>1 year), subchronic (several months), short term ("Short" in table; <30 days), and acute (1 day or less).
cErdei et al. (2003) studied 9- to 11-year-old students symptomatic with respiratory issues, so duration of exposure
was presumed to be years in schools (average exposure concentration is indicated).
dThe differences in lymphocyte subset levels between exposed and unexposed workers reported by Zhang et al.
(2010) were challenged by Mundt et al. (2017) in a reanalysis who did not find evidence of an exposure-response
trend within the exposed group, although the difference between unexposed and exposed subjects was
reconfirmed. The critique by Mundt was responded to in a letter to the editor by the study investigators who
explained that the study was not designed to provide a range of exposures wide enough to evaluate exposure-
response relationships given the expected effect size and sample size in the study (Rothman et al., 2017).
This document is a draft for review purposes only and does not constitute Agency policy.
1-145 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
eThe exposure level is, in general, considered not applicable (N/A), as the comparison presented by Thrasher et al.
(1990) reflected differences in exposure duration (i.e., years of exposure [Yr], as compared to weeks or months
[Mo] of exposure), but there appeared to be minimal differences in concentration from the controls,
the studies by Rager et al. (2014; 2013) were molecular studies (e.g., miRNA) interpreted as high or medium
confidence that provide some indirect evidence of inflammatory changes.
gThis finding (decreased total WBCs) is supported by three studies in humans based on an evaluation by NRC
(2014b): [(Tone et al., 2007; Cheng et al., 2004; Tang and Zhang, 2003)1, but these studies were not evaluated in
this analysis (i.e., they were not indexed in any of the searched databases); additionally, this finding is supported
by a study in mice (Yu et al., 2014) and a study in rats (Brondeau et al., 1990), which are not included above as
they only tested excessive formaldehyde levels (i.e., >20 mg/m3).
hThe authors indicated no changes in "WBC differentials" other than decreased monocytes, but further details NR
(Dean et al., 1984). This test was assumed to include basic granulocyte and lymphocyte counts.
This finding (decreased erythrocytes) is supported by one study in humans based on an evaluation by the NRC
(2014b); [(Yang, 2007)1, but this study was not evaluated in this analysis.
JThis finding (decreased platelets) is supported by two studies in humans based on an evaluation by NRC (2014b);
(Tong et al., 2007; Yang, 2007), but these studies were not evaluated in this analysis. The finding is also
supported by a mouse study testing excessive formaldehyde levels (Yu et al., 2014).
Integrated Summary of Evidence on Immune-mediated Respiratory Conditions, Focusing on
Allergies and Asthma
The general population studies in children and adults provide moderate evidence of an
association between formaldehyde exposure and prevalence of rhinitis or rhinoconjunctivitis, with
a relative risk of approximately 1.2 for formaldehyde exposures of around 0.04-0.06 mg/m3.
Although the effect size is small, these are relatively common conditions. The observation of an
increase in the magnitude of the odds ratio with increasing severity of rhinitis provides coherence
and greater certainty in the evidence fYon etal.. 20191. A stronger association (two-fold risk) was
seen in the only study of eczema and a 3-fold odds of experiencing allergy-like symptoms involving
the eyes, nose and skin within the past week was observed for students exposed to formaldehyde
concentrations in classrooms >0.035 mg/m3 (median 0.045 mg/m3) compared to <0.035 mg/m3
(OR 3.23, 95% CI 1.31, 8.00).
The available general population and occupational studies also provide a moderate level of
evidence of an association between formaldehyde exposure and prevalence of current asthma, as
determined by symptoms or medication use in the past 12 months for asthma in studies with
exposures above 0.05 mg/m3 Notably, a study using an intervention to increase ventilation in
participants' residences observed a decrease of 14-20% in asthma symptoms and medical care
needed during the following year among asthmatic children associated with a 50% reduction in
formaldehyde concentration fLaioie etal.. 20141. However, confounding by coexposures cannot be
excluded. Geometric mean formaldehyde concentrations at baseline were 0.035 mg/m3 and
0.057 mg/m3 in fall/winter and summer, respectively. The two studies examining asthma control or
severity among children with asthma suggest associations may be seen at lower exposures
(e.g., 0.04 mg/m3) in this potentially susceptible population. Relatively strong associations were
seen in studies examining prevalence of current asthma in relation to higher levels of formaldehyde
exposure in occupational settings (exposures above 0.10 mg/m3).
This document is a draft for review purposes only and does not constitute Agency policy.
1-146 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
Sensitivity may also be increased by other attributes as well disease severity. Although
associations with either eczema, prevalence of asthma, or asthma control were either increased or
decreased by a positive atopy status in studies of adults or children, studies in allergen-sensitized
animals suggest that atopy may increase sensitivity to formaldehyde-related asthma endpoints. In
addition, associations with IgE levels or prevalence of asthma symptoms were stronger among
groups exposed to environmental tobacco smoke, although inconsistencies by lifestage were
reported. Relatively strong associations were seen in studies examining prevalence of current
asthma in relation to higher levels of formaldehyde exposure in occupational settings (exposures
above 0.10 mg/m3). Mechanistic studies in animals indicate that formaldehyde exposure can
induce bronchoconstriction with and without allergen sensitization. This heightened
bronchoconstriction response may be due to a combination of increased tachykinins, increased Th2
cytokines and antibodies, and eosinophil recruitment and activation in the lung. Mechanistic
studies of respiratory tissues and the blood provide consistent evidence that formaldehyde
exposure can stimulate a number of immunological and neurological processes that may drive
asthmatic responses; however, a molecular understanding of how formaldehyde exposure favors
asthmatic Th2 responses has not been experimentally established. Separately, the possibility that
formaldehyde exposure might increase the risk or severity of respiratory infections, particularly in
young children, has not been well studied.
Overall, based primarily on a moderate level of human evidence supporting an association
from the available epidemiology studies, with corresponding slight evidence for an effect in animals
based on mechanistic studies in animals supporting biological plausibility, the evidence indicates
that inhalation of formaldehyde likely causes an increased risk of prevalent allergic conditions and
prevalent asthma symptoms, as well as decreased control of asthma symptoms, given appropriate
exposure circumstances (see Table 1-24). The primary basis for this conclusion involves studies of
occupational settings (>0.1 mg/m3) and population studies where formaldehyde concentrations
measured in schools and homes averaged between 0.03 and <0.1 mg/m3.
Table 1-24. Evidence integration summary for effects on immune-mediated
conditions, including allergies and asthma
Evidence
Evidence judgment
Hazard determination
Allergic Conditions
Human
Moderate fo/" Allergic Conditions, based on:
Human health effect studies:
Small elevated risks in five out of six high and medium confidence studies of
prevalence of rhinitis, conjunctivitis, and eczema among adults and children
in residential and school settings with exposures in the range of
0.04-0.06 mg/m3 formaldehyde. Very low formaldehyde concentrations
were measured in the one insensitive null study.
Biological Plausibility (both conditions)\ Studies in humans do not provide
robust or moderate evidence for mechanistic events that clearly support the
The evidence indicates that
inhalation of formaldehyde
likely increases the prevalence
of allergic conditions in humans,
given the appropriate exposure
circumstances3
This judgments is primarily
based on studies of occupational
This document is a draft for review purposes only and does not constitute Agency policy.
1-147 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
development of asthma, although effects in the blood, such as cytokine, cell,
and antibody changes, might contribute
settings (>0.1 mg/m3) and
population studies where mean
formaldehyde concentrations
measured in schools and homes
were between 0.03 and
0.1 mg/m3
Potential Susceptibilities:
Variation in sensitivity is
anticipated depending on
respiratory health, physiologic
changes during pregnancy, age,
and exposure to tobacco smoke
Animal
Sliaht for Immune-Mediated Respiratory Effects based on:
Biological Plausibility.
Robust evidence for mechanistic events exists in relation to
formaldehyde-induced augmentation of responses to allergens and airway
bronchoconstrictor effects in animal models. Although several events
typically associated with asthma were not corroborated (i.e., slight or
inadequate evidence exists for these events), moderate evidence for
mechanistic events exists for stimulation by formaldehyde of important
immunological and neurological processes. These include airway eosinophil
increases and other inflammatory changes in the airways and systemic
circulation that can be reasonably associated with effects on airway
hyperreactivity or other responses relevant to the development of allergic
conditions and, potentially, asthma.
Animal health effect studies: Experimental animal models are generally
considered to be unable to reproduce the overt manifestations of allergic
conditions and are not interpreted to provide direct support.
Other
inferences
• Relevance to humans: The most relevant mechanistic findings in animals
involve neurological and immunological constituents present in both
human and rodent airways.
• MOA: Several incomplete MOAs involving airway inflammatory changes
are considered likely to be involved.
Prevalence of Current Asthma
Humans
Moderate for Asthma, based on:
Human health effect studies:
• Elevated risks in eight medium confidence studies of prevalence of
current asthma in adults and children, change after an intervention to
reduce exposure, or reduced symptom control in children in residential
settings including homes with >0.05 mg/m3 formaldehyde; greater
susceptibility among children
• Inconsistencies in study results appear to be explained by exposure levels.
No elevated risk of current asthma in six high and medium confidence
studies with relatively low exposures (<0.05 mg/m3), but associations with
adequacy of asthma control were observed in one study at this lower
exposure level
• Strongly elevated risks in three medium confidence studies in
occupational settings with exposures from 0.100 to >0.500 mg/m3
Biological Plausibility (both conditions)\ Studies in humans do not provide
robust or moderate evidence for mechanistic events that clearly support the
development of asthma, although effects in the blood, such as cytokine, cell,
and antibody changes, might contribute
The evidence indicates that
inhalation of formaldehyde
likely increases the prevalence
of asthma symptoms in humans,
as well as decreased control of
asthma symptoms, given
appropriate exposure
circumstances3
This judgment is primarily based
on studies of occupational
settings (>0.1 mg/m3) and
population studies where mean
formaldehyde concentrations
measured in schools and homes
were between 0.03 and
0.1 mg/m3
Potential Susceptibilities:
Variation in sensitivity is
anticipated depending on
respiratory health, physiologic
changes during pregnancy, age,
and exposure to tobacco smoke
Animals
Sliaht for Immune-Mediated Respiratory Effects based on:
Biological Plausibility.
In the same way the available mechanistic data are interpreted to provide
slight animal evidence supporting the development of allergic conditions in
humans, this evidence provides slight evidence supportive of asthma.
Animal health effect studies: Experimental animal models are generally
considered to be unable to reproduce the overt manifestations of asthma
and are not interpreted to provide direct support.
This document is a draft for review purposes only and does not constitute Agency policy.
1-148 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
• Relevance to humans: For the animal mechanistic data, while several
events (e.g., amplified bronchoconstriction; eosinophil increases) have an
unclear direct linkage to complex human diseases like asthma, these
Other
findings inform the potential for exposure to result in changes to relevant
Inferences
neurological and immunological constituents present in both human and
rodent airways.
• MOA: Several incomplete MOAs involving airway inflammatory changes
are considered likely to be involved.
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2
1.2.4. Respiratory Tract Pathology
This section describes research on formaldehyde inhalation and pathology endpoints in the
respiratory system. Numerous well-conducted experimental animal studies consistently
demonstrate concentration- and, to a lesser extent, duration-dependent URT hyperplasia and
metaplasia after formaldehyde exposure. Supporting these observations, a set of four studies in
formaldehyde-exposed workers provides consistent findings of an elevated prevalence of nasal
lesions such as hyperplasia and metaplasia. The workers were generally exposed to lower levels of
formaldehyde than those eliciting changes in experimental animals. While the evidence for both of
these nonneoplastic lesions indicates that formaldehyde exposure changes the morphology and
function of the URT tissue, the evidence for metaplasia, in particular, is considered to be the best
representation of a potential health hazard.
In the URT, both hyperplasia and metaplasia are adaptive tissue responses. These cellular
responses help reduce the impact of stressors by changing the structure or function of the locally
affected tissue (Harkema etal.. 20131. Hyperplasia, generally a response to cell injury, involves an
increase in the population of resident cells that results in additional cell layers noticeable by
histology, whereas metaplasia, which typically occurs following prolonged or repeated insults,
results in the replacement of one differentiated cell type with another more resilient cell type
f Harkema etal.. 20131. While hyperplasia and metaplasia may also be relevant, but not necessary,
to the development of cancer (see Section 1.2.5), they are, by themselves, nonneoplastic lesions.
Importantly, metaplasia results in a hardened, drier, and nonciliated skin-like layer (Tomashefski.
20081. Along with the acquisition of a protective, barrier-type phenotype, this metaplastic change
causes a loss of normal tissue function, including reduced mucous secretion and ciliary clearance.
Thus, this loss of normal function is judged to be an adverse outcome in and of itself
(i.e., independent from its potential role in progression to cancer). As an interpretation regarding
adversity is less clear for hyperplasia, this discussion emphasizes the data on squamous metaplasia.
Both hyperplasia and metaplasia are typically associated with cellular proliferation
(Harkema etal.. 2013). As compared to transient increases in cell number, sustained cell
proliferation is required for the formation of hyperplasia. This type of change can be precipitated
by damage to the nasal epithelium, which is evaluated histologically by measures of, for example,
cell loss or necrosis, epithelial degeneration, and erosions. Relatedly, squamous metaplasia is an
This document is a draft for review purposes only and does not constitute Agency policy.
1-149 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
adaptive response to continued toxic insult that involves cellular substitution. Thus, it is useful to
consider these cellular damage-related endpoints in the context of hyperplasia and metaplasia.
While evaluations of necrosis- and cytotoxicity-related pathology are informative to this section,
these endpoints were generally inconsistently measured or poorly reported across the available
studies and are therefore are only summarily discussed. Although hyperplasia and metaplasia
might have been underevaluated or underreported for similar reasons (e.g., most studies focus on
carcinogenic lesions), the potential development of these lesions appears to have been considered
and documented in nearly all the long-term formaldehyde inhalation studies examining URT
histopathology.
Studies that evaluated related outcomes, such as mucociliary flow rates, cellular
proliferation counts based on DNA labeling, and mucosal swelling, are summarized in
Appendix A.5.6). These types of effects were generally evaluated after acute or short-term
exposure and typically represent immediate response repair mechanisms rather than tissue
remodeling (e.g., hyperplasia, metaplasia), the latter of which is often a consequence of longer-term
exposure or sustained injury. Accordingly, those related outcomes are interpreted to be most
relevant to the mechanistic progression of the more overt URT lesions considered in this section,
and they are discussed as such in the MOA analysis. Overall, mechanistic insights from the human
and animal data indicate a clear role for altered mucociliary function or cellular proliferation in the
occurrence of the more overt lesions. Consistent with some of the animal health effect studies,
these mechanistic data also suggest that concentration is likely to be more of a driver of these
effects than duration (noting that duration still contributes).
Given the large number of long-term exposure studies with information on URT pathology
and the focus of the assessment on the effects of lifetime formaldehyde exposure, this section
generally focuses on animal studies of subchronic or chronic exposure, and on human studies of
occupational exposure where exposed employees were generally employed for longer than 5 years.
Exceptions include discussion of shorter-term studies that might inform the potential for
relationships between lesion types and studies specifically considering differences in the exposure
paradigm (e.g., intermittent versus constant exposures) for lesion induction. Dysplastic lesions and
other evidence of carcinogenicity, which are examined in many of the same studies addressed in
this section, are discussed in Section 1.2.5.
Overall, the strength of the evidence for hyperplasia and squamous metaplasia includes
robust evidence from animal studies and moderate human evidence from observational
epidemiology studies, and strong support for a plausible MOA based largely on mechanistic
evidence in animals (supported by more limited, coherent findings in human mechanistic studies).
Therefore, the evidence demonstrates that inhalation of formaldehyde causes respiratory tract
pathology in humans given the appropriate exposure circumstances. The primary support for this
conclusion is based on rat bioassays of chronic exposure, which consistently observed squamous
metaplasia at formaldehyde exposure levels >2.5 mg/m3.
This document is a draft for review purposes only and does not constitute Agency policy.
1-150 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Literature Search and Screening Strategy
Systematic literature searches were conducted separately to identify health effect studies in
humans and in experimental animals. The identification of relevant studies on respiratory tract
pathology in humans and laboratory animals included literature searches in PubMed, Web of
Science, and ToxNet through September 2016 (see Appendix A.5.5 for search details), and a
systematic evidence map updating the literature through 2021 (see Appendix F). Primary research
studies using measurements of formaldehyde in workplace air and histopathological endpoints in
nasal tissue in humans were identified and included. Studies reporting primary research on
formaldehyde exposure and measures of respiratory pathology in several animal species were
identified and included. As stated above, subchronic and chronic exposure durations in either
animals or humans were preferred. The mechanistic evidence informing this health effect was
identified and evaluated as part of the overarching review of mechanistic data relevant to potential
respiratory health effects (see Appendix A.5.6 for details). The bibliographic databases, search
terms, and specific strategies used to search them are provided in Appendix A.5.5 and A.5.6, as are
the specific PECO criteria. Literature flow diagrams summarize the results of the sorting process
using these criteria and indicate the number of studies that were selected for consideration through
2016 (see Appendix F for the identification of newer studies through 2021). The relevant health
effect studies in animals and humans, and the mechanistic data informative to respiratory tract
pathology, were evaluated to interpret the quality and relevance of the study results in regard to
hazard identification (see Appendix A.5.5 and A.5.6).
Methodological Issues Considered in Evaluating Studies
Cross-sectional studies among occupational cohorts were likely influenced by the selection
of the workforce in favor of individuals less responsive to the irritant properties of formaldehyde,
with resulting bias toward null results. Despite this methodological limitation and subsequent
reduction in sensitivity, most of the studies observed increases in histopathological outcomes
among exposed workers, and therefore, confidence in these studies was increased. Nasal biopsies
were taken in four occupational studies; tissues were subsequently stained and cell structure
examined according to variations of the Toriussen etal. f 19791 method. The original Torjussen
method scored morphological characteristics of the nasal epithelium using a whole number
between 0 and 8, with 0 indicating normal epithelium and 8 indicating carcinoma and the midpoint
of four signifying stratified squamous epithelium with a horny layer. Despite the variations of this
scale, in each study the lowest numbers (0 or 1) always indicated normal cell structure while
increasingly higher numbers indicated more disruptive cellular changes. Although the focus of this
section is nonneoplastic histopathologic lesions, the studies compared the means of the total score
between exposed and referent groups. Therefore, the prevalence of dysplasia is presented in the
evidence tables when it was reported. Information regarding workplace temperature and
This document is a draft for review purposes only and does not constitute Agency policy.
1-151 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
humidity, or home environment, all of which may affect nasal pathology, was rarely reported
fArundel etal.. 19861.
Most studies of respiratory pathology in experimental animals used paraformaldehyde or
freshly prepared formalin as the test article, but some studies tested commercial formalin, an
aqueous solution containing both formaldehyde and methanol. The toxicokinetics of these two
chemicals are vastly different, and their toxicities are likely to vary as well. Highly reactive
formaldehyde is mostly captured in the nose, the main site of formaldehyde-induced lesions, and
very little enters the blood stream. Conversely, methanol mostly bypasses the nose but is readily
absorbed in the lungs and then distributed to distal sites, including the blood and other
nonrespiratory tissues, where it can be metabolized to formaldehyde. Inhalation studies of
methanol suggest that URT effects occur at concentrations many times higher than estimates of
methanol concentrations in air, at least those generated from spraying formalin solutions onto
heated glass14 (e.g., >650 mg/m3 in methanol studies by Poon et al. (1995) and Andrews et al.
(19871 versus 5.5 mg/m3 methanol reported by Kamata et al. (19971 in a formalin study testing
formaldehyde levels of 0 and 18.27 mg/m3). Thus, in general, the levels of methanol in formalin
studies are considered unlikely to cause substantial increases in URT lesion severity. While
coexposure to methanol in formalin studies may be a significant confounding factor for systemic
effects, it is not expected to have a substantial influence on formaldehyde-induced respiratory
effects. However, it does introduce the possibility that effective respiratory tract tissue
concentrations of formaldehyde might be slightly higher after inhalation of formalin (due to some
methanol conversion to formaldehyde within the tissue) than after exposure to the same
concentrations of formaldehyde from sources without methanol, which would result in an
overestimate of the effect of formaldehyde exposure. A discussion of the different test articles
(e.g., paraformaldehyde, formalin) used for formaldehyde inhalation studies can be found in
Appendix A.5.1.
For assessing histopathological changes for the different regions of rodent nasal passages,
standard cross-section levels (e.g., Levels I-V) have generally been adopted for consistent analysis
across studies (Merv etal.. 1994: Young. 1981). Although the number and naming of cross-section
levels varied from study to study, the levels always progressed through the nasal cavity from the
area posterior to the nostrils (e.g., Level I or A) to areas anterior to the nasopharynx. Two different
examples of the cross-sectioning procedures in rats are illustrated in Figure 1-14, with other
studies of rats and other rodents employing similar procedures; however, illustrations of the
specific cross-section levels used in each individual study are not included in the evidence tables.
14Even though methanol levels in the air using the generation methods in the other available formalin studies
may be quite different, and possibly significantly higher, than the levels estimated by Kamata et al. (19971
(see Preface for discussion), given the relative insensitivity of the URT to methanol, these crude comparisons
were considered sufficient for interpretations drawn in the context of these URT effects.
This document is a draft for review purposes only and does not constitute Agency policy.
1-152 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Figure 1-14. Example cross-section levels in rat nasal passages used for
histopathological evaluations from Kerns et al. (1983) (left; Levels I-V) and
Kamata et al. (1997) (right; Levels A-E).
It is preferred that studies assessed multiple tissue sections across multiple cross-section
levels to allow for reasonable sampling of the nasal mucosa. Where applicable, histopathological
findings in the nasal mucosa are discussed with reference to these sections, and the specific
structures examined are stipulated in the evidence tables (e.g., nasoturbinates, maxilloturbinates,
or ethmoid turbinates). When data were available, the type of epithelium affected (e.g., respiratory
epithelium) was also noted. Only a few studies evaluated sections of the URT distal to the nasal
cavity, and these evaluations were generally less rigorous (e.g., examining only a single tissue
section) than evaluations of the nasal mucosa and tested much higher formaldehyde
concentrations. Pathological findings in the LRT were generally not identified in higher confidence
studies and are not discussed.
Based on the considerations described above, as well as other potential methodological
issues, the experimental animal studies were evaluated with regard to the utility of their study
results for characterizing hazard (see Appendix A.5.5 for details). Because of the abundance of
studies of respiratory pathology, this section focuses on longer duration (i.e., chronic and
subchronic) studies interpreted with high or medium confidence. Unlike some other sections, this
includes well-performed formalin studies (see above.)
Some studies reported multiple endpoints (e.g., pathological effects and cell proliferation),
which were individually considered. Overall, 22 controlled exposure studies were identified as high
or medium confidence for characterizing respiratory pathology. Studies that reported URT
pathology-related mechanistic information relevant to interpreting the progression of events
leading to overt respiratory tract pathology, including cell proliferation and mucociliary function,
are also discussed (see Appendix A.5.6).
Respiratory Tract Pathology Studies in Humans
A small number of studies were available that reported the results of histological
examinations of nasal tissues from formaldehyde-exposed occupational groups. These are
described in Table 1-25, organized by publication year. Although the evidence was more equivocal
in one study (Bovsen etal.. 1990). the four medium confidence studies examining histopathology
found that exposed participants had a higher average histopathological score than their respective
This document is a draft for review purposes only and does not constitute Agency policy.
1-153 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Toxicological Review of Formaldehyde—Inhalation
comparison group (Ballarin etal.. 1992: Holmstrom et al.. 1989c: Edling etal.. 19881. Average
formaldehyde levels ranged from 0.05 to 0.6 mg/m3. These were cross-sectional studies of current
workers who likely were less sensitive "survivors" of the long-term respiratory irritant effects of
formaldehyde, which would cause survival bias and an attenuation of comparisons between
exposed and comparison groups. Although the studies were limited by probable survival bias, and
in some cases, other limitations resulting in a bias toward the null, a consistent association with
histopathological endpoints was observed. Edling et al. (1988) did not adjust analyses for
differences in smoking prevalence between the exposed and referent groups; smoking prevalence
was higher among participants in the referent group. Therefore, the expected effect on the
association with formaldehyde exposure would again be toward the null. However, the association
observed by Edling et al. (1988) was consistent with those reported by the other studies that did
address potential confounding by smoking status. There was no evidence of a time-dependent
relationship with formaldehyde. Additionally, there was no indication that coexposure with wood
dust or smoking modified the pathological effects of formaldehyde.
The preponderance of evidence shows that the increases in histopathological score levels
were due to a high level of squamous metaplasia among participants exposed to formaldehyde
levels ranging from 0.1 to 2.5 mg/m3. Squamous metaplasia was seen in 32-67% of exposed
participants fBallarin etal.. 1992: Bovsen etal.. 1990: Edling etal.. 19881.
Table 1-25. Formaldehyde effects on respiratory pathology in occupational
settings
Study and design and exposure
Results
Histological analyses
Ballarin et al. (1992)
Italy
Prevalence study
Population: 15 plywood factory workers (mean age 31 yrs,
employment duration 6.8 yrs) compared to 15 university or hospital
clerks matched for age and sex. All nonsmokers.
Exposure: Personal sampling;
8-hrtwa Kominskvand Stroman (1977)
Warehouse (N = 3), 0.39 ± 0.20 mg/m3, range 0.21-0.6 mg/m3
Shearing-press (N = 8), 0.1 ± 0.02 mg/m3, range 0.08-0.14 mg/m3
Sawmill (N = 1), 0.09 mg/m3
Inspirable wood dust: 0.11-0.69 mg/m3, 0.73 in sawmill
Methods: Cytopathology analysis of nasal respiratory mucosa cells
blinded by two readers, scoring and classification analogous to
Toriussen et al. (1979) and Edling et al. (1988); most
Distribution of histological scores of nasal
respiratory mucosa cells
Description Exposed Referent
Normal 0 4 (26%)
Loss of ciliated cells 15 (100%) 10 (67%)
Hyperplasia 6 (40%) 5 (33%)
Squamous metaplasia 10(67%)* 1(6%)
Mild dysplasia 1 (6%) 0
Score (Mean (SD)) 2.3 (0.5)* 1.6 (0.5)
*Mann-Whitney U test (p < 0.01) orx2 test
(p < 0.01)
severe score present assigned. Mean histological scores exposed
compared to referent using Mann-Whitney U test; difference by
exposure group for classification of pathology, x2 test.
Evaluation:3
This document is a draft for review purposes only and does not constitute Agency policy.
1-154 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design and exposure
Results
SB
IB Cf Oth
Overall
Confidence
Medium
4-
h
Inclusion only of current workers raises possibility of healthy worker
survival effect due to irritation effects.
Boysen et al. (1990)
Prevalence survey
Oslo, Norway
Population: 37/74 exposed volunteers from a chemical company
producing formaldehyde (50% of exposed workforce). Mean age 51,
range 27-66 years. Mean years employed 20, range 3-36 years.
37 age-matched referent subjects without overt nasal disease or
occupations associated with nasal cancer. Office staff at two Oslo
chemical companies, hospital laboratory personnel, and outpatients at
the ear, nose, and throat department of hospital. Mean age 49, range
35-66 years.
Exposure: Systematic formaldehyde monitoring after 1980. Before
1980, exposure assessed by plant health officer with knowledge of the
production process, recent measurements, and worker sensations.
Range of formaldehyde 0.5 ppm to >2 ppm.
Methods: Scoring and classification of histologic samples per
Tojussen, 1979 protocol but on a 0-5-point scale by two authors
blinded to clinical or occupational status. Wilcoxon rank sum test
used to compare histological findings in the two groups, x2 test used
to compare the rhinoscopical findings and subjective complaints.
Evaluation:3
Rhinoscopy: 75% of exposed workers and 89% of
controls had normal mucosa. 24% of the exposed
and 8% of the unexposed had hyperplastic nasal
mucosa (difference not statistically significant).
Degree of metaplastic alterations more pronounced
among the exposed workers than in controls
(difference not statistically significant).
Higher prevalence of subjective nasal complaints in
formaldehyde-exposed workers (43%) compared to
5% in unexposed controls (p < 0.01).
Distribution of histological scores
Description
Exposed Referent
SB
IB Cf Oth
Overall
Confidence
Medium
4,
h
Inclusion only of current workers and long duration of employment
raises possibility of healthy worker survival effect due to irritation
effects.
0 Columnar
epithelium
1 Stratified cuboidal
epithelium
2 Mixed stratified
cuboidal/stratifie
d squamous
epithelium
3 Stratified
squamous
epithelium,
nonkeratinizing
4 Stratified
squamous
epithelium,
keratinizing
5 Dysplasia
16
5
17
10
3
1.9/5
0
1.4/5
Holmstrom et al. (1989a); Holmstrom and
Wilhelmsson (1988)
Sweden
Prevalence study
Population: Two exposed groups 170 total; 70 formaldehyde
production workers, Mean age 36.9 years, 87% male, mean duration
employment 10.4 yr. 100 workers exposed to wood dust and
formaldehyde at five furniture factories. Mean age 40.5 years, 93%
male, mean duration employment 16.6 yr. Referent: 36 persons from
local government in the same village as the furniture workers, with no
history of occupational exposure to formaldehyde or wood dust.
Mean age 39.8 years, 56% male, mean duration employment 11.4 yr.
"Slightly" larger number of smokers in the exposed group than control
group, but difference not statistically significant (data not provided).
Exposure: Personal sampling in breathing zone for 1-2 hours in 1985.
Total dust and respirable dust also measured.
Formaldehyde-only nasal specimens had higher mean
score of 2.16 (range 0-4) (p < 0.05, comparison to
referent) while formaldehyde-dust group had mean
score of 2.07 (range 0-6) (p > 0.05, comparison to
referent). Referent group score was 1.56 (range 0-4).
Combining formaldehyde-only and formaldehyde-
dust group mean score of 2.11 (p < 0.05). No
correlation observed between smoking habits and
biopsy score, nor was a correlation found between
the duration of exposure and any histological
changes.
This document is a draft for review purposes only and does not constitute Agency policy.
1-155 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design and exposure
Results
Previous measurements 1979-1984 in chemical company combined
with 1985 values to estimate average annual values for each
participant. Only 1985 values available for wood factories.
Formaldehyde concentration: Chemical Plant: 0.05-0.5 mg/m3, mean
0.26 [SD 0.17 mg/m3]. Furniture Factory: 0.2-0.3 mg/m3, mean 0.25
[SD 0.05 mg/m3]. Referent mean 0.09 mg/m3 (based on four
measurements in four seasons).
Methods: Pretesting questionnaire, histological changes in nasal
mucosa graded by a pathologist blind to exposure according to
ToriUSSen et al. (1979) grading scale of 0-8. 2 tailed t-test for
group comparisons.
Evaluation:3
SB
IB Cf Oth
Overall
Confidence
Medium
4>
h
Inclusion of only current workers and long duration of employment
raises possibility of healthy worker survival effect due to irritation
effects.
Edlingetal. (1988,1987a)
Prevalence Study
Sweden
Population: 75 of 104 exposed male factory workers from three plants
(2 particle board plants and one laminae-processing). Mean duration
of employment: 10.5 years. Mean age: 38 years; range 22-63 years.
35% smokers and 9% ex-smokers. Referents: 25 men with similar age
and smoking habits and no known industrial exposures to
formaldehyde. Mean age: 35 years, range 25-60. 48% smokers and
10% ex-smokers.
Exposure: Past TWA formaldehyde measurements made by plant
industrial hygienists sporadically between 1975 and 1983. Levels of
formaldehyde in air ranged from 0.1 to 1.1 mg/m3, with peaks up to
5 mg/m3. No measurements available before 1975 but estimated
levels higher during the 1960s and early 1970s. Particle board plants
contained low concentrations of wood dust at 0.6-1.1 mg/m3.
Methods: Nasal mucosa histological grading by pathologist blinded to
exposure using ToriUSSen et al. (1979) grading system with 0-8
ranking.
Compared differences in nasal mucosa histological score using
Wilcoxon nonparametric test.
Evaluation:3
Prevalence in exposed of normal nasal mucosa, 75%;
prevalence swollen or dry or both changes, 25%.
Histological scores higher in exposed compared to
referents, mean 2.9 vs. 1.8; (p < 0.05) (Wilcoxon). No
association with years of exposure.
Histological scores in exposed3
Characteristics
Score
%
Normal respiratory
epithelium
Loss of ciliated cells
Mixed cuboid/squamous
epithelium, metaplasia
Stratified squamous
epithelium
Keratosis
Budding of epithelium
Mild or moderate
dysplasia
Severe dysplasia
Carcinoma
0
3
4
1
8
11
2
24
32
3
18
24
4
16
21
5
0
0
6
6
8
7
0
0
8
0
0
aData for referent group were not reported
SB IB Cf
Oth
Overall
Confidence
1 1 1
Medium
4,
Inclusion of only current workers and long duration of employment
(mean 10.5 years) and high prevalence of symptoms raises possibility
of healthy worker survival.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.1 and A.5.5. SB = selection bias; IB = information
bias; Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation.
Direction of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be
This document is a draft for review purposes only and does not constitute Agency policy.
1-156 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
toward the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be
away from the null (i.e., spurious or inflated effect estimate).
Respiratory Tract Pathology Studies in Animals
A large database of well-designed studies has characterized formaldehyde-induced
respiratory tract pathology in mice, hamsters, and monkeys, but primarily in rats. The durations of
these studies range from a few hours to longer than 2 years, and several studies included recovery
periods that explored the reversibility of lesions. While a few studies include the examination of
tissues in other areas of the respiratory tract, most studies focus on pathology in the nasal mucosa.
This synthesis focuses on the incidence of hyperplasia and metaplasia formed after inhaled
formaldehyde exposure. To the extent the available data allow, the discussion separately addresses
the lesion locations along the URT and specifically within the nasal mucosa, the influence of
concentration and exposure duration on lesion formation and lesion persistence, and sex and
species differences in pathology. Because of the abundance of studies that evaluated respiratory
tract pathology, only those studies judged to be of high and medium confidence (see
Appendix A.5.5) are presented in detail in the synthesis and evidence tables below. Likewise, as
animal studies of effects from long-term exposure are most pertinent to lifetime human exposure,
and because some of these lesions can be very slow to develop, long-term studies (preferably
>52 weeks of exposure and follow-up) were generally considered to be more informative.
Accordingly, evidence tables of the experimental animal studies are organized by study duration,
with chronic and subchronic respiratory pathology studies ordered according to species and study
confidence in Tables 1-26 and 1-27, respectively. Short-term studies, generally <1-4 weeks long,
are sometimes discussed in the synthesis, but are only described in detail if they provide insights
unavailable in the longer-term studies, specifically including information on potential species
differences or the relationship between the concentration and duration dependency of lesion
formation (see Appendix A.5.5 for evidence tables of the other short-term studies).
Nasal lesions (i.e., cytotoxicity, hyperplasia, and metaplasia) have been consistently
reported in multiple rodent species and strains, and in monkeys. For hyperplasia and metaplasia,
there were consistent indications of a concentration-response, and to a somewhat lesser extent,
exposure duration-dependent relationships with inhaled formaldehyde. Somewhat surprisingly,
multiple studies report that metaplasia appeared to be more sensitive, prevalent or extensive than
hyperplasia (sometimes pronounced metaplasia was observed in the absence of hyperplasia),
reducing support for a strictly sequential progression of these lesions. The most informative data
on squamous metaplasia (i.e., from long-term medium or high confidence studies), which is
considered to be an adverse effect independent of its potential role in cancer progression, are
illustrated in Figures 1-15 and 1-16.
This document is a draft for review purposes only and does not constitute Agency policy.
1-157 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
O No notable change; size of shape illustrates sample size
• Shadings possible treatment-related effect (see below); size of shape illustrates sample size
O
0
o
o
() ©
©—
• t
O
_o_
o p
O C)
6 0
0
6
o
T5 15
S s
F344 Wistar
[M+F] [M]
2yr 28mos
n=27+ n=26+
% %
Wistar Wistar Wista
[M] [M] [M]
13 wk 3 mos 13 wk
n=44+ n=26+ n=20
Chronic (> lyr) Subchronic (> 3 mos)
Rat
HIGH
Syr.g.
[M]
Life
n=88+
Ch ron ic
Hamstei
T5 15 15
% « «
SD F344 Wistar
[M] [M+F] [M]
Life 28 mos lyr
n=99+ n=32 n=10
Ch ron ic
! !
is
« %
! I
Wistar F344 F344 Wistar
[M] [M] [M+F] [M+F]
13 wk 13 wk 26 wk 13 wk
n=25 n=8 n=36+ n=50
Subchronic
B6C3F1
[M+F]
2 yr
n=25+
B6C3F1 TP53ht
[M+F] [M]
13 wk 8wk
n=20 n=21+
•Chronic Sub. Short
Mouse
Syr.g.
[M+F]
26 wk
n=10
Sub.
Cynom.
[M]
26 wk
n=6
Sub.
HamstefMonke1
Figure 1-15. Squamous metaplasia in medium and high confidence chronic and
subchronic respiratory pathology studies of inhaled formaldehyde.
Studies are organized by study evaluation confidence (see Appendix A.5.5), species, and then duration of
exposure. Shading is indicated as follows: black = statistically significant effects, as indicated by study
authors; gray = increases in incidence in studies without statistical analyses, with dark gray indicating
pronounced changes (incidences of 50-100% were noted for many of these groups) and light gray
indicating subtle changes (generally <25% change compared to controls); see Tables 1-26 through 1-28.
Exposure groups with larger sample sizes are depicted as larger circles. Abbreviations: Syr. g. = Syrian
golden; ht = heterozygotes; Sub. = subchronic; M + F = male and female; wk = week, mos = months, yr = year.
This document is a draft for review purposes only and does not constitute Agency policy.
1-158 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
A.
100-
90-
80-
a, 70
s so H
~a
U 50-1
S? 40"
30-
20-
101
<
0<
0
B.
100-
90-
80-
70-
60-
50-
40-
30-
20
H
a
o
i i i 1 1—
6 9 12 15 18
ft
-a-
o
18
3 6 9 12 15
Formaldehyde concentration (mg/m3)
High confidence Medium confidence
~ Kerns, 1983 (F344; 2yr; n=27-100) O Sellakumar, 1985 (SD; life; n=99-100)
¦ Woutersen, 1989 (Wistar; 2.5yr; n=26-28) O Kamata, 1997 (F344; 2.5yr; n=32)
~ Appelman, 1998 (Wistar; lyr; n~10)
•| d Feron, 1988 (Wistar; 13wk: 2yr evai; n=44-45) Zwart, 1988 (Wistar; 13wk; n=50)
j= S Woutersen, 1989 (Wistar; 3mos: 2yr evai; n=26-30) C Wilmer, 1989 (Wistar; 13wk; n=22-25)
¦ Woutersen, 1987 (Wistar; 13wk; n=20) ¦ Appelman, 1998 (Wistar; 13wk; n=10)
<> Rusch, 1983 (F344; 26wk; n=26-29)
O Andersen, 2010 (F344; 13wk; n=8)
Figure 1-16. Squamous metaplasia incidence in high and medium confidence
rat studies of chronic and subchronic formaldehyde exposure duration.
Incidence data for squamous metaplasia (i.e., of any severity) from the high and medium confidence
studies with >1 year of formaldehyde exposure (Panel A, chronic exposure) or with >3 months of exposure
(Panel B, subchronic exposure). Symbols for chronic studies are outlined in black, while subchronic
studies are outlined in gray. In addition, high confidence studies include black fill, while medium
confidence studies are filled in either white or a combination of white and gray. The size of the points
reflects sample size for that particular exposure group (i.e., larger size = larger n). Notes: this figure does
not present statistical significance; data points at 24.2 mg/m3 (Woutersen et al„ 1987) and 24.6 mg/m3
(Feron et al.. 1988) formaldehyde are not shown (the incidence of squamous metaplasia was
approximately 100% at these levels).
This document is a draft for review purposes only and does not constitute Agency policy.
1-159' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Anatomical location of lesions in the upper respiratory tract
As previously mentioned, the majority of evidence for formaldehyde exposure-induced
pathology in the URT of experimental animals is confined to the nasal cavity, which is discussed in
greater detail in the sections below. This focus on the nasal cavity can be explained, at least in part,
by the historical interest in nasal carcinogenesis.
The evidence for lesions beyond the nasal cavity in rats suggests that concentration is an
important variable in long-term studies. Laryngeal lesions, including hyperplasia and squamous
metaplasia, were observed in Sprague Dawley rats exposed to 18.2 mg/m3 for a lifetime
fSellakumar etal.. 19851 and in male Wistar rats exposed to 24.4 mg/m3, but not to <11.9 mg/m3,
for 13 weeks fWoutersen et al.. 19871. Tracheal lesions (metaplasia and hyperplasia) were
reported in F344 rats after chronic exposure to 17.6 mg/m3 formaldehyde fKerns etal.. 19831.
Similar results were observed in Sprague Dawley rats in a single concentration (18.2 mg/m3)
lifetime study (Sellakumar et al.. 19851. However, no laryngeal or tracheal lesions were observed in
rats exposed to 11.6 mg/m3 for 1 year (Appelman etal.. 19881.
As reported in three studies, even higher concentrations of inhaled formaldehyde may be
necessary for effects beyond the nose in mice. Histopathological changes were not observed in the
trachea or lungs of B6C3F1 mice exposed to 17.6 mg/m3 for 104 weeks in a study that did not
provide quantitative incidence or severity information fKerns etal.. 19831. nor in the larynx of mice
exposed to up to 18.5 mg/m3 for 8 weeks and evaluated at 1 year (Morgan etal.. 2017). However, a
subchronic formalin study observed increases in metaplasia and hyperplasia in the trachea at
>25.1 mg/m3 and in the lung at >49.6 mg/m3 (Maronpot etal.. 1986). These high-concentration
changes were also observed in a low confidence study with limited severity information that
observed squamous metaplasia and hyperplasia in the tracheobronchial epithelium of C3H mice
exposed to >50 mg/m3 for 35 weeks fHorton etal.. 19631.
While it is difficult to draw mechanistic inferences with confidence, these studies suggest
that, in rodents, high levels of formaldehyde might be necessary to exceed the ability of the nose to
scrub formaldehyde from inhaled air and allow formaldehyde to reach sites farther down the
respiratory tract, which would be consistent with rodent toxicokinetic data (Appendix A.2).
Somewhat in contrast to the rodent studies, a single medium confidence study in rhesus
monkeys, which failed to report lesion severity or incidence, observed a loss of goblet cells,
hyperplasia, and metaplasia in the larynx, trachea, and carina, but not in the lungs, after exposure
for <6 weeks to 7.4 mg/m3 formaldehyde (Monticello etal.. 1989). This might suggest that the
monkey nose is less efficient than the rodent nose at scrubbing formaldehyde from inhaled air.
Overall, the evidence indicates the potential for lesions in the larynx and trachea of rats at
sustained high formaldehyde concentrations and in rhesus monkeys at sustained moderate
concentrations. These findings are particularly interesting in the context of future research into
anatomical lesion location following formaldehyde inhalation in nonrodent animal models. The
This document is a draft for review purposes only and does not constitute Agency policy.
1-160 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
remainder of this section will highlight the far more robust evidence of respiratory tract pathology
localized to the nasal cavity.
Duration dependency of nasal lesions
Data from exposed rats, supported by findings in other species, identify a clear relationship
between formaldehyde exposure duration and the development of squamous metaplasia and, to a
lesser extent, hyperplasia. These lesions appear to be at least partially reversible after exposure
ceases (see Tables 1-26 through 1-28 for study details).
As shown in Figure 1-17, the nasal cavities of monkeys and rats are lined with four types of
epithelia—squamous, transitional, respiratory, and olfactory—and there are unique structures that
may be susceptible to pathological change fRenne etal.. 2009: Harkema et al.. 2006: Renne and
Gideon. 2006: Monticello etal.. 1989: Young. 19811. Due to the high reactivity and water solubility
of formaldehyde, nasal metaplasia and hyperplasia have primarily been assessed (and subsequently
observed) in the epithelium lining the anterior regions of rodent nasal passages (typically Levels I,
II, and III) following formaldehyde inhalation exposure, mostly in regions containing respiratory
epithelium.
Figure 1-17. The four epithelial cell populations that line the nasal lateral wall
in monkeys and rats are portrayed in this image.
The cell populations are SE = squamous epithelium, TE = transitional epithelium, RE = respiratory
epithelium, OE = olfactory epithelium. Note that considerably more olfactory epithelium (OE) lines the
intranasal surface in rats than in monkeys. Other abbreviations used in this image are NALT = nasal-
associated lymphoid tissue, et = ethmoturbinate, mt = maxilloturbinate, nt = nasoturbinate, na = naris,
it = incisor tooth, B = brain. Source: Harkema et al. (2006).
SE
TE
RE
OE
Ra'
(N) NALT
This document is a draft for review purposes only and does not constitute Agency policy.
1-161 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Squamous metaplasia
Squamous metaplasia has been observed to occur after chronic, subchronic, and short-term
exposure to inhaled formaldehyde. Overall, the most robust responses (i.e., higher incidence or
severity at lower formaldehyde concentrations) occur following chronic exposure.
Multiple chronic rat studies have reported robust increases in squamous metaplasia
following exposures of approximately 2.5-2.7 mg/m3 (Kamata etal.. 1997: Kerns etal.. 1983:
Battelle. 19821 or 11.3-11.6 mg/m3 fWoutersen et al.. 1989: Appelman et al.. 19881. although some
data suggest that slight increases might be present at lower levels (i.e., 0.4-1.2 mg/m3. Kamata et
al.. 1997: Woutersen et al.. 19891. In studies that compared changes in respiratory and olfactory
epithelia fWoutersen et al.. 1989: Appelman et al.. 19881. squamous metaplasia was observed
almost exclusively in the respiratory epithelium, except perhaps at the highest formaldehyde levels
and with the longest exposure durations [i.e., slight increase in metaplasia at 12.1 mg/m3 after
28 months of exposure in Woutersen et al. (19891], With subchronic exposure, squamous
metaplasia is observed in rat noses at higher concentrations (i.e., >11.3 mg/m3) in high confidence
studies by Appelman et al. (19881. Woutersen et al. (19871. and Feron et al. (19881. the results of
which are supported by consistent observations in two medium confidence studies f Andersen etal..
2010: Zwart etal.. 19881. although these latter studies observed increases at lower exposure levels
(i.e., 2.5-3.7 mg/m3). With short-term exposures ranging from 4.4 to 18.4 mg/m3, observations of
squamous metaplasia in rats across several studies with various methodological limitations provide
supporting evidence (Speit etal.. 2011: Andersen etal.. 2008: Cassee and Feron. 1994: Wilmer etal..
19871. although some findings were not completely consistent with a straightforward
duration-dependency (e.g., Andersen et al. (20081 observed squamous metaplasia with 5 days of
exposure, but not with shorter or longer exposure durations, at 7.4 mg/m3).
The duration-dependency of these lesions in rat studies also appears to be reflected by the
locations at which lesions develop, as well as their severity, possibly in parallel with the increases
resulting from increasing formaldehyde concentration (see additional discussion below). The
association with lesion location is demonstrated by the results of Kerns et al. (19831 which showed
that, in anterior nasal regions (i.e., Level I and II) of F344 rats exposed to >2.5 mg/m3, the incidence
of squamous metaplasia increased from <20 to 100% with increasing duration (i.e., 6-24 months);
however, in posterior nasal regions (i.e., Levels III—V), a duration-dependent increase in incidence
was only observed at 17.6 mg/m3 (Battelle. 19821. In some instances, noted by Kerns et al. (19831.
more posterior lesions were entirely unique to longer exposure durations as compared to shorter
exposures (e.g., Level III at 6.9 mg/m3 only with 24 months of exposure). Regarding severity,
squamous metaplasia was observed to increase (i.e., from slight focal lesions to metaplasia with
keratinization) with exposure duration increases from 13 to 52 weeks of exposure to 11.6 mg/m3 in
Wistar rats (Appelman et al.. 1988). Similarly, at >11.6 mg/m3 in Wistar rats, an increase in the
severity of squamous metaplasia in respiratory epithelium occurred as exposure duration
increased from 4-8 to 13 weeks (Feron etal.. 19881. and at very high formaldehyde levels
This document is a draft for review purposes only and does not constitute Agency policy.
1-162 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
(24.2 mg/m3), exposure duration was associated with an increase in the severity of focal
replacement of olfactory epithelium with respiratory epithelium.
Several studies in rats confirm the important role of exposure duration in lesion
development by demonstrating that the increases in lesions observed with longer-term exposure,
as compared to shorter-term exposure, were not attributable to longer latencies after formaldehyde
exposures began in the studies of longer-term exposure (i.e., since metaplasia, in particular, is
expected to take several weeks to months to develop). In these studies of Wistar rats, nasal lesions
including metaplasia and hyperplasia were consistently investigated at approximately 2 years of
age following formaldehyde exposures of different durations (which began at the same ages, thus
requiring longer periods of nonexposure in the shorter-term studies) fWoutersen etal.. 1989:
Feron etal.. 19881. When animal ages at evaluation and formaldehyde exposure levels were
matched, comparisons of subchronic exposure to chronic exposure fWoutersen et al.. 19891 and of
short-term exposure to subchronic exposure (Feron etal.. 1988) revealed greater incidences or
severity of these lesions with the longer exposure durations.
Rodent species other than rats also exhibit squamous metaplasia, although the
duration-dependence of these lesions has not been as well established. Additionally, compared to
rats, other laboratory rodents may require higher levels (i.e., mice) or exhibit a substantially
reduced response (i.e., hamsters), suggesting that there may be differences in species sensitivity to
formaldehyde-induced squamous metaplasia. Following chronic exposure, slight increases in the
number of mice with metaplasia were observed at 6.9 mg/m3, with more pronounced changes at
17.6 mg/m3 (Kerns etal.. 19831: however, the incidence and severity of these lesions were not
quantified. Similarly, in a subchronic formalin study, squamous metaplasia was observed in all
mice exposed to 12.4 mg/m3 fMaronpotetal.. 19861. Two strains of p53 deficient mice [Trp53
heterozygotes) also developed pronounced metaplasia at both tested concentrations (i.e., 9.23 and
18.45 mg/m3) after only 8 weeks of exposure fMorgan etal.. 20171. with changes that were dose
dependent and exhibited an anterior-to-posterior gradient, similar to findings in rats. Squamous
metaplasia was observed only in 5% of Syrian golden hamsters exposed to 12.3 mg/m3 for a
lifetime (Dalbev. 19821. and no changes were observed after subchronic exposure to 3.6 mg/m3 in
the same strain (Rusch etal.. 19831. although these studies did not provide lesion severity.
Although the few available monkey studies did not report detailed endpoint information,
squamous metaplasia was observed at 3.6 mg/m3 in cynomolgus monkeys following subchronic,
near-constant exposure (i.e., 22 hr/day for 7 d/week), and in rhesus monkeys after short-term
(i.e., 1 or 6 weeks) exposure to 7.4 mg/m3 (Monticello etal.. 1989). The latter study in rhesus
monkeys also supports the findings in rats of an anterior-to-posterior gradient of lesions with
increasing exposure duration, and the general susceptibility of respiratory epithelium. After
exposure to 7.4 mg/m3 for 1 week, mild squamous metaplasia was observed in the respiratory
epithelium of anterior regions (i.e., primarily Level A, the nasal atrium, but also including Levels B
and C); however, with exposure to the same concentration for 6 weeks, the lesions were more
This document is a draft for review purposes only and does not constitute Agency policy.
1-163 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
developed and had progressed to more posterior regions of the nasal cavity (i.e., regions of
olfactory epithelium close to the olfactory/respiratory epithelial interface, and including Levels D
and E) fMonticello etal.. 19891. In another study fRusch etal.. 19831. monkeys exposed to formalin
for 26 weeks had both squamous metaplasia and hyperplasia (these lesions were reported
together) in the middle region of the nasal turbinates, with incidences of 17% at 0.23 mg/m3 and
100% at 3.6 mg/m3. No exposure-related effects were reported for the anterior and posterior nasal
turbinates.
Although uncertainties remain, the reversibility of metaplasia may depend more on
formaldehyde concentration than the duration of exposure. In general, increases in squamous
metaplasia incidence appeared to be a persistent effect at higher levels of exposure (i.e., >11 mg/m3
in rats and >9 mg/m3 in mice), as these lesions were observed many months after formaldehyde
exposure in rat recovery study comparisons by Woutersen et al. (1989) and Feron et al. (1988), and
in two transgenic mouse strains (Morgan etal.. 2017). However, it appears that the magnitude of
this effect, particularly at lower formaldehyde levels (e.g., <6.9 mg/m3), decreases with a recovery
period, as evidenced by significant declines in the incidences of squamous metaplasia (and rhinitis)
in F344 rats and B6C3F1 mice 3 or 6 months after 24 months of exposure f Kerns etal.. 1983:
Battelle. 19821.
In summary, experimental studies, primarily in rats, have demonstrated that formaldehyde
exposure duration clearly influences the incidence, severity, or anatomical location of squamous
metaplasia.
Hyperplasia
As with metaplasia, hyperplasia of the nasal epithelium has been observed across various
durations of exposure. In some studies, hyperplasia was reported as a concurrent lesion with
metaplasia fKamata etal.. 1997: Cassee and Feron. 1994: Reuzel etal.. 1990: Rusch etal.. 19831.
Reliable results from several studies show that chronic formaldehyde exposure of
approximately 11.6-12.1 mg/m3 induces hyperplasia in the nasal epithelium of rats (Woutersen et
al.. 1989: Appelman et al.. 19881. Studies with more limited endpoint information also reported the
formation of hyperplasia following exposure to 7.4-18.2 mg/m3 fMonticello etal.. 1996: Sellakumar
etal.. 19851. Subchronic exposure to formaldehyde also leads to hyperplasia in rat nasal passages
after exposure to 11.9 mg/m3 (Woutersen et al.. 19871 and after exposure to approximately
3.7 mg/m3 as reported in two studies with limited endpoint information fZwartetal.. 1988: Rusch
etal.. 1983). Following short-term exposures in rats to 4.4-18.5 mg/m3, studies with
methodological shortcomings also report the formation of nasal epithelium hyperplasia (Andersen
etal.. 2008: Cassee and Feron. 1994: Wilmer etal.. 1987: Chang etal.. 1983). adding support While
in nearly all cases, hyperplasia was observed in respiratory or transitional epithelium (or, in a few
cases, isolated regions of olfactory epithelium), a single high confidence, short-term study reported
that after 4 weeks of exposure to 18.4 mg/m3, hyperplasia of the epithelium surrounding NALT
(nasal-associated lymphoid tissue) was observed in a majority (87.5%) of F344 rats, but not
This document is a draft for review purposes only and does not constitute Agency policy.
1-164 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
B6C3F1 mice (Kuper etal.. 20111. Overall, comparisons of the formaldehyde concentrations at
which significant increases in hyperplasia are observed across studies of differing exposure
duration do not provide a clear picture of the potential duration dependence of
formaldehyde-exposure-induced hyperplasia.
However, like the results for metaplasia, several rat studies comparing exposures of
differing exposure duration (e.g., chronic versus subchronic) demonstrate that increasing exposure
duration results in increases in the incidence and/or severity of hyperplasia in the respiratory
epithelium when testing the same formaldehyde concentrations and anatomical levels fWoutersen
etal.. 1989: Appelman et al.. 1988: Feron etal.. 1988: Kerns etal.. 19831. This included two high
confidence studies matching the age of the animals at assessment fWoutersen etal.. 1989: Feron et
al.. 19881 to allow identical amounts of time for lesions to develop after the exposures began.
Similarly, some data also indicate that duration can influence the location of the observed
hyperplasia, with an increased frequency of lesions in more posterior locations (i.e., at more
posterior nasal levels or in more posterior structures, such as the trachea) with longer-term
exposure fWoutersen et al.. 1989: Kerns etal.. 19831. However, in the identified rat studies, the
within-study increases in incidence or posterior location with comparatively longer exposures
were generally only observed at high levels of formaldehyde (i.e., >10 mg/m3), preventing clear
interpretations regarding the duration dependence of hyperplasia at lower formaldehyde levels.
The role for duration in the development of hyperplasia in other laboratory animal species
is less clear. Hyperplasia was reported in a chronic mouse study with limited endpoint information
following exposure to 2.5 mg/m3 (Kerns etal.. 19831. with consistent findings in a low confidence,
short-term study at 18.5 mg/m3 (Chang etal.. 19831: however, a medium confidence, short-term
study in transgenic mice failed to observe significant increases in hyperplasia after exposure to
9.23-18.5 mg/m3, despite the presence of pronounced metaplasia fMorgan et al.. 20171.
Interestingly, however, this short-term mouse study did observe increases in nasal osteogenesis
(evidence of bone proliferation in the nasal turbinates) at 18.45 mg/m3 in both strains tested
(Morgan et al.. 2017). In a lifetime study by Dalbev (1982). 5% of hamsters had hyperplasia
following exposure to 12.3 mg/m3; however, hyperplasia did not appear to develop in hamsters
exposed to 3.6 mg/m3 for 26 weeks, although hyperplasia was not specified (i.e., the authors
reported no treatment-related histopathology) fRusch etal.. 19831. In cynomolgus monkeys,
hyperplasia along with metaplasia was reported following subchronic exposure to 3.6 mg/m3
fRusch etal.. 19831. and hyperplasia was also found in rhesus monkeys exposed to 7.4 mg/m3,
although lesion incidence or severity was not reported (Monticello etal.. 1989). When specified,
the hyperplasia observed in mice (Kerns etal.. 1983) and rhesus monkeys (Monticello etal.. 1989)
was generally identified in the anterior nose.
Hyperplasia in rats and mice appears to persist, at least in part fWoutersen etal.. 1989:
Feron etal.. 1988: Kerns etal.. 1983: Battelle. 19821. as with observations of squamous metaplasia.
However, hyperplasia generally appears to be more reversible than metaplasia, even at higher
This document is a draft for review purposes only and does not constitute Agency policy.
1-165 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
formaldehyde concentrations, as evidenced by smaller increases in incidence with a prolonged
recovery following exposure to ~11 mg/m3 formaldehyde fWoutersen etal.. 1989: Feron etal..
19881. Findings in a short-term recovery study in rats fAndersen etal.. 20081. with similar results
observed in a low confidence study in mice f Chang etal.. 19831. suggest that hyperplasia may take
some small amount of time to develop, as lesions progressed in incidence or severity with 18 hours
of recovery after very brief (i.e., days) exposures.
Taken together, formaldehyde exposure duration does appear to have some influence on
the development of hyperplasia, primarily based on studies in rats. However, considering the
notable influence of exposure duration on metaplasia at formaldehyde levels ranging from 2.5 to
2.7 mg/m3 in rat studies fKamata etal.. 1997: Kerns etal.. 19831. the easier reversibility of
hyperplasia, as well as the generally more robust effects of duration on the incidence of metaplasia
as compared to hyperplasia across species, exposure duration appears to be more important to the
development of metaplasia in laboratory animals than to the development of hyperplasia. Overall,
uncertainties remain regarding the relative impact of duration on the development of hyperplasia
(particularly in species other than rats), as compared to the pronounced role for concentration,
particularly at low formaldehyde levels (see additional discussion below).
Necrosis¦, nasal damage, and cytotoxicity
Although possessing methodological limitations, numerous short-term studies and three
long-term studies in rats report overt damage to the nasal epithelium following exposure to
3.9-7.4 mg/m3 (Andersen etal.. 2010: Cassee etal.. 1996: Cassee and Feron. 19941.12 mg/m3
(Wilmer et al.. 19871. or approximately 18.5 mg/m3 (Speitetal.. 2011: Chang etal.. 19831. with
supporting evidence from ultrastructural analyses in a short-term study fMonteiro-Riviere and
Popp. 19861. Consistent observations of nasal tissue damage were reported in rhesus monkeys
fMonticello etal.. 19891 and in a low confidence, mouse study with methodological limitations
(Chang etal.. 1983) following short-term exposure to >7.4 mg/m3. In rhesus monkeys (Monticello
etal.. 1989). loss of cilia and goblet cells was more severe and covered a greater surface of
respiratory epithelium (including extranasal respiratory tract regions), as duration of exposure
increased. As these observations of tissue cytotoxicity generally appear to occur following
exposures of shorter duration than in many of the studies reporting metaplasia or hyperplasia at
similar formaldehyde concentrations, these data may be consistent with the evolution of
hyperplasia and metaplasia from other lesions with increasing exposure duration.
Concentration dependency of nasal lesions
The development of nasal lesions in rodents and monkeys has routinely been shown to
exhibit a strong concentration dependency in terms of incidence, frequency, severity, and location
of the observed lesions. This is particularly true for both squamous metaplasia and hyperplasia in
the respiratory epithelium. Importantly, several studies have reported the occurrence of
This document is a draft for review purposes only and does not constitute Agency policy.
1-166 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
metaplasia in the absence of hyperplasia at a given exposure level (see Tables 1-26 and 1-27 for
study details).
Squamous metaplasia
Although there is a demonstrated exposure duration dependency for the development of
squamous metaplasia, formaldehyde concentration appears to be at least as important, if not more
so. With increasing formaldehyde concentration, squamous metaplasia is observed in more
posterior regions of the nasal tissue, and there is a marked increase in both lesion incidence and
severity.
In a chronic study reporting metaplasia throughout the rat nasal passage (Kerns etal.. 1983:
Battelle. 19821. metaplasia was observed in the anterior nose (i.e., Level I) after exposure to
2.5 mg/m3 and progressed in incidence toward the posterior nose, reaching Level V only after
exposure to 17.6 mg/m3. Consistent observations of the anterior-to-posterior progression of
metaplasia with increasing exposure concentration were reported by another high confidence
chronic study (Woutersenetal.. 19891. These findings are supported by results from a low
confidence chronic study with limited endpoint reporting fMonticello etal.. 19961. as well as by
medium confidence subchronic (Andersen etal.. 20101 and short-term (Speitetal.. 20111 studies.
With a constant duration of exposure, concentration-dependent increases for metaplasia in
rat noses (Level II) after 24 months were reported in a chronic study where 1.1, 62.2, and 100% of
rats were observed to have squamous metaplasia after exposure to 2.5, 6.9, or 17.6 mg/m3,
respectively (Kerns etal.. 1983: Battelle. 19821. Additional studies provide support for a
concentration-dependent increase in squamous metaplasia incidence following chronic and
subchronic exposures in rats and mice f Andersen etal.. 2010: Kamataetal.. 1997: Woutersen etal..
1989: Feron etal.. 1988: Maronpot etal.. 19861. The incidence of squamous metaplasia and
hyperplasia (lesions were reported together) also increased with concentration in rats and
cynomolgus monkeys (Rusch etal.. 1983).
The severity of metaplasia (e.g., from very slight to severe) also increased with
concentration, as reported by subchronic studies (Andersen etal.. 2010: Feron etal.. 1988:
Woutersen etal.. 19871 and a short-term study with a relatively small sample size (Speit etal..
20111. In general, while concentration-dependent increases in more mild instances of metaplasia
are typically observed at concentrations of 2.5 mg/m3 and above (see previous section), moderate
or severe lesions were only observed at the highest formaldehyde concentrations (approximately
12 mg/m3 or more). The available studies demonstrate that formaldehyde exposure concentration
occupies a central role in the development of squamous metaplasia.
Hyperplasia
Concentration-dependent increases in the incidence and severity of hyperplasia have also
been observed in rats with chronic, subchronic, or short-term exposure durations fAndersen et al..
2008: Kamataetal.. 1997: Woutersen et al.. 1989: Appelman et al.. 19881 and with subchronic
This document is a draft for review purposes only and does not constitute Agency policy.
1-167 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
exposure in F344 rats and cynomolgus monkeys (Rusch etal.. 19831. Overall, the concentration
dependence of these lesions, in terms of location, incidence, and severity, closely paralleled the
pattern of changes observed for squamous metaplasia, identifying a strong influence of exposure
concentration on the development of hyperplasia.
Necrosis¦, nasal damage, and cytotoxicity
Results for concentration-dependent cytotoxicity are varied, as reported by
less-than-chronic studies. A subchronic study observed no concentration-dependent increase in
necrosis in the noses of Wistar rats exposed to 1.2 or 2.5 mg/m3 for 13 weeks fWilmer et al.. 19891.
Following <13 weeks of exposure to 0.8-18.5 mg/m3, however, the incidence of necrosis/erosions
in F344 noses generally increased with concentrations of 7.4 mg/m3 and greater f Andersen etal..
20101. Following 4 weeks of formalin exposure from 0.63 to 18.4 mg/m3, degeneration was
observed only after exposure to the highest concentration in F344 rats (Speit etal.. 2011). while
focal thinning and epithelial disarrangement of the respiratory epithelium was observed in Wistar
rats exposed to >12 mg/m3 fWilmer etal.. 19871.
Studies comparing potential differential contributions of duration and concentration
Several animal respiratory pathology studies employed designs that compared intermittent
and continuous exposure scenarios to examine the extent to which Haber's rule [C* t = K; where C
is concentration, t is time, and Kis a constant) applies to formaldehyde-induced nasal pathology. If,
for example, Haber's rule can be strictly applied, similar pathological lesions should result whether
rats are exposed to 12 mg/m3 for 3 hours (12 x 3 = 36) or to 6 mg/m3 for 6 hours (6x6 = 36).
Wilmer et al. (1987) and Wilmer et al. (1989) used continuous and intermittent exposure
scenarios to assess whether lesion formation appears to be influenced more by concentration or
duration of exposure. In Wilmer et al. (1987), male rats were exposed to formaldehyde
5 days/week for 4 weeks. Groups of rats were either continuously exposed for 8 hours/day to
target concentrations of 0, 6, or 12 mg/m3 formaldehyde, or intermittently exposed (30 minutes of
exposure followed by 30 minutes of nonexposure) to 0,12, or 25 mg/m3 formaldehyde (the
analytical concentrations were not reported). Thus, the weekly inhaled concentrations
(concentration x hours x days) were the same for the continuous and intermittent exposure
groups: 0, 240, or 480 mg/m3-h/week. The main difference was that the intermittently exposed
rats were exposed to higher concentrations than the continuously exposed rats. The rats exposed
intermittently to the higher concentrations (12 or 25 mg/m3) had greater nasal cell proliferation
and histopathologic lesions, including squamous metaplasia and basal cell hyperplasia, than did the
rats exposed continuously to the lower concentrations (6 or 12 mg/m3).
Similar results were seen in a 13-week study (Wilmer et al.. 1989) in which groups of male
rats were either continuously exposed for 8 hours/day to target concentrations of 0,1, or 2 mg/m3
formaldehyde, or intermittently exposed (30 minutes of exposure followed by 30 minutes of
nonexposure) to 0, 2, or 5 mg/m3 formaldehyde (again, the analytical concentrations were not
This document is a draft for review purposes only and does not constitute Agency policy.
1-168 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
reported). The rats exposed continuously had greater incidences of diffuse disarrangement, diffuse
necrosis, focal and diffuse basal cell hyperplasia, focal squamous metaplasia, keratinization, and
diffuse goblet cell hyperplasia than the rats exposed intermittently. For some of these lesions, the
incidences were greater in the rats exposed continuously to 2 mg/m3 than to 5 mg/m3, the
interpretation of which is unclear. Overall, the Wilmer et al. studies suggest that in rats exposed for
4 or 13 weeks the extent of nasal lesions and cell proliferation appears to be driven more by
concentration than by duration of exposure or cumulative dose. These findings are consistent with
changes in cell proliferation reported in an acute and a short-term study using similar approaches
fWilmer et al.. 1987: Swenberg etal.. 19831: (see Appendix A.5.6).
While the authors of another subchronic rat study reached similar conclusions, the data did
not fully support a clear concentration over duration driver for the observed effects. Rusch et al.
(1983) compared the findings in their 6-month rat study against the 6-month exposure phase in the
2-year rat study by Kerns et al., as reported in the supporting report by Battelle for CUT (Kerns et
al.. 1983: Battelle. 1982). Rusch et al. (1983) exposed animals 22 hours/day, 7 days/week for a
total of 154 hours/week, compared to 6 hours/day, 5 days/week in the Kerns et al. (1983: 1982)
study, for a total of 30 hours/week; that is, the rats in the Rusch et al. (1983) study were exposed
five times longer than in the Kerns et al. (1983; 19821 study. At 6 months, squamous metaplasia
was observed at 2.5 mg/m3 by Kerns et al. (1983; 19821 versus at 3.6 mg/m3 in the Rusch et al.
(1983) study. However, the incidence was ~60% at 3.6 mg/m3 in Rusch et al. (1983). as compared
to only 20% at 2.5 mg/m3 in the Kerns et al. (1983: 1982) study. In addition, while Kerns et al.
(1983: 1982) did not test lower formaldehyde levels, metaplasia incidence went from 2/38 in
controls to 3/36 at 1.2 mg/m3 in Rusch et al. (1983). introducing the possibility that the study may
have been inadequately powered to detect an effect at lower levels. Regardless, these data do
support the possibility of an increased dependence on concentration, as compared to duration, as
the rats in Rusch et al. (1983) did not appear to be five-fold more sensitive.
In summary, several rat studies suggest that formaldehyde, perhaps similar to mortality
responses following acute exposure to some other local irritants, may not adhere strictly to Haber's
rule for the induction of nasal pathology. Although duration of exposure has a clear and substantial
role for the development of these nasal lesions (see discussion above), the experiments by Wilmer
et al. (1987) and Wilmer et al. (1989) suggest that a power-law function [Cn * t = K) where n is >1
may better represent formaldehyde exposure-induced nasal lesions than the linear C x t = K, at least
when interpreting short-term or subchronic exposure (the exposure scenarios examined by Wilmer
et al.). Although a value for n was not identified for formaldehyde, or for exposure-induced nasal
pathology, in particular, studies of acute exposure to other local irritants and the concentration-
duration dependence for mortality suggest that the value for n, on average, is approximately
This document is a draft for review purposes only and does not constitute Agency policy.
1-169 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
1.8-1.9 (ranging from 0.5 to 4.0).15 It is difficult to speculate where within this range a value for n
might be most applicable to formaldehyde, particularly within the context of respiratory pathology
and long-term exposures (i.e., since these n values are for mortality after acute exposure); however,
based on the data discussed in previous sections, it might be reasonable to expect that an n defined
for associations with hyperplasia should be higher than one defined for metaplasia.
Species and sex differences in respiratory pathology
While most respiratory pathology studies have been conducted in rats, studies conducted
with mice, hamsters, and monkeys have reported interspecies differences in susceptibility
(i.e., lesion incidence and severity), and in the location of lesions. Additionally, differences between
sexes of the same species have also been observed.
Rats have consistently been shown to be more susceptible than mice to the formation of
various nasal lesions after chronic, subchronic, and short-term exposures. A well-conducted
bioassay exposing F344 rats and B6C3F1 mice to 2.5, 6.9, or 17.6 mg/m3 formaldehyde for
24 months reported that squamous metaplasia was observed in rat noses at all exposure levels,
whereas in mice metaplasia was only observed after exposure to the intermediate and high
concentrations. Additionally, lesions observed in mice were less severe than in rats at the same
concentration level. In fact, similar incidences of squamous cell carcinoma were observed in rats
exposed at 6.9 mg/m3 and in mice exposed at 17.6 mg/m3 fKerns etal.. 19831. Likewise, Kuper et
al. (2011) observed hyperplasia of the NALT lymphoepithelium in rats, but not in mice. A possible
explanation for these species disparities is that mice have a greater reflex bradypnea response than
rats and thus inhaled lower doses of formaldehyde than rats. Unfortunately, minute volume and
body temperature were not measured in the 2-year Battelle study or in Kuper et al. (2011), so there
is no way of knowing whether reflex bradypnea played a significant role (see Appendix A.3 for a
discussion on reflex bradypnea).
Rats also show differences with other species. Rats, and, to a lesser extent, mice, appear to
be more sensitive than Syrian hamsters (Appelman etal.. 1988: Rusch etal.. 1983: Dalbev. 1982).
The comparisons to nonrodent experimental models are less clear. Squamous metaplasia and
hyperplasia were specifically found in the anterior, middle, and posterior nasal turbinates of F344
rats, but lesions were predominantly in the middle nasal turbinates of cynomolgus monkeys (Rusch
etal.. 1983) and rhesus monkeys (Monticello etal.. 1989). Monticello et al. (1989) observed lesions
that extended to proximal regions of the URT (outside of the nasal cavity) at lower concentrations
than in the rat studies (7.4 mg/m3, as compared to >15 mg/m3), likely because the monkey nose is
less efficient than the rodent nose at scrubbing formaldehyde from inhaled air.
15Values of n for 11 local irritants as estimated by ten Berge et al. (1986) averaged 1.9 (range 1.0-3.5), while
21 local irritants relying on data in rats or mice, as summarized in Appendix G by California EPA (OEHHA.
2008). averaged 1.8 (range 0.5-4.0). Of potential interest to this assessment, the chemicals included
ammonia (n = 2.0) and acrolein (n = 1.2).
This document is a draft for review purposes only and does not constitute Agency policy.
1-170 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 In addition to differences between species, the formation of histopathological lesions was
2 sometimes observed to differ between sexes, although most studies only examined male animals. A
3 subchronic study in Wistar rats reported that males generally had more severe damage, including
4 metaplasia, to the nasal respiratory, olfactory epithelium, and larynx fWoutersen et al.. 19871.
5 Supportive findings of increased incidence or severity of lesions in males as compared to females
6 was also reported in a second subchronic study of Wistar rats (Zwartetal.. 1988). as well as in
7 mouse studies of subchronic fMaronpot etal.. 19861 and chronic (Kerns etal.. 1983: Battelle. 19821
8 duration. Male rats have a higher metabolic rate and oxygen demand than female rats, and
9 therefore greater minute volumes; thus, these findings might also reflect a greater inhaled dose of
10 formaldehyde in males as compared to females at the concentrations tested.
Table 1-26. Chronic respiratory pathology studies in animals
Reference and study design
Results
Rats
High confidence
Kerns et al. (1983)
Pathological changes"'b
Fischer 344 rats; males and females; 119 to
121/sex/group.
Exposure
duration
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Exposure: Rats were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for up to
24 months. Animals sacrificed at 27 and
30 months had 3- and 6-month periods of
nonexposure, respectively, after 24-months
of exposure.
Test article: Paraformaldehyde.
Actual concentrations were 0, 2.5 (±0.01),
6.9 (±0.02), or 17.6 (±0.05) mg/m3.a
6 months
NRC
Levels 1, II, and III:
purulent rhinitis,
epithelial dysplasia,
and squamous
metaplasia observed
Lesions first noted in
anterior sections
(Levels 1, II, and III) of
nose; changes in
epithelium restricted
to ventral portion of
nasal septum and
distal tips of
nasoturbinates and
maxilloturbinates
Histopathology: 5 midsagittal sections of
nasal turbinates (Levels l-V; see
Figure 1-14) for all animals that died or
were sacrificed at scheduled intervals
(i.e., at month 6,12,18, 24, 27, and 30).
Related studies/earlier reports: Batte lie
12
months
Level ld: purulent
rhinitis, epithelial
dysplasia, and
squamous
metaplasia
observed
NR
(1982,1981); Swenberg et al.
(1980a). See Battelle, 1982 for a more
detailed study report.
Note: transient viral infection at 52 weeks
was noted, but considered unlikely to
influence these findings.
18
months
NR
NR
24
months
Frequency of
metaplasia
exceeded that of
prior sacrifices;
dysplasia and
metaplasia only
observed in Level 1
NR
27
months0
Significant decrease
(p < 0.05) in
frequency of
metaplasia
Levels 1, II, and III:
regression (p < 0.05)
of squamous
metaplasia
Levels IV and V:
regression (p < 0.05)
of squamous
metaplasia
This document is a draft for review purposes only and does not constitute Agency policy.
1-171 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
aSeverity of lesions most intense in Level 1 for all exposure groups.
Exposure-related effects observed in Levels II, III, IV, and V for 6.9- and
17.6-mg/m3 groups. Lesion frequency in exposed groups greater than the <15%
lesion frequency observed for 0 mg/m3 group, where lesions (e.g., dysplasia and
metaplasia) only present in Level 1.
bAuthors defined squamous metaplasia as zones of altered epithelium
characterized by a well-differentiated germinal cell layer (stratum
germinativum) and superficial epithelial layers (stratum spinosum and stratum
corneum). Authors further noted that keratin was only produced in areas of
squamous metaplasia, and that in all exposure groups epithelial dysplasia was
detected earlier than squamous metaplasia.
cChart nine of Kerns et al. (1983) provides graphical representation of the
frequency of squamous metaplasia observed for Levels l-V for all exposure
groups during 24-month exposure and 3-month nonexposure period.
dAt this location, authors observed a transition in the mucosa from normal
nonciliated simple cuboidal epithelium to an epithelial lining several cells thick
and squamoid in appearance. The organization and polarity of the individual
epithelial cells changed from vertical to horizontal with respect to the basement
membrane. The authors termed such alterations as zones of epithelial dysplasia
and noted that similar histomorphological alterations have been called basal
cell hyperplasia and epidermoid metaplasia.
e24 months of exposure and 3 months of nonexposure.
General observations (respiratory epithelium):
17.6 mg/m3—squamous metaplasia with zones of squamous epithelial
hyperplasia and increased keratin production appeared to precede area
of squamous papillary hyperplasia with foci of cellular atypia;
dyspnea and death caused by excessive accumulation of keratin and
inflammatory exudate in lumen of nasal cavity of rats (with and
without carcinomas).
General observations (tracheal pathology):
17.6 mg/m3—rats (frequency NR) sacrificed at 18 months exhibited
multifocal areas of mild epithelial hyperplasia, epithelial
dysplasia, or squamous metaplasia of proximal tracheal mucosa;
similar lesions at a greater frequency (p < 0.05) observed in rats from
24-month sacrifice and unscheduled death groups; tracheal lesions not
observed in postexposure group.
0, 2.5, or 6.9 mg/m3—no significant tracheal lesions observed
Incidence of squamous metaplasia in nasal cavity of rats
Level la
Duration
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
6 months
NAb
4/20
10/20
NA
12 months
NA
7/20
11/20
NA
18 months
0/40
24/40
35/40
38/39
24 months
1/101
91/94
81/82
27/27
27 monthsd
3/19
4/20c
8/19c
5/5
30 months
1/10
2/5
1/8
NR
Level II
6 months
NA
0/20
10/20
NA
12 months
NA
0/20
8/20
NA
18 months
0/40
0/40
24/40
38/39
24 months
0/101
1/94
51/82
27/27
27 months
0/19
0/20
5/19c
5/5
This document is a draft for review purposes only and does not constitute Agency policy.
1-172 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
30 months
0/10
1 0/5
5/8
NR
Level III
6 months
0/20
0/20
0/20
6/20
12 months
0/20
0/20
0/20
10/20
18 months
0/40
0/40
0/40
38/39
24 months
0/101
0/94
9/82
26/27
27 months
0/19
0/20
0/19
4/5
30 months
0/10
0/5
0/8
NR
Level IV
6 months
NA
0/20
0/20
NA
12 months
NA
0/20
0/20
NA
18 months
0/40
0/40
0/40
14/39
24 months
0/101
0/94
1/82
21/27
27 months
0/19
0/20
0/19
1/5C
30 months
0/10
0/5
0/8
NR
Level V
6 months
NA
0/20
0/20
NA
12 months
NA
0/20
0/20
NA
18 months
0/40
0/40
0/40
11/39
24 months
0/101
0/94
0/82
19/27
27 months
0/19
0/20
0/19
0/5c
30 months
0/10
0/5
0/8
NR
aData reported in part in Kerns et al. (1983) and further adapted from Battelle
(1982)b tissue section not available for histopathology; cp < 0.05, regression of
squamous metaplasia 3 months postexposure;d
data for 27 and 30 months
represent incidence after 3 and 6 months of nonexposure, respectively,
following 24 months of exposure.
Woutersen et al. (1989)
3 months of exposure followed by a 25-month observation period with no
Wistar rats; male; 30/group.
Exposure: Rats were exposed to FA in
exposure:
FA-related histological changes generally not observed for Levels IV-VI.
dynamic whole-body chambers
6 hours/day, 5 days/week for 3 or
Histopathological nasal changes after 3 months of exposure and 25-montl
recovery period
28 months.
Incidence of lesions in Levels HI
Test article: Paraformaldehyde.
Actual concentrations were 0, 0.1 (±0.07),
1.2 (+0.22), or 11.3 (+2.0) mg/m3 for 3-
0 mg/m3
0.1 mg/m3
1.2 mg/m3
11.3
mg/m3
Type of lesions (Severity NR)
month exposures and 0, 0.1 (±0.05), 1.2
Respiratory epithelium
(±0.14), or 12.1 (±1.60) mg/m3 for 28-month
Disarrangement
0/2 6a
0/30
0/29
1/26
exposures.1
Squamous metaplasia
3/26
6/30
4/29
17/26
Histopathology: 6 standard cross sections of
Keratinization
0/26
0/30
1/29
2/26
the nose.
Basal cell/pseudoepithelial
1/26
0/30
0/29
4/26
hyperplasia
Note: This study also evaluated the effects
Nest-like infolds/goblet cell
11/26
3/30
15/29
9/26
of FA in a parallel group of rats that had
hyperplasia
undergone bilateral electrocoagulation
Invaginations
3/26
0/30
0/29
0/26
(i.e., damaged nose group) prior to the
Rhinitis
5/26
4/30
3/29
13/26
initiation of FA exposure. Data presented
Olfactory epithelium
here in the Results column are for FA-only
Thinning/disarrangement
0/26
0/30
0/29
0/26
(i.e., undamaged nose group) exposed rats.
Basal cell hyperplasia
0/26
0/30
0/29
0/26
Vacuolation/proteinaceous
0/26
0/30
0/29
0/26
material/numeric atrophy
This document is a draft for review purposes only and does not constitute Agency policy.
1-173 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Replaced by respiratory
epithelium
0/26
0/30
0/29
0/26
denominator represented by the effective number of animals and not the initial
number of animals.
Large variation observed for nest-like infolds/goblet cell hyperplasia; due to lack
of exposure-response, this change was not considered to be exposure-related.
28 months of exposure:
12.1 mg/m3—Incidence of rhinitis elevated in Level 1—VI; other FA-related
histological changes in respiratory epithelium generally found in Level II and III;
lesions observed in olfactory epithelium in Levels III and IV.
Histopathological nasal changes after 28 months of exposure period
Incidence of lesions in Levels l-ll
0 mg/m3
0.1
mg/m3
1.2
mg/m3
12.1
mg/m3
Type of lesions (Severity NR)
Respiratory epithelium
Disarrangement
0/2 6a
0/26
1/28
1/26
Squamous metaplasia
3/26
1/26
6/28
25/26
Keratinization
0/26
1/26
0/28
2/26
Basal cell/pseudoepithelial
hyperplasia
0/26
1/26
2/28
14/26
Nest-like infolds/goblet cell
hyperplasia
5/26
6/26
14/28
4/26
Invaginations
0/26
0/26
1/28
3/26
Rhinitis
2/26
1/26
2/28
18/26
Olfactory epithelium
Thinning/disarrangement
0/26
0/26
0/28
0/26
Squamous metaplasia
0/26
0/26
0/28
0/26
Basal cell hyperplasia
0/26
0/26
0/28
0/26
Vacuolation/proteinaceous
material/numeric atrophy
0/26
0/26
0/28
0/26
Replaced by respiratory
epithelium
0/26
0/26
0/28
0/26
denominator represented by the effective number of animals and not the initial
number of animals.
Highest incidence for nest-like infolds/goblet cell hyperplasia observed for Level
II at 1.2 mg/m3; due to lack of exposure-response, this change was not considered
to be exposure-related.
28 months of exposure (continued):
Histopathological nasal changes after 28 months of exposure period
Incidence of lesions in Level III
0 mg/m3
0.1
mg/m3
1.2
mg/m3
12.1
mg/m3
Type of lesions (Severity NR)
Respiratory epithelium
Disarrangement
4/26a
0/26
2/28
1/26
Squamous metaplasia
0/26
0/26
0/28
13/26
Keratinization
0/26
0/26
0/28
1/26
Basal cell/pseudoepithelial
hyperplasia
1/26
0/26
2/28
7/26
This document is a draft for review purposes only and does not constitute Agency policy.
1-174 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Nest-like infolds/goblet cell
hyperplasia
1/26
2/26
2/28
1/26
Invaginations
0/26
0/26
0/28
0/26
Rhinitis
1/26
0/26
2/28
6/26
Olfactory epithelium
Thinning/disarrangement
1/26
1/26
1/28
7/26
Squamous metaplasia
0/26
0/26
0/28
2/26
Basal cell hyperplasia
3/26
3/26
4/28
3/26
Vacuolation/proteinaceous
material/numeric atrophy
1/26
1/26
3/28
0/26
Replaced by respiratory
epithelium
0/26
0/26
1/28
2/26
denominator represented by the effective number of animals and not the initial
number of animals.
Medium confidence
Appelman et al. (1988)
SPF Wistar rat; male; 20/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for 52 weeks.
sacrificed at 13 weeks.
Slight
0/10
0/10
1/10
9/10a
Test article: Paraformaldehyde.
Moderate/severe
0/10
0/10
0/10
1/10
Actual concentrations were 0, 0.1 (±0.05),
Focal basal cell hyperplasia:
1.2 (±0.18), or 11.6 (±1.60) mg/m3.a
Slight
0/10
0/10
0/10
7/10a
Histopathology. nose (6 standard cross
Moderate/severe
0/10
0/10
0/10
0/10
levels), larynx, trachea, and lungs.
Focal rhinitis
0/10
0/10
0/10
6/10b
Nest-like infolds
0/10
0/10
0/10
0/10
Main limitations: small N; limited reporting
Olfactory epithelium
of lesion severity (note: this 12-month study
was shorter than the other available chronic
Focal
thinning/disarrangement
0/10
0/10
0/10
0/10
studies).
Focal basal cell
0/10
0/10
0/10
0/10
Note: This study also evaluated the effects
hyperplasia
Focal rhinitis
0/10
0/10
0/10
0/10
of FA in a parallel group of rats that had
undergone bilateral electrocoagulation 20
to 26 hours prior to the initiation of FA
exposure (not shown).
Histopathological nasal changes after 13 weeks of exposure (data included
for comparison with 52 weeks of exposure)
0.1
1.2
11.6
Type of lesion
0 mg/m3
mg/m3
mg/m3
mg/m3
Respiratory epithelium
Focal squamous metaplasia:
ap < 0.01; bp < 0.05
Histopathological nasal changes after 52 weeks of exposure
0.1
1.2
11.6
Type of lesion
0 mg/m3
mg/m3
mg/m3
mg/m3
Respiratory epithelium
Squamous metaplasia
Focal
0/10
0/10
0/10
6/10a
Diffuse
0/10
0/10
0/10
0/10
Keratinization
0/10
0/10
0/10
5/10a
Basal cell hyperplasia
Focal
0/10
0/10
0/10
5/10a
Diffuse
0/10
0/10
0/10
5/10a
Focal rhinitis
2/10
0/10
0/10
10/10a
Nest-like infolds
Focal
6/10
2/10
3/10
4/10
Diffuse
2/10
4/10
3/10
0/10
Olfactory epithelium
Thinning/disarrangement
1/10
0/10
0/10
3/10
This document is a draft for review purposes only and does not constitute Agency policy.
1-175 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Focal basal cell
0/10
0/10
0/10
2/10
hyperplasia
Focal squamous
0/10
0/10
0/10
0/10
metaplasia
Loosely arranged
0/10
0/10
0/10
0/10
submucosal connective
tissue
Focal rhinitis
0/10
0/10
0/10
0/10
ap < 0.05
Histopathological changes in larynx, trachea, and lungs were those commonly
found in this strain of rat and were about equally distributed among controls
and exposed groups or were only found in one rat; these changes ultimately
characterized as unrelated to FA exposure.
Kamata et al. (1997)
Fischer 344 rats; male; 32/group.
Exposure: Rats were exposed to FA in
dynamic nose-only chambers 6 hours/day,
5 days/week for 28 months with interim
sacrifices at the end of months 12,18, and
24.
Test article: Formalin (37% FA aqueous
solution containing 10% methanol).
Actual concentrations were 0, 0.40 (±0.09),
2.67 (±0.40), or 18.27 (±2.73) mg/m3.a The
concentration of methanol in the 0 and
18.27 groups was estimated to be 5.5
mg/m3.b A room control served as a no
exposure group.
Histopathology. nasal region (sections from
five anatomical levels, A-E; see Figure 1-14)
and trachea.
Main limitations: formalin; small N for
interim sacrifices; lesion severities NR
Group
Room
control
0 mg/m3
(5.5 mg/m3
MeOH)
0.40 mg/m3
2.67 mg/m3
18.27
mg/m3
(5.5 mg/m3
MeOH)
Squamous cell
metaplasia no
epithelial cell
hyperplasia
No nasal lesions
observed
No nasal lesions
observed
l/32a
(1/5 at
18-month)
5/3 2b
(2/5 at
18-month, 1/5
at 24-month,
2/7 at
28-month)
NR
Epithelial cell
hyperplasia with
squamous cell
metaplasia
No nasal lesions
observed
No nasal lesions
observed
4/32
(1/5 at
24-month, 3/11
at 28-month)
7/32c
(2/5 at
18-month, 1/7 at
28-month, 4/10
of dead)
29/32c
26/32c
(3/5 at
(4/5 at
12-month, 4/5 at
12-month,
18-month, 2/2 at
1/5 at
24-month, 20/20
18-month,
of dead)
1/2 at
24-month,
20/20 of
dead)
Epithelial cell
hyper-
keratosis
No nasal
lesions
observed
No nasal
lesions
observed
No nasal
lesions
observed
1/32
(1/10 of
dead)
Papillary
hyperplasia
No nasal
lesions
observed
No nasal
lesions
observed
No nasal
lesions
observed
No nasal
lesions
observed
2/32
(2/5 at
12-month)
adata reported as group total (i.e., dead animals plus scheduled sacrifices at 12,
18, 24, and 28 months); number in parenthesis represent incidence at sacrifice;
bp < 0.05, compared to 0 mg/m3 group; cp < 0.01, compared to 0 mg/m3 group
Sellakumar et al. (1985)
Sprague Dawley rats; male; 100/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for life.
Observation
0 mg/m3
18.2 mg/m3
Larynx
Hyperplasia
2/99
21/100
Squamous
0/99
4/100
metaplasia
This document is a draft for review purposes only and does not constitute Agency policy.
1-176 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Test article: Paraformaldehyde.
Trachea
Actual concentrations were 0 and 18.2
Hyperplasia
6/99
21/100
(±2.6) mg/m3.a
Squamous
0/99
7/100
Histopathology: multiple sections of the
metaplasia
head (from just behind the nostril to the eye
Nasal Mucosa
orbits) as well as sections of lung (each
Rhinitis (mild to
72/99
74/100
lobe), trachea, and larynx.
severe)
Preliminary study: Albert et al. (1982)
Epithelial or
51/99
57/100
squamous
Main limitations: likely coexposure to
hyperplasia
paraffin oil (kerosene); lesion severities NR
Squamous
5/99
60/100
metaplasia
Additional observations (frequencies NR) from FA exposures included:
exudation in the nasal cavity lumen; necrosis; desquamation of respiratory
epithelial cells of respiratory epithelial covering of naso-maxillary turbinates and
nasal septum; and inflammation of olfactory epithelium lining the ethmoidal
Mice
Medium confidence
Kerns et al. (1983)
B6C3F1 mice; males and females; 119 to
121/sex/group.
Exposure: Mice were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for up to
24 months. Animals sacrificed at 27 and
30 months had 3- and 6-month periods of
nonexposure, respectively, after 24-months
of exposure.
Test article: Paraformaldehyde.
Actual concentrations were 0, 2.5 (±0.01),
6.9 (±0.02), or 17.6 (±0.05) mg/m3.a
Histopathology. 5 midsagittal sections of
nasal turbinates corresponding to the
regions evaluated in rats in this study (levels
l-V; see Figure 1-14) for all animals that
died or were sacrificed at scheduled
intervals (i.e., at month 6,12,18, 24, 27,
and 30).
Earlier reports'. Battelle (1981);
Battelle (1982)
Main limitations: high mortality in all
groups; limited sampling (i.e., sections);
lesion incidence and severity NR
Exposure
duration
12 mos
18 mos
24 mos
27 mosb
Pathological changes"
2.5 mg/m3
ND
ND
Few animals had
serous rhinitis in
Level II, but no
significant nasal
lesions;
hyperplasia
(minimal to
moderate) of
squamous
epithelium lining
nasolacrimal duct
FA-related lesions
ND
6.9 mg/m3
ND
Few micec had
dysplastic changes
associated with
serous rhinitis in
Level II
Few mice had
dysplasia,
metaplasia, or
serous rhinitis in
Level II;
hyperplasia
(minimal to
moderate) of
squamous
epithelium lining
nasolacrimal duct;
focal atrophy of
olfactory
epithelium lining
the
ethmoturbinates
FA-related lesions
ND; regression
observed for
squamous
17.6 mg/m3
Serous rhinitis in Levels
III and V
~90% of mice had
dysplastic and
metaplastic alterations
of nasal mucosa in
Level II with a serous to
purulent change in
nasal exudate
>90% of mice had
dysplastic and
metaplastic changes
associated with
seropurulent rhinitis;
hyperplasia (minimal to
moderate) of
squamous epithelium
lining nasolacrimal
duct, greatest
frequency and
distribution found in
this FA level; focal
atrophy of olfactory
epithelium lining the
ethmoturbinates,
greatest frequency at
this FA level
Dysplastic epithelial
lesions with serous
exudate observed;
squamous metaplasia
This document is a draft for review purposes only and does not constitute Agency policy.
1-177 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
aUnless note
nonexposur
No tracheal
d, severities NR; b24 r
;; cUnless noted, exac
esions were observec
metaplasia and
rhinitis for all
affected Levels
nonths of exposure ar
frequency of lesion
.
in Level II in (~20% of
mice), but not in Levels
III and IV; regression
observed for squamous
metaplasia and rhinitis
for all affected Levels
id 3 months of
\IR.
Hamsters
Medium confidence
Dalbev (1982)
Syrian golden hamsters; male; 132 untreated
controls and 88 exposed.
Exposure: Hamsters were exposed to FA in
dynamic whole-body chambers 5 hours/day,
5 days/week for a lifetime.
Test article: Paraformaldehyde.
Actual FA concentrations were 0 and 12.3
(±5%) mg/m3.a
Histopathology. 2 transverse sections of the
nasal turbinates, longitudinal sections of
larynx and trachea, and all lung lobes cut
along the bronchus.
Main limitations: lesion severities NR
Note: this study also evaluated the effects
of FA on tumorigenicity of
diethylnitrosamine (DEN), either from
concurrent exposures or from DEN then FA
exposures (not shown).
Hamsters exposed at 12.3 mg/m3 had slightly reduced survival (p < 0.05) relative
to controls.
Nasal epithelium:
Hyperplastic lesions
12.3 mg/m3-4/88 (5%)
0 mg/m3-0/132
Metaplastic lesions
12.3 mg/m3-4/88 (5%)
0 mg/m3-0/132
Rhinitis
12.3 mg/m3-21/88 (24%)
0 mg/m3-41/132 (31%)
Abbreviations: FA = formaldehyde, NA = not available, ND = not detected, NR = not reported, SD = standard
deviation.
aStudy authors originally reported FA concentrations in ppm. These values were converted based on
1 ppm = 1.23 mg/m3, assuming 25°C and 760 mm Hg.
bStudy authors did not report methods for specific methanol measurements, but appeared to estimate the
concentration based on the proportion of methanol in the formalin solutions to determine their control group
methanol concentrations (see Preface on assessment methods and organization for relevant discussion of the
uncertainties related to this assumption). Study authors originally reported methanol concentrations in ppm.
These methanol values were converted based on 1 ppm = 1.31 mg/m3.
Table 1-27. Subchronic respiratory pathology studies in animals
Reference and study design
Results
Rats
High confidence
Feron et al. (1988)
Wistar rats; male; 45/group.
4 weeks of exposure followed by observation period of 126 weeks
| Incidence of lesions
This document is a draft for review purposes only and does not constitute Agency policy.
1-178 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Exposure: Rats were exposed to FA in
0
11.3
24.2
dynamic whole-body chambers
mg/m3
mg/m3
mg/m3
6 hours/day, 5 days/week for either 4, 8, or
Focal hyperplasia of respiratory epithelium
13 weeks followed by nonexposure periods
Very slight
0/44
0/44
0/45
of 126,122, or 117 weeks, respectively.
Slight
0/44
3/44
8/45c
Test article: Paraformaldehyde.
Moderate
0/44
0/44
1/45
Actual concentrations were 0,11.3 (±0.25),
Focal stratified squamous metaplasia of respiratory epithelium
or 24.2 (±0.12) mg/m3 for the 4-week
Very slight
3/44
6/44
14/45c
exposed groups; 0,11.6 (±0.21), or 24.2
(±0.11) mg/m3 for the 8-week exposed
groups; and 0,11.9 (±0.15), or 24.4 (±0.09)
mg/m3 for the 13-week exposed groups.3
Histopathology. 6 standard cross levels of
the nose.
Slight
4/44
2/44
19/45c
Moderate
0/44
2/44
3/45
Severe
0/44
0/44
0/45
Rhinitis
7/44
7/44
18/45b
Simple or stratified cuboidal or
0/44
0/44
4/45
squamous metaplasia of epithelium in
Note: only tested high formaldehyde levels
the dorsomedial area where respiratory
and olfactory epithelium joina
Focal replacement of olfactory epithelium by respiratory,
respiratory-like or regenerating olfactory epithelium
Very slight
0/44
0/44
0/45
Slight
1/44
0/44
6/45
Moderate
0/44
0/44
1/45
Severe
0/44
0/44
0/45
aThe changes in this area were scored separately because their origin from either
respiratory or olfactory epithelium was not clear; bp < 0.05; cp < 0.01
8 weeks of exposure followed by observation period of 122 weeks
Incidence of lesions
0
11.6
24.2
mg/m3
mg/m3
mg/m3
Focal hyperplasia of respiratory epithelium
Very slight
0/45
1/44
3/43
Slight
2/45
2/44
12/43c
Moderate
0/45
1/44
0/43
Focal stratified squamous metaplasia of respiratory epithelium
Very slight
8/45
16/44
17/43b
Slight
2/45
1/44
20/43c
Moderate
0/45
0/44
2/43
Severe
0/45
0/44
0/43
Rhinitis
4/45
6/44
22/43b
Simple or stratified cuboidal or
0/45
0/44
17/43c
squamous metaplasia of epithelium in
the dorsomedial area where respiratory
and olfactory epithelium join
Focal replacement of olfactory epithelium by respiratory,
respiratory-like or regenerating olfactory epithelium
Very slight
0/45
0/44
2/43
Slight
0/45
0/44
14/43b
Moderate
0/45
0/44
3/43
Severe
0/45
0/44
1/43
aSee above for explanation; bp < 0.05; cp < 0.01
13 weeks of exposure followed by observation period of 117 weeks
| Incidence of lesions
This document is a draft for review purposes only and does not constitute Agency policy.
1-179 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
0
mg/m3
11.9
mg/m3
24.4
mg/m3
Focal hyperplasia of respiratory epithelium
Very slight
0/45
5/44b
2/44
Slight
1/45
6/44
14/44c
Moderate
0/45
0/44
4/44
Focal stratified squamous metaplasia of respiratory epithelium
Very slight
2/45
10/44b
2/44
Slight
3/45
18/44c
26/44c
Moderate
1/45
5/44
14/44c
Severe
0/45
0/44
1/44
Rhinitis
8/45
11/44
23/44c
Simple or stratified cuboidal or
squamous metaplasia of epithelium in
the dorsomedial area where respiratory
and olfactory epithelium joina
0/45
2/44
23/44c
Focal replacement of olfactory epithelium by respiratory,
respiratory-like or regenerating olfactory epithelium
Very slight
0/45
0/44
1/44
Slight
0/45
0/44
12/44c
Moderate
0/45
0/44
12/44c
Severe
aSee above for explanation; bp < 0.05; cp < 0
0/45
.01
0/44
1/44
Woutersen et al. (1987)
Wistar rats; male and female;
10/sex/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers for
6 hours/day, 5 days/week for 13 weeks.
Test article: Paraformaldehyde.
Actual concentrations were 0,1.2 (±0.00),
11.9 (±0.15), or 24.4 (±0.09) mg/m3.a
Histopathology. sections of the lungs,
trachea, larynx (3 longitudinal) and nose (6
standard cross sections).
[Males] Histological changes in the nose at 13 weeks
Incidence of lesions
0
mg/m3
1.2
mg/m3
11.9
mg/m3
24.4
mg/m3
Respiratory epithelial squamous metaplasia
Diffuse
Slight
0/10
0/10
0/10
0/10
Moderate
0/10
0/10
0/10
5/10a
Severe
0/10
0/10
0/10
5/10a
Focal
Very slight
0/10
1/10
0/10
0/10
Slight
0/10
1/10
6/10a
0/10
Moderate
0/10
0/10
4/10
0/10
Focal respiratory epithelial hyperplasia
Very slight
0/10
0/10
1/10
1/10
Slight
0/10
0/10
6/10a
7/10b
Moderate
0/10
0/10
1/10
0/10
Focal respiratory epithelial disarrangement
Very slight
0/10
0/10
1/10
0/10
Slight
0/10
0/10
3/10
0/10
Moderate
0/10
0/10
1/10
0/10
Focal respiratory epithelial keratinization
Very slight
0/10
2/10
6/10a
1/10
Slight
0/10
0/10
3/10
6/10a
Moderate
0/10
0/10
0/10
1/10
Focal olfactory epithelial thinning
Slight
0/10
0/10
0/10
2/10
Moderate
0/10
0/10
0/10
1/10
Severe
0/10
0/10
0/10
5/10a
Focal olfactory epithelial squamous metaplasia
This document is a draft for review purposes only and does not constitute Agency policy.
1-180 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Slight
0/10
0/10
0/10
4/10
Moderate
0/10
0/10
0/10
4/10
Olfactory epithelial keratinization
Very slight
0/10
0/10
0/10
1/10
Slight
0/10
0/10
0/10
2/10
Rhinitis
0/10
2/10
5/10a
10/10b
Slight submucosal loosely
arranged connective tissue
0/10
0/10
0/10
2/10
Pharyngeal duct
mononuclear cell infiltrate
9/10
10/10
10/10
8/10
Nasolachrymal duct
sinusitis
3/10
6/10
7/10
2/10
Maxillary sinus sinusitis
ap < 0.05; bp < 0.01
7/10
3/10
4/10
2/10
[Females] Histological changes in the nose at 13 weeks
Incidence of lesions
0
mg/m3
1.2
mg/m3
11.9
mg/m3
24.4
mg/m3
Respiratory epithelial squamous metaplasia
Diffuse
Slight
0/10
0/10
0/10
3/10
Moderate
0/10
0/10
0/10
4/10
Severe
0/10
0/10
0/10
3/10
Focal
Very slight
0/10
0/10
1/10
0/10
Slight
0/10
1/10
7/10b
0/10
Moderate
0/10
0/10
2/10
0/10
Focal respiratory epithelial hyperplasia
Very slight
0/10
0/10
2/10
1/10
Slight
0/10
1/10
6/10a
6/10a
Moderate
0/10
0/10
0/10
0/10
Focal respiratory epithelial disarrangement
Very slight
0/10
0/10
2/10
1/10
Slight
0/10
1/10
6/10a
6/10a
Moderate
0/10
0/10
0/10
0/10
Focal respiratory epithelial keratinization
Very slight
0/10
0/10
6/10a
6/10a
Slight
0/10
0/10
2/10
4/10
Moderate
0/10
0/10
0/10
0/10
Focal olfactory epithelial thinning
Slight
0/10
0/10
0/10
2/10
Moderate
0/10
0/10
0/10
2/10
Severe
0/10
0/10
0/10
2/10
Focal olfactory epithelial squamous metaplasia
Slight
0/10
0/10
0/10
3/10
Moderate
0/10
0/10
0/10
1/10
Olfactory epithelial keratinization
Very slight
0/10
0/10
0/10
0/10
Slight
0/10
0/10
0/10
0/10
Rhinitis
0/10
0/10
3/10
2/10
Slight submucosal loosely
arranged connective tissue
0/10
0/10
0/10
4/10
This document is a draft for review purposes only and does not constitute Agency policy.
1-181 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Pharyngeal duct
mononuclear cell infiltrate
10/10
10/10
10/10
10/10
Nasolachrymal duct
sinusitis
3/10
5/10
2/10
4/10
Maxillary sinus sinusitis
1/10
1/10
5/10
0/10
ap < 0.05; bp < 0.01
Lung:
Histological changes (e.g., focal accumulation of alveolar
macrophages) in the lung were considered not to be
exposure related but as common findings in this strain and rat age.
Larynx:
Squamous
24.4 mg/m
11.9 mg/m
I.2 mg/m3
0 mg/m3—
Very slight
24.4 mg/m
II.9 mg/m
I.2 mg/m3
0 mg/m3—
Squamous
24.4 mg/m
II.9 mg/m
I.2 mg/m3
0 mg/m3—
Very slight
24.4 mg/m
II.9 mg/m
1.2 mg/m3
0 mg/m3—
metaplasia (males)
3—3/10, very slight; 1/10, slight; 1/10, moderate
3—no lesions observed
—no lesions observed
no lesions observed
keratinization (males)
3—2/10
3—no lesions observed
—no lesions observed
no lesions observed
metaplasia (females)
3—no lesions observed
3—not examined
not examined
no lesions observed
keratinization (females)
3—no lesions observed
3—not examined
—not examined
no lesions observed
Medium Confidence
Andersen et al. (2010)
Fischer 344; male; 8/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for 1, 4, or
13 weeks. Rats sacrificed immediately
after last exposure.
Test article: Paraformaldehyde.
Actual concentrations reported in the
Results column. Target concentrations
were 0, 0.8, 2.5, 7.4,12.3, or 18.5 mg/m3.a
Histopathology. nasal sections at the nose
-V were not reported.
Target and Actual FA Concentrations
Actual concentration (mg/m3)
for each exposure time
Target
1 week
4 weeks
13 weeks
0
0±0
0±0
0±0
0.8
0.77 ±0.06
0.8 ±0.09
0.83 ± 0.07
2.5
2.5 ±0.0
2.5 ±0.0
2.5 ±0.1
7.4
7.3 ±0.2
7.4 ±0.2
7.4 ±0.2
12.3
12.2 ±0.6
12.3 ±0.7
12.3 ±0.7
18.5
18.9 ±0.1
18.5 ±0.6
18.3 ±0.5
Incidence and severity of nasal squamous metaplasia0
tip and standard cross-section levels (l-V).
FA (target concentrations)
0
0.8
2.5
7.4
12.3
18.5
Main limitations: small N; data for levels
Region
mg/m3
mg/m3
mg/m3
mg/m3
mg/m3
mg/m3
Level I
1 week
4b (l)c
5(1)
8(1.9)
8(1.6)
8(1.5)
6(1.2)
4 weeks
1(1)
6(1)
7(1)
8(1.5)
8(1.7)
8(2.2)
13 weeks
1(1)
2(1)
8(1.1)
8(1.8)
8(1.9)
8 (2.4)
Level II
1 week
0 (NA)
0 (NA)
0 (NA)
6(1.1)
8(1.5)
8(1.5)
This document is a draft for review purposes only and does not constitute Agency policy.
1-182 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
4 weeks
0 (NA)
0 (NA)
0 (NA)
5(1)
8(1.2)
8(1.7)
13 weeks
0 (NA)
0 (NA)
0 (NA)
0 (NA)
8(2.9)
8 (3.4)
Data NR for levels III, IV, and V.
aSquamous metaplasia diagnosed in areas with change in transitional or
respiratory epithelium to squamous epithelium, with or without keratinization;
b8 animals examined at each time point and dose; cAverage severity score
(1 = minimal, 2 = slight/mild, 3 = moderate, 4 = moderately severe).
Incidence of nasal necrosis/erosion
FA (target concentrations)
0
7.4
12.3
18.5
Region
mg/m3
mg/m3
mg/m3
mg/m3
Level 1
1 week
0a
6
8
8
4 weeks
0
3
3
6
13 weeks
0
0
7
4
Level II
1 week
0
0
7
7
4 weeks
0
0
5
8
13 weeks
0
0
0
6
Lesions ND at 0.8 and 2.5 mg/m3.
a8 animals examined at each time point and dose.
Rusch et al. (1983)
Fischer 344 rats; male and female;
20/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers for
22 hours/day, 7 days/week for 26 weeks.
Test article: Unstabilized 5% solution of
formaldehyde (0.03% methanol).
Actual concentrations were 0.23 (±0.02),
1.2 (±0.1), or 3.6 (±0.22) mg/m3.a Controls
exposed to 0.011 (±0.009) mg/m3.
Histopathology: Four sections of lung, one
section of trachea, and three transverse
sections of nasal turbinates (anterior,
middle, and posterior regions) and one
transverse section of ethmoturbinate.
Main limitations: lesion severities were
NR; data only reported for one section;
metaplasia and hyperplasia reported
together.
Microscopic evaluation of lungs and trachea for Groups I, III, V, and VI showed
lesions frequently observed in laboratory animals but not considered
exposure-related. Electron microscopic evaluation for Group I and II animals
(5/sex) did not reveal turbinate, tracheal, or pulmonary ultrastructure changes
associated with treatment.
Observations in middle region of nasal turbinate
Group
Exposure
Squamous
metaplasia and
hyperplasia
Basal cell
hyperplasia
1 (control for
II and III)
0 mg/m3
2/38
0/38
II
0.23 mg/m3
1/38
0/38
III
1.2 mg/m3
3/36
0/36
V (control for
VI)
0 mg/m3
3/39
4/39
VI
3.6 mg/m3
23/37
25/37
For anterior nasal turbinates, no evidence of exposure-related effects with the
possible exception for Group VI. When comparing to Group V, fourfold increase
for the incidences of squamous metaplasia/hyperplasia and basal cell
hyperplasia in Group VI; for posterior nasal turbinates, only Group VI showed
evidence of squamous metaplasia (3/37); no evidence of exposure-related
effects in ethmoturbinates; level of rhinitis comparable in Groups I, II, and III, but
most frequent in Group VI.
Wilmeretal. (1989)
Wistar rats; male; 25/group.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers either
continuously for 8 hours/day, 5 days/week
for 13 weeks or intermittently 8 hours/day
(successive periods of 0.5 hour of exposure
Histopathological changes in respiratory epithelium (cross section
II) observed after 13 weeks of exposure
Incidence of lesions
A
B
C
D
E
Control
1.23
mg/m3
Contin.
2.46
mg/m3
Contin.
2.46
mg/m3
Inter.
4.92
mg/m3
Inter.
Disarrangement
This document is a draft for review purposes only and does not constitute Agency policy.
1-183 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
and 0.5 hour of nonexposure), 5 days/week
Focal
12/25
4/22
8/24
3/23a
8/25
for 13 weeks.
Diffuse
1/25
1/22
0/24
15/23c
ll/25b
Test article: Paraformaldehyde.
Necrosis
Actual concentrations were not
Focal
4/25
3/22
0/24
2/23
3/25
determined. Target concentrations were 0,
Diffuse
0/25
0/22
0/24
2/23
2/25
1.23, or 2.46 mg/m3 for continuous
Basal cell hyperplasia
exposures and 0, 2.46, or 4.92 mg/m3 for
Focal
9/25
4/22
6/24
11/23
10/25
intermittent exposures.3
Histopathology: 6 standard cross sections
Diffuse
4/25
0/22
0/24
4/23
11/25
Squamous metaplasia
of the nose [note: same as Woutersen
Focal
5/25
0/22
1/24
7/23
16/25b
etal. (1989)1
Keratinization
0/25
0/22
1/24
0/23
3/25
Nest-like infolds
Main limitations: analytical concentrations
Focal
5/25
4/22
11/24
14/23b
7/25
and lesion severities were not reported.
Diffuse
0/25
3/22
1/24
0/23
1/25
Goblet cell hyperplasia
Focal
0/25
1/22
1/24
2/23
1/25
Diffuse
5/25
2/22
8/24
13/23a
10/25
Rhinitis
3/25
2/22
3/24
16/23c
8/25
A = 0 mg/m3; B = 1.23 mg/m
continuous (9.8 mg/m3 h/d); C = 2.46 mg/m3
continuous (19.7 mg/m3 h/d); D = 2.46 mg/m3 intermittent (9.8 mg/m3 h/d);
E = 4.92 mg/m3 intermittent (19.7 mg/m3 h/d).
ap < 0.05; bp < 0.01; cp < 0.001.
Zwart et al. (1988)
[Data only reported for cross sections II and III]
Wistar rats; male and female;
3 days:
50/gro up/sex.
Exposure: Rats were exposed to FA in
dynamic whole-body chambers
6 hours/day, 5 days/week for 13 weeks.
Nose:
3.7 mg/m3—Focal basal cell hyperplasia concomitant with loss of cilia
observed at section III, number of rats and sex NR.
Histological changes NR for other groups.
Test article: Paraformaldehyde.
Actual concentrations were 0, 0.37 (±0.02),
13 weeks:
1.2 (±0.10), or 3.7 (±0.27) mg/m3.a
Histopathology: 6 standard cross sections
of the nose [note: same as Woutersen
Nose:
3.7 mg/m3—Histological changes including epithelial disarrangement to
epithelial hyperplasia and squamous metaplasia (with or without
keratinization) found in 37/50 males and 21/50 females. Changes
etal. (1989)1
localized to the anterior part of section
II that is normally covered by
respiratory epithelium.
Main limitations: failed to completely
Histological changes NR for other exposure groups at section II.
report lesion incidence and lesion
No histological changes in respiratory epithelium observed in section III for
severities were not reported.
any rat exposed to FA. Statistically significant differences in the
incidences of inflammatory lesions (e.g., rhinitis, sinusitis, and
aggregates of mononuclear cell infiltrates) in the pharyngeal ducts
observed between control and treatment groups, although
quantitative data NR and exposure-related response was absent.
3.7 mg/m3—Electron microscopic evaluation revealed: changes in nasal septa
epithelium including loss of cilia, but not slender microvilli; strongly
indented and disarranged epithelial eel
nuclei; the presence of small
blood vessels; interdigitations between epithelial cells and the
presence of cilia in intracellular spaces; foci of keratinized squamous
epithelium; and glandularization of goblet cells, which were arranged
in gland-like structures.
0.37 and 1.2 mg/m3—Electron microscopic evaluation of section II showed
no differences except for irregularly shaped and strongly indented
nuclei when compared to controls.
Mice
This document is a draft for review purposes only and does not constitute Agency policy.
1-184 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Medium confidence
Maronpot et al. (1986)
B6C3F1 mice; male and female;
10/sex/group.
Exposure: Mice were exposed to FA in
dynamic whole-body chambers for
6 hours/day, 5 days/week for 13 weeks.
Test article: Formalin (9.2% w/v), assumed
to contain methanol.
Actual concentrations were 2.41 (±0.25),
5.02 (±0.62), 12.4 (±0.80), 25.1 (±1.1), or
49.6 (±3.2) mg/m3.
Histopathology: sections of the nasal
turbinates (3 sections), larynx, trachea, and
lung.
Main limitations: formalin; small N
Lesions after 13 weeks of exposure:
mg/m3:
0
5.02
12.4
25.1
49.6
Nasal cavity
M
F
M
F
M
F
M
F
M
F
Metaplasia,
0/10
0/10
1/10
0/10
10/10
10/10
10/10
10/10
10/10
10/10
squamous
Inflammation,
0/10
0/10
0/10
0/10
4/10
0/10
10/10
8/10
10/10
10/10
seropurulent
No lesions observed after exposure to 2.41 mg/m3.
mg/m3:
0
25.1
49.6
M
F
M
F
M
F
Larynx
Metaplasia,
0/8
0/8
6/9
3/9
10/10
7/8
squamous
Trachea
Metaplasia,
0/10
0/9
3/10
5/10
10/10
10/10
squamous
Hyperplasia,
0/10
0/9
4/10
2/10
2/10
0/10
epithelial
Inflammation,
0/10
0/9
0/10
0/10
8/10
5/10
purulent
Fibrosis,
0/10
0/9
0/10
0/10
9/10
5/10
submucosal
Lung
Bronchus,
0/10
0/10
0/10
0/10
4/10
3/10
metaplasia
squamous
Bronchus,
0/10
0/10
0/10
0/10
3/10
2/10
inflammation
Bronchus, fibrosis,
0/10
0/10
0/10
0/10
2/10
0/10
submucosal
No laryngeal lesions observed after 2.41, 5.02, or 12.4 mg/m3; no tracheal
lesions observed after 2.41, 5.02, or 12.4 mg/m3, except 1/10 females
(squamous metaplasia) after 12.4 mg/m3; no lung lesions after 12.4
mg/m3; data were NR for 2.41 and 5.02 mg/m3.
Hamsters
Medium confidence
Rusch et al. (1983)
Syrian golden hamsters; male and female;
10/sex/group.
Exposure: Hamsters were exposed to FA in
dynamic whole-body chambers for
22 hours/day, 7 days/week for 26 weeks.
Test article: Unstabilized 5% solution of
formaldehyde (0.03% methanol).
Actual concentrations were 0.23 (±0.02),
1.2 (±0.1), or 3.6 (±0.22) mg/m3.a Controls
were exposed to 0.011 (±0.009) mg/m3.
Histopathology: 4 sections of lung, 1
section of trachea, and the hamster
equivalent of the rat turbinate sections
(i.e., 3 transverse sections of nasal
Microscopic evaluation of lungs and trachea for Groups I (controls for Groups II
and III), III (1.2 mg/m3), V (controls for Group VI), and VI (3.6 mg/m3) showed
lesions frequently observed in laboratory animals but not considered exposure
related. Histopathological data for Group II (0.23 mg/m3) not reported.
No evidence of exposure-related effects for the incidence of squamous
metaplasia even at 3.6 mg/m3 exposure level.
This document is a draft for review purposes only and does not constitute Agency policy.
1-185 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
turbinates [anterior, middle, and posterior
regions] and one transverse section of
ethmoturbinate).
Main limitations: lesion incidences NR
(note: only metaplasia was investigated).
Monkeys
Medium confidence
Rusch et al. (1983)
Cynomolgus monkeys; male; 6/group.
Exposure: Monkeys were exposed to FA in
dynamic whole-body chambers for
22 hours/day, 7 days/week for 26 weeks.
Test article: Unstabilized 5% solution of
formaldehyde (0.03% methanol).
Actual concentrations were 0.23 (±0.02),
1.2 (±0.1), or 3.6 (±0.22) mg/m3. Controls
exposed to 0.011 (±0.009) mg/m3.a
Histopathology: 4 sections of lung, 1
section of trachea, and the monkey
equivalent of the rat turbinate sections
(i.e., 3 transverse sections of nasal
turbinates [anterior, middle, and posterior
regions] and one transverse section of
ethmoturbinate).
Main limitations: lesion severities NR;
incidence of squamous metaplasia and
hyperplasia reported together; data
reported for only one nasal section.
Microscopic evaluation of lungs and trachea for Groups I (controls for Groups II
and III), III (1.2 mg/m3), V (controls for Group VI), and VI (3.6 mg/m3) showed
lesions frequently observed in laboratory animals but not considered exposure
related. Histopathological data for Group II (0.23 mg/m3) not reported.
Observations in middle region of nasal turbinate
Group
Exposure
Squamous metaplasia
and hyperplasia
1 (control for II and III)
0 mg/m3
0/6
II
0.23 mg/m3
0/6
III
1.2 mg/m3
1/6
V (control for VI)
0 mg/m3
0/6
VI
3.6 mg/m3
6/6
For anterior and posterior nasal turbinates, no exposure-related effects
reported.
Rhinitis observed in numerous animals from all Groups but with no apparent
exposure-response.
For Group VI, observations of hoarseness, congestion, and nasal discharge were
reported.
Abbreviations: FA = formaldehyde; NR = not reported, SD = standard deviation.
aStudy authors originally reported FA concentrations in ppm. These values were converted based on
1 ppm = 1.23 mg/m3, assuming 25°C and 760 mm Hg.
This document is a draft for review purposes only and does not constitute Agency policy.
1-186 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-28. Selected short-term respiratory pathology studies in animals (see
Appendix A.5.5 for others)
Reference and study design
Results
Rats
High confidence
Kuper et al. (2011)
Fischer rats; males; 8/group.
Exposure: Mice were exposed to FA in
dynamic whole-body chambers 6 hours/day,
Incidence of lesions/changes after 4 weeks
FA (mg/m3)
NALT
0
0.63
1.23
2.48
7.53
12.3
18.4
Size
5 day/week for 4 weeks.
Very small
1
0
0
1
0
2
1
Test article: Formalin (10.21% FA; although
Small
2
1
2
2
3
3
6
NR, the description supports the assumption
Medium
2
7
5
5
5
3
1
that it was freshly prepared).Actual
Large
3
0
1
0
0
0
0
concentrations were 0, 0.63 (±0.06), 1.23
Decreased cellularity
(±0.14), 2.48 (±0.18), 7.53 (±0.42), 12.3
Slight
0
0
0
0
0
0
1
(±0.48), and 18.4 (±0.06) mg/m3.a
Moderate
0
0
0
1
0
0
2
Histopathology: 2 sections of
Germinal center development
nasopharynx-associated lymphoid tissues
Very slight
1
5
3
3
3
3
0
(NALT) and one section of an upper
Moderate
3
0
0
0
0
0
0
respiratory tract-draining lymph node
(i.e., posterior and superficial cervical lymph
Score expanded
total
4
5
3
3
3
3
0
nodes).
Epithelial hyperplasia
Note: small N
Slight
0
0
0
0
0
0
2
Moderate
0
0
0
0
0
0
5
Score expanded
total
ap<0.01.
0
0
0
0
0
0
7a
Incidence of lesions/changes after 4 weeks
FA (mg/m3)
0
0.63
1.23
2.48
7.53
12.3
18.4
Posterior cervical lymph nodes
Germinal center development
Very slight
3
3
2
4
4
5
5
Slight
0
1
2
1
2
0
0
Moderate
1
0
1
0
0
0
0
Marked
1
0
0
0
0
0
0
Very marked
0
1
0
0
0
1
1
Score expanded
totals
5
5
5
5
6
6
6
Superficial cervical lymph nodes
Germinal center development
Very slight
5
3
2
0
3
1
0
Very marked
0
0
1
0
0
0
0
Score expanded
totals
ap< 0.05.
5
3
3
0a
3
1
0a
This document is a draft for review purposes only and does not constitute Agency policy.
1-187 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Medium confidence
Wilmeretal. (1987)
Wistar rats; male; 10/group.
Exposure: Rats were exposed to FA in a
dynamic whole-body chamber either
continuously for 8 hours/day, 5 days/week
for 4 weeks or intermittently 8 hours/day
(successive periods of 0.5 hour of exposure
and 0.5 hour of nonexposure), 5 days/week
for 4 weeks.
Test article: Paraformaldehyde.
Actual concentrations were not determined.
Target concentrations were 0, 6.2, or 12.3
mg/m3 for continuous exposures and 0,
12.3, or 24.6 mg/m3 for intermittent
exposures.1
Histopathology. 6 standard nasal cross
sections.
Main limitations: analytical concentrations
NR; lesion incidence and severities NR
Respiratory epithelium:
Focal thinning and disarrangement of mainly the lateral wall observed in all
animals exposed to 24.6 mg/m3.
Squamous metaplasia and basal cell hyperplasia observed mainly in 12.3 and
24.6 mg/m3.
Rhinitis (minimal to moderate) observed in all groups.
intermittent exposure to 24.6
mg/m3 (98.4 mg/m3-h/day)
>
continuous exposure to 12.3
mg/m3 (98.4 mg/m3-h/day)
intermittent exposure to 12.3
mg/m3 (49.2 mg/m3-h/day)
>
continuous exposure to 6.2
mg/m3 (49.6 mg/m3-h/day)
intermittent exposure to 12.3
mg/m3 (49.2 mg/m3-h/day)
=
continuous exposure to 12.3
mg/m3 (98.4 mg/m3-h/day)
Mice
High confidence
Kuper et al. (2011)
B6C3F1 mice; females; 6/group.
Exposure: Mice were exposed to FA in
dynamic whole-body chambers 6 hours/day,
5 day/week for 4 weeks.
Test article: Formalin (10.21% FA; although
NR, the description supports the assumption
that it was freshly prepared).
Actual concentrations were 0, 0.63 (±0.06),
1.23 (±0.14), 2.48 (±0.18), 7.53 (±0.42), 12.3
(±0.48), and 18.4 (±0.06) mg/m3.a
Histopathology: 2 sections of
nasopharynx-associated lymphoid tissues
(NALT) and one section of an upper
respiratory tract-draining lymph node
(i.e., posterior and superficial cervical lymph
nodes).
Note: small N
Group
Observation
Controls
NALT: varied in size from small to large; scarce germinal
centers
Exposed
Posterior and cervical lymph nodes: no FA-related changes
NALT: no FA-related changes; no significant change in size
compared to controls; scarce germinal centers
Medium confidence
Morgan et al. (2017)
C3B6.129Fl-Trp53tmlBrd (C3B6TP53±) and
B6.129-Trp53tmlBrd (B6 TP53±) mice; males;
24-35/group
Exposure: Mice were exposed to FA in
dynamic whole-body chambers 6 hours/day,
5 day/week for 8 weeks.
Test article: Paraformaldehyde
Incidence (and severity) of noncancer nasal lesions at 32 weeks post-
exposure
FA (mg/m3)
0
9.23
18.45
C3B6 TP53± mice
Squamous Metaplasia
(respiratory epithelium)
0/21
14/21 (1.2)
22/23 (1.5)
Hyperplasia (respiratory
epithelium)
0/21
0/21
1/23 (1.0)
This document is a draft for review purposes only and does not constitute Agency policy.
1-188 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Osteogenesis (turbinate)
0/21
0/21
3/23 (3.0)
B6 TP53± mice
Squamous Metaplasia
(respiratory epithelium)
0/22
13/27(1.0)
17/26(1.5)
Osteogenesis (turbinate)
0/22
1/27(1.0)
1/26(1.0)
Nominal concentrations were 0, 9.23, or
18.45 mg/m3.a
Histopathology: 3 sections of the nasal
cavity and one section of the larynx
Main limitations: somewhat limited
sampling and minor reporting limitations;
potentially short duration (however, lesions
are observed)
Average severity score based on 1= minimal; 2= mild; 3= moderate; 4= marked
No laryngeal lesions were reported
Monkeys
Medium confidence
Monticello et al. (1989)
Rhesus monkeys; males; 3/group.
Exposure: Monkeys were exposed to FA in
dynamic whole-body chambers 6 hours/day,
5 days/week for 1 or 6 weeks.
Test article: Paraformaldehyde.
Actual concentrations were not determined.
Target concentration was 0 or 7.4 mg/m3.a
Histopathology. 5 transverse sections of the
nasal passages (A-E) extending from the
nares to the soft palate. The evaluation also
included cross sections of larynx and mid-
trachea, a frontal section of the carina, and
sections of all lung lobes, which were
trimmed mid-sagitally to include airway
bifurcations.
Main limitations: analytical concentrations
NR; lesion incidence and severities NR
Exposure
Control
7.4 mg/m3
1-week
7.4 mg/m3
6-week
Observations (truncated from original article)
Nasal passages
Four types of epithelium lining rhesus nasal passages were
identified:
(1) stratified squamous in the vestibule (Level A); (2)
transitional (Level A), present in narrow zone just posterior
to vestibule; (3) olfactory in mid-dorsal region (Levels B-D);
and (4) respiratory, the most extensive (Levels B-E) and
present throughout remaining areas.
Extranasal respiratory tract
Typical pseudostratified columnar respiratory epithelium
observed for the larynx, trachea, and major bronchi; mild
inflammatory changes from pulmonary acariasis in one
monkey
Nasal passages
Characteristic changes in respiratory epithelium described
as generally being bilaterally symmetrical and consistent in
nature and severity for all three monkeys in group
Changes included loss of goblet cells and cilia, minimal-to-
mild epithelial hyperplasia with or without early stages of
squamous metaplasia, and an accompanying neutrophilic
inflammatory response
Squamous metaplasia present in various stages;
metaplastic epithelium eroded (mild) in some areas;
neutrophils occasionally found in metaplastic epithelium;
maxillary sinuses exhibited no treatment-related lesions
Extranasal respiratory tract
Lesions of larynx, trachea, and carina were considered mild
and included multifocal loss of cilia; extent of lesions
covering surface area of larynx/trachea of 1-week group
(3.0 ± 1.3%) was minimal compare to 6-week group
(26.0 ± 10.0%); no treatment-related lesions in lungs
Erosions absent; mild squamous metaplasia (more
developed than in 1-week group); maxillary sinuses
exhibited no treatment-related lesions; in two monkeys,
olfactory epithelium exhibited small discrete areas of mild
squamous metaplasia close to olfactory/respiratory
epithelial interface
This document is a draft for review purposes only and does not constitute Agency policy.
1-189 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Extranasal respiratory tract
Lesions included multifocal areas of respiratory mucosa
with loss of cilia and goblet cells, mild epithelial
hyperplasia, and early squamous metaplasia with
occasional squamous cell formation on the surface; no
treatment-related lesions in lungs
Exposure
Morphometric analysis of monkey nasal passages
7.4 mg/m3
1-week
Anterio-posterior severity gradient for percentage of
surface area with treatment-related lesions
7.4 mg/m3
6-week
Of all nasal passage regions, middle turbinate had
greatest percentage of surface area affected
Greater respiratory epithelium surface area with
treatment-related lesions compared with 1-week group
(p < 0.05)
More extensive lesions in the posterior nasal passages
(Levels D-E) and larynx/trachea compared with same
locations in 1-week group (p < 0.05)
7.4 mg/m3
1- and
6-week
Anterior regions (Levels B-C) had highest percentage of
nasal mucosal surface area with treatment-related lesions
Abbreviations: FA = formaldehyde, NA = not applicable, ND = not detected, NR = not reported, SD = standard
deviation, SE = standard error of the mean.
aStudy authors originally reported FA concentrations in ppm. These values were converted based on
1 ppm = 1.23 mg/m3, assuming 25°C and 760 mm Hg.
Evidence on Mode of Action for Respiratory Tract Pathology
Based primarily on studies in experimental animals or acutely exposed human volunteers
(most of these endpoints are difficult to examine in long-term observational epidemiology studies),
induction of histopathological lesions in the respiratory tract following formaldehyde exposure
appears to result, at least in part, from a series of increasingly severe effects, including altered
mucociliary function, damage to the nasal epithelium (e.g., sustained cytotoxicity), and sustained
reparative cell proliferation culminating in a hyperplastic epithelium, or transitioning to an
adaptive, metaplastic tissue (see Figure 1-18; see Appendix A.5.6 for additional details). Consistent
with observations of metaplasia without hyperplasia in many of the rodent health effect studies,
this pathway illustrates that metaplasia may develop following damage to the epithelium in the
absence of hyperplasia (i.e., hyperplasia may not be an essential precursor). All the mechanistic
events and relationships between events in the proposed pathway are based on robust or moderate
evidence, indicating that this is likely a mechanism by which formaldehyde exposure causes
squamous metaplasia. However, because modification of epithelial cell health and function in the
URT can occur via multiple direct and indirect mechanisms following formaldehyde inhalation,
which are expected to vary due to differences in both exposure duration and intensity, there are
likely to be other plausible mechanisms by which formaldehyde exposure could cause this health
This document is a draft for review purposes only and does not constitute Agency policy.
1-190 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
effect. The current understanding provides strong biological support for an association between
formaldehyde exposure and respiratory tract pathology. Additionally, as many of the mechanistic
events in this pathway have been observed in both humans (sometimes indirectly) and
experimental animals, including effects on mucociliary function and cell proliferation, as well as
evidence of elevated oxidative stress, findings from experimental animals are considered relevant
to humans.
Possible Initial Alterations Secondary Alterations
^ oxidative URT protein/ DNA URT mucociliary
stressinURT modification dysfunction
Effector-Level Changes
URT epithelial
damage
URT epithelial
proliferation
Key Hazard Feature
H
Squamous
metaplasia
Legend evidence relationship
jj, Plausiblyan initial if |Robust ~> Robust
effect of exposure
(_) Moderate --> Moderate
~ Key feature of respiratory
tract pathology O Slight Slight
Figure 1-18. Possible mechanistic associations between formaldehyde
exposure and respiratory tract pathology.
An evaluation of the formaldehyde exposure-specific mechanistic evidence informing the potential for
formaldehyde exposure to cause respiratory health effects (see Table 1-29 and Appendix A,5,6) identified
this sequence of mechanistic events as likely to be a mechanism by which formaldehyde inhalation could
cause respiratory tract pathology, specifically squamous metaplasia, although it is assumed that other
plausible pathways explaining this association have yet to be defined.
Some uncertainties remain regarding this pathway. Effects on the mucociliary system are
likely secondary to the production of reactive byproducts in the URT or covalent modification to
mucosal structural components following physical interactions of formaldehyde with proteins in
the mucus, the latter of which at least would be expected to be driven largely by concentration. The
nasal mucociliary apparatus cleans the airways by moving contaminant-laden mucus out of the
URT. When damage to the cilia slows or disrupts the movement of the mucus, formaldehyde or
other reactive molecules dissolved into the mucus may accumulate to a concentration that may be
overtly toxic to the cells beneath the mucus. Thus, alterations to this normally protective apparatus
could allow for greater access of inhaled formaldehyde (and other inhaled chemical and
nonchemical substances] to epithelium lining the nasal passages ( arkcma et al.. 20061.
Conversely, gradual tissue changes following exposure might also lead to resilience (e.g., increases
in epithelial cell barrier function). Unfortunately, animal studies of mucociliary function and other
detailed mechanistic studies characterizing the initial molecular interactions of formaldehyde in the
URT following long-term exposure are unavailable. However, given the formaldehyde removal and
This document is a draft for review purposes only and does not constitute Agency policy,
1-191 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
metabolism processes in the nasal respiratory epithelium (see Appendix A.2), it would generally be
expected that low levels of formaldehyde would be rapidly detoxified in healthy tissues, noting that
changes in mucus flow patterns have been observed at lower formaldehyde levels than those
eliciting URT epithelial lesions (i.e., at <0.3 mg/m3 in exposed humans and >0.6mg/m3 in animals).
Relatedly, while both hyperplasia and metaplasia, which generally represent attempts to
protect the nasal epithelium from further insult, are often correlated with areas of cell proliferation
(see Appendix A.5.6), similar evaluations were not identified for lesions such as necrosis. Although
cell proliferation can occur in response to tissue damage, the concentrations at which cytotoxicity
and tissue damage begin to occur are poorly defined compared to other respiratory tract lesions
(i.e., hyperplasia; metaplasia), partly due to differences in methodology and reporting across
studies. This complicates the interpretation of the potential progression (at least in terms of
concentration) of these URT changes. Regardless, since increases in cell proliferation are largely
adaptive responses to replace damaged and dying cells within the epithelial tissue layer, and
proliferation is typically not observed below 1.23 mg/m3 (note: while proliferation is clearly
increased above ~3.7 mg/m3, results across studies are mixed between 1.23 and 3.7 mg/m3; see
Appendix A.5.6), cellular damage-induced proliferation at similar levels is assumed to represent an
important mechanistic component for the development of URT pathology.
Interestingly, cellular proliferation "rates" (i.e., the available studies labeled dividing cells
only during the last few days of exposures that varied in duration) did not appear to be strongly
influenced by exposure duration (see Appendix A.5.6). Although differences exist, the general
pattern of proliferation was similar across sets of studies exposing rats for either <1 week,
1-6 weeks, or >12 weeks. This similarity adds further support that cellular damage or pathology
resulting in cell proliferation (i.e., hyperplasia) may not be highly dependent on exposure duration;
it remains unclear whether the cumulative proliferative potential (i.e., proliferative events across
the entire duration of exposure) might vary more strongly as a function of exposure duration, or to
what extent this association might hold for lesions that may not be as dependent on proliferation
(e.g., metaplasia). The broader implications of this relationship are discussed elsewhere (see
Sections 1.2.5 (Evidence on MOA for URT cancers) and 2.2.1 and Appendix B.2.2).
In addition, there are potential modifying factors that are not illustrated in Figure 1-18. One
significant uncertainty relates to the potential for inflammatory and immunological changes in the
upper airways (see Sections 1.2.2 and 1.2.3), which generally have been observed only after longer
formaldehyde exposures, to modify the pattern or progression of mechanistic changes leading to
the development of respiratory tract pathology. This understanding is further complicated because
the available data are limited both in terms of understanding the specific initiating events leading to
upper airway inflammatory changes, as well as their ability to clearly define the concentration and
duration requirements for effects on URT immunological processes. As with the other examined
health effects, uncertainties also exist regarding interindividual sensitivity to these effects, with
respiratory health status and sensitivity to allergens expected to be strong modifiers of these
This document is a draft for review purposes only and does not constitute Agency policy.
1-192 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
effects. Nasal lesions are far more severe in rodents with prior nasal damage (e.g., Woutersenetal..
1989: Appelman et al.. 19881. and similar observations have been made in exposed humans fe.g..
Falk etal.. 19941. while changes in mucus flow and related nasal features in allergic individuals
would be expected to modify the more direct effects of formaldehyde on the mucociliary apparatus.
Genetics may also play a role. For example, possibly complementing the hypothesized role of p53
in nasal genotoxicity (see Appendix A.4), two strains of p53 deficient mice (Trp53 heterozygotes)
exhibited pronounced metaplasia after short-term (8-week) exposure fMorgan etal.. 20171:
however, this study did not include metaplasia rates in wild-type mice for comparison16 and there
are no corresponding rat models, which would be presumed to be even more sensitive.
Overall, although uncertainties remain, the mechanistic evidence supports the conclusion
that metaplasia and hyperplasia are likely to result, at least in part, from direct or indirect
(e.g., through disruption of normal mucociliary function) effects on epithelial cell health, which
often appears to involve sustained cellular proliferation, particularly for hyperplasia.
Table 1-29. Mechanistic evidence most informative to the development of
respiratory tract pathology after formaldehyde inhalation
Endpoint
Study-specific findings and confidence
Summary of
evidence
Conclusion
The majority of these mechanistic changes have been discussed in previous sections.
See Table 1-3 for presentation of the evidence for:
URT oxidative stress (moderate)
See Table 1-10 for presentation of the evidence for:
URT protein/DNA modification (robust); URT mucociliary dysfunction (robust); and URT epithelial damage (robust)
¦f URT
Cellular
(epithelial)
Prolifera-
tion
(see
Appendix
A.5.6 for
additional
detail and
discussion)
High or Medium
Human: None (note: indirect data from human studies indicating
an increase in histopathological scores that included hyperplasia
were not specific enough to independently evaluate
proliferation).
Increased cell
proliferation in rats at
all tested durations.
Proliferation increases
were typically observed
in the anterior nasal
cavity at tested levels
>~3.5-4 mg/m3, and
were generally not
observed at <1.23
mg/m3. Sites of
proliferation correlated
with the development
of hyperplasia and
metaplasia, although
the temporal and
exposure levels
specifics of this
association are unclear.
Indirect data from
observations of
Robust
Animal: Acute dose-dependent increases in cell proliferation in
rats, measured primarily by DNA labeling during the final days of
exposure, were consistently observed following acute, short-
term, and subchronic exposure, and generally with a similar
magnitude of responses across durations. Proliferation was
typically highest in anterior regions (e.g., "level 2"), with little
evidence of proliferation at <1.23 mg/m3, mixed findings
between 1.24 and 3.5 mg/m3, and studies generally reporting
increases with exposure at higher levels, particularly with longer
exposure duration. These data are supported by consistent
observations of formaldehyde exposure-induced increases in
hyperplasia in pathology studies, some of which provided
information showing a correlation between acute proliferation
and hyperplasia and metaplasia. The only rat study that
measured exposure longer than 13 weeks suggests that increases
in acute proliferation may begin to decrease in magnitude with
chronic exposure at >6 mg/m3 (Monticello et al., 1996). A
16Lesion frequency or severity in the study by NTP (20171 was not noticably different from the other
available studies of wild-type mice similarly exposed to >9 mg/m3 (i.e., 12.4 and 17.6 mg/m3) for subchronic
[e.g., (Maronpot et al.. 19861] or chronic [e.g., (Kerns et al.. 19831] duration.
This document is a draft for review purposes only and does not constitute Agency policy.
1-193 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Toxicological Review of Formaldehyde—Inhalation
Endpoint
Study-specific findings and confidence
Summary of
evidence
Conclusion
few studies suggest that mice may exhibit less robust responses
than rats, while monkeys may exhibit proliferation in more
posterior nasal regions at >7 mg/m3.
hyperplasia in exposed
animals and humans
are consistent with
these data.
Low
N/A: Sufficient information for 'robust' from high or medium confidence studies.
Integrated Summary of Evidence on Respiratory Tract Pathology
The literature on formaldehyde effects on respiratory tract pathology in animals provides
robust evidence that inhaled formaldehyde exposure can induce histopathologic lesions in the URT
of animals, primarily in the nasal cavity, in a manner dependent on both the concentration and, to a
lesser extent (particularly for hyperplasia), duration of exposure. Based on numerous high and
medium confidence studies of chronic and subchronic exposure duration, formaldehyde exposure
resulted in lesions in the respiratory epithelium, including goblet and basal epithelial cell
hyperplasia, necrosis, and squamous metaplasia (see Tables 1-26 and 1-27). These lesions have
been observed across experimental animal species, including monkeys, mice, and hamsters, but
primarily in rats. In general, rats appear to be more sensitive than mice or hamsters, while the
limited data in monkeys suggest a similar sensitivity to rats with possible differences in lesion
location. While these lesions consistently develop in rodents of both sexes, several studies suggest
an increased susceptibility of males as compared to females, potentially due to differences in
breathing patterns. Presumably due to the high reactivity and water solubility of formaldehyde,
these pathological lesions have been primarily assessed (and subsequently observed) in the
epithelium lining the anterior regions of the rodent nasal passages following formaldehyde
inhalation exposure, mostly in regions containing respiratory epithelium. Generally, at higher
concentrations or longer durations, similar effects are seen in more posterior sections of the nasal
cavity (and sometimes beyond), as well as in the olfactory epithelium. Additionally, lesions
progress in severity (e.g., slight to moderate) at specific anatomical locations (e.g., cross-section
level) with increasing concentration or duration of exposure, indicating cumulative effects. While
several studies support that an increased incidence of nasal lesions such as hyperplasia and
metaplasia persists after cessation of exposure, partial regression (e.g., a reduced severity or
smaller increase in incidence) of these lesions appears to occur, at least in mice and rats.
Although the evidence is more equivocal in one study (Bovsen etal.. 1990). the four human
epidemiology studies examining histopathology found that participants exposed to average
formaldehyde levels between 0.05 and 0.6 mg/m3 had a higher average histopathology score than
their respective comparison group fBallarin et al.. 1992: Holmstrom etal.. 1989c: Edlingetal..
19881. Although the studies were limited by probable survival bias, and in some cases other
limitations that resulted in a bias toward the null, a consistent association with histopathological
endpoints, including squamous metaplasia, was observed. Therefore, the observational human
This document is a draft for review purposes only and does not constitute Agency policy.
1-194 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
data provide moderate evidence that inhaled formaldehyde induces histopathological lesions in the
URT.
Mechanistic insights based on a large amount of animal data (some similar effects were
observed in humans, although the data were sparse) indicate a likely role for altered mucociliary
function or cellular proliferation in the occurrence of these exposure-induced lesions
(see Appendix A.5.6). Overall, the strength of the evidence for hyperplasia and squamous
metaplasia includes robust evidence from animal studies and moderate human evidence from
observational epidemiology studies, and strong support for a plausible MOA based largely on
mechanistic evidence in animals (supported by more limited, coherent findings in human
mechanistic studies), Therefore, the evidence demonstrates that inhalation of formaldehyde
causes respiratory tract pathology in humans given the appropriate exposure circumstances. The
primary basis for this conclusion is rat bioassays of chronic exposure that consistently observed
squamous metaplasia at formaldehyde exposure levels >2.5 mg/m3.
Table 1-30. Evidence integration summary for effects of formaldehyde
inhalation on respiratory pathology
Evidence
Evidence judgment
Hazard determination
Human
Moderate based on:
Human health effect studies:
Of the four occupational studies interpreted with medium confidence (less
sensitive due to healthy survival bias), 3 observed a higher prevalence of
abnormal nasal histopathology, including loss of ciliated cells, hyperplasia,
and squamous metaplasia at concentrations ranging from 0.1-2 mg/m3,
while the remaining (1) study had more equivocal findings.
Biological plausibility:
Mechanistic changes in two studies (one interpreted with medium
confidence) in humans provides evidence of changes in mucociliary
clearance and mucus flow beginning at formaldehyde concentrations of
0.25-0.3 mg/m3.
The evidence demonstrates that
inhalation of formaldehyde
causes respiratory tract
pathology in humans given
appropriate exposure
circumstances3
Primarily based on rat bioassays
of chronic exposure which
consistently observed squamous
metaplasia at formaldehyde
exposure levels >2.5 mg/m3.
Potential Susceptibilities:
Variation in sensitivity may
depend on differences in URT
immunity, allergen sensitivity,
and nasal structure or past
injury (e.g., studies support
increased sensitivity of rodents
with intentionally damaged
nasal cavities), and males may
be more sensitive than females.
Animal
Robust, based on:
Animal health effect studies:
• Consistent evidence of squamous metaplasia and hyperplasia in the nasal
respiratory epithelium across numerous independent studies interpreted
with high or medium confidence, with generally the most sensitive effects
being metaplasia observed after chronic exposure to >2.5 mg/m3
formaldehyde.
• Evidence of both metaplasia and hyperplasia in monkeys, rats, mice, and
hamsters; the data were more limited for monkeys; mice and hamsters
exhibited less sensitivity.
• Multiple studies provided clear evidence of a concentration dependence
for lesion development, as demonstrated by increases in the incidence,
severity, and anatomical location of the observed lesions with increasing
exposure.
Biological plausibility:
Robust or moderate evidence for mechanistic events based predominantly
on experimental animal studies supports a biological progression of changes
that appears to include mucociliary dysfunction, epithelial damage, and
This document is a draft for review purposes only and does not constitute Agency policy.
1-195 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
often cellular proliferation, leading to the eventual development of nasal
lesions, including squamous metaplasia.
Other
inferences
• Relevance to humans: Similarities in the function and properties of the
nasal epithelium across species, as well as similar mechanistic and apical
effects observed in both humans and animals, provide strong support for
the relevance of the findings in experimental animals to humans.
• MOA\ Although it may be incomplete, a MOA involving effects on
mucociliary function and epithelial cell health is well supported and
considered to be a major contributor to these effects.
• Other: Data from animal studies suggest that lesion development may be
driven more by concentration than duration, particularly for hyperplasia.
While estimates for formaldehyde were not identified, estimates for other
irritants indicate that concentration is ~1.8- to 1.9-fold (on average) more
influential than duration regarding exposure-induced mortality after acute
exposure.
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2.
1.2.5. Respiratory Tract Cancers
This section examines the evidence pertaining to the carcinogenic effect of formaldehyde
exposure on the upper respiratory tract (URT) of humans and animals. The specific endpoints
considered in this section include diagnoses of nasopharyngeal cancer, sinonasal cancer, cancers of
the oropharynx and hypopharynx, and laryngeal cancer in exposed humans; experimental animal
studies examining the potential for cancers of the nasal cavity and proximal regions of the URT
(note: the results of several studies that also included examinations of more distal regions of the
respiratory tract are discussed); and mechanistic studies relevant to interpreting potential
carcinogenic effects on the URT. In humans, URT cancers were reviewed independently of one
another based on primary data from case-control and cohort studies (the approximate structural
delineations referred to in the section on human evidence are shown below in Figure 1-19).
Epidemiological findings provide robust evidence for nasopharyngeal cancers (NPCs), and
sinonasal cancer, based on groups with occupational exposure. Epidemiological evidence is slight
for oropharyngeal/hypopharyngeal cancers, and inadequate for laryngeal cancers, respectively.
Evidence for a carcinogenic effect in the URT of humans is further supported by experimental
animal studies. Precancerous lesions (e.g., dysplasia) and tumors (primarily squamous cell
carcinomas) were observed in the nasal cavities of multiple species/strains of rodents. Such
observations in animals were concentration and duration dependent. Mechanistic data suggest that
URT cancers are likely the result of genotoxicity and mutagenicity, cytotoxicity, and cell
proliferation. Together, genotoxicity, cellular proliferation, and cytotoxicity-induced regenerative
proliferation exhibit multiple layers of coherence as a function of species, anatomy, temporality,
concentration, and duration of exposure, and when these factors are integrated, they form a
biologically relevant MOA for formaldehyde-induced URT carcinogenesis.
The evidence demonstrates that formaldehyde inhalation causes nasopharyngeal cancer
(NPC) in humans, given appropriate exposure circumstances, based on robust epidemiological
This document is a draft for review purposes only and does not constitute Agency policy.
1-196 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
evidence of an increased risk of the occurrence of NPCs from studies of occupational formaldehyde
exposure in several geographic locations among different occupational populations representing
diverse exposure settings; robust evidence from long-term bioassays in two animal species
providing consistent and reliable evidence of nasal cancers following exposure; and reliable and
consistent mechanistic evidence in both animals and humans supporting causality. The
nasopharynx, although not typically specified in animal studies, is the region adjacent to the nasal
cavity, where the animal evidence was predominantly observed, providing plausible coherence
between the animal and human data (and thus, the animal evidence is reflected as robust for the
purpose of interpreting human NPC). The evidence is sufficient to conclude that a mutagenic mode
of action of formaldehyde is operative in formaldehyde-induced nasopharyngeal carcinogenicity.
The evidence demonstrates that formaldehyde inhalation causes sinonasal cancer (SNC}
in humans, given appropriate exposure circumstances, based on robust epidemiological evidence of
an increased risk of the occurrence of sinonasal cancer from studies of occupational formaldehyde
exposure in several geographic locations among different occupational populations representing
diverse exposure settings. This evidence is supported by the apical and mechanistic evidence for
nasal cancers across multiple animal species, although some uncertainty remains in the
interpretation of the animal nasal data as wholly applicable to interpreting sinonasal cancer (and
thus, the animal evidence is reflected as moderate for the purpose of interpreting human SNC),
Although uncertainties remain, the nasal cancer MOA, including mutagenicity, is interpreted as
relevant to this cancer type.
- Pharynx
Figure 1-19. Schematic diagram of the human upper respiratory tract
(i.e., nose, nasal cavity, paranasal sinuses, pharynx, larynx), as well as
neighboring structures (from Yokes etal. (19931).
This document is a draft for review purposes only and does not constitute Agency policy,
1-197' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Literature Search and Screening Strategy
The primary databases used for the literature searches were PubMed, Web of Science, and
Toxline, with the last update of the search completed in September 2016 (see Appendix A.4.7, A. 5.9
and A.5.6), and a systematic evidence map updating the literature through 2021 (see Appendix F).
The occurrences of upper respiratory tract (URT) cancers in humans have been described and
grouped according to the International Classification of Disease (ICD) coding rubrics. This review
focused on the specific cancer diagnoses available in the epidemiological literature. The specific
cancers of the URT that are most commonly reported are sinonasal cancers (cancers of the nose and
nasal sinuses), cancers of the pharynx (comprising the nasopharynx, oropharynx and
hypopharynx), and laryngeal cancer. Rarely, cancers of the buccal cavity as a whole are reported,
but as this grouping includes lip, tongue, salivary glands, gums, and the floor of the mouth, which
combine cancers of potentially different etiology and cell origin, the collection of cancers of the
buccal cavity are not reviewed here. Only primary epidemiological studies of specific cancer
endpoints with identified or inferred formaldehyde exposure were included. Additional studies
were identified from review articles and government documents.
Evidence from animal experiments included precancerous lesions (i.e., dysplasia) and
neoplasms (tumors) of the respiratory tract. Animal studies investigating formaldehyde-induced
respiratory carcinogenesis were carried out primarily in rats and to a lesser extent in mice,
hamsters, and nonhuman primates. The most consistent evidence of formaldehyde-induced
respiratory cancers in animals is restricted to the nasal cavity and consists primarily of squamous
cell carcinomas (SCCs). Other neoplasms that have been observed include carcinomas other than
SCCs, sarcomas, papillomas, and adenomas (Kamata etal.. 1997: Monticello etal.. 1996: Morgan et
al.. 1986b: Sellakumar etal.. 1985: Kerns etal.. 19831.
The bibliographic databases, search terms, and specific strategies used to search them are
provided in Appendix A.4.7, A.5.5, and A.5.9 for the cancer outcomes and relevant mechanistic
endpoints. The specific PECO criteria for the human and animal health effects studies are provided
in Appendix A.5.9. Literature flow diagrams summarize the results of the sorting process using
these criteria and indicate the number of studies that were selected for consideration in the
assessment through 2016 (see Appendix F for the identification of newer studies through 2021).
Upper Respiratory Tract Cancers in Human Studies
Each specific type of upper respiratory tract (URT) cancer (nasopharyngeal cancer,
sinonasal cancer, cancers of the oropharynx and hypopharynx, and laryngeal cancer) is reviewed
and evaluated independently in the sections below. For each type of URT cancer, the evidence is
organized by considerations that inform the strength of evidence (e.g., consistency, exposure-
response) and evaluation of the potential for bias and insensitivity in individual studies to affect the
estimates of relative risk. Evidence tables for each type of URT cancer (Tables 1-32 through 1-35)
This document is a draft for review purposes only and does not constitute Agency policy.
1-198 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
are included and are organized first by the study evaluation conclusions (i.e., high, medium, low)
and then by publication year.
Methodological issues and approaches for evaluation
The epidemiology studies generally examined occupational exposure to formaldehyde
either in specific work settings (e.g., cohort studies) or in case-control studies. The considerations
with respect to design, exposure assessment, outcome assessment, potential bias and confounding,
and analysis differ for these different types of studies, and are discussed in more detail in
Appendix A.5.9. Developing an outcome-specific study evaluation for each cancer outcome
encompasses two concepts: minimization or control of bias (internal validity) and sensitivity (the
ability of the study to detect a true effect). Because a single epidemiology study may report on
several different cancer endpoints, the confidence classifications are for the specific cancer results
and are not judgments on the study as a whole except when a study has only a single cancer
endpoint. The distinction here is important in that a study of adequate quality overall may still
report an effect estimate judged to be of low confidence due to the rarity of the cancer outcome, the
rarity of the exposure, or noncritical biases, which are expected to yield effect estimates that
underestimate any true effect.
The diagnosis of cancers in epidemiological studies has historically been ascertained from
death certificates according to the version of the International Classification of Diseases (ICD) in
effect at the time of study subjects'deaths [i.e., ICD-8 and ICD-9: (WHO. 1987a. b)]. The most
specific classification of diagnoses commonly reported across the epidemiological literature was
based on the first three digits of the ICD code without further differentiation. For some cancers, the
reliance of cohort studies on death certificates to detect cancers with relatively high survival may
have underestimated the actual incidence of those cancers, especially when the follow-up time may
have been insufficient to capture all cancers that may have been related to exposure. The potential
for bias may depend on the specific survival rates for each cancer. Five-year survival rates vary
among the selected cancers, from 59.6% for nasopharyngeal cancer (NPC) to less than 50% for
oropharyngeal/hypopharyngeal cancer.
The overwhelming majority of information bias in epidemiological studies of formaldehyde
stems from the use of occupational records to gauge exposures with some degree of exposure
misclassification or exposure measurement error considered to be commonplace. A primary
consideration in the evaluation of these studies is the ability of the exposure assessment to reliably
distinguish among levels of exposure within the study population or between the study population
and the referent population. A large variety of occupations were included within the studies; some
represented work settings with a high likelihood of exposure to high levels of formaldehyde, and
some represented work settings with variable exposures and in which the proportion of people
exposed was quite small. In the latter case, the potential effect of formaldehyde would be "diluted"
within the larger study population, limiting the sensitivity of the study. The exposure-assessment
methods of the identified studies were classified into four groups (A through D), reflecting greater
This document is a draft for review purposes only and does not constitute Agency policy.
1-199 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
or lesser degree of reliability and sensitivity of the measures (see Appendix A.5.9). Outcome-
specific associations based on Group A exposures were considered to be without appreciable
information bias due to exposure measurement error while those based on Groups B-D were
considered to be more biased towards the null.
Additional exposure measurement error may arise in circumstances when the period of
exposure assessment is not well aligned with the period when formaldehyde exposure could induce
carcinogenesis that develops to a detectable stage (incident cancer) or could result in death from a
specific cancer. The cohort studies were evaluated to ensure that they analyzed the analytic impact
of different lengths of "latency periods" (i.e., excluded from the analyses the formaldehyde exposure
most proximal to each individual's cancer incidence or cancer mortality). Analyses that did not
evaluate latency were considered to be more biased towards the null because irrelevant exposure
periods were included fCoggon etal.. 20141.
Studies with small case counts may have little statistical power to detect divergences from
the null but are not necessarily expected to be biased and no study was excluded solely on the basis
of case counts as this methodology would have excluded any study that saw no effect of exposure.
Therefore, cohort studies with extensive follow-up that reported outcome-specific results on a
number of different cancers, including very rare cancers, were evaluated even when few or even no
cases were observed, if information on the expected number of cases in the study population was
provided so that confidence intervals could be presented to show the statistical uncertainty in the
associated effect estimated. For example, Coggon et al. (2014) followed the mortality of 14,008
workers and yet expected only 1.7 deaths from nasopharyngeal cancer in the exposed workers and
observed just one resulting in an unstable estimated RR = 0.38 (95% CI 0.02,1.90). Meyers et al.
(2013) followed the mortality of 11,043 workers and expected only 1.33 deaths from
nasopharyngeal cancer and did not observe any deaths, resulting in an SMR = 0 (95% CI 0, 2.77).
In addition to potential bias, study sensitivity was specifically evaluated; study results with
low sensitivity could result in effect estimates that underestimated a "true" association if it existed.
For example, an outcome-specific effect estimate based on fewer than five observed cases of a
particular cancer would be classified as low based on a lack of sensitivity—even if there were no
appreciable biases. Another example would be a study that might have relied on exposure-
assessment methodologies that were unbiased, but nonspecific in nature so as to yield effect
estimates that were likely biased towards the null and thus underestimates of any true effect.
Finally, cohort studies should have a sufficiently long follow-up period for any exposure-related
cancer cases to develop and be detected and, ideally, allow for analyses of potential cancer latency.
Outcome-specific effect estimates from cohort studies with short follow-up could be considered
uninformative depending on the size of the study population and the baseline frequency of the
cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
1-200 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Nasopharyngeal cancer
Epidemiological evidence
The most specific classification of nasopharyngeal cancer diagnosis that is commonly
reported on death certificates across the epidemiological literature has been based on the first
three digits of the Seventh (i.e., nasopharyngeal cancer ICD-7: 146), Eighth, or Ninth Revision of the
ICD code (i.e., nasopharyngeal cancer ICD-8/9: 147) although some studies did report the
histological type of cancer (i.e., squamous cell carcinoma and nonkeratinizing or undifferentiated
cancer), the histological type is infrequently reported on death certificates.
Evidence describing the association between formaldehyde exposure and the risk of
developing or dying from nasopharyngeal cancer is available from 20 epidemiological studies—
12 case-control studies (Li etal.. 2006: Yang etal.. 2005: Yu etal.. 2004: Hildesheim etal.. 2001:
Armstrong etal.. 2000: Vaughan et al.. 2000: West etal.. 1993: Vaughan. 1989: Roush etal.. 1987b:
Vaughan et al.. 1986a. b; Olsen etal.. 1984) and eight cohort studies (Coggon et al.. 2014: Beane
Freeman etal.. 2013: Meyers etal.. 2013: Siew et al.. 2012: Hauptmann etal.. 2009: Dell andTeta.
1995: Hansen and Olsen. 1995: West etal.. 1993: Malker etal.. 1990: Vaughan. 1989: Roush etal..
1987a: Vaughan et al.. 1986a. b; Olsen etal.. 19841. These are the only primary studies that provide
evidence of the effect of formaldehyde exposure on the risk of dying from nasopharyngeal cancer.
The outcome-specific evaluations of confidence in the precise effect estimate of an association from
each study are provided in Appendix A.5.9. Note that the confidence judgments are for the
confidence in the precise effect estimate of an association from each study—and not a confidence
judgment in the overall study. The distinction here is important in that a study of adequate quality
overall may still report an effect estimate judged to be of low confidence due to the rarity of the
cancer outcome, the rarity of the exposure, or noncritical biases that are expected to yield effect
estimates that underestimate any true effect The results from Li et al. (2006) were classified as not
informative due to the rarity of exposure in both the case and control groups; for details see
Appendix A.5.9. The reported result from a case-control study by Armstrong et al. (2000) was
classified as not informative due, primarily, to the rarity of relevant exposure data as only 8/564
subjects (1.4%) had more than 10 years of potential exposure beyond a 10-year latency period, and
thus the study lacked sensitivity to detect any true effect (see Appendix A.5.9). The results from
Dell et al. (1995) were classified as not informative due to the rarity of exposure in the cohort with
111 men exposed to formaldehyde out of 5932 (1.9%) and there were no observed cases of
nasopharyngeal cancer; for details see Appendix A.5.9. Details of the reported results of high,
medium, and low confidence are provided in the evidence table for nasopharyngeal cancer (see
Table 1-32) following the causal evaluation.
Consistency of the observed association
Seventeen informative studies reported risks of nasopharyngeal cancer among subjects
with formaldehyde exposure based on occupational or residential history. These studies examined
This document is a draft for review purposes only and does not constitute Agency policy.
1-201 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 different populations, in different geographical locations, under different exposure settings and
2 employing different study designs. Importantly, for nasopharyngeal cancer, these studies were
3 conducted in low background risk populations (e.g., Europe and the United States) and high
4 background risk populations (e.g., China and Taiwan). Table 1-31 provides the incidence rates of
5 nasopharyngeal cancer per year by country/region based on the IARC publication Cancer Incidence
6 in Five Continents (Curado etal.. 2007) for each of the 17 studies.
Table 1-31. Age-standardized (world) incidence rates of nasopharyngeal
cancer per 100,000 per year
Study
Country
Region
Incidence rate/year
(per 100,000)
Siew et al. (2012)
Finland
0.3
Coggon et al. (2014)
England and Wales
South and Western
0.4
Hansen and Olsen
(1995)
Denmark
0.4
Malker et al. (1990)
Sweden
0.4
Olsen et al. (1984)
Denmark
0.4
Vaughan et al. (2000)
United States
CT, Detroit, IA, Seattle, UT
0.4-0.7
Meyers et al. (2013)
United States
Georgia and Pennsylvania
0.5-0.6
Beane Freeman et al.
(2013)
United States
National Cancer Registries
0.6
Hauptmann et al.
(2009)
United States
National Cancer Registries
0.6
Vaughan (1989)
United States
Washington
0.6
Roush et al. (1987a)
United States
Connecticut
0.6
Vaughan et al. (1986a)
United States
Washington
0.6
Vaughan et al. (1986b)
United States
Washington
0.6
Yang et al. (2005)
Taiwan3
3.5-8.3
Hildesheim et al.
(2001)
Taiwan3
3.5-8.3
West etal. (1993)
Philippines
5.8
Yu etal. (2004)
China
Hong Kong
17.8
aTaiwan is not included in the IARC publication of cancer incidence rate so data were obtained from Chen et al.
(2002).
This document is a draft for review purposes only and does not constitute Agency policy.
1-202 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Also important for nasopharyngeal cancer is the consideration of histological subtype,
which may be of a keratinizing or nonkeratinizing cell type as the proportion of each cell type varies
in low and high-risk populations. The study results presented in Table 1-32 (by confidence level
and publication date) detail all of the reported associations. Results are plotted in Figure 1-20;
results are grouped by population background risk and arrayed from lowest to highest by the
percentage of cases in each study's results, which were considered likely to be squamous cell
carcinomas.
Fourteen out of 17 studies reported increased risks of nasopharyngeal cancer with at least
one metric of formaldehyde exposure—often with both clear statistical significance and
exposure-response relationships. These included the results of large cohort study of 25,619 U.S.
workers fBeane Freeman etal.. 20131 classified with high confidence, and all four sets of results
classified with medium confidence (see Table 1-32). Nine studies in eight independent populations
reported relative effect estimates greater than three-fold. Yang et al. (2005) reported an OR of 4.29
(95% CI 2.45, 7.51) among cases with the highest cumulative formaldehyde exposure; Yu etal.
(2004) reported a mortality odds ratio (MOR) of 3.75 (95% CI 1.12,12.54) for restaurant workers
in Hong Kong; West et al. (1993.) reported an OR = 4.0 (95% CI 1.3,12.3) among Philippine cases
with greater than 25 years of time since first exposure (TSFE); Roush et al. (1987a) reported an
OR = 4.0 (95% CI 1.3,12.0) among Connecticut cases aged 68+ years with the highest duration of
exposure and 20+ years TSFE; Beane Freeman et al. (2013.) reported an RR = 11.54 (95% CI 1.38,
96.81) for workers with the highest average intensity of exposure; Malker et al. (1990) reported a
standardized incidence ratio (SIR) of 3.9 (95% CI 1.24, 9.40); Vaughan etal. (1986b) reported an
OR = 6.7 (95% CI 1.2, 38.9) for cases living and working in a mobile home; Vaughan (1989)
reported an OR = 31.8 (no CI provided) for the highest duration of working as a carpenter; and
Vaughan et al. (2000). after excluding undifferentiated and nonkeratinizing histological types,
reported an OR = 13.3 (95% CI 2.5, 70) for cases with the highest likelihood of formaldehyde
exposure.
Results showing increased risks were consistently reported in populations from high-risk
areas with endemic Epstein-Barr infection such as Hong Kong (Yu etal.. 2004). Taiwan (Yang etal..
2005: Hildesheim et al.. 2001). the Philippines (West etal.. 1993) as well as in populations from
low/medium-risk areas such as the United States fBeane Freeman et al.. 2013: Vaughan etal.. 2000:
Vaughan. 1989: Roush etal.. 1987a: Vaughan etal.. 1986a. b). Results showing increased risks were
also consistently reported across study populations with different proportions of squamous cell
carcinomas (SCC) (i.e., Hildesheim et al. (2001) and Yang et al. (2005) reported only 9% of their
cases were keratinizing SCC), more heterogeneous mixes of keratinizing and nonkeratinizing
carcinomas [i.e., Malker et al. (1990). (48% keratinizing SCC); Vaughan et al. (2000). (60%);
Vaughan et al. (1986a. b), (78%)], and study populations restricted to only squamous cell
carcinomas fVaughan etal.. 2000: Vaughan. 19891 (100% keratinizing SCC)].
This document is a draft for review purposes only and does not constitute Agency policy.
1-203 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Toxicological Review of Formaldehyde—Inhalation
Of these 17 studies, all but three reported increased risks of nasopharyngeal cancer that
appeared to be associated with exposure to formaldehyde; the three exceptions were the results
from the large occupational cohort studies by Siew et al. (2012), Coggon et al. (2014), and Meyers et
al. (2013.)—all of which were classified with low confidence. One additional study fAndielkovich et
al.. 19951 reported zero cases of NPC among 3,929 U.S. workers exposed to formaldehyde over
83,064 person-years but reported no data on the number of expected cases and thus was not
included here.17 An additional study (Edling etal.. 1987b) reported one case of NPC among 521
Swedish workers exposed to formaldehyde over 7,011 person-years but reported no data on the
number of expected cases and was not included here.18 One possible explanation for the
inconsistency is the rarity of NPC in the populations studied by Siew and by Coggon. Table 1-32
shows that the Finnish population studies by Siew et al. (2012) had a background incidence rate of
0.3 cases per year for each 100,000 people—the lowest of all the available populations reviewed
here. The English and Welch population studied by Coggon et al. (2014) had the second lowest
incidence rate at 0.4 cases per year for each 100,000 people.19 The very low national incidence
rates of NPC can make studies of these populations lack the statistical sensitivity to detect any true
association—even when the number of people being followed appears to be large.
It is important to understand that the statistical power of these cohort studies depends
directly on the number of observed and expected cases. While there are exact methods to compute
the variance of the standardized mortality ratio, the general formula illustrates the dependence on
the case counts. The variance of the standardized mortality ratio is generally a function of the
inverse of the observed and expected case count, specifically, var(SMR) = [# observed cases/(# of
expected cases)2]. Smaller case counts produce larger statistical variances and wider confidence
intervals. Because the SMR is a measure of relative effect bounded between zero and infinity, it
may be more straightforward to consider the width of confidence intervals on the scale of the
natural logarithm, which bounds the estimates symmetrically between negative infinity and
positive infinity. Coggon et al. (2014) expected only 1.7 deaths from nasopharyngeal cancer in the
exposed workers and observed just one resulting in an unstable estimated RR = 0.38 (95% CI 0.02,
1.90); on the natural log scale the ln(RR) = -0.97 (95% CI -3.91, to 0.64). Meyers et al. (2013)
expected only 1.33 deaths and did not observe any deaths, resulting in an SMR = 0 (95% CI 0, 2.77);
on the natural log scale, the ln(RR) = negative infinity (95% CI negative infinity to +1.99). These
effect estimates result in wide confidence intervals. For comparison, the other large cohort study
fBeane Freeman etal.. 20131 expected 4.89 deaths and observed nine deaths from NPC, resulting in
17For Andjelkovich et al. (19951. assuming a rate of NPC for U.S. workers of 0.6 per 100,000 person-years
fCurado etal.. 20071. the expected number of cases would have been 0.33 and the ~SMR = 0 (95% CI 0, 5.99],
18For Edling et al. (1987b], assuming a rate of NPC for Swedish workers of 0.4 per 100,000 person-years
fCurado etal.. 2007], the expected number of cases would have been 0.028 and the ~SMR = 35.71 (95% CI
1.79,176.1].
19For comparison, the background incidence rate in the United States is 0.6 cases per year for each 100,000
people and ranges from 3.5 to 17.8 cases per year for each 100,000 people in the Philippines, Taiwan, and
Hong Kong (see Table 1-31],
This document is a draft for review purposes only and does not constitute Agency policy.
1-204 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
a SMR = 1.84 (95% CI 0.84, 3.49); on the natural log scale, the ln(RR) = negative infinity (95% CI
-0.17,1.25). The NPC results from the Coggon et al. (2014), Meyers et al. (2013.) and Siew et al.
(2012) studies were all considered to lack sensitivity to detect any true effect, which contributed to
their classifications of low confidence.
In summary, the majority of studies from different populations, in different locations,
exposure settings, and using different study designs reported increased risks of nasopharyngeal
cancer associated with formaldehyde exposure. There are reasonable alternative explanations for
the three studies that did not observe an increased risk.
Strength of the observed association
While reported relative effect estimates were consistently elevated above the null value of
one across 14 of the 17 studies, the magnitude of the relative risk estimates varied with the quality
of the exposure assessment Studies with higher quality exposure data that were capable of
stratifying subjects by exposure level, exposure probability, and timing of exposure (including
lagged exposures) generally reported higher relative effect estimates. Nine studies reported
greater than three-fold increased risks of nasopharyngeal cancer that appeared to be associated
with exposure to formaldehyde (Beane Freeman et al.. 2013: Yang etal.. 2005: Yu etal.. 2004:
Vaughan et al.. 2000: West etal.. 1993: Malker etal.. 1990: Vaughan. 1989: Roush etal.. 1987a:
Vaughan et al.. 1986bl. Three studies reported greater than 10-fold increased risks of
nasopharyngeal cancer in the highest exposure categories. These increased risks appeared to be
associated with duration of exposure to formaldehyde after accounting for a latency period (Beane
Freeman etal.. 2013: Vaughan etal.. 2000: Vaughan. 1989). Results from the studies with higher
quality exposure data were judged with greater confidence.
Temporal relationship of the observed association
Two related aspects of time are encompassed in the consideration of temporality. One
aspect is the necessity for the exposure to precede the onset of the disease. In each of the studies,
the formaldehyde exposures among the study participants started before their diagnoses of NPC,
and in the studies that ascertained individual-level exposures, the estimation of formaldehyde
exposures was based on job titles and done in a blinded fashion with respect to outcome status.
The second aspect involves the time course of formaldehyde exposures in relation to the
incidence of NPC and death from NPC. From the epidemiological literature, it is known that there
can be an induction/latency period for some environmental agents and that the induction period
may exceed 10 years. Three studies provided analyses of this temporal relationship showing some
evidence of the effect of time since first exposure on the risk of dying from nasopharyngeal cancer
(Hildesheim etal.. 2001: West etal.. 1993: Roush et al.. 1987b): however, none of them did so by
histological subtype. Hildesheim et al. (2001) reported conflicting evidence of lower risks among
all NPC cases for first exposure to formaldehyde more than 20 years earlier, but higher risks with
greater time since first exposure (TSFE) when analyses were limited to only those who were
This document is a draft for review purposes only and does not constitute Agency policy.
1-205 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
positive for Epstein-Barr virus. Roush et al. (1987b) reported somewhat greater risks among those
first exposed more than 20 years and a stronger such pattern among those considered to be highly
exposed more than 20 years prior to dying of nasopharyngeal cancer. Even higher risks were found
among those with high early exposures and who were 68 years or older at death (OR = 4.0; 95% CI
1.3,12.0), which may imply that TSFE much greater than 20 years carries greater risk. The results
from West et al. (1993) support this assertion; in multivariate analyses, they reported a low odds
ratio for TSFE less than 25 years but higher risks for greater than 25 years (OR = 4.0; 95% CI 1.3,
12.3). In separate analyses controlling only for TSFE to formaldehyde, dust, and exhaust fumes,
West et al. (1993.) reported even higher risk among those first exposed to formaldehyde more than
35 years earlier (OR = 5.6; 95% CI 0.58, 52.9).
The histological subtype and background rate of nasopharyngeal cancer is important in
considering latency as the population studied by Hildesheim et al. (2001) resided in Taiwan (a high
background risk population), and cases were more than 90% nonkeratinizing. In contrast, the
population Roush et al. (1987b) studied was from Connecticut (a low background risk population),
which may have only ~28% nonkeratinizing cases, if consistent with a U.S. study of nasopharyngeal
cancer that included cases from Connecticut fVaughan. 19891. West et al. (1993) studied subjects
from the Philippines where the background rate is intermediate to the high rates of some East
Asians and the low rates in populations of European descent fHildesheim etal.. 19931.
The association between exposure to formaldehyde and risk of nasopharyngeal cancer may
be weaker for nonkeratinizing cases (Vaughan et al.. 2000). This may explain the apparent lack of a
clear latency effect in the Hildesheim et al. (2001) study, which has more than 90% of cases
diagnosed with nonkeratinizing cases. The remaining limited evidence on the time course of death
following initial formaldehyde exposure is consistent with expectation of a lengthy latency period
for cancer development and subsequent deaths.
Exposure-response relationship
In their large population-based case-control study including 196 cases of nasopharyngeal
cancer, Vaughan etal. (2000) clearly demonstrated two important points: (1) that there was an
exposure-response relationship between increased formaldehyde exposure and increased risk of
nasopharyngeal cancer, and (2) that the exposure-response differed by nasopharyngeal cancer
subtype in the U.S. population. Vaughan et al. (2000) reported statistically significant trends for
differentiated squamous cell carcinomas (p = 0.033) and for cases of epithelial carcinoma without
specification of histological type (p = 0.036). However, there was no trend with duration of
exposure to formaldehyde among cases with undifferentiated/nonkeratinizing histology (p = 0.82).
Grouping of all histological subtypes appeared to mask the underlying relationship seen in
squamous cell carcinoma in this study. Excluding nasopharyngeal cancer cases with
undifferentiated or nonkeratinizing histology, Vaughan et al. (2000) reported a clear
exposure-response with increased probability of exposure to formaldehyde with the highest risks
seen in subjects with the highest probability of occupational exposure to formaldehyde (OR = 13.3;
This document is a draft for review purposes only and does not constitute Agency policy.
1-206 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
95% CI 2.5, 70; p = 0.0007). Among those subjects considered to be "definitely exposed," there
were increasing risks of nasopharyngeal cancer with increasing duration of formaldehyde exposure
(p < 0.001) and with increased cumulative formaldehyde exposure (p < 0.001).
Further evidence of exposure-response relationships was reported by Beane Freeman et al.
(2013.) for peak formaldehyde exposures (p = 0.005), and, to a lesser degree, for cumulative
exposures (p = 0.06) and with average intensity of formaldehyde exposure (p = 0.09 ) 20. Other
supporting evidence of an exposure-response relationship between increased exposure to
formaldehyde and increased risk of NPC come from three reports on the same study population in
Washington state fVaughan. 1989: Vaughan et al.. 1986a. b). These studies reported higher risks
with increasing occupational exposures but did not report tests of trend fVaughan etal.. 1986al: for
example, with a 15-year lag, compared to the lowest exposure score, those in the second level had
an OR = 1.7 (95% CI 0.5, 5.7), while those in the third level had an OR = 2.1 (95% CI 0.4,10.0).
These researchers also reported increased risks with length of residence in mobile homes with the
risk peaking among those with more than 10 years of occupancy (OR = 5.5; 95% CI 1.6,19.4)
fVaughan et al.. 1986b). The majority (84%) of mobile homes in the United States at this time were
reported to have mean formaldehyde exposures in excess of 100 ppb, with 22% having mean
exposures in excess of 500 ppb fBrevsse. 19841 as cited in WHO (1989). A qualitative exposure-
response relationship was shown for overall mobile home exposures with the risk of
nasopharyngeal cancer for working in a mobile home but not living in a mobile home (OR = 1.7;
95% CI 0.5, 5.7) being exceeded by the risk of living in a mobile home (OR = 2.8; 95% CI 1.0, 7.9).
However, the greatest risk was reported for living and working in a mobile home (OR = 6.7; 95% CI
1.2, 38.9). Vaughan (1989) also reported increasing risk with duration of employment as a
carpenter after lagging exposures by 15 years to account for cancer latency (x2 trend = 8.65;
p = 0.01 with 2 df)—especially as a carpenter in the construction industry (x2 trend = 14.86;
p = 0.0006 with 2 df). Carpentry is considered to be a formaldehyde-related job since many
products used in construction and building trades involve exposure to formaldehyde (Hildesheim
etal.. 2001: Vaughan etal.. 1986a). Carpentry also involves coexposure to wood dust, which is
likely to be a potential confounder for NPC, as it a potent risk factor. The potential for confounding
by wood dust is evaluated in the following section. Other evidence generally consistent with an
exposure-response relationship was reported by Yu et al. (2004), Hildesheim et al. (2001), and
West et al. (1993.). Yu et al. (2004) reported mortality ORs (MORs) for three levels of increasing
cumulative exposure based on years of union membership compared to none and report MORs of
2.5, 3.41, and 3.75 (95% CI 1.12,12.54). Hildesheim etal. (2001) reported an OR = 1.3 for less than
25 years of cumulative exposure and OR = 1.5 for more than 25 years of cumulative exposure
20Mohner et al. (2019) argued that there might have been a diagnostic bias in coding the specific and non-
specific pharyngeal cancer in the NCI cohort study which could have affected the pharyngeal cancer SMRs;
however, potential administrative miscoding of cancer mortality on death certificates would be independent
of the quantitative estimates of workers' exposures, and any misclassification of diagnostic codes would not
be expected to yield evidence of exposure-response relationships.
This document is a draft for review purposes only and does not constitute Agency policy.
1-207 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
(95% CI 0.88, 2.7, p-trend = 0.10); Westetal. (19931 reported that daily use of antimosquito coils
[which have been shown in experiments to emit formaldehyde concentrations of between 0.87 and
25 ng/m3; see fLiu etal.. 20031] had an OR = 5.9 (95% CI 1.7, 20.1), while less than daily use had an
OR = 1.4.
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias may alter epidemiological findings when participation or follow-up rates are
related to the probability of exposure or the outcome. However, this is an unlikely bias in the
epidemiological studies of nasopharyngeal cancer, as the case-control studies evaluated exposure
status without regard to outcome status and had participation levels of 85-100%. Each of the
cohort studies included at least 72% of eligible participants and lost relatively few participants over
the course of mortality follow-up.
The issue of potential selection bias was relevant to the results from two study populations
—all classified with low confidence (Yang etal.. 20051 and the three Vaughan papers (1989: 1986a.
b). Both Yang et al. (20051 and Vaughan (19891 with Vaughan et al. (1986a. b) used more than 40%
of case interviews completed by next of kin due to cancer mortality among cases and no proxy
respondent was included for the controls. When next-of-kin is used to provide proxy information
on cases, measurement error is likely to be present to some degree. If the quality of those data
differs between cases and controls, this can result in selection bias if any differences are related to
exposure. Hence, EPA considers that there is some risk of selection bias in the results of these
studies (e.g., (Yang etal.. 2005: Vaughan. 1989: Vaughan et al.. 1986a. b).
Information bias may distort findings when subjects' true personal exposures are
inaccurately assigned. Differential misclassification, in which exposure status influences disease
classification (or disease status influences exposure classification), can lead to bias toward or away
from the null (i.e., spurious or "false positive" associations). This scenario is considered unlikely
among these studies of nasopharyngeal cancer mortality because the likelihood of differential
misclassification based on these study designs is low. The assignment of exposure status or
calculation of exposure measures in the case-control and cohort studies was done independently of
knowledge of the cause of death. Therefore, an exposure-related bias in subjects' recall or
reconstruction of their occupational histories seems unlikely.
Another aspect of information bias stems from random measurement error or
nondifferential misclassification. This type of error typically will bias the risk estimate toward the
null, thereby obscuring real effects by underestimating their magnitude. Given the difficulty in
accurately estimating personal exposure over time or in the use of proxies to represent exposure to
formaldehyde, the likelihood of random measurement error is almost certain in many studies. The
implication of such information bias is that the consistently reported increases in risks of
formaldehyde-related mortality may be underestimates and the true risk could be larger than was
demonstrated in these epidemiological studies.
This document is a draft for review purposes only and does not constitute Agency policy.
1-208 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
A third possible scenario for information bias could arise from systematic measurement
error that is nondifferential with respect to disease. Such a scenario would be unusual in a study
with exposure assessment based in industrial settings with extensive industrial hygiene data used
to determine levels of exposure fe.g.. Beane Freeman etal.. 20131. However, a claim was made by
Marsh et al. (2007b; 20021 that the exposure assessment used for the NCI formaldehyde cohort
reported on by Beane Freeman et al. (2013) was 10-fold higher than those estimated by Marsh et al.
(2007b: 20021. If this were true, then the same amount of observed risk in Beane Freeman et al.
(20131 would be apportioned to one-tenth the same exposure, which would yield an exposure-
response 10-fold greater in magnitude. The claim by Marsh et al. (2007b; 20021 suggests a
one-sided uncertainty in the exposure-response reported by Beane Freeman et al. (2013.), which
may be 10 times more potent than reported.
Confounding is a potential bias that could arise if another cause of nasopharyngeal cancer
was also associated with formaldehyde exposure. There does not appear to be any evidence of a
common confounder that would provide an alternative explanation for the consistently observed
association of formaldehyde exposure with increased risk of nasopharyngeal cancer seen across
these studies. Chemicals and other coexposures that have not been independently associated with
nasopharyngeal cancer are not expected to confound results. Other known risk factors for
nasopharyngeal cancer include childhood consumption of Chinese salted fish fYu etal.. 19861. wood
dust (Hildesheim etal.. 2001). smoking, and alcohol consumption (Vaughan. 1996). While these
other exposures may be independent risk factors for nasopharyngeal cancer, consumption of
Chinese salted fish (or other dietary exposures to nitrosamines) and alcohol are unlikely to be
generally related to formaldehyde exposures, and therefore, these other exposures are not
expected to be consistent confounders across all of the studies. Additionally, Epstein-Barr virus is
thought to be a cause of nasopharyngeal cancer due to its ubiquitous presence in nasopharyngeal
cancer cases, but Hildesheim (2001) described Epstein-Barr virus as an effect modifier of the
association between formaldehyde and nasopharyngeal cancer, and not as a confounder.
Wood dust may be an independent risk factor for nasopharyngeal cancer, but three studies
specifically controlled for the potential confounding of the effects of wood dust on the risk of
nasopharyngeal cancer and did not find wood dust to be a confounder (Hildesheim et al.. 2001:
Vaughan et al.. 2000: West etal.. 19931. Similarly, smoking was specifically controlled for in a
number of studies fVaughan et al.. 2000: West etal.. 1993: Vaughan. 1989: Vaughan etal.. 1986a. b)
and was not likely to have been a major confounder of the formaldehyde-associated results. Marsh
et al. (2005) re-evaluated the association between formaldehyde exposure and NPC in the NCI
cohort and reported that the majority of the cases of NPC arose in one of the 10 plants included in
the cohort and that this findings suggested that there might be something specific to the experience
in Plant 1 (in Wallingford, CT) that may have given rise to the excess of NPC cases there - perhaps a
confounder. Marsh et al. (2007b) suggests that silversmithing may be a cause of NPC in Plant 1 and
that the reported association between formaldehyde and NPC may be due to confounding; however,
This document is a draft for review purposes only and does not constitute Agency policy.
1-209 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Beane Freeman et al. (20131 noted that the reported association for formaldehyde on the risks of
NPC did not decrease when analyses adjusted for silversmithing fsee Table 5 of Marsh etal..
2007b). The details of Table 1 in fMarsh and Youk. 20051 show the SMRs for NPC for each of the 10
plants. The two plants with the highest average intensity of formaldehyde exposure had the two
highest SMR estimates for NPC. It is plausible that the observation that the majority of the cases of
NPC in the NCI cohort come from Plant 1 reflects generally higher formaldehyde exposures and
a larger number of people at that plant than at other plants. This overall evidence does not indicate
confounding of the formaldehyde association with increased risk of NPC.
Consistency across multiple studies is demonstrated by a pattern of increased risk in
different populations, exposure scenarios, and time periods. Such consistency makes unmeasured
confounding an unlikely alternative explanation for the observed associations. This consistency
also reduces the likelihood of chance as an alternative explanation by increasing the statistical
strength of the findings through the accumulation of a larger body of similar evidence. The
observations of multiple instances of very strong associations, as well as exposure-response trends
with increased formaldehyde exposure using multiple metrics of exposure similarly, reduce the
likelihood that chance, confounding, or other biases can explain the observed association.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
nasopharyngeal cancer placed the greatest weight on five particular considerations: (1) the
consistency of the observed increases in risk across several studies—including results classified
with high, medium, and low confidence with higher risks among Asian populations that have higher
background rates of nasopharyngeal cancer and reasonable explanations for the lack of findings in
a few studies with very low background rates of nasopharyngeal cancer; (2) the strength of the
association with eight studies reporting at least a three-fold increase in risk; (3) the reported
exposure-response relationships showing that multiple measures of increased exposure to
formaldehyde were repeatedly associated with increased risk of dying from nasopharyngeal
cancer—especially among studies primarily focused on squamous cell carcinomas; (4) a
biologically coherent temporal relationship consistent with a pattern of exposure to formaldehyde
and subsequent death from nasopharyngeal cancer, allowing time for cancer induction, latency, and
mortality; and (5) reasonable confidence that alternative explanations are ruled out, including
chance, bias, and confounding within individual studies or across studies. Consistent observations
of genotoxicity in exfoliated buccal cells or nasal mucosal cells across several occupational studies
involving diverse exposure settings further supports the evidence in humans.
Conclusion
The available epidemiological studies provide robust evidence of an association consistent
with causation between formaldehyde exposure and increased risk of nasopharyngeal
cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
1-210 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
O
•*-»
m
E
+5
if)
LU
O
O
0
c
0
Q.
CL
2
3
a
£ §
0
jo .0
I %
LU LU
LL. IL
W CO
2
100-
5 i i
® .y jK-
Hildesheim
Yang
Yu
I 11 11 i
West Olsen Roush
10-
3-
2-
Beane Freeman
S £
I <3
£ ro
I
s
a
a
•8
g> a) cti ^ gi cn
Vaughan
2000
Vaughan 1986ab; 1989
LU
V
>
.2
0
cn
9% SCC
Unknown % SCC
Asian Populations
Unknown % SCC
European and U.S. Populations
JI 48% SCC -> 78% SCC || 100% SCC |
JL
"3" ^
CM CM CO ,—,
C C C — CO CO ip
— — — in cn
cni in; cm cm co co
¦—¦ 05 — to 1—1 1—¦
1^i CO lO 1 1
S S
cm osi
CO CM CO
t- IN (O
S ST O
CD CO CO
1/3 (/) 10 K UJ 0
I I I 1 I I I I I I I I I 1 I I I
SVA> .vSV A> aV vnVaVaVa\cVAV
VN-
A ^ A AVAVAVAV
y
J?
/
¦if
$
&
oS^
jf
/
A*
Figure 1-20. Epidemiological studies reporting nasopharyngeal cancer risk
estimates.
Results are grouped by population background risk and arrayed from lowest to highest by the percentage
of cases in each study's results that were considered likely to be squamous cell carcinomas (SCC). SMR:
standardized mortality ratio. PMR: proportionate mortality ratio. SPIR: Standardized Proportional
Incidence Ratio. RR: relative risk. OR: odds ratio. MOR: mortality odds ratio. TSFE: time since first
exposure. For each measure of association, the number of exposed cases is provided in brackets
This document is a draft for review purposes only and does not constitute Agency policy,
1-211 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
(e.g., [n = 74]). For studies reporting results on multiple metrics of exposure, only the highest category of
each exposure metric is presented in the figure.
Table 1-32. Epidemiological studies of formaldehyde exposure and risk of
nasopharyngeal cancers
Study
Reference: Beane Freeman et
al. (2013)
Population: 25,619 workers employed
at 10 formaldehyde-using or
formaldehyde-producing plants in the
United States followed from either
the plant start-up or first employment
through 2004. Deaths were identified
from the National Death Index with
remainder assumed to be living. 676
workers (3%) were lost to follow-up.
Vital status was 97.4% complete and
only 2.6% lost to follow-up.
Outcome definition: Death
certificates used to determine
underlying cause of death from
nasopharyngeal cancer (ICD-8: 147).
Histological typing not reported.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in
lowest exposed category. Lagged
exposures were evaluated to account
for cancer latency. Results were
presented for 15-year lag.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Blair et al. (1986)
Hauptmann et al. (2004)
Beane Freeman et al. (2009)
Confidence in effect estimates:3
Exposures
Exposure assessment: Individual-level
exposure estimates based on job titles,
tasks, visits to plants by study
industrial hygienists who took 2,000 air
samples from representative job, and
monitoring data from 1960 through
1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3). Median cumulative
exposure = 0.6 ppm-years (range
0-107.4).
Duration and timing: Exposure period
from <1946 to 1980. Median length of
follow-up: 42 years. Median length of
employment was 2.6 years (range
1 day-47.7 years). Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Peak exposure:
Level 1 (>0 to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Level 1 (>0 to <0.5 ppm)
Level 2 (0.5 to <1.0 ppm)
Level 3 (>1.0 ppm)
Cumulative exposure:
Level 1 (>0 to <1.5 ppm-yrs)
Level 2 (1.5 to <5.5 ppm-yrs)
Level 3 (>5.5 ppm-yrs)
Duration of exposure:
Level 1 (0 years)
Level 2 (>0 to <5 years)
Level 3 (5 to <15 years)
Level 4 (>15 years)
Results: effect estimate (95% CI)
[# of Cases]
[2]
[1]
[0]
[7]
[2]
[1]
[1]
[6]
Cumulative exposure
Unexposed RR = 1.87 (0.30-11.67) [2]
Level 1 RR = 1.00 (Ref. value) [4]
Level 2 RR = 0.86 (0.10-7.70) [1]
Level 3 RR = 2.94 (0.65-13.28) [3]
p-trend (exposed) = 0.06;
p-trend (all) = 0.07
Duration of exposure
Level 1 RR = 1.00 (Ref. value) [4]
Level 2 RR = 0.86 (0.10-7.70) [1]
Level 3 RR = 2.94 (0.65-13.28) [3]
Level 4 RR = 2.53 (0.4-15.0) [not
given]
p-trend (all) = 0.4
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders
and found not be confounders.
External comparisons:
SMRunexposed = 1.45 (0.17-5.25) [2]
SMRexposed = 1.84(0.84-3.49) [9]
Multiple exposure metrics including
peak, average, and cumulative
exposures were evaluated using
categorical and continuous data.
Internal comparisons:
Peak exposure
Unexposed RR = 4.39 (0.36-54.05)
Level 1 RR = 1.00 (Ref. value)
Level 2 RR = NA
Level 3 RR = 7.66 (0.94-62.34)
p-trend (exposed) = 0.005;
p-trend (all) = 0.10
Average intensity
Unexposed
RR = 6.79 (0.55-83.64)
Level 1 RR = 1.00 (Ref. value)
Level 2 RR = 2.44 (0.15-39.07)
Level 3 RR = 11.54 (1.38-96.81)
p-trend (exposed) = 0.09;
p-trend (all) = 0.16
This document is a draft for review purposes only and does not constitute Agency policy.
1-212 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
SB IB a Oth
Overall
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A
[As noted in Appendix A.5.9: There was
no information on smoking, however,
according to Blair et al. (1986),
"The lack of a consistent elevation for
tobacco-related causes of death,
however, suggests that the smoking
habits among this cohort did not differ
substantially from those of the general
population."
Beane Freeman et al. (2013)
report that among a sample of 379
cohort members, they "found no
differences in prevalence of smoking
by level of formaldehyde exposure."]
Reference: Hauptmann et al.
(2009)
Population: 6,808 embalmers and
funeral directors who died during
1960-1986. Identified from registries
of the National Funeral Directors'
Association, licensing boards and
state funeral directors' associations,
NY State Bureau of Funeral Directors,
and CA Funeral Directors and
Embalmers. Deaths were identified
from the National Death Index. Next
of kin interviews conducted for 96%
of cases and 94% of controls.
Outcome definition: Death
certificates used to determine UCOD
from nasopharyngeal cancer (ICD-8:
147).
Design: Nested case-control study
within a prospective cohort mortality
study using two internal comparison
groups; the first composed of those
who had never embalmed (1 case and
55 controls) and the second
composed of those who had fewer
than 500 embalmings (five cases and
83 controls).
Analysis: ORs calculated using
unconditional logistic regression
adjusted for date of birth, age at
death, sex, data source, and smoking.
Lagged exposures were evaluated to
account for cancer latency. These
Exposure assessment: Occupational
history obtained by interviews with
next of kin and coworkers using
detailed questionnaires. Exposure was
assessed by linking questionnaire
responses to an exposure assessment
experiment providing measured
exposure data. Exposure levels (peak,
intensity, and cumulative) were
assigned to each individual using a
predictive model based on the
exposure data. The model explained
74% of the observed variability in
exposure measurements.
Multiple exposure metrics including
duration (mean = 31.3 yrs in cases), #
of embalming, peak, average, and
cumulative exposures were evaluated
using categorical and continuous data.
Duration and timing: Exposure period
from <1932 through 1986. Duration of
exposure was evaluated. Duration is
also a surrogate for time because first
exposure since dates of death was
closely related to cessation of
workplace exposures.
Variation in exposure:
Level 1 Never embalmed
Level 2 Ever embalmed
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
Internal comparisons:
Never embalming: OR =
Ever embalming: OR =
1.00 (Ref. value)
0.1 (0.01-1.2)
[2]
[2]
This document is a draft for review purposes only and does not constitute Agency policy.
1-213 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
results are shown in table 3 of
Hauptmann et al. (2009).
Results from the second internal
comparison group with <500
embalmings were selected to increase
statistical stability. These results are
shown in table 4 of Hauptmann et
al. (2009)
Related studies:
Haves et al. (1990)
Walrath and Fraumeni (1983)
Walrath and Fraumeni (1984)
Note: The original cohorts from these
three original studies were combined
in Hauptmann et al. (2009) and
follow-up was extended so the case-
series overlap and are not
independent. However, the three
original cohorts used external
reference groups for comparison
while Hauptmann et al. (2009)
selected internal controls, which were
independent of the reference groups
used in the original studies.
Confidence in effect estimates:3
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Chemical coexposures are not known
risk factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
SB IB Cf Oth
Overall
Confidence
Medium
Medium ¦i, (low sensitivity)
IB: Exposure Group A
Oth: Low power due to rarity of cases.
Reference: Hildesheim et al.
(2001)
Population: Male and female
Taiwanese aged <75 years newly
diagnosed with nasopharyngeal
cancer identified between July 1991
and January 1995 from two hospitals.
Participation of eligible cases was 99
and 87% for controls.
Outcome definition: Diagnosis of
nasopharyngeal was confirmed by
histological review with >90%
diagnosed with nonkeratinizing and
undifferentiated carcinomas and 9%
with squamous cell carcinoma.
Exposure assessment: Occupational
history obtained from interviews of
cases and controls for jobs held for
>1 year since age 16 and identified job
title, typical activities/duties, type of
industry, and tools and/or materials
used.
Industrial hygienist assigned Standard
Industry Classification/Standard
Occupational Classification codes to
jobs, assigning each a probability and
intensity of exposure on a 0 (not
exposed) to 9 (strong) scale.
Cumulative exposure defined as the
product of average intensity and
duration.
Internal comparisons:
All cases and controls
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. value)
[301]
Level 2 OR = 1.4 (0.93-2.2) [74]
Duration (overall):
Level 1 OR = 1.0 (Ref. value)
[301]
Level 2 OR = 1.3 (0.69-2.3) [31]
Level 3 OR = 1.6 (0.91-2.9) [43]
p-trend (exposed) = 0.08
Duration (excluding 10 vrs before diagnosis):
Level 1 OR = 1.0 (Ref. value)
[307]
This document is a draft for review purposes only and does not constitute Agency policy.
1-214 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Design: Population-based
case-control study of 375 cases of
nasopharyngeal cancer. 325 controls
identified from a random sample of
households from a national household
registration system and matched by
age, sex, and area of residence.
Analysis: RRs estimated by ORs
calculated by logistic regression and
adjusted for age, sex, education, and
ethnicity. An induction period of
10 years was also utilized to account
for latency in evaluating duration of
exposure.
All subjects were tested for the EBV;
subset analysis based on EBV
positivity (360 cases and 94 controls).
EBV seropositives defined as positive
for one of the following anti-EBV
antibodies known to be associated
with nasopharyngeal cancer: viral
capsid antigen IgA, EBV nuclear
antigen one IgA, early antigen IgA,
DNA binding protein IgG, and anti-
DNase IgG.
Related studies:
Yang et al. (2005); Cheng et
al. (1999); Hildesheim et al.
(1997)
Confidence in effect estimates:3
Si IB Cf Oth
LOW (Potential bias toward the
null)
IB: Exposure Group B
Oth: Low sensitivity due to
incomplete control of matching
factors.
Multiple exposure metrics including
average intensity, average probability,
cumulative, years since first exposure,
and age at first exposure were
evaluated.
Duration and timing: Duration and
timing of exposure were evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (no)
Level 2 (yes)
Duration (overall):
Level 1 (none)
Level 2 (<10 years)
Level 3 (>10 years)
Duration (excluding 10 yrs before
diagnosis):
Level 1 (none)
Level 2 (<10 years)
Level 3 (>10 years)
Cumulative exposure:
Level 1 (none)
Level 2 (<25 years)
Level 3 (>25 years)
Time since first exposure:
Level 1 (none)
Level 2 (<20 years)
Level 3 (>20 years)
Age at first exposure:
Level 1 (none)
Level 2 (<25 years)
Level 3 (>25 years)
Other exposures: wood dust, solvents,
and smoking.
[As noted in Appendix A.5.9: The
observed associations were not
materially affected when controlling
for wood dust, solvent exposure, or
smoking.]
Level 2 OR = 1.6 (0.89-3.0)
Level 3 OR = 1.2 (0.67-2.2)
Cumulative exposure:
Level 1 OR = 1.0 (Ref.
[301]
Level 2 OR = 1.3 (0.70-2.4)
Level 3 OR = 1.5 (0.88-2.7)
p-trend (exposed) = 0.10
Time since first exposure:
Level 1 OR = 1.0 (Ref.
[301]
Level 2 OR = 2.3 (0.95-5.8)
Level 3 OR = 1.2 (0.76-2.0)
Age at first exposure:
Level 1 OR = 1.0 (Ref.
[301]
Level 2 OR = 1.3 (0.80-2.0)
Level 3 OR = 3.4 (0.94-12)
[34]
[34]
value)
[29]
[45]
value)
[19]
[55]
value)
[62]
[12]
No notable findings were reported between
formaldehyde exposure and the risk of
nasopharyngeal cancer when considering an
induction period of 10 years.
Authors reported that the observed
associations were not materially affected
when analyses additionally controlled for
wood dust and solvent exposure.
Reference: Hildesheim et al.
(2001)
Exposure assessment: Occupational
history obtained from interviews of
cases and controls for jobs held for
>1 year since age 16 and identified job
title, typical activities/duties, type of
Internal comparisons:
EBV positive subjects
(based on 360 cases and 94 controls)
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. value) [# not
given]
This document is a draft for review purposes only and does not constitute Agency policy.
1-215 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
industry, and tools and/or materials
Level 2 OR = 2.7 (1.2-6.2) [# not
used.
given]
Industrial hygienist assigned Standard
Duration (overall):
Industry Classification/Standard
Level 1 OR = 1.0 (Ref. value) [# not
Occupational Classification codes to
given]
jobs, assigning each a probability and
Level 2 OR = 2.8 (0.83-9.7) [# not
intensity of exposure on a 0 (not
given]
exposed) to 9 (strong) scale.
Level 3 OR = 2.6 (0.87-7.7) [# not
Cumulative exposure defined as the
given]
product of average intensity and
duration.
Duration (excluding 10 vrs before diagnosis):
Level 1 OR = 1.0 (Ref. value) [# not
Multiple exposure metrics including
given]
average intensity, average probability,
Level 2 OR = 4.7 (1.1-20) [# not
cumulative, years since first exposure,
given]
and age at first exposure were
Level 3 OR = 1.7 (0.65-6.0) [# not
evaluated.
given]
Duration and timing: Duration and
Cumulative exoosure:
timing of exposure were evaluated.
Level 1 OR = 1.0 (Ref. value) [# not
given]
Variation in exposure:
Level 2 OR = 4.0 (0.92-17) [# not
Exposure to formaldehyde:
given]
Level 1 (no)
Level 3 OR = 2.2 (0.80-5.8) [# not
Level 2 (yes)
given]
Duration (overall):
Level 1 (none)
Time since first exposure:
Level 2 (<10 years)
Level 1 OR = 1.0 (Ref. value) [# not
Level 3 (>10 years)
given]
Duration (excluding 10 yrs before
Level 2 OR = 2.3 (0.52-10) [# not
diagnosis):
given]
Level 1 (none)
Level 3 OR = 2.8 (1.1-7.6) [# not
Level 2 (<10 years)
given]
Level 3 (>10 years)
Cumulative exposure:
Age at first exposure:
Level 1 (none)
Level 1 OR = 1.0 (Ref. value) [# not
Level 2 (<25 years)
given]
Level 3 (>25 years)
Level 2 OR = 2.6 (1.1-6.5) [# not
Time since first exposure:
given]
Level 1 (none)
Level 3 OR = 3.1 (0.39-24) [# not
Level 2 (<20 years)
given]
Level 3 (>20 years)
Age at first exposure:
No notable findings were reported between
Level 1 (none)
formaldehyde exposure and the risk of
Level 2 (<25 years)
nasopharyngeal cancer when considering an
Level 3 (>25 years)
induction period of 10 years.
Other exposures: wood dust, solvents.
and smoking.
Reference: Vaughan et al.
(2000)
Exposure assessment: Occupational
histories obtained from interviews of
cases and controls and identified job
Internal comparisons:
All histological types:
Exposure to formaldehyde:
This document is a draft for review purposes only and does not constitute Agency policy.
1-216 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Population: Males and females
between the ages of 18 and 74 who
were diagnosed with nasopharyngeal
cancer between April 1987 and July
1993 and identified from five
population-based cancer registries in
the United States. Interviews were
completed for 82% of eligible cases
and 76% of eligible controls.
Outcome definition: Diagnosis of
nasopharyngeal (any histological type)
was based on clinical records from
cancer registries. Histological typing
was reported and included for
analysis with 28% diagnosed with
undifferentiated and nonkeratinizing
carcinomas, 60% with differentiated
squamous cell carcinomas, and 12%
with epithelial carcinomas (not
otherwise specified[NOS]).
Design: Population-based
case-control study of 196 cases of
nasopharyngeal cancer. 244 controls
identified from random digit dialing in
the same geographic regions and
frequency matched by age, sex, and
cancer registry.
Analysis: ORs calculated by logistic
regression and adjusted for age, sex,
race, SEER site, cigarette usage, proxy
status, and education.
An induction period of 10 years was
also utilized to account for latency in
evaluating duration and cumulative
exposure. Results with and without
this 10-year lag period were similar.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward
the null)
IB: Exposure Group B
title, typical activities/duties, type of
industry, and start and stop dates.
Exposure was estimated by industrial
hygienists by linking occupational
history with participants' self-reported
exposure information.
Probability of exposure:
definitely not or unlikely (<10%),
possible (>10 and <50%),
probable (>50 and <90%), and
definite (>90%).
Jobs with potential exposure assigned
estimated concentration levels based
on TWA8: low (<0.10 ppm), moderate
(>0.10 and <0.50 ppm), and high
(>0.50 ppm).
Multiple exposure metrics including
probability of exposure and cumulative
exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (never)
Level 2 (ever)
Maximum exposure:
Level 1 (<0.10 ppm)
Level 2 (0.10 to 0.50 ppm)
Level 3 (>0.50 ppm)
Duration:
Level 1 (1 to 5 years)
Level 2 (6 to 17 years)
Level 3 (>18 years)
Other exposures: Wood dust.
[As noted in Appendix A.5.9: Wood
dust evaluated as an independent risk
factor for NPC controlling for
formaldehyde and it was not a risk
factor in this data set.]
: 1.0
Level 1 OR
[117]
Level 2 OR = 1.3 (0.8-2.1)
Maximum exposure:
Level 1 OR = 1.4 (0.8-2.4)
Level 2 OR = 0.9 (0.4-2.3)
Level 3 OR = 1.6 (0.3-7.1)
p-trend (exposed) = 0.57
Duration:
(Ref.
Level 1 OR = 0.8 (0.4-1.6)
Level 2 OR = 1.6 (0.7-3.4)
Level 3 OR = 2.1 (1.0-4.5)
p-trend (exposed) = 0.07
Epithelial (NOS)
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 3.1 (1.0-9.6)
Maximum exposure:
Level 1 OR = 4.0 (1.2-13.1)
Level 2 OR = 1.5 (0.2-13.9)
Level 3 no cases
p-trend (exposed) = 0.46
Duration:
Level 1 OR = 2.0 (0.4-9.8)
Level 2 OR = 4.0 (0.9-18.6)
Level 3 OR = 4.2 (0.8-21.5)
p-trend (exposed) = 0.036
Differentiated Squamous Cell
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.5 (0.8-2.7)
Maximum exposure:
Level 1 OR = 1.6 (0.8-3.0)
Level 2 OR = 1.2 (0.4-3.3)
Level 3 OR = 2.1 (0.4-12.3)
p-trend (exposed) = 0.32
Duration:
value)
[79]
[60]
[14]
[5]
[24]
[26]
[29]
[12]
[12]
[11]
[1]
Level 1 OR = 0.8 (0.3-2.0)
Level 2 OR = 1.8 (0.7-4.3)
Level 3 OR = 2.5 (1.1-5.9)
p-trend (exposed) = 0.033
Undifferentiated and nonkeratinizing
Exposure to formaldehyde:
[4]
[3]
[5]
[69]
[49]
[35]
[10]
[4]
[12]
[17]
[20]
Level 1 OR = 1.0 (Ref. value)
[36]
Level 2 OR = 0.9 (0.4-2.0)
[18]
Maximum exposure:
Level 1 OR = 1.0 (0.4-2.4)
[14]
Level 2 OR = 0.5 (0.1-3.1)
[3]
Level 3 OR = 1.5 (0.2-14.7)
[1]
p-trend (exposed) = 0.72
Duration:
Level 1 OR = 0.7 (0.3-2.2)
[8]
This document is a draft for review purposes only and does not constitute Agency policy.
1-217 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Level 2 OR = 1.0 (0.2-3.9) [6]
Level 3 OR = 1.2 (0.3-4.8) [4]
p-trend (exposed) = 0.82
Reference: Vaughan et al.
(2000)
Exposure assessment: Occupational
histories obtained from interviews of
cases and controls and identified job
title, typical activities/duties, type of
industry, and start and stop dates.
Exposure was estimated by industrial
hygienists by linking occupational
history with participants' self-reported
exposure information.
Probability of exposure:
definitely not or unlikely (<10%),
possible (>10 and <50%),
probable (>50 and <90%), and
definite (>90%).
Jobs with potential exposure assigned
estimated concentration levels based
on 8-h TWA: low (<0.10 ppm),
moderate (>10 and <50 ppm), and high
(>50 ppm).
Multiple exposure metrics including
probability of exposure and cumulative
exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (never)
Level 2 (ever)
Duration:
Level 1 (1 to 5 years)
Level 2 (6 to 17 years)
Level 3 (>18 years)
Cumulative exposure:
Level 1 (0.05 to 0.40 ppm-yrs)
Level 2 (>0.4 to 1.10 ppm-yrs)
Level 3 (>1.10 ppm-yrs)
Other exposures: Wood dust was
evaluated but not found to be a
confounder.
Internal comparisons:
Excluding undifferentiated
nonkeratinizing histological types
Possible, probable or definite exposure
Exposure to formaldehyde:
and
Level 1 OR = 1.0 (Ref. Value [# not given]
Level 2 OR = 1.6 (1.0-2.8)
[61]
Duration:
Level 1 OR = 0.9 (0.4-2.1)
[16]
Level 2 OR = 1.9 (0.9-4.4)
[20]
Level 3 OR = 2.7 (1.2-6.0)
[25]
p-trend (exposed) = 0.014
Cumulative exposure:
Level 1 OR = 0.9 (0.4-2.0)
[15]
Level 2 OR = 1.8 (0.8-4.1)
[22]
Level 3 OR = 3.0 (1.3-6.6)
[24]
p-trend (exposed) = 0.033
Probable or definite exposure
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. Value) [# not given]
Level 2 OR = 2.1 (1.1-4.2)
[27]
Duration:
Level 1 OR = 2.0 (0.8-5.0)
[12]
Level 2 OR = 3.3 (0.9-11.8)
[9]
Level 3 OR = 1.6 (0.5-5.6)
[6]
p-trend (exposed) = 0.069
Cumulative exposure:
Level 1 OR = 1.9 (0.7-4.9)
[12]
Level 2 OR = 2.6 (0.7-9.5)
[7]
Level 3 OR = 2.2 (0.7-7.0)
[8]
p-trend (exposed) = 0.13
Definite exposure
Exposure to formaldehyde:
Level 1 OR = 1.0 (Ref. Value) [# not given]
Level 2 OR = 13.3 (2.5-70)
[10]
Duration:
Level 1 OR = not reported
[5]
Level 2 OR = not reported
[2]
Level 3 OR = not reported
[3]
p-trend (exposed) <0.001
Cumulative exoosure:
Level 1 OR = not reported
[4]
Level 2 OR = not reported
[2]
Level 3 OR = not reported
[4]
p-trend (exposed) <0.001
This document is a draft for review purposes only and does not constitute Agency policy.
1-218 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Results with and without this 10-year lag
period were similar.
Reference: West et al. (1993)
Population: Male and female Filipinos
between the ages of 11 and 83 years
recruited from the Philippine General
Hospital and diagnosed prior to 1992.
Among 234 suspicious cases, 9%
refused biopsy and were excluded
and 104 were pathologically
confirmed as cases (Hildesheim
et al.. 1992). of which 100%
agreed to participate. All 104 hospital
controls agreed to participate while
only 77% of community controls
agreed to participate (Hildesheim
et al.. 1992).
Outcome definition: Diagnosis of
nasopharyngeal was confirmed by
histological review for all cases.
Histological typing not reported.
Design: Hospital-based case-control
study of 104 predominantly
non-Chinese cases of nasopharyngeal
cancer. 205 controls (104 hospital
and 101 community cases) matched
on gender, age, and hospital or
neighborhood.
Analysis: RRs estimated by ORs were
calculated by conditional logistic
regression and adjusted for
education, years since first exposure
to dust and exhaust fumes, smoking,
antimosquito coils, herbal medicines,
and diet including processed meats
and fresh fish.
Related studies:
Hildesheim et al. (1992)
Confidence in effect estimates:3
SB !& Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward
the null)
Exposure assessment: Occupational
history obtained by interview for all
participants. Occupational exposure to
formaldehyde classified by industrial
hygienist as likely or unlikely.
Multiple exposure metrics including
analysis by length of exposure, length
of exposure lagged 10 years, TSFE, and
age at first exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Time since first exposure:
Level 1 (never)
Level 2 (<25 years)
Level 3 (>25 years)
Antimosquito coil exposure:
Level 1 (never)
Level 2 ( daily)
Length of exposure:
Level 1 (never)
Level 2 (<15 years)
Level 3 (>15 years)
Length of exposure lagged 10 years:
Level 1 (no)
Level 2 (<15 years)
Level 3 (>15 years)
Time since first exposure:
Level 1 (never)
Level 2 (<25 years)
Level 3 (>25 years)
Level 4 (>35 years)
Age at first exposure:
Level 1 (never)
Level 2 (<25 years)
Level 3 (>25 years)
Other exposures: dust and exhaust
exposure, fresh or salted fish
consumption, smoking, antimosquito
coils, and herbal medicines.
Note: Independent testing of six
brands of East Asian mosquito coils
evaluated the emission rates of
carbonyl compounds in the mosquito
smoke and reported that
Internal comparisons:
Multivariate results from Table 4 in West et
al.
Time since first exposure:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.2 (0.41-3.6)
Level 3 OR = 4.0 (1.3-12.3)
Antimosquito coil exposure:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.4 (0.64-2.8)
Level 3 OR = 5.9 (1.7-20.1)
[75]
[12]
[14]
[59]
[24]
[21]
Additional: Bivariate results adjusted only
for dust/exhaust from Table 1
Length of exposure (bivariate):
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 2.7 (1.1-6.6)
Level 3 OR = 1.2 (0.48-3.2)
[75]
[19]
[8]
Length of exposure lagged 10 years
(bivariate):
(Reference value included eight cases and
three controls exposed only in the 10 years
before diagnosis)
Level 1 OR = 1.0 (Ref. value) [83]
Level 2 OR = 1.6 (0.65-3.8) [11]
Level 3 OR = 2.1 (0.70-6.2) [8]
Age at first exposure (bivariate):
Level 1 OR = 1.0 (Ref. value) [75]
Level 2 OR = 2.7 (1.1-6.6) [16]
Level 3 OR = 1.2 (0.47-3.3) [11]
Time since first exposure (bivariate):
Level 1 OR = 1.0 (Ref. value) [75]
Level 2 OR = 1.3 (0.65-3.8) [12]
Level 3 OR = 2.9 (1.1-7.6) [14]
Time since first exposure (bivariate):
Level 4 OR = 5.6 (0.58-52.9) [5]
Authors noted that stronger effects were not
evident among those considered most likely
to have been exposed or most likely to have
been exposed to high doses.
This document is a draft for review purposes only and does not constitute Agency policy.
1-219 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
formaldehyde and acetaldehyde had
the highest emission rates (Liu et
al.. 2003). Among the three
experiments on each of the six brands,
the range of formaldehyde
concentrations was from 0.87 ng/m3
(0.7 ppb) to 25 ng/m3 (20 ppb).
[As noted in Appendix A.5.9. Control
for mosquito coils may have
underestimated the estimated effect
of formaldehyde.]
Exposure assessment: Occupational
history obtained by city directories and
death certificates, which yielded
information on job, industry,
employer, and year of employment.
Exposure classification scheme based
on potential for formaldehyde
exposure, probability of exposure for
each participant and each job-industry
pair, and level of exposure.
Exposure level and timing of exposure:
Level 1 OR = 1.0 (Ref. value) [# not
given]
Level 2 OR = 1.0 (0.6-1.7)
Level 3 OR = 1.3 (0.7-2.4)
Study
IB: Exposure Group C
Cf: Controlling for other sources of
formaldehyde may have
underestimated effect of main
formaldehyde exposures.
Reference: Roush et al. (1987b)
Population: Males identified from the
Connecticut Tumor Registry who died
of any cause during 1935-1975.
Outcome definition: Diagnosis of
nasopharyngeal cancer based on case
registration by the Connecticut Tumor
Registry. Clinical records reviewed for
>75% of cases. Histological typing not
reported.
Design: Population-based
case-control study of 173 male cases
of nasopharyngeal cancer. Controls
Probability of exposure defined as
unexposed, possibly exposed, probably
exposed, or definitely exposed.
Results: effect estimate (95% CI)
[# of Cases]
[21]
[17]
High exposure level and timing of exposure:
Level 1 OR = 1.0 (Ref. value) [# not
given]
Level 2 OR = 1.4 (0.6-3.1) [9]
Level 3 OR = 2.3 (0.9-6.0) [7]
High exposure level and timing of exposure:
Level 3 OR = 4.0 (1.3-12.0) [6]
were 605 males dying in Connecticut
during the same time period,
randomly selected from state death
certificates.
Analysis: ORs calculated by logistic
regression and adjusted for age at
death, year at death, and availability
of occupational information.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward
the null)
IB: Exposure Group C
Level of exposure estimated as zero,
low (<1 ppm), and high (>1 ppm).
Among those probably exposed to
some level of formaldehyde for most
of their working lifetime, the extent
and level of exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure level and timing of exposure:
Level 1 (unexposed)
Level 2 (probably exposed most of
working life)
Level 3 (probably exposed most of
working life and probably
exposed 20+ years before
death)
High exposure level and timing of
exposure:
Level 1 (unexposed)
Additional: Age of Death 68+
This document is a draft for review purposes only and does not constitute Agency policy.
1-220 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Level 2 (probably exposed most of
working life and probably
to high level in some year)
Level 3 (probably exposed most of
working life and probably
exposed to high level
20+ years before death)
Other exposures: Not evaluated as
potential confounders.
[As noted in Appendix A.5.9: Exposure
to wood dust was not found to be a
risk factor for all nasal cancers
(NPC + SNC). This suggests a lower
potential for confounding by wood
dust.l
Reference: Olsen et al. (1984)
Population: Male and females linked
to the Danish Cancer Registry during
1970-1982.
Outcome definition: Diagnosis of
cancer of the nasopharynx based on
ICD code 146 from Registry data. 9%
of nasopharyngeal cases were
sarcomas and 91% were carcinomas.
Sarcomas were excluded but
gender-specific case counts were not
provided for carcinomas.
Design: Population-based
case-control study of 266 cases of
nasopharyngeal cancer. Three
controls per case were selected for
the same distributions of age, sex, and
year of diagnosis as cases.
Analysis: OR calculated using
programs developed by Rothman
and Boice (1979).
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
Exposure assessment: Employment
histories from 1964 maintained by
Danish Cancer Registry. Occupational
exposures estimated by industrial
hygienists based on industries or
occupations considered to have certain
or probably exposure. Authors
reported that 4.2 and 0.1% of control
males and females, respectively, were
exposed to formaldehyde.
Duration and timing: Exposure period
starting at 1964. Exposure to
formaldehyde may have been between
0 and 20 years depending on when
first exposed during the define
exposure period.
Variation in exposure:
Occupational exposure:
Level 1 (no exposure)
Level 2 (ever exposed)
Time since first exposure:
Level 1 (<10 years)
Level 2 (>10 years)
Coexposures: Coexposure evaluated
included: wood dust, paint, lacquer,
and glue.
[As noted in Appendix A.5.9
Wood dust is associated with SNC and
Internal comparisons:
Occupational exposure:
Men [=196 (91% of 215)]
Level 1 RR = 1.0 (Ref. value) [# not
given]
Level 2 RR = 0.7 (0.3-1.7) [# not
given]
Women [=90 (91% of 99)]
Level 1 RR = 1.0 (Ref. value) [# not
given]
Level 2 RR = 2.6 (0.3-21.9) [# not
given]
Time since first exposure:
No evidence of association (data not shown).
MEDIUM (Potential bias toward
the null)
IB: Exposure Group C
was evaluated as a potential
confounder of NPC but was not a risk
factor.]
This document is a draft for review purposes only and does not constitute Agency policy.
1-221 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Reference: Coggon et al. (2014)
Population: 14,008 British men
employed in six chemical industry
factories which produced
formaldehyde. Cohort mortality
followed from 1941 through 2012.
Cause of deaths was known for 99%
of 5,185 deaths through 2000. Similar
cause of death information not
provided on 7,378 deaths through
2012. Vital status was 98.9%
complete through 2003. Similar
information not provided on deaths
through 2012.
Outcome definition: Death
certificates used to determine cause
of deaths from nasopharyngeal
cancer.
Design: Cohort mortality study with
external comparison group with a
nested case-control study.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
Overall
SB
IB Cf
Oth
Confidence
J
LOW
¦
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure is Group B; Lack of
latency analysis.
Oth: Low power due to rarity of cases.
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration
and timing since first exposure were
not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated. Potential
low-level exposure to stvrene.
ethylene oxide, epichlorhydrin,
solvents, asbestos, chromium salts,
and cadmium.
[As noted in Appendix A.5.9: Stvrene is
associated with LHP cancers but not
URT cancers.
Asbestos is associated with URT
cancers, but not this outcome.
Other coexposures are not known risk
factors for this outcome.]
External comparisons:
Exposed:
Observed: 1 deaths
Expected: 1.7 deaths
SMRExposed = 0.59 (0.03-2.90)f
[1]
fEPA derived confidence intervals for the
SMRs using Fischer's Exact method (See
Armitage and Cullis (1971);
Snedecor and Cochran (1980) for
nonzero SMRs and using the Mid-P method
See Rothman and Boice (1979).
Reference: Meyers et al. (2013)
Population: 11,043 workers in 3 U.S.
garment plants exposed for at least
3 months. Women comprised 82% of
the cohort. Vital status was followed
through 2008 with 99.7% completion
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
1984 with 12-73 within each
department. Formaldehyde levels
across all departments and facilities
were similar.
External comparisons:
SMR = 0(0-2.77)
[0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-222 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Outcome definition: Death
certificates used to determine both
the underlying cause of death from
nasopharyngeal cancer (ICD code in
use at time of death). Histological
typing not provided.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates.
Related studies:
Stayner et al. (1985)
Stayner et al. (1988)
Pinkerton et al. (2004)
Confidence in effect estimates:3
SB 13 Cf Oth
Overall
Confidence
Low
Geometric TWA8 exposures ranged
from 0.09-0.20 ppm. Overall
geometric mean concentration of
formaldehyde was 0.15 ppm, (GSD
1.90 ppm). Area measures showed
constant levels without peaks.
Historically earlier exposures may have
been substantially higher.
Duration and timing: Exposure period
from 1955 to 1983. Median duration
of exposure was 3.3 years. More than
40% exposures <1963. Median time
since first exposure was 39.4 years.
Duration and timing since first
exposure were not evaluated for this
cancer.
Variation in exposure: Not evaluated.
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
LOW (Potential bias toward the
null)
IB: Exposure Group A; Lack of latency
analysis.
Oth: Low power due to rarity of cases.
Reference: Siew et al. (2012)
Population: All Finnish men born
during 1906-1945 who participated in
census and were employed in 1970
(n = 1.2 million). Vital status was
"virtually complete."
Outcome definition: Diagnosis of
cancer reported to the Finnish Cancer
Registry.
Design: Prospective national cohort
incidence study with internal
comparison groups.
Analysis: RRs calculated controlling
for sex, age, socioeconomic status,
period of follow-up, and smoking.
Confidence in effect estimates:3
Exposure assessment: Individual-level
exposure estimates based on matching
occupations listed in the census to the
Finnish job-exposure matrix which
covers major occupational exposures
and provided exposure estimates for
formaldehyde.
Duration and timing: Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (none)
Level 2 (any)
Coexposures: Wood dust exposures
were controlled for in analyses.
Internal comparisons:
Exposure to formaldehyde:
Level 1 RR = 1.00 (Ref. value) [144]
Level 2 RR = 0.87 (0.34-2.20) [5]
This document is a draft for review purposes only and does not constitute Agency policy.
1-223 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
Exposure assessment: Occupational
history obtained from interviews of
cases and controls for jobs held for
>1 year since age 16 and identified job
title, typical activities/duties, type of
industry, and tools and/or materials
used.
Industrial hygienist assigned Standard
Industry Classification/Standard
Occupational Classification codes to
jobs, assigning each a probability and
intensity of exposure on a 0 (not
exposed) to 9 (strong) scale.
Cumulative exposure defined as the
product of average intensity and
duration.
Internal Comparisons:
Cumulative exposure:
Level 1 OR = 1.0 (Ref. value) [# not
given]
Level 2 OR = 1.03 (0.60-1.76) [# not
given]
Level 3 OR = 1.31 (0.87-1.97) [# not
given]
Familial cases (n = 502) compared to
population controls (n = 327)
Study
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
IB: Exposure Group D
Oth: Low power due to rarity of
exposure.
Reference: Yang et al. (2005)
Population: Taiwanese men and
women from 325 families which had
two or more nonparent-offspring
family members diagnosed with
nasopharyngeal cancer (other first-,
second-, or third-degree relatives).
Cases were identified from the
national tumor registry.
Outcome definition: Diagnosis of
incident nasopharyngeal cancer was
confirmed by histological review for
all cases (n = 502). An earlier report
on 375 cases from the same series
reported >90% diagnosed with
nonkeratinizing and undifferentiated
carcinomas and 9% with squamous
cell carcinoma Hildesheim et al.
(2001)
Design: Family-based case-control
study of nasopharyngeal cancer.
Cases from high-risk families were
compared to two controls groups.
Initial set of 375 cases reported by
Cheng et al. (1999) had a 99%
occupational questionnaire response
rate. Similar data were available for
60% of new cases (n = 127) with the
remainder considered to be missing at
random. Overall case response rate is
85%.
The Family control groups consisted
of up to five unaffected siblings, the
parents of affected subjects, or
spouses of affected cases' children
(n = 1,944; participation rate not
given). Population controls (n = 327;
88% response rate) were originally
matched to a subset of cases accrued
at an earlier time (n = 375) matched
Duration and timing: Duration was
evaluated as a component of the
cumulative exposure score. The timing
of exposure was not evaluated.
Results: effect estimate (95% CI)
[# of Cases]
Familial cases (n = 502) compared to Family
controls (n = 1,944)
Cumulative exposure (lntensity*duration):
Level 1 OR = 1.00 (Ref. value) [# not
given]
Level 2 OR = 1.30 (0.70-2.39) [# not
given]
Level 3 OR = 4.29 (2.45-7.51) [# not
given]
Variation in exposure:
Intensity scored 0-9
Duration in years
Cumulative exposure
(lntensity*duration):
Level 1 (none)
Level 2 (<25)
Level 3 (>25)
Other exposures: smoking, betel nut
use, wood exposure, and salted fish
consumption which were not
controlled for in the analysis.
[As noted in Appendix A.5.9: In this
study, smoking was inversely
associated with NPC. Since smoking is
positively associated with
This document is a draft for review purposes only and does not constitute Agency policy.
1-224 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
on age, sex and residence (Cheng
et al.. 1999). The same population
controls and cases were later
augmented with additional cases to
encompass the total of 502 cases.
Analysis: For the Family controls, ORs
were calculated by conditional logistic
regression matched on family. For
the Population controls, OR's were
calculated by unconditional logistic
regression controlling for age and sex;
however, while population controls
were originally matched on residence,
residence was not controlled for in
this later analysis.
Related studies:
Hildesheim et al. (2001);
Cheng et al. (1999);
Hildesheim et al. (1997)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias using next
of kin only among the cases which
may result in poorer quality exposure
data and a bias toward the null.
Cf: Negative confounding possible.
Oth: Low sensitivity due to
incomplete control of matching
factors.
formaldehyde, there may be negative
confounding by smoking in this study.]
Reference: Yu et al. (2004)
Population: Deceased male and
female restaurant workers who died
during 1986-1995 and were
registered as union members by four
major Chinese-style restaurant
workers' unions in Hong Kong
(n = 1,225).
Outcome definition: Underlying cause
of death from nasopharyngeal cancer
(ICD-9: 147) obtained from the Hong
Exposure assessment: Occupational
history obtained from union records.
Waiters, waitresses and kitchen
workers presumed to be exposed to
formaldehyde based on independent
studies of air quality from the kitchen
exhausts of Hong Kong restaurants
(Ho etal.. 2006b: EHS
Consultants Ltd.. 1999)
Internal Comparisons:
Male and female (Waiters and waitresses)
Note:
Ho et al. (2006b) reported time-
averaged formaldehyde concentrations
Wait staff cases compared to kitchen worker
controls
MOR = 2.53 (1.01-6.36) [21]
Male only (Waiters)
Wait staff cases compared to kitchen worker
controls
MOR = 2.61 (1.02-6.69) [17]
External Comparisons:
This document is a draft for review purposes only and does not constitute Agency policy.
1-225 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Kong Census and Statistics
Department (n = 29). Cause of death
available for more than 80% of
restaurant workers. Histological
typing not reported.
Design: Mortality odds ratio where
cases are deaths from nasopharyngeal
cancer and controls are deaths from
all other causes of death after
excluding cancer. Internal control
group composed of other deceased
kitchen workers. External control
group composed of all noncancer
deaths from the general population in
Hong Kong.
Analysis: Mortality odds ratios
(MORs) based on the internal control
group were calculated by logistic
regression controlling for sex, age at
death, year of death, and place of
origin. For the external control group,
MORs were calculated by logistic
regression controlling for sex, age at
death, and year of death.
Related studies:
Hoetal. (2006a)
EHS Consultants Ltd. (1999)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null)
IB: Exposure Group C; Latency not
evaluated.
Cf: Potential confounding by smoking.
at Chinese restaurants in Hong Kong
were reported as high as 249 ppb (306
Mg/m3).
The Hong Kong Environmental
Protection Department survey of
indoor air at local restaurants reported
a mean formaldehyde concentration of
162 ng/m3 with a high value of 975
Mg/m3 (EHS Consultants Ltd..
1999).
Duration and timing: Duration of
exposure was evaluated based on
length of restaurant union
membership.
Variation in exposure:
Cumulative exposure:
Level 1 (none)
Level 2 (<15 yrs union membership)
Level 3 (16-24 yrs union
membership)
Level 4 (>25 yrs union membership)
Other exposures: not evaluated. Wait
staff exposed to other sources of
formaldehyde such as environmental
tobacco smoke, furniture, carpeting,
and room partitions made of plywood
and fiberboard, which are not shared
by kitchen staff.
[As noted in Appendix A.5.9: Smoking
was evaluated as a potential
confounder because 49% of staff
smoked compared to 27% of
population, but it was insufficient to
explain the observed effects.]
Male and female (Waiters and waitresses)
Wait staff cases compared to general Hong
Kong male and female population controls
MOR = 3.28 (2.08-5.16) [21]
Male only (Waiters)
Wait staff cases compared to general Hong
Kong male population controls
MOR = 3.02 (1.82-5.00) [17]
Male only (Waiters)
Cumulative exposure:
Level 1 MOR = 1.00 (Ref. Value) [3,225]
Level 2 MOR = 2.50 (1.14-5.49) [7]
Level 3 MOR = 3.41 (1.56-7.45) [7]
Level 4 MOR = 3.75 (1.12-12.54) [3]
Female only (Waitresses)
Wait staff cases compared to general Hong
Kong female population controls
MOR = 4.58 (1.63-12.86) [4]
Reference: Hansen and Olsen
(1995)
Population: 2,041 men with cancer
who were diagnosed during
1970-1984 and whose longest work
experience occurred at least 10 years
before cancer diagnosis. Identified
from the Danish Cancer Registry and
matched with the Danish
Supplementary Pension Fund.
Ascertainment considered complete.
Exposure assessment: Individual
occupational histories including
industry and job title established
through company tax records to the
national Danish Product Register.
Subject were considered to be exposed
to formaldehyde if: (1) they had
worked in an industry known to use
more than 1 kg formaldehyde per
employee per year and (2) subject's
longest single work experience (job) in
External comparisons:
Overall (exposure to formaldehyde >10 years
prior to cancer diagnosis)
SPIR = 1.3 (0.3-3.2) [4]
This document is a draft for review purposes only and does not constitute Agency policy.
1-226 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
Pension record available for 72% of
cancer cases.
Outcome definition: Nasopharyngeal
cancer (ICD-7:146) listed on Danish
Cancer Registry file. Histological
typing not reported.
Design: Proportionate incidence study
with external comparison group.
Analysis: Standardized proportionate
incidence ratio calculated as the
proportion of cases for a given cancer
in formaldehyde-associated
companies relative to the proportion
of cases for the same cancer among
all employees in Denmark. Adjusted
for age and calendar time.
Confidence in effect estimates:3
that industry since 1964 was >10 years
prior to cancer diagnosis.
Duration and timing: Exposure period
not stated. Based on date of diagnosis
during 1970-1984, and the
requirement of exposure more than
10 years prior to diagnosis, the
approximate period was 1960-1974.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated for
potential confounding
[As noted in Appendix A.5.9: While
other coexposures were not evaluated,
the overall correlation between
coexposures in multiple occupational
industries is likely to be low.]
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null)
IB: Exposure Group D
Oth: Low power due to rarity of cases.
Reference: Malker et al. (1990)
Population: Employed Swedish men
newly diagnosed with nasopharyngeal
cancer identified during 1961-1979
registered by the Swedish Cancer-
Environment Registry.
Outcome definition: Microscopic
confirmation obtained for 99.6% of
nasopharyngeal cases. Squamous cell
carcinomas constituted 48% of cases
with 37% classified as unspecified
carcinomas, 5% transitional cell
carcinomas, and 3%
adenocarcinomas.
Design: Population-based
standardized incidence ratio study of
471 incidence cases of
nasopharyngeal cancer compared to
expected number of cases among
men in occupational groups defined
by employment in 1960.
Exposure assessment: Occupations
presumed to be exposed to
formaldehyde.
Duration and timing: Duration and
timing of exposure were not evaluated.
Variation in exposure: Occupation and
industry
Coexposures: Not evaluated as
potential confounders.
[As noted in Appendix A.5.9: Wood
dust is associated with URT cancers
and would likely be positively
correlated with formaldehyde
exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.]
External comparisons:
Occupation
Glassmakers
SIR = 6.2 (1.58-16.87)f
Bookbinders
SIR = 6.1 (1.55-16.59)f
Shoemakers
SIR = 3.8 (1.39-8.42)f
Industry
Shoe repair
SIR = 4.0 (1.47-8.87)f
Fiberboard plant
SIR = 3.9 (1.24-9.40)f
[3]
[3]
[5]
[5]
[4]
fEPA derived CIs using the Mid-P Method
(See Rothman and Boice, 1979)
This document is a draft for review purposes only and does not constitute Agency policy.
1-227 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Analysis: SIRs calculated as the ratio
of observed to expected cases of
nasopharyngeal cancer.
Related studies:
Malker et al. (1990)
Confidence in effect estimates:3
SB IE Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
IB: Exposure Group D; Latency not
evaluated.
Cf: Potential confounding.
Reference: Vaughan (1989)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System during
1980-1983. Participation for all cases
was 68.7 and 80.0% for controls.
Outcome definition: Diagnosis of
nasopharyngeal cancer based on
review of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates. Nonsquamous cell
cancers were excluded from the
study.
Design: Population-based,
case-control study of 21 cases with
nasopharyngeal cancer. 552 controls
were identified by random digit
dialing in same geographic area.
Analysis: ORs were calculated by
logistic regression and adjusted for
age, gender, and race. Induction
periods were evaluated.
Related studies:
Vaughan et al. (1986a,
1986b)
Confidence in effect estimates:3
Exposures
Variation in exposure: Occupation and
industry
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder.
~50% of cases interviews completed by
next of kin. May result in poorer
quality exposure data and a bias
toward the null.]
Results: effect estimate (95% CI)
[# of Cases]
Exposure assessment: Presumed
exposure to formaldehyde. Interview-
based information on lifetime
occupational history byjob type and
industry.
Occupations evaluated for both no lag
and 15-year lag time between recent
exposure and diagnosis.
Duration and timing: Duration and
timing of exposure were evaluated.
Duration:
Level 1 (unexposed)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Other exposures: Not evaluated as
potential confounders.
fAs noted in Appendix A.5.9: Wood
Internal comparisons:
Carpenter (lagged 15 years)
All Industries:
OR = 4.5 (1.1-18.7) [3]
All Industries by Duration:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.6 (not provided)
Level 3 OR = 12.4 (not provided)
Chi2 trend = 8.65 (p = O.Olf
Carpenter (lagged 15 years)
Construction industry:
OR = 6.8 (1.6-28.2) [3]
Construction by Duration:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 2.1 (not provided)
Level 3 OR = 31.8 (not provided)
Chi2 trend = 14.86 (p = 0.0006)f
Food Service (lagged 15 years)
All Industries:
OR = 1.8 (0.6-5.7) [4]
All Industries by Duration:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.6 (not provided)
Level 3 OR = 4.0 (not provided)
Chi2 trend = 1.65 (p = 0.44)f
Food Service (lagged 15 years)
Retail Trade:
OR = 1.9 (0.5-6.9) [3]
This document is a draft for review purposes only and does not constitute Agency policy.
1-228 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
SB IB Cf Oth
Overs!)
Confidence
Low
LOW \1/ (Potential bias toward the
null;
Low sensitivity)
IB: Exposure Group D
SB: Potential selection bias using next
of kin only among the cases which
may result in poorer quality exposure
data and a bias toward the null.
Oth: Low power due to rarity of cases.
Retail Trade by Duration:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.4 (not provided)
Level 3 OR = 9.3 (not provided)
Chi2 trend = 2.21 (p = 0.33)f
fEPA computed p-value assuming 2 d.f.
Reference: Vaughan et al.
(1986a)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System during
1980-1983. Participation for all cases
was 68.7 and 80.0% for controls.
Outcome definition: Diagnosis of
nasopharyngeal cancer based on
review of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates. Histological typing
not reported; however, according to
Vaughan (1989), 6 cases were
nonsquamous cell cancers.
Design: Population-based,
case-control study of 27 cases with
nasopharyngeal cancer. 552 controls
were identified by random digit
dialing in same geographic area.
Analysis: ORs were calculated by
logistic regression and adjusted for
cigarette smoking and ethnic origin.
Induction periods were evaluated.
Related studies:
Vaughan (1989); Vaughan et
al. (1986b)
Confidence in effect estimates:3
Exposure assessment: Interview-based
information on lifetime occupational
exposure to formaldehyde with cases,
next of kin, and controls. Exposure
from available hygiene data, NIOSH
and other data, and NCI job-exposure
linkage system.
Multiple exposure metrics including
intensity, # of years exposed, and
exposure score based on the sum of
# years spent per job weighted by
estimated formaldehyde level were
evaluated. Exposure score calculated
for both no lag and 15-year lag time
between recent exposure and
diagnosis.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Intensity of exposure:
Level 1 (background)
Level 2 (low)
Level 3 (medium or high)
Number of years exposed:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Exposure score (no lag):
Level 1 (0 to 4)
Level 2 (5 to 19)
Level 3 (>20)
Exposure score (15-year lag):
Level 1 (0 to 4)
Level 2 (5 to 19)
Level 3 (>20)
Internal comparisons:
Intensity of exposure:
Level 1 OR = 1.0 (Ref. value) [16]
Level 2 OR = 1.2 (0.5-3.3) [7]
Level 3 OR = 1.4 (0.4-4.7) [4]
Number of years exposed:
Level 1 OR = 1.0 (Ref. value) [16]
Level 2 OR = 1.2 (0.5-3.1) [8]
Level 3 OR = 1.6 (0.4-5.8) [3]
Exposure score (no lae):
Level 1 OR = 1.0 (Ref. value) [21]
Level 2 OR = 0.9 (0.2-3.2) [3]
Level 3 OR = 2.1 (0.6-7.8) [3]
Exposure score (15-vear lag):
Level 1 OR = 1.0 (Ref. value) [21]
Level 2 OR = 1.7 (0.5-5.7) [4]
Level 3 OR = 2.1 (0.4-10.0) [2]
Additional:
Excluding Next of Kin Interviews
Exposure score (no lag):
Level 1 OR = 1.0 (Ref. value) [# not
given]
Level 2 OR = 1.1 (0.2-5.5) [# not
given]
Level 3 OR = 2.2 (0.4-10.8) [# not
given]
Exposure score (15-vear lag):
Level 1 OR = 1.0 (Ref. value) [# not
given]
Level 2 OR = 1.4 (0.3-7.3) [# not
given]
Level 3 OR = 3.1 (0.6-15.4) [# not
given]
[15]
This document is a draft for review purposes only and does not constitute Agency policy.
1-229 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of Cases]
SB
IB Cf Oth
Overall
Confidence
Low
1
h
LOW (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias using next
of kin only among the cases which
may result in poorer quality exposure
data and a bias toward the null.
Other exposures: Not evaluated as
potential confounders.
[As noted in Appendix A.5.9: Wood
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder. However, as this is a
case-control study the correlation
between formaldehyde and wood dust
is expected to be small and thus wood
dust would not be expected to be a
confounder.]
Reference: Vaughan et al.
(1986b)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System between 1980
and 1983. Participation for all cases
was 68.7 and 80.0% for controls.
Outcome definition: Diagnosis of
nasopharyngeal cancer based on
review of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates. Histological typing
not reported; however, according to
Vaughan (1989), 6 cases were
nonsquamous cell cancers.
Design: Population-based,
case-control study of 27 cases with
nasopharyngeal cancer. 552 controls
were identified by random digit
dialing in same geographic area.
Analysis: ORs were calculated by
multiple logistic regression and
adjusted for cigarette smoking and
ethnic origin.
Related studies:
Vaughan (1989); Vaughan et
al. (1986a. 1986b)
Confidence in effect estimates:3
Exposure assessment: Interview-based
information on lifetime occupational
history and residential history from
cases, next of kin, and controls.
Multiple exposure metrics including
type of dwelling (i.e., mobile home)
and use of particleboard or plywood
were evaluated.
Duration and timing: Exposure period
since 1950. Duration of exposure was
evaluated.
Variation in exposure:
Lived in a mobile home:
Level 1 (no)
Level 2 (yes)
Lived in a mobile home (lagged
15 years):
Level 1 (no)
Level 2 (yes)
Years of residence in mobile home:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Years of exposure to particleboard or
plywood:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Mobile home exposures (lagged
15 years):
Level 1 (none)
Level 2 (occupation only)
Level 3 (mobile home only)
Level 4 (both)
Note: The majority (84%) of mobile
homes in the United States at about
this time were reported to have mean
Internal comparisons:
Lived in mobile home:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 3.0 (1.2-7.5)
[19]
[8]
Lived in mobile home (lagged 15 years):
Level 1 OR = 1.0 (Ref. value) [24]
Level 2 OR = 3.0 (0.8-11.2) [3]
Years of residence in mobile home:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 2.1 (0.7-6.6)
Level 3 OR = 5.5 (1.6-19.4)
Years of exposure to particleboard or
plywood:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 1.4 (0.5-3.4)
Level 3 OR = 0.6 (0.2-2.3)
[19]
[4]
[4]
[17]
[6]
[4]
Mobile home exposures (lagged 15 years):
Level 1 OR = 1.0 (Ref. value) [15]
Level 2 OR = 1.7 (0.5-5.7) [4]
Level 3 OR = 2.8 (1.0-7.9) [6]
Level 4 OR = 6.7 (1.2-38.9) [2]
Additional:
Excluding Next of Kin Interviews
Lived in mobile home:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 2.8 (0.9-8.8)
[15]
[10]
[5]
This document is a draft for review purposes only and does not constitute Agency policy.
1-230 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Study
SB ib a
MH-
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias using next
of kin only among the cases which
may result in poorer quality exposure
data and a bias toward the null.
Cf: Low potential for confounding.
Exposures
formaldehyde exposures in excess of
100 ppb, with 22% having mean
exposures in excess of 500 ppb
(Breysse (1984) as cited in WHO
Coexposures: Not evaluated.
Information on occupational exposures
provided in Vaughan et al. (1986a).
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder. However, as this is a
case-control study the correlation
between formaldehyde and wood dust
is expected to be small and thus wood
dust would not be expected to be a
confounder.]
Results: effect estimate (95% CI)
[# of Cases]
[As noted in Appendix A.5.9: Wood
Overall
Confidence
LOW
(1989).
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9. SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Sinonasal cancer
Epidemiological evidence
The most specific classification of sinonasal cancer diagnosis commonly reported across the
epidemiological literature has been based on the first three digits of the Seventh, Eighth or Ninth
Revision of the ICD code (i.e., Malignant neoplasm of nose, nasal cavities, middle ear and accessory
sinuses ICD-7/8/9: 160), although some studies did report the histological type of cancer
(i.e., squamous cell carcinoma and adenocarcinoma).
Evidence of an association between formaldehyde exposure and the risk of developing or
dying from sinonasal cancer was available from 20 epidemiological studies—7 case-control studies
(Mavr etal.. 2010: D'Errico etal.. 2009: Pesch etal.. 2008: Luce etal.. 2002: Teschke etal.. 1997:
Roush etal.. 1987b: Olsen and Asnaes. 1986) and 12 cohort studies (Coggon etal.. 2014: Beane
Freeman etal.. 2013: Meyers etal.. 2013: Siew et al.. 2012: Takobsson etal.. 1997: Hansen and
Olsen. 1995: Hayes etal.. 1990: Bertazzi et al.. 1989: Stroup etal.. 1986: Levine etal.. 1984a:
Walrath and Fraumeni. 1984.19831. One study, fLuce etal.. 20021. combined 12 other case-control
studies in a pooled analysis of occupational exposures using a common protocol of standardized
This document is a draft for review purposes only and does not constitute Agency policy.
1-231 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
questionnaires and standardized exposure classifications.21 The results of this pooled analysis of
original primary data across studies fLuce etal.. 20021 are included in place of those from the 12
individual studies that are listed under "Related studies" in Table 1-33 for Luce et al. (2002). The
outcome-specific evaluations of confidence in the precise effect estimate of an association from
each study are provided in Appendix A.5.9. Three sets of reported results from Mayr et al. (2010),
d'Errico et al. (2009). and Harrington and Oakes (1984) were classified as uninformative due to
multiple biases and uncertainties; for details see Appendix A.5.9. Details of the reported results of
these studies are provided in the evidence table for sinonasal cancer (see Table 1-33) following the
causal evaluation.
Consistency of the observed association
Seventeen informative studies reported risks of sinonasal cancer among study subjects with
formaldehyde exposure based on occupational history. These studies examined different
populations, in different locations, under different exposure settings, and used different study
designs. For sinonasal cancer, it is important to consider the histological subtype or types in each
report (squamous cell carcinoma, adenocarcinoma, or mixed). The study results presented in
Table 1-33 (by confidence level and publication date) detail all of the reported associations. One
additional study (Andjelkovich et al.. 1995) reported zero cases of SNC among 3,929 U.S. workers
exposed to formaldehyde over 83,064 person-years but reported no data on the number of
expected cases and thus was not included here. 22
Sinonasal cancer is exceedingly rare with expected rates of 0.6 cases per 100,000 people
each year (Curado et al.. 2007). Many of these cohort studies lacked the statistical sensitivity to
detect an association with formaldehyde; eight of 12 cohort studies reported zero cases in their
study populations and all but one cohort study (Beane Freeman et al.. 2013) were classified with
low confidence. For such rare cancers, case-control studies can often be the most informative study
design.
Of the nine studies that did observe cases of sinonasal cancer, results from six reported
increased risks of sinonasal cancer that appeared to be associated with exposure to
formaldehyde—four of six sets of results had been classified with medium confidence (Beane
Freeman etal.. 2013: Luce etal.. 2002: Roush etal.. 1987b: Olsen and Asnaes. 1986) and two with
low confidence (Teschke etal.. 1997: Hansen and Olsen. 1995). Each of the other three sets of
21Note the pooled study by Luce et al. (2002) includes data from 12 publications and thus represents
substantially more information than a single result. The references for the source data are: Leclerc et al.
fl994"l: Luce etal. fl993"l: Magnani etal. fl993"l: Combaetal. f!992a"l: Combaetal. f!992b"l: Luce et al.
fl992"l: Zheng etal. fl992"l: Vaughan and Davis f!9911: Bolm-Audorff et al. fl990"l: Vaughan fl989"l: Haves et
al. (1986b): Hayes etal. (1986a): Merler etal. (1986): Vaughan et al. (1986a. 1986b): Hardell etal. (1982):
Mack and Preston-Martin (Unpub. Data presented in Luce etal. (2002)): Brinton etal. (1985): Brinton et al.
Q984I
22For Andjelkovich et al. (1995). assuming a rate of SNC for U.S. workers of 0.6 per 100,000 person-years
(Curado etal.. 2007). the expected number of cases would have been 0.33 and the ~SMR = 0 (95% CI 0, 5.99).
This document is a draft for review purposes only and does not constitute Agency policy.
1-232 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
results that did not report some increase in risk associated with formaldehyde exposure had been
in the group classified with low confidence, in part due to their lack of sensitivity to detect a true
effect fCoggon et al.. 2014: Siew etal.. 2012: Pesch etal.. 20081.
As discussed in a following section on the potential for confounding, wood dust is a very
strong risk factor for sinonasal cancer and because coexposure to wood dust may also be correlated
with formaldehyde exposures (e.g., in carpentry and other woodworking occupations), wood dust
could have been a potent confounder that might have caused the reported effects of formaldehyde
to appear inflated due to positive confounding. However, the evaluation of studies in
Appendix A.5.9 screened each set of results for potential confounding by wood dust and retained
only those results that either controlled for coexposures to wood dust using statistical adjustment
in regression analyses or by restricting analyses to workers without coexposure to wood dusts
fBeane Freeman etal.. 2013: Luce etal.. 2002: Teschke etal.. 1997: Hansen and Olsen. 1995: Roush
etal.. 1987b: Olsen and Asnaes. 1986). or those results from studies that were unlikely to have had
occupational coexposure to wood dusts (Coggon et al.. 2014: Siew etal.. 2012: Teschke etal.. 1997).
As can be seen in Table 1-33, and in Figure 1-21, which shows the medium confidence
studies, associations were stronger for adenocarcinomas than for squamous cell carcinomas.
However, both histological cell type groupings, and a mixed-type group, yielded results which were
consistently elevated—with a clear demonstration of statistical significance for the
adenocarcinomas.
In summary, the majority of these studies of different populations, in different locations,
exposure settings, and using different study designs reported increased risks of sinonasal cancer
associated with formaldehyde exposure that was unlikely to have been confounded by coexposure
to wood dust
Strength of the observed association
While reported relative effect estimates were largely elevated above the null value of unity
(1.0) across the sets of results that detected cases of sinonasal cancer, the magnitude of the relative
effect estimates varied with the quality of the exposure assessment and stratification by histological
cell type. The adenocarcinoma results classified with medium confidence reported three-fold (and
higher) increased risks of sinonasal cancer that appeared to be associated with higher exposure to
formaldehyde after controlling for wood dust (Luce etal.. 2002: Hansen and Olsen. 1995: Olsen and
Asnaes. 19861. Olsen and Asnaes (1986) reported results among men for adenocarcinoma adjusted
for wood dust and among those never exposed to wood dust: for ever vs never exposed to
formaldehyde, the RR adjusted for ever being exposed to wood dust was 2.2 (95% CI 0.7, 7.2; 17
exposed cases) while the RR for formaldehyde among men never exposed to wood dust was 7.0
(95% CI: 1.1, 43.9; one exposed case after excluded men ever exposed to wood dust). Further
restricting formaldehyde exposures to those first exposed more than 10 years prior to cancer
incidence, the RR was 9.5 (95% CI 1.6, 57.8; one exposed case). Luce et al. (2002) reported
increased risks for men with the highest cumulative formaldehyde exposure adjusted for wood
This document is a draft for review purposes only and does not constitute Agency policy.
1-233 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
dusts (OR = 3.0; 95% CI 1.5, 5.7; 91 cases) and for women (OR = 5.8; 95% CI 1.7,19.4; five cases).
Hansen and Olsen (1995), a low confidence study, reported that for formaldehyde exposures more
than 10 years prior to cancer incidence, the Standardized Proportional Incidence Ratio was 3.0
(95% CI 1.4, 5.7; nine cases). One adenocarcinoma study that was classified with low confidence
and was not able to report results by level of formaldehyde exposure, found a decreased risk of
sinonasal cancer among woodworkers ever exposed to formaldehyde [Pesch etal. (2008):
OR = 0.46; 95% CI 0.14,1.54], Pesch et al. (2008) was the only case-control study of sinonasal
cancer that relied on prevalent cases and included cases accrued over a 10-year period. Since the
controls in Pesch et al. (2008) were accident victims who were frequency matched on age
(<60 vs. 60+ years), it is possible that the prevalent cases available at the time of the study could
have been selected for survival, which may have resulted in a downward bias and may explain the
inverse findings for this study.
The squamous cell carcinoma study results classified with medium confidence reported
1.5-to 2-fold increased risks of sinonasal cancer that appeared to be associated with higher
exposure to formaldehyde after controlling for wood dust (Luce etal.. 2002: Olsen and Asnaes.
19861. although one study result classified with low confidence found no association between
sinonasal cancer in the 5% of cases "ever" exposed to formaldehyde fSiewetal. f20121: OR = 0.97;
95% CI 0.47, 2.00).
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants started
prior to their diagnoses of sinonasal cancer. Three studies provided analyses of the temporal
relationship showing some evidence of the effect of TSFE on the risk of dying from sinonasal cancer
(Luce etal.. 2002: Roush etal.. 1987b: Olsen and Asnaes. 1986). Lagging formaldehyde exposures
by 10 or 20 years to account for cancer latency increased the observed effects only slightly for
adenocarcinoma results (Luce etal.. 2002: Olsen and Asnaes. 1986) and for mixed cell type cancers
(Roush etal.. 1987b): but not for squamous cell carcinomas (Olsen and Asnaes. 1986). It is notable
that for nasopharyngeal cancer in the tissue adjacent to the sinonasal tissues, the effect of latency
on the temporal relationship between formaldehyde exposure and cancer mortality was generally
longer than 25 years. Only one study of sinonasal cancer examined a lag of 20 years fLuce etal..
2002). and none examined the effect of an even longer latency. If the effect of exposure on the
occurrence of sinonasal cancer took longer than the 20 years, then differences in results between
lagged and unlagged exposure analyses would be consistent with the available epidemiological
data.
Exposure-response relationship
Exposure-response relationships were not typically examined in these studies, most likely
due to the rarity of cases in all of the studies except in the large, pooled study of information from
12 publications fLuce et al.. 20021: see Table 1-33 for details). No results showing associations with
This document is a draft for review purposes only and does not constitute Agency policy.
1-234 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
duration of exposure were reported, but Luce et al. (20021 did state that even though their studies
reported primarily on cumulative exposure, "All exposure variables (probability, maximum level,
and duration) were associated with the risk of adenocarcinoma." The majority of studies reported
only comparisons of exposed versus unexposed subjects. Hansen and Olsen (1995) did report an
increase in risk among formaldehyde-exposed blue-collar worker (OR = 3.0; 95% CI 1.4, 5.7)
compared to exposed white-collar workers whose likely formaldehyde exposures were considered
to have been lower (OR = 0.8; 95% CI 0.02, 4.4). Luce et al. (2002) pooled 196 cases of sinonasal
adenocarcinoma and 432 cases of squamous cell carcinoma and was able to contrast risks in three
levels of exposure probability with the risk in the unexposed. An exposure-response relationship
for adenocarcinoma, controlling for coexposure to wood dust, was observed for both men and
women (see Table 1-33) with the highest risks among those with the highest probability of
exposure. The OR among men with the highest cumulative exposure was 3.0 (95% CI 1.5, 5.7),
while it was 5.8 (95% CI 1.7,19.4) among women. Among men with adenocarcinoma, the odds
ratios adjusted for wood dust increased from OR = 0.7 (95% CI: 0.3,1.9; six cases) among those
with 'low' cumulative exposure, to OR = 2.4 (95% CI: 1.3, 4.5; 31 cases) among those with 'medium'
cumulative exposure, to OR = 3.0 (95% CI: 1.5, 5.7; 91 cases) among those with 'high' cumulative
exposure.
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias is an unlikely bias in the epidemiological studies of sinonasal cancer as the
case-control studies evaluated exposure status without regard to outcome status and most had
participation levels of 85-100%, although one case-control study of prevalent cases accrued over
long periods of time had lower participation levels (67% in Pesch et al. (2008)). The cohort study
(Hansen and Olsen. 1995) included 72% of eligible participants. Selection biases could obscure a
truly larger effect of formaldehyde exposure in analyses based on "external" comparisons with
mortality in the general population (Hansen and Olsen. 1995). but would not influence analyses
using "internal" or matched comparison groups (Pesch etal.. 2008: Luce etal.. 2002: Roush etal..
1987b: Olsen and Asnaes. 1986). Information bias from the use of indirect exposure measures is
unlikely to have resulted in bias away from the null, however random measurement error or
nondifferential misclassification is almost certain to have resulted in some bias toward the null
among these studies of sinonasal cancer.
Confounding is a potential bias that could arise if another cause of sinonasal cancer were
also associated with formaldehyde exposure. Chemicals and other coexposures that have not been
independently associated with sinonasal cancer are not expected to confound results. Other known
risk factors for sinonasal cancer include wood dust (Hansen and Olsen. 1995: Olsen and Asnaes.
1986). smoking, and alcohol consumption (Vaughan. 1989). While smoking and alcohol may be
independent risk factors for sinonasal cancer they are unlikely to be related to formaldehyde
exposure and therefore unlikely to be across-the-board confounders. Wood dust, however, is a
potential confounder as many wood-related jobs also have exposures to formaldehyde and the
This document is a draft for review purposes only and does not constitute Agency policy.
1-235 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
association between wood dust exposure and sinonasal cancer is extremely strong, with relative
risks greater than 30-fold fOlsen andAsnaes. 19861.
Wood dust may be an independent risk factor for sinonasal cancer; however, the majority of
investigators presented analytic results for formaldehyde among workers who were either not
exposed to wood dusts fHansen and Olsen. 1995: Olsen and Asnaes. 19861. or else controlled for
the potential confounding of the effects of wood dust on the risk of sinonasal cancer and did not
find wood dust to be a confounder (Luce etal.. 20021.
Consistency across multiple studies is demonstrated by a pattern of increased risk in
different populations, exposure scenarios, and time periods. Such consistency makes unmeasured
confounding an unlikely alternative explanation for the observed associations. This consistency
also reduces the likelihood of chance as an alternative explanation by increasing confidence in the
statistical strength of the findings through the accumulation of a larger body of similar evidence.
The observations of multiple instances of very strong associations in different settings reduce the
likelihood that chance, confounding, or other biases can explain the observed associations.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
sinonasal cancer placed the greatest weight on four particular considerations: (1) the consistency of
the elevated risk across studies (particularly for adenocarcinoma)—including four sets of results
classified with medium confidence—one of which represents a large pooled analysis of 12
case-control studies with considerably more cases and with greater detail on formaldehyde
exposures; (2) the strength of the association with two results classified with medium confidence
reporting at least a three-fold increase in risk for adenocarcinoma with lower associations for
squamous cell carcinoma; (3) the exposure-response relationship in a large pooled analysis of 12
case-control studies showing increased exposure to formaldehyde was associated with increased
risk of sinonasal cancer among people with little, or no exposure to wood dust or in analyses that
controlled for wood dust; (4) reasonable confidence that alternative explanations have been
addressed, including chance, bias, and confounding within individual studies or across studies,
although many of the analyses lacked precision due to the rarity of sinonasal cancer. Consistent
observations of genotoxicity in exfoliated buccal cells or nasal mucosal cells across several
occupational studies involving diverse exposure settings further supports the evidence for
sinonasal carcinogenicity in humans.
This evidence was judged to be near the borderline of robust evidence and moderate
evidence, but one additional consideration increased confidence that the evidence was robust. The
large, pooled analysis using a case-control study design especially suited to identify associations for
this extremely rare cancer (Luce etal.. 20021 was considered to be especially informative in
identifying the effects of formaldehyde on the risks of sinonasal cancer and provided clear evidence
of an association of increased risks of sinonasal cancer with formaldehyde exposure - especially for
adenocarcinoma.
This document is a draft for review purposes only and does not constitute Agency policy.
1-236 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 Conclusion
2 The available epidemiological studies provide robust evidence of an association consistent
3 with causation between formaldehyde exposure and increased risk of sinonasal cancer.
0)
•*->
re
E
-H
(/)
LU
o
a>
a>
a:
100-1
10-
3-
2-
1-
£ 0.1-
f
.¦&
s
§.
Beane Freeman
Luce
Luce
Squamous Cell Carcinomas
I I
Mixed cell type Carcinomas
Adenocarcinomas
—|—
c0>
v»°
^ ^ ^
c0>
Figure 1-21. Highest (medium) confidence epidemiological studies reporting
sinonasal cancer risk estimates.
Results are grouped by histological type as squamous cell carcinomas, mixed cell types, or
adenocarcinoma. SMR: standardized mortality ratio. RR: relative risk. OR: odds ratio. TSFE: time since
first exposure. For each measure of association, the number of exposed cases is provided in brackets
(e.g., [n = 4]). For studies reporting results on multiple metrics of exposure, only the highest category of
each exposure metric is presented in the figure. Note that two studies (Luce et al., 2002; Olsen and
Asnaes, 1986) reported separate results for squamous cell carcinoma and adenocarcinoma and appear
This document is a draft for review purposes only and does not constitute Agency policy.
1-237 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
twice in the figure. Also note that the pooled analysis by Luce et al. (2002) includes data from 12
publications and thus represents substantially more information than a single set of results (see
Table 1-33 for details).
This document is a draft for review purposes only and does not constitute Agency policy.
1-238 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-33. Epidemiological studies of formaldehyde exposure and risk of
sinonasal cancers
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Beane Freeman et
al. (2013)
Population: 25,619 workers employed
at 10 formaldehyde-using or
formaldehyde-producing plants in the
United States followed from either the
plant start-up or first employment
through 2004. Deaths were identified
from the National Death Index with
remainder assumed to be living. 676
workers (3%) were lost to follow-up.
Vital status was 97.4% complete and
only 2.6% lost to follow-up.
Outcome definition: Death certificates
used to determine underlying cause of
death from nasal cancer (ICD-8: 160).
Histological typing not reported.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in
lowest exposed category. Lagged
exposures were evaluated to account
for cancer latency. Results were
presented for 15-year lag.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Blair et al. (1986)
Hauptmann et al. (2004)
Marsh et al. (2007a)
Beane Freeman et al. (2009)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM • (No appreciable bias)
IB: Exposure Group A
Exposure assessment: Individual-level
exposure estimates based on job
titles, tasks, visits to plants by study
industrial hygienists who took 2,000
air samples from representative job,
and monitoring data from 1960
through 1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3). Median cumulative
exposure = 0.6 ppm-years (range
0-107.4).
Multiple exposure metrics including
peak, average, and cumulative
exposures were evaluated using
categorical and continuous data.
Duration and timing: Exposure period
from <1946 to 1980. Median length
of follow-up: 42 years. Median length
of employment was 2.6 years (range
1 day-47.7 years). Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Peak exposure:
Level 1 (>0 to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Level 1 (>0 to <0.5 ppm)
Level 2 (0.5 to <1.0 ppm)
Level 3 (>1.0 ppm)
Cumulative exposure:
Level 1 (>0 to <1.5 ppm-yrs)
Level 2 (1.5 to <5.5 ppm-yrs)
Level 3 (>5.5 ppm-yrs)
Duration of exposure:
Level 1 (0 years)
Level 2 (>0 to <5 years)
Level 3 (5 to <15 years)
Level 4 (>15 years)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders
and found not be confounders.
fAs noted in Appendix A.5.9: There
was no information on smoking,
Internal comparisons:
Peak exposure
Unexposed RR = 5.67 (0.41-78.89) [2]
Level 1 RR = 1.00 (Ref. value) [1]
Level 2 RR = 1.53 (0.09-24.68) [1]
Level 3 RR = 1.29 (0.08-21.23) [1]
p-trend (exposed) > 0.5;
p-trend (all) = 0.37
Average intensity
Unexposed RR = 4.31 (0.48-38.67) [2]
Level 1 RR = 1.00 (Ref. value) [2]
Level 2 RR = 1.47 (0.13-16.50) [1]
Level 3 RR = N/A [0]
p-trend (exposed) > 0.50;
p-trend (all) = 0.23
Cumulative exposure
Unexposed RR = 3.90 (0.41-37.06) [2]
Level 1 RR = 1.00 (Ref. value) [2]
Level 2 RR = 1.22 (0.11-14.11) [1]
Level 3 RR = N/A [0]
p-trend (exposed) > 0.50;
p-trend (all) = 0.28
External comparisons:
SMRunexposed = 1.93 (0.23-6.98) [2]
SMRExposed =0.90(0.18-2.62) [3]
This document is a draft for review purposes only and does not constitute Agency policy.
1-239 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Oth: Low power due to rarity of cases.
however, according to Blair et al.
(1986), "The lack of a consistent
elevation for tobacco-related causes
of death, however, suggests that the
smoking habits among this cohort did
not differ substantially from those of
the general population."]
Reference: Luce et al. (2002)
Population: Males and females from
seven different countries diagnosed
with sinonasal cancer during
1968-1990.
Outcome definition: Diagnoses
originally assessed in 12 studies. 195
cases were adenocarcinomas (169
men and 26 women) and 432 were
squamous cell carcinomas (330 men
and 102 women).
Design: Pooled analysis of 12
case-control studies that included 627
total cases of sinonasal cancer and
3,136 controls (2,349 men and 787
women).
Analysis: ORs calculated by
unconditional logistic regression.
Adenocarcinoma results in men
adjusted for age, study, and
cumulative exposure to wood and
leather dust. All other results adjusted
for age and study.
Related studies:
Zheng et al. (1992)
Luce et al. (1992)
Luce et al. (1993)
Leclerc et al. (1994)
Bolm-Audorff et al. (1990)
Comba et al. (1992a); Comba
et al. (1992b)
Magnani et al. (1993)
Merler et al. (1986)
Haves et al. (1986b)
Hayes et al. (1986b); Hayes
et al. (1986a)
Hardell et al. (1982)
Exposure assessment: Detailed
occupational history information
gathered from interview
questionnaires provided the basis for
developing an individual's index of
exposure to formaldehyde. Standard
occupational classification codes and
standard industrial classification codes
were used to develop a job-exposure
matrix in conjunction with available
industrial hygiene data. With the
given occupational history information
of the subjects and the job-exposure
matrix, a semiquantitative index of
cumulative exposure was determined
for each individual calculated as the
sum of the job-specific products of
probability, level, and duration of
exposure over the total work history.
Subjects fell into one of four
categories of probable exposure
(unexposed, low exposure, medium
exposure, or high exposure) based
upon the job-exposure matrix.
Duration and timing: Latency was
evaluated with 10 and 20-year lags in
exposure with somewhat higher
effects. Results here are without
lagged exposures.
Variation in exposure:
Cumulative exposure:
Level 1 (unexposed)
Level 2 (low)
Level 3 (medium)
Level 4 (high)
Coexposures: Exposures to other
compounds were identified and
evaluated as potential confounders.
Other occupational exposures
potentially affecting the risk estimates
were controlled for including wood
dust, leather dust, textile dust, flour
dust, coal dust, crystalline silica,
Internal comparisons:
Adenocarcinoma
Men (Adjusted for wood dust)
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 2 OR = 0.7 (0.3-1.9) [6]
Level 3 OR = 2.4 (1.3-4.5) [31]
Level 4 OR = 3.0 (1.5-5.7) [91]
Women (Not adjusted for wood dust)
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 2 OR = 0.9 (0.2-4.1) [2]
Level 3 no cases
Level 4 OR = 6.2 (2.0-19.7) [5]
Women (Adjusted for wood dust)
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 4 OR = 5.8 (1.7-19.4) [5]
Squamous cell carcinoma
Men (Adjusted for wood dust)
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 2 OR = 1.2 (0.8-1.8) [43]
Level 3 OR = 1.1 (0.8-1.6) [40]
Level 4 OR = 1.2 (0.8-1.8) [30]
Women (Not adjusted for wood dust)
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 2 OR = 0.6 (0.2-1.4) [6]
Level 3 OR = 1.3 (0.6-3.2) [7]
Level 4 OR = 1.5 (0.6-3.8) [6]
Additional:
Authors reported that as an additional check
for potential residual confounding, the
formaldehyde adenocarcinoma results for
men were further adjusted for wood dust
and that the results were not markedly
changed.
Among women the result for high probability
of formaldehyde exposure was slightly
diminished (OR = 5.8; 95% CI: 1.7-19.4).
This document is a draft for review purposes only and does not constitute Agency policy.
1-240 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Vaughan et al. (1986a,
1986b)
Vaughan and Davis (1991)
asbestos, and man-made vitreous
fibers.
Vaughan (1989)
Mack and Preston-Martin (unpub.
data)
Brinton et al. (1985); Brinton
et al. (1984)
Confidence in effect estimates:3
SB
IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward the
null)
IB: Exposure Group C
Reference: Roush et al. (1987b)
Population: Males identified from the
Connecticut Tumor Registry who died
of any cause during 1935-1975.
Outcome definition: Diagnosis of
sinonasal cancer based on case
registration by the Connecticut Tumor
Registry. Clinical records reviewed for
>75% of cases. Histological typing not
reported.
Design: Population-based case-control
study of 198 male cases of sinonasal
cancer. Controls were 605 males
dying in Connecticut during the same
time period, randomly selected from
state death certificates.
Analysis: ORs calculated by logistic
regression and adjusted for age at
death, year at death, and availability
of occupational information.
Exposure assessment: Occupational
history obtained by city directories
and death certificates, which yielded
information on job, industry,
employer, and year of employment.
Exposure classification scheme based
on potential for formaldehyde
exposure, probability of exposure for
each participant and each job-industry
pair, and level of exposure.
Probability of exposure defined as
unexposed, possibly exposed,
probably exposed, or definitely
exposed.
Level of exposure estimated as zero,
low (<1 ppm), and high (>1 ppm).
Among those probably exposed to
some level of formaldehyde for most
of their working lifetime, the extent
and level of exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure level and timing of exposure:
Level 1 (unexposed)
Level 2 (probably exposed most of
working life)
Internal comparisons:
Exposure level and timing of exposure:
Level 1 OR = 1.0 (Ref. value) [# not given]
Level 2 OR = 0.8 (0.5-1.8) [21]
Level 3 OR = 1.0 (0.5-1.8) [16]
[9]
[7]
Level
1
OR = 1.0
Level
2
OR = 1.0
Level
3
OR = 1.5
This document is a draft for review purposes only and does not constitute Agency policy.
1-241 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Confidence in effect estimates:3
SB IB Of Oth
Overall
Confidence
Medium
MEDIUM ¦i, (Potential bias toward the
null)
IB: Exposure Group C
Level 3 (probably exposed most of
working life and probably
exposed 20+ years before
death)
High exposure level and timing of
exposure:
Level 1 (unexposed)
Level 2 (probably exposed most of
working life and probably
to high level in some
year)
Level 3 (probably exposed most of
working life and probably
exposed to high level
20+ years before death)
Coexposures: Not evaluated.
fAs noted in Appendix A.5.9: Exposure
to wood dust was not found to be a
risk factor for all nasal cancers
(NPC + SNC). This suggests a lower
potential for confounding by wood
dust.]
Reference: Olsen and Asnaes
(1986)
Population: Identified from the Danish
Cancer Registry between 1970 and
1982. Exposures to formaldehyde and
wood dust were identified too rarely
to allow for risk estimation.
Outcome definition: Diagnosis of
cancer of the nasal cavity (ICD-7 160.0)
or sinuses (ICD-7 160.2-160.9) was
histologically confirmed. Of all male
cases for cancer of the nasal cavity and
paranasal sinuses (n = 310), 69% were
squamous cell carcinoma and
lymphoepithelioma, 13% were
adenocarcinoma, 6% were sarcoma,
5% were malignant melanoma, and 7%
were of other histological type.
Design: Case-control study of 254 men
with sinonasal cavity and paranasal
cancers (215 with squamous cell
carcinoma/lymphoepithelioma and 39
with adenocarcinomas). 2,465
controls with other cancers matched
for gender, age, and year of diagnosis.
Exposure assessment: Employment
histories from 1964 maintained by
Danish Cancer Registry estimated by
industrial hygienists. Occupational
exposures estimated by industrial
hygienists based on industry or
occupations considered to have
certain or probably exposure. Authors
reported that 4.2% of control males
exposed to formaldehyde.
Multiple exposure metrics including
known exposure and duration since
first exposure were evaluated.
Duration and timing: Exposure period
starting at 1964.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (Unexposed)
Level 2 (Exposed)
Exposure to formaldehyde and wood
dust:
Level 1 (unexposed to either)
Level 2 (exposed to formaldehyde
and unexposed to wood
dust)
Internal comparisons:
Adenocarcinoma
Exposure to formaldehyde controlling for
wood dust:
Level 1 RR = 1.0 (Ref. value)
Level 2 RR = 2.2 (0.7-7.2)
[10]
[17]
Exposure to formaldehyde and wood dust:
Level 1 RR = 1.0 (Ref. value) [8]
Level 2 RR = 7.0 (1.1-43.9) [1]
Level 3 RR = 24.0 (7.6-75.6) [2]
Level 4 RR = 39.5 (22.0-70.8) [16]
>10 years since 1st exposure to formaldehyde
and wood dust:
1.0 (Ref. value) [6]
9.5 (1.6-57.8) [1]
36.8 (13.5-96.0) [3]
44.1 (22.2-87.8) [11]
Level 1 RR =
Level 2 RR =
Level 3 RR =
Level 4 RR =
Squamous cell carcinoma and
lymphoepithelioma
Exposure to formaldehyde controlling for
wood dust:
Level 1 RR = 1.0 (Ref. value)
Level 2 RR = 2.3 (0.9-5.8)
[113]
[13]
This document is a draft for review purposes only and does not constitute Agency policy.
1-242 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Analysis: The Mantel-Haenszel
summary estimates of the relative risk
were used to account for possible
confounding since the subjects were
stratified according to several
variables.
Related studies:
Olsen and Jensen (1984)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM ¦i, (Potential bias toward the
null)
IB: Exposure Group C
Level 3 (unexposed to
formaldehyde and
exposed to wood dust)
Level 4 (exposed to both)
>10 years since 1st exposure to
formaldehyde and wood dust:
Level 1 (unexposed to either)
Level 2 (exposed to formaldehyde
and unexposed to wood
dust)
Level 3 (unexposed to
formaldehyde and
exposed to wood dust)
Level 4 (exposed to both)
Coexposures: Exposure to wood dust
was identified and evaluated as a
potential confounder and as an effect
modifier.
Exposure to formaldehyde and wood dust:
Level 1 RR= 1.0 (Ref. value) [113]
Level 2 RR = 2.0 (0.7-5.9) [4]
Level 3 no cases
Level 4 RR = 1.6 (0.8-3.3) [9]
>10 years since 1st exposure to formaldehyde
and wood dust:
RR = 1.0 (Ref. value)
RR= 1.4 (0.3-6.4)
Level 1
Level 2
Level 3
Level 4
[81]
[2]
no cases
RR= 1.8 (0.7-4.4)
[6]
Reference: Coggon et al. (2014)
Population: 14,008 British men
employed in six chemical industry
factories which produced
formaldehyde. Cohort mortality
followed from 1941 through 2012.
Cause of deaths was known for 99% of
5,185 deaths through 2000. Similar
cause of death information not
provided on 7,378 deaths through
2012. Vital status was 98.9% complete
through 2003. Similar information not
provided on deaths through 2012.
Outcome definition: Death certificates
used to determine cause of deaths
from nasal cancer. Histological typing
not reported.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration
was evaluated as "more," or "less,"
than one year only among the 'High'
exposure group. Timing since first
exposure was not evaluated.
Variation in exposure:
Highest exposure level attained
Level 1 (Background)
Level 2 (low/moderate)
Level 3 (High)
Coexposures: Not evaluated.
Potential low-level exposure to
stvrene. ethylene oxide,
epichlorhydrin, solvents, asbestos,
chromium salts, and cadmium.
[As noted in Appendix A.5.9: Stvrene
is associated with LHP cancers but not
URT cancers.
Asbestos is associated with URT
External comparisons:
Overall:
SMR = 0.71 (0.09-2.55)
Exposed:
Level 1
Level 2
Level 3
SMR = 1.08 (0.03-6.01)
SMR = 1.01 (0.03-5.62)
SMR = 0(0-4.03)
[2]
[1]
[1]
[0]
cancers, but not this outcome.
Other coexposures are not known risk
factors for this outcome.]
This document is a draft for review purposes only and does not constitute Agency policy.
1-243 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
IB: Exposure is Group B; lack of latency
analysis.
Oth: Low power due to rarity of cases.
Reference: Meyers et al. (2013)
Population: 11,043 workers in 3 U.S.
garment plants exposed for at least
3 months. Women comprised 82% of
the cohort. Vital status was followed
through 2008 with 99.7% completion
Outcome definition: Death certificates
used to determine both the underlying
cause of death from nasal cancer
(ICD-code in use at time of death).
Histological typing not provided.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates.
Related studies:
Pinkerton et al. (2004)
Stayner et al. (1985)
Stayner et al. (1988)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group A; lack of latency
analysis.
Oth: Low power due to rarity of cases.
Exposures
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
1984 with 12-73 within each
department. Formaldehyde levels
across all departments and facilities
were similar. Geometric TWA8
exposures ranged from 0.09 to
0.20 ppm. Overall geometric mean
concentration of formaldehyde was
0.15 ppm, (GSD 1.90 ppm). Area
measures showed constant levels
without peaks. Historically earlier
exposures may have been
substantially higher.
Duration and timing: Exposure period
from 1955 to 1983. Median duration
of exposure was 3.3 years. More than
40% exposures <1963. Median time
since first exposure was 39.4 years.
Duration and timing since first
exposure were not evaluated for this
cancer.
Variation in exposure: Not evaluated.
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
[As noted in Appendix A.5.9: There
was no information on smoking in this
analysis, however, according to
Leclerc et al. (1997), "the
overall prevalence of cigarette
smokers was ... similar to those
reported in a 1980 survey of adult
Americans, in which 29.2% of females
and 38.3% of males over the age of 20
were current cigarette smokers."
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
SMR = 0(0-3.89) [0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-244 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Therefore, confounding was
considered to be unlikely.
Reference: Siew et al. (2012)
Population: All Finnish men born
during 1906-1945 who participated in
census and were employed in 1970
(n = 1.2 million). Vital status was
"virtually complete."
Outcome definition: Diagnosis of nasal
squamous cell cancer reported to the
Finnish Cancer Registry.
Design: Prospective national cohort
incidence study with internal
comparison groups.
Analysis: RRs calculated controlling for
sex, age, socioeconomic status, period
of follow-up, and smoking.
Confidence in effect estimates:3
Exposure assessment: Individual-level
exposure estimates based on
matching occupations listed in the
census to the Finnish job-exposure
matrix which covers major
occupational exposures and provided
exposure estimates for formaldehyde.
Duration and timing: Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (none)
Level 2 (any)
Coexposures: Wood dust exposures
were controlled for in formaldehyde
analyses.
Internal comparisons:
Exposure to formaldehyde:
Level 1 RR = 1.00 (Ref. value)
Level 2 RR = 0.97 (0.47-2.00)
[158]
[9]
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure Group D
Oth: Low power due to rarity of
exposure.
Reference: Pesch et al. (2008)
Population: Male workers insured by a
liability insurance association for the
German wood-working industries with
an occupational disease during
1994-2003. Of 129 cases of sinonasal
adenocarcinoma identified, 86 cases
(67%) agreed to participate (including
29 next of kin). 204 controls (75%)
participated (including 69 next of kin).
Outcome definition: Cases were ever
employed in German wood industries
and diagnosed with histopathologically
confirmed sinonasal adenocarcinoma.
Design: Insurer-based case-control
study of 86 cases of sinonasal
adenocarcinoma. Controls were 204
Exposure assessment: Occupational
history information gathered from
structured questionnaires. Because
next of kin information on exposure to
wood additives was considered poor,
the probability of exposure to
formaldehyde was rated by an expert
team as none, low, medium, or high.
In Germany, legislation or new
formulations altered potential
formaldehyde exposure in 1985 (likely
lowering them). Final analyses
classified exposure as unexposed, any
probability of exposure before 1985,
or any probability of exposure in 1985
or afterwards.
Internal comparisons:
Exposure level:
Level 1
OR = 1.0 (Ref. value)
[39]
Level 2
OR = 0.46 (0.14-1.54)
[8]
Level 3
OR = 0.94 (0.47-1.9)
[39]
This document is a draft for review purposes only and does not constitute Agency policy.
1-245 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
workers with accidents between home
and work or falls during working shifts.
Controls were frequency matched on
age with 60 years as the stratification
point.
Analysis: ORs calculated using logistic
regressions controlling for age (<60 vs.
60+), region, interviewee, and average
wood dust exposure. All temporal
exposure variables were lagged by
5 years.
Confidence in effect estimates:3
Duration and timing: Duration of
formaldehyde exposure was not
evaluated.
Variation in exposure:
Exposure level:
Level 1 (unexposed)
Level 2 (any exposure <1985)
Level 3 (any exposure >1985)
Coexposures: Wood dust exposures
were controlled for in formaldehyde
analyses.
SB IB Cf Oth
Overall
Confidence
Low
1
h
LOW (Potential bias toward the
null)
IB: Exposure Group B; latency
evaluated only for 5 years.
SB: Potential selection issue due to use
of prevalent cases.
Reference: Jakobsson et al.
(1997)
Population: 727 male employees of
two plants producing stainless steel
sinks and saucepans employed at least
one year during 1927-1981 with
minimum 15-year follow-up.
Outcome definition: Incidence of
sinonasal cancer from the Swedish
Tumor Registry (ICD-7:160).
Design: Cohort incidence study with
external comparison group.
Analysis: SIRs calculated using sex,
age, and calendar-year-expected
number of cases from the national
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
Exposure assessment: Workers grind
stainless steel with grinding plates
made of formaldehyde resins which
may release formaldehyde when
heated during grinding operations.
Duration and timing: Occupational
exposure preceding death during
1927-1981. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Coexposures may have
included chromium, nickel, and
abrasive dusts including silicon
carbide, aluminum oxide, silicon
dioxide, and clay.
[As noted in Appendix A.5.9: Nickel
and chromium are associated with
URT cancers and would likely be
positively correlated with
formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.
External comparisons:
Observed: 0
Expected: 0.5
SIR = 0(0-8.0)
[0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-246 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure Group D
Cf: Potential confounding
Oth: Low power due to rarity of cases.
Other coexposures are not known risk
factors for these outcomes.
No mention of exposure to wood
dust.]
Reference: Teschke et al.
(1997)
Population: 48 incident cases of nasal
cancers (31% female) older than
19 years and registered by the British
Columbia Cancer Agency during
1990-1992. Controls were randomly
selected from age and sex strata of
voter lists of the same time period
(frequency matched).
6 of original 54 cases (11%) were
excluded for lack of interview as were
36 of 195 eligible controls (18%).
Outcome definition: Incidence of
sinonasal cancer from the British
Columbia Cancer Agency (ICD-0:160.0,
160.2,160.9). Histological types: 23
squamous cell carcinomas (48%),
seven melanomas, seven lymphomas,
two adenocarcinomas (4%), two
adenoid cystic carcinomas, and seven
other histologies with one case each.
Design: Population-based case-control
study of nasal cancer.
Analysis: ORs controlled for sex, age,
and smoking.
Confidence in effect estimates:3
SB IB a Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null;
Low sensitivity)
IB: Exposure Group C
Cf: Potential confounding by acid
mists.
Oth: Low power due to rarity of
exposure.
Exposure assessment: Detailed
occupational history information
gathered from interview
questionnaires.
57 Occupational groups assessed.
Investigators discussed that textile
workers, pulp and paper mill workers,
and chemical and biological laboratory
personnel may have formaldehyde
exposures.
Duration and timing: Duration of
exposure was not evaluated. Timing
of exposure was evaluated for nasal
cancer with results for 20-year latency
presented.
Variation in exposure:
Ever employed in occupational group:
Level 1 (never)
Level 2 (ever)
Coexposures: Not evaluated.
fAs noted in Appendix A.5.9: Potential
confounders for these outcomes
include chlorophenols. acid mists.
dioxin. and perchloroethvlene and
would likely be positively correlated
with formaldehyde exposure.
However, on acids mists are
associated with URT cancers.
Potential for confounding is unknown
but could have inflated the observed
effect.]
External comparisons:
All histological types:
Textile workers (all)
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 7.6 (1.4-56.6) [6]
Textile workers (most recent 20 years
removed)
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 5.0 (0.8-43.0) [4]
Pulp and paper mill workers (all)
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 3.1 (0.4-25.4) [3]
Pulp and paper mill workers (20-yr lag)
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 3.1 (0.4-25.4) [3]
Chemical and biological lab workers (all)
Level 1 OR = 1.0 (Ref. value) [8]
Level 2 OR = 0.7 (0.1-4.0) [2]
Chemical and biological lab workers (20-yr
lag)
Level 1 OR = 1.0 (Ref. value) [7]
Level 2 OR = 0.9 (0.1-5.3) [2]
Squamous cell carcinoma:
Textile workers (all)
Level 1 OR = 1.0 (Ref. value) [not given]
Level 2 OR = 5.3 (0.2-5.3) [not given]
This document is a draft for review purposes only and does not constitute Agency policy.
1-247 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Hansen and Olsen
(1995)
Population: 2,041 men with cancer
who were diagnosed during
1970-1984 and whose longest work
experience occurred at least 10 years
before cancer diagnosis. Identified
from the Danish Cancer Registry and
matched with the Danish
Supplementary Pension Fund.
Ascertainment considered complete.
Pension record available for 72% of
cancer cases.
Outcome definition: Nasal cavity
cancer (ICD-7:160) listed on Danish
Cancer Registry file. Of all male cases
(n = 13), histological types of nasal
cavity tumors included four squamous
cell carcinomas, three
adenocarcinomas, one adenoid cystic
carcinoma, one melanoma, and one
unknown type. Tumors of the
maxillary sinus included two
squamous cell carcinomas and one
anaplastic carcinoma. Overall, there
were six squamous cell carcinomas
(46%) and two adenocarcinomas
(15%).
Design: Proportionate incidence study
with external comparison group.
Analysis: Standardized proportionate
incidence ratio calculated as the
proportion of cases for a given cancer
in formaldehyde-associated
companies relative to the proportion
of cases for the same cancer among all
employees in Denmark. Adjusted for
age and calendar time.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
1
Exposure assessment: Individual
occupational histories including
industry and job title established
through company tax records to the
national Danish Product Register.
Subject were considered to be
exposed to formaldehyde if: (1) they
had worked in an industry known to
use more than 1 kg formaldehyde per
employee per year; and (2) subjects
longest single work experience (job) in
that industry since 1964 was >10 years
prior to cancer diagnosis.
All subjects were stratified based on
job title as either low exposure (white
collar worker), above background
exposure (blue collar worker), or
unknown (job title unavailable).
Duration and timing: Exposure period
not stated. Based on date of diagnosis
during 1970-1984, and the
requirement of exposure more than
10 years prior to diagnosis, the
approximate period was 1960-1974.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (unknown)
Level 2 (low formaldehyde
exposure)
Level 3 (formaldehyde exposure,
no wood dust)
Level 4 (formaldehyde and wood
dust exposure)
Coexposures: Exposure to wood dust
was evaluated as a potential
confounder of sinonasal cancer.
Authors excluded wood dust exposed
Cases from Level 3 analyses.
External comparisons:
Overall (exposure to formaldehyde >10 years
prior to cancer diagnosis)
SPIR = 2.3 (1.3-4.0) [13]
Exposure to formaldehyde:
Level 1 SPIR = 1.0 (0.03-6.1) [1]
Level 2 SPIR = 0.8 (0.02-4.4) [1]
Level 3 SPIR = 3.0 (1.4-5.7) [9]
Level 4 SPIR = 5.0 (0.5-13.4) [2]
LOW (Potential bias toward the
null)
IB: Exposure Group D
This document is a draft for review purposes only and does not constitute Agency policy.
1-248 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Hayes et al. (1990)
Population: 4,046 deceased U.S. male
embalmers and funeral directors,
derived from licensing boards and
funeral director associations in 32
states and the District of Columbia
who died during 1975-1985. Death
certificates obtained for 79% of
potential study subjects (n = 6,651)
with vital status unknown for 21%.
Outcome definition: Death certificates
and licensing boards used to
determine cause of death from
sinonasal cancer (ICD-8: 160).
Design: Proportionate mortality
cohort study with external comparison
group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
Overall
SB IB
Ct
irt-h
Confidence
1
Low
1
LOW ^(Potential bias toward the null;
Low sensitivity)
IB: Exposure Group A; latency not
evaluated.
Oth: Potential undercounting of cases.
Low power due to rarity of cases.
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative. Exposure based on
occupation which was confirmed on
death certificate. Authors
subsequently measured personal
embalming exposures ranging from
0.98 ppm (high ventilation) to
3.99 ppm (low ventilation) with peaks
up to 20 ppm.
Authors state that major exposures
are to formaldehyde and possibly
glutaraldehyde and phenol.
Duration and timing: Occupational
exposure preceding death during
1975-1985. Of 115 deaths from LHP
cancer, 66 (57%) were aged
60-74 years. Duration and timing
since first exposure were not
evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
External comparisons:
Observed: 0 cases
Expected: 1.7 cases
PMR = 0(0-1.76) +
Additional:
By Race
White
[0]
PMR = 0(0-2.00)+ [0]
Non-White PMR = 0 (0-14.98) + [0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
Reference: Bertazzi et al. (1986)
Population: 1,332 male workers ever
employed in the plant between 1959
and 1980. Deaths were identified
from vital statistics offices. Vital status
was 98.6% complete.
Outcome definition: Nasal cancer
listed as cause of death on death
certificates.
Exposure assessment: Individual-level
exposure estimates based on
occupational histories. Over the
whole cohort, approximately 28% of
person time was estimated to be
exposed to formaldehyde.
Duration and timing: Occupational
exposure preceding death during
1959-1980. Duration and timing since
External comparisons:
Observed: 0
Expected: 0.0327
SMR = 0 (0-91.61) t
[0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
This document is a draft for review purposes only and does not constitute Agency policy.
1-249 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
age, and calendar-year-expected
number of deaths from the local
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
IB: Exposure Group B
Oth: Low power due to rarity of cases.
Reference: Stroup et al. (1986)
Population: 2,239 white male
members of the American Association
of Anatomists from 1888 to 1969 who
died during 1925-1979. Death
certificates obtained for 91 with 9%
lost to follow-up.
Outcome definition: Cancer of the
nasal cavity and sinuses listed as cause
of death on death certificates.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; Low sensitivity)
IB: Exposure Group A; latency not
evaluated
Oth: Low power due to rarity of cases.
Exposures
first exposure were not evaluated for
nasal cancer.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9: Other
exposures included stvrene. xylene,
toluene, and methyl isobutyl ketone.
Styrene is associated with LHP cancers
but not URT cancers.
Other coexposures are not known risk
factors for this outcome.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure during 1925-1979. Median
birth year was 1912. By 1979, 33% of
anatomists had died. Duration and
timing since first exposure were not
evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
[Benzene is not associated with URT
cancer.]
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
SMR = 0(0-7.2) [0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-250 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Levine et al. (1984a)
Population: 1,477 male undertakers
first licensed during 1928-1977 with
mortality follow-up from 1950 to
1977.
Vital status was 96% complete with
cause of death available for 94%.
Outcome definition: Cancer of the
nasal cavity and sinuses listed as
underlying cause of death on death
certificates (ICD-8: 160).
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
age, and calendar-year-expected
number of deaths from the Canadian
population.
Confidence in effect estimates:3
SB
IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null;
Low sensitivity)
IB: Exposure Group A; latency not
evaluated.
SB: Healthy worker effect
Oth: Low power due to rarity of cases.
Reference: Walrath and
Fraumeni (1984)
Population: 1,007 deceased white
male embalmers from the California
Bureau of Funeral Directing and
Embalming who died during
1925-1980. Death certificates
obtained for all.
Outcome definition: Nasal cancer
listed as cause of death on death
certificates.
Design: Proportionate mortality
cohort study with external comparison
group.
Exposures
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure during 1928-1977. Duration
and timing since first exposure were
not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure preceding death during
1916-1978. Birth year ranged from
1847-1959. Median age of death was
62 years. Most deaths were among
embalmers with active licenses.
Duration and timing since first
exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
Results: effect estimate (95% CI)
[# of cases]
Observed: 0
Expected: 0.2
PMR = 0(0-14.98)+ [0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
PMR = 0(0-4.99)+ [0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
1979)
External comparisons:
Observed: 0
Expected: 0.6
This document is a draft for review purposes only and does not constitute Agency policy.
1-251 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IE Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
SB: Potential selection bias: due to
incomplete death certificate
ascertainment.
IB: Exposure Group A; latency not
evaluated.
Oth: Low power due to rarity of cases.
Reference: Walrath and
Fraumeni (1983)
Population: 1,132 deceased white
male embalmers licensed to practice
during 1902-1980 in New York who
died during 1925-1980 identified from
registration files. Death certificates
obtained for 75% of potential study
subjects (n = 1,678).
Outcome definition: Nasal cancer
listed as cause of death on death
certificates.
Design: Proportionate mortality
cohort study with external comparison
group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB a Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
SB: Potential selection bias: due to
incomplete death certificate
ascertainment.
Exposures
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing:
Occupational exposure preceding
death during 1902-1980. Median
year of birth was 1901. Median year
of initial license was 1931. Median
age at death was 1968. Expected
median duration of exposure was
37 years. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Results: effect estimate (95% CI)
[# of cases]
PMR = 0(0-5.99)+ [0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
External comparisons:
Observed: 0
Expected: 0.5
This document is a draft for review purposes only and does not constitute Agency policy.
1-252 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
IB: Exposure Group A; latency not
evaluated.
Oth: Low power due to rarity of cases.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Oropharyngeal /Hypopharyngeal cancer
Epidemiological evidence
Oropharyngeal and hypopharyngeal cancer is commonly reported across the
epidemiological literature based on the Seventh, Eighth, or Ninth Revision of the ICD code (ICD-
7/8/9: 146 and ICD-7/8/9: 148, respectively). Two studies reported specifically on
hypopharyngeal cancer risks fMarsh et al.. 2007b: Laforest et al.. 20001. and one study reported
specifically on oropharyngeal cancer risks (Marsh et al.. 2007b). The results from five other studies
(of three populations) allowed for grouping these two adjacent tissue sites for analyses to examine
the risks of pharyngeal cancers below the nasopharynx (Marsh etal.. 2002: Gustavsson et al.. 1998:
Vaughan. 1989: Vaughan etal.. 1986a. b).
Overall, evidence describing an association between formaldehyde exposure and the risk of
developing or dying from oropharyngeal/hypopharyngeal cancer was available from nine reports
on six distinct study populations—four reports on three cohort studies fCoggon etal.. 2014: Meyers
etal.. 2013: Marsh etal.. 2007b: Marsh etal.. 2002) and five reports on three case-control studies
(Laforest etal.. 2000: Gustavsson et al.. 1998: Vaughan. 1989: Vaughan et al.. 1986a. b). No studies
with data specific to these pharyngeal cancer sites were excluded. The outcome-specific
evaluations of confidence in the precise effect estimate of an association from each study are
provided in Appendix A.5.9). Details of the reported results of high, medium, and low confidence are
provided in the evidence table for oropharyngeal/hypopharyngeal cancer (see Table 1-34)
following the causal evaluation.
Consistency of the observed association
The nine papers describing six populations reported the risks of
oropharyngeal/hypopharyngeal cancer among study subjects who had a high likelihood of
formaldehyde exposure (e.g., based on occupational history). The study results presented in
Table 1-34 (by confidence level and publication date) detail all of the reported associations. Results
are plotted in Figure 1-22 with results grouped by cancer site as "Oropharyngeal only,"
"Undifferentiated oropharyngeal/hypopharyngeal," or "Hypopharyngeal only."
This document is a draft for review purposes only and does not constitute Agency policy.
1-253 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Based on results for overall SMRs for all workers (both exposed and unexposed) compared
to external referent populations in three cohort studies (all classified with medium confidence), the
effect estimates were generally elevated and ranged in magnitude between 1.1 and 2.01, but none
had sufficient statistical power to exclude the null. The effect estimate for oropharyngeal cancer
alone was 1.95 fMarsh et al.. 2007bl: 95% CI 0.63, 4.56); for the combination of oropharyngeal and
hypopharyngeal cancer, the effect estimates were 1.1 (Meyers et al.. 2 013): 95% CI 0.40, 2.39) and
1.29 (Coggon etal.. 2014): 95% CI 0.76, 2.05), respectively; and for hypopharyngeal cancer alone
the effect estimate was 2.01 (Marsh et al.. 2007b): 95% CI 0.87, 3.96). The only case-control study
results classified with medium confidence fLaforestetal.. 20001 reported effect estimates by the
probability of exposure with an OR = 1.35 for "Ever" exposure to formaldehyde associated with
hypopharyngeal cancer (95% CI 0.86, 2.14), but for cases with >50% probability of formaldehyde
exposure the OR was 3.78 (95% CI 1.50, 9.49). The results from the two case-control studies
classified with low confidence (Gustavssonetal.. 1998). and the three Vaughan reports (Vaughan.
1989: Vaughan et al.. 1986a. b) were largely surrounding the null.
Subgroup analyses provide some indication of increased risk when a latency period was
accounted for. Increased risks of oropharyngeal/hypopharyngeal cancer were also reported by
Marsh etal. (2002) among workers with at least 10 years of formaldehyde exposure (SMR = 2.48;
95% CI 0.63, 6.75)—especially for those with at least 10 years of exposures greater than 0.2 ppm
(SMR = 4.94; 95% CI 1.25,13.38). After excluding those with <10% probability of being exposed to
formaldehyde, Laforest et al. (2000) found that for those with at least 20 years of exposure, the OR
was 2.70 (95% CI 1.08, 6.73).
Overall, the findings were heterogeneous. Results from the two case-control studies
classified with low confidence Gustavsson et al. (1998) and the Vaughan papers fVaughan. 1989:
Vaughan et al.. 1986a. b) did not show increased risks, although Gustavsson et al. (1998) did not
assess differences by exposure concentration or duration. The Vaughan analyses fVaughan. 1989:
Vaughan et al.. 1986a. b) did examine differences in exposures but did not observe consistently
increased risks. As with the Gustavsson et al. (1998) study, the Meyers et al. (2013) cohort study
did not assess differences in exposure concentration or duration and found only a minimally
increased risk. Coggon et al. (2014) did report results for duration greater than 1 year but did not
observe consistently increased risks, and Vaughan et al. (1986b) did not observe an increased risk
of oropharyngeal/hypopharyngeal cancer for living more than 10 years in a mobile home (although
the corresponding OR for NPC was 5.5). Two other medium confidence results from Marsh et al.
(2002) and Laforest et al. (2000) did observe increased risks associated with >10 and >20 years of
exposure duration.
Strength of the observed association
Summary effect estimates (SMR or RR) ranged from 1.01 fGustavsson et al.. 19981 to
slightly more than a doubling of the relative effect estimates fMarsh etal.. 2007bl. Only one study
fMarsh et al.. 2007bl reported a summary effect estimate (for cancers of the oropharynx,
This document is a draft for review purposes only and does not constitute Agency policy.
1-254 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
hypopharynx and unspecified pharynx) that excluded the null (OR = 1.98; 95% CI 1.17, 3.15). The
magnitude of the relative effect estimates varied but did not appear to depend on the specific non-
nasopharyngeal cancer site. Marsh et al. (2002) provided specific SMRs for oropharyngeal (ICD-9:
146), hypopharyngeal (ICD-9: 148), and "pharyngeal cancer, unspecified" (ICD-9: 149), which were
very similar at 1.95, 2.01, and 2.11 respectively. Exposure level-specific estimated risks ranged
from 0.8 for the highest residential duration of exposure to particleboard (Vaughan et al.. 1986b)
up to 4.94 for workers exposed to concentrations of formaldehyde greater than 200 ppb for more
than 10 years.
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants started
before their diagnoses of oropharyngeal/hypopharyngeal cancer. Only one study fVaughan etal..
1986a) reported results for formaldehyde exposure lagged by 15 years to account for latency and
did not find higher risks. It is notable that for nasopharyngeal cancer in the tissue neighboring the
oropharynx, the latency between formaldehyde exposure and cancer mortality was generally
longer than 25 years (see Section 1.2.5 Nasopharyngeal cancer); thus, studies without similar
follow-up time and appropriately lagged exposure may be insufficiently sensitive.
Marsh et al. (2002) reported on the effect of time since first employment in a formaldehyde-
related occupation as a proxy for latency. Those data (see Table 1-34) indicate that the risk of
workers with 20-29 years at a chemical plant producing or using formaldehyde had an SMR = 1.50
(95% CI 0.48, 3.61), while workers with more than 30 years' tenure had a higher risk (SMR = 2.69;
95% CI 1.31, 4.94). Extended duration of exposure can also be a reasonable proxy for latency.
Compared to unexposed workers, Laforest et al. (2000) reported increasing risks with increasing
duration of exposure for all workers (regardless of their probability of exposure) reaching an
OR = 1.51 (95% CI 0.78, 2.92) for those with more than 20 years' exposure to formaldehyde with an
even more pronounced effect of extended duration among those workers with the higher
probabilities of exposure (OR = 2.70; 95% CI 1.08, 6.73).
Exposure-response relationship
Only three study populations were available for evaluating exposure-response relationships
between formaldehyde and increased risk of oropharyngeal/hypopharyngeal cancer. The paired
studies by Vaughan et al. (1986a, b) did not show evidence of an exposure-response relationship
with the same exposure metrics as they did for nasopharyngeal cancer. Conversely, Laforest et al.
(2000) reported a clear exposure-response trend for increasing probability of formaldehyde
exposure (p < 0.005) and for increasing duration of formaldehyde exposure among subjects with at
least 10% probability of exposure (p < 0.04), with some indication of a trend with increasing
cumulative exposure (p < 0.14). Marsh et al. (2002) also found higher risks at higher durations of
exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
1-255 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias is an unlikely bias in the epidemiological studies of
oropharyngeal/hypopharyngeal cancer as the cohort study followed by Marsh et al. (2007b; 20021
included 98% of eligible participants and lost relatively few participants over the course of
mortality follow-up, and the case-control study by Laforest et al. (2000) evaluated exposure status
without regard to outcome status and had participation levels of 80% for cases and 86% for
controls. Information bias is unlikely to have resulted in bias away from the null; however, random
measurement error or nondifferential misclassification is almost certain to have resulted in some
bias toward the null among these studies of oropharyngeal/hypopharyngeal cancer. For example,
regarding one particular analysis from Marsh et al. (2002.), the authors reported risks for exposure
greater than 700 ppb of formaldehyde that might have been useful for comparison with the risk for
exposure of greater than 200 ppb; however, by comparing risk above 700 ppb to risk among
"unexposed" workers (with exposures ranging from 0 to 699 ppb), information bias was likely
induced, which may have attenuated that risk and made the inclusion of this result unsuitable for
exposure-response evaluation.
Confounding is a potential bias that could arise if another cause of
oropharyngeal/hypopharyngeal cancer is also associated with formaldehyde exposure. There does
not appear to be any evidence of confounding that would provide an alternative explanation for the
observed association of formaldehyde exposure with increased risk of
oropharyngeal/hypopharyngeal cancer seen across these studies. Chemical and other coexposures
that have not been independently associated with oropharyngeal/hypopharyngeal cancer are not
expected to confound results. Other known risk factors for oropharyngeal/hypopharyngeal cancer
include smoking and alcohol consumption fVaughan. 19961. While these other exposures may be
independent risk factors for oropharyngeal/hypopharyngeal cancer, smoking and alcohol
consumption are unlikely to be generally related to occupational and residential formaldehyde
exposures and are therefore unlikely to be across-the-board confounders. This is especially true for
studies comparing risks within a cohort of workers who may be more similar to each other in
smoking status than they are compared to an external population. This is relevant to the NCI cohort
which Beane Freeman (2013) noted had a high prevalence of current or former smokers across all
levels of formaldehyde exposure (i.e., smoking prevalence was likely independent of formaldehyde
exposure and not a confounder). However, the Marsh reports on one plant in this cohort fMarsh et
al.. 2007b: Marsh etal.. 2002) compared the risk of those workers to an external population which
might have had lower prevalences of smoking allow for a greater potential for confounding by
smoking in those reports.
Overall, the findings were heterogeneous with no association observed in study results of
low confidence and a mix of positive associations and null findings in study results of medium
confidence. For oropharyngeal/hypopharyngeal cancer, the lack of consistency weakens the
etiologic conclusion. However, the observations of increased risks across multiple medium
This document is a draft for review purposes only and does not constitute Agency policy.
1-256 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
confidence results, as well as two identified exposure-response relationships with increased
duration of formaldehyde exposure is suggestive of an association.
Causal evaluation and conclusion
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
oropharyngeal/hypopharyngeal cancer placed the greatest weight on three particular
considerations: (1) the observations of increased risks in two medium confidence studies with the
ability to evaluate multiple metrics of formaldehyde exposure, but little other evidence of increases
in risk across one other medium and two low confidence results; (2) the variable strength of the
association across studies and metrics with several results near the null and two medium
confidence studies reporting three-fold to five-fold increases in risk among groups with the highest
exposure probability or duration; and (3) exposure-response relationships using multiple metrics
of exposure from one study showing that increased exposure to formaldehyde was associated with
increased risk of developing oropharyngeal/hypopharyngeal cancer. Although consistent
observations of genotoxicity in exfoliated buccal cells or nasal mucosal cells have been observed
across several occupational studies, these data were not interpreted as sufficient to further
strengthen the judgment on the human evidence of cancers of the oropharynx and, more indirectly,
the hypopharynx.
Conclusion
The available epidemiological studies provide slight evidence of an association between
formaldehyde exposure and increased risk of oropharyngeal/hypopharyngeal cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
1-257 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
a>
+-»
rc
E
<~-»
LU
o
o
it
LU
O
>
_TO
O
OH
100i
10 -
3
2
1
0.1
| Coggon"
Oro-
only
I IL
Undifferentiated Oro- and Hypopharynx
| Hypopharyx only|
in
§
§3
CO
lO
cn
CO
CO
CD
o
ll
c
to
CN
in
h-
CO
CN
>
CO
'5)
II
c
ii
l:
II
c
II
c
II
c:
CO
n
II
C
ii
c
CD
II
ll
II
ll
II
II
¦g
"5
c
IT)
c.
c
c
C
c
c
c
CO
CO
O)
Tit
c
CO
CO
o
OJ
1—1
fN
CO
T—
T—
CO
CO
CO
CO
CO
CO
(N
CN
c\i
¦^r
q
T—
00
II
O
T—'
O
o
T-'
CD
II
ll
II
II
ii
r—
ll
11
o
Cd
II
II
II
II
ll
II
II
a:
a:
CC
01
CH
II
II
CC
IT
ii
2
q:
a:
Cd
0£
CE
a:
ir
2
2
a:
a:
S
2
tr
o
o
O
o
O
O
o
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-34. Studies of formaldehyde exposure and risk of cancer of
oropharynx/hypopharynx
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Coggon et al. (2014)
Population: 14,008 British men
employed in six chemical industry
factories which produced
formaldehyde. Cohort mortality
followed from 1941 through 2012.
From Coggon et al. (2003), cause of
death was known for 99% of 5,185
deaths through 2000. Similar cause of
death information not provided on
7,378 deaths through 2012. Vital
status was 98.9% complete and only
1.1% lost to follow-up through 2003.
Similar information not provided on
deaths through 2012.
Outcome definition: Death
certificates used to determine cause
of deaths from pharyngeal cancer
minus deaths from nasopharyngeal
cancer.
Design: Cohort mortality study with
external comparison group with a
nested case-control study.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward
the null)
IB: Exposure is Group B; lack of
latency analysis.
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration
was evaluated as more, or less, than
one year only among the "High"
exposure group. Timing since first
exposure was not evaluated.
Variation in exposure:
Duration of "High" exposures
Level 1 (Background)
Level 2 (<1 year)
Level 3 (1 year or more)
Coexposures: Not evaluated. Potential
low-level exposure to stvrene.
ethylene oxide, epichlorhydrin,
solvents, asbestos, chromium salts,
and cadmium.
[As noted in Appendix A.5.9: Styrene is
associated with LHP cancers but not
URT cancers.
Asbestos is associated with URT
cancers, but not this outcome.
Other coexposures are not known risk
factors for this outcome.]
External comparisons:
For NPC (p.1,307):
I observed case with exposure
above background vs. 1.7
expected.
For all pharyngeal cancers (see Table 3 in
Coggon et al.):
17 cases observed in all subjects vs.
14.1 expected.
II cases with exposures above
background v. 9.2 expected.
Therefore, for OH PC:
10 observed cases with exposure
above background vs. 7.5
expected.
16 observed cases in all subjects vs.
12.4 expected.
SMRan subjects = 1.29 (0.76-2.05)+ [16]
SMRExposed = 1.33 (0.68-2.38)+ [10]
Internal comparisons:
Since the 1 NPC case had "low/Moderate
exposure," the all-pharyngeal-cancer results
in Table 6 in Coggon et al. (2014) for
"High exposure" are OHPC.
Duration of 'High' exposures
Level 1 OR = 1.00 (Ref. value) [10]
Level 1 OR = 0.63 (0.13-3.03) [3]
Level 2 OR = 0.81 (0.22-3.05) [6]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
Reference: Meyers et al. (2013)
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
External comparisons:
SMR = 1.1 (0.40-2.39)
[6]
This document is a draft for review purposes only and does not constitute Agency policy.
1-259 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Population: 11,043 workers in 3 U.S.
garment plants exposed for at least
3 months. Women comprised 82% of
the cohort. Vital status was followed
through 2008 with 99.7% completion.
Outcome definition: Death
certificates used to determine both
the underlying cause of death from
nasopharyngeal cancer (ICD code in
use at time of death). Histological
typing not provided.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates.
Related studies:
Pinkerton et al. (2004)
Stayner et al. (1985)
Stayner et al. (1988)
Exposures
1984. Geometric TWA8 exposures
ranged from 0.09-0.20 ppm. Overall
geometric mean concentration of
formaldehyde was 0.15 ppm, (GSD
1.90 ppm). Area measures showed
constant levels without peaks.
Historically earlier exposures may have
been substantially higher.
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
Results: effect estimate (95% CI)
[# of cases]
Duration and timing: Exposure period
from 1955 to 1983. Median duration
of exposure was 3.3 years. More than
40% exposures <1963. Median time
since first exposure was 39.4 years.
Duration and timing since first
exposure were not evaluated for this
cancer.
Variation in exposure: Not evaluated.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward
the null)
IB: Exposure Group A; latency not
evaluated.
Reference: Marsh et al.
(2007b): Marsh etal. (2002)
is described on the next pages.
Population: 7,328 workers employed
at formaldehyde-using plant in the
United States followed from 1945
through 2003. Vital status was
identified from the National Death
Index, private businesses, or state and
local agencies, and was 98% complete
and 1.4% lost to follow-up. Among
the deceased, the cause of death was
available for 95.2%.
Exposure assessment: Worker-specific
exposure from job-exposure matrix
based on available sporadic sampling
data from 1965 to 1987, job
descriptions, and verbal job
descriptions by plant personnel and
industrial hygienists.
Exposures ranked on a 7-point scale
with exposure range assigned to each
rank. 17% of jobs validated with
company monitoring data; remaining
83% based on professional judgment.
Assumed pre-1965 exposure levels
same as post-1965 levels.
External comparisons:
Oropharyngeal cancer
U.S. referent SMR = 1.95 (0.63-4.56) [5]
County referent SMR = 1.71 (0.56-4.00) [5]
Hypopharvneeal cancer
U.S. referent SMR = 2.01 (0.87-3.96) [3]
County referent SMR = 1.88 (0.81-3.70) [3]
Pharyngeal cancer excluding nasopharyngeal
U.S. referent SMR = 1.98 (1.17-3.15) [16]
County referent SMR = 1.71 (1.01-2.72) [16]
This document is a draft for review purposes only and does not constitute Agency policy.
1-260 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
Exposure assessment did not include
the same industrial hygiene sampling
conducted by Stewart et al. (1986)
used in the Beane Freeman (2013;
2009) analyses which included this
plant.
Study
This population was from one plant
from Beane Freeman et al. (2009).
Outcome definition: Death
certificates used to determine
underlying cause of death from
oropharyngeal/hypopharyngeal
cancer according to the ICD-9 codes
(146,148).
Design: Cohort mortality study with
external comparison groups.
Analysis: SMRs calculated by dividing
the number of observed deaths by the
number of expected deaths. Expected
deaths were the product of death rate
(at national, state, or local level) and
person-years accumulated by all the
members of the cohort. SMRs made
age, race, gender, and period specific
to reduce bias and to generate tabular
information by these variables.
Mortality was compared with death
rates in two Connecticut counties and
the United States. These results are
shown in Table 2 in Marsh et al.
(2007b).
Related studies:
Hauptmann et al. (2004)
Marsh et al. (2002; 1996; 1994)
Confidence in effect estimates:3
SB
IB
Cf
Oth
Overall
Confidence
Medium
MEDIUM
/ (Potential bias toward
the null; low sensitivity)
IB: Exposure Group B; lack of latency
analysis.
Oth: Low power due to rarity of cases.
Reference: Marsh et al. (2002)
Population: 7,328 workers employed
at formaldehyde-using plant in the
United States followed from 1945
through 1998. Vital status was
identified from the National Death
Index, private businesses, or state and
Exposure estimates generated by this
method were 10 times lower on
average than those estimated by the
NCI.
Multiple exposure metrics including,
known exposure, average intensity and
cumulative exposures were evaluated.
Results: effect estimate (95% CI)
[# of cases]
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure: None.
Coexposures: Coexposures previously
identified in Marsh et al. (1996)
included product and nonproduct
particulates and airborne pigments.
[As noted in Appendix A.5.9: Marsh et
al. (2002) attempted to evaluate
smoking but data were incomplete.
No other potential confounders were
evaluated.
Beane Freeman et al. (2013; 2009)
evaluated 11 potential confounders
among a set of 10 plants that included
this one and did not find any
confounding.]
Exposure assessment: Worker-specific
exposure from job-exposure matrix
based on available sporadic sampling
data from 1965 to 1987, job
descriptions, and verbal job
descriptions by plant personnel and
industrial hygienists. Exposures
ranked on a 7-point scale with
exposure range assigned to each rank.
External comparisons:
Oropharyngeal cancer
U.S. referent SMR = 2.17 (0.71-5.07) [5]
County referent SMR = 1.80 (0.58-4.19) [5]
Hypopharvngeal cancer
U.S. referent SMR = 2.25 (0.46-6.58) [3]
County referent SMR = 1.52 (0.31-4.43) [3]
This document is a draft for review purposes only and does not constitute Agency policy.
1-261 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
local agencies, and was 98.4%
complete and 1.6% lost to follow-up.
This population was from one plant
from Beane Freeman et al. (2009).
Outcome definition: Death
certificates used to determine
underlying cause of death from
oropharyngeal/hypopharyngeal
cancer according to the ICD-9 codes
(146,148).
Design: Cohort mortality study with
external comparison groups.
Analysis: SMRs calculated by dividing
the number of observed deaths by the
number of expected deaths. Expected
deaths were the product of death rate
(at national, state, or local level) and
person-years accumulated by all the
members of the cohort. SMRs made
age, race, sex, and period specific to
reduce bias and to generate tabular
information by these variables.
Mortality was compared with death
rates in two Connecticut counties and
the United States. These results are
shown in Table 2 in Marsh et al.
(2002).
Related studies:
Beane Freeman et al. (2013; 2009)
Marsh et al. (2007b; 1996; 1994)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM >1/ (Potential bias toward
the null; low sensitivity)
IB: Exposure Group B; lack of latency
analysis.
Oth: Low power due to rarity of cases.
Multiple exposure metrics including,
known exposure, average intensity and
cumulative exposures were evaluated.
Results: effect estimate (95% CI)
[# of cases]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure {from Table 3 in
Marsh et al. (2002)):
For all variations in exposure:
Level 1 (unexposed)
Exposure to formaldehyde:
Level 2 (exposed)
Duration of exposure to formaldehyde:
Level 2 (0 to <1 years)
Level 3 (1 to 9 years)
Level 4 (>10 years)
Cumulative exposure to formaldehyde:
Level 2 (0 to <0.004 ppm-yrs)
Level 3 (0.004 to 0.219 ppm-yrs)
Level 4 (>0.22 ppm-yrs)
Average intensity exposure:
Level 2 (0 to <0.03 ppm)
Level 3 (0.03 to 0.159 ppm)
Level 4 (>0.16 ppm)
Exposure to formaldehyde >0.2 ppm:
Level 2 (exposed)
Duration of exposure to >0.2 ppm:
Level 2 (0 to <1 years)
Level 3 (1 to 9 years)
Level 4 (>10 years)
Coexposures: Coexposures previously
identified in Marsh et al. (1996)
Exposures
17% of jobs validated with company
monitoring data; remaining 83% based
on professional judgment. Assumed
pre-1965 exposure levels same as
post-1965 levels.
Exposure assessment did not include
the same industrial hygiene sampling
conducted by Stewart et al. (1986)
used in the Beane Freeman (2013;
2009) analyses which included this
plant,
Exposure estimates generated by this
method were 10 times lower on
average than those estimated by the
NCI.
Pharyngeal cancer, unspecified
U.S. referent SMR = 2.11 (0.85-4.35) [7]
County referent SMR = 1.89 (0.76-3.89) [7]
Oropharvngeal/Hvpopharvngeal cancer
Exposure to formaldehyde:
Level 1 SMR = 1.24* (0.21-4.10)+
Level 2 SMR = 1.83* (1.02-3.05)+
Duration of formaldehyde exposure:
Level 1 SMR = 1.24* (0.21-4.10)+
Level 2 SMR = 1.75* (0.77-3.46)+
Level 3 SMR = 1.58* (0.40-4.32)+
Level 4 SMR = 2.48* (0.63-6.75)+
[2]
[13]
[2]
[7]
[3]
[3]
Cumulative exposure to formaldehyde:
Level 1 SMR = 1.24* (0.21-4.10)+ [2]
Level 2 SMR = 3.20* (1.17-7.10)+ [5]
Level 3 SMR = 1.28* (0.40-3.07)+ [4]
Level 4 SMR = 1.56* (0.50-3.77)+ [4]
Average intensity exposure:
Level 1 SMR = 1.24* (0.15-4.49)+
Level 2 SMR = 1.96* (0.72-4.33)+
Level 3 SMR = 1.91* (0.49-5.20)+
Level 4 SMR = 1.69* (0.62-3.74)+
Exposure to formaldehyde >0.2 ppm:
Level 1 SMR = 1.51* (0.21-4.10)+
Level 2 SMR = 2.01* (1.02-3.05)+
[2]
[5]
[3]
[5]
[6]
[9]
Duration of exposure to >0.2 ppm:
Level 1 SMR = 1.51* (0.21-4.10)+ [6]
SMR = 1.72* (0.47-4.16)+ [4]
SMR = 1.30* (0.22-4.29)+ [2]
SMR = 4.94* (1.25-13.38)+
Level 2
Level 3
Level 4
[3]
*Note: EPA derived SMRs for the
combination of oropharyngeal,
hypopharyngeal and unspecified pharyngeal
cancer by subtracting the number of
observed and expected nasopharyngeal
cancer from the same counts for all
pharyngeal cancers.
This document is a draft for review purposes only and does not constitute Agency policy.
1-262 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
included product and nonproduct
particulates and airborne pigments.
Reference: Marsh et al. (2002)
Exposure assessment: Worker-specific
exposure from job-exposure matrix
based on available sporadic sampling
data from 1965 to 1987, job
descriptions, and verbal job
descriptions by plant personnel and
industrial hygienists. Exposures
ranked on a 7-point scale with
exposure range assigned to each rank.
17% of jobs validated with company
monitoring data; remaining 83% based
on professional judgment. Assumed
pre-1965 exposure levels same as
post-1965 levels.
Exposure estimates generated by this
method were 10 times lower on
average than those estimated by the
NCI.
Multiple exposure metrics including,
known exposure, average intensity,
and cumulative exposures were
evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in work history (from Table 3
in Marsh et al. (2002)]:
For all variations in exposure:
Level 1 (unexposed)
Exposure to formaldehyde:
Level 2 (exposed)
Work history:
Level 1 (short-term workers:
<1 year)
Level 2 (long-term workers:
1+ year)
Year of hire:
Level 1 (1941-1946)
Level 2 (1947-1956)
Level 3 (1957+)
Duration of employment:
Level 1 (unexposed)
Level 2 (<1 year)
Level 3 (1+ years)
External comparisons:
Exposure to formaldehyde >0.7 ppm:
Level 1 SMR = 1.86* (1.01-3.16)+ [12]
Level 2 SMR = 1.46* (0.37-3.98)+ [3]
Duration of exposure to >0.7 ppm:
Level 1 SMR = 1.86* (1.01-3.16)+
Level 2 SMR = 1.49* (0.25-4.93)+
Level 3 SMR = 1.41* (0.07-6.95)+
Work history:
Level 1 SMR =
Level 2 SMR =
Year of hire:
Level 1 SMR =
[1]
Level 2 SMR =
Level 3 SMR =
2.82* (1.31-5.37)+
1.70* (0.74-3.37)+
[12]
[2]
[1]
[8]
[7]
0.46* (0.11-10.73)+
2.49* (1.35-4.23)+ [12]
1.14* (0.19-3.78)+ [2]
Duration of employment:
Level 1 SMR = 1.83* (0.85-3.47)+
Level 2 SMR = 1.77* (0.56-4.27)+
Level 3 SMR = 1.62* (0.41-4.41)+
Time since first employment:
Level 1 SMR = 0.82* (0.14-2.71)+
Level 2 SMR = 1.50* (0.48-3.61)+
Level 3 SMR = 2.69* (1.31-4.94)+
[8]
[4]
[3]
[2]
[4]
[9]
*Note: EPA derived SMRs for the
combination of oropharyngeal,
hypopharyngeal and, unspecified pharyngeal
cancer by subtracting the number of
observed and expected nasopharyngeal
cancer from the same counts for all
pharyngeal cancers.
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice.
1979)
This document is a draft for review purposes only and does not constitute Agency policy.
1-263 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Time since first employment:
Level 1 (<20 year)
Level 2 (20-29 years)
Level 3 (30+ years)
Reference: Laforest et al.
(2000)
Population: Males diagnosed with
primary hypopharyngeal squamous
cell cancers between January 1989
and May 1991 and identified through
15 French hospitals. Interviews
completed for 79.5% of eligible cases
and 86% of eligible controls.
Outcome definition: Diagnosis of
laryngeal and hypopharyngeal cancers
was histologically confirmed.
Design: Hospital-based case-control
study of 201 hypopharyngeal cancers.
296 hospital controls frequency
matched on age.
Analysis: ORs were calculated by
unconditional logistic regression and
adjusted for age, alcohol, and
smoking. Induction periods of 5,10,
and 15 years was also utilized to
account for latency in evaluating risk.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
Medium \|/ (Potential bias toward the
null)
IB: Exposure Group C
Exposure assessment: Occupational
history obtained by interview.
Exposure assessment based on job-
exposure matrix that included level
and probability of exposure, duration,
and cumulative exposure to
formaldehyde.
Multiple exposure metrics including
known exposure, probability of
exposure, and cumulative exposure
were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
All subjects
Exposure to formaldehyde:
Level 1 (never exposed)
Level 2 (ever exposed)
Probability of exposure:
Level 1 (never exposed)
Level 2 (<10%)
Level 3 (10 to 50%)
Level 4 (>50%)
Duration of exposure:
Level 1 (never exposed)
Level 2 (<7 years)
Level 3 (7 to 20 years)
Level 4 (>20 years)
Cumulative exposure:
Level 1 (never exposed)
Level 2 (low, <0.02)
Level 3 (medium, 0.02 to 0.09)
Level 4 (high, >0.09)
Subjects with a probability of exposure
>10%
Exposure to formaldehyde:
Level 1 (never exposed)
Level 2 (ever exposed)
Duration of exposure:
Level 1 (never exposed)
Level 2 (<7 years)
Level 3 (7 to 20 years)
Level 4 (>20 years)
Cumulative exposure:
Level 1 (never exposed)
Level 2 (low)
Internal comparisons:
All subjects
Exposure to formaldehyde:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 1.35 (0.86-2.14) [83]
Probability of exposure:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 1.08 (0.62-1.88) [42]
Level 3 OR = 1.01 (0.44-2.31) [15]
Level 4 OR = 3.78 (1.50-9.49) [26]
p-trend (all) <0.005
Cumulative exposure:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 1.03 (0.51-2.07) [23]
Level 3 OR = 1.57 (0.81-3.06) [32]
Level 4 OR = 1.51 (0.74-3.10) [28]
Duration of exposure:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 1.09 (0.50-2.38) [18]
Level 3 OR = 1.39 (0.74-2.62) [37]
Level 4 OR = 1.51 (0.78-2.92) [28]
Subjects with a probability of exposure >10%
Exposure to formaldehyde:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 1.74 (0.91-3.34)
Cumulative exposure:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 0.78 (0.11-5.45)
Level 3 OR = 1.77 (0.65-4.78)
Level 4 OR = 1.92 (0.86-4.32)
p-trend (all) < 0.14
Duration of exposure:
Level 1 OR = 1.00 (Ref. value)
[118]
Level 2 OR = 0.74 (0.20-2.68)
Level 3 OR = 1.65 (0.67-4.08)
Level 4 OR = 2.70 (1.08-6.73)
p-trend (all) <0.04
[41]
[3]
[13]
[25]
[6]
[19]
[16]
This document is a draft for review purposes only and does not constitute Agency policy.
1-264 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Level 3 (medium)
Level 4 (high)
Other exposures: asbestos, coal dust,
leather dust, wood dust, flour dust,
silica, and textile dust.
[As noted in Appendix A.5.9: Of these,
only coal dust significantly increased
the risk of hypopharyngeal cancer in
this study but coal dust and asbestos
were controlled for in the OHPC
analysis.]
Introduction of induction times as described
did not substantially change the results.
Reference: Gustavsson et al.
(1998)
Population: Males between the ages
of 40 and 79 years residing in Sweden
identified by hospitals reports or
regional cancer registries during
1988-1990. Interviews completed for
90% of cases and 85% of controls.
Outcome definition: Diagnosis of
cancer of the pharyngeal caner based
on ICD-9 codes 146 (oropharynx) and
148 (hypopharynx) but not including
code 147 (nasopharynx) on weekly
reports from departments of
otorhinolaryngology, oncology, and
surgery and from regional cancer
registries.
Design: Community-based,
case-control study of 138 cases of
squamous cell carcinoma of the
oropharynx/hypopharynx. 641
controls were randomly identified
from population registers and
frequency matched by region and age.
Analysis: RRs were calculated by
unconditional logistic regression and
adjusted for region, age, drinking, and
smoking.
Confidence in effect estimates:3
SB ! B
Cf
Oth
Overall
Confidence
Louv
Exposure assessment: Occupational
history obtained by interview and
yielded information on all jobs held
>1 year, starting and stopping times,
job title, tasks, and company. Histories
reviewed by industrial hygienist who
coded jobs based on intensity and
probability of exposure to 17
occupational factors.
Exposure assessments estimated
intensity on a 4-point scale and
probability of exposure as point
estimates. Cumulative exposure
calculated as the product of exposure
intensity, probability of exposure, and
duration of exposure, and by adding
contributions over entire work history.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (never)
Level 2 (ever)
Other exposures: polycyclic aromatic
hydrocarbons, asbestos, general dust,
wood dust, quartz, metal dust, oil mist,
welding fumes, manmade mineral
fibers, paper dust, textile dust,
hexavalent chromium, phenoxy acids,
nickel, acid mist, and leather dust.
[As noted in Appendix A.5.9: Of these,
only leather dust was a risk factor but
only five cases were exposed.]
Internal comparisons:
Exposure to formaldehyde:
Level 1 OR = 1.00 (Ref. value) [# not given]
Level 2 OR = 1.01 (0.49-2.07) [13]
This document is a draft for review purposes only and does not constitute Agency policy.
1-265 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure Group B
Latency not evaluated.
Oth: Low power due to rarity of
exposure.
Reference: Vaughan (1989)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System during
1980-1983. Participation for all cases
was 68.7 and 80.0% for controls.
Outcome definition: Diagnosis of
nasopharyngeal cancer based on
review of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates. Nonsquamous cell
cancers were excluded from the
study.
Design: Population-based, case-
control study of 183 cases with oro
pharyngeal/hypopharyngeal cancer.
552 controls were identified by
random digit dialing in same
geographic area.
Analysis: ORs were calculated by
logistic regression and adjusted for
gender, cigarette smoking, and
alcohol. Induction periods were
evaluated.
Related studies:
Vaughan et al. (1986a,
1986b)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
Exposure assessment: Presumed
exposure to formaldehyde.
Interview-based information on
lifetime occupational history by job
type and industry.
Occupations evaluated for both no lag
and 15-year lag time between recent
exposure and diagnosis.
Duration and timing: Duration and
timing of exposure were evaluated.
Variation in exposure: Occupation and
industry
Duration:
Level 1 (unexposed)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Other exposures: Not evaluated.
[As noted in Appendix A.5.9: Wood
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder. However, as this is a
case-control study the correlation
between formaldehyde and wood dust
is expected to be small and thus wood
dust would not be expected to be a
confounder.]
Internal comparisons:
Carpenter (lagged 15 years)
All Industries:
OR = 1.3 (0.5-3.4)
All Industries by Duration:
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 0.6 (not given)
Level 3 OR = 2.2 (not given)
Carpenter (lagged 15 years)
Construction industry:
OR = 1.8 (0.7-4.8)
[11]
[10]
Construction by Duration:
Level
1
OR = 1.0
Level
2
OR = 0.7
Level
3
OR = 6.2
LOW (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias due to use
of next of kin.
This document is a draft for review purposes only and does not constitute Agency policy.
1-266 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Vaughan et al.
(1986a)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System between 1980
and 1983. Participation for all cases
was 69 and 80% for controls.
Interviews completed for 71% of cases
and 83% of controls.
Outcome definition: Diagnosis of
oropharynx/hypopharynx cancer (ICD
codes 146 and 148) based on review
of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates.
Design: Population-based, case-
control study of 205 incident cases
with cancer of the
oropharynx/hypopharynx including
unspecified pharyngeal sites. 552
controls were identified by random
digit dialing in same geographic area.
Analysis: ORs were calculated by
logistic regression and adjusted for
cigarette smoking, alcohol
consumption, sex, and age. An
induction period of 15 years was also
utilized to account for latency in
evaluating exposure score.
Related studies:
Vaughan et al. (1986b)
Confidence in effect estimates:3
SB
IB Cf Oth
Overall
Confidence
Low
1
J
U
LJ
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias due to use
of next of kin.
Exposures
Exposure assessment: Interview-based
information on lifetime occupational
exposure to formaldehyde with cases,
next of kin, and controls. Exposure
from available hygiene data, NIOSH
and other data, and NCI job-exposure
linkage system.
Results: effect estimate (95% CI)
[# of cases]
Multiple exposure metrics including
intensity, # of years exposed, and
exposure score based on the sum of
# years spent per job weighted by
estimated formaldehyde level were
evaluated. Exposure score calculated
for both no lag and 15-year lag time
between recent exposure and
diagnosis.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Intensity:
Level 1 (background)
Level 2 (low)
Level 3 (medium)
Level 4 (high)
Number of years exposed:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Exposure score (no lag):
Level 1 (0 to 4)
Level 2 (5 to 19)
Level 3 (>20)
Exposure score (15-year lag):
Level 1 (0 to 4)
Level 2 (5 to 19)
Level 3 (>20)
Coexposures: Not evaluated.
fAs noted in Appendix A.5.9: Wood
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder. However, as this is a
case-control study the correlation
between formaldehyde and wood dust
is expected to be small and thus wood
dust would not be expected to be a
confounder.]
Internal comparisons:
Intensity:
Level 1 OR = 1.0 (Ref. value)
[147]
Level 2 OR = 0.8 (0.5-1.4) [41]
Level 3 OR = 0.8 (0.4-1.7) [13]
Level 4 OR = 0.6 (0.1-2.7) [4]
Number of years exposed:
Level 1 OR = 1.0 (Ref. value)
[147]
Level 2 OR = 0.6 (0.3-1.0) [32]
Level 3 OR = 1.3 (0.7-2.5) [26]
Exposure score (no lae):
Level 1 OR = 1.0 (Ref. value)
[170]
Level 2 OR = 0.6 (0.3-1.2) [14]
Level 3 OR = 1.5 (0.7-3.0) [21]
Exposure score (15-vear lag):
Level 1 OR = 1.0 (Ref. value)
[174]
Level 2 OR = 0.9 (0.4-1.8) [16]
Level 3 OR = 1.3 (0.6-3.1) [15]
This document is a draft for review purposes only and does not constitute Agency policy.
1-267 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Vaughan et al.
(1986b)
Population: Males and females
between the ages of 20 and 74 years
residing in a 13-county area identified
by the Washington State Cancer
Surveillance System between 1980
and 1983. Participation for all cases
was 68.7 and 80.0% for controls.
Interviews completed for 71% of cases
and 83% of controls.
Outcome definition: Diagnosis of
oropharynx/hypopharynx cancer (ICD
codes 146 and 148) based on review
of hospital medical records,
surveillance of private radiotherapy
and pathology practices, and state
death certificates.
Design: Population-based, case-
control study of 205 incident cases
with cancer of the
oropharynx/hypopharynx including
unspecified pharyngeal sites. 552
controls were identified by random
digit dialing in same geographic area
with one control per case randomly
selected from all the eligible persons
in the household and frequency
matched for gender and age.
Analysis: ORs were calculated by
multiple logistic regression and
adjusted for cigarette smoking,
alcohol consumption, sex, and age.
Related studies:
Vaughan et al. (1986a)
Confidence in effect estimates:3
Overall
SB IB Cf
Oth
Confidence
1 1 1
Low
H *
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group D
SB: Potential selection bias due to use
of next of kin.
Exposures
Exposure assessment: Interview-based
information on lifetime occupational
history and residential history from
cases, controls, and next of kin for
deceased cases.
Results: effect estimate (95% CI)
[# of cases]
Multiple exposure metrics including
type of dwelling (i.e., mobile home)
and use of particleboard or plywood
were evaluated.
Duration and timing: Exposure period
since 1950. Duration of exposure was
evaluated.
Variation in exposure:
Residence in mobile home:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Years of exposure to particleboard or
plywood:
Level 1 (0 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Coexposures: Not evaluated.
Information of occupational exposures
provided in Vaughan et al.
(1986a)
[As noted in Appendix A.5.9: Wood
dust is associated with risk of sinonasal
cancer and was not evaluated as a
confounder. However, as this is a
case-control study the correlation
between formaldehyde and wood dust
is expected to be small and thus wood
dust would not be expected to be a
confounder.]
Internal comparisons:
Residence in mobile home:
Level 1 OR = 1.0 (Ref. value)
[177]
Level 2 OR = 0.9 (0.5-1.8) [21]
Level 3 OR = 0.8 (0.2-2.7) [7]
Years of exposure to particleboard:
Level 1 OR = 1.0 (Ref. value)
[137]
Level 2 OR = 1.1 (0.7-1.9) [40]
Level 3 OR = 0.8 (0.5-1.4) [28]
This document is a draft for review purposes only and does not constitute Agency policy.
1-268 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Laryngeal cancer
Epidemiological evidence
Evidence describing an association between formaldehyde exposure and the risk of
developing or dying from laryngeal cancer was available from 18 studies—13 cohort studies
(Coggon et al.. 2014: Beane Freeman etal.. 2013: Meyers etal.. 2013: Band etal.. 1997: Takobsson et
al.. 1997: Andielkovich etal.. 1995: Hansen and Olsen. 1995: Hansen etal.. 1994: Hayes etal.. 1990:
Stroup etal.. 1986: Levine etal.. 1984a: Walrath and Fraumeni. 1984.19831 and five case-control
studies fShangina etal.. 2006: Berrino etal.. 2003: Laforest etal.. 2000: Gustavsson et al.. 1998:
Wortlev etal.. 19921. Two reported results were classified as uninformative. Berrino et al. (2003)
was classified as uninformative due to likely confounding by highly correlated coexposures, one of
which was a stronger risk factor for laryngeal cancer in that study than was formaldehyde
(i.e., solvents). Hansen et al. (19941 was classified as uninformative due to likely information bias
stemming from the rarity of exposure among cases in that cohort. The outcome-specific
evaluations of confidence in the precise effect estimate of an association from each study are
provided in Appendix A.5.9. Details of the reported results of high, medium, and low confidence are
provided in the evidence table for laryngeal cancer (see Table 1-35) following the causal evaluation.
Consistency of the observed association
The results of the 16 informative studies were not consistent. The study results presented
in Table 1-35 (by confidence level and publication date) detail all of the reported associations. Only
one set of results was classified with high confidence fBeane Freeman et al.. 20131. and those
results surrounded the null with a modest increase in risk overall with SMR = 1.23
(95% CI 0.91,1.67), and at the highest level of average intensity of exposure a RR = 1.73
(95% CI 0.83, 3.60), and conversely, a modest decrease in risk at the highest level of peak exposure
with RR = 0.72 (95% CI 0.32,1.65), and a stronger decreased risk at the highest level of duration of
exposure with RR = 0.33 (95% CI 0.10,1.11). Of the five sets of results classified with medium
confidence (Coggon etal.. 2014: Shangina etal.. 2006: Laforest etal.. 2000: Wortlev etal.. 1992:
Hayes etal.. 19901. only two reported clearly increased risks; Shangina et al. (2006) showed an
association with the highest level of cumulative exposure (OR = 3.12, 95% CI 1.23, 7.91) and
Wortley et al. (1992.) showed an association among those with at least 10 years of exposure and the
highest peak exposures (OR = 4.3, 95% CI 1.0,18.7). Coggon etal. (2014) found modestly increased
risk for the cohort as a whole (SMR = 1.22, 95% CI 0.76,1.84) and higher risks among those
workers who had ever been "highly" exposed (SMR = 1.96, 95% CI 0.98, 3.50). They did not find
greater risk among those who had been "highly" exposed for more than 1 year (SMR = 1.30,
This document is a draft for review purposes only and does not constitute Agency policy.
1-269 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
95% CI 0.39, 4.38). The results from Laforest et al. (2000) and Hayes et al. (19901 did not show
consistently increased risks. The study results classified with low confidence were consistently
around the null. Results are plotted in Figure 1-23.
Strength of the observed association
Summary effect estimates for the association between formaldehyde exposure and the
relative effect estimates of developing or dying from laryngeal cancer ranged from 0.33 to 4.3 and
generally clustered around the null. The study results classified with low confidence were all
limited to summary estimates without examination of exposures levels within the exposed study
subjects. The results classified with medium confidence differentiated the risks by levels of
exposure, and these results showed somewhat higher effect estimates among the most highly
exposed groups, but these effect estimates were largely less than a doubling of risk. There were
two results of medium confidence that reported more than a tripling of risk (Shangina et al.. 2006:
Wortlev etal.. 1992). However, the one set of results classified with high confidence (Beane
Freeman et al.. 2013) did not report a consistent pattern of increased risk.
Specificity of the observed association
Only the specific diagnosis of laryngeal cancer was considered here. The most specific level
of laryngeal cancer diagnosis that is commonly reported across the epidemiological literature has
been based on the first three digits of the Eighth or Ninth Revision of the ICD code (i.e., Laryngeal
cancer ICD-9: 161).
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants started
prior to their diagnoses of laryngeal cancer and in the studies that ascertained individual-level
exposures, the estimation of formaldehyde exposures was based on job titles and done in a blinded
fashion with respect to outcome status. While several of the studies did report results with lagged
exposures to account for potential latency effects, none of the 16 studies provided details of
analyses of a temporal relationship between the timing of exposure using different lags and the
diagnoses of laryngeal cancer or deaths from laryngeal cancer. However, Shangina et al. (2006) did
state that a 20-year lag in exposure was assessed but did not report those details for formaldehyde;
and Wortley et al. (1992) reported that a 10-year lag in exposure 'only slightly' increased the
estimated effects.
Exposure-response relationship
The strongest evidence of an exposure-response was reported by Shangina et al. (2006).
who found that among cases of "Ever" exposed to formaldehyde, the OR = 1.68 (95% CI 0.85, 3.31),
those cases with the highest tertile of cumulative exposure had an OR = 3.12 (95% CI 1.23, 7.91).
Shangina et al. (2006) also reported suggestions of trends for increased risk with increasing tertiles
This document is a draft for review purposes only and does not constitute Agency policy.
1-270 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
of duration of exposure (p < 0.06) and with increasing tertile of cumulative exposure (p < 0.07).
Wortley et al. (1992.) also found higher risks among the most highly exposed with an OR = 4.3 (95%
CI 1.0,18.7). However, Beane Freeman et al. (2013.) did not find consistent evidence of an
exposure-response relationship for increasing peak exposure (p > 0.5), for increasing average
intensity (p = 0.44), but did find a significant trend (p = 0.02) with cumulative exposure that may
have been decreasing in nature with lower risks at higher exposures.
Potential impact of selection bias; information bias; confounding bias, and chance
For laryngeal cancer, the reliance of cohort studies on death certificates to detect cancers
with relatively high survival underestimated the actual incidence of those cancers. Five-year
survival rates are about 60% (see Appendix A.5.9). This may have resulted in undercounting of
incident cases and underestimates of effect estimates in cohort studies compared to general
populations. Selection bias could have somewhat obscured a truly larger effect of formaldehyde
exposure on the risk of death from laryngeal cancer and may explain the preponderance of effect
estimates near the null. The case-control studies Shangina et al. (2006). Laforest et al. (2000).
Gustavsson et al. (1998), and Wortley et al. (1992), because they recruited incident cases, were less
prone to such a bias. Information bias may distort findings when subjects' true personal exposures
are inaccurately assigned. Random measurement error typically results in a bias toward the null,
thereby obscuring any real effect by underestimating the effect's magnitude. Confounding is
another potential bias that could arise if another cause of laryngeal cancer was statistically
associated with formaldehyde exposure. However, there does not appear to be any evidence of
negative confounding that could have obscured a real but unobserved effect. Overall, bias is
considered to be an unlikely alternative cause for the isolated reports of increased risks of laryngeal
cancer associated with formaldehyde exposures.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
laryngeal cancer placed the greatest weight on five particular considerations: (1) the suggestive
associations reported for two medium confidence studies (Shangina et al.. 2006: Wortley etal..
1992): (2) the suggestive exposure-response relationships of increased risk with increased
formaldehyde exposure, specifically by Shangina et al. (2006), but the lack of support for exposure-
response from other studies including the single set of results classified with high confidence which
found a significant downward trend in risks with increasing exposure (Beane Freeman et al.. 2013):
(3) the moderate survival rate for laryngeal cancer (60%), which may indicate that mortality data
are not as good a proxy for incidence data for this cancer type; and (4) the absence of evidence to
evaluate the potential risk to sensitive populations or lifestages.
This document is a draft for review purposes only and does not constitute Agency policy.
1-271 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Conclusion
The available epidemiological studies provide indeterminate evidence to assess the
potential for an association between formaldehyde exposure and an increased risk of
laryngeal cancer.
(TS
E
¦H
111
•l-i
o
o
!£
LU
%
TO O
IS LU
Sen a>
i i
Laforest
Sh.
O) O)
8 S
W X 0)
I % £
& 1 £
S g a
lis
1 i ;
Beane
Freemai
-I r
Coggon
•i
^ — — "5
li ' " CD CD c
c CM Ol ^ ft
_ I
3 £ £ £ =
o P CI CO CO o
(N CM CD CD CD
CO If) CO
„ t5
_ sr - a
00 CO CO
C — '—' c
tt ^ a: o: it
S 5 5 1 S
0.0.(0(0(0
II II II II
N CO CO ^
f- h- CO ¦<—
ii a: gz m
a: S a. 2E s s tt
o CO CO CO CO CO CO
T
T
& t$* S5
//
//v'Vv-
V V
~
>0 A
Figure 1-23. Epidemiological studies reporting laryngeal cancer risk
estimates.
For each measure of association, the number of exposed cases is provided in brackets (e.g., [n = 1]). For
studies reporting results on multiple metrics of exposure, only the highest category of each exposure
metric is presented in the figure. Note that the confidence intervals for Band et al. (1997) are 90%, not
95%. Abbreviations: SMR = standardized mortality ratio; RR = relative risk; OR = odds ratio.
SPIR = standardized proportional incidence ratio.
This document is a draft for review purposes only and does not constitute Agency policy,
1-272 ' DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-35. Epidemiological studies of formaldehyde exposure and risk of
laryngeal cancer
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Beane Freeman et
al. (2013)
Population: 25,619 workers employed
at 10 formaldehyde-using or
formaldehyde-producing plants in the
United States followed from either the
plant start-up or first employment
through 2004. Deaths were identified
from the National Death Index with
remainder assumed to be living. 676
workers (3%) were lost to follow-up.
Vital status was 97.4% complete and
only 2.6% lost to follow-up.
Outcome definition: Death
certificates used to determine
underlying cause of death from
laryngeal cancer (ICD-8:161).
Histological typing not reported.
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in
lowest exposed category. Lagged
exposures were evaluated to account
for cancer latency. Results were
presented for 15-year lag.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Hauptmann et al. (2004)
Beane Freeman et al. (2009)
Confidence in effect estimates:3
SB IB a Oth
Overall
Confidence
High
Exposure assessment: Individual-level
exposure estimates based on job titles,
tasks, visits to plants by study
industrial hygienists who took 2,000 air
samples from representative job, and
monitoring data from 1960 through
1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3). Median cumulative
exposure = 0.6 ppm-years (range
0-107.4).
Multiple exposure metrics including
peak, average, and cumulative
exposures were evaluated using
categorical and continuous data.
Duration and timing: Exposure period
from <1946 to 1980. Median length of
follow-up: 42 years. Median length of
employment was 2.6 years (range
1 day-47.7 years). Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Peak exposure:
Level 1 (>0 to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Level 1 (>0 to <0.5 ppm)
Level 2 (0.5 to <1.0 ppm)
Level 3 (>1.0 ppm)
Cumulative exposure:
Level 1 (>0 to <1.5 ppm-yrs)
Level 2 (1.5 to <5.5 ppm-yrs)
Level 3 (>5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders.
Internal comparisons:
Peak exposure
Unexposed RR = 0.79 (0.25-2.48) [6]
Level 1 RR = 1.00 (Ref. value) [17]
Level 2 RR = 1.52 (0.76-3.05) [16]
Level 3 RR = 0.72 (0.32-1.65) [9]
p-trend (exposed) > 0.50;
p-trend (all) >0.50
Average intensity
Unexposed RR = 0.89 (0.29-2.75) [6]
Level 1 RR = 1.00 (Ref. value) [21]
Level 2 RR = 1.25 (0.57-2.76) [9]
Level 3 RR = 1.73 (0.83-3.6) [12]
p-trend (exposed) = 0.44;
p-trend (all) = 0.39
Cumulative exposure
Unexposed RR = 0.67 (0.22-2.00) [6]
Level 1 RR = 1.00 (Ref. value) [29]
Level 2 RR = 1.01 (0.49-2.11) [10]
Level 3 RR = 0.33 (0.10-1.11) [3]
p-trend (exposed) = 0.02;
p-trend (all) = 0.03
External comparisons:
SMRunexposed =0.93(0.42-2.08)
SMRExposed = 1.23
(0.91-1.67) [42]
[6]
HIGH •
IB: Exposure: Group A
This document is a draft for review purposes only and does not constitute Agency policy.
1-273 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Coggon et al. (2014)
Population: 14,008 British men
employed in six chemical industry
factories that produced
formaldehyde. Cohort mortality
followed from 1941 through 2012.
Cause of deaths was known for 99% of
5,185 deaths through 2000. Similar
cause of death information not
provided on 7,378 deaths through
2012. Vital status was 98.9%
complete through 2003. Similar
information not provided on deaths
through 2012.
Outcome definition: Death
certificates used to determine cause
of deaths from laryngeal cancer.
Design: Cohort mortality study with
external comparison group with a
nested case-control study.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
SB IB Cf Oth
Ove ral 1
Confide nee
Medium
MEDIUM >1/ (Potential bias toward the
null)
IB: Exposure: Group B; lack of latency
analysis.
Coexposures: Not evaluated. Potential
low-level exposure to stvrene.
ethylene oxide, epichlorhydrin,
solvents, asbestos, chromium salts,
and cadmium.
Results: effect estimate (95% CI)
[# of cases]
[As noted in Appendix A.5.9: Stvrene is
associated with LHP cancers but not
URT cancers.
Asbestos is associated with URT
cancers, including laryngeal cancer.
Authors stated that the extent of
coexposures was expected to be low.
Potential for confounding may be
mitigated by low coexposures.
Other coexposures are not known risk
factors for this outcome.]
Exposure assessment: Occupational
histories obtained by interview and
yielded information on all jobs held
>lyear. A general questionnaire
obtained information of job titles,
tasks, industries, starting and stopping
times, full-time/part-time status,
working environments, and specific
exposures. A specific questionnaire
Exposures
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration
was evaluated as more, or less, than
one year only among the "High"
exposure group. Timing since first
exposure was not evaluated.
Variation in exposure:
Highest exposure level attained
Level 1 (Background)
Level 2 (low/moderate)
Level 3 (High)
Duration of "High" exposures
Level 1 (Background)
Level 2 (<1 year)
Level 3 (1 year or more)
External comparisons:
SMR = 1.22 (0.76-1.84)
Highest exposure level attained
Level 1 SMR = 0.33 (0.04-1.20)
Level 2 SMR = 1.40 (0.64-2.66)
Level 3 SMR = 1.96 (0.98-3.50)
Internal comparisons:
Highest exposure level attained
Level 1 OR = 1.00 (Ref. value)
Level 2 OR = 1.20 (0.53-2.73)
Level 3 OR = not given
Duration of'
Level 1
Level 1
Level 2
High" exposures
OR = 1.00 (Ref. value)
OR = 2.02 (0.65-6.27)
OR = 1.30 (0.39-4.38)
[22]
[2]
[9]
[11]
[14]
[17]
[22]
[14]
[14]
[8]
Internal comparisons:
Exposure to formaldehyde:
Level 1 OR = 1.00 (Ref. value)
Level 2 OR = 1.68 (0.85-3.31)
Duration of exposure:
p-trend (all) = 0.06
Cumulative exposure:
[298]
[18]
Reference: Shangina et al.
(2006)
Population: Males between the ages
of 15 and 79 years residing in four
European countries that were
diagnosed with laryngeal cancer
during 1999-2002 and identified by
This document is a draft for review purposes only and does not constitute Agency policy.
1-274 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
study centers in Romania, Poland,
Russia, and Slovakia.
Outcome definition: Diagnosis of
laryngeal cancer was histologically or
cytologically confirmed and included
topographic subcategories from ICD-0
code C32 (glottis, supraglottis,
subglottis, laryngeal cartilage,
overlapping lesion of the larynx, and
larynx, unspecified).
Design: Multicenter case-control
study of 316 laryngeal cancer cases.
728 hospital controls were frequency
matched by age.
Analysis: ORs were calculated by
unconditional logistic regression and
adjusted for age, country, tobacco
smoking, and alcohol consumption.
An induction period of 20 years was
also utilized to account for latency in
evaluating risk.
Confidence in effect estimates:3
Overall
SB IB
Ct
iifh
Confidence
1
Medium
1
MEDIUM (Potential bias toward the
null)
IB: Exposure: Group C
Oth: Low power due to rarity of
exposure.
was completed for employment in
defined jobs or industries.
Exposure assessment based on expert
judgment of reported task
descriptions. Exposure scored
according to intensity, frequency, and
confidence.
Multiple exposure metrics including
known exposure and cumulative
exposure were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (never)
Level 2 (ever)
Cumulative exposure (tertiles):
Level 1 (Tertile 1 unspecified)
Level 2 (Tertile 2 unspecified)
Level 3 (>22,700 mg/m3-hrs)
Duration of exposure (tertiles):
Level 1 (Tertile 1 unspecified)
Level 2 (Tertile 2 unspecified)
Level 3 (Tertile 3 unspecified)
Definitions for levels of exposure for
duration of exposure to formaldehyde
and cumulative exposure not provided
by authors except for the lower bound
of Tertile 3 for cumulative exposure.
Other exposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9: Other
exposures that were found to be risk
factors included dusts of "hard alloys"
(16 cases) and chlorinated solvents (15
cases).
Hard-alloy dust and chlorinated
solvents were each found in fewer
than 6% of cases, the correlation
between them is considered to be
small enough to make confounding
unlikely.]
Level 1 Unspecified
Level 2 Unspecified
Level 3 OR = 3.12 (1.23-7.91)
p-trend (all) = 0.07
Duration of exposure:
Level 1 Unspecified
Level 2 Unspecified
Level 3 Unspecified
p-trend (all) = 0.06
No notable findings were reported between
formaldehyde exposure and the risk of
laryngeal cancer when considering an
induction period of 20 years.
Reference: Laforest et al.
(2000)
Exposure assessment: Occupational
history obtained by interview.
Exposure assessment based on job-
exposure matrix that included level
Internal comparisons:
All subjects
Exposure to formaldehyde:
This document is a draft for review purposes only and does not constitute Agency policy.
1-275 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
and probability of exposure, duration,
and cumulative exposure to
formaldehyde.
Multiple exposure metrics including
known exposure, probability of
exposure, and cumulative exposure
were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
All subjects
Exposure to formaldehyde:
Level 1 (never exposed)
Level 2 (ever exposed)
Probability of exposure:
Level 1 (never exposed)
Level 2 (<10%)
Level 3 (10 to 50%)
Level 4 (>50%)
Duration of exposure:
Level 1 (never exposed)
Level 2 (<7 years)
Level 3 (7 to 20 years)
Level 4 (>20 years)
Cumulative exposure:
Level 1 (never exposed)
Level 2 (low, <0.02)
Level 3 (medium, 0.02 to 0.09)
Level 4 (high, >0.09)
Subjects with a probability of exposure
>10%
Exposure to formaldehyde:
Level 1 (never exposed)
Level 2 (ever exposed)
Duration of exposure:
Level 1 (never exposed)
Level 2 (<7 years)
Level 3 (7 to 20 years)
Level 4 (>20 years)
Cumulative exposure:
Level 1 (never exposed)
Level 2 (low)
Level 3 (medium)
Level 4 (high)
Other exposures: asbestos, coal dust,
leather dust, wood dust, flour dust,
silica, and textile dust.
[As noted in Appendix A.5.9: Of these,
none significantly increased the risk of
Results: effect estimate (95% CI)
[# of cases]
Population: Males diagnosed with
primary laryngeal squamous cell
cancers between January 1989 and
May 1991 and identified through 15
French hospitals. Interviews
completed for 79.5% of eligible cases
and 86% of eligible controls.
Outcome definition: Diagnosis of
laryngeal was histologically confirmed.
Design: Hospital-based case-control
study of 296 laryngeal cancers. 296
hospital controls frequency matched
on age.
Analysis: Ors were calculated by
unconditional logistic regression and
adjusted for age, alcohol, and
smoking. Induction periods of 5,10,
and 15 years was also utilized to
account for latency in evaluating risk.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Corrfi dence
Medium
MEDIUM (Potential bias toward the
null)
IB: Exposure: Group C
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.14 (0.76-1.70)
[102]
Probability of exposure:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.16 (0.73-1.86) [58]
Level 3 OR = 1.12 (0.55-2.30) [23]
Level 4 OR = 1.04 (0.44-2.47) [21]
Cumulative exposure:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.12 (0.62-2.01) [35]
Level 3 OR = 1.44 (0.79-2.63) [38]
Level 4 OR = 0.87 (0.45-1.67) [29]
Duration of exposure:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.42 (0.75-2.68) [35]
Level 3 OR = 1.09 (0.62-1.96) [37]
Level 4 OR = 0.96 (0.52-1.76) [30]
Subjects with a probability of exposure >10%
Exposure to formaldehyde:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.17 (0.63-2.17)
Cumulative exposure:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 0.68 (0.12-3.90)
Level 3 OR = 1.86 (0.76-4.55)
Level 4 OR = 0.91 (0.42-1.99)
Duration of exposure:
Level 1 OR = 1.00 (Ref. value)
[194]
Level 2 OR = 1.68 (0.60-4.72)
Level 3 OR = 0.86 (0.33-2.24)
Level 4 OR = 1.14 (0.47-2.74)
[44]
[4]
[17]
[23]
[15]
[14]
[15]
Introduction of induction times as described
did not substantially change the results.
This document is a draft for review purposes only and does not constitute Agency policy.
1-276 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
laryngeal cancer in this study but coal
dust was controlled for in the laryngeal
cancer analysis.]
Reference: Wortley et al.
(1992)
Population: Males and females
between the ages of 20 and 74 years
residing in western Washington who
were diagnosed with laryngeal cancer
between September 1983 and
February 1987 and identified through
the cancer surveillance system of the
Fred Hutchinson Cancer Research
Center. Interviews completed for
80.8% of eligible cases and 80% of
eligible controls.
Outcome definition: Diagnosis of
cancer of the larynx based on ICD
codes 161.0-161.9 from cancer
registry data.
Design: Population-based case-control
study of 235 cases of laryngeal cancer.
547 controls identified from random
digit dialing and were selected for the
same distributions of age and sex to
the cases.
Analysis: ORs were calculated by
multiple logistic regression and
adjusted for smoking, drinking, age,
and education. An induction period of
10 years was also utilized to account
for latency in evaluating duration and
exposure score.
Confidence in effect estimates:3
SB
IB Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward the
null)
IB: Exposure: Group C
Exposure assessment: Occupational
history obtained by interview for all
jobs held for >6 months and included
job titles, description of tasks
performed, and industry. Job titles
analyzed by duration of exposure
(<9 year and >10 years).
Exposures assessment based on
job-exposure matrix. Industrial
hygienists classified jobs into four
levels of exposure to formaldehyde
based on judgment of both likelihood
and degree of exposure.
Exposure score calculated as the
weighted sum of years with exposure,
with weight based on level of exposure
code. Exposure codes defined as:
0 = no, 1 = low, 2 = medium, and
3 = high.
Multiple exposure metrics including
peak exposure (subject's highest
exposure code) and exposure score
were evaluated.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Peak exposure:
Level 1 (none)
Level 2 (low)
Level 3 (medium)
Level 4 (high)
Duration:
Level 1 (<1 years)
Level 2 (1 to 9 years)
Level 3 (>10 years)
Exposure scores:
Level 1 (<5)
Level 2 (5 to 19)
Level 3 (>20)
Peak and Duration:
Level 1 (none)
Level 2 (med/high and >10 years)
Level 3 (high and >10 years)
Other exposures: asbestos, chromium.
nickel, cutting oils, and diesel fumes.
High-risk occupations (e.g., mechanics,
Internal comparisons:
Peak exposure:
Level 1 OR = 1.0 (Ref. value)
[177]
Level 2 OR = 1.0 (0.6-1.7)
Level 3 OR = 1.0 (0.4-2.1)
Level 4 OR = 2.0 (0.2-20)
Duration:
Level 1 OR = 1.0 (Ref. value)
[182]
Level 2 OR = 0.8 (0.4-1.3)
Level 3 OR = 1.3 (0.6-3.1)
Exposure scores:
Level 1 OR = 1.0 (Ref. value)
[201]
Level 2 OR = 1.0 (0.5-2.0)
Level 3 OR = 1.3 (0.5-3.3)
Peak and Duration
[42]
[14]
[2]
[27]
[26]
[18]
[16]
[177]
Level 1 OR = 1.0 (Ref. value)
Level 2 OR = 4.2 (0.9-19.4) [not given]
Peak and Duration
Level 1 OR = 1.0 (Ref. value) [177]
Level 2 OR = 4.2 (0.9-19.4) [not given]
Level 3 OR = 4.3 (1.0-18.7) [not given]
No notable findings were reported between
formaldehyde exposure and the risk of
laryngeal cancer when considering an
induction period of 10 years.
This document is a draft for review purposes only and does not constitute Agency policy.
1-277 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
carpenters, painters, textile machine
operators) likely had coexposures to
unidentified substances.
[As noted in Appendix A.5.9: This is a
case-control study the correlation
between formaldehyde and those
potential confounders is expected to
be small, and thus, wood dust would
not be expected to be a confounder.]
Reference: Hayes et al. (1990)
Population: 4,046 deceased U.S. male
embalmers and funeral directors,
derived from licensing boards and
funeral director associations in 32
states and the District of Columbia
who died during 1975-1985. Death
certificates obtained for 79% of
potential study subjects (n = 6,651)
with vital status unknown for 21%.
Outcome definition: Death
certificates and licensing boards used
to determine cause of death from
laryngeal cancer (ICD-8:161).
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IS Cf Oth
Overall
Confidence
Medium
MEDIUM (Potential bias toward the
null)
IB: Exposure Group A; latency not
evaluated
Oth: Potential undercounting of cases.
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative. Exposure based on
occupation which was confirmed on
death certificate. Authors
subsequently measured personal
embalming exposures ranging from
0.98 ppm (high ventilation) to
3.99 ppm (low ventilation) with peaks
up to 20 ppm.
Authors state that major exposures are
to formaldehyde and possibly
glutaraldehyde and phenol.
Duration and timing: Occupational
exposure preceding death during
1975-1985. Of 115 deaths from LHP
cancer, 66 (57%) were aged
60-74 years. Duration and timing
since first exposure were not
evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene, stains,
chlorinated hydrocarbons, dioxane,
and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
External comparisons:
PMR = 0.64 (0.26-1.33)
[7]
This document is a draft for review purposes only and does not constitute Agency policy.
1-278 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Meyers et al. (2013)
Population: 11,043 workers in 3 U.S.
garment plants exposed for at least
3 months. Women comprised 82% of
the cohort. Vital status was followed
through 2008 with 99.7% completion.
Outcome definition: Death
certificates used to determine both
the underlying cause of death from
laryngeal cancer (ICD code in use at
time of death).
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates.
Related studies:
Pinkerton et al. (2004)
Stayner et al. (1985)
Stayner et al. (1988)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null; low sensitivity)
IB: Exposure: Group A; lack of latency
analysis.
Oth: Low power due to rarity of cases.
Reference: Gustavsson et al.
(1998)
Population: Males between the ages
of 40 and 79 years residing in Sweden
identified by hospitals reports or
regional cancer registries during
1988-1990. Interviews completed for
90% of cases and 85% of controls.
Outcome definition: Diagnosis of
laryngeal cancer based on ICD-9 codes
on weekly reports from departments
of otorhinolaryngology, oncology, and
Exposures
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
1984. Geometric TWA8 exposures
ranged from 0.09 to 0.20 ppm. Overall
geometric mean concentration of
formaldehyde was 0.15 ppm, (GSD
1.90 ppm). Area measures showed
constant levels without peaks.
Historically earlier exposures may have
been substantially higher.
Exposure assessment: Occupational
history obtained by interview and
yielded information on all jobs held
>1 year, starting and stopping times,
job title, tasks, and company. Histories
reviewed by industrial hygienist who
coded jobs based on intensity and
probability of exposure to 17
occupational factors.
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
SMR = 0.77 (0.21-1.97) [4]
Exposure assessments estimated
intensity on a 4-point scale and
probability of exposure as point
estimates. Cumulative exposure
calculated as the product of exposure
Internal comparisons:
Exposure to formaldehyde:
Level 1 RR = 1.00 (Ref. value) [# not
given]
Level 2 RR = 1.45 (0.83-2.51)
[23]
Duration and timing: Exposure period
from 1955 to 1983. Median duration
of exposure was 3.3 years. More than
40% exposures <1963. Median time
since first exposure was 39.4 years.
Duration and timing since first
exposure were not evaluated for this
cancer.
Variation in exposure: Not evaluated.
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
This document is a draft for review purposes only and does not constitute Agency policy.
1-279 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
intensity, probability of exposure, and
duration of exposure, and by adding
contributions over entire work history.
Duration and timing: Duration of
exposure was evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (never)
Level 2 (ever)
Study
surgery and from regional cancer
registries.
Design: Community-based,
case-control study of 157 cases of
squamous cell carcinoma of the
larynx. 641 controls were randomly
identified from population registers
and frequency matched by region and
age.
Analysis: RRs were calculated by
unconditional logistic regression and
adjusted for region, age, drinking, and
smoking.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null)
IB: Exposure: Group B; latency not
evaluated.
Cf: Potential confounding
Oth: Low power
Reference: Band et al. (1997)
Population: 30,157 male workers in
the pulp and paper industry with at
least 1-year employment accrued by
January 1950. Followed through
December 1982. Loss to follow-up
was less than 6.5% for workers
exposed to the sulfate process (67% of
original cohort of 30,157) and less
than 20% for workers exposed to the
sulfite process.
Outcome definition: Cause of death
obtained from the National Mortality
Database based on ICD version in
effect at time of death and
standardize to ICD-9 version. Larynx:
ICD-9 161.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
race, age, and calendar-year-expected
Other exposures: polycyclic aromatic
hydrocarbons, asbestos, general dust,
wood dust, quartz, metal dust, oil mist,
welding fumes, manmade mineral
fibers, paper dust, textile dust,
hexavalent chromium, phenoxy acids,
nickel, acid mist, and leather dust.
[As noted in Appendix A.5.9: Asbestos
and metal dust were both stronger risk
factors for laryngeal cancer so there is
a potential for confounding.]
Variation in exposure:
No variation in formaldehyde exposure
was reported. Results presented by
pulping process (sulfate vs. sulfite) but
neither process uses formaldehyde
which is used in paper making.
Results: effect estimate (95% CI)
[# of cases]
[12]
Workers only in sulfite process
All workers
SMR = 1.78 (90% CI 0.78-3.52) [8]
Work duration <15 years
TSFE <15 years
SMR = 2.46 (90% CI 0.10-11.63) [1]
TSFE >15 years
SMR = 2.13 (90% CI 0.72-4.87) [4]
TSFE >15 years
SMR = 0.93(90% CI 0.04-4.38) [1]
Exposure assessment: Occupational
data limited to hire and termination
dates for all workers and type of
chemical process of pulping (sulfate vs.
sulfite). No job-specific data available.
Presumed exposure to formaldehyde
known to be used in the plant.
Formaldehyde is known to be an
exposure risk for pulp and paper mill
workers: job-specific median
exposures ranging from 0.04 to
0.4 ppm with peaks as high as 50 ppm
(Korhonen et al., 2004).
Duration and timing: Duration and
timing since first exposure were not
evaluated.
External comparisons:
All workers
SMR = 1.01 (90% CI 0.58-1.63)
Work duration >15 years
This document is a draft for review purposes only and does not constitute Agency policy.
1-280 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
numbers of deaths from the Canadian
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null)
IB: Exposure: Group C
Cf: Potential confounding
Coexposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9: Potential
confounders for these outcomes
include chlorophenols, acid mists,
dioxin. and perchloroethvlene and
would likely be positively correlated
with formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.]
Reference: Jakobsson et al.
(1997)
Population: 727 male employees of
two plants producing stainless steel
sinks and saucepans employed at least
1 year during 1927-1981 with
minimum 15-year follow-up.
Outcome definition: Incidence of
laryngeal cancer from the Swedish
Tumor Registry (ICD-7:161).
Design: Cohort incidence study with
external comparison group.
Analysis: SIRs calculated using sex,
age, and calendar-year-expected
number of cases from the national
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure: Group D
Oth: Low power due to rarity of cases.
Exposure assessment: Workers grind
stainless steel with grinding plates
made of formaldehyde resins, which
may release formaldehyde when
heated during grinding operations.
Duration and timing: Occupational
exposure preceding death during
1927-1981. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9: Nickel
and chromium are associated with URT
cancers and would likely be positively
correlated with formaldehyde
exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.
Other coexposures are not known risk
factors for these outcomes.]
External comparisons:
SIR = 0.7 (0-3.9)
[1]
Reference: Andjelkovich et al.
(1995)
Population: 3,929 automotive
industry iron foundry workers
exposed from 1960 to 1987 and
followed through 1989.
Exposure assessment: Individual-level
exposure status (Yes/No, Quartile)
based on review of work histories by an
industrial hygienist.
Exposure assessment blinded to
outcome.
External comparisons:
SMRunexposed =0.70(0.01-3.91) [1]
SMRExposed =0.98(0.11-3.53) [2]
This document is a draft for review purposes only and does not constitute Agency policy.
1-281 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Outcome definition: Underlying cause
of death obtained from Social Security
Administration, Pension Benefit
Informations, and National Death
Index)
Larynx: ICD 161
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex-,
age-, race-, and calendar-year-specific
U.S. mortality rates.
Confidence in effect estimates:3
Ove ra 11
SH
IH
a
iM-h
Confidence
Low
_
_
LOW (Potential bias toward the
null)
IB: Exposure: Group B; Latency not
evaluated
Cf: Potential confounding
Oth: Low power due to rarity of cases.
Independent testing of iron foundries b
NIOSH reported a range from 0.02 ppm
to 18.3 ppm (cited in WHO (1989)
Env. Health Criteria 89: Formaldehyde)
Duration and timing: Duration and
timing since first exposure were not
evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9: Nickel
and chromium are associated with URT
cancers and would likely be positively
correlated with formaldehyde
exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.
Other coexposures are not known risk
factors for these outcomes.]
Reference: Hansen and Olsen
(1995)
Population: 2,041 men with cancer
who were diagnosed during
1970-1984 and whose longest work
experience occurred at least 10 years
before cancer diagnosis. Identified
from the Danish Cancer Registry and
matched with the Danish
Supplementary Pension Fund.
Outcome definition: Cancer of the
larynx (ICD-7: 161) listed on Danish
Cancer Registry file.
Design: Proportionate incidence study
with external comparison group.
Analysis: Standardized proportionate
incidence ratio calculated as the
proportion of cases for a given cancer
in formaldehyde-associated
companies relative to the proportion
of cases for the same cancer among
Exposure assessment: Individual
occupational histories including
industry and job title established
through company tax records to the
national Danish Product Register.
Subjects whose longest work
experience was >10 years prior to
cancer diagnosis were considered
potentially exposed to formaldehyde.
All subjects were stratified based on
job title as either low exposure (white
collar worker), above background
exposure (blue collar worker), or
unknown (job title unavailable).
Duration and timing: Exposure period
since 1964.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9: While
other coexposures were not evaluated,
the overall correlation between
Overall (exposure to formaldehyde >10 years
prior to cancer diagnosis)
SPIR = 0.9 (0.6-1.2) [32]
This document is a draft for review purposes only and does not constitute Agency policy.
1-282 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
coexposures in multiple occupational
industries is likely to be low.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Study
all employees in Denmark. Adjusted
for age and calendar time.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
1
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group D
Reference: Stroup et al. (1986)
Population: 2,239 white male
members of the American Association
of Anatomists from 1888 to 1969 who
died during 1925-1979. Death
certificates obtained for 91% with 9%
lost to follow-up.
Outcome definition: Laryngeal cancer
(ICD-8: 161) listed as cause of death
on death certificates.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null)
IB: Exposure Group A; latency not
evaluated
SB: Healthy worker effect.
Oth: Low power due to rarity of cases.
Reference: Levine et al. (1984a)
Population: 1,477 male undertakers
licensed with the Ontario Board of
Funeral Services from 1928 to 1957
who died during 1950-1977. Vital
status was followed through 1977
Duration and timing: Occupational
exposure preceding death during
1925-1979. Median birth year was
1912. By 1979, 33% of anatomists had
died. Duration and timing since first
exposure were not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene, stains,
chlorinated hydrocarbons, dioxane,
and osmium tetroxide.
Results: effect estimate (95% CI)
[# of cases]
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
External comparisons:
SMR = 0.4 (0-2.0)
[1]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure preceding death during
External comparisons:
Observed: 1
Expected: 1.0
SMR = 1.00 (0.05-4.93)+
[1]
This document is a draft for review purposes only and does not constitute Agency policy.
1-283 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
with 96% completion and only 4% lost
to follow-up.
Outcome definition: Death
certificates used to determine cause
of death from cancer of the larynx
(ICD-8: 161).
Design: Retrospective cohort
mortality study with external
comparison group.
Analysis: Ontario mortality rates for
<1950 not available for SMR
calculations. Expected deaths were
determined by applying age- and
calendar year-specific mortality rates
of Ontario men to the 1950 through
1977 experience of the cohort.
Confidence in effect estimates:3
SB
IB Cf Oth
Overall
Confidence
Low
Low >1/ (low sensitivity;
potential bias toward the null)
IB: Exposure Group A; latency not
evaluated
SB: Healthy worker effect.
Oth: Low power due to rarity of cases.
Reference: Walrath and
Fraumeni (1984)
Population: 1,007 deceased white
male embalmers from the California
Bureau of Funeral Directing and
Embalming who died during
1925-1980. Death certificates
obtained for all.
Outcome definition: Laryngeal cancer
listed as cause of death on death
certificates.
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Exposures
1950-1977. Duration and timing since
first exposure were not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene, stains,
chlorinated hydrocarbons, dioxane,
and osmium tetroxide.
Duration and timing: Occupational
exposure preceding death during
1916-1978. Birth year ranged from
1847 to 1959. Median age of death
was 62 years. Most deaths were
among embalmers with active licenses.
Duration and timing since first
exposure were not evaluated.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Results: effect estimate (95% CI)
[# of cases]
+EPA derived CIs using the Mid-P Method
(See Rothman and Boice, 1979)
PMR = 0.77 (0.13-2.54)+ [2]
+EPA derived CIs using the Mid-P Method
(See Rothman and Boice, 1979)
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
External comparisons:
Observed: 2
Expected: 2.6
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
This document is a draft for review purposes only and does not constitute Agency policy.
1-284 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null;
low sensitivity)
SB: Potential selection bias: due to
incomplete death certificate
ascertainment.
IB: Exposure Group A; latency not
evaluated
Oth: Low power due to rarity of cases.
Reference: Walrath and
Fraumeni (1983)
Population: 1,132 deceased white
male embalmers licensed to practice
during 1902-1980 in New York who
died during 1925-1980 identified
from registration files. Death
certificates obtained for 75% of
potential study subjects (n = 1,678).
Outcome definition: Laryngeal cancer
listed as cause of death on death
certificates.
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/ (Potential bias toward the
null;
Low sensitivity)
SB: Potential selection bias: due to
incomplete death certificate
ascertainment.
IB: Exposure Group A; latency not
evaluated
Oth: Low power due to rarity of cases.
Exposures
Duration and timing:
Occupational exposure preceding
death during 1902-1980. Median year
of birth was 1901. Median year of
initial license was 1931. Median age at
death was 1968. Expected median
duration of exposure was 37 years.
Duration and timing since first
exposure were not evaluated.
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc,
and ionizing radiation.
Anatomists may also be coexposed to
stains, benzene, toluene xylene, stains,
chlorinated hydrocarbons, dioxane,
and osmium tetroxide.
Results: effect estimate (95% CI)
[# of cases]
PMR = 0.50 (0.10-1.94)+ [2]
+EPA derived CIs using the Mid-P Method
(See Rothman and Boice. 1979)
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Anatomists may also be coexposed to
stains, benzene, toluene, xylene,
stains, chlorinated hydrocarbons,
dioxane, and osmium tetroxide.
Radiation exposure likely to be poorly
correlated with formaldehyde.
Benzene is not associated with URT
cancer.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
External comparisons:
Observed: 2
Expected: 3.4
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9:
This document is a draft for review purposes only and does not constitute Agency policy.
1-285 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Respiratory Tract Cancers in Animal Studies
This section covers precancerous lesions (i.e., dysplasia) and neoplasms (tumors) of the
respiratory tract in animal experiments, with most of the available studies focusing on the
development of squamous cell carcinomas (SCCs) in the nasal cavity. Considering the long duration
necessary for the development of these cancers, the evidence tables of the experimental animal
studies are organized by study duration, specifically focusing on chronic exposure (>1 year) and
subchronic exposure (>3 months) with long-term follow-up (typically assessed after >1 year).
These studies are further organized by study confidence and species in Table 1-36.
Animal studies investigating formaldehyde-induced respiratory carcinogenesis were
carried out primarily in rats and to a lesser extent in mice, hamsters, and nonhuman primates.
While the most consistent evidence of formaldehyde-induced respiratory cancers in animals is
restricted to the nasal cavity and consists primarily of squamous cell carcinomas (SCCs), other
neoplasms that have been observed include carcinomas other than SCCs, sarcomas, papillomas, and
adenomas (Kamataetal.. 1997: Monticello etal.. 1996: Morgan et al.. 1986b: Sellakumar et al..
1985: Kerns etal.. 19831. Nasal tumors are rare in both mice and rats fBrown. 19901. thus any
consistent increase in incidence is notable. Although dysplastic lesions, as well as hyperplasia and
squamous metaplasia (see Section 1.2.4), have been observed posterior to the nasal cavity,
respiratory tract tumors in these regions have not been reported to be significantly increased by
formaldehyde treatment In chronic studies in rats, carcinogenic effects generally first occur
around 12 months in high exposure groups, with increased tumor incidence and decreased latency
correlating with increasing exposure concentrations. Two subchronic studies with an extended
period of observation also reported an increase in tumor incidence.
Although the bioassays in mice, hamsters, and rats represent similar exposure
concentrations and duration of exposure, clear species differences in the severity of lesions are
present. Hamsters display little histopathological change whereas rats exhibit gross toxicity and
even increased mortality. Mice exhibit a range of effects on the respiratory epithelium, but not to
the severity observed in rats. There are significant species differences in the anatomical structure
of the airways, and in oral/nasal breathing patterns, including reflex bradypnea (see Appendix A.3
for discussion), all of which may influence areas of formaldehyde absorption or flux into the tissue.
The differential toxicity of formaldehyde on the URT in animals may also be due to localized
differences in mucus flow and production, as well as differences in the expression or distribution of
enzymes involved in formaldehyde detoxification. Overall, as discussed below, inhalation exposure
to formaldehyde in experimental animals induces nasal cancer and dysplasia with increasing
incidence as a function of exposure duration and concentration at the POE.
This document is a draft for review purposes only and does not constitute Agency policy.
1-286 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Methodological issues considered in evaluation of studies
This section describes histopathological evidence reporting the induction of carcinomas,
other neoplasms, and dysplasia in the respiratory tract of experimental animals after formaldehyde
exposure. The discussion emphasizes observations of malignant tumors (e.g., adenocarcinomas
and carcinomas and squamous cell carcinomas (SCCs), which were those most commonly
observed) as representing the most advanced stage of rodent tumor malignancy. Other neoplasms
were reported in the database, including adenomas and papillomas. While these neoplasms also
represent abnormal changes to the respiratory tissue, the use of benign lesions to characterize
potential human cancer risk is more straightforward when chemical-specific data are available to
associate such lesions with the development of more malignant lesions along relevant progression
pathways. For example, while squamous cell papillomas are benign lesions that could progress to
become malignant SCCs in various rodent tissues, this progression through a benign papillomatous
stage may not occur in rat nasal passages, whereas SCCs may arise directly from hyperplastic or
dysplastic tissue (McConnell etal.. 19861. Conversely, nasal polypoid adenomas (representing a
different cellular lineage from those developing into SCCs) may progress to adenocarcinomas,
which represent the more advanced stage in this cancer continuum. While benign and malignant
rodent tumors are considered neoplasms, dysplasia is an example of a dedicated, preneoplastic
lesion which may progress to neoplasia, and is therefore informative to the potential for human
carcinogenesis. However, dysplasia itself is not cancer per se, but simply one possible stage along
the presumed continuum of progressive changes characteristic of epithelial carcinogenesis. Thus,
this section prioritizes discussion of incidence data for malignant tumors, representing the most
advanced and rare lesions relevant to informing human cancer hazard; discussion of other
neoplasms or dysplasia is presented separately, as supporting evidence.
This section describes the incidence, location, and severity of these lesions. Although,
generally, the study authors cited in this section did not provide statistical comparisons for the
reported lesions data, given the rarity of these neoplasms in unexposed animals (SCCs in
particular), any observations of malignant tumors in the respiratory tract are considered to be
biologically relevant, abnormal changes. Potential relationships between lesions or the potential
for progression of benign lesions to malignant tumors are presented in the MOA discussion that
follows. Other respiratory tract lesions that may be relevant to cancer development include
hyperplasia and squamous metaplasia, which were discussed in Section 1.2.4.
All subchronic or chronic studies (and an 8-week exposure study in potentially vulnerable
mice) in experimental animals that included histopathological evaluations of respiratory tract
tissues (i.e., nose/nasal cavity, larynx, trachea, lung) were considered and evaluated (see
Appendix A.5.9), noting that evaluations of the pharynx or mouth were uncommon in these studies,
probably because experimental rodents are obligate nose-breathers). Histopathological
evaluations used standard cross-section levels of the nasal passages that paralleled the evaluations
of respiratory tract pathology described in Section 1.2.4 (see Figure 1-14 for example cross-section
This document is a draft for review purposes only and does not constitute Agency policy.
1-287 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
levels). This section focuses on studies of high and medium confidence. Studies interpreted with
low confidence for these particular endpoints are briefly summarized, but excluded from the
evidence tables: This includes all subchronic exposure studies that did not include a follow-up
period to allow for the development of respiratory tract cancers, such that the total experimental
duration from first exposure to terminal sacrifice was >12 months (24 months of observation is
preferred).
Synthesis of respiratory tract cancer in animals
Squamous cell carcinomas
Squamous cell carcinomas (SCCs) are the most consistently observed respiratory tract
cancer in mice and rats exposed to formaldehyde. These malignant tumors likely arise from
squamous cells, a type of differentiated epithelial cell that also comprises the majority of the
epidermis ("skin" cells). Formaldehyde-induced SCCs are restricted to the nasal cavity and have not
been observed in any other region of the respiratory tract The most useful and abundant SCC data
(i.e., the large majority of studies interpreted with medium or high confidence) are from studies of
exposed rats. Following exposure of rats to formaldehyde for 2 years, an increase in SCCs was
observed in 5 of 6 studies (see Table 1-36 and Figure 1-24). These tumors were detected in
exposed male and female Fischer 344 (F344) and Sprague Dawley rats, but findings in Wistar rats
were less clear (see discussion below). Overall, SCCs were not reproducibly detected below 6
mg/m3 formaldehyde in rats; however, none of the available rat studies tested exposure between 3
and 6 mg/m3, introducing uncertainty. Reflecting the rarity of these tumors [rat background
incidence averages <0.3% fBrown etal.. 19911], the incidence in control groups across the chronic
formaldehyde exposure studies in rats was 0%. Generally, the incidence increased to around 1% at
approximately 7 mg/m3 formaldehyde, and further increased to around 40% as formaldehyde
concentrations neared 18 mg/m3 (Note that for purpose of comparison across studies, Table 1-36
reports incidence rates unadjusted for mortality; see Section 2.2.1 for mortality-adjusted rates.
Unadjusted rates are generally underestimates; for example, the adjusted cumulative incidence rate
in female rats exposed for 24 months at 17.6 mg/m3 by Kerns et al. (1983) was reported at 87%).
The data as reported in Kerns et al. (1983) and Monticello et al. (1996) were corrected in a
memorandum issued by the CUT Centers for Health Research, which had sponsored or conducted
these studies fBermudez. 20041. The corrected data are noted in separate rows in Table 1-36. The
correction for Kerns et al. (1983) in the CUT memo (2004) indicates the number of animals
examined instead of the number of animals in the experiment. The corrections for Monticello et al.
(1996) issued in the CUT memo (2004) arise from an examination by CUT scientists of tissues for
an additional group of 94 rats from the study that had not been previously examined (as explained
This document is a draft for review purposes only and does not constitute Agency policy.
1-288 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
inConollv etal.. 2 0 0 3).23 These tissues were from the 12-, 18-, and 24-month time points and were
distributed approximately evenly across the six exposure concentrations.
23Conolly et al. (2003) modeled the dose-response for squamous cell carcinoma (SCC) data by combining the
data from Kerns et al. (1983) and Monticello et al. (1996) and the data from these 94 rats. The individual
animal data pertaining to the combined data are reported in the Appendix in Conolly et al. (2003). EPA's
dose-response analysis used the combined data.
This document is a draft for review purposes only and does not constitute Agency policy.
1-289 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-36. Squamous cell carcinoma (SCC) incidence in rats3 exposed to
formaldehyde for £2 years
Strain
Sex
Formaldehyde concentration rangeb (specific mg/m3 examined)
0
0 < x < 3
3 < x < 6
6 < x < 9
9 < x < 12
12 < x < 15
15 > x >
18.5
High confidence
Kerns et
al. (1983)
F344
M
0/118
0/118 (2.5C)
—
1/119
(6.9)
—
—
51/117 (17.6)
F
0/114
0/118 (2.5)
—
1/116
(6.9)
—
—
52/115 (17.6)
Corrected
Bermudez (2004)
M and F
0/237
0/239
—
2/235
—
—
83/225
Monticell
o et al.
(1996)
F344
M
0/90
0/90 (0.9);
0/90 (2.5)
—
1/90 (7.4)
—
20/90(12.2)
69/147 (18.4)
Corrected
Bermudez (2004)
M and F
0/104
0/221
—
1/108
22/103
79/161
Wouters
en et al.
(1989)
Wistar
M
0/26
1/26 (0.1);
1/28 (1.2)
—
—
—
1/26 (12.1)
—
Medium confidence
Holmstro
m et al.
(1989c)
Sprague
Dawley
F
0/15
—
—
—
—
—
1/16 (15.3)
Kamata
et al.
(1997)
F344
M
0/32
0/32 (0.4);
0/32 (2.7)
—
—
—
—
13/32 (18.3)
Sellakum
ar et al.
(1985)
Sprague
Dawley
M
0/99
—
—
—
—
—
38/100 (18.2)
Formaldehyde range (mg/m3)
0
0 < x < 3
3 < x < 6
6 < x < 9
9 < x < 12
12 < x < 15
15 > x > 18.5
Total rats examined
Range of percentage
incidenced/study
494
0-0%
534
0-3.8% e
0
325
0.8-1.1%
0
116
3.8-22.2%
527
6.3-46.9%
F344: Fischer 344; M: Male; F: Female; — Concentrations in this range were not examined.
aThis table is restricted to experimental studies in rats, given toxicokinetic differences across species. A mouse (Kerns et al.,
1983) and hamster (Dalbev, 1982) study also meet confidence and exposure duration criteria.
This document is a draft for review purposes only and does not constitute Agency policy.
1-290 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
bThese ranges were arbitrarily chosen to cover the available data and do not have a biological basis.
cThe specific concentration(s) of formaldehyde tested in the study is in parentheses,
incidence rates are unadjusted for mortality.
eBoth SCCs in this concentration range are from Woutersen et al. (1989). which did not observe any increases in SCCs at much
higher formaldehyde concentrations in Wistar rats, reducing confidence in these findings.
"O
<~>
CD
U
4—'
U
i/>
to
3
*+-
o
u
•M
C
HZ
CD
T>
ro
¦M
&-
U
o
c
£
\o
ON
¦M
o
£
5CH
45'
40'
35'
30'
£ 25'
20'
15'
10'
A
V
~
A
~
~
Kerns et al., 1983
(male F344; high)
Kerns et al., 1983
(female F344; high)
Monticello et al., 1996
(male F344; high)
Woutersen et al., 1989
(male Wistar; high)
Kamata et al., 1997
(male F344; medium)
Sellakumar et al., 1985
(male SD; medium)
Holmstrom et al., 1989
(female SD; medium)
JA.
A
A
~
~
5 10 15
Formaldehyde concentration (mg/m3)
20
Figure 1-24. Nasal SCCs in rats exposed to formaldehyde for at least 2 years.
Incidence data for squamous cell carcinomas from the high and medium (unfilled shapes) confidence
studies evaluating formaldehyde exposures of at least 2 years.
1 The data suggest that rats of different strains may vary in their sensitivity to
2 formaldehyde-induced SCCs. The only rat study with 2 years of formaldehyde exposure that did not
3 observe an association of SCCs with increasing formaldehyde exposure was conducted in Wistar
4 rats (Woutersen etal.. 19891. Although the authors reported a single SCC in each of the treatment
5 groups (no SCCs were observed in controls), these tumors may not have been related to
This document is a draft for review purposes only and does not constitute Agency policy,
1-291' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
formaldehyde exposure as the incidence did not change at higher formaldehyde levels and
observations of SCCs occurred at far lower concentrations than in any other rat studies. Consistent
with this potential resistance of Wistar rats to formaldehyde-induced SCCs observed by Woutersen
et al. (1989), an earlier study from the same laboratory examining Wistar rats at identical
formaldehyde concentrations did not detect any SCCs fAppelman et al.. 19881: however, the earlier
study only exposed and observed animals for 12 months, substantially reducing its ability to detect
cancers. Two additional experiments from the same laboratory examined whether subchronic
formaldehyde exposure with follow-up for more than 2 years resulted in SCCs in Wistar rats
fWoutersen et al.. 1989: Feron etal.. 19881. Both of these studies observed a single SCC induced in
response to formaldehyde exposure at approximately 11 mg/m3, with an increased incidence of
formaldehyde-induced SCCs to 3 of 44 in the study that tested a higher exposure of 24.4 mg/m3
fFeron etal.. 19881. The <4% incidence in Wistar rats exposed to approximately 11 mg/m3 in these
studies contrasts with the 22% incidence observed at this level in F344 rats by Monticello et al.
(19961. Taken together, although some of the data with a sufficient duration of observation suggest
that formaldehyde exposure can induce a low incidence of SCCs in Wistar rats (Woutersen etal..
1989: Feron etal.. 19881. these findings indicate that this strain may be resistant to
formaldehyde-induced nasal SCCs, as compared to F344 and Sprague Dawley rats.
The effects of long-term formaldehyde exposure in species other than rats are less well
studied, but the available data suggest that rats may be the most sensitive laboratory rodents. The
only mouse study testing exposure of at least 2 years (Kerns etal.. 1983) provided support for the
consistent observations of SCCs in formaldehyde-exposed rats. In this well-conducted (i.e., high
confidence) study, SCCs were observed at 17.6 mg/m3, but not at 6.9 or 2.5 mg/m3 (incidence in
controls was 0%). The incidence at 17.6 mg/m3 was <2% (2/120), in contrast with the >40%
incidence detected in F344 rats exposed to similar formaldehyde concentrations by the same study
authors fKerns etal.. 19831. The authors also reported that the SCCs in rats were more invasive
and severe than those observed in mice. These differences could reflect the use of a mouse strain
that might be insensitive to these effects, similar to the above discussion of Wistar rats, but the
differences more likely reflect a decreased response due to a lower inhaled dose of formaldehyde
resulting from differences in breathing patterns and irritant responses across species (see
Appendices A2 and A3). In contrast, no respiratory tract tumors were observed in Syrian golden
hamsters exposed to 12.3 mg/m3 of formaldehyde for a lifetime fDalbev. 19821. although no other
exposure levels were tested to inform whether this species or strain may also be less sensitive than
exposed F344 and Sprague Dawley rats, and exposed mice.
In rats and mice, SCC formation appears to be dependent on both the formaldehyde
concentration and the duration of exposure and observation. Specifically, higher formaldehyde
exposure levels tend to be associated with both an increased incidence and an earlier onset of
tumor formation. An example of this was demonstrated in a follow-up to the Kerns et al. (1983)
study by Monticello et al. (1996). Monticello et al. (1996) reported that the incidence of SCCs in rats
This document is a draft for review purposes only and does not constitute Agency policy.
1-292 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
exposed to 18.4 mg/m3 formaldehyde was 47%, with the first tumor noted at 12 months. The
incidence of SCCs in the 12.2 mg/m3 exposure group was lower, at 22%, and the tumor latency was
longer, with the first SCC observed at 18 months. Of the 90 rats exposed at 7.4 mg/m3 for
20 months, only one SCC was noted, and no SCCs were detected at 0, 0.85, or 2.52 mg/m3 over
28 months fMonticello etal.. 19961. Initial observations of SCCs varied across the available rat
studies, and the study design sometimes prevented an accurate determination of the timing
(e.g., microscopic examinations may have been conducted every 6 months, every year, or only after
2 years). However, the first tumor generally was not observed before 12 months of observation,
and often took 16 months or longer to develop (see Table 1-37). Consistent with this long latency,
SCCs observed in mice took 2 years to develop fKerns etal.. 19831. and no URT neoplasms were
observed during 8 months of observation in a short-term, low confidence (i.e., due to its 8-week
exposure duration and <1 year follow-up) study of potentially sensitive mice fMorgan etal.. 20171.
In light of these observations, subchronic and shorter-term exposure studies without a long
duration of follow-up are not expected to be capable of detecting formaldehyde-induced SCCs. In
studies where interim sacrifices were performed and described, longer durations of exposure were
generally associated with an increased incidence, severity, and sometimes more posterior location,
of the induced SCCs fMonticello etal.. 1996: Kerns etal.. 19831. These data suggest that longer
formaldehyde exposure duration is correlated with a greater incidence and severity of SCCs.24
The large bioassay of Kerns et al. (1983) in F344 rats showed no overt differences in the
development of SCCs across sexes (i.e., 51/117 in males vs. 52/117 in females at 17.6 mg/m3).
There is some evidence to suggest that male rodents may be more sensitive to these effects. For
example, only 1 of 16 female Sprague Dawley rats exposed to 15.3 mg/m3 developed SCCs
fHolmstrom etal.. 1989bl. whereas slightly higher levels (18.2 mg/m3) of formaldehyde in another
study of male Sprague Dawley rats fSellakumar et al.. 19851 induced more than six times as many
SCCs (38/100). In addition, only male mice (2/120), but not female mice (0/120), developed SCCs
in a chronic study (Kerns etal.. 1983). However, these suggestions of differential sensitivity
between sexes are not easily interpreted given the small sample sizes (Holmstrom et al.. 1989b)
and a low incidence of SCCs in exposed mice fKerns etal.. 1983).
The locations of the induced SCCs were consistent with both the distribution of inhaled
formaldehyde and locations of other formaldehyde-induced nasal pathologies (see Section 1.2.4),
with SCCs arising from the epithelium lining the airway and not from the underlying glandular
epithelium. These tumors were most commonly observed in anterior regions of the nasal cavity,
although higher exposure levels sometimes resulted in progression of SCCs to more posterior
locations. Morgan et al. (1986b) mapped the location of formaldehyde-induced SCCs from the
24While some data exist to suggest that SCCs can be induced following subchronic formaldehyde exposure
when observations continue for more than 2 years fWoutersen et al.. 1989: Feron et al.. 19881. definitive
experiments in rats that are sensitive to the development of SCCs have not been performed (e.g., comparing
SCC incidence in Sprague Dawley or F344 rats exposed for shorter durations and followed up for >2 years
versus rats exposed to the same concentrations for >2 years with no additional follow up).
This document is a draft for review purposes only and does not constitute Agency policy.
1-293 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Kerns et al. (19831 study. In F344 rats, the majority of animals had single tumors, with a little
under 20% of each sex with tumors developing multiple neoplasms. More than half (57%) of the
SCCs occurred on the lateral side of the nasoturbinate and adjacent lateral wall at the front of the
nose (Levels I and II; see Table 1-37); approximately 25% were located on the midventral nasal
septum (Levels II and III); and about 10% were on the dorsal septum and roof of the dorsal meatus
(Levels I, II, and III). A small number (3%) were found on the maxilloturbinate (Levels II and III),
which only involved the medial aspect Similar observations were reported for other studies of
F344 rats (Monticello etal.. 1996) and B6C3F1 mice (Kerns etal.. 1983). Locations of SCCs in
Sprague Dawley and Wistar rats were not as specifically reported in the available studies, but were
generally similar, primarily affecting the respiratory epithelium lining the septum and
nasoturbinates fWoutersen et al.. 1989: Sellakumar etal.. 19851.
Other malignant neoplasms
Although the data on other neoplasms are far less robust than those related to SCCs,
formaldehyde inhalation also appears to induce other types of malignant nasal tumors. The
incidence of these other neoplasms was typically only one, or rarely two, animals in an exposed
group (never in controls); however, it is considered highly unlikely that these are incidental, as
these rare neoplasms only developed after exposure to the highest formaldehyde concentrations,
typically those above 17 mg/m3 (see Table 1-37). As with SCCs, these neoplasms were limited to
the nasal cavity. Carcinomas, which derive from epithelial tissues, were reported in several studies
with an observation period greater than 2 years, consistent with the pronounced effect of inhaled
formaldehyde on the nasal epithelium. A single nasal carcinoma was observed in both male and
female F344 rats fKerns etal.. 19831. a mixed carcinoma was observed in male Sprague Dawley rats
(Sellakumar etal.. 1985). and a carcinoma in situ was observed in male Wistar rats exposed to
24 mg/m3 fFeron etal.. 19881. but not <12.1 mg/m3 fWoutersen et al.. 1989: Appelman etal.. 1988:
Feron etal.. 1988). failed to develop any of these other malignant tumors.
Nonmalignant neoplasms
Several other benign tumors of the respiratory tract have been reported following
formaldehyde exposure in rats, but not in other species. These tumors parallel findings for the
other observed tumors, in that they are restricted to the nasal cavity and generally take more than
12 months to develop. Overall, these tumors appear to represent an erratic growth of the nasal
epithelial tissue (i.e., adenomas and papillomas), with the exception being an ameloblastoma
observed at 24 mg/m3 formaldehyde (Feron etal.. 1988). a tumor that presumably secondarily
infiltrated the nasal cavity. In male Sprague Dawley rats, 10% of animals (10/100) exposed to
18.2 mg/m3 for their lifetime developed nasal polyps or papillomas (Sellakumar etal.. 1985: Albert
etal.. 1982). while approximately the same percentage of male F344 rats (3/3 2) exposed to a near-
identical formaldehyde concentration (18.3 mg/m3) developed squamous cell papillomas (Kamata
etal.. 19971. Polypoid adenomas have also been consistently observed in response to
This document is a draft for review purposes only and does not constitute Agency policy.
1-294 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
formaldehyde exposure. Similar to SCCs, and in contrast to the other malignant tumors discussed
above, these neoplasms may be inducible at formaldehyde concentrations below 12 mg/m3, and
perhaps even below 7 mg/m3, although the data are somewhat more variable as compared to the
SCC data (see Table 1-37). Polypoid adenomas were increased compared to controls in male Wistar
rats exposed to 11.3 mg/m3 fWoutersen etal.. 19891 or 24.2 mg/m3 fFeron etal.. 19881 for
3 months with follow-up to >2 years, and in chronically exposed F344 rats (Monticello etal.. 1996:
Kerns etal.. 19831. The responses in F344 rats occurred primarily in males and were reported at
concentrations as low as 2.5 mg/m3 (Kerns etal.. 19831. although interpretation of the incidence
data across exposure levels is not straightforward. Taken together, the data indicate that benign
epithelial tumors in the nasal cavity can be induced by formaldehyde exposure.
Dysplasia
Similar to observations of nasal tumors, the incidence of dysplasia in long-term
formaldehyde inhalation studies in rats and mice (i.e., chronic or subchronic exposure with
observation periods of >12 months) increased in severity and occurred in more distal portions of
the nasal cavity with both formaldehyde concentration and duration. Whereas the rat nasal tumor
data consistently demonstrated that tumors are restricted to the nasal cavity, one study reported
that F344 rats (which appear to be sensitive to these effects) also exhibited mild dysplasia in the
trachea fKerns etal.. 19831. although the tracheal lesions were not observed when rats exposed for
2 years were left unexposed for 3 months. The study authors did not observe any tracheal lesions
in mice (Kerns etal.. 19831. Epithelial dysplasia of the nasal cavity was first noted at 12 months in
rats exposed to concentrations as low as 2.5 mg/m3, and in a "few" mice after 18 or 24 months of
exposure at concentrations as low as 6.9 mg/m3 formaldehyde fKerns et al.. 19831. However, after
24 months of exposure to 17.6 mg/m3 formaldehyde, the incidence of nasal dysplasia was
significantly increased in rats and mice, with greater than 90% of mice exhibiting this lesion fKerns
etal.. 1983). The study authors noted that the identification of dysplasia in this study may have
been termed metaplasia or hyperplasia by other study authors (Kerns etal.. 1983). suggesting that
this may represent a sensitive estimate of dysplasia. In another study, a female Sprague Dawley rat
exposed to 15.3 mg/m3 formaldehyde for a lifetime also developed dysplasia of the nasal
epithelium fHolmstrom etal.. 1989bl. In line with the nasal tumor data, studies of Wistar rats and
hamsters did not identify dysplastic lesions (see Table 1-37).
Conclusions
Tumors of the respiratory tract (predominantly SCCs but including other epithelial and
nonepithelial tumors) were consistently observed in mice and several strains of rats, but not in
hamsters, exposed to formaldehyde concentrations above approximately 6-7 mg/m3.
Precancerous dysplastic lesions were induced in rats and mice, sometimes at lower formaldehyde
concentrations than those at which malignant tumors were observed. The dysplasia and neoplasms
were predominantly localized to anterior regions of the nasal respiratory epithelium, although the
This document is a draft for review purposes only and does not constitute Agency policy.
1-295 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
lesions progressed to more posterior locations with increasing duration and concentration of
formaldehyde exposure, with one study reporting that dysplasia can develop in portions of the
proximal trachea in rats (note: all tumors were limited to the nasal cavity). These lesions were
never observed in other respiratory tract regions, such as the larynx and lung, and they generally
only developed in animals that were observed for longer than 12 months. Studies of subchronic
formaldehyde exposure without follow-up consistently failed to observe dysplasia or neoplasms in
the nose, trachea, larynx, or lungs across a range of formaldehyde concentrations in rats (Wilmeret
al.. 1989: Appelman et al.. 1988: Feronetal.. 1988: Zwartetal.. 1988: Woutersen etal.. 1987: Rusch
etal.. 19831 and mice fMaronpotetal.. 19861. and at lower formaldehyde levels (<3.65 mg/m3) in
hamsters and cynomolgus monkeys fRusch etal.. 19831. Studies with a long observation period
were not identified to inform the possibility of cancer development in nonhuman primates exposed
to formaldehyde. The development of these lesions, particularly SCCs, depended on the duration of
observation, and based on an increasing incidence and severity of lesions in animals exposed for
longer periods of time, the formaldehyde exposure duration. Most notably, the lesion incidence, as
well as the tumor invasiveness and latency, was reproducibly shown to worsen with increasing
formaldehyde exposure level.
Table 1-37. Respiratory tract cancer—chronic and subchronic (with long-term
follow up) exposure in rats, mice, and hamsters
Reference and study design3
Results
Chronic exposure
High confidence
Rats
Kerns et al. (1983)
Malignant tumors
Rats: F344; males and females; 119 to
121/sex/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for up to 2 yr
(recovery: 27 and 30 months) at 0, 2.5, 6.9,
or 17.6 mg/m3
Histopathologyb\ 5 sections of nasal
turbinates (Levels l-V) for animals that died
or at interval sacrifices (i.e., at months 6,12,
18, 24, 27, and 30)
Related study/earlier reports: Battelle
(1982, 1981); [interim findings presented
in Swenberg et al. (1980b)l
Note: viral infection reported
(sialodacryoadenitis) at approximately
weeks 52-53 (Kerns et al., 1983); the
authors attributed transient decreases in
body weight to this infection. This infection
was not interpreted to affect the reliability
of the cancer incidence data, in part
mg/m3
0
2.5
6.9
17.6
Squamous cell carcinoma 0
Male
0/118
0/118
1/119
51/117
Female
0/114
0/118
1/116
52/115
Nasal carcinoma
Male
0/118
0/118
0/119
l/117b
Female
0/114
0/118
0/116
1/115
Carcinosarcoma
Male
0/118
0/118
0/119
1/117
Female
0/114
0/118
0/116
0/115
Undifferentiated carcinoma or sarcoma
Male
0/118
0/118
0/119
2/117b
Female
0/114
0/118
0/116
0/115
Other Neoplasms
Polypoid adenoma
Male
1/118
4/118
6/119
4/117
Female
0/114
4/118
0/116
1/115
This document is a draft for review purposes only and does not constitute Agency policy.
1-296 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results
because dysplasia and other lesions were
already present at 12 months (when the
infection began)
Epithelial Dysplasia
6 months
_C
-
-
_d
12 months
_C
Level 10
18 months
_c
NR
Level l-lllf
Level l-V*
24 months
_c
Level 1
aSCCs became clinically observable in females at ~12 months, and in males at ~14
months; most appeared to originate in the nasoturbinates
bA rat in this group also had SCC
cLesion frequency (dysplasia or metaplasia) of <15% at 0 mg/m3 (Level I)
d Although formaldehyde-induced lesions were identified in Level l-lll, the
authors did not specify them as dysplasia
eSquamoid epithelial lining several cells thick with polarity changed from vertical
to horizontal was noted and termed dysplasia, but authors acknowledged related
changes can be termed hyperplasia or metaplasia
'Dysplasia was most intense in Level I. Exposure-related effects were observed in
Levels l-lll and l-V at 6.9- and 17.6-mg/m3, respectively, although the specific
timing for these lesions was not provided; note: dysplasia was consistently
detected earlier than squamous metaplasia
Trachea: at 17.6 mg/m3, minimal-to-mild dysplasia at 18 months, with greater
frequency (p < 0.05) in 24-month and unscheduled deaths groups; trachea
lesions not observed in postexposure group or at lower levels
Monticello et al. (1996)
Rats: F344; male; 90-147/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for up to
24 months at 0, 0.85, 2.52, 7.40,12.2, or
18.4 mg/m3
Histopathologyb\ 6 sections of the nasal
cavity
Malignant tumors in the nasal cavity"
Squamous
carcinoma6
cell
Adenocarcinoma
Rhabdomyosarcoma
0, 0.85, or
2.52 mg/m3
0/90
0/90
0/90
7.4 mg/m3
1/90
(1%)
0/90
0/90
12.2 mg/m3
20/90
(22%)
1/90
1/90
18.4 mg/m:
69/147
(47%)
1/147
1/147
Other neoplasms
Polypoid adenoma
0/90
0/90
5/90
(6%)
14/147
(10%)
Spontaneous buccal SCCs were observed at 0, 2.52, and 18.4 mg/m3
bSCCs that could be localized were identified most often in the anterior or
posterior lateral meatus 1/90,12/90,17/147 or 0/90,12/90, 9/147
corresponding to 7.4,12.2, and 18.4 mg/m3); SCCs were also observed in the
mid- and dorsal septum, as well as the maxilloturbinates, but only at 18.4
mg/m3; however, most tumors were too large to localize and these often
eroded through nasal bone and invaded the subcutis of the dermis. Tumors
began appearing ~1 yr at 18.4 mg/m3 and ~1.5 yr at 12.2 mg/m3
No tumors observed beyond the respiratory tract
Sellakumar et al. (1985)
Rats: Sprague Dawley; male; 99-100/group
Test article: Paraformaldehyde (slurry in
paraffin oil)
Exposure: 6 hr/d, 5 d/wk for lifetime at 0 or
18.2 mg/m3 [Note: prior reporting of levels
during first 588 days at 17.5 mg/m3
(Albert et al.. 1982)1
Histopathologyb\ multiple sections of the
head (from just behind the nostril to the eye
orbits), lung, trachea, and larynx
Related study. Albert et al. (1982)
Colony Control
Air sham
18.2 mg/m3
Malignant tumors in the nasal mucosa
Squamous cell carcinoma3
0/100
0/99b
38/100
Adenocarcinoma
0/100
0/99
0/100
Mixed carcinoma
0/100
0/99
1/100
Fibrosarcoma
0/100
0/99
1/100
Other neoplasms in the nasal mucosa
Polyp or papillomas | 0/100 | 0/99 | 10/100
Predominantly moderate/well differentiated, keratin obstructed lumen;
latency to tumor formation was approximately 603-645 days
No tumors observed in the trachea or lungs
This document is a draft for review purposes only and does not constitute Agency policy.
1-297 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results
Woutersen et al. (1989)
Rats: Wistar; male; 30/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for 28 months at 0,
0.1,1.2, or 12.1 mg/m3
Histopathologyb\ 6 nasal cross sections
Note: experiments with nasal damage prior
to exposure are not presented here
Malignant tumors
0 mg/m3
0.1 mg/m3
1.2 mg/m3
12.1 mg/m3
Squamous cell
carcinoma
0/26
1/26
1/28
1/26
Adenosquamous
carcinoma
0/26
0/26
0/28
0/26
Adenocarcinoma
0/26
0/26
0/28
0/26
Note: the specific locations of these tumors was not described
Mice
Kerns et al. (1983)
Mice: B6C3F1; males and females;
119 to 121/sex/group
Exposure: 6 hr/d, 5 d/wk for up to
24 months (recovery at 27 and 30 months)
at 0, 2.5, 6.9, or 17.6 mg/m3
Test article: Paraformaldehyde
Histopathologyb\ 3 sections of nasal
turbinates, defined as Levels II, III, and V for
all animals that died or were sacrificed at
scheduled intervals (i.e., at month 6,12,18,
24, 27, and 30)
Earlier reports: Battelle (1982, 1981)
Main limitations: Lesion incidence NR; only
three nasal sections examined
Malignant tumors
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
SCCs at
24 months3
0/~120
(both sexes)
0/~120
(both sexes)
0/~120
(both sexes)
2/~120 male
0/~120 female
Dysplasia
b
12 months
-
-
-
-
18 months
-
-
Level II: "few"
Level II (~90%)
24 months
-
-
Level II: "few"
>90%
Recovery
(27 months)
-
-
none
yes (incidence
and level NR)
aSCCs were not observed prior to 24 months (p > 0.05); both SCCs originated
from nasoturbinates; the number of mice evaluated was not specified, but
assumed ~120 based on 119-121 mice/group
bUnless noted, exact frequency of lesion NR, and sex not specified
No tracheal lesions were observed
Medium confidence
Rats
Appelman et al. (1988)
Rats: SPF Wistar; male; 10/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for 52 weeks at
0.12,1.2, or 12.1 mg/m3
Histopathologyb\ nose (6 standard cross
levels), larynx, trachea, and lungs
Note: experiments with nasal damage prior
to exposure are not presented here
Main limitations: 1-year short duration to
allow for cancer development
No dysplasia or nasal neoplasms were observed in nose, larynx, trachea, or lungs
with exposure up to 12.1 mg/m3 for up to 1 year (assumed, based on
histopathological evaluation of these tissues, although the study authors did not
specifically state these conclusions)
Holmstrom et al. (1989b)
Rats: Sprague Dawley; female; 15-16/group
Test articles: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for 104 weeks at 0
or 15.3 mg/m3
Histopathologyb\ 5 sections of the nose
from the vestibulum to the posterior
ethmoturbinatic region, and the lungs
Note: data on wood dust combined with
formaldehyde exposure not evaluated
Malignant tumors
Air control
15.3 mg/m3
Squamous Cell Carcinoma
0/15
l/16a
Dysplasia
0/15
l/16b
aObserved after 21 months after exposure
bAn addition two rats exhibited pronounced squamous metaplasia with
keratinization (7 more exhibited squamous metaplasia)
Note: Mortality was similar in both groups
This document is a draft for review purposes only and does not constitute Agency policy.
1-298 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results
Main limitations; Limited reporting; some
health issues noted
Kamata et al. (1997)
Months (interim sac.)
Dead
All
Rats: F344; male; 32/group
Test article: Formalin (methanol control)
Exposure: nose-only 6 hr/d, 5 d/wk for up to
28 months at 0, 0.40, 2.67, or 18.27 mg/m3
(methanol—0,18.27 mg/m3 groups,
estimated at 5.5 mg/m3, presumed from
percentage methanol in formalin)
Histopathologyb\ nasal region (sections
from five anatomical levels) and trachea
Main limitations: small sample size; use of
formalin (uncertainties, such as possible
differences in tissue formaldehyde due to
methanol, remain despite inclusion of a
methanol control)
12
18
24
28
Squamous cell carcinomas at 18.27 mg/m3 0
SCCs
0/5
1/5
0/2
0/0
12/20
13/32b
Other malignant tumors at 18.27 mg/m3 a
Unclassified sarcoma
0/5
0/5
0/2
0/0
0/20
0/32
Sarcoma
0/5
0/5
0/2
0/0
1/20
1/32
Other neoplasms at 18.27 mg/m3a
Squamous cell papilloma 10/5 11/5 10/2 10/0 12/20 13/32
aNo nasal tumors were observed at 0, 0.4, or 2.67 mg/m3 (note: 1 unclassified
sarcoma found in a dead room control group rat); average latency across groups
varied from 603 and 645 days
Significant at p < 0.01, compared with the 0 mg/m3 group.
Note: Most tumors were located in levels B and C (see diagram in left column);
large tumors invaded the subcutis through the nasal bones
No tumors were observed in the trachea
Hamsters
Dalbev (1982)
Hamsters: Syrian golden; male; 132
untreated controls and 88 exposed
Test article: Paraformaldehyde
Exposure: 5 hr/d, 5 d/wk for a lifetime atO or
12.3 mg/m3
Histopathologyb\ Two transverse sections of
the nasal turbinates, and sections of the
larynx, trachea, and lungs
Main limitations: minimal sampling,
histological evaluation, and reporting
Note: mixture experiment not evaluated
No tumors reported in the nose, larynx, lungs, or trachea with a lifetime of
exposure to 12.3 mg/m3
Note: study authors indicated formaldehyde exposure at 36.9 mg/m3 amplified
diethylnitrosamine-induced respiratory tumors.
Subchronic exposure with long-term follow-up
High confidence
Rats
Woutersen et al. (1989)
0 mg/m3
0.1 mg/m3
1.2 mg/m3
11.3 mg/m:
Rats: Wistar; male; 30/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for 3 months at 0,
0.1,1.2, or 11.3 mg/m3; sacrificed at
28 months
Histopathologyb\ 6 nasal cross sections
Note: short duration of exposure
Malignant tumors
Squamous cell carcinoma
0/26
0/30
0/29
1/26
Carcinoma in situ
0/26
0/30
0/29
0/26
Other neoplasms
Polypoid adenoma | 0/26
Note: cross-section locations not specifiec
0/30
0/29
1/26
Medium confidence
Rats
This document is a draft for review purposes only and does not constitute Agency policy.
1-299 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results
Feron et al. (1988)
0 mg/m3
~11.5 mg/m3
~24 mg/m3
Rats: Wistar; male; 45/group
Test article: Paraformaldehyde
Exposure: 6 hr/d, 5 d/wk for up to 13 weeks
at 0, 11.3-11.9, or 24.2-24.4 mg/m3;
sacrificed at 130 weeks
Histopathologyb\ 6 standard cross levels of
the nose.
Main limitations: Limited reporting; short
duration of exposure
Malignant tumors
Squamous cell carcinoma:
4-wk exposure
(wk sacrificed indicated)
0/44
0/44
1/45
(wk 106)
8-wk exposure
2/45
(wk 94,
130)
1/44
(wk 130)
1/43
(wk 121)
13-wk exposure
0/45
1/44
(wk 82)
3 or 4/44a
(wk 63,112,114,
NR)
Other malignant tumors with 13 wk exposure b:
Carcinoma in situ:
0/45
0/44
1/44 (wk 81)
Other neoplasms
Ameloblastoma:
0/45
0/44
1/44 (wk 73)
Polypoid adenoma:
4 wk exposure
0/44
0/44
1/45 (wk 110)
8 wk exposure
0/45
0/44
1/43 (wk 100)
13 wk exposure
al SCC was classified as a "cyst
palate, and which the authors
bcarcinomas other than SCC w«
0/45
c SCC," which
did not associ
:re not observ
0/44
may have been
ate with exposure
red with <13 wk e
0/44
lerived from the
*
xposure
Abbreviations: NR = not reported; F = Fischer; hr = hour(s); d = day(s); wk = week(s); yr = year(s).
Analytical formaldehyde levels are presented and, unless otherwise noted, whole-body exposures were used.
bThe studies used the same sectioning levels described for noncancer lesions in Section 1.2.4 (see Figure 1-14).
Evidence on Mode of Action for Upper Respiratory Tract Cancers
Formaldehyde exposure has been associated with elevated incidence of carcinomas in
human URT tissues, with the strongest evidence for nasopharyngeal and sinonasal tumor formation
(Tables 1-32 and 1-33). Formaldehyde inhalation reproducibly induces squamous cell carcinomas
(SCC) in the nasal passages of F344, Sprague Dawley, and Wistar rats (obligate nose-breathers), as
well as polypoid adenomas (PA); SCCs and PAs are both rare tumors in rats, with background
frequencies of <0.3% and <0.04%, respectively fPoteracki and Walsh. 1998: Chandra etal.. 1992:
Brown etal.. 19911. SCCs were also elevated in the anterior nasal passages of chronically exposed
B6C3Fi mice [background frequency of 0/2,818; (Brown etal.. 1991)]. but not in hamsters.
Formaldehyde-associated SCCs and PAs originate in the nasoturbinates, maxilloturbinates, or
lateral wall of the nasal cavity, and likely arise from the same target cell population (i.e., the nasal
respiratory or transitional epithelium). The neoplastic response to formaldehyde exposure in rat
nasal epithelium appears to be complex; SCC incidence is dramatically induced at exposure levels
associated with other proliferative epithelial pathology, increasing from 1% at 7 mg/m3 to 60% at
18 mg/m3 in chronically exposed F344 rats. In contrast, relatively low frequencies of PAs are
induced at concentrations ranging from 2.5 tol8 mg/m3, with PA incidence increasing moderately
to a maximum of 10% at 18 mg/m3 (see Table 1-37). SCCs and PAs are similarly induced in Sprague
This document is a draft for review purposes only and does not constitute Agency policy.
1-300 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Dawley rats, and although nasal tumor incidence may be somewhat lower in Wistar rats, studies in
the latter strain provide some evidence of tumor induction following subchronic exposure with
lifetime follow-up.
Following inhalation exposure at analogous POE tissues in humans (nasal, buccal, and
nasopharyngeal epithelium), nonhuman primates (nasal and extranasal respiratory and
transitional epithelium, larynx, trachea, and carina), and rodents (nasal respiratory and transitional
epithelium), evidence exists supporting the evaluation of a cancer mode of action (MOA). Among a
variety of influential forces, two primary mechanistic considerations appear to contribute, both
directly and indirectly, to tumorigenesis resulting from formaldehyde exposure at POE tissues:
genotoxicity-associated mutagenicity, and cytotoxicity-induced regenerative proliferation.
Furthermore, formaldehyde may stimulate nasal epithelial cell proliferation to some extent, even in
the absence of frank tissue cytotoxicity. Instead of considering independent, sequential series of
key events for each of these mechanistic considerations, evidence for genotoxicity and
mutagenicity, cellular proliferation (independent from tissue pathology), and cytotoxicity-induced
regenerative tissue proliferation is evaluated in an integrated manner, whereby hypothesized
mutagenesis and increased cellular turnover initiate and then augment URT carcinogenesis as a
function of exposure duration, periodicity, and tissue dose. This approach is consistent with the
observation that, while mitogenesis can drive rodent tumor prevalence, it may not supplant the
contribution of mutagenicity to chemically induced carcinogenesis (Ames and Gold. 1990).
Much of the available evidence relevant to these mechanistic considerations is discussed in
detail in the prior sections on URT cancer data in human and animal studies, as well as in
Sections 1.2.3 and 1.2.4, and in Appendices A.4 and A.5.6. Herein, these findings are summarized
and integrated into a proposed cancer MOA network to serve as a framework for the evidence
evaluation and MOA analysis (see Figures 1-25-1-27). The evidence is synthesized with an
emphasis placed on observations from humans and experimental animals repeatedly exposed to
formaldehyde via the inhalation route, evaluated following the Bradford Hill considerations (U.S.
EPA. 2005a). and conclusions are discussed in the context of URT carcinogenesis proceeding via
this hypothesized, integrated cancer MOA. While evidence from biochemical investigations or cells
cultured in vitro is not exhaustively described, pertinent observations are presented when useful in
providing a mechanistic interpretation to effects described in vivo, when the available in vivo
evidence is limited or nonexistent, or does not inform the effect under consideration. Only results
from studies reporting some quantitative estimate of formaldehyde exposure concentration were
synthesized, due to a general abundance of information relevant to the mechanistic considerations,
and relative paucity of studies failing to provide formaldehyde exposure estimates. Evidence
informing other modulating or modifying effects such as immune dysfunction and oxidative stress,
DNA repair inhibition, and epigenetic alterations are also discussed briefly (for more detail see
Appendices A.4 and A.5.6), while evidence for systemic genotoxicity and immune system effects
outside the URT as relevant to carcinogenesis are primarily discussed elsewhere (see Section 1.3.3,
This document is a draft for review purposes only and does not constitute Agency policy.
1-301 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Evidence on mode of action for LHP cancers). While these factors may contribute significantly at
various stages of URT carcinogenesis to the mechanistic considerations described above, the
limited available data preclude evaluating their independent contribution to the formaldehyde URT
cancer MOA. Likewise, while various aspects of this analysis may be directly relevant to
formaldehyde exposure by other routes, or cancer at other (i.e., distal) tissue locations, this
discussion is focused on cancers atPOE tissues (i.e., the URT) following inhalation exposure.
Summary of genotoxicitv and mutagenicity
This overall summary is relevant to MOA interpretations for both URT cancers (this section)
and lymphohematopoietic cancers (see Section 1.3.3). Formaldehyde is a direct-acting chemical
that has been shown to be genotoxic or mutagenic in a variety of in silico and in vitro test systems;
experimental animals including mice, rats, and monkeys; as well as in humans. Formaldehyde
exposure typically induces genotoxicity, mutagenicity, or related endpoints in a concentration- and
duration-dependent manner, including deletions and point mutations; DNA-protein and DNA-DNA
crosslinks (DPX and DDC, respectively) and DNA mono (hmDNA) adducts; clastogenic-related
effects such as micronuclei (MN) and chromosomal aberration (CA) formation, as well as sister
chromatid exchanges (SCEs), single-strand and double-strand breaks (SSBs, DSBs, respectively);
and unscheduled DNA synthesis (UDS), DNA repair inhibition, and cellular transformation. For a
comprehensive description of the evidence on formaldehyde genotoxicity, see Appendix A.4, which
includes a summary table of genotoxicity endpoints investigated across the test systems most
relevant to human inhalation exposure and, when possible, separates the results into respiratory-
versus nonrespiratory-related tissues or systems.
This evaluation emphasizes the experiments interpreted to best inform the potential for
genotoxicity in humans following inhalation exposure to formaldehyde, and therefore focuses on in
vivo studies in mammalian species. In addition, the relative importance of the specific genotoxic
endpoints was considered when prioritizing results in the synthesis of epidemiological evidence for
genotoxicity. For example, it has been shown that increased frequency of CAs and MN are
associated with increased cancer mortality, and these endpoints are considered by EPA to be highly
relevant to the assessment of genotoxicity in humans (Bonassi et al.. 2008: Bonassi etal.. 2007: U.S.
EPA. 2005a: Bonassi et al.. 2004bl SSBs and DSBs in DNA indicate genetic instability and are also
considered by EPA to be highly relevant to the assessment of genotoxicity for humans, while
increased frequencies of sister chromatid exchange (SCE) are less strongly associated with cancer
mortality (Bonassi et al.. 2004a).
Inhaled formaldehyde primarily encounters cellular macromolecules atPOE tissues,
including both nasal and buccal epithelial cells in humans, while preferentially affecting the nasal
epithelium in rodents, which are obligate nose-breathers. In these barrier tissues, formaldehyde
can interact directly with DNA, resulting in DPX and DDC, DNA mono (hmDNA) adducts, SSBs, MN,
and CAs. Furthermore, cells in the lower respiratory tract (LRT) and tissues distal to the initial
This document is a draft for review purposes only and does not constitute Agency policy.
1-302 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
point of exogenous formaldehyde exposure, such as peripheral blood lymphocytes (PBLs), are also
potential targets of formaldehyde genotoxicity.
Neither DPX nor hmDNA adduct levels have been assessed specifically in nasal or buccal
tissues from formaldehyde-exposed human workers, although occupational exposure to
formaldehyde was associated with a significant exposure- and duration-related increase in DPX
formation in PBLs. Formaldehyde-induced DPXs in the URT of rats and nonhuman primates in a
dose-responsive manner across several studies. The predominant location of DPX formation varied
due to anatomical differences in the nasal physiology and breathing patterns of primates versus
rodents; however, the distribution of DPXs in rat nasal tissues corresponded to sites of tumor
incidence, cell proliferation, and cytotoxicity. hmDNA monoadducts have been observed in the
nasal epithelium of rats and the maxilloturbinate regions of rhesus monkeys following
formaldehyde exposure, as well as in cell-free systems, and cultured cell lines including human
nasal epithelial cells.
The majority of occupational studies have associated formaldehyde exposure with
increased MN formation in human nasal or buccal epithelial cells, predominantly forming
centromere-negative micronuclei suggesting clastogenic effects. Although no MN in nasal tissues
were observed in one short-term, high-dose rodent inhalation study, MN were consistently induced
in different mammalian cells in vitro. In addition, long-term occupational exposure was associated
with significantly increased MN in PBLs, and aneugenicity appears to be the predominant effect in
peripheral tissues (see Section 1.3.3). Exposure to formaldehyde also was associated with
significantly increased CAs in PBLs of human workers, as well as in rodents from a short-term,
high-dose study. Formaldehyde also induced CAs in rat pulmonary lavage cells, as well as hamster
and primary human cells in vitro. Exposure-related increases in SSBs were observed in rat nasal
tissues in one experimental study and in several studies of PBLs from exposed workers and
rodents. Occupational exposure to formaldehyde caused increased mutant p53 protein expression
in the serum of exposed workers, while cell lines derived from formaldehyde-induced rat nasal
SCCs showed p53 mutations. Across the available database, formaldehyde consistently induces
various endpoints consistent with mutagenicity, such as base pair mutations, deletions, insertions
and point mutations, SCEs, SSBs, UDS, and DNA repair inhibition in various cells in vitro, in
experimental animal models in vivo, as well as in exposed humans.
Formaldehyde is genotoxic. This conclusion is supported by several streams of evidence
including observations of CAs, MN, and SSBs in exposed humans across a range of studies,
occupations, and exposure scenarios, with supporting, similar findings in exposed rodents and in
vitro systems, and consistent observations of DPXs detected in multiple experimental systems,
showing a pattern of concentration-dependent increases. Together, these multiple streams of
evidence (from human, animal, in vitro and nonmammalian systems) converge to clearly indicate
that formaldehyde is genotoxic in most systems tested, is mutagenic in systems specifically
This document is a draft for review purposes only and does not constitute Agency policy.
1-303 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
evaluating genetic or chromosomal mutations, and exhibits strong evidence for mutagenicity in the
URT of rodents and humans following inhalation exposure.
Summary and integration of mechanistic pathways into a cancer mode of action
The evidence pertaining to URT carcinogenesis following formaldehyde exposure was
assembled into a putative URT cancer MOA network highlighting the potential contributions of
genotoxicity and cytotoxicity-induced regenerative proliferation (see Figure 1-25), as well as
incorporating the influences of underlying chronic inflammation and epigenetic activity as prime
examples of other considerations that can interact with and further modify the primary
mechanisms propelling formaldehyde-induced URT cancer, in addition to potentially contributing
independently. Table 1-38 presents a concordance summary view of the available evidence (Meek
etal.. 20141. illustrating the exposure concentration and duration required to either elicit or
amplify formaldehyde-associated effects in the URT of F344 rats (the model species most sensitive
to SCC development with the most diverse and robust data set available). These rat data are
informative of the mechanistic pathways of primary concern, including genotoxicity endpoints as
an indicator of mutagenic potential; reports of tissue pathology including hyperplasia, squamous
metaplasia, dysplasia, and necrosis; cellular DNA synthesis as an indicator of epithelial proliferation
rate (independent of cause); as well as formaldehyde-associated tumor induction (see Section 1.2.5,
Respiratory Tract Cancers in Animal Studies). These interrelated streams of evidence are
summarized separately (below) and then integrated into a composite MOA, which is evaluated in
subsequent sections.
This document is a draft for review purposes only and does not constitute Agency policy.
1-304 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Figure 1-25. An integrated cancer mode-of-action (MOA) network for the URT.
Various effects occur in a manner dependent upon duration and magnitude of formaldehyde (FA)
inhalation exposure. Primary mechanistic considerations in call-out boxes are described in the following
tables and figures (blue/genetic damage, see Table 1-39; green/formaldehyde-induced proliferation
without damage, see Table 1-40; red/tissue and cellular damage, see Tables 1-40 and 1-41) with evidence
identified from the formaldehyde database as possibly informative of molecular mechanisms. These
mechanistic considerations or modifying factors are consistent with those factors described as cancer
hallmarks, enabling, or key characteristics of carcinogens (Smith et al., 2016; Hanahan and Weinberg.
2011).
This document is a draft for review purposes only and does not constitute Agency policy,
1-305' DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-38. Concordance of temporal and dose-response relationships among
formaldehyde effects induced in F344 rat nasal epithelium in vivo
F344 Rats
Time (months)
Time (months) .
0-3
4-12
13-28
0-3
4-12
13-28
Genotoxicity°
Necrosisb
Exposure
(mg/m3)
0-2
+
ND
ND
-
-
-
2-7
%
•
++
ND
ND
-/+
-
-
>7
+++
ND
ND
++
+
+
Hyperplasia and/or metaplasiac
DNA synthesisd-e
Exposure
(mg/m3)
0-2
i
-
-
+
-/+
_f
_f
2-7
-/+
+
++
+
_f
_f
If
>7
+
++
+++
+++
++f
++f
Tumorigenesis
(polypoid adenoma)9
Tumorigenesis
(squamous cell carcinoma)9
Exposure
(mg/m3)
0-2
-
-
-
-
-
-
2-7
*
»
-
-
+
-
-
-/+
>7
-
-
++
-
+
+++
Male F344 rats were the most widely evaluated sex/strain/species/evaluated, but observations were comparable
between rat sexes, where available. The presence or absence of treatment-related effects across all available
studies (as determined by EPA review) in or near the nasal anterior lateral meatus (ALM, where specified,
generally within Level II), were depicted as follows: indicates the absence of effects; "ND" indicates no data
available for the specified endpoint/dose/time combination; -/+ indicates an equivocal response, or evidence
limited to the highest extreme of the exposure range indicated; +, ++, +++ indicate the presence of an
exposure-related effect, with symbol number corresponding to increasing magnitude, incidence, or severity,
relative to concurrent controls and other exposure level/duration entries within an effect category
(see Section 1.2.4 and Appendix A.4).
includes DNA-protein and DNA-DNA crosslinks or increases in N2-hmdG DNA adducts attributed to exogenous
formaldehyde exposure.
bDirect evaluation necrosis was not frequently reported, and apoptosis has not been directly measured; significant
exposure-related tissue destruction was inferred from pathological determination of necrosis, erosion,
disarrangement, or atrophy of the nasal epithelium.
Tissue reactive or adaptive responses to irritant or cytotoxic effects were determined by evaluating hyperplasia or
squamous metaplasia (typically combined in reporting by study authors) of the nasal respiratory or transitional
epithelium; however, the biochemical stimulus of this tissue reaction remains unclear, as such areas of
hyperplasia could also include areas of dedicated preneoplastic foci.
dDNA label incorporation as a measure of proliferation at the individual cell level in the ALM was measured by
incorporation of Brdll, [3H]-thymidine or [14C]-formaldehyde into DNA, and reported as an index normalizing
affected (positive) cells as a fraction of the total respiratory epithelium (see detailed summary in Appendix A.5.6).
This document is a draft for review purposes only and does not constitute Agency policy.
1-306 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
eDNA synthesis has been evaluated following both continuous and intermittent exposures; while effects of
continuous exposure are depicted herein for purposes of drawing comparisons across similar exposure scenarios,
intermittent exposure may be also informative for some human exposure scenarios.
'Results from a single study reporting rat nasal epithelial cell DNA label incorporation following 26, 52, or 78 weeks
of exposure (Monticello et al., 1996).
gBoth polypoid adenomas (PA) and squamous cell carcinomas (SCC) were described as likely arising from the
respiratory or transitional epithelium, typically on or near the ALM. However, SCCs were typically associated with
areas of hyperplasia or squamous metaplasia, whereas PAs were not.
Formaldehyde directly adducts DNA and proteins, causing dose-responsive increases in
DNA-protein (DPX) or DNA-DNA (DDC) crosslinks, as well as DNA mono deoxyguanosine (hmdG)
adducts (see Table 1-38, also see Appendix A.4). Evidence from humans and rodents suggests that
formaldehyde exposure can lead to increasing levels of reactive oxidative species (ROS) and
possibly inhibit cellular detoxification mechanisms (see Appendix A.5.6), which would be expected
to further exacerbate oxidative damage to cellular constituents and DPX formation. Following these
initial effects, single-strand DNA breaks could be created more frequently, and DNA repair could be
inhibited, possibly leading to an accumulation of genetic damage at the chromosome
(clastogenicity) and sequence level (gene mutations). While the specific nature of persistent
genetic damage leading to URT cancer following formaldehyde exposure is unclear, heritable
changes in genetic material are a prerequisite step for carcinogenesis following a mutagenic mode
of action. The observations most relevant to genotoxic effects and sequelae to URT neoplasia are
summarized inTable 1-39.
Table 1-39. Genotoxicity and mutagenicity in the upper respiratory tract
Observations from the available in vivo database
(see Appendix A.4for details)3*
Exposure level
(mg/m3)c
Statistical
associations'1
Human
Acute or short-term exposure: controlled
• No effect or limited T* on MN incidence in nasal/buccal
epithelial tissue
<1, or
17 mg/m3-hrse
NR
Subchronic exposure: repeat environmental (pathology and medical
students)
• 1" MN incidence in nasal and buccal epithelium, stronger
association in centromere-negative MN
0.5-2
[0.07-5]
NR and -/+
assoc. w/1" CE
Chronic exposure: repeat occupational/environmental
• 1" Binucleation, but not nuclear bud or MN frequency, in buccal
epithelium from furniture workers
0.04-0.1
[NR]
+ assoc. w/1" [C]
No assoc. w/1" D
• 1" MN frequency in nasal epithelium from workers
0.1-1
[0.05-5]
NR
• 1" MN frequency in buccal epithelium from anatomy/pathology
faculty and staff, laboratory or factory workers
0.2-NR
[0.05-5]
+ assoc.
exposed:referent
This document is a draft for review purposes only and does not constitute Agency policy.
1-307 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Observations from the available in vivo database
(see Appendix A.4for details)"*
Exposure level
(mg/m3)c
Statistical
associations'1
+ association w/1" D
Nonhuman primate
Acute or short-term exposure: controlled
• 1" DPX in the nasal mucosa; larynx, trachea, and/or carina;
maxillary sinuses and lower respiratory tract of rhesus monkeys
>0.9; >2; 7
- assoc. w/1"
distance from POE
• 1" Exogenous FA13 CD2-N2-hmdG adducts and DPXs in
maxilloturbinates of cynomolgus monkeys
>2
+ assoc. w/1" [C]
Rodent
Acute or short-term exposure: controlled
• 1" DPX in the nasal epithelium; no effect in bronchoalveolar
lavage fluid or nasal olfactory mucosa of F344 rats
>0.4;
>18
- assoc. w/1"
distance from POE
• 1" Exogenous FA13 CD2-N2-hmdG adducts and DPXs in nasal
epithelium of F344 rats
>0.9
+ assoc. w/1" [C], D
Subchronic exposure: controlled
• 1" DPX in the nasal epithelium of F344 rats
>0.9
- assoc. w/1"
distance from POE
• No effect on MN incidence in nasal epithelium of F344 rats
<18
NR
aTreatment-associated increase (1^), micronucleus (MN), DNA-protein crosslinks (DPX), DNA monomethyl
deoxyguanosine adducts resulting from exogenously administered formaldehyde (FA13 CD2-N2-hmdG), single-
strand DNA breaks (SSBs).
bThe earliest duration reported by the study authors to elicit the specified effect is noted for controlled exposure
studies, or the mean duration reported in epidemiological studies; multiple values are provided in cases where
the study authors described only a range of exposure durations, or to represent a range of average durations
from a collection of similar epidemiological or experimental reports.
c For experimental studies, lowest effective concentrations (LEC) are presented, while for individual
epidemiological studies, mean exposures are listed, otherwise the range of mean exposures is presented to
represent a collection of studies reporting similar effects, with the overall range reported in individual studies or
collections in [ ]; determinations were made by EPA review considering potentially biologically relevant effects
that were attributed by the study authors to formaldehyde exposure; ">" indicates that higher exposures were
evaluated that also indicated an exposure-related effect. Where no effect was reported, the highest ineffective
concentrations (HIC), or ranges of exposure are indicated; "<" indicates that concentrations lower than the HIC
were also evaluated.
dResults of association, regression, correlation, or trend analysis as reported by study authors; "NR" indicates that
either associations were not evaluated or that no significant associations (assoc.) were reported; positive (+),
weakly positive (-/+) associations, inverse association (-); with (w/), exposure duration (D), cumulative exposure
(CE), exposure concentration ([C]), apical portal of entry (POE).
eThis study employed a complex and variable exposure protocol, with individuals experiencing 17 mg/m3-hours of
cumulative formaldehyde exposure distributed throughout a period of 40 hours over 10 workdays (2 weeks).
'Results presented from respiratory or transitional epithelial tissue generally described as located in "Level II" of
the anterior rodent nasal passages, including the nasal lateral meatus, septum, naso- and maxilloturbinates, as
described in Section 1.2.4.
This document is a draft for review purposes only and does not constitute Agency policy.
1-308 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
In addition to directly damaging DNA, formaldehyde inhalation can cause a number of
pathological cellular changes in the URT, such as inhibited mucous flow and decreased ciliary beat,
rhinitis and inflammation, ciliastasis, cilia loss, and possibly sporadic epithelial proliferation at low-
to-moderate exposure levels that elicit marginal increases in frank tissue toxicity as evidenced by a
lack of necrosis, epithelial degeneration, or squamous metaplasia in the nasal passageways
(see Section 1.2.4). Any molecular mechanisms responsible for such respiratory epithelial
proliferation remain to be determined, but could include some of the cytokines and eicosanoids
associated with URT inflammation and leukocyte extravasation, epigenetic activation, or
suppression of cell cycle regulatory machinery through changes in gene regulation, including
miRNA, loss of contact-inhibition signaling, or even direct stimulation of epithelial mitosis via
adduction of growth factor-signaling mediators (see Appendix A.5.6 for the evidence available on
some of these potential events). Accelerated cell cycle progression can increase the rate of random
genotoxic events in proliferating cells (indirect genotoxicity), which—if improperly repaired due to
insufficient delay in G1 phase, failure to arrest in S phase, or deficiency of DNA repair machinery—
could lead to heritable mutations and eventually URT neoplasia (Branzei and Foiani. 20081. Tissue
stem cell proliferation rate and the contribution of this random or "background" mutagenesis to
human lifetime cancer risk has been proposed to be significant for a variety of tissues (Tomasetti
and Vogelstein. 20151. although the relevance, magnitude, and scope are still under debate (Rozhok
etal.. 2015: Wild etal.. 2015: Wodarz and Zauber. 2015). Experimentally, the magnitude of
formaldehyde-induced DNA synthesis is dramatically increased as a function of concentration and,
to a lesser extent, duration, reaches maximal levels after 1-3 months with short-term or subchronic
exposure, and then appears to diminish in the only study that looked at changes after exposure
longer than 13 weeks (see Appendix A.5.6). Observations from direct DNA labeling studies are
summarized in Table 1-40 (scenarios involving cytotoxic exposures are described below).
Table 1-40. Direct measurements of DNA synthesis in the upper respiratory
tract
Observations from the available in vivo database
(see Appendix A.5.6 for details on proliferation)aM
Exposure level
(mg/m3)c
Statistical
associations'1
Nonhuman primate
Acute—subchronic exposure: controlled
• 'T* Epithelial cell proliferation in nasal and extranasal transitional and
respiratory epithelium of rhesus monkeys
7
- assoc. w/^ D,
distance from POE
Rodente
Acute exposure: controlled
• 'T* Epithelial cell proliferation in nasal septum, lateral meatus, or
turbinates of Wistar rats; in the anterior nose (not otherwise specified)
in Sprague Dawley rats
>4; >3
NR; NR
This document is a draft for review purposes only and does not constitute Agency policy.
1-309 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
Toxicological Review of Formaldehyde—Inhalation
Observations from the available in vivo database
(see Appendix A.5.6 for details on proliferation)aM
Exposure level
(mg/m3)c
Statistical
associations'1
• 'T* Epithelial cell proliferation in the nasal lateral meatus, or
maxilloturbinates in F344 rats
>7
- assoc. w/^ D
+ assoc. w/^ CEf
• 'T* Epithelial cell proliferation in the nasal lateral meatus, or
nasoturbinates in B6C3Fi mice
>15
- assoc. w/^ D
+ assoc. w/^ CEf
Subchronic exposure: controlled
• 'T* Epithelial cell proliferation in nasal septum, turbinates, or lateral
meatus of Wistar rats
>4
+ assoc. w/^ [C] and
not CE
• 'T* Epithelial cell proliferation in the nasal lateral meatus, septum,
and/or turbinates of F344 rats
>3-75
- assoc. w/^ distance
from POE
+ assoc. w/^ [C], D
Chronic exposure: controlled
• 'T* Epithelial cell proliferation in the nasal lateral meatus in F344 rats
>12
- assoc. w/^ D,
distance from POE
aTreatment-associated increase (1^).
bThe durations reported by the study authors to elicit the specified effect are noted for controlled exposure
studies; multiple values represent different durations from several experimental reports,
lowest effective concentrations (LEC) are presented for experimental studies, as determined by EPA review
considering potentially biologically relevant effects that were attributed by the study authors to formaldehyde
exposure; ">" indicates that higher exposures were evaluated which also indicated an exposure-related effect.
dResults of association, regression, correlation, or trend analysis as reported by study authors; "NR" indicates that
either associations were not evaluated or that no significant associations (assoc.) were reported; positive (+) or
inverse association (-); with (w/), exposure duration (D), cumulative exposure (CE), exposure concentration ([C]),
apical portal of entry (POE).
eResults presented from respiratory or transitional epithelial tissue generally described as located in "Level II" of
the anterior rodent nasal passages, including the nasal lateral meatus, septum, naso- and maxilloturbinates,
whereas "Level I" commonly included the high-flux region and nose tip, as described in Section 1.2.4 and
Appendix A.2.
these associations are for "Level I" epithelial cells; only exposure concentration ([C]) was positively associated
with cells in "Level II."
gLEC reported varied among reports from different authors and following exposures of different durations.
At higher, cytotoxic exposure levels, regenerative tissue proliferation concomitant with and
resulting from cytotoxic epithelial pathology (including squamous hyperplasia, metaplasia, and
dysplasia, with or without evidence of frank necrosis; discussed in Section 1.2.4) occurs in an
exposure concentration- and duration-dependent manner. The relative contribution of exposure
concentration and duration to this process may not be equal, particularly for events that segue from
hyperplasia (exposure duration appears to be substantially more important to the development of
metaplasia in laboratory animals than to the development of hyperplasia; see Section 1.2.4);
however, specific data defining the relative contributions are unavailable. Metaplasia or
hyperplasia is induced at moderate to high exposure levels after even short-term exposure, and
extending the duration generally both increases the severity of nasal tissue pathology observed and
decreases the exposure concentration necessary to elicit significant cytotoxicity (see Section 1.2.4).
This document is a draft for review purposes only and does not constitute Agency policy.
1-310 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
Pathological indications of significant epithelial necrosis in F344 rats are primarily reported
following exposure to relatively high concentrations, with similar results in Wistar or Sprague
Dawley rats, although occasionally necrosis is reported at more moderate exposure levels. Under
these conditions, the tissue rhinitis/inflammation, macromolecule adduction, or inhibition of
cellular function is presumably severe enough, possibly in conjunction with tissue glutathione
(GSH) depletion, to trigger cell death and significant regenerative pathology in the nasal respiratory
or transitional epithelium. Together, these effects can increase damage from all sources to cellular
constituents (e.g., membrane lipids and proteins, cytosolic proteins, DNA), and amplify genotoxicity
while simultaneously decreasing the capacity for and fidelity of DNA repair. Thus, both direct and
indirect effects of formaldehyde exposure at these levels can feed forward to increase
insurmountable cellular toxicity. Cytotoxicity and death of more sensitive cells in the respiratory
epithelial tissue compartment could select for and trigger compensatory proliferation among more
resistant cells in the population, possibly including the division and differentiation of local
pluripotent stem cells, all of which may replicate to replenish the damaged nasal mucosa. The
magnitude of these tissue proliferative effects may also fluctuate as the result of epithelial tissue
responses to chronic, continuous (i.e., metaplastic differentiation to a squamous phenotype) versus
episodic (variable pathology) exposure scenarios. In this manner, formaldehyde exposure may
accelerate proliferation as a field effect at the epithelial tissue level, causing genotoxicity and
mutagenesis in both actively proliferating (direct and indirect genotoxicity) and more quiescent
cells (direct genotoxicity only). Observations relevant to cytotoxic tissue pathology and
regenerative proliferation are summarized in Table 1-41.
Table 1-41. Epithelial pathology, cytotoxicity, and regenerative proliferation
in the upper respiratory tract
Observations from the available in vivo database
(see Appendix A. 5.6 for details)"*
Exposure level
(mg/m3)c
Statistical
associations'1
Human
Acute Exposure: Controllede
• 'T* Nasal mucosal membrane swelling; nasal and throat irritation
>0.07; >0.3
NR;
+ assoc. w/^ [C]
• \1/ Nasal mucociliary function, mucus flow rate; 'T* rhinitis and
permeability index
>0.3; >0.5
No assoc. w/D; NR
Chronic Exposure: Repeat Occupational/Residential
• \1/ Nasal patency (airway volume)
0.01
[0.003-0.02]
- assoc. w/dust, N02,
mold
• 'T* General symptoms of rhinitis, URT irritation, or inflammation
0.05-1
[0.01-2]
+ assoc. w/^ [C],
No assoc. w/D
• \1/Nasal mucociliary function
0.3
[0.05-0.5]
No assoc. w/D
This document is a draft for review purposes only and does not constitute Agency policy.
1-311 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Observations from the available in vivo database
(see Appendix A. 5.6 for details)"*
Exposure level
(mg/m3)c
Statistical
associations'1
• 'T* Nasal hyperplasia, keratinization, or squamous metaplasia
0.3-NR
[0.02-2.5]
No assoc. w/D
+ assoc. w/age >50
Nonhuman Primate
Acute Exposure: Controlled
• \1/ Cilia content and 'T* hyperplasia or squamous metaplasia in nasal
epithelium, nasopharynx, and larynx of rhesus monkeys
7
- assoc. w/^ distance
from POE
Subchronic Exposure: Controlled
• 'T* Squamous metaplasia and hyperplasia in nasal epithelium,
nasopharynx, and larynx of rhesus monkeys
7
+ severity w/^ D
- assoc. w/^ distance
from POE
• 'T* Squamous metaplasia and hyperplasia in nasal turbinates of
cynomolgus monkeys
>4
+ assoc. w/^ [C]
Rodent}
Acute Exposure: Controlledg
• 'T* Nasal rhinitis, hyperplasia, or squamous metaplasia in Wistar rats
4
NR
• \1/ Microvilli content in nasal epithelial cells, \|/ nasal mucociliary
function, flow rate; 'T* nasal squamous metaplasia of F344 rats
>3; >7
- assoc. w/^ [C], D;
NR
• 'T* Nasal squamous metaplasia or hyperplasia in Swiss-Webster or
B6C3Fi mice
>4
NR
Subchronic Exposure: Controlled
• 'T* Nasal rhinitis, hyperplasia, or squamous metaplasia; \|/ cilia content
of nasal septa epithelium in Wistar rats
>4; 4
+ assoc. w/^ [C] and
not CE; NR
• 'T* Nasal hyperplasia or squamous metaplasia in F344 rats
>7-12
- assoc. w/^ distance
from POE
• 'T* Nasal squamous metaplasia and seropurulent inflammation in
B6C3Fi mice
>12
NR
Chronic Exposure: Controlled
• 'T* Nasal rhinitis, hyperplasia, or squamous metaplasia in Wistar and
F344 rats
>1 and >3
NR and
+ assoc. w/^ [C], D
• 'T* Nasal squamous metaplasia (but not rhinitis or hyperplasia) in
Sprague Dawley rats
18
NR
• 'T* Nasal rhinitis, hyperplasia; nasal squamous metaplasia and dysplasia
in B6C3F1 mice
>3; >12
NR; NR
aTreatment-associated increase (1^), treatment-associated decrease (4/), hours (hrs), upper respiratory tract (URT).
bThe earliest duration reported by the study authors to elicit the specified effect is noted for controlled exposure
studies, or the mean duration reported in epidemiological studies; multiple values are provided in cases where
the study authors described only a range of exposure durations, or to represent a range of average durations
from a collection of similar epidemiological or experimental reports.
Tor experimental studies, lowest effective concentrations (LEC) are presented, while for individual epidemiological
studies, mean exposures are listed, otherwise the range of LECs or mean exposures are presented to represent a
collection of studies reporting similar effects, with the overall range reported in individual epidemiological studies
This document is a draft for review purposes only and does not constitute Agency policy.
1-312 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
or collections shown in brackets ([ ]); determinations were made by EPA review considering potentially
biologically relevant effects that were attributed by the study authors to formaldehyde exposure; ">" indicates
that higher exposures were evaluated that also indicated an exposure-related effect.
dResults of association, regression, correlation, or trend analysis as reported by study authors; "NR" indicates that
either associations were not evaluated or that no significant associations (assoc.) were reported; positive (+),
inverse association (-); with (w/), exposure duration (D), cumulative exposure (CE), exposure concentration ([C]);
apical portal of entry (POE).
eDue to the abundance of acute exposure human studies, only those rated as Tier I or IIA are summarized, as
described in Appendix A.5.6.
'Results presented from respiratory or transitional epithelial tissue generally described as located in "Level II" of
the anterior rodent nasal passages, including the nasal lateral meatus, septum, naso- and maxilloturbinates, as
described in Section 1.2.4.
gDue to the abundance of acute exposure rodent studies, only those rated as Tier I or II are summarized, as
described in Appendix A.5.6.
Relationships among the various events discussed above are integrated into a mechanistic
network depicted in Figure 1-26, along with the modifying factors of chronic airway inflammation,
oxidative stress, and epigenetic effects, which are also likely to stimulate or enhance URT
tumorigenesis. Together, these primary mechanistic events and modifying factors form potential
adverse outcome pathways (AOP), which are illustrated as a network of interconnected events
[adverse outcome network (AON)], with some duplication of events across individual pathways for
clarity (see Figure 1-27). These figures highlight various interactions among mechanistic elements
for which some evidence exists in the formaldehyde database. They also facilitate the discussion
and evaluation of this evidentiary support The figures are not intended to illustrate every possible
relationship among various aspects of formaldehyde toxicity and do not represent an attempt to
exhaustively list all possible carcinogenic mechanisms. Furthermore, the understanding of how
such signaling circuits actually operate in human carcinogenesis is still fragmentary and the current
subject of intense study (Weinberg. 2014). The following section serves to evaluate the supporting
evidentiary data pertaining to the events depicted in these figures.
This document is a draft for review purposes only and does not constitute Agency policy.
1-313 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Rhinitis /
inflammation
Ciliastasis /
cilia loss f j
Leukocyte
recruitment,
dysfunction ,
•T" DNA strand
breaks
<
t NALT /""N
hyperplasia
,ry x
^ 1* epithelial cell\. V
* s s /r ** v
damage, death
o ' V
1s lipid, protein, \ 7i . ¦>
DNA oxidation
7^0;
f' clastogenicity
<
n
Allergy, asthma,
sensitization
^ H proliferation
©4^ DNA repair /""N rate
capacity/fidelity vjl/
1 f",
gene mutations
>( )<-
\ 1" hmdG
adducts
I Sustained airway
I ¦ n ' <-
I
Neoplasia
O
A
¦ > i Epigenetic effects
0
Mechanistic
consideration
©
Reliable evidence;
Plausibly a direct effect
0
Toxicological sequelae
of exposure
r_-j
Modifying factor
->
Association for FA
demonstrated
Association plausible
Figure 1-26, Mechanistic relationships relevant to URT carcinogenesis.
Integration of the molecular evidence available for the spectrum of formaldehyde- [FA-] related health
effects pertinent to upper respiratory tract carcinogenesis summarized in the previous sections.
Endpoints are depicted with varying degrees of support (with solid lines representing evidence from
exposure in vivo, or consistent findings across multiple types of in vitro evidence). The identification of
"reliable evidence" and related conclusions depicted in this figure are based primarily on evaluations
conducted elsewhere (i.e., robust or moderate evidence described in Appendices A.4 and A.5.6).
Plausible relationships are illustrated in a manner consistent with the cancer MOA schematic in
Figure 1-25, including the hallmarks and enabling characteristics of cancer outlined therein.
This document is a draft for review purposes only and does not constitute Agency policy,
1-314' DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
[Metabolism]
Electro-
philic
Molecular
Cellular Responses
Tissue/Organ
Organism
Initiating Events
Responses
Responses
DNA
adducts
Protein
adducts
t DNA
strand (-
breaks |
5
Cellular Mitogenesls
¦J, DNA
repair
fidelity
1
Stimulat.
prolifer.
Undeter.
cyto-
toxicity
Genotoxicity and
Mutagenicity
_4_£
Altered
Dysreg.
DNA
~
cell cycle
content
prog.
Colored boxes correspond to key
characteristics of carcinogens
I 1 Plausibly direct effect, key
' ' consideration for pathway
~
Primary mechanistic consideration in
HCHO carcinogenicity
- i Other relevant, less supported or
- 1 later sequelae
. Well-supported linkage following
formaldehyde exposure
Plausible linkage following
formaldehyde exposure
Linkage likely to provide stimulation
as part of a feedback-loop
Cytotoxicity and Regenerative
Proliferation
URT
neoplasia
1" Death
"T Regen.
growth
microenv.
4, DNA
Altered
repair
DNA
fidelity
content
Dysreg.
cell cycle
prog.
' inflam- .
I I
• mation |
Modifying
Factors
t
Oxidative
Stress
Epigen.
alteration
Figure 1-27. Network of adverse outcome pathways relevant to URT
carcinogenesis.
Integration of the possible key events in pathways describing the role of genotoxicity and mutagenicity,
cellular mitogenesis, and cytotoxicity and regenerative tissue proliferation in URT carcinogenesis following
formaldehyde exposure. Endpoints are depicted with varying degrees of support (with solid lines
representing evidence from exposure in vivo, or consistent findings across multiple types of in vitro
evidence), with plausible relationships as hashed arrows, and possible feed-back loops illustrated as
dotted reverse-facing blue lines. Boxes of varying colors represent events associated with related groups
of key characteristics of carcinogens (Smith et al.. 2016); electrophilicity, genotoxicity, and DNA repair
elements are in blue, cell death and proliferation elements are in green, while the influence of chronic
inflammation, oxidative stress, and epigenetic alterations are depicted as factors modifying the network
in orange, purple, and yellow, respectively.
1 Evaluation of experimental support for the hypothesized mode of action
2 Genotoxicity
3 DNA-protein crosslinks (DPXs) were significantly elevated in the respiratory tracts of
4 rhesus monkeys after 3 days of inhalation exposure, with lowest effective concentrations (LEC)
5 increasing with anatomical distance from the apical POE, from 0.9 mg/m3 in the nasal turbinates, to
6 2 mg/m3 in the larynx, trachea, and carina (pooled samples), and 7 mg/m3 in maxillary sinuses and
7 lungs (Casanova et al.. 1991). demonstrating direct genotoxicity as an early effect in tissues
This document is a draft for review purposes only and does not constitute Agency policy,
1-315' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
analogous with sites of tumor formation in humans. In rats, increased DPX levels from exogenous
formaldehyde were observed in the nasal lateral, medial, and posterior meatus fCasanova etal..
19941 or the entire nasal cavity of rats after >0.86 mg/m314C-formaldehyde inhalation (Casanovaet
al.. 19891. following single and multiple inhalation exposures over 0.25-81 days. Exogenous DPXs
resulting from exposure to 13C, d2-labeled formaldehyde were reported in nasal passages from both
nonhuman primates and rats. In rat nasal passages, DPX levels accumulated several-fold following
28 days of exposure to 2.5 mg/m3 and remained largely unchanged following 7 days of recovery
postexposure (different time points were not evaluated in nonhuman primate studies. Lai et al..
20161. Interestingly, while DPX levels increased by 2-fold to 30-fold over control levels from 0.9 to
18 mg/m3 in rat nasal passages fNTP. 2010: Liteplo and Meek. 20031. the rate of DPX formation per
unit of formaldehyde exposure (DPX/ppm exogenous formaldehyde) increased to a plateau at
7 mg/m3, where it remained constant from 7 to 18 mg/m3 fSwenberg etal.. 2013: Casanova-
Schmitz etal.. 1984b). In both rhesus monkeys and F344 rats, DPX incidence was inversely
associated with increasing anatomical distance from apical POE (Casanova and Heck. 1997:
Casanova et al.. 1994: Casanova etal.. 1991: Casanova etal.. 1989: Lam etal.. 1985: Casanova-
Schmitz etal.. 1984b: Casanova-Schmitz and Heck. 19831. While increased DPX formation in
human peripheral white blood cells (WBCs) has been positively associated with duration of
exposure to concentrations >0.3 mg/m3 [fLin etal.. 2013: Shah am etal.. 2003: Shaham etal.. 1997:
Shaham etal.. 1996): see Appendix A. 4], DPX levels have not been evaluated in analogous human
POE tissues (i.e., nasal, buccal, or nasopharyngeal epithelium).
Bulky DNA adducts, such as DPX, can block progression of the DNA polymerase complex,
possibly contributing to genotoxicity or cell death in the URT (for further discussions see
Appendices A.4 and A.5.6; fWongetal.. 2012: Heck and Casanova. 199911. After a single exposure
in rats, the inhibition of DNA replication due to DPX blockage was also predicted to be significant at
>7 mg/m3 fHeck and Casanova. 19991. While DNA replication was thought to be only marginally
affected after a single exposure to lower concentrations (<1% at 1 mg/m3 in rats), this effect may
increase in magnitude or impact with the accumulation of DPXs and DNA adducts resulting from
repeated exposure, as discussed below. Although the mechanisms regulating these effects remain
undetermined, exposures >7 mg/m3 are associated with increasingly severe epithelial pathology,
cell death, and hyperproliferation in rat nasal passages following subchronic exposure, as well as
dramatic increases in SCC formation after chronic exposure (see discussions of the specific animal
evidence in Sections 1.2.4 and 1.2.5).
In addition to forming crosslinks, biochemical investigations have demonstrated that
formaldehyde can react with DNA to form predominantly N6-hydroxymethyl-deoxyadenosine
(N6-hmdA) and N2-hydroxymethyl-deoxyguanosine (N2-hmdG) adducts, with dA adducts more
abundant than dG (Cheng etal.. 2008: Zhong and Hee. 2004: Beland etal.. 1984). While both DNA
adducts have been detected in various tissues in vivo, likely resulting from endogenous
formaldehyde reactivity, studies administering deuterium-labeled formaldehyde (13C, d2) have
This document is a draft for review purposes only and does not constitute Agency policy.
1-316 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
detected labeled N2-hmdG, but not N6-hmdA, in the URT epithelium of both rodents and nonhuman
primates (see Table 1 42; fLu etal.. 2012: Lu etal.. 2011: Lu etal.. 2010bl: see Appendix A.4; fYu et
al.. 2015b: Swenberg etal.. 2013: Moeller etal.. 20111. as well as human HeLa cells in culture fLu et
al.. 20121. The inability to detect 13C, d2-N6-hmdA was surprising, since 13C, d2-N2-hmdG is reliably
quantifiable following low levels of exposure, and increases in an exposure-dependent manner in
both rodents and nonhuman primates (Yu etal.. 2015b: Swenberg etal.. 2013): the reason for the
apparent absence of 13C, d2-N6-hmdA adducts formed by reaction with exogenous formaldehyde
remains unknown (see Appendix A.2). N2-hmdG adducts resulting from exogenous exposure were
positively associated with exposure concentration in the nasal maxilloturbinates of cynomolgus
monkeys after 2 days, with an LEC of 2 mg/m3 fMoeller etal.. 20111. and also in the nasal
epithelium of F344 rats after 1 to 28 days, with an LEC of 0.86 mg/m3 fYu etal.. 2015b: Lu etal..
2011: Lu etal.. 2010bl. However, formaldehyde exposure up to 0.37 mg/m3 in F344 rats failed to
induce DPXs or hmDNA adducts in the nasal epithelium or in systemic tissues (Leng etal.. 2019). As
with DPXs, rat nasal N2-hmdG adduct formation was also positively associated with exposure
duration, with adducts accumulating to levels >5 times higher after 28 days of exposure to
2.5 mg/m3 compared with single exposures; different time points were not evaluated in nonhuman
primate studies fYu etal.. 2015b: Swenberg etal.. 2013: Lu etal.. 2010bl. No studies have assessed
the formation of exogenous hmDNA adducts in any tissues from humans exposed to formaldehyde.
Together with the above, acute exposure in rats and nonhuman primates appears to be
sufficient to significantly increase formation of DPXs at an LEC of approximately 0.86 mg/m3 and
exogenous N2-hmdG adducts at LECs of 0.86 and 2 mg/m3 in analogous nasal tissues from both
species. The observation that both DPXs and N2-hmdG adducts are positively associated with
exposure concentration in both nonhuman primates and rats fLai etal.. 2016: Yu etal.. 2015b:
Swenberg etal.. 2013: Lu etal.. 2011: Moeller etal.. 2011: Lu etal.. 2010bl. and that they
accumulate in rat nasal passages with repeat exposure fLai etal.. 2016: Yu etal.. 2015bl. is
consistent with the hypothesis that DPXs may undergo spontaneous hydrolysis to form N2-hmdG
adducts (Yu etal.. 2015b). While some DPXs may undergo hydrolysis to form N2-hmdG adducts
following exogenous formaldehyde exposure, other DPXs appear to be quite stable in vivo; it may
be these latter DPXs that play a more important role in formaldehyde-mediated respiratory tract
mutagenicity and carcinogenicity fLai etal.. 2016: NRC. 20111.
In addition to DNA adducts, strand breaks and cytogenetic endpoints have also been
observed following formaldehyde exposure, and such damage can lead to heritable mutations,
deletions, amplification, or chromosomal abnormalities if not successfully repaired. While DNA
strand breaks have not been evaluated in apical POE tissues from rats or nonhuman primates, DNA
SSB incidence was significantly increased in a concentration-dependent manner in both lung
epithelial cells and PBLs from Sprague Dawley rats after 14 days of exposure to >6 mg/m3, in the
absence of significant protein or lipid oxidation in lung tissue fSul etal.. 2007: Im etal.. 20061.
corresponding with increased lung cell apoptosis observed following 28 days of exposure to
This document is a draft for review purposes only and does not constitute Agency policy.
1-317 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
>7 mg/m3 (Avdin etal.. 20141. Likewise, while strand breaks have not been measured in adult
human URT tissues, increased SSBs have been reported in PBLs following occupational exposure to
>0.3 mg/m3 fAvdin etal.. 2013: Lin etal.. 2013: Costa etal.. 2008. see Appendix A. 41.
Unlike DNA stand-breaks, clastogenicity (in particular, MN formation) has been evaluated in
human URT tissues. Acute, controlled exposures in healthy human volunteers yielded equivocal
results; furthermore, MN incidences fell dramatically in both tissues during 21 days of
postexposure monitoring (Zeller etal.. 2011: Speitetal.. 20071. Binucleation only, a proposed early
event in MN formation, was elevated in buccal tissues from workers repeatedly exposed to low
formaldehyde levels (meanlocation-specific concentrations of 0.04-0.11 mg/m3; fPeteffi etal..
20151. Although MN incidence was not significantly elevated in rat URT tissues after 28 days of
exposure to <18 mg/m3 (see Table 1-42) fSpeitetal.. 2011: Neuss etal.. 20101. the majority of
human studies have reported significant MN induction in the buccal epithelium after 5-35 years of
occupational exposures to higher concentrations, averaging >0.2 mg/m3 (see Table 1-42) (Costa et
al.. 2019: Aglan andMansour. 2018: Ladeira etal.. 2013: Ladeiraetal.. 2011: Viegas etal.. 2010:
Burgaz etal.. 2002: Burgaz etal.. 20011. and in the nasal epithelium of adults after an average of
7-11 years at >0.1 mg/m3 f Costa etal.. 2008: Ye etal.. 2005: Ballarin et al.. 19921. Results in
students from shorter- duration classroom exposures (60-90 days) to 0.5-2 mg/m3 have been
lower in magnitude and less consistently positive, showing a stronger association between
cumulative exposure and buccal versus nasal MN incidence and a stronger association with
centromere-negative MN incidence, consistent with MN formation following DNA strand breakage
(Yingetal.. 1997: Titenko-Holland etal.. 1996: Suruda etal.. 19931. This hypothesized mechanism
is consistent with the gene expression profile of human B-lymphoblastoid cells (Tk6) directly
exposed to cytotoxic concentrations of formaldehyde in vitro, with transcript changes more akin to
DNA-alkylating clastogenic agents than aneugenic spindle poisons fKuehner et al.. 20131. In buccal
epithelium from human students or factory workers, MN incidence was positively correlated with
exposure duration (p < 0.01) following exposure to 0.06-0.6 mg/m3 for >1 year (Viegas et al..
2010). and positively correlated with cumulative exposure in male (p = 0.01) or male + female
(p = 0.06) student populations exposed to 0.5-2 mg/m3 for 90 days (Titenko-Holland etal.. 1996:
Suruda etal.. 19931. Compared with the evaluations of URT tissues, cytogenetic endpoints have
been more frequently evaluated in PBLs from occupational exposure cohorts (for further
discussion, see Section 1.3.3 Evidence on Mode of Action for Lymphohematopoietic Cancers and
Appendix A.4). Most of the studies conducted over the past 20 years have reported increased PBL
MN incidence in formaldehyde-exposed humans, including the majority of studies reporting
formaldehyde-associated increases in buccal or nasal MN incidence (Kirsch-Volders et al.. 2014).
Together with the above, the existing evidence consistently supports the association of MN
induction in nasal and buccal tissue from human cohorts occupationally exposed to formaldehyde,
in a manner temporally, biologically, and dose-responsively concordant with observations of
This document is a draft for review purposes only and does not constitute Agency policy.
1-318 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
nasopharyngeal and sinonasal carcinogenesis across a range of exposure scenarios and
concentrations.
Similar MN induction in epithelial cells of the URT has also been associated with increased
human cancer risk in other populations fRamirez and Saldanha. 2002: Lippman etal.. 19901.
Independent of formaldehyde exposure, a strong correlation between POE (buccal) and systemic
(PBL) MN incidence has also been reported in samples collected from >6,500 healthy human
subjects across 10 countries (r = 0.86; fKirsch-Volders etal.. 2014: Ceppietal.. 20101. suggesting
that increases in PBL genotoxicity are relevant to human URT cancer risk, although the magnitude
of MN induction in buccal cells is typically less than in PBLs fHolland etal.. 20081. Elevated PBL MN
and nuclear bud incidence, such as that observed in cohorts of formaldehyde-exposed workers, are
predictive for lung cancer risk in smokers fFenech etal.. 2011: El-Zein etal.. 20061 and are
associated with increased cancer incidence in otherwise healthy individuals fKirsch-Volders et al..
2014: Bonassi etal.. 2008: Holland etal.. 2008: El-Zein etal.. 2006): see Section 1.3.3 Evidence on
Mode of Action for Lymphohematopoietic Cancers). Parallel increases in buccal and PBL MN
incidence have also been observed in human workers chronically exposed to wood dust, another
URT carcinogen fRekhadevi et al.. 20091. Similarly, in radon-exposed miners, a 1% increase in the
frequency of aberrant PBLs was associated with a 60% increase in lung cancer risk fSmerhovskv et
al.. 2002: Smerhovskv et al.. 20011. Together, this evidence supports associations between local
and peripheral clastogenicity and between tissue clastogenicity and human respiratory
carcinogenesis.
The mutation profile of formaldehyde-induced rodent tumors has not been well
characterized, and it is unclear which of the various genotoxic endpoints elicited by formaldehyde
exposure may lead to permissive mutations in either rodent or human URT carcinogenesis. P53
mutations were specifically evaluated in SCCs isolated from the nasal passages of F344 rats
following 2 years of exposure to 18 mg/m3 formaldehyde fWolfetal.. 1995a: Recio etal.. 19921. and
in hyperplastic nasal tissues following 90 days of exposure to similar concentrations (Meng etal..
2010). While not detected in hyperplastic epithelium, the p53 mutations at codon 271 detected in
five of the 11 rat URT SCCs have also been described in human URT cancers (Wolf etal.. 1995a:
Audrezetetal.. 1993: Recio etal.. 1992: Hollstein etal.. 1991). At 18 mg/m3, nasal squamous
metaplasia preceding or concomitant with hyperplasia is significantly elevated early after first
exposure (within 7 days; see Section 1.2.4), prior to the emergence of dysplasia at 365 days, in the
nasal regions of F344 rats, which eventually harbor SCC after 330-548 days fKamataetal.. 1997:
Monticello etal.. 1996: Kerns etal.. 1983). The absence ofp53 mutations in reactive nasal mucosa
after 90 days of exposure is consistent with p53 mutations acting as a selective or permissive factor
acquired during the latter stages of formaldehyde-initiated carcinogenesis, facilitating increased
genetic instability and the progression of nascent neoplasms to SCCs, which emerge months later
fHanahan and Weinberg. 2011. 20001. Perhaps consistent with this potential temporal
relationship, a recent study of short-term (i.e., 8-week) exposure to high levels of formaldehyde in
This document is a draft for review purposes only and does not constitute Agency policy.
1-319 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
two strains of p53 deficient mice failed to observe any treatment-related increases in nasal tumors
at 32 weeks post-exposure, despite pronounced metaplasia fMorgan etal.. 20171. Additional study
using longer-term exposures, ideally in rat models (as mice are demonstrably less sensitive), would
help clarify the role of p53 in URT carcinogenesis.
The proportion of human URT SCCs exhibiting p53 mutations is similar to that reported in
formaldehyde-elicited rat URT SCC (~45%), and codons orthologous to those with mutations in rat
nasal SCC are also mutated in human URT SCC (Catalogue of Somatic Mutations in Cancer [COSMIC]
build v73; filters: upper aerodigestive tract, all subtissues, carcinoma, squamous cell; accessed 10
July, 2015; http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/). However, this has not
been examined specifically in formaldehyde-exposed humans. The observation that formaldehyde-
induced rat URT carcinomas share similar p53 mutations with cancers in analogous human tissues
suggests that rat and human URT tissues may be subjected to similar initiating or selective
biological processes, which further supports the relevance of rodent URT tumors in informing
human cancer risk.
Summary:
Genotoxicity in the respiratory or transitional epithelium temporally and dose-responsively
precedes and anatomically coincides with sites of significant SCC and PA induction (see
Section 1.2.5) in rats following chronic formaldehyde exposure as a function of increasing
concentration (NTP. 2010: Liteplo and Meek. 2003). In both rats and nonhuman primates, nasal
DPX and exogenous formaldehyde N2-hmdG adducts were elevated in an exposure concentration-
or duration-related manner after 1-28 days of experimental exposure to formaldehyde
concentrations > 0.9 mg/m3 within the range of average occupational exposures associated with
increased DPXs in human PBLs (0.5-4 mg/m3) after various durations of exposure
(see Appendix A.4) and increased MNs in human nasal (0.1-1 mg/m3) or buccal tissue
(0.2-0.5 mg/m3) after >5 years (Appendix A.4). Human mortality risks from nasopharyngeal
cancer were also elevated with both increasing exposure concentration and duration, with elevated
risks evident at concentrations >1.23 mg/m3 and after ~20 years following first exposure (see
Section 1.2.5). The coherence of strong and consistent evidence for genotoxicity spans multiple
evidence types from exposed humans to relevant model systems and species, in analogous POE and
surrogate tissues, incorporating pertinent aspects of dose-response and temporality (i.e., preceding
other mechanistic events), all of which strongly supports a role for direct DNA damage leading to
mutagenicity in formaldehyde-induced URT carcinogenesis.
Cellular proliferation
Studies employing labeled nucleotides or analogs have reported increased epithelial cell
proliferation in the nasal and extranasal passageways of rhesus monkeys after 7 or 42 days of
exposure to 7 mg/m3, concurrent with increased tissue hyperplasia and metaplasia in the nasal
epithelium, nasopharynx, and larynx (see Section 1.2.4 and Appendix A.5.5). Acute exposure
This document is a draft for review purposes only and does not constitute Agency policy.
1-320 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
(1-9 days) to similar concentrations also stimulated epithelial proliferation in the anterior nasal
passages of F344, Wistar, and Sprague Dawley rats, while only exposures to >15 mg/m3 increased
proliferation in similar tissue from B6C3Fi mice. This difference in exposure concentrations
required to induce proliferation in nasal epithelium across rodent species may result from the
increased reflex bradypnea observed in mice compared to similarly exposed rats. Respiratory
minute volumes of mice acutely exposed to 15-18 mg/m3 decrease such thatthey are roughly
equivalent to a 7 mg/m3 exposure in rats (see Appendix A.3) (Swenberg et al.. 20131. This
difference in rodent physiology between mice and rats is also consistent with the reported SCC
incidence of 1-2% following chronic exposure to 18 and 7 mg/m3, respectively (see Section 1.2.5),
and with the apparent resistance of mice to formaldehyde-elicited cytotoxic nasal pathology (see
Section 1.2.4).
In Wistar rats, proliferation was increased in the anterior nasal passages after 28 or 90 days
of exposure with an LEC of 4 mg/m3, a concentration not frequently evaluated in other species
(see specific evaluations of proliferation in Appendix A.5.6) (Wilmer etal.. 1989: Zwartetal.. 1988:
Wilmer et al.. 1987). In F344 rats, cellular proliferation was induced to a similar extent after
90 days at >12 mg/m3 f Andersen etal.. 2010: Monticello etal.. 19961 or 7 mg/m3 in some studies
f Casanova et al.. 19941. A lesser magnitude of proliferation was also apparent following exposure
to >3 mg/m3 f Andersen etal.. 2010: Mengetal.. 2010: Monticello etal.. 19961. In both strains,
some evidence suggests increases in proliferation may occur at 0.8-2.5 mg/m3 (Andersen etal..
2010: Meng etal.. 2010: Casanova et al.. 1994: Zwartetal.. 1988) although this was inconsistent
across studies (see Appendix A.5.6). While proliferation in the anterior nasal passages may appear
to be stimulated to a greater extent at slightly lower exposure levels in Wistar versus F344 rats
(due in part to choice of exposure concentrations evaluated), the strain sensitivity to nasal SCC
induction was reversed: nasal tumors were present in only 4% of Wistar rats after 28 months of
exposure to 12 mg/m3, while 22% of F344 rats developed tumors after 24 months of exposure to
the same concentration (see Section 1.2.5; (Monticello etal.. 1996: Woutersenetal.. 1989). This
pattern also appears in PA incidence, where PAs were reported in ~1% (1 rat) of Wistar rats
exposed to 11 mg/m3 for <28 months (with lifetime observations), versus 6% of F344 rats exposed
to 12 mg/m3 for 24 months (Monticello etal.. 1996: Woutersen etal.. 1989: Feron etal.. 1988).
Unlike the differences seen with Wistar rats, incidence of both nasal SCCs and PAs appear to be
generally similar between Sprague Dawley and F344 rats exposed to 18 mg/m3 for 24-28 months
(see Section 1.2.5), although the limited evidence in Sprague Dawley rats precludes a comparison of
URT proliferation with F344 rats following repeat exposure (see Table 1-26). While limited, the
available data suggest that some strain differences exist in the URT tumor response in Wistar
versus F344 rats, while proliferation appears to be similarly induced in both rat strains.
Integrating across all available studies, the magnitude of proliferation induced in F344 rats
was generally similar following exposure durations of 4-90 days (see Appendix A.5.6). In the single
study available reporting URT epithelial proliferation in rats following chronic as well as
This document is a draft for review purposes only and does not constitute Agency policy.
1-321 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
subchronic exposures, the proliferation response declined between 45 and 90 days, most strikingly
at 7 mg/m3, and then decreased gradually throughout 548 days of continuous exposure fMonticello
et al.. 19961. An inverse association between nasal epithelium DNA synthesis and exposure
duration was reported between 7 and 42 days of exposure in rhesus monkeys fMonticello etal..
19891. suggesting that a proliferative peak may have been reached fairly rapidly in primates
(<7 days).
Investigations into the relative mitogenic versus cytotoxic consequences of formaldehyde
exposure in vitro have revealed that while significant cytolethality was observed at >1 mM in
cultured human colon carcinoma (HT-29), T lymphocyte (Jurkat E6-1) and umbilical vein
endothelial cells (HUVEC) fSaito etal.. 2005: Tvihak etal.. 20011. lower and more physiologically
relevant dose levels (0.1 mM; see Appendix A.2) induced proliferation in both HT-29 and HUVEC
cells, and to a greater extent in the neoplastic HT-29 cells compared with the nonneoplastic HUVEC
(Tvihak etal.. 2001). However, >0.1 mM induced endoplasmic reticulum (ER) stress and increased
the ratio of proapoptotic to antiapoptotic markers in both human lung carcinoma (A549; (Lim etal..
2013) and lymphoblast cell lines, with greater sensitivity observed in DNA repair deficient cells
fRen etal.l (see Appendix A.5.6). Increased sensitivity to formaldehyde-induced cell death has
been consistently reported in eukaryotic cell lines deficient in excision, DNA crosslink, or
chromosomal breakage repair fMchale etal.. 2014: Ren etal.. 2013: Noda etal.. 2011: Rosado etal..
2011: de Graaf et al.. 2009: Ridpath etal.. 2007). suggesting that unresolved genotoxicity could
contribute to some of the cytotoxicity observed with increasing levels of formaldehyde exposure.
Formaldehyde-stimulated cell cycle progression may be highly context dependent and only
observed in circumstances where the concomitant genotoxicity and low-level toxicity (e.g., ER
stress) are adequately controlled. This variable proliferation response in vitro is consistent with
some in vivo observations of increased epithelial proliferation in the nasal passages of F344 rats
following subchronic exposure at subcytotoxic exposure levels (~0.8-3 mg/m3; see Section 1.2.4
and a specific proliferation analysis in Appendix A.5.6). However, nasal epithelial proliferation in
the absence of cytotoxic nasal pathology was not consistently observed, and cell-density adjusted
cellular proliferation indices correlate well with tumor formation following chronic exposures to
>7 mg/m3, concentrations that induced significant epithelial pathology in rodent nasal passages
(see Section 1.2.4).
Summary:
Nasal epithelial cell proliferation was positively associated with the induction of squamous
metaplasia and necrosis or epithelial erosion in F344 rats (Andersen et al.. 2010) and correlated
with SCC incidence as a function of both anatomical location and exposure concentration following
exposures <19 mg/m3 for up to 548 days (Swenberg etal.. 2013: Monticello etal.. 1996). The
mutually permissive relationship between chemical carcinogenicity and epithelial cell proliferation
has been described for several respiratory tract carcinogens and rodent models of human cancers
fMonticello etal.. 19931. Such a relationship can accelerate the acquisition of traits consistent with
This document is a draft for review purposes only and does not constitute Agency policy.
1-322 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
a current understanding of the carcinogenic process (Goodson etal.. 2015: Sonnenschein and Soto.
2013: Hanahan and Weinberg. 20111. as exemplified in the well-described etiology of mutagen-
induced rat mammary gland tumorigenesis fRusso etal.. 19901. The available data suggest that
formaldehyde may elicit some mitogenicity at low-to-moderate exposures through an unknown
cellular mechanism independent from the regenerative tissue proliferation associated with
cytotoxicity following exposure to higher concentrations (see Figures 1-25-1-27). However, the
limited evidence supporting proliferation as an effect independent from cytotoxic tissue pathology
is not strong or consistent; furthermore, while the database contains several reports evaluating
cellular proliferation at a molecular level (i.e., DNA nucleotide analog incorporation), it suffers from
a dearth of molecular evaluations on other cellular functions, such as markers of toxicity, cell cycle
regulation, or death, which prevents a more precise delineation of mitogenic effects at a cellular
level from compensatory proliferation at a tissue level.
URT cytotoxicitypathology
In humans, nasal airway function may be impaired at average exposures as low as
0.01 mg/m3, suggesting that pathological URT changes occur even at low exposures
(see Table 1-42) (Norback et al.. 20001. while increasingly severe nasal histopathology (including
hyperplasia, keratinization, and metaplasia) is associated with average chronic exposures
>0.3 mg/m3 (see Table 1-42) fBallarin et al.. 1992: Bovsen etal.. 1990: Holmstrom et al.. 1989c:
Edling etal.. 1988: Odkvistetal.. 1985). The incidence of distinct dysplasia, a dedicated
preneoplastic lesion, was elevated in study participants with higher average chronic exposure,
ranging from 0.1 to 3 mg/m3 (see Section 1.2.4). Human nasal and throat irritation and cytotoxicity
was positively associated with exposure concentrations >0.2 mg/m3 in controlled acute exposure
trials or after a single 8-hr work shift (see Table 1-42) (Priha etal.. 2004: Kulle etal.. 19871 and
average exposure to 0.05-1 mg/m3 in occupational cohort studies fHolness and Nethercott. 1989:
Horvath etal.. 1988). Consistent with these observations, fluctuation in ciliary beat frequency was
also reported in primary human nasal cells exposed to 0.5-3 mg/m3 following differentiation into a
functional ciliated epithelium and cultured on an air-liquid interface (ALI) in vitro fWangetal..
2014). However, unlike the positive association between human MN induction and exposure
duration, or the clear relationship between rat squamous metaplasia induction and formaldehyde
exposure duration (see Section 1.2.4), no significant associations were reported between exposure
duration and various indications of human nasal mucosal pathology (see Table 1-42).
Similar to observations following chronic human exposure, the incidence of squamous
metaplasia and hyperplasia in the nasal turbinates of cynomolgus monkeys was also positively
associated with exposure concentrations >1 mg/m3 fRusch etal.. 19831. Although lesion severity in
rhesus monkeys was positively associated with extending exposure duration from 7 to 42 days at
7 mg/m3 fMonticello etal.. 19891. this observation is not necessarily discordant with the human
data set, which generally evaluated pathology resulting from chronic durations as a function of
differences in years of exposure versus days, as was evaluated in the nonhuman primates.
This document is a draft for review purposes only and does not constitute Agency policy.
1-323 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Nonhuman primates may be more resistant to nasal irritation and cytotoxicity than humans, as
squamous metaplasia and hyperplasia were observed following 42 days exposure to 7 mg/m3 in
rhesus monkeys fMonticello etal.. 19891. or 180 days of exposure to 4 mg/m3 to cynomolgus
monkeys, with 1 of 6 monkeys affected at 1 mg/m3 (vs. 0/12 in controls), and no effects observed at
0.2 mg/m3 fRusch etal.. 19831. although no studies have evaluated exposure durations directly
analogous to chronic human exposure.
In F344 rats, nasal mucociliary function and flow rate decreased in an exposure
concentration- and duration-associated manner following acute exposures to >3 mg/m3 (Morgan et
al.. 1986a: Morgan et al.. 1986cl. Incidence or severity of squamous metaplasia also increased in
both a duration- and concentration-dependent manner following exposures >3 mg/m3 fKerns etal..
19831: all effects were inversely associated with increasing distance from the apical POE (Casanova
etal.. 19941. Nasal pathology in Wistar rats was positively associated with exposure concentration,
but not cumulative exposure, following subchronic exposures (Wilmer et al.. 1989.1987). This
result is consistent with similar relationships reported between DNA synthesis rates and exposure
concentration in the same anatomical regions (i.e., Level II) in both Wistar and F344 rats
(see Table 1-42) fWilmer et al.. 1989: Zwartetal.. 1988: Wilmer etal.. 1987: Swenberg et al.. 19861.
Generally, formaldehyde exposure elicited similar pathology and ultrastructural changes in the
analogous nasal passages of both nonhuman primates and rats (see Section 1.2.4). F344 rats
appear to be similarly sensitive to the onset of nasal cytotoxicity induced by chronically inhaled
formaldehyde compared with nonhuman primates, since a similar duration of exposure
(180-365 days) induced nasal squamous metaplasia or hyperplasia in both species at >3 mg/m3,
while higher concentrations of >7-12 mg/m3 were generally required to induce similar pathology
following shorter durations (30-90 days; see Table 1-42). However, nasal damage in nonhuman
primates (rhesus monkeys) became more developed, covered the URT epithelium to a greater
extent, progressed to posterior nasal regions, and involved the larynx/trachea in less time
(1.5 months) and at lower exposure levels (Monticello et al.) than similar changes observed in rats
(Kerns etal.. 1983). Likewise, nasal squamous metaplasia in cynomolgus monkeys was detected in
all animals exposed to 4 mg/m3 after 6 months (Rusch etal.. 1983). while a comparable prevalence
of analogous pathology in F344 rats required exposure to 18 mg/m3 and >18 months to develop
(see Section 1.2.4).
Other rodent species appear to be less sensitive to formaldehyde-induced nasal dysplasia,
SCC and PA (in order of decreasing sensitivity): F334 and Sprague Dawley rats > Wistar rats >
B6C3F1 mice > hamsters (see Section 1.2.5). Necrosis, inflammation, hyperplasia, or squamous
metaplasia were observed in the anterior nasal passages of F344 rats, Wistar rats, and B6C3Fi mice
after short-term high-concentration exposures, as well as in the posterior nasal cavity of F344 rats
after 6 months, and in the larynx/trachea after 18 months of exposure to 18 mg/m3, although
tumors of the larynx or trachea have not been associated with formaldehyde exposure in rodents
(see Section 1.2.4). Conditions that induced nasal dysplasia in rats and mice consistently resulted
This document is a draft for review purposes only and does not constitute Agency policy.
1-324 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
in SCC formation after an additional 6-12 months of exposure, whereas neither dysplasia nor SCCs
were observed in hamsters (see Section 1.2.5). While formaldehyde-associated benign PAs and
malignant SCCs may share similar tissue level origins (i.e., the transitional or respiratory but not
olfactory epithelium), this reflects a neoplastic fate arising from morphologically different epithelial
populations and does not imply that PAs are precursor lesions to SCC. In the rodent nasal cavity,
SCCs are thought to arise directly from hyperplastic or dysplastic tissue (i.e., atypical squamous
metaplasia) and do not necessarily progress through a benign tumor intermediate (McConnell etal..
19861.
Summary:
Progressive tissue cytotoxicity and induction of proliferative pathological lesions in the URT
respiratory or transitional epithelium temporally and dose-responsively precede and anatomically
coincide with sites of significant SCC and PA induction (see Section 1.2.4) in rats following chronic
formaldehyde exposure as a function of increasing concentration fNTP. 2010: Liteplo and Meek.
2003). Similar lesions were also observed in the URT of nonhuman primates exposed up to
180 days, which appeared to progress farther along the primate respiratory tract. In humans, some
indications of URT cellular toxicity have been reported at very low concentrations, with
hyperplasia, keratinization, and metaplasia observed following chronic exposures >0.3 mg/m3,
which are concentrations approximately 10-fold lower than those eliciting similar effects in
experimental animal models. Together, strong and consistent evidence exists associating URT
epithelial pathology-driven tissue proliferation with SCC induction in rodent experimental models.
Along with limited information from both nonhuman primates and occupationally exposed humans,
these observations support a significant role for regenerative tissue proliferation in URT
carcinogenesis associated with formaldehyde exposures high enough to induce cytotoxic URT
pathology.
Summary of evidence supporting the primary mechanistic considerations:
In F344 rats chronically exposed to formaldehyde, there is a clear temporal,
dose-responsive, and biological relationship in the appearance of exposure-related genotoxicity,
sustained epithelial damage, cellular proliferation, and eventual SCC or PA development, consistent
with similar relationships evident in analogous URT tissues from both the nonhuman primate and
human databases. Furthermore, the chronic formaldehyde exposure concentrations reported to
elicit nasal cytotoxic pathology appear to be higher in the rats and nonhuman primates evaluated
experimentally (>3 mg/m3), compared with the results from human epidemiological cohorts
(>0.3 mg/m3; see Table 1-42), whereas formaldehyde-associated genotoxicity has been induced in
analogous POE tissues from rats, nonhuman primates, and humans exposed to similar
formaldehyde concentrations (see Table 1-42). Together, genotoxicity, cellular proliferation, and
cytotoxicity-induced tissue regenerative proliferation exhibit multiple layers of coherence as a
function of species and anatomy, temporality, concentration, and duration of exposure. When
This document is a draft for review purposes only and does not constitute Agency policy.
1-325 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
integrated, this evidence forms a biologically relevant MOA for formaldehyde exposure-induced
URT carcinogenesis fU.S. EPA. 2005al.
Other factors modifying the mode of action
Oxidative stress, immune disease, and dysfunction
Increased rhinitis, nasal irritation, URT inflammation, and some indications of increased
oxidative stress were observed in human cohorts after environmental or occupational exposures at
the lower end of the range of average formaldehyde exposures associated with nasal hyperplasia
and metaplasia. Rhinitis has been observed following subchronic or longer exposure in F344 rats
and B6C3F1 mice, as well as chronically exposed human workers, and some observations suggest
that oxidative stress may in part evolve as an effect secondary to the activation of inflammatory
leukocytes in the human respiratory tract (see Section 1.2.3 and Appendix A.5.6). The prevalence of
allergic conditions and asthma symptoms are increased in both children and adults exposed to
formaldehyde, suggesting that immune dysfunction occurs to some extent in respiratory tract
tissues following formaldehyde exposure (see Section 1.2.3 Immune-mediated Conditions). These
observations may imply a decreased functional activity of immune effector cells. Whether these
effects are due to immunosuppression, inappropriate polarization, or exposure-related cytotoxicity,
such immune dysfunction could promote a chronic inflammatory environment and permit cancer
progression (Tia etal.. 2014: Coussens et al.. 2013a. b; Balkwill etal.. 2012: Mantovani et al.. 2008).
In experimental rodent studies, depletion of nonprotein sulfhydryls (NP-SH, primarily GSH)
increased DPX formation in the nasal mucosa of F344 rats following formaldehyde exposure to
>1 mg/m3 fCasanova and Heck. 19871. while GSH coadministration attenuated increases in DPX
formation in systemic tissues from formalin-exposed BALB/c mice [Ye etal. (2013a): see also
Appendix A.4 and A.5.6], Although alterations in cellular GSH content may affect DPX formation
and the mutagenic potential of formaldehyde exposure, it is unclear whether formaldehyde
exposure itself will reduce URT glutathione levels in rodents. For example, even though glutathione
reductase activity was decreased in the rat URT following short-term exposure to >4 mg/m3, total
non-NP-SH content actually increased (Cassee etal.. 1996). A few other rodent studies have
reported increased oxidative stress from the lower respiratory tract (LRT) following short-term
exposures; however, data on oxidative stress endpoints from evaluation of URT tissues is limited,
and it remains unclear whether LRT responses indicate analogous responses in URT passages (see
Appendix A.5.6). In vitro, cellular GSH concentration was inversely correlated with formaldehyde
cytotoxicity in human oral fibroblast cells and rathepatocytes (Nilsson etal.. 1998: Ku and Billings.
1984). In conditions where GSH was sufficiently decreased, formaldehyde inhibited mitochondrial
respiration and led to increased lipid peroxidation and ROS production (IARC 88; (Tengetal..
20011. which could trigger NF-kB activation f Zhang etal.. 2013al and thus initiate an inflammatory
signaling cascade. While formaldehyde may directly deplete cellular GSH pools to some extent, the
resulting impact on cellular cytotoxicity can be amplified by other sources of oxidative stress fSaito
This document is a draft for review purposes only and does not constitute Agency policy.
1-326 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
et al.. 20051. Taken together, formaldehyde exposure may exacerbate oxidative stress primarily
resulting from inflammation, cytotoxicity, or sulfhydryl depletion, which could further augment
DPX-mediated genotoxicity as well as increasing ROS-mediated genetic instability and cell death.
This could result in an amplification of both direct and indirect mutagenicity in the nasal
epithelium.
Tumor immunosurveillance may play an important role specifically in limiting human
nasopharyngeal carcinoma development; for example, patients with acquired immune deficiency
syndrome (AIDS) are at significantly higher risk of developing both nonkeratinizing (commonly
associated with Epstein-Barr virus [EBV] infection) as well as keratinizing nasopharyngeal
carcinoma fShebl etal.l. In vitro, formaldehyde attenuates the perforin secretion and cell lytic
activity of cultured mouse and human natural killer (NK) cells at subcytotoxic concentrations (Kim
etal.. 2013a: Li etal.. 2013bl. which would limit NK-mediated destruction of infected epithelial cells
and prolong URT infection, possibly inhibiting any tumor-suppressive function of these cytotoxic
lymphocytes. Consistent with this theory, 2 weeks of formaldehyde exposure attenuated both NK
cell numbers and activity in the lungs of both naive and tumor-bearing mice. This attenuation was
associated with enhanced malignancy, growth, and neutrophil involvement of lung metastases
formed by injected syngeneic melanoma cells fKim etal.. 2013al. Additional evidence for other
formaldehyde-induced immune dysfunction comes from allergic sensitization studies and reports
of exacerbated immune-mediated airway hyperresponsiveness presensitized rodents (see
Section 1.2.3). Further, evidence exists to suggest the possibility that formaldehyde exposure may
alter immune cell phenotypes, maturation and survival at a systemic level (see relevant mechanistic
discussions in Sections 1.2.3 and 1.3.3); however, few studies have examined such evidence
specifically within respiratory tissues, and those testing endpoints that might otherwise be most
informative to this possibility fZhao etal.. 202Oal had methodological limitations that prevent clear
interpretation. Together, however, the available data suggest that formaldehyde exposure may
induce immune suppression or dysfunction in both experimental animals and humans, which could
reduce the effectiveness of local immunosurveillance in suppressing tumor progression and
metastasis, thus enabling URT carcinogenesis (Hanahan and Weinberg. 2011. 20001.
In summary, nasal infection and allergic symptoms are exacerbated in humans following
exposure to fairly low formaldehyde levels, concomitant with or preceding epithelial tissue distress,
inflammation, and preneoplastic lesion formation. Chronic inflammation is highly relevant to and
positively associated with human risk of respiratory tract cancers; however, the specific
mechanistic relationships between formaldehyde-induced inflammation, immune dysfunction,
infection, allergy, oxidative damage, and URT cancer remain unclear.
DNA repair inhibition
The primary effects of formaldehyde interactions with DNA are N2-hmdG adducts, DPXs and
DDCs, and strand breaks, and repair of such formaldehyde-mediated genotoxicity appears to be
crucial to cell survival. Consistent with this hypothesis, DNA repair genes are rapidly induced in rat
This document is a draft for review purposes only and does not constitute Agency policy.
1-327 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
nasal mucosa following acute or subchronic exposure in vivo (Rager etal.. 2014: Andersen etal..
2008: Hester etal.. 20051 and human B-lymphoblastoid cells in vitro fKuehner etal.l.
The primary mechanism for repair of N2-hmdG adducts is unclear. While nucleotide or base
excision repair (NER/BER) may be responsible, the removal of small DNA adducts species may also
result from nonspecific cellular processes fBrooks and Zakhari. 2014: Lindahl. 19931. The
existence of two phases in the elimination of formaldehyde N2-hmdG adducts from the rat nasal
mucosa in vivo also supports a role for multiple removal mechanisms (Swenberg et al.. 20131. DPXs
are unlikely candidates for direct removal via excision repair in mammalian cells, although a
fraction of smaller crosslink products (likely DDCs) may be removed via NER activity or proteolysis
(see Appendices A.4 and 5.6 for detailed discussions). DPXs are more likely repaired via activity of
the BRCA/Fanconi anemia family (FANC) proteins, components of the homologous recombination
repair pathway, which regulate DPX repair following chronic or lower formaldehyde
concentrations in mammalian cells and can attenuate the formation of DSBs and some
chromosomal abnormalities (see Appendix A.4) (Ren etal.. 2013: Rosado etal.. 2011: Nakano etal..
20091. If unresolved, DPXs could lead to SSBs, DSBs, various cytogenetic abnormalities, and
genomic instability f Kumari et al.. 2 015: Brooks and Zakhari. 2014: Kirsch-Volders etal.. 2014: Ren
etal.. 2013: Langevin etal.. 2011: Noda etal.. 2011: Nakano etal.. 2009: Ridpath et al.. 20071.
Additionally, DNA repair pathways are differentially engaged as a function of damage location in
relation to DNA replication machinery, supporting a role for the context of DNA damage in
determining the manner of its resolution (de Graaf et al.. 2009).
In cultured human fibroblasts, exogenous formaldehyde directly interfered with
DNA-binding damage sensor complex recruitment to DNA adducts and inhibited the repair of DNA
lesions induced by either ultraviolet light or cisplatin adduction fLuch etal.. 20141. consistent with
similar observations in other human tissues and cells (see Appendix A.4 for a detailed discussion).
This interaction also inhibited the migration and function of BER, and consequently inhibited the
repair of oxidative DNA lesions. These results suggest that formaldehyde may inhibit excision
repair by directly interfering with the DNA damage detection apparatus, which could delay the
recognition and repair of DNA damage induced by both formaldehyde as well as other agents.
However, any direct impact on the BRCA/FANC-mediated DNA repair pathway, which is likely to be
responsible for removing formaldehyde-induced DPXs following chronic exposure, remains to be
elucidated.
Members of the X-ray repair cross-complementing gene (XRCC) family serve as scaffolding
proteins for the repair of single- and double-strand DNA breaks, including those caused by
oxidative or UV-induced DNA damage (Kirsch-Volders etal.. 2014). Despite several correlations
between XRCC polymorphisms and increased sensitivity to formaldehyde-induced genotoxicity in
human tissues and cells, the role for XRCC family proteins in regulating formaldehyde mutagenicity
remains unclear (see Appendix A.4 for a detailed discussion). The molecular mechanisms by which
formaldehyde causes MN are also unknown, but incomplete repair of DNA-protein or DNA-DNA
This document is a draft for review purposes only and does not constitute Agency policy.
1-328 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
crosslinks, and the consequent stress from stalled replication forks, could result in DNA strand
breaks and possibly centromere-negative MN formation fBrooks and Zakhari. 2014: Kirsch-Volders
etal.. 2014: Nakano etal.. 20091. Taken together, the available data suggest that formaldehyde
exposure may inhibit the detection and repair of lesions resulting directly from formaldehyde-DNA
interactions, as well as genotoxicity resulting from other sources, and may thereby accelerate tissue
carcinogenesis by exacerbating both direct and indirect mutagenesis. However, the available data
are insufficient to determine any independent contribution of such interference in DNA repair to
URT carcinogenesis.
Epigenetics and toxicogenomics
Changes in message RNA (mRNA) transcript levels from pathways relevant to URT
carcinogenesis (e.g., cell cycle, proliferation signaling, apoptosis, and DNA repair) have been
reported in URT tissues following formaldehyde exposure, possibly mediated by microRNA
(miRNA) regulation, changes in DNA/histone modifying marks including methylation, acetylation
and formylation, or by responses to cellular toxicity and tissue distress (see Appendix A.5.6 for a
detailed discussion). After repeated exposure, mRNA levels for genes involved in growth signaling
pathways increased in a concentration- or duration-related manner in F344 rats (Rager etal.. 2014:
Andersen etal.. 20101. and some of these pathway perturbations were also reported in nonhuman
primates fRager etal.. 20131.
In nasal tissues from acutely exposed nonhuman primates, significant induction of
miR-125b and suppression of miR-29a were observed (Rager etal.. 2013: Swenbergetal.. 20131.
Expressions of several candidate mRNA targets of miR-125b were also decreased in this study,
consistent with miR-125b induction, including two that were also reported to be affected in
subchronically exposed rats (Andersen etal.. 20101 (see Appendix A.5.6). In analogous rat nasal
tissues, expression of several members from the growth-suppressing miRNA family let-7 decreased
following subchronic exposure (Rager etal.. 2014). consistent with observations from exposed
A549 lung carcinoma cells (Rager etal.. 2011). Decreased expression of let-7 family members was
found in nasopharyngeal carcinomas compared with healthy tissue (Li etal.. 2011). and this effect
has been reported to promote proliferative and oncogenic cellular signaling pathways in
respiratory tract cancers flakopovic etal.. 20131. Despite the numerous significant changes in
miRNA expression levels reported following formaldehyde exposure, miR-203 was the only target
reported to be similarly affected (decreased) in analogous nasal tissue from both rats and
nonhuman primates (Rager etal.. 2014: Rager etal.. 2013) (see Appendix A.5.6). Overall, changes
in expression of these miRNAs are generally consistent with observations in human lung, prostate,
breast, and bone marrow cancers (Garzon etal.. 2009: Ma and Weinberg. 2008: Fabbri etal.. 2007).
The abundance of highly significant changes in specific targets within individual arrays or
experiments, but limited concordance across expression array data sets or species, is not unusual;
however, it greatly complicates interpretation and integration of various data streams fWeinberg.
20141.
This document is a draft for review purposes only and does not constitute Agency policy.
1-329 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
DNA methylation and histone modification can promote carcinogenesis through steric
regulation of enhancer/promoter binding and transcription factor-DNA association, thereby
affecting gene transcription fVaissiere etal.. 20081. DNA methylation was globally decreased in
human bronchial epithelial cells exposed to formaldehyde in vitro for up to 24 weeks, which may
have been mediated by the down-regulation of de novo methyltransferase genes fLiu etal.).
Formaldehyde may affect gene transcription via posttranslational modification (PTM) of histone
proteins, in part by directly adducting unmodified lysine residues in histones to form
N6-formyllysine, thus preventing acetylation of this residue fEdrissi etal.. 2013a: Lu etal.. 20081.
Such irreversible adduction could interfere with transcriptional activation, nucleosome
organization fWisniewski et al.. 20081. and DNA lesion repair activity fLuch etal.. 20141. Levels of
these formylated lysine adducts increase in a concentration-dependent manner in the URT of rats
exposed to >0.9 mg/m3 fEdrissi etal.. 2013bl. levels at which increased DPXs are also observed
(see Table 1-39, and Appendix A.4). In addition, exogenous formaldehyde can induce histone
phosphorylation through activation of MAP kinase signaling in vitro fYoshida and Ibuki. 20141. In
A549 cells, as histone serine phosphorylation increased, lysine acetylation levels correspondingly
decreased, providing an additional (indirect) mechanism by which exogenous formaldehyde
attenuates histone acetylation and potentially modulates gene transcription. c-Jun N-terminal
protein kinase (JNK) was the primary regulator of this histone phosphorylation, which led to
elevated nuclear c-Fos and c-Jun protein expression (Shi etal.. 2014: Yoshida and Ibuki. 2014).
Together, c-Fos and c-Jun comprise the transcription factor AP-1, which can play an early role in
human respiratory tract carcinogenesis (Karamouzis etal.. 2007). Likewise, increased histone
phosphorylation may be an important mechanism specifically in human nasopharyngeal
carcinogenesis fLi etal.. 2013al. suggesting that these epigenetic effects may play a causal role in
human URT cancer formation.
The existing evidence illustrates myriad time- and concentration-dependent effects
following formaldehyde exposure, indicating the potential for both direct and indirect impacts on
transcriptional activity, in addition to inhibiting protein translation via miRNA dysregulation. What
is lacking, however, are conceptual paradigms and computational strategies for integrating systems
and cancer biology data streams (Weinberg. 2014). While provocative, in the absence of direct
hypothesis evaluation and more explicit phenotypic anchoring, the causal contribution of
epigenetic effects to URT carcinogenesis cannot be evaluated independently from the primary
mechanistic considerations outlined above.
Mode of action evidence integration and summary of analysis
Prolonged inflammation or irritation to the nasal mucosal surface has been associated with
squamous metaplasia of the respiratory or transitional epithelium following exposure to infectious
agents such as fungi or bacteria, but such exposures did not result in neoplasia fBrown etal.. 1991:
Monticello etal.. 1990bl. Likewise, chemical URT irritants such as dimethylamine, glutaraldehyde,
ethylacrylate, hydrogen chloride, and chlorine gas cause rhinitis, inflammation, and cytotoxicity
This document is a draft for review purposes only and does not constitute Agency policy.
1-330 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
leading to squamous metaplasia or hyperplasia, but do not induce rat nasal tumors following
chronic exposure fNRC. 2014b: Mcgregor et al.. 2006: Wolfet al.. 1995b: Buckley etal.. 1985:
Sellakumar etal.. 1985: Albert etal.. 19821. However, a number of genotoxic chemicals that also
induce pathological changes in the rat nasal epithelium similar to formaldehyde (e.g., acetaldehyde,
acrolein, 4-[N-methyl-N-nitrosamino]-l-[3-pyridyl]-l-butanone [NNK] and 1,2-epoxybutane) also
induce nasal tumors including SCCs and PA-like lesions (NTP. 2011: U.S. EPA. 2003: Monticello et
al.. 1993: Monticello etal.. 1990b: NTP. 1988: Woutersen et al.. 19861. The comparison between
formaldehyde and glutaraldehyde is particularly informative, as similar rat nasal cytotoxic
pathology (e.g., squamous metaplasia, hyperplasia, inflammation) is elicited by exposure to both
aldehydes fHester etal.. 20051. and yet glutaraldehyde exposure does not induce rat nasal tumors
even after 24 months of exposure, while such tumors are induced following >12 months of
formaldehyde exposure fMcgregor et al.. 20061. It has been proposed that glutaraldehyde exposure
causes more epithelial cell death in the nasal mucosa compared with formaldehyde, possibly
resulting in part from the greater inability of cells to repair or otherwise resolve any
glutaraldehyde-DNA adducts (Mcgregor etal.. 2006: Hester etal.. 20051. The observation that a
more effectively cytotoxic but less effectively mutagenic agent, glutaraldehyde, induces similar
cytotoxicity-induced regenerative URT pathology to formaldehyde, yet appears unable to elicit rat
URT tumors, suggests that cytotoxicity-induced regenerative proliferation alone is insufficient to
induce URT carcinogenesis resulting from formaldehyde exposure.
The underlying balance between formaldehyde-associated cytotoxicity and genotoxicity
may not only be responsible for the induction of these rare URT tumors in rats, but may also be key
to the difference in phenotype between formaldehyde-induced nasal squamous metaplasia and that
normally encountered in the aging rat. Gamma-glutamyl transpeptidase activity, present in normal
and metaplastic epithelium in unexposed animals, is absent in the frequently atypical squamous
metaplasia associated with formaldehyde exposure fDinsdale etal.. 1993: Brown etal.. 19911. Such
atypical squamous metaplasia (i.e., dysplasia) has been noted as a possible precursor to SCC in the
rat URT (Monticello etal.. 1990b). Together with the above, several lines of evidence converge to
support the conclusion that while inflammation, squamous metaplasia, or hyperplasia alone are
clearly not sufficient to induce nasal cancer in rats (Monticello etal.. 1993). the amplified cellular
proliferation occurring in regenerating tissues may be a mechanism by which genotoxicity-induced
DNA mutation rates are augmented, facilitating neoplastic transformation. The marked increase in
formaldehyde-initiated clones observed in vitro following growth stimulation by
12-0-tetradecanoylphorbol-13-acetate (TPA) in two-stage transformation studies (Boreiko and
Ragan. 1983: Ragan and Boreiko. 1981) is also consistent with this conceptual model.
Strong and consistent evidence for formaldehyde-induced direct genotoxicity and
mutagenicity comes from studies in mammalian cell lines, controlled inhalation studies in rodents
and nonhuman primates, and occupationally exposed humans, wherein mutagenicity anatomically
coincides with and temporally precedes URT tumorigenesis. Strong and consistent evidence
This document is a draft for review purposes only and does not constitute Agency policy.
1-331 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Toxicological Review of Formaldehyde—Inhalation
associates URT tissue pathology of increasing severity and regenerative proliferation with
squamous cell carcinoma (SCC) formation in experimental rodent studies at moderate-to-high
exposure levels, consistent with some measurements of cytotoxicity reported in analogous nasal or
buccal tissues from formaldehyde-exposed humans (see Table 1-43). Experimental evidence also
links polypoid adenoma (PA) formation to formaldehyde exposure in several rat strains that also
develop SCCs, and limited evidence associates increased PA incidence across a range of exposure
concentrations in F344 rats. Limited evidence from a subset of experimental rodent studies also
supports nasal epithelial cell proliferation in the absence of significant epithelial tissue pathology
following acute, discontinuous, or moderate concentration exposure scenarios; however, while
even intermittent proliferative stimuli could promote the growth of both nascent and malignant
clones, the specific role for formaldehyde-induced cellular proliferation as an effect independent
from either concomitant genotoxicity or tissue pathology remains undetermined. Evidence
supporting the URT cancer MOA depends not only on temporality, duration, and concentration of
exposure, but also anatomical location within the URT (i.e., incidence or severity of all primary
mechanistic considerations decreases following an anterior-to-posterior gradient within the URT).
While significant evidence supports some association between formaldehyde exposure and
immune disease or dysfunction, including chronic inflammation and increased oxidative stress, the
existing database is not sufficient to evaluate the independent contribution of these effects to URT
carcinogenesis. Likewise, while formaldehyde appears to inhibit various cellular DNA repair
pathways, the independent contribution of this effect to URT carcinogenesis remains to be
determined.
There is sufficient evidence to conclude that formaldehyde induces URT carcinogenicity via
at least two primary mechanistic considerations: genotoxicity-associated mutagenicity and
cytotoxicity-induced regenerative proliferation. By means of its fundamentally mutagenic activity,
formaldehyde damages DNA and increases the mutational burden of the URT mucosa when this
damage is not adequately repaired, while mucosal cytotoxicity creates a tissue microenvironment
driving continuous proliferation, facilitating the accumulation of mutations arising from both direct
and indirect genotoxicity, thereby increasing the rate at which initiated clones are formed as well as
stimulating the expansion of existing neoplastic colonies (see Table 1-43). The involvement of both
genotoxicity- and cytotoxicity-induced proliferation in the URT MOA is internally consistent with
the available formaldehyde evidence, and is also externally consistent with the described activities
of other reported URT toxins and carcinogens.
This document is a draft for review purposes only and does not constitute Agency policy.
1-332 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-42. Summary considerations for upper respiratory tract (URT)
carcinogenesis
Hypothesized
mechanistic
event
Experimental support for
mechanistic event
Human relevance
Weight-of-evidence
conclusion and biological
plausibility
Direct genotoxicity
and mutagenicity
(see Table 1-39)
• 'T* MN incidence in URT
mucosa from human students
and workers following
subchronic-to-chronic
exposure
• ^ DPX and/or hmdG adducts
in URT tissues of rhesus or
cynomolgus monkeys,
following acute exposure
• ^ DPX or hmdG adducts and
accumulation in URT tissues of
F344 rats following acute to
subchronic exposure
• No effect on MN incidence
URT tissues of F344 rats follow
subchronic exposure
Yes. Markers of direct
genotoxicity correspond
anatomically and temporally
with subsequent URT
neoplasia in experimental
animal models, are consistent
with increased MN induction
following exposure in
humans, and are presumed
relevant to human
carcinogenesis.
Strong and consistent evidence for
formaldehyde-induced direct
genotoxicity and mutagenicity
exists from both experimental
animal models and human
molecular epidemiology to support
a significant role for mutagenicity
in URT carcinogenesis.
Cytotoxicity-
induced
regenerative
proliferation
(see Tables 1-40
and 1-41)
• \1/ Nasal mucociliary function,
'T* nasal hyperplasia,
keratinization and/or
squamous metaplasia, URT
rhinitis, irritation, and
inflammation in humans
following acute to chronic
exposure
• \1/ Nasal cilia content, 'T*
hyperplasia and squamous
metaplasia in URT tissues from
monkeys following acute to
subchronic exposure
• Associated with 'T* URT cell
proliferation in rhesus
monkeys
• \1/Nasal mucociliary function,
'T* nasal rhinitis, hyperplasia
and squamous metaplasia
and/or dysplasia in various rat
strains and B6C3F1 mice
following acute to chronic
exposure
• Associated with 'T* URT cell
proliferation rats and mice
Yes. Increasing incidence or
severity of URT dysfunction or
pathology is positively
associated with formaldehyde
exposure in humans,
nonhuman primates, and
rats. A continuum of similar
epithelial pathology is
observed across affected
species at POE tissues, and
therefore the resulting
increased cellular turnover
observed in experimental
models is presumed relevant
to human carcinogenesis.
Strong and consistent evidence
exists which associates the nasal
epithelial pathology-driven
proliferation with SCC abundance
following formaldehyde exposure
in rodent experimental models to
support a significant role for
regenerative proliferation in URT
carcinogenesis.
Cellular
mitogenesis in the
absence of
cytotoxic tissue
pathology
(see Table 1-41)
• Clear evidence of 'T* URT cell
proliferation under conditions
also resulting in tissue
pathology in rhesus monkeys
• Exposure to subcytotoxic
concentrations not evaluated
• Clear evidence of 'T* URT cell
proliferation under conditions
Yes. Cellular proliferation
may be increased at lower
exposures and/or following
shorter durations of exposure
than that eliciting tissue
pathology, which suggests
that mitogenesis may be
directly stimulated by
Limited and inconsistent evidence
associates cellular proliferation
with formaldehyde exposures
below those eliciting cytotoxic
pathology in the rat nasal
epithelium, which precludes a
determination as to the
This document is a draft for review purposes only and does not constitute Agency policy.
1-333 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
Hypothesized
mechanistic
event
Experimental support for
mechanistic event
Human relevance
Weight-of-evidence
conclusion and biological
plausibility
also resulting in tissue
pathology in Wistar and F344
rats (>4 mg/m3)
• Suggestive evidence of 'T* URT
cell proliferation under
conditions not clearly causing
tissue pathology (<4 mg/m3;
see Appen dix A.5.6)
formaldehyde exposure.
Proliferation is expected to
accelerate and enhance
carcinogenesis in both
humans and model systems,
and is therefore presumed
relevant to human
carcinogenesis.
importance of this phenomenon in
URT carcinogenesis.
Oxidative stress,
immune disease
and dysfunction in
the URT (see
Appendix A.5.6)
• 'T* LRT infection frequency,
inflammation, allergic
outcomes in children; 'T*
leukocyte activation, allergy
symptoms, chronic URT
inflammation and \|/ infection
resistance in adult workers
following subchronic-chronic
exposure
• 'T* LRT oxidative stress,
markers of inflammation and
leukocyte recruitment in rats
and mice; 'T* airway wall
thickening or remodeling in
mice and rats following OVA
sensitization
• 'T* Malignancy and neutrophil
involvement of lung
metastases, \|/ lung NK cell
numbers and activity in
C57BL/6 mice
Yes. Nasal infection, markers
of persistent inflammation
and/or immune dysfunction
are positively associated with
a range of formaldehyde
exposure in both humans and
rodents. Oxidative stress and
chronic inflammatory
diseases, including
immunosuppression, are
presumed relevant to human
carcinogenesis. The
relevance of other immune
system dysfunctions to
human carcinogenesis, such
as allergy, is less clear.
While significant evidence exists
supporting oxidative stress, chronic
inflammation and various immune
dysfunctions following
formaldehyde exposure in humans
and experimental animal models
(see Appendix A.5.6), the evidence
supporting associations between
these effects and URT
carcinogenesis is insufficient to
evaluate the contribution of these
effects independently in either
humans or experimental animal
models.
Mode of action conclusions for URT cancers
Support for the hypothesized mode of action in experimental animal models
Strong, consistent evidence from rodent and nonhuman primate models supports the role
for both direct (i.e., potentially DPX or hmDNA adduct-associated) mutagenicity, as well as indirect
genotoxicity, mutagenicity, and regenerative proliferation resulting from respiratory tissue
pathology, in rodent URT carcinogenesis. DNA labeling studies in rodent nasal epithelium suggest
that cell division may also accelerate in response to marginally cytotoxic tissue concentrations
resulting from short-term, lower level, or discontinuous exposure scenarios, although this evidence
was neither strong nor consistent across similar studies and model systems. Observations of
mutagenicity, cytotoxic epithelial pathology, and proliferation correspond histologically,
anatomically, temporally, and dose-responsively with subsequent SCC and PA formation, consistent
with contribution of both mutagenesis and regenerative proliferation to rodent URT carcinogenesis
following formaldehyde exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
1-334 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Relevance and applicability of the hypothesized mode of action to human cancer
Mutagenicity is presumed to be a relevant component of URT carcinogenesis in humans,
supported by strong evidence of direct genotoxicity in both rodent and nonhuman primate
experimental models and consistent observations of direct genotoxicity and mutagenicity from
human epidemiological studies. Increased nasal epithelial cell proliferation (in rats and nonhuman
primates) coincides anatomically with dysplastic lesions found in tissues from similar species, as
well as with progressive, proliferative lesions in the nasal/buccal epithelium and nasopharynx of
chronically exposed humans. This cross-species concordance, combined with the observation that
cellular proliferation may be induced at lower exposures or following shorter durations of exposure
than those eliciting tissue metaplasia, suggests that cellular proliferation in the presence of
marginal tissue toxicity may also be potentially relevant to human URT carcinogenesis, as this
episodic exposure scenario may be more frequently encountered in human populations than the
continuous, chronic high-level exposures traditionally employed in rodent cancer bioassays.
Increasing incidence or severity of nasal dysfunction and progressive pathology is associated with
escalating formaldehyde exposure concentration or duration in humans, nonhuman primates, and
rats. While POE tissue sensitivity to formaldehyde toxicity may quantitatively differ in humans
versus rats and other rodents, qualitatively similar nasal dysfunction and pathology consistent with
preneoplastic stages of cancer progression are observed across analogous tissues from all affected
species, and therefore conclusions derived from these model systems are presumed relevant to
human URT carcinogenesis. Given this presumed relevance, the potential for an increased
susceptibility of specific human populations to developing URT cancers can be informed by both the
human data and relevant mechanistic evidence from experimental model systems
(see Section 1.4.1).
In general, URT findings in animals are found to be relevant to the URT cancer types and
locations observed in humans despite significant differences in the occurrence of the individual
cancer types. Firstly, site concordance is not required (U.S. EPA. 2005a). Secondly, the lack of a clear
site-specific correspondence may be attributed to large interspecies differences in anatomy and
airflow which in turn dictates formaldehyde distribution.
Regarding human NPC, the observed formaldehyde exposure-induced nasal tumors and
mechanistic changes in animals are considered directly applicable to interpreting changes in the
human nasopharynx. The nasopharynx is part of the nasal cavity and a recognized target of inhaled
nasal toxicants across species (Chamanza and Wright. 2015).
Similarly, the URT MOA is considered relevant and applicable to the interpretation of
human SNC, although some uncertainties remain. Across species, the sinuses are positioned close
to the nasal cavity and encounter inspired air (Reznik. 1990). Analyses of sinonasal cancer cases
indicate that most sinonasal cancers are squamous cell carcinomas (the primary tumor type in
animals) and the upper nasal cavity is generally the primary site of tumor occurrence, in more than
40% of cases (the maxillary sinus is the next most common site), although it is often difficult to
This document is a draft for review purposes only and does not constitute Agency policy.
1-335 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
pinpoint the exact anatomical location from which the cancers developed (Duttaetal.. 2015:
Llorente etal.. 2014: Turner and Reh. 20121. While these similarities support the relevance of the
animal data to human SNC, it is necessary to consider the anatomy of the rodent and human URT
given the importance of the distribution of inhaled formaldehyde and, as compared to the
nasopharynx and other parts of the nasal cavity, a reduced flow of inspired air reaches sinonasal
regions and the sinuses specifically (via narrow channels from the nasal cavity) (Kumar etal.. 2016:
Xiong etal.. 20081. Although tumors in the sinonasal regions of exposed rodents or monkeys were
not observed, this may be partially explained by differences in anatomy. Specifically, while humans
have four paranasal sinuses, rodents and monkeys only have one, and the sinus in rodents is much
smaller, thus presenting a smaller target for potential cancer development and a reduced capacity
for detection as compared to in humans. Additional uncertainties in drawing interpretations across
species include differences in airflow and tissue/cellular composition, which cannot be easily
evaluated. Taken together, while there is some uncertainty in the applicability of the MO A to SNC,
the mechanistic evidence (as well as the evidence on nasal cancers in animals) is interpreted as
applicable to and supportive of human SNC.
The hypopharynx and oropharynx, and to a greater extent the larynx, are more distal from
the POE than the nasopharynx and sinonasal tissues. Oronasal breathing in humans, as compared to
nasal-only breathing in rodents, may suggest a greater relevance of tissue sites close to the oral
cavity for human exposure; thus, mechanistic changes in rostral parts of the URT (i.e., the nasal
cavity) in animals may be more relevant to human oropharyngeal cancer. In general, however,
based on the known reactivity and distribution of inhaled formaldehyde, a greater level of
uncertainty in the applicability of the animal nasal findings is inferred for these human cancer
types, most notably laryngeal cancer, as compared to NPC or SNC.
Utility of mechanistic data for informing hazard quantification decisions
Since strong and consistent evidence supports the contribution of both direct genotoxicity
and mutagenicity as well as cytotoxicity-induced regenerative proliferation as primary mechanistic
considerations relevant to the pathogenesis of formaldehyde-associated URT cancer in rodents,
mechanistic data relevant to these endpoints may be useful for informing quantification of nasal
cancers in experimental animals following chronic formaldehyde exposure. In particular,
quantitative evaluation of these mechanisms may inform a biological response basis for guiding
dose-response extrapolations of rodent SCCs, as described in Section 2.2.1.
Integrated Summary of Evidence for Upper Respiratory Tract Cancers
Table 1-43 summarizes the evidence integration judgments and supporting rationale for the
individual URT cancers.
Epidemiological findings provide robust evidence for nasopharyngeal cancers (NPCs), based
on groups with occupational exposure. Consistent increases in NPC risk were reported by
numerous high and medium confidence studies involving occupational exposure to formaldehyde
This document is a draft for review purposes only and does not constitute Agency policy.
1-336 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
among diverse populations in different geographic locations and exposure settings that accounted
for expected temporal relationships for cancer induction and progression, with several reporting a
large magnitude of relative risk (RR >3). A dose-response gradient was reported for various
measures of exposure, including cumulative exposure, duration of exposure, and peak exposure.
Robust evidence for nasal cancers is provided from studies in experimental animals (rats and mice).
In animals, the incidence of lesions, as well as the tumor invasiveness and latency, was reproducibly
shown to worsen with increasing formaldehyde exposure level. The distribution of tumors was
dependent on duration of exposure as well as formaldehyde concentration. Mechanistic changes
associated with the development of cancer in the nasal cavity were consistently observed in
humans and experimental systems, including genotoxicity, epithelial damage and proliferation, and
eventual cancer development in relevant URT tissues. The mechanistic changes and URT lesions
exhibited a temporal and dose-response relationship coherent with carcinogenesis and supportive
of a mutagenic MOA (see Evidence on MOAfor upper respiratory tract cancers). The observed
formaldehyde exposure-induced nasal tumors and mechanistic changes in animals are considered
directly relevant to changes in the human nasopharynx (the nasopharynx is part of the nasal cavity
and a recognized target of inhaled nasal toxicants). Thus, based on robust human evidence, robust
animal evidence, and mechanistic evidence supporting a mutagenic MOA for NPC, the evidence
demonstrates that formaldehyde inhalation causes nasopharyngeal cancer in humans, given
appropriate exposure circumstances. This conclusion is primarily based on studies of groups
exposed to occupational formaldehyde levels and coherent findings in animals, with tumors in
rodents generally only observed at formaldehyde concentrations above 6 mg/m3.
Epidemiological findings also provide robust evidence for sinonasal cancer (SNC), based on
groups with occupational exposure. The robust judgment for SNC is supported by a smaller set of
epidemiological studies than for NPC, although a large, pooled analysis of 12 casecontrol studies
included a large number of cases and greater detail on formaldehyde exposures, which increased
confidence. This study observed an increasing trend in risk for adenocarcinoma with higher
cumulative exposure among men and women in analyses that controlled for key confounders
including exposure to wood dust. The studies were conducted in different geographic locations and
exposure settings that accounted for expected temporal relationships for cancer induction and
progression. Rodent nasal cancers and related mechanistic changes in the nasal cavity are
considered relevant to human SNC (see discussion in Evidence on MOAfor upper respiratory tract
cancers), although some uncertainty in their applicability to SNC, as compared to NPC remains, and
thus judgments of both robust and moderate animal evidence were considered. Ultimately, given
this uncertainty in applicability, while the animal and mechanistic evidence cited for NPC is judged
as informative and supportive for interpreting SNC, including providing sufficient support for a
mutagenic MOA for this cancer type, the animal evidence overall is interpreted as moderate rather
than robust. Based on robust human evidence, moderate animal evidence, and mechanistic evidence
supporting a mutagenic MOA for SNC, the evidence demonstrates that formaldehyde inhalation
This document is a draft for review purposes only and does not constitute Agency policy.
1-337 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 causes sinonasal cancer in humans, given appropriate exposure circumstances. This conclusion is
2 primarily based on studies of groups exposed to occupational formaldehyde levels.
3 For oropharyngeal/hypopharyngeal cancers, the human evidence is slight, based on data
4 from highly exposed workers, and slight animal evidence is provided from relevant observations of
5 preneoplastic lesions and mechanistic changes. Taken together, the evidence suggests, but is not
6 sufficient to infer, that formaldehyde inhalation might cause oropharyngeal/hypopharyngeal
7 cancers given appropriate exposure circumstances.
8 The human and animal evidence is indeterminate for laryngeal cancers and, overall, the
9 evidence is inadequate to determine whether formaldehyde inhalation may cause this cancer.
Table 1-43. Evidence integration summary for effects of formaldehyde
inhalation on URT cancers
Evidence
Evidence judgment
Hazard determination
Nasopharyngeal cancer (NPC)
Human
evidence
Robust, based on:
Human health effect studies:
•Consistent increases in risk across numerous high, medium and low
confidence studies
•Very strong associations (eight studies reported at least a threefold increase
in risk for some exposure categories, three of the eight were of high or
medium confidence, direction of potential bias toward the null)
• Evidence of exposure-response relationships across multiple measures of
increased exposure
• A temporal relationship consistent with causality (i.e., allowing for cancer
induction, latency and mortality)
Biological Plausibility:
Although not as strong as the animal database of mechanistic studies,
mechanistic evidence from human studies indicates a clear biological
relationship with genotoxicity, epithelial damage and proliferation, and
eventual cancer development in relevant URT tissues
The evidence demonstrates
that formaldehyde inhalation
causes nasopharyngeal cancer
in humans, given appropriate
exposure circumstances3
Primarily based on studies of
groups of workers exposed to
occupational formaldehyde
levels, coherent findings in
animals (with tumors in rodents
generally only at formaldehyde
levels above 6 mg/m3), and a
well-supported MOA for nasal
tumor development
Animal
evidence
Robust, based on:
Animal health effect studies:
• Tumors of the respiratory tract (predominantly nasal squamous cell
carcinomas, SCCs, but including other epithelial and nonepithelial tumors)
were consistently observed in mice and in several strains of rats in
numerous high and medium confidence studies, but not in hamsters,
generally at formaldehyde levels above 6 mg/m3.
• The lesions progressed to more posterior locations with increasing duration
and concentration of formaldehyde exposure
• The development of these lesions, particularly the SCCs, depended on the
duration of observation and, based on an increasing incidence and severity
of lesions in animals exposed for longer periods of time, the formaldehyde
exposure duration. Most notably, the lesion incidence, as well as the tumor
invasiveness and latency, was reproducibly shown to worsen with increasing
formaldehyde exposure level.
Biological Plausibility: Mechanistic changes consistent with cancer
development in nasal tissues were observed across species, including rats,
mice, and monkeys. In F344 rats chronically exposed to formaldehyde, a clear
Potential Susceptibilities: There
is very little evidence to
evaluate the potential risk to
sensitive populations and/or
lifestages. However, several
animal studies suggest that
prior damage to the nasal
epithelium might increase the
development of cancer in these
damaged regions.
This document is a draft for review purposes only and does not constitute Agency policy.
1-338 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
temporal, dose-responsive, and biological relationship was observed in the
appearance of genotoxicity, sustained epithelial damage, cellular proliferation,
and eventual tumor development.
Other
Inferences
• Relevance of the animal evidence to human NPC. The types of findings were
consistent and coherent across species (including humans). Although site
concordance is not essential ("U.S. EPA. 2005al considering the anatomy
of the rodent and human URT and the importance of the distribution of
inhaled formaldehyde, the observed formaldehyde exposure-induced nasal
tumors and mechanistic changes in animals are considered directly relevant
to changes in the human nasopharynx.
• /WO/4: Together, genotoxicity, cellular proliferation, and cytotoxicity-
induced regenerative proliferation exhibit multiple layers of coherence as a
function of species, anatomy, temporality, concentration, and duration of
exposure, and when integrated, form a biologically relevant MOA for
formaldehyde-induced URT carcinogenesis (U.S. EPA. 2005a). While the
chronic formaldehyde exposure concentrations reported to elicit nasal
cytotoxic pathology appear to be higher in the rats and nonhuman
primates evaluated experimentally (>4 mg/m3), compared with the results
from human epidemiological cohorts (>0.3 mg/m3), formaldehyde-
associated genotoxicity has been induced in analogous POE tissues from
rats, nonhuman primates and humans exposed similarly (<0.9 mg/m3).
Sinonasal cancer (SNC)
Human
evidence
Robust, based primarily on:
Human health effect studies:
• Consistent increases in risk across a set of medium and low confidence
studies; four (2 medium and 2 low confidence) studies reporting at least a
threefold increase in risk, primarily for adenocarcinoma, including the
largest study, a pooled analysis of 12 case-control studies, demonstrating a
clear exposure-response relationship.
• Increased risk of lower magnitude reported by two other medium
confidence studies.
• Null results in 3 insensitive low confidence studies.
Biological Plausibility: The human mechanistic evidence cited for NPC is
informative and supportive for interpreting the biological plausibility of SNC
(see discussion in MOA analysis).
The evidence demonstrates
that formaldehyde inhalation
causes sinonasal cancer in
humans, given appropriate
exposure circumstances3
Primarily based on studies of
groups of workers exposed to
occupational formaldehyde
levels. Although less certain
than the support provided for
NPCs, animal and MOA
evidence provide support for
the human evidence.
Potential Susceptibilities: There
is very little evidence to
evaluate the potential risk to
sensitive populations and/or
lifestages. However, several
animal studies suggest that
prior damage to the nasal
epithelium might increase the
development of cancer in these
damaged regions.
Animal
evidence
Moderate, based on:
Animal health effect studies:
(Same evidence base as for NPC above; see "Other inferences, relevance of the
animal evidence to human SNC" for justification)
• Note: tumors were not reported in the maxillary sinus of exposed animals
Biological Plausibility:
(Same mechanistic evidence base as for NPC above)
• Although infrequently examined, studies that measured noncancer lesions
in the maxillary sinus did not detect treatment-related respiratory tract
pathology, although cell proliferation was observed (see Section 1.2.4).
• Although also poorly studied, some mechanistic changes consistent with
the MOA for nasal cancers, including increased DPX in the monkey maxillary
sinus, have been observed.
This document is a draft for review purposes only and does not constitute Agency policy.
1-339 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Other
Inferences
• Relevance of the animal evidence to human SNC. The types of findings were
consistent and coherent across species (including humans). The strong
animal and mechanistic evidence for nasal cancers across species is
interpreted to provide moderate evidence supportive of sinonasal cancer (a
judgment of moderate rather than robust reflects some uncertainty in
interpreting the nasal cavity findings in animals as fully applicable to human
sinonasal cancer specifically; see discussion in MOA analysis).
• /WO/4: Similar to the inference above, although there is uncertainty in the
application of the identified MOA to SNC, the evidence overall is interpreted
to provide reasonable support for the mutagenic MOA asapplicable to SNC.
Oropharyngeal/ Hypopharyngeal cancer (OHPC)
Human
evidence
Slight, based on:
Human health effect studies:
• Increased risks in two of three medium confidence studies that evaluated
multiple metrics of exposure and reported three- to fivefold increases in
those highly exposed, including one which demonstrated clear exposure-
response relationships across several metrics
• However, little evidence of increases in risk (near the null) across one
medium and two low confidence results
Biological Plausibility: Although cells from exposed humans in tissues closely
apposed to the oropharynx and, more indirectly, the hypopharynx (e.g., buccal
cells) demonstrate mechanistic changes consistent with the development of
cancer, including genotoxicity, these data were not interpreted as sufficient to
further strengthen the human evidence judgment beyond slight.
The evidence suggests, but is
not sufficient to infer, that
formaldehyde inhalation might
cause oropharyngeal
/hypopharyngeal cancer, given
appropriate exposure
circumstances'5
Animal
evidence
Slight, based on:
Animal health effect studies:
• While most findings in animals were localized to the nasal cavity, some data
suggest that changes in more caudal (e.g., in the trachea) regions, including
evidence of dysplasia (a dedicated pre-neoplastic lesion) in one study, can
occur with very high formaldehyde exposures and/or different breathing
patterns (e.g., oronasal breathing in monkeys).
• Changes in the more caudal URT tissues most relevant to OHPC were
generally less direct indicators of cancer development, were less severe, or
occurred only at very high exposure levels.
Biological Plausibility: Mechanistic changes within caudal portions of the
rodent and monkey URT have been observed, and oronasal breathing in
humans (contrasting nasal-only breathing in rodents) infers an increased
potential relevance of mechanistic changes in rostral (anterior) regions of the
rodent to human OHPC. However, this was not interpreted as sufficient to
further strengthen the evidence judgment beyond slight.
Other
inferences
• Relevance of the animal evidence to human OHPC. While cancer site
concordance is not reauired for hazard determination (U.S. EPA. 2005al
given the known reactivity and distribution of inhaled formaldehyde, a
lesser level of confidence in the applicability of the animal nasal findings is
inferred for OHPC as compared to NPC or SNC.
• MOA: While aspects of the MOA for nasal cancers, including NPC and SNC,
may be operant for OHPC, the evidence overall is not interpreted to
provide reasonable support for a MOA that is relevant to OHPC.
Laryngeal cancer
Human
Indeterminate, based on:
There is inadequate evidence
This document is a draft for review purposes only and does not constitute Agency policy.
1-340 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
Human health effect studies:
• Suggestive associations reported in two medium confidence studies
• Inconsistent evidence on exposure-response relationships
• The moderate survival rate for laryngeal cancer mav indicate that mortalitv
data are not as good a proxy for incidence.
Biological Plausibility: Human mechanistic data specifically related to this
cancer type are lacking.
to determine whether
formaldehyde inhalation may
be capable of causing laryngeal
cancer in humans
Animal
Indeterminate, based on:
Animal health effect studies:
• No studies observed tumors in the rodent or monkey larynx, nor were
preneoplastic lesions such as dysplasia detected.
Biological Plausibility: The evidence for mechanistic changes specifically within
the larynx included findings in rodents and monkeys consistent with the MOA
for nasal cancers, specifically noncancer lesions (e.g., tissue damage,
hyperplasia, and squamous metaplasia) and genotoxicity (i.e., increased DPX).
Although these mechanistic changes alone could support a judgment of slight,
in the absence of experimental confirmation (or a biological understanding)
that these mechanistic changes are likely to lead to cancer or preneoplastic
lesions at sublethal formaldehyde concentrations, the animal evidence was
judged as indeterminate.
Other
inferences
• Relevance of the animal evidence to human laryngeal cancer. The
mechanistic changes observed in similar regions of the rodent and monkey
URT are considered relevant to the interpretation of laryngeal cancer.
• MOA: No potential MOA was identified for this cancer type.
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2.
bGiven the uncertainty in this judgment and the available evidence, this assessment does not attempt to define what might be
the "appropriate exposure circumstances" for developing this outcome.
1.3. SYNTHESIS OF EVIDENCE FOR NONRESPIRATORY EFFECTS
This section synthesizes research on nervous system effects (see Section 1.3.1),
developmental and reproductive toxicity (see Section 1.3.2), and cancer effects beyond the
respiratory tract (see Section 1.3.3), specifically in the lymphohematopoietic (LHP) system. Very
little information has been reported concerning cancer associations at other nonrespiratory sites
(e.g., brain; see Appendix A.5.9 for details). Evidence relevant to assessing carcinogenicity is
synthesized for LHP cancer subtypes in Section 1.3.3 (i.e., myeloid leukemia, lymphatic leukemia,
multiple myeloma, and Hodgkin lymphoma; note: non-Hodgkin lymphoma was not systematically
evaluated: see Appendix A.5.9).
1.3.1. Nervous System Effects
Numerous studies suggest that formaldehyde inhalation might result in noncancer nervous
system effects; however, the evidence across studies is generally weak and the database is
incomplete. Few studies in humans are available; formaldehyde exposure was reported to be
associated with neurobehavioral deficiencies as indicated by poorer performance in tests of
short-term memory and psychomotor responses, and with the motor neuron disease, amyotrophic
This document is a draft for review purposes only and does not constitute Agency policy.
1-341 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
lateral sclerosis (ALS). Observations in rodents include altered performance in tests of locomotion
and anxiety, and in learning and memory tests. In many of these animal neurobehavioral studies, a
confounding factor was introduced when test animals were exposed to the known neurotoxicant,
methanol, in formalin solutions. Experimental animal studies without methanol coexposure
suggest that repeated formaldehyde exposure may lead to amplified behavioral responses to
certain challenges (e.g., pharmacological), possibly through persistent modifications to neural
pathways. Similarly, studies from one laboratory suggest that developmental exposure to
formaldehyde at concentrations well above those causing adverse effects on the respiratory system
(see Sections 1.2.1-1.2.4) results in long-lasting changes in brain structure. To date, none of these
potential nervous system changes are supported by an experimentally verified mechanistic
hypothesis outlining how formaldehyde might elicit neurotoxicity without systemic distribution.
Overall, a definitive association between formaldehyde inhalation and neurotoxicity could not be
concluded. Most of the available experiments had significant study design deficiencies and
corroboration across the database was incomplete; thus, overall, the evidence suggests, but is not
sufficient to infer, the potential for formaldehyde inhalation to cause nervous system effects in
humans (i.e., based on slight evidence from human or animal health effect studies). Additional
research is needed to draw a more certain evidence integration judgment
Literature Search and Screening Strategy
Studies in humans or experimental animals examining the potential nervous system effects
of formaldehyde exposure were retrieved in a comprehensive systematic literature search of
PubMed, Web of Science, and ToxNet through September 2016 (see Appendix A.5.7), and a
systematic evidence map updating the literature through 2021 (see Appendix F). Human
(observational epidemiology or controlled exposure) studies of neurobehavioral tests or specific
neurological diseases were included. Studies of symptoms that may be associated with nervous
system effects (e.g., headache, fatigue) were not included due to the highly subjective nature of
these endpoints as compared to the other available data (these measures were primarily based on
self-administered questionnaires that varied in type and specificity), and because many of the
commonly reported symptoms are not necessarily specific to effects on the nervous system. In vivo
inhalation animal exposure studies were included, but in vitro studies and studies of other
exposure routes (e.g., oral, injection), including a multitude of studies using formaldehyde exposure
(typically hind paw or forepaw injections) as a model to study nociceptive (pain) behaviors in
rodents, were not included. These experiments are considered unlikely to reproduce the
distribution of formaldehyde and its metabolites following inhalation exposures (i.e., inhaled
formaldehyde has negligible distribution beyond the POE [see Appendix A.2], whereas other
exposure routes may allow for substantial distribution to nervous system tissues). In addition,
most of the oral and injection exposure experiments are confounded by methanol in the aqueous
formaldehyde formulations, reducing the ability of these experiments to attribute any observed
effects to formaldehyde. Unlike formaldehyde, methanol, a known neurotoxicant, is transported in
This document is a draft for review purposes only and does not constitute Agency policy.
1-342 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
the blood to nervous system tissues. In vitro studies possess the same limitations (i.e., direct
formaldehyde interaction with cells from nervous system tissues and methanol confounding).
Finally, studies examining nervous system effects (e.g., memory loss; neurodegeneration)
associated with increases in endogenous formaldehyde levels in the brain were identified by the
literature search but not deemed PECO-relevant. These studies were not included in this evidence
synthesis because formaldehyde inhalation does not appear to cause appreciable changes in
formaldehyde levels in nonrespiratory tissues such as the brain and no hypothesis currently exists
to explain how inhaled formaldehyde would affect endogenous formaldehyde levels in the CNS (see
Appendix A.2). However, similar to other health effects (see Section 1.3.3), studies suggesting that
CNS effects can result from reduced function of enzymes responsible for clearing formaldehyde
from relevant tissues (e.g., downregulated ALDH2 in fAi L. 2019: Tan etal.. 20181. highlight an area
of interest to future studies on potential susceptibility to inhaled formaldehyde exposure.
The bibliographic databases, search terms, and specific strategies used to search them are
provided in Appendix A.5.7, as are the specific PECO criteria. Appendix A.5.7 includes a literature
flow diagram that summarizes the results of the sorting process using these criteria and indicates
the number of studies that were selected for consideration in the assessment through 2016 (see
Appendix F for the identification of newer studies through 2021). These studies in animals and
humans were evaluated to interpret the quality and relevance of the study results for use in
interpreting the potential for formaldehyde exposure to cause neurotoxicity (see Appendix A.5.7).
Methodological Issues Considered in Evaluation of Studies
A key consideration for interpreting nervous system effects following formaldehyde
inhalation involves possible coexposure to methanol when aqueous formaldehyde solutions are
used as the test article. Findings in experimental studies describing the effects of formalin but not
controlling for methanol, and studies failing to indicate the formaldehyde source, are identified
throughout this section and automatically characterized as low confidence (at best); these studies
contribute very little weight to the evidence integration conclusions pertaining to the potential for
formaldehyde exposure to induce nervous system effects. Evaluation of the exposure protocol,
including consideration of the potential impact of irritant or odorant effects on behavioral
measures, was emphasized during study evaluations, contributing to the identification of some
studies as not informative for characterizing hazard. The database of studies evaluating the
potential for formaldehyde inhalation exposure to cause nervous system effects included very few
studies interpreted with medium or high confidence. Overall, studies were primarily of low
confidence and the majority of identified studies were interpreted as not informative for at least one
of the outcomes examined.
Nervous System Effects in Human Studies
The identified studies describing results of neurobehavioral tests, as well as the occurrence
or mortality from neurological disease are described in this section. These studies are summarized
This document is a draft for review purposes only and does not constitute Agency policy.
1-343 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Toxicological Review of Formaldehyde—Inhalation
in Tables 1-44 and 1-45. The tables are organized by study design (observational, acute controlled
exposure), confidence in study results, and publication year.
While several observational epidemiology and controlled exposure studies report nervous
system impairment in humans following exposure to formaldehyde, there are notable limitations in
the available data and the results from some of the studies are potentially confounded by
coexposures. Specifically, data from both observational and experimental studies showed an
association between formaldehyde exposure and impaired performance in neurobehavioral tests of
memory, dexterity, and psychomotor function (Lang etal.. 2008: Kilburn and Warshaw. 1992: Bach
etal.. 1990: Kilburn etal.. 1989: Kilburn etal.. 19871. In prospective studies from one research
group, Weisskopf et al. (20091 and Roberts et al. (20151 both noted an association between
formaldehyde exposure and death from the fatal motor neuron disease, ALS, in different study
populations in the United States; a separate case-control study from another research group in
Sweden also identified an association among individuals younger than 65 years of age, but not in
the overall analysis using national registry data (Peters etal.. 20171. A national registry-based
case-control study in Denmark by the same research group in the United States also observed an
association f Seals etal.. 20171. but a subsequent analysis using the same cases examining joint
effects by multiple health and chemical risk factors observed an inverse association in both men
and women, although only the latter reached statistical significance fBellavia etal.. 20211. Two
other studies failed to identify an association (Pinkertonetal.. 2013: Fang etal.. 2009). All of the
studies were limited by uncertainty in individual exposure assignments, except for the study by
Pinkerton et al. (2013). which evaluated a cohort of garment workers with known formaldehyde
exposure and detailed information on employment history. The cohort studies were limited by a
very low number of exposed cases.
This document is a draft for review purposes only and does not constitute Agency policy.
1-344 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Pinkerton (2013)
high; SMR
al t duration T "TSFC
Fang (2009)
medium; OR
Weisskopf (2009)
medium. RR
all T Drob. T duration all Fr> T duration
nr^i 1 nni—
Roberts (2015)
medium; HR; males
t probability
Bellavia (2021)
medium; OR
Peters (2017) \ Seals (2017,
medium; OR | medium; OR
T intensity lp all <75yr all T levels M F ever
i nnm—ir^ nnm nn n
10t
£ 1...
£
I
0.1-
140 255
8 2 3 3 1 0 7 20 2 9 9 7 8 5 4 36 22 4 5 13 55 43 2 2 51 47 2 47 39 2 323 I 30 53 422 I
_l I i i « » » ' « » ' ¦ ' ' I I I ¦ ¦ ¦ i i ¦ i ¦ i i i i i i I i i i I
ALS diagnosis or death
[cases or deaths in exposed group above x-axis]
Figure 1-28. Human studies of medium or high confidence examining the
potential for formaldehyde exposure to cause ALS.
Seven epidemiological studies of medium or high confidence were identified, all of which examined
potential associations with amyotrophic lateral sclerosis (ALS) [notes: a medium confidence, acute
controlled exposure study of neurobehavior, Lang et al. (2008), is not presented; results from Roberts et
al. (2015) are only presented for males; all results in females were null], Estimates of risk (i.e., odds ratios
[ORs], standardized mortality ratios [SMRs], relative risks [RRs], or hazard ratios [HRs]), 95% confidence
intervals (CIs), and number of exposed cases or deaths are presented for different comparisons within the
studies, including full cohort (e.g., ever/never exposed) comparisons (unlabeled) and comparisons across
multiple groups by: increasing duration, probability (prob.), time since first exposure (TFSE) [note: null
results comparing date of first exposure in Pinkerton et al. (2013) are not shown], or age-restricted
(e.g., younger than 65 years: <65). Different shapes reflect different research groups. Other
abbreviations: FD = full cohort comparison excluding persons not providing duration information;
Ip = maximum intensity in persons with a high probability of exposure compared to controls; M = males;
F = females; all = overall (full cohort comparisons).
This document is a draft for review purposes only and does not constitute Agency policy,
1-345' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Neurobehavioral tests
A series of epidemiology studies examined neurobehavior in histology technicians using
standardized test batteries designed to assess higher brain functions fKilburn and Warshaw. 1992:
Kilburn etal.. 1989: Kilburn etal.. 19871 (see Table 1-44). It is important to note that the majority
of formaldehyde exposure in this occupation is from formalin (containing methanol), which
introduced bias due to confounding of unknown magnitude and thus reduced the reliability of the
results for interpreting the effects of formaldehyde exposure. All of these studies were ultimately
considered to be of low confidence during study evaluation. Decreased performance in multiple
tests of memory and tests of dexterity, balance, coordination, motor control, and reaction time was
observed with increased daily hours of formaldehyde exposure fKilburn et al.. 1989: Kilburn etal..
19871. Although these workers were also exposed to solvents that can affect behavior (e.g., xylene),
hours of daily exposure to solvents was only correlated with decreased performance in a single
memory test (Kilburn etal.. 1989: Kilburn et al.. 19871. The effects of formaldehyde exposure on
neurobehavior were not verified when a comparable test battery was performed in a slightly larger
(350 versus 305 technicians), but possibly overlapping, study fKilburn and Warshaw. 19921. In
addition, a smaller group (n = 19) tested yearly over a 4-year period did not experience worsening
effects with continued work exposure, but this analysis did not specifically address formaldehyde
exposure fKilburn and Warshaw. 19921. These latter results suggest a lack of worsening effects
with cumulative exposure, but they did not incorporate a consideration of the relative magnitude of
exposure (e.g., hours of daily exposure to formaldehyde).
Three acute, controlled exposure studies evaluated performance in standardized
neurobehavioral tests (see Table 1-44). Two of these studies included multiple tests assessing
concentration, short-term memory, and motor control fBach etal.. 1990: Andersen and Molhave.
19831. while the third focused on decision reaction time fLang etal.. 20081. Although Bach et al.
(1990) reported decreased performance in multiple neurobehavioral tests following controlled
exposures at >0.480 mg/m3, particularly in workers with previous chronic formaldehyde exposure,
the exposure groups were not well matched for a number of variables relevant to test performance,
most of the responses were not concentration dependent, and distractibility due to possible
irritation cannot be ruled out (irritation measurements were subjective). In contrast to these
results, Andersen and Molhave (1983) indicated that they found no effects of exposure on
performance in cognitive tests, but the supporting data were not provided. Increased decision
reaction times in response to visual, auditory, or combined visual/auditory stimuli were observed
with exposure to 0.369 mg/m3 formaldehyde by Lang et al. (2008): the motor component of the
reaction times was unaffected by exposure. These increases were not observed at higher exposure
levels and did not exhibit the same dose-response pattern as effects on irritation; thus, additional
experiments are needed to better explain the findings.
Taken together, the epidemiological and human-controlled exposure studies provide mixed
results suggesting that formaldehyde exposure might be associated with deficits in performance in
This document is a draft for review purposes only and does not constitute Agency policy.
1-346 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 neurobehavioral tests related to memory, coordination, and motor control. However, the reliability
2 of these results is unclear and additional experiments are needed to clarify the potential
3 contributions of variables that are known to affect these measures, but which were poorly
4 controlled in these studies, including coexposures to neurotoxicants, irritation, and differences in
5 population characteristics such as age or education.
Table 1-44. Summary of alterations in neurobehavioral tests in relation to
formaldehyde exposure in observational epidemiology and controlled
exposure studies
Reference and study design
Exposure measures
Results
Observational epidemiology studies
Reference: Kilburn et al. (1989);
Kilburn et al. (1987) (United States)
Survey, n = 305 female histology
technicians attending histology
conference in Boston (167 of 658 in 1982,
25.4% or Anaheim (218 of 704, 31%, in
1983. Age 23-78 years, mean 40 years.
Work duration, mean 17 years. Seventy-
nine female referent laboratory
technicians in Los Angeles (participation
rate not reported).
Outcome: Neurobehavioral battery (10
tests) administered in 1 hour by trained
personnel.
Analysis: Multiple regression,
formaldehyde (hours) controlling for age,
education, smoking, home solvent
exposure and number of cover-slipped
slides.
Evaluation:3
SB IB
Cf Oth
Overall
Confidence
¦
Low
Potential selection bias (could be
influenced by perceived exposure and
effects), limited detail presented in
results.
Self-reported estimated formaldehyde
exposure (average 4.3 hr/d) and xylenes
(average 112 cover-slipped slides).
Most recent exposures were at least
several days prior.
Hr formaldehyde/day correlated with
number of slides/day, p < 0.05.
Source of formaldehyde is most likely
formalin (containing methanol).
Statistically significant association
(p < 0.05) between hr/d formaldehyde
exposure:
Recall memory (stories): One of two
tests
Visual memory (diagram): One of three
tests
Associative memory (digit span): One of
two tests
Dexterity (pegboard): One of one test
Balance (sharpened Romberg): One of
one test
Perceptual motor speed (trail making):
One of two tests
Age associated with performance
decrements in nine tests; solvent
exposure (# of slides cover-slipped)
associated with one test (p < 0.05)
No association with formaldehyde
observed for choice reaction time,
peripheral nerve function, or spatial
relation tests.
Reference: Kilburn and Warshaw
(1992) (United States)
Prospective study; histology technicians
attending histology conferences between
1982 and 1987; 19 histology technicians
tested yearly across 4 years (46-50 years
old); 299 technicians tested 2-3 times
across 4 years (44-47.9 years old); 350
histology technicians tested once
(38-40.4 years old); sex not reported.
Duration of formaldehyde exposure up
to 37 years.
Self-rated exposure scales.
Source of formaldehyde is most likely
formalin (containing methanol).
For single test analysis (n = 250),
formaldehyde exposure was not
associated with age-related change in
performance in tests encompassing
memory, cognition, pattern recognition,
dexterity, decision-making, motor
speed, or balance (beta and SE not
provided; reported as not statistically
significant). No decline seen in smaller
group (n = 19) tested across 4 years.
This document is a draft for review purposes only and does not constitute Agency policy.
1-347 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Exposure measures
Results
Outcome: 2-3 h neurobehavioral
battery; testers blinded to exposure
status.
Analysis: Multiple regression, adjusting
for age. Other variables considered were
sex, years of employment, smoking, and
nonoccupational exposures.
Evaluation:3
SB IB
Cf
Oth
Overall
Confidence
1 1
1 1
Low
1 1
1 1
Potential selection bias, limited detail
presented in results. Longitudinal
analysis limited by sample size and did
not specifically address formaldehyde
exposure.
Acute, controlled exposure studies
Reference: Lang et al. (2008)
(Germany)
N = 21 (of 26 volunteers selected based
on screening; five left study), 10 women,
11 men (results were combined), age 19-
39 years, healthy nonsmokers.
Exposure order randomly assigned;
double blinded. Ten 4-hour exposures,
one per day, over 10 days.
Outcome: Reaction times (Vienna Test
System) to visual and acoustic stimuli
measured before and after exposures.
Evaluation: Medium confidence.
Tested immediately after exposure.
Four hours in groups of four.
Formaldehyde levels3: Clean air, 0.185,
0.369, and 0.615 mg/m3; additional
0.369 and 0.615 mg/m3 with peaks up
to 1.23 mg/m3. Additional 0.0, 0.369,
and 0.615 mg/m3 with ethyl acetate
introduced as a "mask" for
formaldehyde. (Analytical
concentrations achieved were
measured, but not reported.)
Formaldehyde generated from
paraformaldehyde; ethyl acetate at 12-
16 ppm (irritant threshold of EA
reported at 20 ppm, identified from
scientific literature).
'T* in decision reaction time upon visual
stimulus at 0.3 and 0.3+ethyle acetate
(data presented graphically, p < 0.05).
'T* in decision reaction time upon
acoustic or audio-visual stimulus at
0.3 ppm only (data presented
graphically, p < 0.05; comparison group
for contrast not stated).
The motor speed component of the
decision reaction time was unaffected
by exposure.
Andersen and Molhave (1983)
(Denmark)
N = 16 healthy students, age 30-33,
68.8% male, 31.2% smokers, groups of
four over 4 days.
Exposure order determined by Latin
square design, blinding not indicated.
Outcome: Numerical addition: tested
3x/d (once in clean air; twice during
exposure); multiplication: tested lx/d
during exposure; card punching: tested
2x/d (once in clean air; once during
exposure).
Evaluation: Low confidence.
Tested during exposure; results not
reported.
Five hours; 0.3, 0.5,1.0, and 2.0 mg/m3
(analytical concentrations achieved
were not reported: indicated as within
20% of target concentrations).
Formaldehyde generation via thermal
depolymerization of paraformaldehyde,
dynamic chamber.
The study authors reported no change
in performance in addition (speed and
accuracy), multiplication, or transfer of
numbers to punch cards, but data were
not provided.
Reference: Bach et al. (1990)
(Denmark)
32 with occupational exposure to
formaldehyde (>5yr); age 18-64 years;
Formaldehyde concentrations
0, 0.15, 0.4, and 1.2 mg/m3 [analytical
concentrations achieved: 0.04, 0.21,
0.48, and 1.10 mg/m3].
Occupational group showed significantly
\1/ performance on the digit symbol test
(p < 0.025 for pooled exposure groups,
0, 0.15, and 0.4 compared to 1.2
This document is a draft for review purposes only and does not constitute Agency policy.
1-348 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Exposure measures
Results
selected from 108 workers (recruitment
and selection not described). Referent
group (n = 29 from 546 selected
randomly from a population registry);
attempted frequency matching by
average age, education, and smoking
prevalence but workers had higher
smoking prevalence and lower education
(detailed demographic data not
reported). Formaldehyde-exposed
excluded from referent group.
Exposure order by balanced Latin square
design; double blinded—Furfuryl
mercaptan (coffee aroma) used to mask
odor.
Outcome: Four performance tests twice
during exposure.
Evaluation: Low confidence.
Education and smoking imbalance in
workers and referents; tested during
acute exposure.
5.5 hr (0.5 hr pre-exposure in chamber
and gradual increase in formaldehyde).
Formaldehyde vapor generation not
reported; however, assumed to be from
depolymerization of paraformaldehyde
based on protocols used in the same
exposure chamber as reported by a
coauthor (Andersen and Molhave,
1983).
mg/m3); controls showed an inverse
relationship; digit span (p < 0.025) for
total digit sum in one of the six test
components—lowest scores in 0.4
mg/m3 group, and graphic continuous
line test (p < 0.05 only for the 0.4 mg/m3
group); effects were not dose-related.
Addition test: Dose-related performance
decrements (4< # of additions and
reaction time).
Data were presented graphically.
Matching was not completely
successful; due to last-minute
substitutions, the exposed workers,
particularly the 1.2 mg/m3 group, had a
lower education and different
proportion of smokers; the 1.2 mg/m3
group had a lower average age and
fewer smokers overall. Exposure groups
were not comparable.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.7). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
bFormaldehyde levels in the study converted to mg/m3 from ppm.
Nervous system disease
In a large and well-designed, prospective study of risk factors associated with amyotrophic
lateral sclerosis (ALS) mortality, years of self-reported exposure to formaldehyde was associated
with a 2.5-fold (95% CI 1.58, 3.86) increased mortality risk when examined across individuals
reporting duration data (this information was available for 22 of the 36 cases reporting
formaldehyde exposure) (Weisskopf et al.. 2009) (see Table 1-45). The overall risk was no longer
significantly elevated when individuals who reported exposure but did not report duration were
included in the analysis (all 36 cases; RR = 1.34; 95% CI 0.93,1.92). Risk increased with increasing
duration of formaldehyde exposure, with a fourfold risk seen with >10 years of exposure (13 cases).
In total, Weisskopf etal. (2009) followed 987,229 people and identified 1,156 ALS deaths (1,120 of
these cases reported that they were not exposed to formaldehyde), but formaldehyde intensity was
not assessed, and the duration of exposure was self-reported. A second study from the same
research group also identified some evidence of an association between formaldehyde exposure
and ALS death in a national study (Roberts etal.. 2015). An odds ratio (OR) of 4.43 was observed
among individuals with a high probability, high intensity exposure, based on only two cases of ALS;
no cases were observed among individuals with high probability, medium intensity exposure.
Formaldehyde exposure assignments were made by industrial hygienists using a job-exposure
This document is a draft for review purposes only and does not constitute Agency policy.
1-349 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
matrix with estimates of intensity and probability of exposure for the most recent job held by
participants, although duration was not assessed. More recently, two registry-based studies in
Sweden and Denmark observed associations of similar magnitude between ALS diagnosis and
occupational formaldehyde exposure analyzing all incident ALS cases occuring over a 20- to almost
30-year period. Both studies used a job-exposure matrix developed for the Nordic Occupational
Cancer Study (NOCCA) with exposure data specific to each country. The Swedish study observed no
association in the entire analytic group of blue-collar workers and farmers, however an odds ratio
of 1.28 (95% CI 1.02,1.61) was observed when the analysis was restricted to persons younger than
65 years of age fPeters etal.. 20171. In Denmark, occupational exposure to formaldehyde was
associated with ALS incidence in the entire cohort (RR 1.3, 95% CI 1.2,1.4) and associations of the
same magnitude were observed across all exposure quartiles in comparison to nonexposed fSeals
etal.. 20171. Hence neither study observed an (exposure-response trend. Also, the potential effect
of confounding by smoking on the formaldehyde—ALS association (Wang etal.. 2011: Armon.
20091 was not addressed. Paradoxically, the direction of the association was reversed when
investigators used a machine learning method to select joint predictors and interaction terms and
then included these health and chemical risk factors for ALS in the model fBellavia etal.. 20211. An
OR of similar magnitude but less precise than that reported by Peters et al. (2017) (OR = 1.3; 95%
CI 0.5, 3.2) was observed for participants with a high probability of exposure in a small case-control
study, although no association with exposure duration was observed (Fang etal.. 2009). Although
the longitudinal design of the prospective studies makes it unlikely that the association between
formaldehyde exposure and ALS death is attributable to some types of bias, a study with detailed
evaluations of formaldehyde exposure (probability, frequency) and duration of exposure in the
exposed populations failed to confirm an association fPinkerton etal.. 20131. Exposure in the
cohort of garment workers fPinkerton etal.. 20131. in particular, was more certain, based on
monitoring data in the 1980s, year of hire, and years of employment However, all of the studies,
except Peters et al. (2017) and Seals et al. (2017) were limited by small numbers of exposed cases,
which leads to decreased sensitivity to detect an association that might exist, or decreased stability
in effect estimates. Overall, evidence is emerging that formaldehyde exposure may pose a hazard
for ALS, but there is a large degree of uncertainty due to the mixed nature of the findings. As risk
factors for increased risk of ALS are complex and poorly defined, it remains possible that the
findings of Weisskopf et al. (2009), and the less robust but supportive findings by Roberts et al.
(2015), Peters et al. (2017) and Seals et al. (2017), identify a true risk of formaldehyde exposure.
However, additional research designed to address the identified limitations would help to clarify
these study results.
This document is a draft for review purposes only and does not constitute Agency policy.
1-350 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-45. Summary of human studies of nervous system disease risk in
relation to formaldehyde exposure
Reference and study design
Exposure measures
Results
Observational epidemiology studies
Reference: Pinkerton et al. (2013) (United
States) Prospective cohort, 11,098 garment
workers (82% women) exposed to
formaldehyde-treated fabric for >3 mo. (late
1950s to early 1980s).
Outcome: Vital status through 2008, underlying
cause of death, ICD-10 G12.2, ICD-9 335.2, ICD-8
348.0 and ICD-7 356.1.
Analysis: Life-table analysis based on U.S.
population, excluded missing birth date (n = 55),
deaths (n = 8), lost to follow-up prior to date file
begin date (n = 13); SMRs and 95% CI, adjusted
for age, calendar time, sex, race; no information
on smoking.
Evaluation:3
SB IB Cf Oth
Overall
Confidence
High
Small number of cases.
Monitoring in 1980s,
geometric mean 0.15 ppm
(GSD 1.9 ppm), constant
levels across departments
and facilities, year of first
exposure (42% before
1963), time since first
exposure (median
39.4 years) and exposure
duration (median
3.3 years); no other
exposures associated with
ALS.
Amytrophic lateral sclerosis mortality
N = 11,022, 414,313 person-years at risk; eight
ALS deaths; mortality for COPD and lung cancer
in cohort was similar or greater than national
rates (Meyers et al., 2013) indicating that
possible confounding by smoking would be in
direction away from the null, not a concern for
these null results.
All eight deaths were recorded due to ALS in
death certificates.
Deaths SMR (95% CI)
Overall 8
Yr of 1st exposure
Before 1963 5
1963-70
0.89 (0.38, 1.75)
>1973
Duration
<3 yr
3-9 yr
10+ yr
TSFEa
<10 yr
10-19 yr
20+ yr
0.84 (0.27,
1.29(0.27,
0.00 (0.00,
0.61 (0.07,
1.17(0.24,
0.94 (0.19,
3.50 (0.09,
0.00 (0.00,
0.89 (0.36,
1.96)
3.78)
4.92)
2.21)
3.41)
2.75)
19.52)
4.19)
1.83)
aTSFE: time since first exposure
Reference: Bellavia et al. (2021)
(Denmark)
Population-based case-control
Cancer cases, 1982-2009, from Seals et al.
(2017) with complete data for several health
factors and environmental risk factors previously
linked with ALS (N = 1086). Controls, 100 per
case matched on being alive on index date for
case diagnosis, same birth year and sex
(N = 111,507). Excluded individuals with less
than 5 years work experience.
Outcome: see Seals et al. (2017)
Analysis: Selected joint predictors and
interactions using boosted regression trees and
Logic regression, which were included in a
logistic regression model adjusting for age, SES,
and geography. Model used a 3-year lag.
Evaluated diabetes, obesity, physical/ stress
trauma, CVD (1977-2009) and lead, diesel
exhaust and solvents.
Evaluation:3
see Seals et al. (2017)
Formaldehyde exposure
metric was ever/never
exposed. Anticipate
exposure misclassification
and large variation in
prevalence and intensity of
exposure across
individuals. In men,
correlations between
formaldehyde, diesel
exhaust and solvents were
0.22 and 0.41, respectively
(Phi coefficients)
Amytrophic lateral sclerosis
Ever formaldehyde
Exposed Controls Cases OR (95% CI)
N (%) N (%)
Men 43,760(0.64) 422(0.63) 0.87
(0.73, 1.04)
Women 28.100(0.65) 255(0.61) 0.86
(0.84,0.89)
Logistic regression mutually adjusting for age,
SES, and geography, diesel exhaust (male),
solvents, trauma, CVD, diesel*CVD (male),
solvents*trauma (male), diesel*trauma (male),
and diesel*solvents (male), lead (female),
lead*solvents (female) and
trauma*formaldehyde (female).
This document is a draft for review purposes only and does not constitute Agency policy.
1-351 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Exposure measures
Results
SB IB Cf Oth
n
Overall
Confidence
Medium
Uncertainty regarding exposure assessment;
adequacy of 3-year lag is unknown
Reference: Seals et al. (2017) (Denmark)
Population-based case-control study, Registry-
based case identification using the Danish
National Patient Register, 1982-2009 (3650
incident cases). Controls obtained from Central
Person Registry (All Denmark residents since
1968), 4 per case matched on sex, age, and no
ALS diagnosis in Hospital Register as of date of
diagnosis for matched case (index date).
Outcome: Cases identified from Danish National
Patient Register, discharge diagnosis ICD-8 348.0
or ICD-10 G12.2. Case definition was 1st
diagnoses on or after 1/1/1982-12/31/2009.
Analysis: Conditional logistic regression adjusted
for age, sex, index date, SES, marital status and
residence. No information on smoking status.
Evaluation:3
SB IB Cf Oth
n
Overall
Confidence
Medium
Uncertainty regarding exposure assessment;
adequacy of 3-year lag is unknown
Occupational histories
obtained from Danish
Pension Fund databases.
Used NOCCA (Nordic
Occupational Cancer
Study)- Danish JEM for
periods 1960-74,1975-84,
and 1985 and after. Inputs
year and industry code and
outputs prevalence of
exposure for each job
along with expected
exposure level (ppm) in
exposed. The JEM has not
been validated to estimate
levels. Cumulative
expected exposure
calculated (prevalence
multiplied by expected
level) summed over jobs
and time (3- and 5-year
lags). Exposure
misclassification expected
due to variation of tasks
within industries.
Amytrophic lateral sclerosis
RR (95% CI)
Exposure Controls Cases
N (%) N (%)
None 10,934(75)2582(71) 1.0 (ref)
Ever 3666(25) 1068(29) 1.3(1.2,
1.4)
Quartiles (mg/m3)
<0.016 935(6.4) 262(7.2) 1.3(1.1,
1.5)
0.016-0.1 976(6.7) 272(7.5) 1.2(1.1,
1.4)
0.1-0.34 873(6.0) 268(7.3) 1.4(1.2,
1.6)
>0.34
1.5)
882(6.0) 266(7.3) 1.3(1.1,
Reference: Peters et al. (2017) (Sweden)
Nested case-control study, 5,020 patients
diagnosed with ALS between 1991 and 2010 and
25,100 Swedish controls (5 per ALS case)
matched by birth year and sex, alive on case's
date of diagnosis; source population born
1901-1970 and included in the 1990 Swedish
Population and Household Census (includes
persons living in Sweden for >1 year).
Outcome: Cases identified from National Patient
Register (primary or secondary diagnosis)
through 2010 (ICD-9 335C; ICD-10 G12.2).
Analysis: Conditional logistic regression with
adjustment for education and other 11
exposures examined; restricted to individuals
with at least one occupation registered in any of
the censuses, occupations listed in censuses
10 years before diagnosis, and either blue collar
workers or farmers (2,647 cases, 13,378
controls).
Evaluation:3
Individual occupational
histories obtained from
1970, 1980, and 1990
censuses; Swedish version
of Nordic Occupational
Cancer Study JEM
(industrial hygienist
estimates of prevalence
and level of specific
exposures at specific
calendar times).
Dose-response: exposure
metric calculated:
prevalence multiplied by
annual mean level of
exposure in a specific
occupation at the time of a
census, averaged over all
three censuses,
dichotomized at mean level
in controls.
Amytrophic lateral sclerosis
Cases Control OR (95% Ci)
Restricted analytic sample (2,647 cases)
All 323 1,579 1.07
(0.92-1.25)
Exposure metric (mg/m3)
Not 659 3,341
exposed
<0.013 30
1.0
(Reference)
185 0.89
(0.58-1.36)
>0.013 53 210 1.31
(0.86-1.99)
Restricted to individuals <65 years old at
diagnosis (1,014 cases)
All 140 576 1.28
(1.02-1.61)
This document is a draft for review purposes only and does not constitute Agency policy.
1-352 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Exposure measures
Results
SB IB Cf Oth
n
Overall
Confidence
Medium
Uncertainty regarding exposure assessment.
Reference: Roberts et al. (2015) (United
States)
Prospective cohort, 1,469,235 occupational
workers (46% women); National Longitudinal
Mortality Study (NLMS) restricted to age 25+ at
initial survey. Participants provided follow-up
from survey until 2011 or death.
Outcome: NLMS records matched to the
National Death Index (1979-2011) with
underlying cause of death as ALS: ICD-10 G12.2
or ICD-9 335.2.
Analysis: HRs estimated for each exposure level
using survival analyses with age as the time
variable, separate models for men and women,
adjusted for education, race/ethnicity, and
income.
Evaluation:3
SB IB Cf Oth
n
Overall
Confidence
Medium
Uncertainty regarding exposure assessment,
including the influence of duration, particularly
in light of the use of a one-time survey at
enrollment; very small number of exposed cases
(n = 2 in jobs with high probability and intensity
of formaldehyde exposure).
Note: same laboratory, data handling, and
analysis methods as Weisskopf et al. (2009).
Exposure matrix by
industrial hygienists at the
National Cancer Institute
(see Wang et al., 2009)
was constructed based on
participant survey at
enrollment regarding their
last or most recent job; no
information or adjustments
for other potential
exposures.
Amytrophic lateral sclerosis mortality
N = 757 total ALS deaths (472 deaths in men,
with 100 exposed cases and 12,930,240 total
person-years in men).
Duration not evaluated.
No information on mortality from smoking-
related disease or smoking in the general
cohort.
Deaths matched to ALS in death certificates.
No increased risk of ALS in women (data not
shown): authors attribute this to occupation
role.
ALS deaths in men
Deaths HR (95% CI)
Intensity
Unexposed
Low
Medium
High
372
55
43
2
1.0 (Reference)
0.99 (0.74, 1.30)
0.63 (0.44, 0.90)
1.53 (0.4, 5.80)
Intensity, restricted to probability = high
Unexposed
372
1.0 (Referent)
Low
0
-
Medium
0
-
High
2
4.43(1.16, 16.85)
Probability
Unexposed
372
1.0 (Reference)
Low
51
0.85 (0.63, 1.15)
Medium
47
0.76 (0.54, 1.06)
High
2
2.98(0.78, 11.30)
Probability, follow-up to age 75 only
Unexposed
332
1.0 (Reference)
Low
41
0.79(0.57,1.11)
Medium
40
0.66 (0.44, 0.99)
High
2
4.13(1.09, 15.69)
Probability, aged 50-75 at enrollment
Unexposed
197
1.0 (Reference)
Low
31
1.00 (0.67,1.49)
Medium
27
0.75 (0.47, 1.19)
High
2
4.76(1.16, 19.49)
Probability analyses excluding the first 5 years
of follow-up or restricted to men aged 35-75 at
enrollment, or to those employed at
enrollment, are not shown (results were similar
to the overall probability analysis).
Reference: Fang et al. (2009) (United States)
Case-control study, 111 cases and 256 controls;
sequential ALS cases recruited, 1993-1996,
from two major referral centers; cases and
controls lived in New England at least 50% of
year, mentally competent, English speakers; 71%
Occupational history by
structured questionnaire;
industry, occupation,
frequency, and duration;
jobs held before ALS
diagnosis or 2 years before
Amytrophic lateral sclerosis
Association of ALS risk with occupational
formaldehyde exposure (109 cases, 253
controls)
Controls Cases OR (95% CI
Never3
204
89
Ref.
This document is a draft for review purposes only and does not constitute Agency policy.
1-353 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Exposure measures
Results
0.8
o
Ln
1.5)
0.6
(0.1,
2.8)
0.7
(0.3,
1.6)
1.3
o
Ln
3.2)
0.50
of eligible cases participated; controls by
random telephone screening, frequency
matched on sex, age (three groups), and region,
76% of eligible (256 of 270 completed
questionnaires).
Outcome: Diagnoses by board-certified
specialists in motor neuron disease using World
Federation of Neurology El Escorial criteria
(Brooks, 1994).
Analysis: Unconditional logistic regression
models; tested linear trend with lifetime
exposure days, probability, and weighted
exposure duration (four categories); adjusted for
age, sex, area of residence, smoking
(ever/never), and education.
Evaluation:a
interview (controls);
formaldehyde-exposed
occupations identified a
priori by industrial
hygienist; calculated
life-time hours of exposure
to formaldehyde weighted
by probability of exposure
in specific jobs.
Ever
49
Exposure Probability15
0-1 7
27
15
20
2
9
9
Trend p-value
Weighted exposure duration (hr)c
<10,000 14 7
19
SB IB Cf Oth
M
Overall
Confidence
Medium
10,001-
40,000
>40,000 16
Trend p-value
1.1
(0.4, 2.8)
0.8
(0.3,1.9)
0.7
(0.2,2.0)
0.45
>60,000
3.0
(0.7,12.9)
Uncertainty regarding exposure assessment;
small number of exposed cases.
aReferent was group with no previous,
occupational exposure to formaldehyde
bHighest probability ever experienced.
cWeights were 0.5,1, and 2 for probabilities
0-1,1, and 2.
Additional analysis.
Reference: Weisskopf et al. (2009) (United
States)
Prospective cohort, 987,229 men and women.
American Cancer Society Cancer Prevention
Study II. No major illness at baseline in 1982.
Follow-up from 1989 through 2004.
Outcome: Cause of death obtained for >98% of
known deaths; underlying or contributing cause.
ICD-9 (1989-1998) code 335.3; ICD-10
(1999-2004) code G12.2 (ALS represents >98%
of these categories).
Analysis: Cox proportional hazards modeling,
adjusted for age, sex, smoking, military service,
education, alcohol, occupation (farmer, lab
technician, machine assembler, programmer),
vitamin E use, and the other chemical (and X-
rays) exposures assessed at baseline.
Evaluation:3
Self-report (at baseline,
1982) of current or past
regular exposure to
formaldehyde (and
duration); data on 10 other
types of chemicals and
X-ray exposure also
collected.
Source(s) of formaldehyde
exposure were not
defined; likely to be
occupational settings.
Amytrophic lateral sclerosis mortality
1,156 ALS deaths; mortality rate 11.3 and 6.7
per 100,000 person-years in men and women,
respectively.
N cases
exposed
RR
(95% CI)
Full cohort
36
1.34
(0.93, 1.92)
With
duration3
22
2.47
(1.58, 3.86)
<4 years
4
1.5
(0.7, 4.2)
4-10
5
2.1
(0.9, 5.4)
>10
13
4.1
(2.2, 7)
CIs estimated from graph
RR between other exposures and ALS ranged
from 0.68 to 1.44.
a"With duration" indicates the subset of the full
cohort after excluding individuals not providing
duration information.
SB IB Cf Oth
n
Overall
Confidence
Medium
Uncertainty regarding exposure assessment.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.7). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
This document is a draft for review purposes only and does not constitute Agency policy.
1-354 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
Abbreviations: ALS = amyotrophic lateral sclerosis; COPD = chronic obstructive pulmonary disease; GSD = geometric standard
deviation; CI = confidence interval; SMR = standardized mortality ratio.
Nervous System Effects in Animal Studies
Numerous experimental animal studies report findings of neurobehavioral and structural
alterations following formaldehyde inhalation. This section discusses these studies according to
the type of evaluation(s) performed, specifically by studies of neuropathology (see Table 1-46),
studies examining potential sensitization of the nervous system (see Table 1-47), tests of general
motor-related behaviors (see Table 1-48), and tests of learning and memory (see Table 1-49). The
evidence tables are organized by study confidence and the first author's last name.
As discussed below, much of the available data are difficult to interpret due to potential
coexposures (e.g., methanol), possible mischaracterization of irritation-related behaviors as central
nervous system- (CNS)-mediated effects, unreported or inadequate study design methods, and
unclear dose-response relationships. The neurobehavioral effects reported following formaldehyde
inhalation include changes in assays testing motor function, anxiety, habituation, learning and
memory, and chemical sensitization in adult animals fLICM. 2008: Malek etal.. 2004: Sorg et al..
2004: Malek etal.. 2003a. b, c; Usanmaz etal.. 2002: Sorg etal.. 2001b: Pitten etal.. 2000: Sorg and
Hochstatter. 1999: Sorg etal.. 1998: Boiaetal.. 19851. Nociception was unaffected in one study
(Sorg etal.. 1998). Several studies also indicate neuropathology or behavioral effects following
developmental formaldehyde exposure (Sarsilmaz etal.. 2007: Asian etal.. 2006: Songur etal..
2003: Sheveleva. 1971): no corresponding information in human studies is available for children.
In addition to these studies evaluating specific effects on the nervous system, one
subchronic study fWoutersen etal.. 19871 and three chronic studies fAppelman et al.. 1988: Tobe et
al.. 1985: Kerns etal.. 19831 designed to assess the general toxicity or carcinogenicity of
formaldehyde reported general behavioral effects (e.g., uncoordinated locomotion) following
exposure to high levels of formaldehyde (>12 mg/m3). In these studies, no overt changes in
absolute brain weight, brain histopathology, or performance in simple tests of nervous system
function were observed (data not shown). These general toxicity and carcinogenicity studies were
not specifically designed to assess nervous system function and did not report many of the relevant
procedural details or, in most cases, the specific quantitative results. Thus, a confidence rating was
not assigned to these experiments and they are not discussed further. Aside from these cursory
examinations and one subchronic experiment with brief, 10-minute, daily formaldehyde exposures
(Pitten etal.. 2000). the remaining animal studies of the potential for nervous system effects due to
formaldehyde inhalation relied on exposures of acute or short-term duration; extrapolation of these
effects to long-term exposure scenarios is difficult. Figure 1-29 presents all of the medium or low
confidence experimental animal studies identified (no high confidence studies were identified),
whereas the data from the medium confidence animal studies are summarized in greater detail in
Figure 1-30.
This document is a draft for review purposes only and does not constitute Agency policy.
1-355 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
O Non-significant • Statistically significant The size of the shape illustrates the sample size
20
15
¦a
o
+3 io H
iy
a-
¦H
C
Q)
U
c
O
u
20
Adult
Sensiti
zatio n
Rats
20 d
n>15
Adult
Activity
Confidence:
MEDIUM
.
Rats
Pl-30
n=6
Devel.
Path.
Ht
o
O
to
i/)
Q.
D
Cl.
<
~
cn
cn
cn
r-.
cn
QJ
cn
"nT
d-
w
>
ro
_C
(LI
>
o
>
0)
a
>
o
X
_c
OO
OJ
-C
o3
0/1
to
o
l/l
ra
Rats
Gl-19
n=15
Devel. | Adult
Sensitization
tf
• o
CLl
Mice
Rats
Rats
21 d
Gl-19
<1 d
n=6
n=15
n=8
Devel.
:
ro —
ra
7d 5 n=15 n=5 n=15
Adult
Learning/
Memory
LOW
Figure 1-29. Nervous system effects in animal studies.
As no high confidence experimental animal studies were identified, the available studies are organized by
medium and low confidence study evaluation interpretations (see Appendix A.5.7), then by endpoint,
then by timing of exposure (i.e., developmental [devel.] or adult). Filled symbols indicate statistically
significant effects, and the size of the points reflecting the sample size for that formaldehyde exposure
group (larger size = larger n). The low confidence experiments are shown on a gray background, as the
identified study limitations substantially reduce confidence in the reliability of the results; these low
confidence experiments contribute very little to the weight of evidence for nervous system effects. Note:
"Activity" refers to motor-related behaviors (e.g., open field activity). *The studies by Asian et al. (2006)
and Sarsilmaz et al. (2007) report data from the same cohort of exposed rats.
This document is a draft for review purposes only and does not constitute Agency policy.
1-356 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1.5n
1.0- -
ro
u
NP
SV
m
CTl
+1
01
CuO
c
TO
2 0.5H
o
timing:
measure:
duration:
animals:
mg/m
L
i
at PND30 • at PND90
•
Dentate gyrus
PND1-30; n=5
f
t
* *
at PND30 ; at PND90
CA3 Region
PND1-30; n=5
Wistar rats [male]
T 1 1 1-
14.8
7.4 14.8 7.4 14.8 7.4 14.8 7.4 14.8
Hippocampal neuron counts
early : late
vertical count
7 d; n=15-16
"T*
1
early : late
vertical count
20 d; n=20-24
SD rats [female]
¦6
•5
•4
3
1.0
-0.5
I I 1
1.23 1.23 1.23 1.23
Cocaine-induced activity3
Legend: A Asian etal., 2006 ~ Sarsilmaz et al., 2007 0 Sorg et al., 1998
Figure 1-30. Medium confidence animal studies of nervous system effects.
The evidence for nervous system effects reported in medium or high confidence experimental animal
studies is arrayed (note: no high confidence studies were identified). Two studies examined
developmental neuropathology using stereological methods after postnatal exposure to 7.4-14,8 mg/m3
formaldehyde in a single cohort of rats (Sarsilmaz et al., 2007: Asian et al., 2006), while a third study
evaluated sensitization-type responses in adult rats at 1.23 mg/m3 (Sorg et al., 1998). 1Results are
displayed as fold change from control animals (control responses at 1 are illustrated as a dashed line),
with variability in both the controls and treatment groups represented by the quotient (ratio) of the 95%
CI, as calculated based on the method described by E.C. Fieller (Cox and Ruhl, 1966), which assumes
Gaussian distributions. aChanges in vertical activity induced by stimulation with cocaine exposure
following formaldehyde inhalation for 7 or 20 days and several days ("early") or several weeks ("late") of
nonexposure are shown; the authors did not observe any changes in cocaine-induced horizontal activity
(not shown). *p < 0.05, as reported by study authors. Note: all results were estimated from data
presented graphically using Grab It!™, Datatrend Software.
1 Neuropathology
2 Several studies examined the effects of formaldehyde inhalation on brain neuropathology.
3 Evidence of changes in brain structure and neuron number following developmental exposure to
4 >7.38 mg/m3 formaldehyde has been described in three publications from one laboratory
This document is a draft for review purposes only and does not constitute Agency policy,
1-357' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
(Sarsilmaz etal.. 2007: Asian etal.. 2006: Songur etal.. 20031 (see Table 1-46). Two of these
studies f Sarsilmaz etal.. 2007: Asian etal.. 20061 were evaluations of the same cohort of animals.
No overt changes in CNS pathology have been reported following subchronic or chronic
formaldehyde exposures in adult rats at concentrations ranging from 0.369 to 18.5 mg/m3 fPitten
etal.. 2000: Appelman et al.. 1988: Tobe etal.. 1985: Kerns etal.. 19831. although the methods
employed in the adult animal studies were far less sensitive than those used by Sarsilmaz et al.
(20071 and Asian et al. (20061.
Neuropathological alterations were evident in male rats following exposure to 7.38 or
14.8 mg/m3 formaldehyde from postnatal day (PND) 1 to PND 30. Specifically, in the cornu
ammonis (CA) region of the hippocampus, a 4% (at 7.38 mg/m3) or 22% (at 14.8 mg/m3;
statistically significant) decrease in the number of neurons in the pyramidal cell layer was observed
at PND 30, and statistically significant, 8-9%, decreases were still observable at both
concentrations at PND 90 (Sarsilmaz etal.. 2007). Although the morphology of the cell nuclei
determined by cresyl violet staining was indicated as normal in all regions of the hippocampus at
PNDs 30 and 90 in Sarsilmaz et al. (2007) and Asian et al. (2006). these decreased cell counts were
consistent with separate observations of robust increases (59-322%) in the number of pyknotic
(i.e., dying) CA neurons at PNDs 30 and 60 in Songur et al. (2003.). A decrease in cell number is
considered an adverse effect and a specific indicator of toxicity. The decreased magnitude of
neuronal loss at PND 90 as compared to PND 30 (Sarsilmaz etal.. 2007). along with a separate
observation that pyknotic CA neuron counts were no longer elevated at PND90 (Songur etal..
2003). suggest some measure of recovery or adaptation 60 days after exposures were terminated.
Notably, hippocampal cell number exhibits a natural decrease between PNDs 30 and 90, as
demonstrated by Sarsilmaz et al. (2007) and Asian et al. (2006).
Changes in the hippocampal dentate gyrus (DG) cell number and in volumetric measures
were less clear. A significant increase in DG volume was observed at >7.36 mg/m3 formaldehyde at
PND 30, without any accompanying changes in cell number (Asian etal.. 2006). The authors
attributed this finding to possible formaldehyde-triggered inflammation during postnatal growth of
the DG, which continues until ~PND 28; however, this hypothesis was not evaluated by
immunostaining. At PND 90, although DG cell number was decreased at 14.8 mg/m3, DG volume
and cell number were elevated at 7.36 mg/m3. In contrast to decreases in cell number, an increase
in cell number is not necessarily adverse. Although CA cell counts were decreased, the volume of
the pyramidal cell layer on PND 30 was increased at 7.38 mg/m3 but decreased at 14.8 mg/m3;
neither exposure group was significantly different from controls on PND 90. Changes in brain
hemisphere volume [decreased at PND 30 and increased at PND 90; (Sarsilmaz etal.. 2007)]
suggest formaldehyde-induced structural changes or inflammation in nonhippocampal regions, or
altered ventricular parameters, as the changes were not consistent with volume changes in the DG
or CA regions. Volume changes can provide nonspecific measures of neural health. Although these
changes are sometimes associated with regional atrophy and degeneration, they are also sensitive
This document is a draft for review purposes only and does not constitute Agency policy.
1-358 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
to variations such as changes in neuron size or changes in the size or number of nonneuronal cells.
Thus, decreased cell number is a more specific indicator of toxicity.
Exposure from PND 1 to PND 30 covers a sensitive window of hippocampal development, as
a large percentage of hippocampal neurons, particularly in the DG, are generated or mature
(e.g., establish permanent connections) during the early postnatal period. In addition, the
stereological methods used by Asian et al. (2006) and Sarsilmaz et al. (2007) are extremely
sensitive and unbiased by design (e.g., sampling is random and systematic). These methods were
not applied in any other studies, highlighting a key uncertainty in the database. The specific
exposure window or methods employed could explain the general lack of overt neuropathological
effects in rats exposed as adults. Importantly, these developmental studies did not appear to
evaluate possible effects on nursing dams (i.e., dam health and behavior), who appear to have been
exposed along with the pups from PND 1 to PND 14. It is plausible that the high-level exposures
could lead to nutritional changes that influence measures of structural brain development. Pup
health, which was affected at PND 30 (i.e., decreased body weight) but not PND 90 in the study by
Songur et al. (2003). was not reported in the other two studies. However, CA neuron loss was still
evident at PND 90 when no body-weight differences were evident fSongur et al.. 20031. An
additional significant limitation of these studies is that the sample size is very small considering
that the analyses were performed on a pup basis rather than a litter basis, as would be preferred.
Specifically, although 5-6 pups/group were analyzed, because litter effects may influence these
measures, the data are better evaluated as representing only N = 3 litters (the authors indicate two
pups were assessed from each of the three litters). Litter data were not available to determine
whether such analyses would result in a greater or lesser magnitude of response, further
complicating interpretation.
Complete recovery of the observed neuropathology following developmental exposure was
not observed. Partial recovery was apparent, but examinations did not continue long enough to
detect whether or when the observed pathology completely resolves. This supports the possibility
that formaldehyde may cause long-lasting or permanent neuroanatomical changes in the brain
following early-life exposure, which would substantiate characterizing it as a nervous system
hazard according to Agency guidelines (U.S. EPA. 1998). However, these stereological data reflect a
single cohort of exposed animals, and the study deficiencies described above limit the ability to
attribute the results to formaldehyde exposure alone. In addition, the limited data supporting these
effects were derived from studies only testing high-level formaldehyde exposure (i.e., well above
levels demonstrated to affect the respiratory system; see Sections 1.2.1-1.2.4), introducing
additional uncertainties. Thus, the potential for developmental neuropathology remains a
significant concern, and this represents an area in need of further research.
This document is a draft for review purposes only and does not constitute Agency policy.
1-359 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-46. Developmental neuropathology in experimental animal studies
Reference and study design
Results (percentage change from control) and exposure levels
Medium confidence
Reference: Asian et al. (2006)
Rat (Wistar); N = 3 litters (5 male
pups)
0, 7.38, or 14.8 mg/m3aPND 1-PND 30
Test article: paraformaldehyde
Main limitations: Small sample size;
potential for litter effects; note: same
cohort as Sarsilmaz et al. (2007).b
(Importantly, all data were analyzed on a pup basis rather than on a litter basis.)
0 7.38 14.8 0 7.38 14.8
Total DG cell number assessed by stereology:
at PND 30: 0 3 0% at PND 90: 0 10* -12%*
Note: DG cell morphology was normal at PND 30 and PND 90.
Volume of the DG assessed by stereology:
at PND 30: 0 9* 8%* at PND 90: 0 13* -1%
Reference: Sarsilmaz et al.
("20071
Rat (Wistar); N = 3 litters (5 male
pupsb)
0, 7.38, or 14.8 mg/m3a
PND 1-PND 30
Test article: paraformaldehyde
Main limitations: Small sample size;
potential for litter effects; note: same
cohort as Asian et al. (2006lb.
(Importantly, all data were analyzed on a pup basis rather than on a litter basis.)
0 7.38 14.8 0 7.38 14.8
Total CA cell number assessed by stereology:
at PND 30: 0 -4C -22%* at PND 90: 0 -9* -8%*
Note: CA cell morphology was normal at PND 30 and PND 90
CA volume assessed by stereology:
at PND 30: 0 15* -28%* at PND 90: 0 -7 10%
Hemisphere volume assessed by stereology:
at PND 30: 0 -3* -7%* at PND 90: 0 24* 5%*
Reference: Songur et al. (20031
Rat (Wistar); N = 3 litters (6 male
pups)
0, 7.38, or 14.8 mg/m3a
PND 1-PND 30
Test article: paraformaldehyde
Main limitations: Small sample size;
potential for sampling bias and litter
effects.
Low confidence
(Importantly, all data were analyzed on a pup basis rather than on a litter basis.)
at PND 30
at PND 60
at PND 90
0
7.38
14.8
0
7.38
14.8
0
7.38 14.8
CA1 pyknotic neurons:
0
59*
74%*
0
5
54%
0
20 -6%
CA2 pyknotic neurons:
0
322*
336%*
0
65*
72%
0
18 9%
CA3 pyknotic neurons:
0
273*
291%*
0
128
60%*
0
60 -19%
Body weight:
0
-12*
-21%*
0
-4
-9%*
0
-2 -5%
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: DG = dentate gyrus; PND = postnatal day; CA = cornu ammonis.
*p < 0.05 versus control exposure; formaldehyde levels are underlined.
formaldehyde levels in the study (converted to mg/m3 from ppm) were interpreted from the methods to
represent the achieved mean analytical levels, although the range of measured concentrations was not reported.
bSex and cohort information provided to EPA by personal communication (Kaplan, 2014, 2012).
indicated as -19% by study authors in text but estimated by EPA at -4% from data displayed graphically.
1 Neural sensitization
2 Research suggests that formaldehyde exposure might induce sensitization-like properties in
3 neuronal networks (Sorg etal.. 2004: Usanmaz etal.. 2002: Sorgetal.. 2001b: Sorg and Hochstatter.
4 1999: Sorg etal.. 1998: Sheveleva. 1971) (see Table 1-47). Behavioral sensitization in animals can
5 be initiated by drugs affecting the mesolimbic dopamine system (e.g., cocaine, morphine). Although
6 the mechanisms are not fully understood, repeated, low-level exposures to certain chemicals and
7 other stimuli have been hypothesized to cause a persistent modification to brain signaling, possibly
This document is a draft for review purposes only and does not constitute Agency policy.
1-360 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
due to altered dopamine levels in limbic circuits (Bell etal.. 1999: Bell etal.. 1992: Antelman et al..
19801. Subsequent re-exposure to the conditioned chemical or stimulus, or challenge with other
sensitizing agents, may result in amplified neural responses. These responses can be manifest as,
for example, increased impulsivity, motor activity, or CNS excitability.
Possible sensitization manifest as amplified cocaine-induced locomotor activity and
conditioned fear responses, as well as disrupted sleep patterns, has been reported by one group of
researchers following repeated exposure to formaldehyde at 1.23-2.46 mg/m3 (Sorg etal.. 2004:
Sorg etal.. 2001b: Sorg and Hochstatter. 1999: Sorgetal.. 19981. In the study interpreted with the
highest confidence (medium confidence), although cross-sensitization to cocaine was not observed
in female rats exposed to formaldehyde for 7 days, 4 weeks of exposure led to increased cocaine-
induced vertical activity (with no difference in horizontal activity) when tested at 2-4 days (early
withdrawal) and 4-6 weeks (late withdrawal) after cessation of exposure fSorg etal.. 19981.
Sleep-wakefulness patterns, which are regulated in part by dopaminergic signaling (Dzirasa etal..
20061. were disrupted in male rats (females were not tested) after a 1-week withdrawal from
formaldehyde inhalation (Sorgetal.. 2001b): however, these results were limited by incomplete
reporting (see Table 1-47). The study authors hypothesized that formaldehyde exposure may be
causing a persistent stress response in the animals.
Several weeks following exposure to >1.23 mg/m3 formaldehyde for 20 days, rats
previously trained in a fear conditioning paradigm (a neutral odor was paired with footshock)
tended to spend more time immobilized ("freezing") in the presence of the odor than did
air-exposed controls, although these differences were not statistically significant (Sorg and
Hochstatter. 1999). The authors concluded that the formaldehyde-treated rats had more difficulty
than controls in extinguishing the fear response to the conditioned odor; however, as these changes
were noted in response to odor cues, it is unclear whether formaldehyde preconditioning may have
altered the sensitivity of the respiratory tract to odor. Overt damage of the nasal mucosa is not
expected at these formaldehyde levels, and airway irritation at these levels is expected to be
resolved two weeks after exposure (see Section 1.2.1), making causation by physical irritation
unlikely. As these data could be related to observations suggesting increased anxiety following
exposure (as discussed in the next subsection), the results identify the need to systematically test
whether formaldehyde inhalation preconditioning influences responses related to limbic system
function using olfactory-independent stimuli, and to compare any findings with responses caused
by other stressors (e.g., restraint stress; chemicals with strong irritant odors, but no CNS action).
Equivocal evidence of increased CNS excitability following formaldehyde exposure has been
reported in a few studies. Proconvulsant activity following acute formaldehyde exposures in mice
was observed at 2.21-7.87 mg/m3 (Usanmaz et al.. 2002). but not at higher exposure levels or when
formaldehyde was administered for longer durations (2-3 weeks). A critical component of
sensitization was not included in this study, namely, a period of latency between the stimulus and
challenge. These data are difficult to interpret because of an inability to distinguish between a
This document is a draft for review purposes only and does not constitute Agency policy.
1-361 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Toxicological Review of Formaldehyde—Inhalation
"wet-dog shake" due to an irritating odor and that due to a preconvulsive movement Changes in
pentylenetetrazole-induced seizures reported by Usanmaz et al. (2002) were also not easily
interpreted, as no discernible pattern could be identified (e.g., seizure incidence was decreased at
18.2 mg/m3 and seizure intensity was increased at 2.21 mg/m3). In a developmental study,
exposed pregnant dams displayed a significant reduction (12%) in the threshold of neuromuscular
excitability at 4.92 mg/m3, whereas neuromuscular excitability was unchanged in rat offspring
exposed in utero (Sheveleva. 1971). However, the details of the study methods, including latency
between exposure and testing in dams, were not provided. It is unclear whether reflex
bradypnea-related responses would affect these types of measures (e.g., via transient tissue
hypoxia). No other developmental studies examining these types of effects have been identified.
Overall, the data indicate the potential for an effect, but the evidence is insufficient to conclude that
formaldehyde exposure causes neural excitation or acts as a proconvulsant
In some studies, it is unclear how the observed sensitization-type responses can be fully
separated from potential confounders, such as responses due to irritation (the levels used are likely
to elicit some irritant aversion responses) or sensitivity to the formaldehyde odor. Odor detection
and irritation responses in rodents and humans differ. In general, odor detection of formaldehyde
occurs at slightly lower concentrations than irritation-related responses, with human thresholds
reported at 0.068-0.135 mg/m3 fBerglund etal.. 2012: Berglund and Nordin. 19921. An alternative
explanation for some of the observed effects is that formaldehyde exposure, and the irritation
associated with exposure, is uncontrollable or inescapable, which has the potential to modify stress
and brain reward responses (Sorgetal.. 1996). This is in contrast to situations of controllable
stress expected to be encountered by formaldehyde-exposed humans. Additionally, explanations
for sex-dependent differences in potential sensitization responses have yet to be explored. Overall,
the human relevance of, and the formaldehyde-independent contributions to, the observed
sensitization responses in rodents require additional research, including studies clarifying human
sensitization-type responses to chemical irritants and well-controlled animal studies designed to
mimic the human condition.
This document is a draft for review purposes only and does not constitute Agency policy.
1-362 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-47. Neural sensitization in experimental animal studies
Reference and study design
Results3 and exposure levels
Medium Confidence
Reference: Sorg et al. (1998)
Rat (Sprague-Dawley); N = 15-16 (7d)
or 20-24 (20d) females
0 or 1.23 mg/m3b
[Actualc:0 or 0.779-1.76]
7 or 20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Blinding NR;
description of methods incomplete.
Cocaine-induced vertical activity following 20-day exposures:
Early withdrawal
_0 1.23
Saline-induced activity (counts):
Cocaine-induced activity (counts):
Percentage change in activity by
cocaine:
333
1,233
370% 1,040%*
333
3,467*
Late withdrawal
0 1.23
231
1,983
858%
231
3,372*
1,460%*
No changes in cocaine-induced activity were noted after 7 days of exposure and no
changes in horizontal activity were noted after 20 days of exposure.
No changes in nociception (hot plate test) were noted after 7 or 20 days of exposure.
Low confidence
Reference: Sheveleva f!971)
Rat (Strain NR); N = 15/sex
0, 0.492, or 4.92 mg/m3e
[Actual: 0,1.24, 3.09, or 6.20]
GD1-GD19
Test article: Not reported
Main limitations: Test article and
endpoint evaluation methods NR.
Neuromuscular excitability in dams: (
No changes in offspring neuromuscular excitability.
0.492 4.92
-7
-19%*
Reference: Sorg and Hochstatter
C19991
1.23
Rat (Sprague-Dawley); N = 4-8
females
0 or 1.23 mg/m3b
[Actual: not reported]
20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Unclear impact of
altered olfactory detection or cocaine
injection; note: formalin use as an
aversive odor was deemed irrelevant.
Cocaine (10 mg/kg)-induced horizontal activity (as percentage change in induced
activity):
Cocaine-induced activity 2-4 days after air or formaldehyde, as
compared to cocaine-induced activity prior to exposure: 198 407%*
Fear-conditioned responses to odor (as percentage change from nonshockedf:
Freezing in the context used for shock training: 433* 476%*
Freezing with the conditioned odor 2 days later: 45 127%*
Freezing with the conditioned odor 12 days later: 54 181%*
*p < 0.05, as compared to no shock condition in the same exposure group (f-test).
[Notes: Statistically significant differences in direct comparisons of the control and
HCHO pre-exposed groups were not observed for any fear conditioning tests {N = 4).]
Reference: Sorg et al. f2001bl
Rat (Sprague-Dawley); N = 6/sex
0 or 2.46 mg/m3b
[Actual: not reported]
20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Description of
methods incomplete; no
preformaldehyde exposure
comparisons.
Sleep patterns, as assessed by EEG/EMG in Males 7 days after exposure:
[Dark: 1—12h/Light: 13-24h phase5]:
l-6h
7-12h
13-18h
19-24h
0 2.46
0 2.46
0 2.46
0 2.46
0 -30%
0 -25%
0 -16%
0 -18%
0 -25%
0 -21%
0 -10%
0 -18%
0 37%
0 59%
0 9%
0 12%
Number of waking episodes:
Number of NREMS episodes:
Duration of waking episodes:
*Significant treatment effects noted for each measure above by 2-way ANOVA.
No changes in REMS episodes or duration of NREMS episodes were noted.
[Note: a 15-min challenge with 37% formalin odor abolished all differences.]
This document is a draft for review purposes only and does not constitute Agency policy.
1-363 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results3 and exposure levels
Reference: Sorg et al. (2004)
Rat (Sprague Dawley); N = 7-8/sex
0 or 2.46 mg/m3b
[Actual: 0 or 2.66]
20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Unclear Influence of
changes in olfactory detection.
Freezing responses to a conditioned stimulus (CS, odor)h in males:
Day 1
Day 2
Day 3
Day 4
Day 5
Renewal'
0
2.46
0
2.46
0 2.46
0
2.46
0
2.46
0 2.46
Unpaired:
0
64%*
0
19%
0 7%
0
76%*
0
0%
0 86%
Paired:
0
26%
0
12%
0 5%
0
22%
0
50%*
0 47%
No changes observed in response to the context alone (footshock or novel).
No change in female conditioned fear behaviors (to context orCSJ.
Reference: Usanmaz etal.
f20021
Mouse (Balb/C); N = 6 Sex NR
0, 2.21, 3.94, 7.87, 11.9, or 18.2
mg/m3j: 3 hours
0 or 3.94 mg/m3: 2 weeks
0 or 2.46 mg/m3: 3 weeks
Test article: Paraformaldehyde
Main limitations: Tested immediately
after exposure; blinding NR.
CNS excitability after a 3-hour exposure:
Percentage incidence of wet-dog
shakek:
Percentage incidence of seizures':
Seizure intensity (median vales):
Seizure threshold (seconds to onset):
No significant effects on seizure mortality.
No significant effects on CNS excitability after 2-3 weeks of exposure.
0
2.21
3.94
7.87
11.9
18.2
0
63*
67*
60*
25
17%
91
82
ND
60
ND
33%*
4
6*
ND
4
ND
1
74
83
ND
104
ND
110%
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: GD = gestational day; NREMS = nonrapid eye movement sleep; CS = conditioned stimulus; ND = not
determined; NR= not reported; EEG/EMG= electroencephalogram/electromyelogram; CNS = central nervous
system.
*p < 0.05 vs. control exposure unless otherwise indicated; formaldehyde levels are underlined.
aData presented as percentage change from control, unless otherwise indicated.
bFormaldehyde levels in the study converted to mg/m3 from ppm.
cActual mean analytical concentrations achieved.
d2-4 days after discontinuing exposure, rats were given cocaine and evaluated for 2 hr (early withdrawal); an
additional cocaine challenge and locomotor assessment were conducted 4-6 wk later (late withdrawal),
formaldehyde levels in the study (converted to mg/m3 from mg/L) represented the achieved analytical levels.
'Context = in the context the shock was delivered, rats receiving shock training vs. those not shocked were
compared at 1 day after training; conditioned odor = comparison as in "context" 2 or 12 days after training except
in a novel context and with the odor used for shock training (orange oil) present. Values and statistical analyses
are compared against nonshocked rats within the same treatment group.
gData were recorded for 6-hour periods beginning at dark phase for 24 hours; percentage change from air controls
for each period is presented; air and formaldehyde groups were significantly different by two-way ANOVAs.
hSeveral weeks after treatment an orange oil odor (CS) was either Paired (with CS presentation) or Unpaired
(separately and randomly from CS presentation) with footshocks, then testing performed over subsequent days
'CS presented in a second, completely novel context.
formaldehyde levels in the study (converted to mg/m3 from ppm) were interpreted from the methods to
represent the achieved mean analytical levels, although the range of measured concentrations was not reported.
kWet-dog shake, a possible pro-convulsive sign, is a shuddering motion in rodents that can be induced
pharmacologically with agents that affect glutamatergic and/or serotonergic signaling.
'Seizures were induced by injection of pentylenetetrazole.
1 Tests of general motor-related behaviors
2 This section encompasses a range of behavioral tests examining general locomotion
3 (without pharmacological manipulation) as the output These tests span a range of test
4 environments and testing conditions, and the observed responses often involve contributions from
5 multiple specific behavioral processes (e.g., motor function, anxiety, arousal, olfaction, acclimation
This document is a draft for review purposes only and does not constitute Agency policy.
1-364 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
to the test environment, etc.) that can be difficult to disentangle. Motor-related tests designed to
examine learning and memory processes are discussed separately in the next section.
Animal studies that included protocols of sufficient duration to specifically assess changes
in motor function fSorg etal.. 2001b: Sorg etal.. 19981 either did not observe effects of
formaldehyde inhalation alone fSorg etal.. 19981 or were complicated by irritant effects when
tested during exposure (Sorg etal.. 2001b). However, open field activity testing following
formaldehyde exposure revealed decreased ambulatory activity in rats and mice, as well as
elevated anxiety and reduced habituation to the test environment in nearly all available studies
fMaleketal.. 2004. 2003a. b; Usanmaz etal.. 2002: Boiaetal.. 1985: Sheveleva. 19711 (see
Table 1-48). Open field testing is a commonplace test that can be standardized and reproducible
fBroadhurst. 19691. but which often involves a somewhat arbitrary interpretation of different
behavioral features. The short testing duration used in open field tests (typically 3-5 minutes) is
not of sufficient length to accurately assess motor function, and the results are also affected by the
initial anxiety of the animals to the novel test environment Thus, with these tests (which vary by
laboratory), it can be difficult to separate changes in motor function and interpretation of olfactory
and visual cues from changes due to exploration of a novel environment and anxiety due to open
spaces and bright light (e.g., increased anxiety correlates with decreased ambulation in these tests).
A second test (typically 24 hours later) measures the level of habituation or learned familiarity to
the test environment. Due primarily to prominent exposure-quality issues (Malek etal.. 2004.
2003a. b; Sheveleva. 1971) or significant study design concerns (Usanmaz etal.. 2002: Boja etal..
1985: Sheveleva. 1971). all of the data suggesting effects of exposure on motor-related behaviors
are derived from low confidence studies (see Appendix A5.7), limiting their interpretability.
Consistent decreases in open field locomotor activity in male mice and rats of both sexes
were observed at formaldehyde concentrations as low as 0.123 mg/m3 (with rats exhibiting
enhanced sensitivity) when assessed shortly after a single, acute formaldehyde exposure fMalek et
al.. 2004. 2003a. b) or after exposure for 1 week (Li etal.. 2016): however, these studies employed
formalin exposures. From the current studies it remains unclear whether these changes persist
more than a few hours after exposure, noting that motor activity testing (not open field tests) did
not reveal changes several weeks after exposure (Sorg etal.. 1998). A portion of this immediate
response in male mice may be due to increased anxiety, as decreases in crossed inner squares
occurred at notably lower levels than decreases in crossed peripheral squares (anxious animals
tend to spend less time in the open and bright areas at the center of the field), suggesting an
elevated stress response after acute exposure (Malek etal.. 2004): however, this increased anxiety
was not confirmed in a second, short-term study (Li etal.. 2016). which actually reported evidence
of a decrease in anxiety in both open field and elevated plus maze tests at 1.23 mg/m3. Although, no
changes were observed at 2.46 mg/m3 and changes in plus maze activity were not observed in rats
that were similarly exposed fSorg etal.. 19981. Perhaps relatedly, short-term exposure of mice to
>1 mg/m3 resulted in dose-dependent increases in immobility time in the forced swim test fLi etal..
This document is a draft for review purposes only and does not constitute Agency policy.
1-365 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
2016). a stress-related test of "behavioral despair" (Porsoltetal.. 19771. When habituation to the
open field was tested 24 hours after exposure, formaldehyde-treated rats and mice did not
demonstrate the same degree of habituation as control animals fMalek etal.. 2004. 2003a). In male
rodents, the degree of habituation was reduced compared to controls. In contrast, formaldehyde-
treated female rats demonstrated robust increases (50-150%) in activity at all formaldehyde
exposure levels (>1.23 mg/m3), suggesting not only reduced habituation, but also delayed
hyperactivity in these animals. These mixed results suggest a general effect on behavior across a
range of tests of general motor-related behaviors, but the specifics of this effect(s) remain difficult
to interpret and require clarification in studies with better-controlled formaldehyde exposures.
A serious concern that changes may be due to irritation and related phenomena (e.g., reflex
bradypnea; distractibility) is raised for three of the studies which evaluated behaviors during or
immediately after exposure to formaldehyde at concentrations expected to cause irritation
(Usanmaz etal.. 2002: Boja etal.. 1985). Decreased activity from 0to 24 hours after exposure to
6.15 mg/m3 formaldehyde was reported using a minimally informative protocol developed for
observations of rat pups (Boia etal.. 1985). with activity defined as the percentage of time "active"
(i.e., not sleeping or immobile). Consistent with the pattern of alterations to habituation reported
by Malek et al. (2004, 2003a), after several days of daily exposure and activity testing, vertical
activity measured during exposure to 2.46 mg/m3 formaldehyde was depressed in male rats (on
exposure days 12-20) and increased in female rats (on exposure days 5 and 20), as compared to
controls (Sorgetal.. 2001b). Usanmaz etal. (2002) notedunexplainable formaldehyde sensitivity
(gastrointestinal impairment and decreased weight gain), causing them to discontinue the study, at
exposures as low as 2.5 mg/m3 for 3 weeks, which would be expected to confound their findings of
decreased activity. Owing primarily to the timing of the behavioral tests, none of the observed
changes in activity can be clearly attributed to formaldehyde-induced effects on the nervous
system.
Reduced spontaneous mobility at PND 30 was observed in pups exposed in utero to 0.492
or 4.92 mg/m3 (Sheveleva. 1971). In contrast, concentration-related increases in mobility were
observed in these pups at PND 60 (an increased level of spontaneous mobility was also observed in
dams at 4.92 mg/m3), with the female pups exhibiting enhanced sensitivity. Increases in activity
which persist into adulthood following developmental exposure are of concern. However, the
methodology was insufficiently described and the significance of these formaldehyde-induced,
bidirectional changes in the activity of young animals, which were dependent either on the delay
between exposure and testing or the postnatal age at testing, is unclear.
Overall, the data from basic tests of motor-related behaviors suggest an effect in
formaldehyde-exposed rodents. This response may be short lived, and, at least in open field tests,
rats seem to be more sensitive to changes following formaldehyde exposure than mice (which
would be consistent with the known toxicokinetic differences across species; see Appendix A.2) and
females seem to exhibit a different pattern of responses than their male counterparts. Somewhat
This document is a draft for review purposes only and does not constitute Agency policy.
1-366 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 differing results across some of the studies, particularly in tests other than open field activity
2 (i.e., elevated plus maze and forced swim test), together present a complicated picture of these
3 potential effect(s). More importantly, however, no studies using methanol-free formaldehyde and
4 other appropriate methodology were available to clarify and confirm the findings of behavioral
5 changes from this set of low confidence studies.
Table l-48.Tests of motor-related behaviors in experimental animal studies
Reference and study design
Results3 and exposure levels
Medium confidence (activity); low confidence (elevated plus maze)
Reference: Sorg et al. f 19981
Rat (Sprague-Dawley); N = 15-24
females
0 or 1.23 mg/m3b
[Actuals 0 or 0.779-1.76]
7 or 20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Description of
methods incomplete; activity could be
affected, and plus maze data are likely
affected, by prior manipulations; total
plus maze activity NR; blinding NR.
No change in horizontal or vertical activity were noted following saline injections
2-4 days or 4-6 weeks after discontinuing formaldehyde exposures.
Note: activities were measured over a 2-hour period after allowing the rats to
acclimate to the test environment.
No statistically significant changes in elevated plus maze performance were noted.
Note: percentage open arm entries and percentage time spent in open arms were
decreased 24 and 39%, respectively after 7 days [p = 0.06 for percentage time];
percentage time in open arms was increased 21% after 20 days, but this did not
approach statistical significance.
Low confidence
Reference: Boia et al. f 19851
Rat (Sprague Dawley); N = 8 males
0 or 6.15b mg/m3c
[Actuald: not reported]
1-2 days (switching paradigm)
Test article: Paraformaldehyde
Main limitations: Tested immediately
after exposure; uncommon protocol.
Percentage time "active" versus "inactive"e during exposui
at 30 min.
e relative to
at 60 min.
lir controls:
at 120 min.
Day 1 HCHO (Day 1 exposed): -34%*
Day 2 HCHO (Day 1 and 2 exposed): -76%*
Day 2 HCHO (only Day 2 exposed): -58%
24h post HCHO (only Day 1 exposed): -30%
Boja etal. (1985)
-66%*
-70%*
-80%
-80%
-77%*
24%
122%
72%f
Reference: Li et al. (2016)
0 1.23 2.46
0 1.23 2.46
Mouse (Kunming: outbred Swiss
albino); N = 15 males
0,1.23, or 2.46 mg/m3c
[Actual: levels confirmed]
7 days (2 hours/day)
Test article: Formalin
Main limitations: Formalin; blinding
NR for tests other than forced swim;
possible influence of multiple
behavioral tests in the same animals.
Open Field Activity (2-hr postexposure):
Total Distance: 0 -3.15 -18.7*
Total Crossings: 0 -4.02 -20.9*
Percentage Center Q 3g 0* _1L5
Time:
Forced Swim (after plus maze):
Immobility Time: 0 42.3 87.6*
Note: Statistically significant differences in b
mg/m3 (-3.7%, as compared to + 1.82% in cc
Elevated Plus Maze (after open field):
Total Distance: 0 0.70 -3.00
Number of Entries: 0 -14.5 -12.1
Percentage Open Q 2Q g* _433
Arm Time:
ody-weight gain were observed at 2.46
jntrols).
This document is a draft for review purposes only and does not constitute Agency policy.
1-367 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results3 and exposure levels
Reference: Malek et al. (2003a)
Males
Females
Rat (LEW.1K); N = 15/sex
0 1.23 3.08 6.15
0
1.23
3.08
6.15
0,1.23, 3.08, or 6.15 mg/m3c
Open field activity and behaviors at 2 hours postexposure:
[Actual: 0,1.24, 3.09, or 6.20]
Locomotion:
0 -63* -22* -
41%*
0
-72*
-30*
-36%*
2 hours
Grooming:
0 -47 -23* -
34%*
0
4
-17*
-62%*
Test article: Formalin
Air sniffing:
0 103* 118* 104%*
0
1
-23*
22%*
Main limitation: Formalin.
Floor sniffing:
0 105* 51* 84%
0
-2
56
79%
Wall climbing:
0 -22* -22* -
26%*
0
-8
-14
16%
Rearing:
0 28 32 2%
0
58*
74*
42%
[Note: an excessive level of variability
Note: No changes in defecation.
was noted for this study, possibly due
to an erroneous indication of data as
Habituation to the open field at 26 hours postexposure (Trial 2/Trial lh):
Mean ± SE in this study, in contrast to
Locomotion:
1
00
0
1
4^
4^
1
UJ
Ln
*
I
21%*
-78
140*
42*
38%*
Mean ± SD in the other studies by
Air sniffing:
161 -31* -13* 12%*
78
48*
174*
43%
Malek et al.(2004. 2003b. Cl.l
Climbing:
95 -10* -14* 38%*
73
118
105
46%
Rearing:
-14 6* 9* 24%*
34
3
6*
-8%
Note: No consistent changes in grooming, floor sniffing, or defecation.
Reference: Malek et al. f2003bl
Males
Females
Rat (LEW.1K); N = 10/sex
0 0.123 0.615 6.15
0
0.123
0.615
6.15
0, 0.123, 0.615 or 6.15 mg/m3c
Open field activity and behaviors at 2 hours postexposure:
[Actual: 0, 0.160, 0.590, or 6.37]
locomotion:
0 -17* -48* -
-65%*
0
-5
-19*
-39%*
2 hours
Air sniffing:
0 8* -22* -
-55%*
0
21*
14*
-11%*
Test article: Formalin
Floor sniffing:
0 -23* -39* -
-64%*
0
-5
-23*
-27%*
Main limitation: Formalin.
Wall climbing:
0 21* -55* -
-72%*
0
54*
-4
-34%*
Rearing:
0 -57* -75* -
-59%*
0
44*
-35*
-24%
Note: No consistent changes in grooming or defecation
Reference: Malek et al. (2004)
Open field activity and
Mouse (AB); N = 20 Males
behaviors at 2 hours
Habituation to the open field
0,1.35, 2.83 or 6.40 mg/m3c
postexposure:
at 26 hours postexposure:
[Actual: 0, 1.37, 2.84, or 6.64]
2 hr (Percentage control)
26 hr (Trial 2/Trial lh
2 hours
0 1.35 2.83 6.40
0
1.35
2.83
6.40
Test article: Formalin
Crossed inner squares: 0 -26* -38* -
53%*
-70
-62
-57
-40%
Main limitation: Formalin.
Crossed outer squares: 0 5 -12
49%*
-24
-25
-10
41%
Total crossed squares: 0 -7 -22* -
51%*
-41
-36
-24
15%
Air sniffing:
0 11 -16* -
58%*
-29
-38
-23
52%
Floor sniffing:
0 26* 2 9%
3
-40*
-38*
-23%*
Grooming:
0 -11 -11 -
18%
5
96*
45*
82%*
Rearing:
0 -22* -37* -44%*
3
-11*
8*
21%*
Note: No consistent changes in wall climbing or defecation.
Reference: Sheveleva ("19711
Males
Females
Rat (Strain NR); N = 15/sex
0 0.492 4.92
0
0.492
4.92
0, 0.492, or 4.92 mg/m3'
Spontaneous mobility in offspring and dams:
[Actual: 0,1.24, 3.09, or 6.20]
at PND 30:
0 -48* -2%
0
-36*
-44%*
GD1-GD 19
at PND 60:
0 16 32%
0
42
291%*
Test article: Not reported
in dams:
NA NA NA
0
-46
89%*
Main limitations: Test article and
endpoint evaluation details NR.
This document is a draft for review purposes only and does not constitute Agency policy.
1-368 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results3 and exposure levels
Reference: Sorg et al. (2001b)
Total vertical activity during formaldehyde exposure:
Rat (Sprague-Dawley); N = 7-8/sex
Males: 4- at exposure days 12-20 (-25 to
-55%*)
0 or 2.46 mg/m3c
Females: T* at exposure days 5 (133%*) and 20 (98%*)
[Actual: not reported]
20 days (5 days/week)
Test article: Paraformaldehyde
Main limitations: Activity tested
during exposure; description of
methods incomplete.
Reference: Usanmaz et al.
(2002)
0 2.21
3.94 5.54
7.87
11.9
18.2
Mouse (Balb/C); N = 6 (sex NR)
Open field activity immediately after a 3-hour exposure:
0, 2.21, 3.94, 5.54, 7.87, 11.9, or 18.2
Horizontal activity: 0 -10
-16 -28
-35*
-69*
-91%*
mg/m3j: 3 hours
Vertical activity: 0 -26
* -43* -48s
-48*
*
fO
00
1
-88%*
0 or 2.46 mg/m3:1 or 3 weeks
0, 2.46, or 3.94 mg/m3: 2 weeks
Open field activity and body-weight gain after 1- to 3-week exposures:
Test article: Paraformaldehyde
1 week
2 weeks
3 weeks
Main limitations: Tested immediately
0 2.46
0 2.46
3.94
0
2.46
after exposure; blinding NR.
Horizontal activity: 0 -28%*
0 -3
-40%*
0
-23%
Vertical activity: 0 -37%*
0 -1
-44%*
0
-32%*
Body-weight gain: 0 33%
0 0
-150%*
0
-280%*
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: HCHO = formaldehyde; SE = standard error; SD = standard deviation; GD = gestational day; NR= not
reported; PND = postnatal day.
*p < 0.05 vs. control exposure; formaldehyde levels are underlined.
aData presented as percentage change from control, unless otherwise indicated.
Additional exposure groups of 12.3 and 24.6 mg/m3 were indicated, but data were not reported and thus, not
included.
formaldehyde levels in the study converted to mg/m3 from ppm.
dActual mean analytical concentrations achieved.
eActive (e.g., grooming, eating, climbing, ambulating, etc.) versus inactive (i.e., immobile, sleeping).
'Statistical comparisons to air-air group not performed.
gLocomotion = crossed squares; M = changes were observed in males; F = changes were observed in females.
hValues presented as Trial 2 (26 hr) vs. Trial 1 (2 hr) performance in same group; * for comparisons within Trial 2.
'Formaldehyde levels in the study (converted to mg/m3 from mg/L) represented the achieved analytical levels.
'Formaldehyde levels in the study (converted to mg/m3 from ppm) were interpreted from the methods to
represent the achieved mean analytical levels, although the range of measured concentrations was not reported.
kOpen field activity in the short-term studies is inferred to have been conducted immediately following exposure.
1 Tests of learning and memory
2 Five studies have examined the effects of inhaled formaldehyde on learning and memory
3 processes in experimental animals see Table 1-49). All of the studies are expected to have
4 significant coexposures due to the formaldehyde generation methods (see Appendix A.5.7), and
5 thus, the effects cannot be attributed to formaldehyde inhalation alone. In addition, many of the
6 dose-response relationships are difficult to interpret and the results are occasionally inconsistent
7 Decreased performance in short-term spatial memory tasks following exposure to
8 formaldehyde has been observed in rats across two studies from coauthors in the same research
This document is a draft for review purposes only and does not constitute Agency policy.
1-369 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
institute (Malek etal.. 2003c: Pitten etal.. 2000). as well as in three mouse studies (LICM. 20081.
These testing paradigms involve components of memory, orientation, reward seeking, stress,
olfactory and visual information processing, and motor function. In the rat studies, increased error
rate and increased latency in a water maze were observed after short-term exposures
to >0.123 mg/m3 and >0.615 mg/m3, respectively fMalek et al.. 2003cl. although the results were
not entirely consistent across all trial days. Similarly, very brief (10-minute) formaldehyde
exposures over a prolonged duration (90 days) resulted in an increased number of errors and
longer running times in a land-based maze at >3.06 mg/m3 (Pitten etal.. 2000). with an increasing
magnitude of change with increasing trial days, which suggests an additive effect of exposure. In
general, excluding the latency measures reported by Malek et al. (2003c), all exposed rats were
equally impaired across a broad range of exposures; no explanation for this lack of a dose-response
relationship is presently available. These observations are supported by potentially related
findings in mice exposed for 1 week to similar levels of formaldehyde (i.e., 2.46 to 3 mg/m3);
specifically, exposed mice exhibited decreased performance in the Morris water maze (Mei etal..
2016) and decrements in a test of recognition memory, the novel object test (Li etal.. 2016).
However, it is difficult to attribute these decrements to formaldehyde exposure due to notable
methodological limitations (e.g., the use of formalin and the lack of observer blinding for these
nonautomated measures raise substantial concerns). In addition, the data from both studies
suggest possible complicating effects on behaviors other than learning or memory in the mice
exposed to formaldehyde [i.e., in Mei et al. (2016). exposed mice did not exhibit improved
performance across training trials and swimming tracks suggest that they avoided the target
quadrant completely during the probe trial; in Li et al. (2016). even in the absence of a novel object,
exposed mice spent approximately half the time exploring objects during training than did
controls]. Although vision and olfaction were not evaluated in these rodent studies, possible effects
on these functions are not expected to influence performance in the studies by Malek et al. (2003c),
Mei et al. (2016). and Li et al. (2016). or by Pitten et al. (2000). as assessments occurred 2-3 or
22 hours after exposure(s), respectively. In contrast, supportive observations in mice (LICM. 2008)
are considered even less reliable due to the short, 30-minute delay before testing following
exposure to formaldehyde and other potential contaminants (formaldehyde was released from
wood baseboard) at levels that are likely to induce irritation-related responses.
In rats, the increases in maze latency are most likely reflective of the increased number of
errors in treated animals as errors usually increase the distance traveled, and thus the time
required, for completion of the trial. However, in the absence of data on path length or motor speed
in all three of the maze-based studies, it is unclear whether hyperactivity of the
formaldehyde-exposed animals may have been present (e.g., increased swim time and increased
number of errors due to exposed animals swimming faster in circular or back-and-forth patterns).
In the study by Malek et al. (2003c), increased swim speed is indeed evident at 0.123 mg/m3 in
females: despite making approximately four more errors than control rats on trial days 4, 5, and 8,
This document is a draft for review purposes only and does not constitute Agency policy.
1-370 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
they still had significantly shorter swimming times. Recovery following exposure was only
assessed by Pitten et al. (2000), who observed that performance was still impaired 4 weeks after
exposures had ended.
While the study authors interpreted these results to suggest deficits in the retention of a
previously learned task or in remembering a previously explored object, these studies had
significant methodological shortcomings. Thus, sole attribution of the decreases in performance to
formaldehyde-induced impairment, and specifically to impairment of memory or orientation,
cannot be concluded. Although two developmental studies evaluating learning and memory
processes following formaldehyde exposure were identified fLiao etal.. 2010: Senichenkova.
1991a), data from these studies were not considered useful for the purposes of hazard
characterization (see Appendix A.5.7). Overall, while the available data suggest a potential effect on
behavior in tests of learning or memory, which may or may not reflect effects on those specific
cognitive processes, no studies using methanol-free formaldehyde and other more appropriate
methodology were available to clarify and confirm the findings of behavioral changes from this set
of low confidence studies.
Table 1-49. Tests of learning and memory in experimental animal studies
Reference and study design
Results (as indicated) and exposure levels
Low confidence
Reference: Li et al. (2016)
0 1.23 2.46
Mouse (Kunming: outbred Swiss
albino); N = 15 males
0,1.23, or 2.46 mg/m3a
[Actual: levels confirmed]
7 days (2 hours/day)
Test article: Formalin
Main limitations: Formalin; blinding
NR; possible influence of multiple
behavioral tests performed in the
same animals.
Novel Object Training and Testing (~ 2 days postexposure):
Training exploration (time ± SEM) of Left identical object: 94 ± 14 99 ± 25 55 ± 10
Training exploration (time ± SEM) of Right identical „„ „
,. ® ^ v ' 5 98 ±20 88 ± 23 51 ±9
object:
Familiar object exploration (seconds) 24-hr posttraining: 69.8 47.0, 61.8
Novel object exploration (seconds) 24-hr posttraining ^03* ^
(*p < 0.05 versus familiar object exploration time):
Discrimination Index [(novel object time-r total
43.3 32 7 -12.0
time) - (familiar object time -r total time) x 100]:
Notes: Statistically significant differences in body-weight gain were observed at 2.46
mg/m3 (-3.7%, as compared to + 1.82% in controls). The study authors did not
provide comparisons of total exploratory activity (Left + Right object) during training.
Reference: LICM (2008)
0 13
Mouse (Kun Ming: outbred Swiss
albino); N = 5 males
0,1, or 3 mg/m3
[Actual3: 0.020, 0.990, or 3.03]
7 days beginning at ~PND 42
Test article: Wood baseboard
Main limitations: Undefined mixture
exposure; possible impact of irritation.
Escape latency across training trial days in the Morris water mazeb:
Latency (percentage from control for averaged trial days): 0 32 74%*c
Note: Magnitude of change was unrelated to duration of exposure.
Performance during probe trial test:
Time spent in the target quadrant (percentage from ^ ^ ^
controls): 0
Note: Only controls spent significantly more time in the target quadrant.
This document is a draft for review purposes only and does not constitute Agency policy.
1-371 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results (as indicated) and exposure levels
Reference: Mei et al. (2016)
Training trials escape latency (sec.; *p < 0.05: Dunnett's post hoc tests on
Mouse (Balb/c); N = 8 males
ANOVAf:
0 or 3 mg/m3a
Control
3 mg/m3
[Actual: confirmed, 3.04 ± 0.13 mg/m3]
Day 1: 58.2
56.7
7 days (8 hours/day)
Day 2: 55.4
55.0
Test article: Formalin
Day 3: 55.7
52.2
Main limitations: Formalin; blinding
Day 4: 49.4
51.4
NR; details of behavioral protocols NR.
Day 5: 38.0
52.1*
Day 6: 36.3
50.4*
Day 7: 33.1
50.7*
Probe trial test performance on Day Se:
Control 3 me/m3
Mean (+ SE) swim distance (cm) in target quadrant:
316 (±42) 154
* (± 16)
Mean (+ SE) time (sec) in target quadrant:
27.5 (± 3.4) 10.0* (± 0.9)
Reference: Malek et al. f2003cl
Latency and number of errors in a water maze:
Rat (LEW.1K); N = 15/sex
Maze errors (as #)
Swim time (as percentage control)
0, 0.123, 0.615, or 6.64 mg/m3d
Males
Females
Males
Females
10 days
0 .12 .62 6.6
0
.12 .62 6.6
0
.12 .62
6.6
0 .12
.62 6.6
Test article: Formalin
Day 1: 7 8 8 8
8
7 8 8
0
-5 -8
* 0
0 -7*
-6* -5*
Main limitations: Formalin; protocol
Day 2: 6 7 6 6
8
7* 8 6*
0
-1 3
8*
0 -4
-2 4*
deficiencies, including blinding NR.
Day 3: 5 5 6* 7*
4
6* 7* 8*
0
-2 14
* 4
0 4
8* -2
Day 4: 2 5* 5* 6*
1
6* 5* 6*
0
-11 29* 14*
0 -24*
16* 14*
Day 5: 1 4* 3* 5*
1
4* 4* 5*
0
-11 -2
23*
0 -13*
-9 -1
Day 6: 1 5* 4* 5*
0
5* 5* 5*
0
6 37* 111*
0 -2
17* 88*
Day 7: 0 5* 4* 5*
0
4*
0
6 38
t g4*
0 12*
11* 62*
Day 8: 0 3* 3* 3*
0
4* 3* 3*
0
-3 -8
41*
0 -20*
-8 15*
Day 9: 0 3* 3* 3*
0
3* 3* 4*
0
3 17
* 64*
0 18*
11* 46*
Day 10: 0 3* 2* 3*
0
3* 2* 3*
0
-3 21
* 73*
0 15
17* 49*
Reference: Pitten etal. (2000)
Latency and number of errors in a land maze:
Rat (Wistar); N = 5-8/sexf
Latency (as percentage
Errors (as percentage
0, 3.06, or 5.55 mg/m3d
control)®
control)
90 days (Note: only 10 minutes/day
0
3.06
5.55
0
3.06
5.55
exposures)
Exposure wk 0:
0
-6
4%
0
-39
-7%
Test article: Formalin
Exposure wk 2:
0
8
21%
ND
ND
ND
Main limitation: Formalin.
Exposure wk 4:
0
30
51%
0
70
91%
Exposure wk 6:
0
48
76%
ND
ND
ND
Exposure wk 8:
0
75*
113%*
0
116
112%
Exposure wk 10:
0
94*
143%*
ND
ND
ND
Exposure wk 12:
0
128*
185%*
0
153*
184%*
2 wks postexposure:
0
168*
241%*
ND
ND
ND
4 wks postexposure:
0
215*
303%*
0
72
89%
No CNS pathology or changes in body weight were observed.
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: SEM = standard error of the mean; PND = postnatal day; ND = not detected.
*p < 0.05 vs. control exposure (unless otherwise indicated); formaldehyde levels are underlined.
aActual mean analytical concentrations achieved.
bMorris water maze: Four trials/day during training; Probe trial involved removal of the platform on Day 7.
Significant differences between the 0 and 3 mg/m3 groups by multiple comparison testing (LICM, 2008).
formaldehyde levels in the study (converted to mg/m3 from ppm) represented the achieved analytical levels.
eData digitized using Grab It!™, Datatrend Software.
fMale and female data were pooled for comparisons; no differences between sexes were noted.
gAverage seconds estimated from points along the fitted linear regression curves presented by Pitten et al. (2000).
This document is a draft for review purposes only and does not constitute Agency policy.
1-372 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Evidence on Mode of Action for Nervous System Effects
Little mode of action (MOA) information regarding potential nervous system effects
following formaldehyde inhalation is available. To date, there are no definitive data supporting a
specific mechanism for effects on nervous system structure or function. As appreciable amounts of
formaldehyde are not expected to reach the systemic circulation or CNS to elicit direct effects, any
potential mechanisms would need to be indirect Thus, this section focuses on mechanisms that
might secondarily result from alterations to the respiratory system (see Appendix A.5.6). As such,
only data from formaldehyde inhalation studies are discussed, and confidence in the findings based
on individual study evaluations is emphasized (see Appendices A.5.6 and A.5.7). Although none has
been confirmed experimentally, several biologically plausible, but speculative sequences of
mechanistic changes that might support indirect effects can be hypothesized based on the available
formaldehyde-specific data, including:
1) Repeated activation of sensory nerves (e.g., trigeminal, vagal) causing sensitization or
neurogenic inflammation leading secondarily to effects on neuronal populations unrelated
to pain and irritation pathways—based primarily on three medium fAhmed etal.. 2007:
Fujimaki et al.. 2004b: Kulle and Cooper. 1975) and one low confidence (Tsukahara etal..
2006) studies
Repeated stimulation of sensory nerve fibers relaying information related to formaldehyde
exposure to neuronal nuclei might eventually lead, indirectly, to lasting changes in centrally located
neurons or soluble factors; however, specific data assessing this possibility, and the downstream
consequences of such potential changes, remain unexamined. Formaldehyde inhalation has been
shown to increase the electrical activity of trigeminal nasal afferents at concentrations below
1 mg/m3 (Kulle and Cooper. 1975). which appears to cause neurogenic inflammation, a process
whereby stimulation of sensory nerve endings causes localized (e.g., into airway tissue) release of
neuropeptides (e.g., the tachykinin, substance P) that elicit local inflammatory responses (see
discussion in Section 1.2.1). In addition to the "axon reflex" that can be induced upon sensory nerve
stimulation (causing a localized release of factors), if the stimulus is of sufficient intensity or
duration, signaling along ascending pathways from these afferents can continue, and eventually
might lead to central sensitization where the excitability or responsiveness of afferent nerve fibers
is enhanced (Woolf and Salter. 2000).
While changes in neuronal nuclei associated with ascending pathways related to pain and
irritation signals seems likely following formaldehyde inhalation, there are no data or hypotheses
available to inform how this might indirectly affect other neuronal nuclei. Regardless of the
unexplainable connection between sensory nerve stimulation and changes in presumably unrelated
neuronal nuclei, hippocampal neurochemical changes which appear to be related to neurogenic
inflammation, were observed in the absence of neuronal injury in a series of subchronic
formaldehyde inhalation studies by Fujimaki and colleagues at formaldehyde levels as low as
0.1 mg/m3 (Ahmed etal.. 2007: Tsukahara etal.. 2006: Fujimaki et al.. 2004a). Importantly, these
This document is a draft for review purposes only and does not constitute Agency policy.
1-373 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
effects were generally only observed after stimulation with foreign materials known to cause an
allergic response. Although the evidence related to potential neurogenic inflammation has been
primarily observed in the airways, some factors released as a result of this process can be long-
lived, and receptors for these upregulated cytokines and neuropeptides, including substance P, are
prevalent throughout the CNS fDouglas etal.. 20081. These data suggest the possibility that sensory
nerve stimulation of sufficient duration and intensity, perhaps particularly in allergic individuals,
might eventually result in lasting changes in CNS regions that regulate behaviors unrelated to pain
or irritation responses. However, dose-response relationships for the observed mechanistic
changes were unclear and data are not available to inform some of the essential logical connections
that would be necessary to connect peripheral stimulation to these central changes. An additional
uncertainty with this hypothesized relationship is lack of understanding whether and to what
extend this potential mechanism might be involved following chronic exposure. For example,
although a related chemical, capsaicin, also causes neurogenic inflammation, no neurogenic
inflammatory response to subsequent stimuli is observed following long-term exposure to
capsaicin because tachykinins become depleted from sensory neurons (Kashiba etal.. 1997:
Cadieux etal.. 19861. Further, no data are available to inform human relevance and some suggest
responses might differ across species (e.g., distribution of substance P receptors in the brain can
differ across species fRigbv etal.. 200511.
2) Neuronal activation following stimulation of the olfactory epithelium leading, indirectly, to
alterations in neuronal targets unrelated to olfaction or, directly, to alterations in
olfactory-dependent behaviors—based primarily on one high fHavashi et al.. 20041. one
medium fBoia etal.. 19851. and one low confidence fZhangetal.. 20141 study
Formaldehyde is not only a chemical irritant, it is also an odorant, and its odor is typically
detectable at lower levels than those causing irritation. Repeated and prolonged stimulation of
neuronal olfactory receptors in the nasal epithelium at posterior regions of the upper respiratory
tract (URT) might affect neurons along ascending pathways related to olfaction; however, similar to
the hypothesis presented above, no data exist to describe how such changes could indirectly affect
neurons or neuronal regions unassociated with olfaction. Hayashi et al. (2004) reported that
subchronic, but not acute, formaldehyde exposure increases the activity of periglomerular (PG)
cells in the main olfactory bulb (OB). Increases in the number of tyrosine hydroxylase (TH)+ PG
cells were observed at >0.1mg/m3, with no differences in PG cell number or size of the OB
(indicating increased TH synthesis in TH" PG cells rather than new cell formation). These changes
might be related to observed decreases in the synapse protein, SNAP25, in the OB after periodic
exposure (twice daily 30-minute exposures for 14 days) to high levels of formaldehyde fZhang et
al.. 20141. although these latter results are interpreted with low confidence. The results in Hayashi
et al. (2004) appear to highlight sensory-induced adaptive properties of the OB in relation to
dopaminergic function (TH is an essential enzyme for dopamine synthesis). OB dopamine affects
odor detection and can affect odor-related behaviors (e.g., impaired learning was observed with
This document is a draft for review purposes only and does not constitute Agency policy.
1-374 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
increased dopamine D2 receptor signaling by Escanilla et al. (200911. Thus, it is considered
plausible that formaldehyde exposure could modify rodent behaviors with an olfactory component
(e.g., motor-related behaviors; learning and memory in land maze tests); however, the potential for
human behaviors, which are far less reliant on odorant signals, to be significantly impacted is
unlikely.
It is unknown whether the adaptive changes observed in OB neurons result in alterations in
neural circuitry. To date, no electrophysiological experiments have been conducted to specifically
address the potential for an association between formaldehyde exposure and CNS
electrophysiological changes. From the OB, olfactory signals are typically conveyed to higher order
neurons, including those in the amygdala, hypothalamus, and olfactory areas of the entorhinal and
piriform cortex. Possibly in relation to this, there is some suggestion of altered dopaminergic or
serotonergic signaling in the hypothalamus with high-level formaldehyde exposures [6.15 mg/m3;
(Boja etal.. 1985)]. but these changes (increased dopamine and 5-HIAA, a serotonin metabolite)
were only evaluated acutely following exposure, have not been linked to behavioral changes, and
contrast somewhat with suggestive observations of decreases in TH-positive cells across several
brain regions at lower levels fLi etal.. 20161. In addition, it remains speculative to infer that
changes in olfaction-related ascending pathways after formaldehyde exposure might modify neural
cell populations that are likely to be unrelated to those specific olfactory neuronal circuits. Overall,
the cascade of events surrounding these adaptive changes remains unknown.
3) Altered hypothalamus-pituitary-adrenal gland (HPA) axis signaling (possibly linked to
events above) causing persistent, stress-induced changes in behaviors—based primarily on
one high (Sorg etal.. 2001a) and one medium confidence (Sari etal.. 2004) study
Stress can be a strong modifier of behavior, particularly at early lifestages. Sorg et al.
(2001a: 1996) have suggested that behavioral sensitization to formaldehyde may be linked to
alterations in HPA axis control of corticosterone or sensitization of limbic circuitry following
repeated exposure. In support of this hypothesis, elevated numbers of corticotropin-releasing
hormone (CRH)+ neurons in the hypothalamus (at 0.49 mg/m3) and adrenocorticotropic hormone
(ACTH)+ cells in the pituitary gland (at 0.1 mg/m3) were observed after subchronic formaldehyde
exposure fSari etal.. 20041. while increased serum corticosterone (at 0.86 mg/m3) was evident
after exposure for only 4 weeks fSorg etal.. 2001al. These findings may be related to evidence
suggesting depressed hippocampal glucocorticoid responses at 2.46 mg/m3 from a single
short-term (7 day), low confidence study (Li etal.. 2016). CRH and ACTH represent precursor steps
in the release of glucocorticoids into the circulation following HPA axis stimulation, and
corticosterone is the rodent glucocorticoid equivalent of Cortisol in humans. Reported disruptions
in sleep behavior [observed at 2.46 mg/m3 formaldehyde by (Sorg etal.. 2001b)] may also be linked
to HPA axis dysfunction fBucklev and Schatzberg. 20051. In addition to highlighting the potential
for formaldehyde-induced effects on allergy-related responses to impact the HPA axis, Sari et al.
(2004) hypothesized that these stress-related responses might have resulted from neural
This document is a draft for review purposes only and does not constitute Agency policy.
1-375 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
sensitization via amplification of CNS circuits with repeated exposure; however, as previously
mentioned, no well-conducted formaldehyde inhalation studies assessing electrophysiological
endpoints were identified. Although formaldehyde exposure appears to be correlated with HPA
axis-associated changes, no studies describe exactly how these CNS-regulated HPA responses could
be modified by formaldehyde, highlighting a critical information gap. Importantly, the available
studies are unable to rule out the possibility that the stress responses might be caused by the
animal exposure-specific phenomenon of "inescapable stress" highlighted in Sorg et al. (19961. The
available studies have not fully examined the temporal profile of these changes (acute stress
responses are not necessarily adverse), and no studies have demonstrated that formaldehyde-
induced stress leads to persistent neurobehavioral changes, functional alterations (e.g., through
impaired neurogenesis), or neuroanatomical changes.
4) Changes in neuronal health and function due to indirect CNS oxidative stress or excitatory
changes (possibly linked to events described above)—based primarily on two medium
fSongur etal.. 2008: Ahmed et al.. 20071 and three low confidence fMei etal.. 2016: LICM.
2008: Songur etal.. 20031 studies
Markers of oxidative stress in the CNS are commonly associated with altered neuronal
health and behavior. Songur et al. (2008) hypothesized that formaldehyde exposure may cause
persistent brain changes via oxidative damage. Although a linkage between altered redox balance
and hippocampal neuropathology was not tested in the stereological studies from this laboratory
fSarsilmaz etal.. 2007: Asian etal.. 20061. an earlier study fSongur et al.. 20031 observed reversible
upregulation of hippocampal heat shock protein 70, an oxidative stress-responsive protein. Several
other studies using molecular endpoints also support that formaldehyde inhalation may disrupt
brain oxidative stress responses (i.e., increased malondialdehyde and nitric oxide levels; decreased
superoxide dismutase activity and glutathione levels), particularly in the cerebellum, following
high-level formaldehyde exposures in juvenile rats [at 7.36-14.7 mg/m3 in (Songur etal.. 20081]
and adult mice [at ~3 mg/m3 in Mei et al. (2016)]. Songur et al. (2008) observed effects that
persisted up to 60 days post-exposure. Lower level exposures (e.g., 0.123 mg/m3) for up to
24 hours did not cause changes in brain 80HdG: dG ratios fMatsuoka etal.. 20101. The evidence for
oxidative stress in the brain could be related to prolonged increases in inflammatory mediators in
the blood after formaldehyde exposure, including reactive oxygen species, hormones, or other
factors (see Appendix A.5.6); however, this potential linkage has not been tested. Relatedly,
changes in oxidative stress markers might reflect effects on excitatory neurotransmission.
Specifically, acute formaldehyde inhalation has been shown to increase expression of NMDA
receptor subunits (e.g., NR2B) in nasal tissue fHester etal.. 20031 and forebrain regions fLICM.
20081. while subchronic exposure in rats sensitized to allergen increased NMDA receptor
expression f Ahmed etal.. 20071 but not protein levels fTsukahara etal.. 20061. However, the
cause(s) and functional consequences of these reported molecular increases have not been
examined. In general, an explanation for oxidative stress-related changes in the absence of
This document is a draft for review purposes only and does not constitute Agency policy.
1-376 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
systemic distribution of formaldehyde or very high formaldehyde exposure levels is unavailable,
limiting the feasibility of this potential mechanism.
Overall, no MOA for potential formaldehyde-induced nervous system effects is available.
Integrated Summary of Evidence on Nervous System Effects
Numerous human and animal studies were available and, although multiple lines of
evidence suggest that some concern for nervous system effects following formaldehyde inhalation
is warranted, major deficiencies in study conduct were identified and the database is considered
incomplete. No experimentally supported MOA is available to explain how formaldehyde inhalation
could cause nervous system effects, although some potentially relevant mechanistic changes in the
brain have been observed in well-conducted studies. Summary evaluations of the evidence for
potential nervous system effects of formaldehyde inhalation exposure are provided in Table 1-50.
In human studies, evidence of an association between formaldehyde exposure and ALS was
suggested across four studies in different populations by two separate groups of researchers.
Positive associations observed in a large prospective study were somewhat corroborated by a few
(but not most) comparisons in the other studies, noting that some associations were based on a
very small number of cases or secondary analyses. However, three of the studies had uncertainties
in the assignment of individual exposure to formaldehyde and two of the four did not observe a
dose-response relationship when the data were stratified by estimated formaldehyde levels. In
addition, the results were not verified in another study in a different population, which had greater
certainty in individual exposure assessments. Based on these uncertainties, the currently available
human evidence is interpreted as slight. Importantly, however, the unexpected nature of the
observed associations between formaldehyde exposure and this rare and fatal disease across a
growing number of studies (the first association was reported in 2009, with some corroborating
evidence in 2015 and 2016) identifies an urgent need for additional research. As no experimental
animal or mechanistic studies specific to this effect were identified (i.e., indeterminate), overall the
evidence suggests, but is not sufficient to infer, that formaldehyde inhalation might cause the fatal
human disease, ALS, but additional study is needed for a stronger judgment This is primarily based
on epidemiological studies in occupational settings (presumably higher levels of exposure);
however, there were notably uncertainties in the studies' exposure assessments.
Although numerous studies reported changes in behavior following formaldehyde
exposure, the evidence was not considered adequate to support a causal hazard conclusion, as it
was primarily based on rodent studies with notable methodological limitations, with more limited
supporting data from studies in humans. Effects in learning and memory tests, and performance in
tests of motor-related behaviors, were relatively consistent across the available animal data, and
several human studies reported coherent, but more marginal, changes in related tests. However,
the available experiments had significant methodological deficiencies and, overall, the data were
not attributable to formaldehyde alone. Based on the methodological limitations of the available
studies, both the human and animal evidence for effects in learning and memory tests, and on
This document is a draft for review purposes only and does not constitute Agency policy.
1-377 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
motor-related behaviors, is considered slight. Although no established MOA exists for changes in
these behaviors, several well-conducted studies reporting molecular and structural effects in
relevant brain regions (e.g., limbic structures; cerebellum) provide some biological plausibility for
these effects. Taken together, it was judged that the evidence suggests, but is not sufficient to
infer, that formaldehyde exposure might cause these potential behavioral effects.
Somewhat separate from the other reported behavioral effects, formaldehyde inhalation in
rodents was also reported to be associated with sensitization-related changes in behavior. While
several animal studies of varying quality observed amplified behavioral responses after
formaldehyde exposure, interpretation of the results is unclear. Additional data are needed to rule
out any potential influence from factors other than formaldehyde exposure. No human studies
were available to inform this endpoint (i.e., inadequate). In addition, although some biological
plausibility is provided by neurochemical and hormonal changes that may be consistent with such
effects, without mechanistic information to verify that formaldehyde exposure alone resulted in
these effects (e.g., supporting a reasonable MOA or ruling out alternative explanations), the animal
findings are considered slight. As uncertainties also exist in the relevance of these tests to human
exposure scenarios, based on the data overall, it was judged that the evidence suggests that
formaldehyde might cause neural sensitization-related behavioral changes.
Thus, based on the available database of studies, it was concluded that the available
evidence suggests, but is not sufficient to infer, that formaldehyde inhalation might cause
behavioral effects. The primary support for this conclusion is from low confidence studies in
experimental animals, many of which reported effects at <1 mg/m3. Given that this judgment
relates to multiple manifestations of potential behavioral toxicity (i.e., learning or memory; motor-
or anxiety-related activity; and neural sensitization), with some findings reported at low-exposure
levels, this represents a significant data gap.
Data from experimental animal studies also suggest that excessive formaldehyde inhalation
(levels >7 mg/m3) may cause developmental neurotoxicity. The evidence most informative to this
potential health effect was a medium confidence study (i.e., two publications on the same
experiment) examining neuropathological changes in rats; a few low confidence studies reporting
somewhat equivocal evidence for developmental effects other than neuropathology did not
contribute. While the methods used in this study to evaluate developmental neuropathology were
sensitive and designed to minimize bias, and the endpoint (persistently decreased neuron number)
is adverse, of clear concern to humans, and without comparable data to the contrary, there were
notable uncertainties introduced by the study design that warrant replication of the results. These
include a very small sample size [n = 3 litters), as well as lack of control for potential litter effects.
As some mechanistic changes in the hippocampus and related brain regions after developmental
exposure have been reported in well-conducted studies, indirect effects of formaldehyde exposure
on the CNS have some demonstrated plausibility. In the absence of confirmatory studies (e.g., in
other species; by other laboratories; using more informative study designs), the evidence for effects
This document is a draft for review purposes only and does not constitute Agency policy.
1-378 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
in animals is considered slight. No studies in children were available to inform developmental
neurotoxicity (i.e., inadequate). Overall, the evidence suggests, but is not sufficient to infer, that
formaldehyde inhalation might cause developmental effects on the nervous system, primarily based
on a set of neuropathology studies from the same laboratory. The primary support for this
judgment is from animal studies of neuropathology following developmental exposure to >7 mg/m3
of formaldehyde. Given the potential for children to be exposed to formaldehyde, this area
represents a research need.
Overall, conclusive evidence of a nervous system health hazard in humans exposed to
formaldehyde was not identified. Given that, across a number of studies, the evidence suggests,
but is not sufficient to infer, that formaldehyde inhalation might cause multiple manifestations of
nervous system health effects in humans given relevant exposure circumstances (see Table 1-50),
and the general lack of comprehensive and rigorous experiments across the database, additional
study is warranted.
Table 1-50. Evidence integration summary for nervous system effects after
formaldehyde inhalation3
Evidence
Evidence judgment
Hazard Determination
Amyotrophic Lateral Sclerosis (ALS)
Human evidence
Slight, based on:
Human health effect studies:
• Strong association in one medium confidence study, with more limited
support from three additional medium confidence studies (including two
studies from the same researchers).
• However, no association in one high confidence study.
• Effects were from large, well-conducted longitudinal or retrospective studies.
• However, there was uncertainty in individual exposure assessments, lack of
exposure-response trends in studies with adequate data to examine,
inconsistency in associations with duration, and effect estimates based on a
very small number of exposed cases.
Biological plausibility. No relevant mechanistic studies in humans were
identified, and this effect is surprising (i.e., plausibility is lacking) without
systemic distribution.
The evidence suggests,
but is not sufficient to
infer, that formaldehyde
inhalation might cause
increases in ALS
incidence or mortality,
given the appropriate
exposure circumstances'5
Primarily based on
occupational studies
(presumably higher
levels of exposure),
generally with uncertain
exposure assessments.
(Note: Confirmatory
effects in a medium
confidence human study
with a reasonable
number of exposed cases
and more certain
measures of exposure
would be expected to
adjust this to evidence
indicates [likely])
Potential susceptibilities:
ALS disproportionately
affects males, which
Animal evidence
• Indeterminate, based on: No available animal studies address this outcome.
Other inferences
• Relevance to humans: The effect was observed in humans.
• MOA: No verified MOA exists for how formaldehyde could elicit effects in
motor neuron-related systems without systemic distribution. Additional
study into the potential involvement of systemic oxidative stress
(see Appendix A.5.6) is warranted, given research interest in associations
between elevated oxidative stress and ALS progression.
This document is a draft for review purposes only and does not constitute Agency policy.
1-379 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
were the focus of most
of the available
formaldehyde studies.
Developmental Neurotoxicity
Human evidence
• Indeterminate, based on: No available human studies address this outcome.
The evidence suggests,
but is not sufficient to
infer, that formaldehyde
inhalation might cause
developmental
neurotoxicity, given the
appropriate exposure
circumstances'5
Based on a small set of
studies from one
laboratory that exposed
postnatal rats to
formaldehyde
concentrations
>7 mg/m3.
(Note: confirmatory
effects in a medium
confidence animal study
from another laboratory
or in another species,
particularly one testing
lower exposure levels,
would be expected to
adjust this to evidence
indicates [likely].)
Potential susceptibilities:
The available data relate
to postnatal exposure; it
is unknown whether
other lifestages might
exhibit even greater
sensitivity.
Animal evidence
Slight for developmental neurotoxicity, based on:
Animal health effect studies:
• Developmental neuropathology in one medium confidence (reported in two
papers) and one low confidence study of the male rat hippocampus (less
convincing evidence on other endpoints from other low confidence studies
did not contribute).
• No conflicting evidence (i.e., no comparable evaluations).
• The stereological methods used minimize bias, and multiple indications of
toxicity persisted 60 days after exposure.
• However, the studies were conducted by a single laboratory, had a low
sample size, were analyzed on a pup (not litter) basis, and only tested
formaldehyde levels >7 mg/m3 (which complicates interpretation).
Biological plausibility. Several studies with well-conducted exposures (including
developmental exposure) demonstrate molecular and neurochemical changes
in relevant (i.e., limbic) brain regions at lower concentrations, providing
plausibility.
Other inferences
• Relevance to humans: Uncertainty regarding the relevance of the animal
evidence exists, as the studies only tested high levels of formaldehyde
expected to cause strong irritant effects that may not occur in humans;
otherwise, rodent neuropathology is relevant to humans and is adverse.
• MOA: No verified MOA exists for how formaldehyde could elicit CNS effects
without systemic distribution, although evidence related to several indirect
mechanisms of potential relevance was identified.
Neurobehavioral Effects
Human evidence
Indeterminate for neural sensitization, based on: No available human studies
address this outcome.
Sliaht for effects in tests of motor-related behaviors, based on:
Human health effects studies:
• Effects in two low confidence studies and slight effects (near equivocal; not
dose-dependent) in one medium confidence study.
The evidence suggests,
but is not sufficient to
infer, that formaldehyde
inhalation might cause
multiple manifestations0
of potential behavioral
toxicity, given the
This document is a draft for review purposes only and does not constitute Agency policy.
1-380 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Animal evidence
Other inferences
• No effect in one low confidence study.
• Effects were observed across demographics and behavioral tests.
• However, likely coexposures were not always evaluated, and data are
primarily based on acute exposure.
Slight for effects in tests of learning or memory, based on:
Human health effects studies:
• Effects in three low confidence, independent studies.
• No effect in one low confidence study.
• Effects were related to duration of exposure across studies.
• However, the studies had significant coexposures or poorly comparable
groups, and no dose-dependent effects were observed with controlled
exposure.
Biological plausibility (for any of the above behaviors): No relevant human
studies identified.
Slight for neural sensitization, based on:
Animal health effects studies:
• Effects in one medium confidence and five low confidence studies across two
species (rats and mice).
• No contrary results.
• Some studies show that responses increase with increasing duration of
exposure and persist weeks after exposure.
• However, behaviors may be complicated by possible olfaction, irritation, and
stress responses specific to animal exposure scenarios that were untested.
Slight for changes in Tests of motor-related behaviors, based on:
Animal health effects studies:
• Effects in seven low confidence studies across laboratories in both sexes of
rats and mice (multiple strains).
• No effect in one medium confidence study.
• Most responses were dose-dependent and one study reported effects
persisting for weeks.
• However, every study had test article deficiencies or was complicated by
irritation-related responses, and few tests assessed a discrete function
(e.g., motor activity).
Slight for changes in Tests of learning or memory, based on:
Animal health effects studies:
• Effects in five low confidence studies from multiple research laboratories
across various durations of exposure and in both sexes of rats and mice
• No contrary results
• Effect magnitude increased with repeated exposure, and was sometimes
dose-dependent (in two studies) and persisted weeks after exposure (in one
subchronic study)
• However, all studies had test article deficiencies, and most did not evaluate
motor activity as a contributing factor.
Biological plausibility (for any of the above behaviors): Several studies with well-
conducted exposures demonstrate molecular and neurochemical changes in the
brain at comparable or lower formaldehyde levels. Specifically, for sensitization,
animal evidence of changes to circulating stress hormones provides additional
plausible support.
Relevance to humans: For neural sensitization, translatability to human
exposure scenarios and adversity in humans remains unclear, requiring further
study. For the other behavioral changes, the commonly used tests and the
appropriate exposure
circumstances'5
Primarily based on a
number of low
confidence studies in rats
and mice, many of which
observed effects after
formaldehyde exposure
<1 mg/m3.
(Notes: Confirmatory
effects supporting neural
sensitization in one
medium confidence
study from another
laboratory alongside
mechanistic confirmation
of the human relevance
and adversity of the
animal findings would be
expected to adjust to
evidence indicates
[likely]; as the data for
other types of behavioral
effects are only based on
low confidence studies, it
is expected that
confirmatory effects of
behavioral changes other
than neural sensitization
in multiple medium
confidence studies would
be needed to adjust this
to evidence indicates
[likely].)
Potential susceptibilities:
Unknown, as
well-conducted
developmental studies of
these effects were not
identified.
This document is a draft for review purposes only and does not constitute Agency policy.
1-381 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
changes observed at levels not expected to induce irritation are considered
relevant to humans and potentially are adverse.
• MOA (for any of these centrally mediated effects)\ No verified mechanism
exists for how formaldehyde could elicit CNS effects without systemic
distribution; however, several lines of evidence exist to support the potential
for indirect effects on the CNS.
• Other. The duration- and timing-dependence of these potential effects is
unknown, as most of the data are from acute and short-term exposure
(i.e., no chronic studies; one subchronic study of 30 min. daily exposures) of
formaldehyde levels >7 mg/m3 (which complicates interpretation).
Abbreviations: ALS = amyotrophic lateral sclerosis; MOA = mode of action; CNS = central nervous system.
aln addition, a single, cursory experiment on nociception was identified; this evidence was considered inadequate.
bGiven the uncertainty in this judgment and the available evidence, this assessment does not attempt to define what might be
the "appropriate exposure circumstances" for developing this outcome.
cThe available evidence suggests, but is not sufficient to infer, that formaldehdye might cause each of the evaluated
manifestations of potential behavioral toxicity (i.e., neural sensitization, tests of motor-related behaviors, and tests of learning
and memory), either individually or as encompassed by the broader category of neurobehavioral tests.
1.3.2. Developmental and Reproductive Toxicity
Studies in humans, and a number of animal studies have reported effects of inhaled
formaldehyde on pre- and postnatal development and on the female and male reproductive
systems. Three studies evaluated residential exposure during pregnancy and fetal and infant
growth measures, including ultrasonographic biometric measures, birth weight and head
circumference, and postnatal growth The most common outcome reported by occupational
epidemiology studies was an elevated spontaneous abortion risk in different industries, with strong
associations seen in the highest exposure categories. Further, maternal and paternal formaldehyde
exposure was associated with decreased fecundity,25 indicated by a longer time to achieve a
pregnancy, in two studies of employees in the woodworking industry (out of a total set of three
studies). The associations among female workers may reflect either toxicity to the reproductive
system of the mother (ability to achieve and support the pregnancy) or the developing fetus.
Together, the findings among women provide moderate evidence of developmental or female
reproductive toxicity. In animal studies, there is indeterminate evidence for manifestations of
developmental toxicity (i.e., decreased survival, decreased growth, or increased evidence of
structural anomalies) or female reproductive toxicity (ovarian and uterine pathology, ovarian
weight, and hormonal changes). All available studies were of low confidence, primarily due to
exposure-quality concerns (i.e., the use of formalin, or an uncharacterized test substance).
Two studies of exposure to male workers from one research group provide slight evidence
that formaldehyde exposure is associated with lower total and progressive sperm motility, and
delayed fertility and spontaneous abortion. The epidemiological observations are supported by
robust evidence from experimental studies in animals that used paraformaldehyde to expose the
animals. Across this set of studies, coherent evidence for a range of effects on the male
25The capacity to conceive and deliver a baby.
This document is a draft for review purposes only and does not constitute Agency policy.
1-382 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
reproductive system was demonstrated, including quantitative histopathological effects in the
testes and epididymides, decreased serum testosterone (T), decreased sperm count and motility,
and increased sperm morphological abnormalities. However, limitations in the animal study
database for male reproductive toxicity include a general lack of functional measures in the
available studies and no studies that tested formaldehyde levels below 6 mg/m3, warranting
additional study.
Overall, the evidence indicates that inhalation of formaldehyde likely causes increased risk
of developmental or female reproductive toxicity in humans, given the appropriate exposure
circumstances. This conclusion is based on moderate evidence in observational studies finding
increases in time-to-pregnancy (TTP) and spontaneous abortion risk among women with
occupational formaldehyde exposures. The evidence in animals is indeterminate, and a plausible,
experimentally verified MOA explaining such effects without systemic distribution of formaldehyde
is lacking. Likewise, the evidence indicates that inhalation of formaldehyde likely causes
increased risk of reproductive toxicity in men, given the appropriate exposure circumstances, based
on robust evidence in animals that presents a coherent array of adverse effects in two species
testing formaldehyde concentrations >6 mg/mg3, and slight evidence from observational studies of
occupational exposure, and no plausible, experimentally verified MOA explaining such effects
without systemic distribution of formaldehyde. However, some support for indirect effects in
rodents is provided by relevant mechanistic changes in male reproductive organs.
Literature Search and Screening Strategy
The primary databases used for the literature search were PubMed, Web of Science, and
ToxNet, with the last update of the search completed in September 2016 (see Appendix A.5.8), and
a systematic evidence map updating the literature through 2021 (see Appendix F). This included
the identification of studies of specific health outcomes and particular exposure scenarios in studies
of exposed humans, studies of reproductive and developmental toxicity in animals with exposure to
inhaled formaldehyde, and relevant mechanistic data. Animal studies conducted with other routes
of exposure (e.g., oral, IP injection) were excluded because such studies would likely result in target
organ concentrations of formaldehyde and its metabolites that would not be anticipated with
inhalation exposures. The majority of health outcomes assessed in epidemiology studies of
inhalation exposure that were included for further evaluation were studies of fecundability26
(e.g., TTP), reproductive parameters in males, spontaneous abortion, and birth outcomes.
Outcomes assessed in animal toxicology studies that were included in the assessment were
developmental toxicity (prenatal survival, fetal and postnatal growth, and malformations), male
reproductive toxicity (sperm count and morphology, testes and epididymal weight and
histopathology, and functional measures), and female reproductive toxicity (hormone levels,
ovarian and uterine weight and histopathology, and early embryo loss). Functional developmental
26A couple's probability of conception in one menstrual cycle.
This document is a draft for review purposes only and does not constitute Agency policy.
1-383 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
outcomes (i.e., developmental neurotoxicity) were addressed in the sections on the nervous system
(see Section 1.3.1).
The bibliographic databases, search terms, and specific strategies used to search them are
provided in Appendix A.5.8, as are the specific PECO criteria. A literature flow diagram summarizes
the results of the sorting process using these criteria and indicates the number of studies that were
selected for consideration in the assessment through 2016 (see Appendix F for the identification of
newer studies through 2021). These studies in animals and humans were evaluated to interpret
the quality and relevance of the study results regarding hazard identification (see Appendix A.5.8
and below for details).
Methodological Issues Considered in Evaluation of Studies
A variety of different approaches to the assessment of occupational exposure were used in
the epidemiological literature. These ranged from more specific, highly informative measures such
as estimates of job-exposure matrix (JEM)-based TWA concentrations (based on job-specific
formaldehyde measurements and the proportion of time spent at the job reported by participants)
to measures subject to greater misclassification error, such as the self-reported use of specific
products or chemicals, or assignment to exposures by supervisors. Four studies reported by three
independent research groups assigned exposure levels to individual participants using area-level
formaldehyde measurements fWang etal.. 2015: Wang etal.. 2012: Taskinen etal.. 1999: Seitz and
Baron. 1990). Of these, three studies of wood workers used JEMs to increase the accuracy of their
exposure estimates (Wang etal.. 2015: Wang etal.. 2012: Taskinen et al.. 1999).
In the absence of formaldehyde measurements, studies assigned exposure to individuals
based on self-reporting (workprocesses Zhu etal.. 2005: Steele and Wilkins. 1996: Tohn etal.. 1994:
Saurel-Cubizolles etal.. 1994: Taskinen etal.. 1994: Axelsson et al.. 1984). an informed source
fHemminki etal.. 1985: Hemminki et al.. 19821 or occupation/industry codes from census data
combined with expert knowledge of industry-wide concentrations (Lindbohm etal.. 1991). The
studies that collected information about jobs or tasks with a higher probability of formaldehyde
exposure, and the amounts or frequency of exposure, were less likely to be limited by exposure
misclassification (Tohn etal.. 1994: Taskinen etal.. 1994). In two studies of hospital staff,
Hemminki et al. (1985; 19821 identified staff who worked in specific departments and requested
information about chemical exposures, including formaldehyde used as a sterilizing agent, from
their supervising nurses. Supervisors were asked to assign exposures for specific periods
pertaining to the first trimester of identified births that had occurred over several preceding years
(Hemminki etal.. 1985: Hemminki et al.. 1982). In one of these studies (Hemminki et al.. 1985).
hospital staff were categorized as exposed if they used the sterilizing agent or merely used
instruments sterilized with the agent. No information about the amount or frequency of sterilant
use was incorporated in the estimates. Although relying on the nurses' supervisors for exposure
information could reduce the possibility of recall bias, the actual level and frequency of exposure
for some individuals categorized as exposed to formaldehyde may have been very low. Some
This document is a draft for review purposes only and does not constitute Agency policy.
1-384 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
exposure categories were quite broad, including individuals exposed infrequently to low levels
fZhu etal.. 2006. 2005: Steele and Wilkins. 19961. Exposure misclassification and the classification
of individuals with probable low or infrequent exposure as exposed likely resulted in an
underestimate of the effect estimate and was a major limitation in these and other studies
designated as low confidence fZhu etal.. 2006. 2005: Lindbohm etal.. 1991: Hemminki etal.. 1985:
Hemminki etal.. 1982).
A key consideration for the interpretation of developmental and reproductive outcomes
associated with inhalation exposures to formaldehyde in experimental studies was the potential for
coexposure to methanol, a known developmental and reproductive toxicant fU.S. EPA. 20131. when
the test article was an aqueous solution of formaldehyde. Studies that used formalin but did not
control for methanol, and studies that did not characterize the formaldehyde source, are identified
throughout this section. Such studies were assigned a low confidence rating and contributed little
to the synthesis of evidence regarding formaldehyde effects on development or the reproductive
system.
Developmental and Reproductive Effects in Human Studies
The observational studies of reproductive toxicity or pregnancy outcomes evaluated
associations with exposure during pregnancy in three studies and with occupational exposure
among cosmetologists, woodworkers, laboratory workers, and hospital staff. The evidence
regarding TTP, spontaneous abortion, pre- and post-natal growth and other birth outcomes, and
male reproductive toxicity was synthesized, and the studies summarized in Tables 1-51 through
1-54, ordered by the level of confidence in the study result (i.e., high, medium, or low) and then by
publication date.
Time to pregnancy and subfertilitv
TTP is a measure of fertility and has been characterized in terms of number of menstrual
cycles that occurred prior to conception. TTP of greater than 12 months of unprotected intercourse
is indicative of infertility (Wilcox. 2010 p. 1231. Increased TTP might result from potential effects
on gametogenesis, transport, fertilization, migration, implantation, or survival of the embryo (Baird
etal.. 19861. Thus, the measure reflects a potential impact on multiple biological processes,
possibly in both partners, and can be sensitive to the detection of events early during pregnancy
that usually cannot be easily detected in population-based studies. Because it is evaluated in
number of months or menstrual cycles, TTP is informative regarding exposures with impacts over
shorter time periods (e.g., <1 year). TTP is not a measure of infertility as these studies only include
women who became pregnant and had a live birth.
One medium confidence study (Taskinen etal.. 1999) and one low confidence study (Zhuet
al.. 2005) were identified that evaluated effects on TTP in relation to maternal exposure to
formaldehyde (see Table 1-51). TTP was retrospectively ascertained using self-completed
questionnaires fTaskinen etal.. 19991. Taskinen et al. (1999) used an appropriate analytical
This document is a draft for review purposes only and does not constitute Agency policy.
1-385 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
approach, involving the comparison of fecundability27 among four exposure groups. The
association of maternal formaldehyde exposure with TTP became significantly increased in the
highest exposure group with an 8-hour TWA (TWA8) exposure of 0.27 mg/m3. The fecundability
density ratio (FDR) for individuals in the highest formaldehyde exposure category compared to
nonexposed individuals, adjusting for potential confounders and phenol exposure was 0.57 (95% CI
0.37, 0.85). The FDRs for organic solvents, dusts, wood dusts, and phenols in models that adjusted
for potential confounders, including formaldehyde as a coexposure, were all greater than 0.90
(p > 0.05). Therefore, the observed association with formaldehyde was not explained by these
other exposures because they were not associated with longer TTP. FDR was lowest among 17 of
the 39 highly exposed women who did not wear gloves (FDR = 0.51; 95% CI 0.28, 0.92), suggesting
that dermal exposure contributed to increased risk of increased TTP. In addition to the detailed
exposure assignments, Taskinen et al. (1999) reduced the potential for selection bias by recruiting
from female members of a woodworkers union who had been employed at least six months prior to
their pregnancy. Thus, selection into the study was not conditional on being currently employed in
the industry at the time of the study.
Table 1-51. Epidemiology studies describing effects on time to pregnancy in
relation to formaldehyde exposure
Study and design
Results
Reference: Taskinen et al. (1999)
Retrospective cohort study, Finland
Population: Women (n = 3,772), recruited from a woodworkers' union and
other businesses involving wood processing, 1,094 women eligible (born
between 1946 and 1975, had a live birth at age 20-40 years during 1985-
1995, had worked in the wood processing industry for at least 1 month, and
had first employment in the wood processing industry beginning at least
6 months before the index pregnancy). The first eligible pregnancy was the
index pregnancy. Information about personal characteristics, pregnancies,
and exposures was collected from mailed questionnaires; response rate 64%.
After other exclusions (primarily infertility history, unknown TTP, and
contraceptive failure), the final sample included 602 women. Period of recall
of TTP period: 1-11 years.
Exposure: Questionnaire on exposure to specific agents including hours/week
during TTP period. Mean daily exposure to formaldehyde was based on
measurements taken at the factories where the women worked during the
early 1990s or, if measurements unavailable, from comparable industries.
Sampling protocol was not described. Formaldehyde concentrations were
obtained from comparable industries for 46, 31, and 61% of women in low,
medium, and high exposure categories, respectively.
Formaldehyde concentration in factories by exposure category:
Low mean 0.07 ppm (0.086 mg/m3)*, range 0.01 to 0.3 ppm (0.012 to
0.37 mg/m3);
TTP by formaldehyde category
N FDRa 95% CI
Not 367 1.00
Exposed
Low 119 1.09 0.86,1.37
Medium 77 0.96 0.72,1.26
High 39 0.64 0.43,0.92
a Fecundability density ratio adjusted for
employment, smoking, alcohol consumption,
irregular menstrual cycles, and number of
children (recent contraceptive use not found
be a confounder).
TTP among women with high formaldehyde
exposure, by glove use
N FDRa 95% CI
Gloves 22 0.79 0.47,1.23
No gloves 17 0.51 0.28,0.92
aFecundability density ratio adjusted for
employment, smoking, alcohol consumption,
irregular menstrual cycles, and # children.
27Fecundability is the probability of a couple conceiving in 1 month, calculated as the average number of
menstrual cycles to achieve a pregnancy for a group divided by the total number of cycles experienced in the
group.
This document is a draft for review purposes only and does not constitute Agency policy.
1-386 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Medium mean 0.14 ppm (0.17 mg/m3), range 0.05 to 0.4 ppm (0.062 to
0.49 mg/m3);
High mean 0.33 ppm (0.41 mg/m3), range 0.15 to 1.0 ppm (0.18 to 1.2 mg/m3)
Other chemicals with measurements: phenol, organic solvents, wood dust,
other dusts.
Methods: Analysis: discrete proportional hazards regression; outcome, FDR,
ratio of average incidence density of pregnancies in exposed compared to
employed, unexposed women); for covariates in model, see results;
significance assessed by likelihood ratio test.
Evaluation:3
TTP among women with high formaldehyde
exposure and phenol (when included in same
model)3
N FDRb 95% CI
Phenol 68
Formaldeh NRC
yde
1.56
0.57
0.93,2.53
0.37,0.85
SB
IB
Cf ah
Overa 11
Confidence
MecJum
aAII women exposed to phenols were also
exposed to formaldehyde, but not vice versa.
Fecundability density ratio adjusted for
employment, smoking, alcohol consumption,
irregular menstrual cycles, and # children.
cNot reported.
Expect some error in individual exposure assignments.
Reference: Zhu et al. (2005)
Cohort study, Denmark
Population: Exposed were female laboratory technicians, identified through
the Danish National Birth Cohort, who had only held one job (n = 1,069); 1st
interview in June 1997-February 2003 (at week 12-25 of gestation); excluded
women with endometriosis, ovarian or cervical cancer, unplanned or partly
planned pregnancies, and included only 1st pregnancy in study period for each
woman (final n = 829, 77.5% of initial study cohort); 8.6% >35 years old, 13.9%
smoker during 1st trimester; 29.3% previous spontaneous abortion. Referents
were teachers identified in same manner; n = 6,250 (73.9% of initial cohort of
8,461); 12.7% >35 years old, 20.1% smoker during 1st trimester; 31.1%
previous SA
Exposure: Queried at gestation week 12-25 (median week 17). Self-report on
laboratory work processes during pregnancy and 3 months before, including
frequency and use of protective measures.
El calculated as exposure level x frequency of work contact, using scores for
exposure level and frequency:
Formaldehyde exposure level (low = 1, medium = 2), assigned by study
researchers as follows:
Low: human blood and tissue processing, work with experimental animals,
work with microorganisms; medium: preparation of slides for microscopy. No
work processes were identified considered to involve high exposure to
formaldehyde.
Frequency: everyday = 4, several times per week = 3, several days per
month = 2, and rarely = 1.
Exposure Index categories: 1-5 and >6
Methods: Self-report of TTP (4 categories: 0-2 months, 3-5 months,
6-12 months, and >12 months); Fecundability ratios analyzed using discrete-
time survival analysis (complementary log-log link); comparisons between
laboratory technicians and referents (teachers) and among laboratory
technicians; covariates in model see results.
Evaluation:3
Fecundability ratio for 1st pregnancies among
829 laboratory technicians, by formaldehyde
exposure index
El
N
cFR
aFRa
95% CI
1-5
112
1.0
0.92
0.69,1.22
>6
74
1.18
1.03
0.74,1.43
aaFR: adjusted for maternal age, gravidity,
smoking, prepregnancy BMI, and paternal
job (also evaluated history of spontaneous
abortion and alcohol consumption).
Fecundability ratios for 1st pregnancies: labor;
technicians compared to teachers
N cFR aFRb 95%
Teacher
Lab
technician
6,250 1.00 1.00
829
1.01 0.98
0.86
1.13
bFRa: adjusted for maternal age, gravidity,
smoking, prepregnancy BMI, and paternal job
(also evaluated history of spontaneous aborti
and alcohol consumption).
SB
IE Cf Oth
Overall
Confidence
Law
This document is a draft for review purposes only and does not constitute Agency policy.
1-387 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Categorized TTP (decreased precision), missed pregnancies that ended before
1st interview.
Variation in probability or intensity of formaldehyde exposure possible for
work processes across different types of labs and high likelihood of exposure
misclassification (likely underestimating the effect estimate), did not account
for large proportion of participants who used protective measures to prevent
inhalation exposure. JEM was not validated for formaldehyde.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.8). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: TTP = time to pregnancy; CI = confidence interval; El = exposure index; JEM = job-exposure matrix;
FDR = fecundability density ratio; BMI = body mass index.
'Converted study exposure values are presented in [italics]. Conversion factors for formaldehyde in air (at 25°C):
1 ppm = 1.23 mg/m3.
Results from low confidence studies are shaded; these findings are considered less reliable.
Spontaneous abortion
Two medium confidence studies provide evidence (see Table 1-52) that formaldehyde
exposure to female workers is associated with an increased risk of spontaneous abortion. A third
low confidence study contributed information about exposure-response patterns, which was
included as a consideration in the synthesis. These studies examined diverse occupational groups
exposed to different combinations of chemical exposures and products containing formaldehyde
(wood working, cosmetology, research laboratories). Relatively high odds ratios (ORs) of 2-3.5 in
the highest exposure categories were observed (Taskinen et al.. 1999: Tohn etal.. 1994: Taskinen et
al.. 1994). Studies of hospital, nursing, or medical employees generally did not report an
association with formaldehyde exposure, although these low confidence studies tended to use less
precise exposure assessment methods, a major limitation that reduced the sensitivity of these
studies.
All of the studies defined spontaneous abortion, also called miscarriage, as a pregnancy loss
before the 20th week of gestation. Spontaneous abortions were ascertained retrospectively,
primarily using questionnaires, and in several studies these self-reports were included for analysis
only if they could be verified using additional information. Some studies included all eligible
spontaneous abortions recalled by participants (Taskinen etal.. 1999: Steele and Wilkins. 1996).
These studies had greater sensitivity (ascertained early pregnancies prior to clinical recognition).
Validity studies indicate that recall of previous spontaneous abortions is relatively complete,
particularly for losses that occurred after the 8th week of gestation (>80% of recorded spontaneous
abortions were recalled) (Wilcox and Hornev. 1984). Other studies identified spontaneous
abortions directly from a hospital discharge register (Lindbohm etal.. 1991: Hemminki et al.. 1985).
an approach that avoids the limitations of recall bias but is prone to underascertainment of early
recognized losses that do not merit medical attention (Wilcox. 2010).
This document is a draft for review purposes only and does not constitute Agency policy.
1-388 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
All of the studies focused their exposure assessments on the first trimester of pregnancy
(women). The assignment of formaldehyde exposure during this period of susceptibility for
spontaneous abortion fWilcox and Hornev. 19841 was less certain for two low confidence studies,
possibly resulting in misclassification and reduced study sensitivity fSteele and Wilkins. 1996:
Lindbohm et al.. 19911.
Two medium confidence studies conducted analyses or provided details to evaluate
potential confounding by coexposures and found that formaldehyde exposure posed an
independent risk. One study adjusted for other coexposures in the workplace that also posed a
possible risk of spontaneous abortion flohn etal.. 19941. In this evaluation of cosmetologists, an
adjusted OR of 2.1 was reported for use of formaldehyde-based disinfectants (95% CI 1.0, 4.3).
Taskinen etal. (1999) evaluated previous spontaneous abortions reported by female woodworkers,
all of whom had a live birth, using unconditional logistic regression, and adjusted for age,
employment, smoking, and alcohol consumption. No associations were observed for exposure to
phenol, organic solvents, wood, and other dusts. Because formaldehyde was the only exposure
associated with spontaneous abortion, these other work exposures were not confounders in this
analysis. Potential confounding was identified to be a limitation for a study of laboratory
technicians fTaskinen et al.. 19941. This study observed a strong association between formalin
exposure at a frequency of 3-5 days per week and spontaneous abortion (OR = 3.5; 95% CI 1.3, 7.5),
but most of the participants exposed to formalin also reported exposure to xylene, which also was
strongly associated with spontaneous abortion (OR = 3.1; 95% CI 1.3, 7.5). Although potentially
confounded by xylene, the results of this study were compared to those of John et al. (19941 and
Taskinen et al. (19991 to assess a potential bias away from the null. Other studies did not provide
information to evaluate confounding by coexposures and did not provide risk estimates adjusted
for coexposures.
ORs for spontaneous abortion risk in relation to maternal formaldehyde exposure are
plotted in Figure 1-31 and are grouped by industry. The three studies indicate that maternal
formaldehyde exposure is associated with risk of spontaneous abortion among woodworkers,
laboratory workers, and cosmetologists (Taskinen et al.. 1999: Tohn etal.. 1994: Taskinen etal..
19941. Two studies evaluated multiple exposure groups and found that stronger associations were
observed among women in the highest exposure groups (OR range 3.2-3.5). Although Taskinen et
al. (1994) did not control for xylene exposure, which also was associated with spontaneous
abortion risk, the magnitude of the OR among laboratory workers with the most frequent exposure
was comparable to the two higher confidence studies.
This document is a draft for review purposes only and does not constitute Agency policy.
1-389 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Taskinen 1999
Taskinen 1994
10n
~o
Q)
3-
2-
1-
Cosmetology
(< 0.06 mg/m3)
Wood workers
; 0.01 - 1.23 mg/m3)
Laboratory Workers
( « 0.01 - 8.6 mg/m3)
i^
¦3-
(N ®
ii E
"= I
o >
^ c
(N (1)
II E
Og
™ c
CO
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-52. Epidemiology studies describing effects on spontaneous abortion
in relation to formaldehyde exposure
Study and design
Results
Reference: John et al. (1994) United States
Case-control study
Population: 6,202 of 8,356 women (74%) in North Carolina cosmetology
license registry responded to screening questionnaire; 1,249 of 1,696
women (74%) with eligible pregnancy (most recent pregnancy for which
last menstrual period occurred between April 1983 and March 1988)
completed detailed questionnaire. Data obtained on 191 of 267 eligible
spontaneous abortions, and 1,058 of 1,429 eligible live births (1,696 total
abortions and live births); 87% white, 92% high school education, 65%
income <$20,000, mean age 25.9 years.
Exposure: Self-reported exposure through mailed questionnaire to
formaldehyde-based disinfectant products during first trimester.
Other measures of exposure intensity: number of customers, number and
type of chemical services performed per week, number of hours per day
spent standing, disinfection products used, and glove use.
Methods: Three spontaneous abortions were excluded because no
positive pregnancy test or subsequent medical care was reported.
Women working >35 hrs/week as cosmetologists, with or without use of
formaldehyde disinfectants, were compared to women working in other
jobs (referent) during first trimester, and cosmetologists working with
formaldehyde disinfectants were compared with those who did not.
Multivariate unconditional logistic regression.
Evaluation:3
SB 11 Cf Oth
llyorall
[ l
Medium
Selection of most recent eligible pregnancy (potential
underascertainment); no ambient measurements; adjustment for previous
pregnancy loss may introduce bias.
Spontaneous abortions in 7.8% of most recent
pregnancies; mean gestational age for
spontaneous abortion: 9.8 weeks.
Spontaneous abortion among women working
full-time (>35 hr/week) during 1st trimester
# SA ORa 95% CI
Other jobs
Cosmetology work and
no formaldehyde-based
disinfectant use
Cosmetology work and
use of
formaldehyde-based
disinfectant
26
16
51
1.0
0.8
Referent
0.4, 1.6
1.7 1.0,3.0
aAdjusted for mother's age at conception,
previous pregnancy loss, and cigarette smoking.
Spontaneous abortion among women working
full-time (>35 hr/week) as cosmetologists during
1st trimester
formaldehyde # SA ORa 95% CI
disinfectant use
No
Yes
14
47
1.0
2.1
1.0, 4.3
aAdjusted for variables listed above and other
work exposures (hours worked, hours standing,
chemical services, formaldehyde-based
disinfectant, alcohol-based disinfectant, and nail
sculpturing).
ORs increased with standing >8 hours a day and
the number of chemical services/week.
Previous pregnancy loss, >3 pregnancies, and
cigarette smoking were more prevalent among
women with spontaneous abortion.
Reference: Taskinen et al. (1999)
Retrospective cohort study, Finland
Population: Women (n = 3,772), recruited from a woodworkers' union and
other businesses involving wood processing. 1,094 women eligible (born
between 1946 and 1975, had a live birth at age 20-40 years during 1985-
1995, had worked in the wood processing industry for at least 1 month,
and had first employment in the wood processing industry beginning at
least 6 months before the index pregnancy). The first eligible pregnancy
was the index pregnancy. Information about personal characteristics,
For 52 pregnancies with report of previous
spontaneous abortion and same place of
employment for both events (95% CI)
Exposure OR 95% CI
Low
Medium
High
2.4
1.8
3.2
1.2, 4.8
0.8, 4.0
1.2, 8.3
Organic solvents, dusts, wood dusts, and phenols
were not associated with spontaneous abortions.
This document is a draft for review purposes only and does not constitute Agency policy.
1-391 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
pregnancies, and exposures was collected from mailed questionnaires;
response rate 64%. After other exclusions (primarily infertility history,
unknown TTP, and contraceptive failure), the final sample included 602
women.
Exposure: Questionnaire on exposure to specific agents
including hours/week during the period pertaining to TTP. Exposures
during critical exposure period(s) for spontaneous abortion were not
estimated. Mean daily exposure to formaldehyde was based on
measurements taken at the factories where the women worked during
the early 1990s or, if measurements unavailable, from comparable
industries. Sampling protocol was not described.
Formaldehyde concentrations were obtained from comparable industries
for 46, 31, and 61% of women in low, medium, and high exposure
categories, respectively.
Formaldehyde concentration in factories by exposure category:
Low mean 0.07 ppm (0.086 mg/m3)a, range 0.01 to 0.03 ppm (0.012 to
0.37 mg/m3);
Medium mean 0.14 ppm (0.17 mg/m3), range 0.05 to 0.4 ppm (0.062 to
0.49 mg/m3);
High mean 0.33 ppm (0.41 mg/m3), range 0.15 to 1.0 ppm (0.18 to
1.2 mg/m3)
Other chemicals with measurements: phenol, organic solvents, wood dust,
other dusts.
Methods: Self-reported spontaneous abortions occurring prior to the
index pregnancy and at the same workplace were evaluated.
Unconditional logistic regression, ORs, adjusted forage, employment,
smoking, and alcohol; # exposed cases not reported.
Evaluation:3
SB IE tf Qh
m
Overall
Confidence
Medum
Uncertainty regarding exposure measurements with regard to critical
exposure period(s) for spontaneous abortion; excluded women with no
live birth (missing spontaneous abortions to women with no live births).
Reference: Taskinen et al. (1994)
Finland, Retrospective case-referent
Population: Sampled from payroll of state lab personnel (1970,
1975-1986), Finnish Union of Laboratory Assistants (1987), and Register
of Employees Occupationally Exposed to Carcinogens (1979-1986)
Exposure: Self-reported exposure from mailed questionnaire.
Substances listed in questionnaire or open-ended question
Frequency:
Rare: 1-2 days/week
Frequent: 3+ days/week
Reviewed by two occupational hygienists blinded to case status;
8/10 cases and 5/7 referents exposed to formalin were also exposed to
xylene.
Methods: Participants responded to mailed questionnaire regarding
occupational exposure, health status, medications, contraception use,
smoking, and alcohol consumption during 1st trimester (824
returned/1,000 mailed (82.4%)). Sample linked to Hospital Discharge
Register and database of spontaneous abortions treated at hospital
outpatient clinics, 1973-1986. Cases: 206 women aged 20-34 years with
one spontaneous abortion during study period; 329 referents: 2/case
Spontaneous abortion risk by frequency of
formaldehyde exposure
Exposure
Cases/
Referent
OR
95% CI
Employed
0.9
0.5,
1.7
Laboratory
1.4
0.9,
2.2
Formalin
1-2 days/wk
12/28
0.7
0.3,
1.4
3-5 days/wk
11/8
3.5a
1.1,
11.2
ap < 0.05
Other substances also were associated with
spontaneous abortion during 1st trimester; xylene
3-5 days/week (OR 3.1; 95% CI 1.3, 7.5), toluene
3-5 days/week (OR 4.7; 95% CI 1.4,15.9).
This document is a draft for review purposes only and does not constitute Agency policy.
1-392 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
selected from registered births and not a case, matched on age
(24 months) and year of end of pregnancy. Logistic regression for
matched data adjusting for parity, previous miscarriage, febrile diseases
during pregnancy, used contraception at beginning of pregnancy, alcohol
consumption, and employment status.
Evaluation:3
5B
10 Ci Oth
Overall
Confidence
1 1
Low
¦
||
t
1 1
Adjustment for parity and previous miscarriage may introduce bias; lack of
adjustment for xylene, an exposure associated with the spontaneous
abortion and formalin exposure. Evaluation of increasing frequency of use
a strength.
Reference: Steele and Wilkins (1996) United States
Population: 85% of 2,978 eligible women graduating from U.S. colleges of
veterinary medicine during 1970-1980, mean age 36.1 years, 96.2%
White; 1,444 women reported 3,098 pregnancies, 2,375 after graduation.
Exposure: Self-reported job exposure to specific listed chemical or
physical agents (yes, no, don't know). Exposed pregnancy defined if
estimated time of conception was during the reported years of a job for
which exposure also was reported.
Definitions of exposure:
1. Job classification associated with the index pregnancy (type of clinical
practice). Referent pregnancies: women unemployed when pregnancies
began.
2. Specific chemical and physical agents. Referent: employed women
reporting no exposure to that agent or unemployed while pregnant.
Thirteen exposure categories examined: disinfectants, antibiotics, animal
insecticides, formaldehyde, non-DES hormones, solvents, radiation,
diethylstilbestrol, nonhalothane anesthetics, halothane, antineoplastics,
heavy metals, and ethylene oxide.
Methods: Self-reported (via mailed questionnaire in 1987) pregnancy and
employment history. Evaluated eligible pregnancies (live births, induced
abortions, spontaneous abortions) in relation to postgraduate
employment. Spontaneous abortion defined as fetal death prior to
20 weeks. Unconditional multiple logistic regression of spontaneous
abortion in relation to clinical practice type or self-reported exposures
adjusting for maternal age, gravidity, history of spontaneous abortion,
history of smoking, and alcohol use.
Evaluation:3
264 (11.1%) spontaneous abortions.
Analysis limited to women holding only one job at
the time of conception (1,813 pregnancies).
Spontaneous abortions in veterinarians with
self-reported exposure to formaldehyde,
adjusted3 OR (95% CI)
Clinical Exposed OR 95% CI
practice pregnancies (N)
All types
All small
animal
172
115
0.9
1.1
0.6,1.5
0.6, 2.0
3adjusted for age, history of spontaneous
abortion, gravidity, smoker, drinker.
5B IE £f Oh
Overa 11
Canfideme
Low
No information on intensity and frequency of formaldehyde exposure,
which would likely be variable among veterinarians (exposure
misclassification-decreased sensitivity). Adjustment for gravidity and
previous spontaneous abortion may introduce bias.
Reference: Hemminki et al. (1982) Finland
Retrospective cohort
Adjusted spontaneous abortion rate (total
pregnancies (N) and adjusted rate) among
women not exposed and exposed to
formaldehyde during pregnancy
This document is a draft for review purposes only and does not constitute Agency policy.
1-393 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Population: Female nursing staff working in sterilizing units (exposed) or
auxiliary units (referent) in all (approx. 80) general hospitals; 50 exposed
pregnancies, 1,100 unexposed pregnancies.
Exposure: Exposure to sterilizing agents (formaldehyde, ethylene oxide,
glutaraldehyde) at beginning of pregnancy (1960-1980) assigned by
supervising nurse. Blind to case status; 50 formaldehyde-exposed
pregnancies out of 545 total exposed group (9%).
No air monitoring conducted.
Methods: Questionnaire mailed to current supervising nurses to identify
nurses exposed to chemical sterilizing agents and nurses not exposed to
sterilizing agents, X-rays, or anesthetic gases; response in exposed 91.6%;
referent 90.6%.
Spontaneous abortions, 1960-1980, identified via questionnaire sent to
nurses (self-report); compared to Finland hospital discharge register,
1973-1979.
Spontaneous abortion rate (compared to total pregnancies, live births,
induced abortions, spontaneous abortions), logistic regression adjusting
for age, parity, decade of pregnancy, smoking habits, alcohol, and coffee
consumption.
Evaluation:3
Agent
Not Exposed
N Rate
Exposed
N Rate
HCHOa 1,100 8.3
50
8.4
aSome individuals used more than one
sterilizing agent
Adjusted rates among women exposed to
ethylene oxide were higher 16.1% versus 7.8%,
p<0.01.
SB IB Cf Oth
Overall
Confidence
Low
Adjustment for parity may introduce bias. Assumed sterilant use was same
throughout period; no information on intensity and frequency of
formaldehyde exposure (exposure misclassification-decreased sensitivity);
no adjustment for other sterilants.
Reference: Hemminki et al. (1985) Finland
Case-control study
Population: Pregnancies during 1973-1979 among women who worked in
anesthesia surgery, intensive care, operating room or internal medicine
departments of a general hospital.
Exposure: Exposure assessment via questionnaire sent to head nurses at
all general hospitals in Finland. For each study subject, requested
occupation and exposure (yes, no) to any of the listed substances during a
stated 3-month period (1st trimester); blind to case status.
Listed substances were anesthetic gases (nitrous acid, halothane, other),
sterilizing agents (ethylene oxide, glutaraldehyde, formaldehyde),
disinfectant soaps (requested names), cytostatic drugs, and X-rays.
Included information about job: shiftwork, night shift, rotating etc.
Occupation identified during 1st trimester for 87.1% cases and 87.8%
controls. Information on employment and exposure obtained for 81% of
casexontrol sets.
No air monitoring conducted.
Methods: Spontaneous abortions identified by linking Finnish Hospital
Discharge Register with Central Register of Health Care Personnel; 217
cases identified from register as treated for spontaneous abortion
1973-1979 (ICD8 643 & 645).
Controls (n = 571) were nurses who gave birth to a healthy infant
1973-1979 and other pregnancies who were not cases. Selected three
controls per case, matched on age (± 1.5 years), among nurses from same
hospital as case. Relationships between spontaneous abortion and
formaldehyde analyzed using an unmatched crude analysis.
Spontaneous abortion
Crude rate (# cases/# all pregnancies): 8.3%; not
different from Finnish rate: 8.4%
Exposed pregnancies (#) (at least once per week)
among cases and controls (unadjusted OR)
Agent Cases Controls OR
# % # %
HCHO
3.7
24
5.2
0.6
Exposure defined as whether subject used
sterilizing agent or sterilized instruments
This document is a draft for review purposes only and does not constitute Agency policy.
1-394 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Evaluation:3
SB IB Cf Oth
Overall
Confidence
Low
No information on intensity or frequency (exposure misclassification-
decreased sensitivity); very small number of exposed cases.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.8). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: SA = spontaneous abortion; OR = odds ratio; CI = confidence interval; HCHO = formaldehyde.
Birth outcomes
The epidemiology literature is very limited regarding formaldehyde exposure and birth
outcomes (see Table 1-53). One birth cohort study reported decreases of 0.044 and 0.056 in the z-
scores for birth weight and head circumference, respectively, with each 1 |ig/m:i unit increase in
formaldehyde concentration measured in the mother's homes at 34 weeks gestation (Franklin etal..
20191. Gestational age was not associated with exposure. The median concentration in the homes
was 0.0028 mg/m3 and 23.3% of samples were below the LOD in this relatively small study.
Another pregnancy cohort study in South Korea observed lower birth weights associated with
increasing formaldehyde concentration measured at mid to late pregnancy (mean concentrations
were 0.08 mg/m3), although the associations were of greater magnitude for total volatile organic
compounds, which were correlated with formaldehyde levels (Chang etal.. 2017). Another study of
pregnant women in the southeastern United States, rated as low confidence, reported an
association of biparietal diameter, suggestive of intrauterine growth retardation, with personal
formaldehyde exposure >0.037 mg/m3, both measured in the second trimester (Amiri and Turner-
Henson. 2017). Preterm birth and low birth weight were not associated with exposure to high
formaldehyde concentrations among a cohort of male woodworkers in China fWang etal.. 20121.
An elevated association with congenital malformations and maternal exposure was
reported by a limited set of low confidence studies among female hospital or laboratory workers
(Zhu etal.. 2006: Hemminki etal.. 1985). The precision of the ORs was low, as indicated by the
wide CIs generally overlapping 1.0. In addition, the studies evaluated associations for all or major
malformations grouped together. These outcomes may be etiologically distinct, so this lack of
specificity limits the ability to interpret these results. The probability or frequency of exposure to
formaldehyde likely was low in these studies, which would have limited the ability to detect
differences across various exposure groups for these rare outcomes fHemminki etal.. 1985: Ericson
etal.. 19841.
This document is a draft for review purposes only and does not constitute Agency policy.
1-395 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-53. Epidemiology studies describing effects on prenatal growth and
births outcomes in relation to formaldehyde exposure
Study and design
Results
Reference: Franklin et al. (2019)
Birth cohort study, Australia
Population: Pregnant women, all nonsmokers, recruited prior to
18 weeks gestation. 305 of 373 recruited, 81.7% participation; Birth
data available for 262 live births. N=129 males and N=133 females,
gestational age 38.97 weeks (6 infants born at 36-37 weeks).
Exposure: Air monitoring in homes at 34 weeks gestation, 7-day
sampling duration using validated passive samplers in bedroom and
living room. LOD 2.4 |ag/m3; used LOD/2 for values 0.03 ppm (0.037 mg/m3),
(p< 0.013).
Multiple linear regression adjusted for race. Maternal
age and fetal sex were not associated.
This document is a draft for review purposes only and does not constitute Agency policy.
1-396 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Exposure: Personal exposure during 2nd trimester, vapor monitor
badges, 24-hour period, detection limit 0.003 ppm.
Mean (SD) 0.04 (0.06) ppm; 0.049 (0.074) mg/m3
Methods: Ultrasonographic biometry during 2nd trimester for head
circumference, abdominal circumference, femur length, biparietal
diameter, estimated fetal weight, and ratio of abdominal
circumference to femur length. Measurements in mm converted to
percentiles using gestational age and the Hadlock formulas.
Evaluation:3
Other biometric measures were not associated with
formaldehyde exposure.
SB IB Cf Oth
m
Overall
Confidence
Low
Low participation rate with no comparisons of participants with
nonparticipants raises concern for selection bias. Small sample size
with reduction in sensitivity. Reference population for BPD measure
was not appropriate for >50% of participants. Potential incomplete
control for smoking; collection methods and timing were not
described.
Reference: Hemminki et al. (1985)
Case-control study, Finland
Population: Pregnancies during 1973-1979 among women who
worked in anesthesia surgery, intensive care, operating room, or
internal medicine departments of a general hospital.
Exposure: Exposure assessment via questionnaire sent to head nurses
at all general hospitals in Finland. Reported occupation for each name
and whether exposed to listed substance during a stated 3-month
period (1st trimester); blind to case status.
Substances were anesthetic gases (nitrous acid, halothane, other),
sterilizing agents (ethylene oxide, glutaraldehyde, formaldehyde),
disinfectant soaps (requested names), cytostatic drugs, and X-rays.
Included information about job: shiftwork, night shift, rotating etc.
Occupation identified during 1st trimester for 87.1% cases and 87.8%
controls.
No air monitoring conducted.
Methods: Congenital malformations identified by linking with Register
of Congenital Malformations; 46 cases 1973-1979.
Controls were nurses who gave birth to a healthy infant 1973-1979
and other pregnancies were not cases. Selected three controls per
case, matched on age (± 1.5 years), among nurses from same hospital
as case. Congenital malformation controls: 128.
Evaluation:3
Congenital Malformations
Exposed pregnancies (E) (at least once per week) and
total pregnancies (T) among cases and controls
(unadjusted OR)
Agent Cases Controls OR
E/T % E/T %
HCHO
3/34
5/95
5.3
1.8
Exposure defined as whether subject used
sterilizing agent or used sterilized instruments
(only one nurse sterilized instruments)
SB IB Cf Oth
Overall
Confidence
Low
No information on intensity or frequency (exposure misclassification-
decreased sensitivity); very small number of exposed cases.
This document is a draft for review purposes only and does not constitute Agency policy.
1-397 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Reference: Zhu et al. (2006)
Cohort study
Population: Source: Danish National Birth Cohort; 30-40% of all
pregnant women in Denmark, 1st interview June 1997-February 2003;
1,025 of 1,069 pregnancies of laboratory technicians with one job at
interview and 1st pregnancy; excluded induced abortions,
hydatidiform mole, or unknown outcomes of pregnancy (95.9% of
eligible); 9.7% >35 years old, 14.9% smoker during 1st trimester;
27.7% previous spontaneous abortion. Referent: 8,037 of 8,461
teachers; 14.6% >35 years old, 22.1% smoker during 1st trimester;
29.6% previous spontaneous abortion.
Exposure: Queried at gestation week 11-25 (median week 16).
Self-report on laboratory work processes during pregnancy and
3 months before including frequency and use of protective measures.
JEM: El = Exposure level times Frequency of work contact
Exposure level: low (1), medium (2), and high (3); assigned by study
researchers
For formaldehyde: low: human blood and tissue processing, work with
experimental animals, work with microorganisms; medium:
preparation of slides for microscopy. No work processes were
identified with high exposure to formaldehyde.
Frequency: everyday (4), several times per week (3), several days per
month (2), and rarely (1); El categorized into two levels: 1-5 and >6.
Methods: Cohort linked to National Hospital Register and Medical
Birth Register, Cox regression and hazard ratios for late fetal loss and
congenital malformations; laboratory technicians compared to
teachers and comparisons within laboratory technicians. Adjusted for
maternal age, history of spontaneous abortion, gravidity,
prepregnancy BMI, smoking, paternal laboratory job, alcohol
consumption, child's sex (some models).
Evaluation:3
ORs for 1st pregnancies among 991 laboratory
technicians by formaldehyde exposure category (N,
adjusted OR, [95% CI]).
Exposure Index
£6
0 1-5
"Major" malformation
20,1.0 20, 1.2 (0.6,2.1) 16, 1.5 (0.8, 2.9)
Unexposed technicians were exposed to other work
processes.
SB IB Cf Oth
Overall
Confidence
Low
Variation in probability or intensity of formaldehyde exposure
possible for work processes across different types of labs, did not
account for large proportion of participants who used protective
measures to prevent inhalation exposure. JEM was not validated for
formaldehyde.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.8). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: OR = odds ratio; El = exposure index; BMI = body mass index; JEM = job-exposure matrix.
1 Male reproductive toxicity
2 Two studies (one medium and one low confidence) of male woodworkers in China from one
3 research group reported associations with lower sperm motility (total and progressive), delayed
4 fertility and spontaneous abortion (Wang etal.. 2015: Wangetal.. 2012). Eligible participants were
This document is a draft for review purposes only and does not constitute Agency policy.
1-398 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
of Han Chinese ethnicity and were occupationally exposed for at least 24 months. A detailed
exposure assessment involved formaldehyde measurements and individual information regarding
workplace, work tasks, time spent at work tasks, and duration of employment. Progressive motility
and total motility were inversely associated with formaldehyde exposure index, a cumulative
measure of exposure, and a strong association with this exposure metric also was observed in
logistic models of below-normal values of these motility measures. For example, ORs of 2.58 and
3.41 were found for progressive motility less than 32% in the low and high exposure groups,
respectively, compared to the community-based referent group. Lindbohm et al. (19911 reported
no association with spontaneous abortion identified from a nationwide hospital discharge register
in relation to male formaldehyde exposure assessed using census data. There was a high likelihood
of exposure misclassification using this assessment method, which reduced the sensitivity of the
study (i.e., judged as low confidence) to identify an association with developmental endpoints. In
another study, no statistically significant differences in sperm counts or percentage of abnormal
sperm were observed in an underpowered, low confidence study of autopsy workers (Ward et al..
19841 (see Table 1-54).
Table 1-54. Epidemiology studies describing male reproductive toxicity in
relation to formaldehyde exposure
Study and design
Results
Reference: Wang et al. (2015) China
Regression analysis of sperm parameters and
Prevalence
formaldehyde exposure index
Population: Woodworkers; N = 124 participated (62.3%), N = 10 with
P
95% CI
missing semen data, aged 23-40, Chinese Han ethnicity, occupational
Volume (mL)a
-0.02
-0.08, 0.03
exposure at least 24 months; excluded men living in newly built or
Concentration
-0.02
-0.19, 0.14
recently remodeled house, men with genital malformations or other
(106/mL)a
chronic disease; N = 81 (40.5%) recruited referent group age-matched,
Total sperm count3
-0.20
-0.68, 0.29
male Han volunteers from same area (salesmen and clerks), N = 5 with
Sperm progressive
-0.19
-0.25, -0.12
missing semen data.
motility (%)b
Exposure: Sampling: 25-minute samples at three times on one workday,
Total motilityb,c
-0.23
-0.30, -0.16
same day as questionnaire. Exposure information based on workplace,
aRelative percentage change
work tasks, work duration, and time (referenced Wang et al., 2012).
bAbsolute change
Exposure index based on formaldehyde concentration (mean of three
Progressive motility plus nonprogressive motility
samples) multiplied by exposed work time during work day and
exposure duration (years). Two categories with cutpoint at median.
No association with kinematic parameters
Concentrations: Exposed 0.22-2.91 mg/m3, exposure index 4.54-
195.08, median 56.55; referent 0-0.02 mg/m3. Measurement and
Logistic regression of below-normal values of
adjustment for other contaminants was not described (e.g., phenols).
sperm parameters and formaldehyde exposure
Methods: Semistructured interview questionnaire, genital examination,
index (below and above median, compared to
semen collection (2-7 days after abstinence), and analysis (within
referent (N = 76)
2 weeks of formaldehyde sampling); parameters were semen volume,
Low {N = 57)
High
sperm concentration, total sperm count, sperm progressive motility,
(N = 57)
total sperm motility, and kinematic parameters (WHO, 2010). Linear
Semen volume
1.83
2.28
regression Ln-transformed semen parameters and formaldehyde
(<1.5 mL)
(0.63, 5.36)
(0.75, 6.91)
exposure and logistic regression of abnormal semen parameters.
Concentration
1.67
1.25
Models adjusted for age, BMI, education, income, smoking, alcohol, and
(<15 x 106/mL)
(0.33, 8.43)
(0.21, 7.35)
abstinence duration.
Total sperm count
1.59
1.73
Evaluation:3
(<39 x 106)
(0.45, 5.61)
(0.49, 6.15)
This document is a draft for review purposes only and does not constitute Agency policy.
1-399 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
SB IB
Cf
Oth
Overa II
Confidence
Medium
¦
Progressive
motility (<32%)
Total motility
(<40%)
2.58
(1.11, 5.97)
3.21
(1.24, 8.28)
Other workplace exposures in woodworking industry (solvents) have
been associated with sperm motility but not accounted for; however,
otherwise strong design and analysis, including evaluation of increasing
exposure-response relationship.
3.41
(1.45, 7.92)
4.84
(1.83,
12.81)
Reference: Wang et al. (2012). Retrospective cohort, 2007-2009
China
Population: Woodworkers; 302 eligible of 1,035 married men, aged
23-40, Chinese Han ethnicity, occupational exposure at least 24 months;
excluded 733 couples living in newly built or recently remodeled house
before and during pregnancy, couples who never tried to conceive,
couples with genital malformations or other chronic disease, wives with
occupational exposure to reproductive toxicants, pregnancies before
husband's formaldehyde exposure and data incomplete; 305 of 816
recruited referent group age-matched, married male Han volunteers
from same area (salesmen and clerks)
Exposure: Mean daily exposure for each worker: Reported workplace,
work tasks, and hour per day exposed to formaldehyde; concentration
monitored three times during different periods.
Daily exposure index: Mean formaldehyde concentration times
proportion of exposed work time during work day multiplied by 100
[cited exposure assessment by Taskinen et al. (1999)1.
Daily mean concentration categorized in low (n = 151) and high
(n = 151), equal number in each group.
Formaldehyde sampling details not provided (concentrations, sampling
protocols, sampling locations, etc.). TWA formaldehyde concentrations
were not reported. Measurement and adjustment for other
contaminants was not described (e.g., dust, phenols)
Methods: Semistructured interview questionnaire. Most recent
pregnancy; TTP: # months of unprotected intercourse leading to
pregnancy; spontaneous abortion defined as termination of pregnancy
prior to 20th week gestation; preterm: <37 weeks; low birth weight:
2,500 g; major structural birth defects.
Spontaneous abortion
Evaluation:3
SB
IB
Cf Oth
Ove ra 11
Confidence
1 1
1
Medium
1
Other workplace exposures in woodworking industry (solvents) have
been associated with spontaneous abortion but not accounted for;
Analysis of most recent pregnancy: possible selection for live births
(time-lapse bias) and impact of gravidity on spontaneous abortion
Time-to-pregnancy
Ove ra 11
SB
IB
et
Oth
Confidence
Medium
OR (95% CI) associated with paternal formaldehyde
exposure
Exposed: High: Low
Referent
TTP >12
2.83
2.29
months
(1.08, 7.41)
(0.78, 6.77)
Spontaneous
1.92
1.78
abortion
(1.10, 3.33)
(0.88, 3.62)
Preterm birth
1.25 (0.55,
0.85
2.84)
(0.28, 2.60)
Low birth
1.26
1.0
weight
(0.59, 2.66)
(0.37, 2.74)
Birth defects
2.61
1.26
(0.79, 8.65)
(0.33, 4.78)
Significant covariates: BMI, alcohol
Significant covariates: Cigarette smoking
Significant covariates: Education
Significant covariates: Alcohol
Logistic regression model adjusted for confounders
identified through univariate analyses.
Confounders considered: age, BMI, education,
income, smoking, alcohol, and frequency of
intercourse.
The numbers of exposed and referent cases were
not presented.
This document is a draft for review purposes only and does not constitute Agency policy.
1-400 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
Exposure levels not reported (but robust assessment method).
Dichotomized TTP in analysis (low sensitivity).
Reference: Lindbohm et al. (1991) Finland; Registry linkage
Population: All Finnish women with diagnosis of spontaneous abortion
(ICD—8 643, 645), induced abortion (ICD-8 640-642), or birth (ICD-8
650-662) between 1973 and 1982 were identified using the nationwide
Hospital Discharge Register and hospital outpatient records.
Information on occupation and industry of women and their husbands,
and SES (women only), was obtained from Finnish national censuses
from 1975 to 1980. Excluded pregnancies among women <12 years or
>50 years of age, and those lacking data on occupation, industry, or SES.
Final study population included 99,186 pregnancies ending Jan. 1-Dec.
31, 1976 or May 1, 1980-Apr. 30,1981.
Exposure: Job-exposure classification developed by two industrial
hygienists using combinations of occupation and industry with similar
type of exposure. Identified jobs held during census period close to
period of susceptibility. List of toxic agents associated with job groups
developed using air sampling data from Finnish occupational health
agency and register of employees occupationally exposed to
carcinogens.
Exposure categories:
1. Not exposed
2. Potential, low: jobs with low levels but high prevalence of exposure,
jobs without exposure data but in register of occupational exposure to
carcinogens, or jobs with high level but unknown prevalence of
exposure
3. Moderate or high: jobs with levels >TLV, or periodically >TLV and high
prevalence
Paternal exposure to any mutagenic agent:
Not exposed: 87,616
Potential, low: 9,930
Moderate/high: 1,640
Methods: Logistic regression models were used to evaluate association
between spontaneous abortion and paternal occupation or industry
during period of susceptibility (spermatogenesis 80 days prior to
conception, or 1st trimester).
Evaluation:3
SB IB Cf Oth
Ove ra 11
Confidence
Low
Industry/occupation coding has low specificity; potential exposure
misclassification and imprecise assignment of exposure period to period
of spermatogenesis relevant to identified pregnancy.
Spontaneous abortion rate 8.8% (including induced
abortions in denominator).
Spontaneous abortion risk by paternal exposure
to formaldehyde3
Group N Cases ORb 95% CI
Not
exposed
Potential,
low
Mod/High
87,616 7,772 1.0
1,212 110 1.1 0.9, 1.4
596 54 1.0 0.8, 1.4
aAmong 25 evaluated exposures.
bAdjusted for maternal age, socioeconomic status,
and maternal exposure to potential reproductive
hazards.
Paternal exposures to solvents (petroleum
refineries), rubber production solvents, rubber
chemicals, and ethylene oxide were associated with
increased odds of spontaneous abortion (p < 0.05).
Reference: Ward et al. (1984) Texas
Population: Exposed: 11 male pathologists and coworkers at university
autopsy service. Matched referent: 11 staff and students in medical
branch; matched on sex, age, tobacco, alcohol, and recreational drug
use.
Exposure: Area and personal breathing zone samples; exposures
episodic, maximum 5.8 ppm (7.13 mg/m3),* LOD = 0.12 mg/m3
TWA 0.61-1.32 ppm (0.75-1.62 mg/m3)
Methods: Morning semen samples every 2-3 months. Sperm counts
and morphology (percentage abnormal); three samples per subject at 2-
Sperm abnormalities (mean [SD]) by
exposure group
Exposed Referent
Count3
percentage
abnormal
62.9 (49.9) 87.4 (75.0)
44.5 (13.4) 53.5 (16.2)
3 millions/cc of semen
Differences between exposed and referent were
reported to be not statistically significant.
This document is a draft for review purposes only and does not constitute Agency policy.
1-401 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
Study and design
Results
to 3-month intervals; mean value analyzed; Pearson correlation
coefficients.
Evaluation:3
Ove ra 11
SB
IB
et
Oth
Confidence
Low
¦
Small sample size; uncertainty regarding reliability of morphology
scoring.
Evaluation of sources of bias or study limitations (see details in Appendix A.5.8). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: BMI = body mass index; TWA = time-weighted average; SD = standard deviation.
Converted study exposure values are presented in (italics). Conversion factors for formaldehyde in air (at 25°C):
1 ppm = 1.23 mg/m3.
Developmental and Reproductive Effects in Animal Studies
This section provides a separate discussion of the available experimental animal studies on
developmental toxicity, female reproductive toxicity, and male reproductive toxicity, which are
separately summarized in Tables 1-55,1-56, and 1-57, respectively. For each of these three
categories of health effects, the discussion is organized based on the types of endpoints evaluated,
and the evidence tables are organized by endpoint, study confidence (if applicable; see
Appendix A.5.8 for details), species, and lowest formaldehyde exposure level tested.
Two of the studies that assessed developmental toxicity evaluated a standard battery of
developmental endpoints following inhalation exposure of formaldehyde to rats on gestation days
(GDs) 6-15 fMartin. 19901 or GD 6-20 fSaillenfaitetal.. 19891 (i.e., during [at a minimum] the
period of major organogenesis in the rat). Both of these studies had limitations. Martin (1990)
employed robust exposure methods, but failed to report methodological details and quantitative
results. In contrast, Saillenfait et al. (1989) was well reported, but rodents were exposed to
formalin (including 10% methanol), which introduces substantial uncertainty regarding the role of
formaldehyde in the observed effects. Importantly, of these two studies, only Saillenfait et al.
(1989) identified adverse developmental outcomes. There are also reports identifying
developmental effects resulting from formaldehyde exposures administered throughout gestation
to rats (Monfared. 2012: Kum etal.. 2007: Senichenkova and Chebotar. 1996a: Senichenkova.
1991a: Kitaev etal.. 1984: Sheveleva. 1971: Gofmekler et al.. 1968: Pushkina etal.. 1968). Evidence
that inhalation exposures to formaldehyde might affect the female reproductive system in rats is
limited to three studies that are considered to be low confidence (Wang etal.. 2013: Maronpot etal..
1986: Kitaev etal.. 19841. However, all of the available animal studies of female reproductive
This document is a draft for review purposes only and does not constitute Agency policy.
1-402 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
toxicity and developmental toxicity had serious methodological limitations, most notably poor
methods used in conducting formaldehyde exposures, and are all interpreted with low confidence.
Additionally, studies in rodents demonstrated that formaldehyde adversely affects the male
reproductive system after inhalation exposures of varied durations. Some of the studies were
considered as high to medium confidence fVosoughi etal.. 2013: Vosoughi etal.. 2012: Ozen etal..
2005: Ozen etal.. 2002: Sarsilmaz etal.. 1999): however, all of the available medium and high
confidence studies exposed animals to high formaldehyde concentrations (>5 mg/m3]. The other
available studies, including many testing lower formaldehyde levels, had methodological limitations
that resulted in their consideration as low confidence studies fHan etal.. 2015: Zhou etal.. 2011a:
Zhou etal.. 2011b: Golalipour etal.. 2007: Xing etal.. 2007a: Zhou etal.. 2006: Appelman etal..
19881. Studies examining developmental immunotoxicity following gestational exposure and
developmental neuropathology following postnatal exposure were discussed previously (see
Sections 1.2.3 and 1.3.1, respectively).
Developmental toxicity
The formaldehyde database contains results of studies that evaluated effects on pre- or
postnatal development following inhalation exposures (see Table 1-55; Figure 1-32). The evidence
table is organized by several major manifestations of developmental toxicity fU.S. EPA. 19911:
survival, growth, and morphological development (Functional developmental toxicity is not
addressed here.) Because all of the developmental toxicology studies have limitations that result in
low confidence ratings, studies within each category are presented in alphabetical order by author
in the table. The results of these studies are presented in Figure 1-32.
This document is a draft for review purposes only and does not constitute Agency policy.
1-403 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
mE 10
"SB
£
c
o
*«p
ro
i—
+¦>
c
® 1
U -L
c
o
u
O)
"O
>
-C
2
CO
E
0.1
W)
5
0.01
O Non-significant • Shading: statistically significant (black) or > 10% change (gray) from controls
o
=0:
ZQi
Q _Q_
T
f
O • •
Design:
Species:
Endpoirrt:
Confidence:
4 mos. G6-15
Pregest.
n>5
n=25
G6-20 Gl-19 Gl-19 Gl-19
n=25 n=46 n=29 n=15
Rats
Fetal Survival
O
o o
C5 O
o
:
Premat. G1-P21 G6-15
-G21 or Pl-42 n=25
n=12 n=6
G6-20 Gl-19 Gl-19
n=25 n=46 n=15
Rats
Fetal and Postnatal Growth
LOW
G6-16
n=10
Mice
C)
o C)
?5=i=
o
o
G6-15 G6-20 Gl-19 Gl-19
n=25 n=25 n=46 n=29
Rats
Fetal Morphology
Figure 1-32. Animal studies evaluating the effects of formaldehyde inhalation
exposure on developmental toxicity.
Low confidence animal studies of developmental toxicity are presented. As no high or medium
confidence experimental animal studies were identified (see Appendix A.5.8), the available studies are
organized by endpoint, then species, then by timing of exposure (e.g., premating [premat.] or
pregestational [pregest.]; gestational [g= gestational day]; or postnatal [p = postnatal day] exposure).
Filled shapes indicate statistical significance, as indicated by the study author (black), or >10% change
from control groups (gray). The size of the points reflecting the sample size for that particular exposure
group (larger size = larger n). The low confidence experiments are shown on a gray background, as the
identified study limitations substantially reduce confidence in the reliability of the results; these low
confidence experiments contribute very little to the weight of evidence for developmental toxicity.
This document is a draft for review purposes only and does not constitute Agency policy.
1-404 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Fetal survival
Decreased prenatal survival following developmental exposures was observed as increased
preimplantation loss by Kitaev et al. (1984) at 1.5 mg/m3 and by Sheveleva (1971) at 0.5 mg/m3 or
increased postimplantation loss at 0.5 mg/m3 by Senichenkova and Chebotar (1996b). The
evidence for these outcomes across the available studies is inconsistent For example, only Kitaev
et al. (1984). Senichenkova (1991b). and Sheveleva (1971) treated the dams during the
preimplantation period (i.e., GD 0-6 in rats) and specifically indicated that preimplantation loss
was examined. Kitaev et al. (1984) found degenerated embryos on GD 3, but not GD 2 (which could
reasonably have been the result of continued exposure of the embryos to stressors resulting from
formaldehyde exposure, and may not have been an inconsistency in response); however, increased
preimplantation loss was not observed by Senichenkova (1991b). The increased postimplantation
loss reported by Senichenkova and Chebotar (1996a) was not observed by Senichenkova (1991b).
in spite of the fact that these two studies used the same procedures and exposure levels, nor was it
reported by Sheveleva (1971). Saillenfait et al. (1989). or Martin (1990). The reason for these
varied responses is unknown, although they might have been influenced by differences in study
protocols or study conduct that are not transparently elucidated in the publications. Because of
limitations in the description of methods or results for most of these studies, it is not possible to
conduct an in-depth evaluation of this issue.
Fetal and postnatal growth
Evidence of decreased or delayed fetal or early postnatal growth was noted in a number of
studies, but a consistent pattern of response was difficult to identify due to differences in study
protocols and study quality. Following gestational formaldehyde exposure, significant 24-32%
decreases in fetal body weight (accompanied by alterations in placental weight and ultrastructural
conformation of the placenta) were observed in mice at exposure levels of >5.68 mg/m3 by
Monfared (2012). Saillenfait et al. (1989) reported significant fetal weight decreases in rats of 5%
at 24.6 mg/m3 and of 19-21% at 49.2 mg/m3. However, fetal weight deficits were not noted by
Martin (1990) at exposure levels up to 12.3 mg/m3 or by Sheveleva (1971) at 5 mg/m3. Conversely,
significantly increased fetal body weight was noted in some studies following gestational exposure
to comparatively lower exposure levels of formaldehyde, e.g., Gofmekler et al. (1968) (7% and 13%
increased fetal weight at 0.012 and 1 mg/m3, respectively) and Senichenkova (1991b) (a 5%
increase at 0.5 mg/m3). It is possible that such findings might be more subtle signals for
developmental disruption of metabolic regulation and function. At 7.38 mg/m3, Kum et al. (2007)
found significant 31% decreases in rat pup weights at 3 weeks of age following in utero and
lactational exposures and significant 14% decreases at 6 weeks of age (i.e., around the time of
puberty) following 6 weeks of exposure starting at birth. Body weight decreases (9%) in young
adult rats after 6 weeks of exposure starting at 4 weeks of age did not reach statistical significance.
Notably, the same outcome did not occur when adult rats on the study were treated for 6 weeks.
This document is a draft for review purposes only and does not constitute Agency policy.
1-405 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
These findings suggest the possibility of a life stage-related susceptibility to formaldehyde
exposures. Gofmekler et al. (1968) reported significantly decreased neonatal relative liver and lung
weights (~5 and 20%, respectively) following gestational exposures to >0.012 mg/m3. A 2-3-day
increase in the mean postnatal day on which incisor eruption occurred, another indicator of
delayed postnatal growth, was reported in rat pups that had been exposed in utero to 0.5 mg/m3
(Senichenkova. 1991a).
Fetal morphological development
Morphological alterations of fetuses exposed in utero were reported in three studies
(Senichenkova and Chebotar. 1996a: Senichenkova. 1991a: Saillenfaitetal.. 1989). Senichenkova
(1991b) and Saillenfait et al. (1989) observed delayed skeletal ossification of various bones, some
of which are generally consistent with developmental delays, at 0.5 and 49.2 mg/m3, respectively.
However, Senichenkova (1991b) noted significantly increased metatarsal and metacarpal
ossification centers; this finding suggests more advanced ossification states rather than a delay in
development and is consistent with the finding of increased fetal weights in that study.
Senichenkova (1991b) also reported an increase in litters with uncharacterized internal organ
anomalies at 0.5 mg/m3. The only outcome specific to reproductive system development was a
reported ~20% increase in "cryptorchidism" by Senichenkova and Chebotar (1996a) and
Senichenkova (1991b) at 0.5 mg/m3; this was interpreted as evidence of a delay in fetal (i.e., 1st
stage) testes descent No study in the available database specifically examined the second stage of
postnatal testes descent in pups. Thus, there is no evidence to determine if the observed effect
represented a developmental delay or if it was related to disruptions in male reproductive tract
ontogeny, which is dependent on normal levels of fetal testicular testosterone and on the
expression of insulin-like hormone-3 (insl3) in fetal Leydig cells (Klonisch et al.. 2004). This
abnormality was not observed in any other study in the formaldehyde database; however, no single
or multigeneration reproduction studies were available, and it is with this type of protocol that
such a finding would more likely be detected. Martin (1990) did not report any structural
anomalies resulting from inhalation exposures during gestation up to exposure levels of
12.3 mg/m3.
The potential influence of maternal toxicity on developmental findings was considered in
the review of the available data. For several studies, information on maternal toxicity was not
reported fMonfared. 2012: Senichenkova and Chebotar. 1996b: Senichenkova. 1991al although for
these studies, it is not known whether (1) maternal toxicity was not assessed or (2) maternal
toxicity was assessed, but results were not reported. Kum et al. (2007) measured maternal body
and liver weight but found no treatment-related effects. In Kitaev et al. (1984). increased
luteinizing hormone (LH) or follicle-stimulating hormone (FSH) levels were observed in dams at 0.5
and 1.5 mg/m3, with compromised preimplantation survival noted at the highest exposure level.
Although the maternal hormonal alterations could have been related to the embryo loss, there was
no confirmation in other studies. Gofmekler et al. (1968) noted increased gestation duration at
This document is a draft for review purposes only and does not constitute Agency policy.
1-406 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
0.012 and 1 mg/m3, with corollary evidence of increased newborn body and organ weights at those
exposure levels. Sheveleva (1971) reported evidence suggesting maternal toxicity at 5 mg/m3,
including a decreased threshold of neuromuscular excitability, increased rectal temperature, and
increased hemoglobin in dams; however, developmental toxicity (i.e., increased preimplantation
loss) was observed at both 0.5 and 5 mg/m3. Martin et al. (1990) reported significantly decreased
maternal weight gain and food consumption only at the highest exposure level (12.3 mg/m3), but
no developmental toxicity was observed in the study. In the Saillenfait et al. (1989) study,
significantly decreased maternal body-weight gain was observed only at the highest exposure level
(49.2 mg/m3); however, significantly decreased fetal weight was observed at both 24.6 and
49.2 mg/m3. Thus, in the limited developmental toxicity database available for evaluation, there
was little evidence that maternal toxicity was a major contributing factor to observations of
developmental toxicity.
Overall, the database for the evaluation of developmental toxicity (survival, growth, and
morphological alterations) consisted of weak (low confidence) studies that had methodological
limitations, primarily lack of information about the test substance or the described use of formalin,
with known or presumed methanol coexposures. Effects on fetal survival, pre- or postnatal growth,
or morphological alterations were observed in several studies and sometimes more than one
rodent species, and maternal toxicity did not appear to be a confounding influence. However,
inconsistencies in response were also observed, and clear dose-response relationships were not
discernable. Additional experiments using stronger study designs are needed to more thoroughly
assess the effect of formaldehyde exposure on development.
This document is a draft for review purposes only and does not constitute Agency policy.
1-407 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-55. Summary of developmental effects observed in animal studies
following inhalation exposure to formaldehyde
Reference and study design3
Results'3 and exposure levels (mg/mB)
Low confidence (all animal studies of developmental toxicity)
Fetal survival
Reference: Kitaev et al. (1984)
Rats (Wistar), 200 females total
4 hr/day, 5 days/wk, for 4 months
0, 0.5 or 1.5 mg/m3
Test article: Not characterized
Maternal tox: Altered LH and FSH levels in
treated dams
Main limitations: Test article NC; limited
description of methods.
Number (percentage) degenerated
embryos GD 2 (n = 5-8)
Number (percentage) degenerated
embryos GD 3 (n = 5-9)
0 0.5 1.5
2(5.1) 3(3.8) 5(10.2)
3(4.4) 4(9.1) 10(14.9)
Reference: Martin (1990)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-15
0, 2.46, 6.15, 12.3 mg/m3
Test article: Paraformaldehyde
Maternal tox: Significantly decreased
maternal body-weight gain and food
consumption at 12.3 mg/m3
Main limitations: Inadequate reporting of
methods and quantitative results.
Report states that there was no evidence of decreased fetal survival; no data were
presented.
Reference: Saillenfait et al. (1989)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-20
0, 6.15, 12.3, 24.6, or 49.2 mg/m3
Test article: Formalin
Maternal tox: Significantly decreased
maternal body-weight gain at 49.2 mg/m3
Main limitation: Formalin.
Mean total fetal loss/litterc
0 6^15 123 24^6 49^2
-33 0 0 0%
Reference: Senichenkova (1991b)
Rats (white mongrel), 137 dams total, =46
dams/group
4 hr/day, GD 1-19 (C-section GD 20)
0 or 0.5 mg/m3
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article NC; exposure
generation, animal strain/source, #
dams/group, maternal tox NR; limited
description of methods.
Number (percentage)
preimplantation loss
Number (percentage)
postimplantation loss
Mean preimplantation loss
Mean postimplantation loss
0 0.5
38/381 (10.0) 25/304 (8.2)
26/343 (7.6) 12/279 (7.3)
-3%
-15%
Reference: Senichenkova and Chebotar
(1996a)
Rats (mongrel, strain not reported),
29/group
4 hr/day, GD 1-19 (C-section GD 20)
0 or 0.5 mg/m3
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article, exposure
generation, animal strain/source, #
dams/group, maternal tox NR; limited
description of methods.
Mean postimplantation lossc
0 05
29%
This document is a draft for review purposes only and does not constitute Agency policy.
1-408 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/mB)
Reference: Sheveleva (1971)
Rats (mongrel, strain not reported),
15/group terminated GD 20, 6/group littered
4 hr/day, GD 1-19
0, 0.5, or 5 mg/m3
Test article: Not characterized
Maternal tox: Decreased threshold of
neuromuscular excitability, rectal
temperature, and hemoglobin in dams at 5
mg/m3
Main limitations: Test article NC; exposure
generation, animal strain/source NR; limited
description of methods.
Mean preimplantation lossc
Mean postimplantation lossc
0 0,5 5
50 70%
0 0%
Fetal and postnatal growth
Reference: Gofmekler et al. (1968)
Rats (strain not specified), 12 females/group
Continuous exposure 10-15 days prior to
mating and throughout gestation
0, 0.012, or 1 mg/m3
Test article: Not characterized
Maternal tox: Increased duration of
gestation at both dose levels
Main limitations: Test article NC, exposure
generation, animal strain/source NR; limited
description of methods; limited reporting.
Mean newborn weight (g)
Mean relative neonatal lung
weight (mg/10 g BW)
Mean relative neonatal liver
weight (mg/10 g BW)
0 0.012 1
7* 13%*
-20* -19%*
-5* -6%*
Reference: Kum et al. (2007)
Rats (Sprague Dawley), 6/group
8 hr/day, 7 days/wk, for 6 weeks
starting at GD 1, PND 1, Wk-4, or Adult
0 or 7.38 mg/m3
Test article: Formalin
Maternal tox: Not reported
Main limitations: Formalin; limited
description of methods; maternal tox NR.
Decreased pup weight (g) (3-wk
old pups that were exposed in
utero and during lactation)
Decreased pup weight (g) (6-wk
old pups that were exposed during
lactation and for 3 weeks
postweaning)
Decreased young adult weight (g)
(10-wk old young adults that were
exposed starting at 4-weeks of
age)
Mature adult weight (g) (6 weeks
of exposure to adult rats)
0 7.38
-31%*
-14%*
-9%
7%
Reference: Martin (1990)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-15
0, 2.46, 6.15, 12.3 mg/m3
Test article: Paraformaldehyde
Maternal tox: Significantly decreased
maternal body-weight gain and food
consumption at 12.3 mg/m3
Main limitations: Inadequate reporting of
methods and quantitative results.
Report states that fetal weights were not affected by treatment; no data were
presented.
Reference: Monfared (2012)
Mice (Balb/C), 10/group
8 hr/day, GD 6-16 (C-section GD 17)
0, 5.68, 11.38, or 22.76 mg/m3
Mean fetal weight (g)
Mean placental weight (g)
0 5.68 11.38 22.76
-24* -27* -32%*
35* 57* 39%*
This document is a draft for review purposes only and does not constitute Agency policy.
1-409 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/mB)
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article NC; maternal
tox: NR.
Thickness of placental
trophoblastic basement
membrane (nm)
Thickness of placental labyrinth
interhemal membrane (|am)
148* 177* 203%*
45* 42* 49%*
Reference: Saillenfait et al. (1989)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-20
0, 6.15, 12.3, 24.6, or 49.2 mg/m3
Test article: Formalin
Maternal tox: Significantly decreased
maternal body-weight gain at 49.2 mg/m3
Main limitation: Formalin.
Mean fetal body weight/litter-
males
Mean fetal body weight/litter-
females
0 6J5 123 24J> 49^
-1 -2 -5* 21%*
10-3 19%*
Reference: Senichenkova (1991b)
Rats (white mongrel), 137 dams total, =46
dams/group
4 hr/day, GD 1-19 (C-section GD 20)
0 or 0.5 mg/m3
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article NC; exposure
generation, animal strain/source, #
dams/group, maternal tox NR; limited
description of methods.
Mean fetal body weight (g)
Mean fetal length (mm)
Mean day of upper incisor
eruption
Mean day of lower incisor
eruption
0 OJj
5%*
0%
17%*
25%*
Reference: Sheveleva (1971)
Rats (mongrel, strain not reported),
15/group terminated GD 20, 6/group littered
4 hr/day, GD 1-19
0, 0.5, or 5 mg/m3
Test article: Not characterized
Maternal tox: Decreased threshold of
neuromuscular excitability, rectal
temperature, and hemoglobin in dams at
5 mg/m3
Main limitations: Test article NC; exposure
generation, animal strain/source NR; limited
description of methods.
Mean fetal weight (g)
Mean fetal length (mm)
0 0.5 5
0 3%
0 0%
Fetal morphological development
Reference: Martin (1990)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-15
0, 2.46, 6.15, 12.3 mg/m3
Test article: Paraformaldehyde
Maternal tox: Significantly decreased
maternal body-weight gain and food
consumption at 12.3 mg/m3
Main limitations: Inadequate reporting of
methods and quantitative results.
Fetal incidences of major malformations,
minor external and visceral anomalies,
and minor skeletal anomalies.
Report states that fetal incidences
were not affected by treatment; no
data presented.
Reference: Saillenfait et al. (1989)
Rats (Sprague Dawley), 25/group
6 hr/day, GD 6-20
0, 6.15, 12.3, 24.6, or 49.2 mg/m3
Test article: Formalin
Maternal tox: Significantly decreased
maternal body-weight gain at 49.2 mg/m3
Unossified sternebrae
[fetal(litter) incidence]
Unossified sternebrae
[fetal percentage]
Unossified sternebrae
[litter percentage]
0 6J5 123 24^6 49^2
3(3) 1(1) 6(3) 6(3) 15(7)
0.9 0.4 1.9 2 4.4%
This document is a draft for review purposes only and does not constitute Agency policy.
1-410 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/mB)
Main limitation: Formalin.
12.5 4.8 13 14.3 29.2%
Reference: Senichenkova (1991b)
Rats (white mongrel), 137 dams total, =46
dams/group
4 hr/day, GD 1-19 (C-section GD 20)
0 or 0.5 mg/m3
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article NC; exposure
generation, animal strain/source, #
dams/group, maternal tox NR; limited
description of methods.
Mean percentage fetuses with
cryptorchidism
Number of litters with internal organ
anomalies
Mean number of litters with internal
organ anomalies
Number (percentage) embryos with
ossification centers in the hyoid bone
Mean number of metacarpal bone
centers
Mean number of metatarsal bone
centers
0 0.5
20%*
2 8%
914%*
145(100) 61(91)*
13%*
9%*
Reference: Senichenkova and Chebotar
(1996a)
Rats (mongrel, strain not reported),
29/group
4 hr/day, GD 1-19 (C-section GD 20)
0 or 0.5 mg/m3
Test article: Not characterized
Maternal tox: Not reported
Main limitations: Test article NC; exposure
generation, animal strain/source, #
dams/group, maternal tox NR; limited
description of methods.
Mean percentage litters with
hydronephrosis
Mean percentage litters with
cryptorchidism
0 05
5%
21%
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: GD = gestational day; LH = luteinizing hormone; FSH = follicle-stimulating hormone; NC = not
characterized; NR = not reported.
aStudies with gestational or lactational exposures and evaluation of pre- or postnatal developmental outcomes are
included in this table.
bResponse relative to control for mean data, or incidence data,
incidence data not reported.
^Statistically significant difference from control value, as reported by the study author.
Study exposure levels converted from ppm to mg/m3 are presented in italics (1 ppm = 1.23 mg/m3).
1 Female reproductive toxicity
2 Information on female reproductive toxicity in the formaldehyde database is minimal (see
3 Table 1-56; Figure 1-33). For the three low confidence studies that noted effects on the female
4 reproductive system, the test substance was either not characterized fWang et al.: Kitaev et al..
5 1984) or was reported to be formalin (Maronpotetal.. 1986).
This document is a draft for review purposes only and does not constitute Agency policy.
1-411 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
O Non-significant • Shading: statistically significant (black) or > 10% change (gray) from control groups
E
"5
E
c
o
'¦p
0)
v
c
o
u
-C
a)
2
ro
£
ZD
3
10
o.i
Design:
Species:
Endpoint:
Confidence:
<-
4 mos.
N=5-9
4 mos.
N=5-9
Rats
Hormone Levels
60 d
n=10
4 mos.
N=5-9
60 d
n=10
Rats
4, Ovary wt.
60 d
n=10
Rats
13 wk
n=10
Mice
Ovarian or Uterine Pathology
1
2
3
4
5
6
Figure 1-33. Animal studies evaluating female reproductive toxicity.
As no high or medium confidence experimental animal studies were identified (see Appendix A.5.8), the
available studies are organized by endpoint, species, and then by duration of exposure. Shading indicates
statistically significant (black) or >10% change (gray) from controls, and the size of the points reflects the
sample size for that exposure group (larger size = larger n). The low confidence experiments are shown
on a gray background, as the identified study limitations substantially reduce confidence in the reliability
of the results; these low confidence experiments contribute very little to the weight of evidence for
female reproductive toxicity.
Uterine and ovarian hypoplasia was observed by Maronpot et al. (1986) in 100% of the
mice on study at 49.2 mg/m3 following 13 weeks of inhalation exposure; the incidence of these
findings was zero at the next lower exposure level of 24.6 mg/m3. Histopathological evaluation
conducted by Wang et al. (20131 did not confirm these findings, but identified a significant decrease
in the number and size of mature ovarian follicles with a concomitant increase in the number of
atretic follicles, and disruptions in structural integrity of the ovary in rats after 8 weeks of
This document is a draft for review purposes only and does not constitute Agency policy.
1-412 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
formaldehyde exposure. Kitaev et al. (19841 reported a 56% increase in relative ovarian weight,
accompanied by increased blood LH and FSH levels (11 and 36%, respectively) and significantly
increased ovulation (not shown in evidence table), at the lowest dose tested (0.5 mg/m3) in rats
following 4 months of inhalation exposure; these findings are suggestive of a treatment-related
disruption of the hypothalamic-pituitary-ovarian (HPO) axis. At the highest dose tested in the same
study (1.5 mg/m3), ovarian weights and LH levels were decreased by 33 and 17%, respectively, as
compared to control, and FSH levels were statistically significantly increased (191%); these
findings might represent evidence of direct ovarian toxicity and the consequences of disturbed
early embryo development in addition to effects on the HPO axis. However, a lack of information
about sample collection and analytical methods render it difficult to interpret these data with
confidence. The nonmonotonic effect on ovarian weight observed by Kitaev et al. (1984) was not
corroborated by Wang et al. (2013.). The hormonal alterations observed by Kitaev et al. (1984)
could have been related to increased preimplantation loss observed in that study or indicative of an
adverse effect on female reproductive system integrity. Other evidence of hormonal disruption,
such as 12% decreased estradiol (E2) levels observed by Wang et al. (2013). might have been
related to the ovarian histopathology observed in that study.
Overall, as only low confidence animal studies of female reproductive toxicity were
available, this points to the need for further evaluation of the female reproductive system following
formaldehyde inhalation exposure, including an assessment of overall female reproductive
function.
Table 1-56. Summary of female reproductive effects observed in animal
studies following inhalation exposure to formaldehyde
Reference and study design3
Results'3 and exposure levels (mg/m3)
Low confidence (all animal studies of female reproductive toxicity)
Reference: Kitaev et al. (1984)
0
0.5
2.46
Rats (Wistar), 200 females total
4 hr/day, 5 days/wk, for 4 months
Mean relative ovary
weightc
0
56
-33
0, 0.5 or 1.5 mg/m3
Test article: Not characterized
Mean blood LH (mg/mL)c
0
11
-17
Main limitations: Test article NC; limited
Mean blood FSH (mg/mL)c
0
36
191*
description of methods.
Number (percentage)
degenerated embryos GD
2 (n = 5-7)
2(5.1)
3(3.8)
5(10.2)
Number (percentage)
degenerated embryos GD
3 (n = 5-9)
3 (4.4)
4(9.1)
10 (14.9)
*p<0.05
This document is a draft for review purposes only and does not constitute Agency policy.
1-413 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Reference: Maronpot et al. (1986)
Mice (B6C3F1), 10/sex/group
6 hr/day, 5 days/wk, for 13 weeks
0, 2.46, 4.92,12.3, 24.6 or 49.2 mg/m3
Test article: formalin
Main limitations: Formalin; limited
reporting of methods and results.
Ovarian hypoplasia
Uterine hypoplasia
0 2A6 432 123 24J> 49,2
0/10 NE NE NE 0/10 10/10
0/10 NE NE NE 0/9 9/9
Reference: Wang et al. (2013)
Rats (SD), 10 females/group
8 hr/day, 7 days/wk, for 60 days
0, 0.5, 2.46 mg/m3
Test article: Not characterized
Main limitations: Test article NC.
Mean serum E2 (ng/L)c
Mean ovarian weight (g)c
0 05 2;46
0 -2 -12
0 -2 -8
Ovarian histopathological findings at 2.46 mg/m3d:
Number and size of mature follicles significantly decreased
Number of atretic follicles increased
Vascular congestion, interstitial edema, structure disorder
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: NE = not evaluated.
aStudies that evaluated female reproductive system toxicity are included in this table. Studies are organized by
endpoint, species, and lowest dose tested.
bResponse relative to control for mean data, or incidence data.
cData digitized using Grab It!™, Datatrend Software,
incidence data not reported.
Study exposure levels converted from ppm to mg/m3 are presented in italics (1 ppm = 1.23 mg/m3).
Male reproductive toxicity
Fourteen studies in rodents assessed effects on the male reproductive system following
inhalation formaldehyde exposure (see Table 1-57; Figure 1-34); although eight of the studies had
substantial methodological limitations, 13 of the 14 studies demonstrated treatment-related effects.
Of the available studies, only those by Vosoughi et al. (2013; 20121 (both of which reported data
from the same cohort of mice; see footnote in Table 1-57), Ozen et al. (2005: 2002). Appelman et al.
(1988). Sapmaz et al. (2018). and Sarsilmaz et al. (1999) administered paraformaldehyde to the
test animals and provided adequate characterization of the exposure paradigm. The results of
these paraformaldehyde studies are interpreted with high fVosoughi etal.. 2013: Vosoughi et al..
2012: Ozen etal.. 2005: Ozen etal.. 20021 and medium fSapmaz etal.. 2018: Sarsilmaz etal.. 19991
confidence; however, the results of the remaining studies in this section are considered much less
reliable (i.e., low confidence), based in part upon deficient exposure criteria. Evaluations of male
reproductive toxicity in the more reliable (e.g., medium and high confidence) studies are
constrained by a complete lack of testing at lower formaldehyde concentrations. Specifically, one
medium confidence study (Sapmaz etal.. 2018) tested a single concentration of 6.15 mg/m3 and one
medium confidence study fOzen etal.. 20051 tested concentrations >6 mg/m3, while the remainder
of the medium fSarsilmaz etal.. 19991 and high fVosoughi etal.. 2013: Vosoughi etal.. 2012: Ozen et
This document is a draft for review purposes only and does not constitute Agency policy.
1-414 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
al.. 20021 confidence studies only examined concentrations >12 mg/m3. These high levels of
formaldehyde could introduce additional complications to interpretation, including potential reflex
bradypnea. In this regard, Ozen et al. (2005), the only well-conducted study testing formaldehyde
levels <12 mg/m3, and Sarsilmaz etal. (1999) noted clinical signs of respiratory irritation or altered
breathing rate, while Ozen et al. (2002) and Vosoughi et al. f2013: 20121 did not report such
observations. Sapmaz et al. (2018) did not report observations consistent with reflex bradypnea at
6.15 mg/m3.
The evidence table is organized by outcomes of male reproductive toxicity, in order of the
strength of the evidence: histopathology, sperm measures, gonadotropic hormone measures, organ
weights, and reproductive function. Within each category, the studies are organized by high to low
confidence, and then alphabetically within a confidence category. The available animal studies of
male reproductive toxicity are illustrated in Figures 1-34 and 1-35, with Figure 1-34 presenting all
of the studies and Figure 1-35 presenting in greater detail the studies interpreted with medium or
high confidence.
This document is a draft for review purposes only and does not constitute Agency policy.
1-415 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
A.
O Non-significant
• Statistically significant
Design:
Q
o
Did
n=G
10 d
n=12
Species: Rats Mice
Endpoint: Pathology
Confidence:
10 d
n=12
Mice
Sperm
91 d
n=6
10 d
n=12
Rats Mice
-i- Serum T
13 wk
n=7
10 d
n=12
Rats
..1-
Organ
wt.
4 wk
n=10
4 wk
n=5-7
Rats
Pathology
13 wk
n=5-7
4 wk
n=10
4- Org.
wt.
B.
O Non-significant • Shaded symbols: Statistically significant (black) or severe pathology noted without quantification (gray)
.
T
T
1
0
#
. . . <
—• •—
* •
I
:
—
«
>
.
<
9
9
•
-
v v-/
2
-c
a>
2
E
0.1
Species:
Endpoint:
Figure 1-34. Animal studies evaluating male reproductive toxicity.
This document is a draft for review purposes only and does not constitute Agency policy.
1-416 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
The available studies are organized into high or medium confidence (panel A) and low confidence (panel
B) study evaluation interpretations (see Appendix A.5.8), then by endpoint, and then by species. Shaded
symbols indicate statistically significant effects (unless otherwise noted), as reported by the study
authors, and the size of the points reflects the sample size for that exposure group (larger size = larger n).
The low confidence experiments (panel B) are shown on a gray background, as the identified study
limitations substantially reduce confidence in the reliability of the results; these low confidence
experiments contribute very little to the weight of evidence for male reproductive toxicity.
mg/m3: 12.2 24.4 12.3 24.6 12.3 24.6
I T
1.0--
-i
6.2 12.3
12.3 24.6
6.2 12.3 12.3 24.6 6.2 6.2 12.3 24.6
4wk 13wk
12.3
—I—
24.6
0.5
+l
CD
D)
C
nj
.C
o
5
o
-5-
*
-5-
*
13wk
rats,n= 7
high
Ozen
(2002)
4wk
rats,n=10
medium
Sarsilmaz
(1999)
relative weight
10d
mice,n=12
high
Vosoughi
(2013)
absolute
13wk
rats,n=6
high
Ozen
(2005)
10d
mice,n=12
high
Vosoughi
(2013) 3
13wk
rats,n=6
high
10d
mice,n=12
high
*
4 or 13wk
rats,n=5-7
medium
4wk
rats,n=10
medium
Sapmaz Sarsilmaz
>>
Ozen Vosoughi
(2005) (2013) (2018)b (1999)
diameter thickness Leydig
cell#
seminiferous tubule measures
f 3r
10d
mice,n=12
high
Vosoughi
(2013)
Testis Weight
Serum T
Histopathology
Sperm count
Figure 1-35. Medium and high confidence animal studies evaluating male
reproductive toxicity.
The available high and medium confidence studies are arrayed and organized by endpoint. 1Results are
displayed as fold change from control animals (control responses at 1 are illustrated as a dashed line),
with variability in both the controls and treatment groups represented by the quotient (ratio) of the 95%
confidence intervals (CI), as calculated based on the method originally described by E.C. Fieller (Cox and
Ruhl, 1966), which assumes Gaussian distributions. aThe serum T measure at 24 hr is presented from
Vosoughi et al. (2013). bSeminiferous tubule diameter was not significantly affected by formaldehyde
exposure (p > 0.05) in Sapmaz et al. (2018), although in addition to the reduced thickness shown above,
the authors also reported a significantly reduced percentage of intact tubules at both formaldehyde
exposure timepoints (i.e., 71.1% in controls; 42.2% with 6.2mg/m3 at 4 weeks; and 17.2% with 6.2 mg/m3
at 13 weeks). Notes: * = author-reported statistical significance (p < 0.05). Vosoughi et al. (2013) reflects
results from both the 2012 and 2013 studies (2013; 2012), which report data from the same cohort of
mice; Ozen et al. (2005; 2002) and Sarsilmaz et al. (1999) are studies from the same research group.
1 Testes and epididymides histopathology
2 Quantitative and qualitative histopathological findings in the testes of adult male rodents
3 following from 10 days to 18 weeks of inhalation exposure were reported in two high confidence
4 studies fVosoughi etal.. 2013: Vosoughi etal.. 2012: Ozen etal.. 20051 and two medium confidence
5 studies fSarsilmaz etal.. 19991 that used paraformaldehyde, and in five low confidence studies that
6 used formalin (Han etal.. 2013: Zhou etal.. 2011a: Zhou etal.. 2011b: Golalipour etal.. 2007: Zhou
This document is a draft for review purposes only and does not constitute Agency policy.
1-417 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
et al.. 20061. Alterations in germ cell number and integrity, statistically significant reductions in
germinal epithelium thickness or seminiferous tubule diameter (5-30%), tubular atrophy, markers
of disrupted spermatogenic process, and Leydig cell damage were observed. Epididymal findings
(e.g., decreased tubule diameters or atrophy, epithelial alterations, or absence of sperm) in Zhou et
al. (2011b) also indicated a disruption of spermatogenesis. One low confidence study in mice
treated for 13 weeks (Maronpotetal.. 1986) did not report any lesions of the male reproductive
tract. Notably, while this study used formalin as the test article, this limitation would be expected
to bias the study toward observing an effect; thus, there is no credible rationale for this negative
outcome. However, evidence of treatment-related testicular pathology in the high confidence
mouse study by Vosoughi et al. (2013.; 20121 suggests that the absence of effects in Maronpot et al.
(1986) is probably not attributable to a difference in species response, although any potential
influence of animal strain on response is unknown.
Sperm measures
A significantly decreased sperm count of 44-49% was observed at 35 days posttreatment in
a study of mice exposed to >12.2 mg/m3 paraformaldehyde for 10 days fVosoughi etal.. 2013:
Vosoughi et al.. 20121. In rats, 10 mg/m3 formalin exposure significantly decreased sperm count by
38% with a 2-week exposure fZhou etal.. 201 lal and 77% with a 4-week exposure fZhou etal..
20Hb), demonstrating an increase in the magnitude of the response as the duration of exposure
increased, with the exposure concentration level remaining constant Zhou et al. (2011a) reported
a significant 13% reduction in sperm count at 2.46 mg/m3 after 60 days of formalin exposure,
consistent with the interrelationship among concentration, exposure duration, and magnitude of
response. These data provide evidence of the downstream effects of disruptions to
spermatogenesis that are observed histopathologically.
In the same studies, sperm motility was significantly decreased (by 40-46%) in mice
(Vosoughi etal.. 2013: Vosoughi etal.. 2012) and by 13-17% in rats (Zhou etal.. 2011a: Zhou etal..
2011b) at exposure levels >10 mg/m3 paraformaldehyde or formalin, respectively, and significant
abnormal sperm morphology was observed at the same exposure levels (Vosoughi etal.. 2013:
Vosoughi etal.. 2012: Zhou etal.. 2006). Statistically significant increases in abnormal sperm were
also observed by Xing et al. (2007b) after 4 weeks of formalin exposure at exposure levels
>20 mg/m3. The alterations in sperm count, motility, and morphology reported by Vosoughi et al.
(2013; 20121 achieved statistical significance at 35 days (but not at 24 hours) postexposure,
demonstrating a biologically plausible temporal delay in the outcomes associated with disruption of
spermatogenesis. Altered sperm measures are considered biomarkers of reduced fertility;
however, with the exception of the high exposure study by Xing et al. (2007) that identified a male-
mediated reduction in viable conceptuses, the formaldehyde database does not include any studies
that specifically assessed fertility measures.
This document is a draft for review purposes only and does not constitute Agency policy.
1-418 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Hormone measures
Two high confidence studies that exposed rodents to paraformaldehyde fVosoughi etal..
2013: Vosoughi etal.. 2012: Ozen etal.. 20051 found significant decreases in serum testosterone
(T). Vosoughi et al. f2013: 20121 exposed mice to paraformaldehyde for 10 consecutive days and
reported 32-49% decreases at 24 hours post-exposure and 10-15% decreases at 35 days
postexposure. While this might suggest postexposure recovery or a compensatory process, there
are no other studies that tested this possibility. Ozen et al. (20051 noted significant 6-9%
decreases in serum T after exposing rats for 91 days to paraformaldehyde. Zhou et al. (2011a), a
low confidence formalin study in rats, demonstrated nonsignificant decreases (up to 6%) in serum
T after 60 days of exposure. The decreased serum testosterone levels observed by Ozen et al.
(2005), Vosoughi et al. f2013: 20121. and Zhou et al. f2011al are biologically consistent with the
Leydig cell pathology observed by Vosoughi et al. (2013: 2012) and Sarsilmaz et al. (1999) because
Leydig cells are the primary source of testosterone production in the testes. No other studies
evaluated alterations in serum T levels following formaldehyde exposure.
Vosoughi et al. (2013: 2012) also reported a significant 15% decrease in serum LH at
24 hours postexposure but not at 35 days postexposure. In the same study, FSH levels were not
affected at the 24-hour and 35-day assessment times.
Testes and epididymides weights
A treatment-related effect on testes weight is suggested by the available data. However,
even though a number of studies examined testes and epididymides weights, the findings were
neither consistent nor easily interpretable. Statistically significant decreased mean testes or
epididymal weight of >20% magnitude was reported in three low confidence rat studies with
inhalation exposures to 5-10 mg/m3 formalin for 2 or 4 weeks duration fHan etal.. 2013: Zhou et
al.. 2011b: Zhou etal.. 20061. Conversely, testis or epididymal weights were not decreased in two
studies: one high confidence study that exposed mice to paraformaldehyde for 10 days at up to
24.4 mg/m3 (Vosoughi et al.. 2 013: Vosoughi etal.. 2012) and one low confidence study that
exposed rats for 60 days to 2.46 mg/m3 formalin (Zhou etal.. 2011a). It is possible that these two
studies did not detect effects on testes weight due to either the short exposure duration or the low-
exposure level used, respectively.
Slight decreases in relative (to body weight) testes weight data in rats resulting from 12.2
or 24.4 mg/m3 paraformaldehyde exposures were reported by Ozen et al. (2002) and Sarsilmaz et
al. (1999). high and medium confidence studies in rats, respectively. Findings at 4 weeks of
exposure in each study were similar, with <3% decreases in relative testes weights (although
statistical significance was reported by Ozen et al. (2002). Notably, following 13 weeks of exposure,
Ozen et al. (2002.) reported significant relative testes weight decreases compared to control of up to
10%, suggesting that there was a duration-related component to the response. A significant
increase in mean relative (to body weight) testes weight following 53 weeks of paraformaldehyde
This document is a draft for review purposes only and does not constitute Agency policy.
1-419 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
exposure was reported for a low confidence study by Appelman et al. (19881: however, no
quantitative data were presented in the study report Appelman et al. (1988) attributed the relative
testes weight increase to decreased body weights. Due to the absence of data on body weight, the
veracity of this interpretation could not be assessed. The use of relative testes weights is typically
not preferred for assessment of reproductive toxicity because testes weight has been shown to be
generally conserved across 5-30% decreases in body weight (OECD. 2013). Insufficient
information (on either the mean testes or body weights used in deriving the relative weight values)
was provided in Ozen et al. (2002). Sarsilmaz et al. (1999). and Appelman et al. (1988) to fully
evaluate the magnitude of the absolute testes weight effects.
Overall, the database for the evaluation of male reproductive toxicity (histopathology,
sperm measures, gonadotropic hormone measures, organ weights, and reproductive function)
included multiple high or medium confidence studies that provided coherent evidence of toxicity
spanning biochemical, cellular, tissue, and functional levels. These findings were supported by
evidence of male reproductive system toxicity in seven of eight of the remaining low confidence
studies, although the interpretability of these findings is questionable, primarily due to a lack of
information about the test substance or the described use of formalin. Specifically, effects on testes
and epididymides histopathology were observed in a high confidence study in mice fVosoughi etal..
2013: Vosoughi etal.. 20121 and another in rats fOzen etal.. 20051. a medium confidence study in
rats (Sarsilmaz etal.. 1999). and five low confidence studies in rats. The histopathological outcomes
were supported by evidence of reduced serum testosterone in the two high confidence studies,
alterations in sperm measures (count, motility, and morphology) in the high confidence study in
mice (Vosoughi etal.. 2013: Vosoughi etal.. 2012) and four other low confidence studies in rodents,
thus demonstrating downstream consequences of the testes and epididymides histopathological
lesions. Data on testes and epididymides weights provided some limited supportive information
from several low confidence studies, and from a medium and a high confidence study (Ozen et al.
(2002) and Sarsilmaz et al. (1999). respectively), although the results were difficult to interpret
Uncertainties remain due to a complete lack of high or medium confidence studies testing exposure
levels <6 mg/m3, and observations potentially consistent with the occurrence of reflex bradypnea
at >6 mg/m3 in two of the studies. However, the observed responses to high levels of formaldehyde
provided a coherent pattern of effects in well-conducted studies performed across two
international laboratories, using two rodent species, and varied durations, and, in some cases,
demonstrating clear concentration-dependent responses of exposure. None of the studies in the
database conducted an in-depth assessment of male reproductive function (e.g., including mating or
fertility) or evaluated outcomes attributable to early-life exposures (such as would be assessed in a
multigeneration reproduction study).
This document is a draft for review purposes only and does not constitute Agency policy.
1-420 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-57. Summary of male reproductive effects observed in animal studies
following inhalation exposure to formaldehyde
Reference and study design3
Results'3 and exposure levels (mg/m3)
Testes and epididymides histopathology
High confidence
Reference: Ozen et al. (2005)
Rats (Wistar), 6 males/group
8 hr/day, 5 days/wk, for 91 days
0, 6.15, or 12.3 mg/m3
Test article: Paraformaldehyde
Mean seminiferous tubule diameters
(|am) (n = 100 randomly selected
tubules/group)
0 6J5 123
-23* -26%*
Reference: Vosoughi et al. (2013); Vosoughi
et al. (2012)c
Mice (NMRI), 12 males/group
8 hr/day, 10 days
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Histopathological findings in treated males at 35 days postexposure11
Testes: seminiferous tubule atrophy
Testes: increased space between germ cells
Testes: degeneration of Leydig cells
Testes: disintegration of seminiferous epithelial cells
Testes: degeneration of a number of seminiferous tubules
Histopathological measurements:
Mean seminiferous tubule diameter
(|am)-24 hr postexposure
Mean seminiferous tubule diameter
(|am)-35 days postexposure
0 12J 24J
-6 -7%*
-11* -13%*
Medium confidence
Reference: Sapmaz et al. (2018)
Rats (Sprague-Dawley), 7 males/group
8 hr/day, 5 days/wk, for 4 or 13 weeks
0 or 6.15 mg/m3
Test article: Paraformaldehyde
Main limitations: Lack of detailed reporting on
quantitative analyses of histopathology.
Histopathological assessments:
Mean germinal epithelial thickness
Mean seminiferous tubule diameter
Percent intact tubules
6.15 6.15
0 (4wk) (13wk)
-33.7%* -62%*
-5.2% -2.2%
71.7% 42.2%* 17.2%*
Reference: Sarsilmaz et al. (1999)
Rats (Wistar), 10 males/group
8 hr/day, 5 days/wk, for 4 weeks
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Main limitations: Inadequate information for
quantitative analysis of histopathology data,
Mean Leydig cell quantity (100 sections
total)
Leydig cell nuclear damage (picnotic,
karyoretic, karyolitic) (percentage
normal)
0 123 24^6
-5* -6%*
-6 -22%
This document is a draft for review purposes only and does not constitute Agency policy.
1-421 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Low confidence
Reference: Golalipour et al. (2007)
Rats (Wistar), 7 males/group
18 weeks formaldehyde exposure
(1) 4 hr/day, 4 days/wk
(2) 2 hr/day, 4 days/wk
(3) 2 hr/day, 2 days/wk
0 or 1.85 mg/m3
Test article: Not characterized
Main limitations: Test article NC; open air
exposures; N = 4/group.
Histopathological findings in formaldehyde exposure group (3)d:
Increased spaces between germ cells in seminiferous tubules
Disrupted association between Sertoli and germinal cells
Histopathological findings in formaldehyde exposure group (2)d:
Decreased germ cells and increased thickness of basal membrane in 75% o
seminiferous tubules
Histopathological findings in formaldehyde exposure group (l)d:
Severe decrease in germ cells in >85% of seminiferous tubules
Arrested spermatogenesis
Histopathological measurements across study groups:
Control 1 and exposure paradigm (1-3)
Mean seminiferous tubule diameter (urn)
Mean seminiferous tubule height (|im)
iC} 111 (21 M
-19* -8* -5%*
-21* -16* -12%'
Reference: Zhou et al. (2011b)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 4 weeks
0, 0.5, 5, or 10 mg/m3
Test article: Not characterized
Main limitations: Test article NC; exposure
generation NR; static chamber used; limited
reporting of study results and group data.
Histopathological findings at 5 and 10 mg/m3 d
Testes: seminiferous tubule atrophy
Testes: decreased spermatogenic cells
Testes: oligospermic lumina
Histopathological measurements:
Mean seminiferous tubule diameter (|im)
0 05 5 10
-4 -28* -30%*
Reference: Zhou et al. (2006)
Rats (Sprague Dawley), 10 males/group
(1) 0 (gavage saline);
(2) 10 mg/m3,12 hr/day, 2 weeks;
(3) 10 mg/m3,12 hr/day, 2 weeks, plus 30
mg/kg-day oral vitamin E
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Histopathological findings observed in formaldehyde exposure group (2)c
Atrophy of seminiferous tubules
Decreased spermatogenic cells
Disintegrated and sloughed seminiferous epithelial cells
Edematous interstitial tissue with vascular dilation and hyperemia
Azoospermic seminiferous tubule lumina
Reference: Zhou et al. (2011a)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 60 days
0, 0.5, or 2.46 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Histopathological findings'1
Testes: seminiferous tubule atrophy
Testes: spermatogenic cells decreased
Testes: oligozoospermic lumina
Epididymis: oligozoospermic lumina
Histopathological measurements across exposure groups:
This document is a draft for review purposes only and does not constitute Agency policy.
1-422 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Mean seminiferous tubule diameter
(nm)
Mean epididymal tubular diameter
(caput, |am)
Mean epididymal tubular diameter
(cauda, [am)
0 05 2A6
-2 -7%*
-1 0%
1 -2%
Reference: Zhou et al. (2011b)
Rats (Sprague Dawley), 12 males/group
8 hr/day, 7 days/wk, for 4 weeks
0, 0.5, or 10 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Histopathological findingsd
Atrophy of epididymal tubules
Disintegration of epididymal epithelium
Disorganization and denaturalization of epididymal epithelial cells
Epididymis: hyperemia of interstitial vasculature
Epididymis: oligozoospermic lumina
Reference: Maronpot et al. (1986)
Mice (B6C3F1), 10/sex/group
6 hr/day, 5 days/wk, for 13 weeks
0, 2.46, 4.92,12.3, 24.6 or 49.2 mg/m3
Test article: Formalin
Main limitations: Formalin; limited reporting of
methods and results.
Testes histopathology
No observed effect of
treatment
Sperm measures
High confidence
Reference: Vosoughi et al. (2013): Vosoughi
et al. (2012)c
Mice (NMRI), 12 males/group
8 hr/day, 10 days
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Postexposure assessments. 24 hr:
Mean epididymal sperm count (106/mL)
Mean progressive motility (%)
Mean immotile sperm (%)
Sperm viability (%)
o 12^ 2M
-18 -22%
-7 -18%
33 56%*
-8 -14%*
Mean normal morphology (%)
-7
-7%
Postexposure assessments. 35 davs:
Mean sperm count (106/mL)
Mean progressive motility (%)
Mean immotile sperm (%)
Sperm viability (%)
Mean normal morphology (%)
-44*
-40*
129*
-26*
-13*
-49%*
-46%*
170%*
-34%*
-16%*
This document is a draft for review purposes only and does not constitute Agency policy.
1-423 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Low confidence
Reference: Xing Sv (2007)
Mice (unspecified strain), 7 males/group
2 hr/day, 6 days/wk, for 4 weeks
0, 20.79, 41.57, or 83.15 mg/m3
Test article: Not characterized
Main limitations: Test article NC; exposure
generation, strain NR; high exposure levels.
Percentage abnormal sperm
0 208 41^6 83^2
6.5 9.5* 14.3* 16.2
Reference: Zhou et al. (2011a)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 60 days
0, 0.5, or 2.46 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean epididymal sperm count (x 106)
Mean percentage motile sperm
Mean percentage abnormal sperm
0 05 2A6
-2 -13%*
-3 -4%
1 4%*
Reference: Zhou et al. (2011b)
Rats (Sprague Dawley), 12 males/group
8 hr/day, 7 days/wk, for 4 weeks
0, 0.5, or 10 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean epididymal sperm count (x 106)6
Mean percentage motile sperm0
0 0.5 10
3 -77%*
-1 -14%*
Reference: Zhou et al. (2006)
Rats (Sprague Dawley), 10 males/group
(1) 0 (gavage saline);
(2) 10 mg/m3,12 hr/day, 2 weeks;
(3) 10 mg/m3,12 hr/day, 2 weeks, plus 30
mg/kg-day oral vitamin E
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean epididymal sperm count (107/g
epididymal wt)
Mean percentage motile sperm
Mean percentage abnormal sperm
111 121 131
-38* -16%
-17* -11%
13* 6%
Hormone measures
High confidence
Reference: Ozen et al. (2005)
Rats (Wistar), 6 males/group
8 hr/day, 5 days/wk, for 91 days
0, 6.15, or 12.3 mg/m3
Test article: Paraformaldehyde
Mean (terminal) serum T (nmol/L)
(n = 6)
0 6^15 123
-6* -9%*
Reference: Vosoughi et al. (2013): Vosoughi
et al. (2012)c
Mice (NMRI), 12 males/group
8 hr/day, 10 days
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Postexposure assessments:
Mean serum T (ng/mL), 24 hr
Mean serum T (ng/mL), 35 days
Mean serum LH (ng/mL), 24 hr
Mean serum LH (ng/mL), 35 days
Mean serum FSH (ng/mL), 24 hr
Mean serum FSH (ng/mL), 35 days
0 12^ 2AA
-32* -49%*
-10* -15%*
-15%*
-5%
-5%
-5%
This document is a draft for review purposes only and does not constitute Agency policy.
1-424 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Low confidence
Reference: Zhou et al. (2011a)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 60 days
0, 0.5, or 2.46 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean (terminal) serum T (nmol/L)0
0 05 2A6
-1 -6%
Testes and epididymides weights
High confidence
Reference: Ozen et al. (2002)
Rats (Wistar), 7 males/group
8 hr/day, 5 days/wk, for 4 weeks or 13 weeks
0,12.2, or 24.4 mg/m3
Test article: Paraformaldehyde
Mean relative testes weight (4 wks)
(n = 7)
Mean relative testes weight (13 wks)
(n = 7)
0 12^2 2AA
-2* -3%*
-8* -10%*
Reference: Vosoughi et al. (2013): Vosoughi
et al. (2012)c
Mice (NMRI), 12 males/group
8 hr/day, 10 days
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Postexposure assessments:
Mean testes weight (mg), 24 hre
Mean testes weight (mg), 35 days0
0 12^2 2AA
2 7%
-1 0%
Medium Confidence
Reference: Sarsilmaz et al. (1999)
Rats (Wistar), 10 males/group
8 hr/day, 5 days/wk, for 4 weeks
0,12.3, or 24.6 mg/m3
Test article: Paraformaldehyde
Main limitations: Inadequate information for
quantitative analysis of histopathology data.
Mean relative testes weight
0 12^ 2AA
-1 -4%
Low confidence
Reference: Appelman et al. (1988)
Rats (Wistar), 40 males/group
6 hr/day, 5 days/wk, for 13 or 52 weeks
0, 0.123, or 12.3 mg/m3
Test article: Paraformaldehyde
Main limitations: No indication if histopathology
performed on male reproductive organs;
quantitative testes weights not presented.
Mean relative testes weight, 53 wks
Significant increase at 10 ppm
(12.3 mg/m3) reported (no
data were presented); effect
was attributed by study
author to decreased body
weight.
Reference: Zhou et al. (2011b)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 4 weeks
0, 0.5, 5, or 10 mg/m3
Test article: Not characterized
Main limitations: Test article NC; exposure
generation NR; static chamber used; limited
reporting of study results and group data.
Mean testes weight (g)e
0 0.5 5 10
-3 -24* -21%*
This document is a draft for review purposes only and does not constitute Agency policy.
1-425 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design3
Results'3 and exposure levels (mg/m3)
Reference: Zhou et al. (2006)
Rats (Sprague Dawley), 10 males/group
(1) 0 (gavage saline);
(2) 10 mg/m3,12 hr/day, 2 weeks;
(3) 10 mg/m3,12 hr/day, 2 weeks, plus 30
mg/kg-day oral vitamin E
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean testes weight (g)e
111 121 131
-22* -3%
Reference: Zhou et al. (2011a)
Rats (Sprague Dawley), 10 males/group
8 hr/day, 7 days/wk, for 60 days
0, 0.5, or 2.46 mg/m3
Test article: Not characterized
Main limitations: Test article NC, exposure
generation NR; static chamber used.
Mean testes weight (g)
Mean epididymis weight (g)
0 05 2A6
-1 -3%
4 -2%
Reference: Zhou et al. (2011b)
Rats (Sprague Dawley), 12 males/group
8 hr/day, 7 days/wk, for 4 weeks
0, 0.5, or 10 mg/m3
Test article: Not characterized
Main limitations: test article, exposure
generation NR; static chamber used.
Epididymis weight (g)e
0 0.5 10
-2 -31%*
Reproductive function
Low confidence
Reference: Xing Sv (2007)
Mice (unspecified strain), 7 males/group, mated
with untreated females
2 hr/day, 6 days/wk, for 4 weeks
0, 20.79, 41.57, or 83.15 mg/m3
Test article: Not characterized
Main limitations: Test article NC; exposure
generation, strain NR.
Mean live fetuses/litter
Mean percentage resorptions0
0 208 41^6 83^2
-3 -12 -18%:
7* 8* 10%*
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: NR = not reported; NC = not characterized; T = testosterone; LH = luteinizing hormone;
FSH = follicle-stimulating hormone.
aStudies that evaluated male reproductive system toxicity are included in this table. Studies are organized by
endpoint, species, and lowest dose tested.
bResponse relative to control for mean data, or incidence data.
cVosoughi et al. (2013; 2012) reported histopathology and sperm measure data for the same low-exposure group
study animals. However, serum LH and FSH data were presented only in Vosoughi et al. (2012) and serum T and
testes weight data were presented only in Vosoughi et al. (2013).
incidence data not reported.
eData digitized using Grab It!™, Datatrend Software.
^Statistically significant difference from control value, as reported by the study author.
Study exposure levels converted from ppm to mg/m3 are presented in italics (1 ppm = 1.23 mg/m3).
This document is a draft for review purposes only and does not constitute Agency policy.
1-426 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Evidence on Mode of Action for Developmental and Reproductive Effects
Mode of action (MOA) information for potential developmental and reproductive toxicity
associated with formaldehyde exposures is limited. No definitive data have been identified that
fully support a specific MOA for developmental outcomes, or for alterations in male or female
reproductive system conformation or function. Because it is considered unlikely that formaldehyde
is distributed via systemic circulation to the reproductive organs, this section discusses potential
mechanisms by which formaldehyde exposures might indirectly affect reproductive outcomes
following toxic insult at the portal of entry. Mechanistic events associated with respiratory health
effects (see Sections 1.2.1-1.2.4 and Appendix A.5.6) were considered. Biological mechanisms that
could plausibly be associated with developmental and reproductive toxicity are discussed, based
upon consideration of experimental animal data that included inhalation exposures to
formaldehyde. These include: oxidative stress and neuroendocrine-mediated effects (alterations of
adrenergic or gonadotropic hormones). Although additional study is needed to better define and
verify these potential mechanisms, they could be operant in several primary outcomes that have
been noted across toxicology or epidemiology studies with inhalation exposures to formaldehyde:
developmental delays, fetal loss, and effects on sperm quality and quantity.
1) Effects on the reproductive system that are due to indirect oxidative stress, possibly linked
to inflammatory responses following formaldehyde exposures (evidence from two high and
two low confidence studies (Zhou et al.. 2011b: Zhou etal.. 2006: Ozen etal.. 2005: Ozen et
al„ 20021
Oxidative stress/damage by reactive oxygen species (ROS) has been hypothesized to play a
role in reproductive and developmental toxicity (Wells and Winn. 1996: Tuchau etal.. 1992: Fantel
and Macphail. 1982). Markers of increased oxidative stress have been identified in the blood
following formaldehyde inhalation exposures (see Section 1.2.3), and thus, this could also be
occurring in peripheral tissues. Plausibly, inflammatory mediators, ROS, or other factors observed
in the blood could be operant in reproductive or developmental outcomes by indirectly eliciting
responses in the reproductive system or in the developing fetus.
ROS-related outcomes have been detected in cells and tissues distal from the POE, notably
in the male reproductive system, where testicular and epididymal toxicity and effects on sperm
have been observed. In a high confidence study in rats, Ozen et al. (2002) investigated the
mechanism of oxidative stress associated with testes toxicity by assessing testicular iron, copper,
and zinc levels. Zinc and copper levels were reduced in the rat testes, consistent with an increase in
testicular ROS. A medium confidence study in rats (Sapmaz etal.. 2018) identified a statistically
significant decrease in glutathione peroxidase (GSH-Px) activities and a statistically significant
increase in malondialdehyde (MDA) levels, A low confidence study fOzen etal.. 2008: Zhou etal..
2006) investigated biomarkers of oxidative stress as a potential MOA for testicular toxicity
following inhalation exposures of rats to formaldehyde. Significant effects on antioxidants and
redox enzymes were observed: decreases in superoxide dismutase (SOD), GSH-Px, and glutathione
This document is a draft for review purposes only and does not constitute Agency policy.
1-427 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
(GSH), as well as an increase in the oxidative stress biomarker, MDA. The authors also
demonstrated the protective effect of coadministration with the antioxidant vitamin E fZhou etal..
20061 on decreased testes weight, biochemical alterations, histopathological effects, or on sperm
count, motility, and morphology. Zhou et al. (2011b), another low confidence study from the same
research laboratory, demonstrated significantly decreased SOD and GSH-Px activities and
significantly increased MDA levels in the epididymides of rats exposed to formaldehyde. No studies
have been identified that specifically evaluated the generation of ROS in fetuses following maternal
inhalation exposures to formaldehyde, which would be directly informative to this potential
relationship.
Chemical or physical stress has been shown to increase the synthesis of heat shock protein
70 (Hsp70), which is involved in protein folding and repair fCraig and Schlesinger. 19851.
regulation of apoptosis fTakavama etal.. 20031. and it is synthesized during normal
spermatogenesis (Dix etal.. 1997: Dix. 1997). Additionally, testicular heat shock protein
immunoreactivity has been associated with human infertility (Werner et al.. 1997). Ozen at al.
(2005). a high confidence study, reported the detection of increased Hsp70 in spermatogenic cells
from the seminiferous tubules of rats following 13 weeks of inhalation exposure to formaldehyde.
The increase in testicular Hsp70 could reflect a response to chemical (formaldehyde) stress to the
respiratory system, but no mechanisms exist to explain this potential association. Regardless, the
role of heat shock proteins in mammalian fetal development is well-recognized (Walsh et al.. 1997).
It has also been proposed that oxidative stress resulting from formaldehyde exposure could
result in epigenetic consequences to the male reproductive system (Duong etal.. 2011). Tunc and
Tremellen (2009) reported that oxidative stress to sperm DNA has resulted in hypomethylation in
infertile men. Abnormal methylation of a key spermatogenic gene is associated with defective
sperm fNavarro-Costa et al.. 20101. This represents a hypothetical indirect mechanism by which
formaldehyde could influence methylation in sperm DNA and alter male fertility. None of the
studies reporting sperm alterations or related measures (see previous sections) examined the
potential role of sperm methylation in these outcomes.
2) Neuroendocrine-mediated mechanisms: disruption of the hypothalamus-pituitary-adrenal
gland (HPA) axis or hypothalamic-pituitary-gonadal (HPG) axis (evidence from three high,
one medium, and one low confidence studies—fVosoughi etal.. 2013: Vosoughi etal.. 2012:
Sari etal.. 2004: Ozen etal.. 2002: Sorg etal.. 2001a: Kitaev etal.. 1984)
A stress-induced mechanism might contribute to adverse outcomes on the reproductive
system and development in the absence of systemic distribution of formaldehyde.
Disruption of the HPA axis: Stressors such as chemical exposure can cause increased
secretion of CRH in the hypothalamus, ACTH in the anterior pituitary gland, and adrenal
corticosteroids in the adrenal gland fSmith and Vale. 20061. In support of this hypothesis, a high
confidence study, Sorg et al. (2001a). demonstrated an increase in blood corticosterone levels after
inhalation exposure to formaldehyde. Additionally, Sari et al. (2004). a medium confidence study,
This document is a draft for review purposes only and does not constitute Agency policy.
1-428 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
reported effects of inhalation formaldehyde exposures to mice on CRH neurons in the
hypothalamus and ACTH cells in the pituitary gland. The effects of stress on disruptions to
reproductive function and outcome in humans are well-recognized fNegro-Vilar. 1993: Barnea and
Tal. 1991: Mcgradv. 19841. The preoptic area of the hypothalamus is considered a potential site of
integration between the HPA axis and gonadal steroid hormones fSmith and Vale. 20061.
Disruption of the HPG axis: A steroidal endocrine-mediated mechanism would be consistent
with outcomes observed in some of the reproductive and developmental epidemiology and
toxicology studies. Developmental delays can result from effects on the maternal HPG axis.
Hormone levels in pups were not measured in any identified studies; however, there are three
studies in adult animals that have directly tested for changes in reproductive hormones after
formaldehyde exposure. Kitaev et al. (1984), a low confidence study, observed serum FSH increases
and LH decreases after inhaled formaldehyde in adult female rats. Alterations in hormone levels
could compromise pregnancy maintenance. Another potentially endocrine-mediated outcome, lack
of ovarian luteal tissue in females exposed to formaldehyde, was reported in a low confidence study
by Maronpot et al. (19861. In males, alteration of the HPG axis by formaldehyde exposure could
also be theoretically operant. Two high confidence inhalation studies with formaldehyde, Vosoughi
et al. f2013: 20121 and Ozen et al. (2002), reported significant serum testosterone level decreases,
accompanied by histopathological evidence of seminiferous tubule depletion. Vosoughi et al.
(2013: 2012) also reported a significant decrease in serum LH at 24 hours after inhalation
formaldehyde exposure. This is notable because the initiation and maintenance of spermatogenesis
in rodents and primates require LH stimulation (Plant and Marshall. 20011. Reduced testosterone
levels might also contribute to sperm quality and quantity decrements.
These two potential mechanisms are not necessarily mutually exclusive. If verified, they
could be shown to be acting alone for certain endpoints (in which case the others may not be
operant) or in concert for others. Nevertheless, as stated above, no definitive data have been
identified that define an MOA(s) explaining how developmental or reproductive outcomes might
occur following inhalation exposure to formaldehyde.
Integrated Summary of Evidence on Developmental and Reproductive Toxicity
Hazard conclusions integrating the evidence of developmental and reproductive hazards in
humans and animals were drawn for two categories: female reproductive or developmental toxicity
(TTP, spontaneous abortion, birth outcomes, fetal survival, growth, and malformations), and male
reproductive toxicity (see Table 1-58). Specifically, for the purposes of this assessment and based
on the outcomes reported in the epidemiological literature, female reproductive toxicity and
developmental toxicity were considered as a group because it is difficult to distinguish the
underlying events that may have resulted in either a delayed recognized pregnancy or fetal loss.
This document is a draft for review purposes only and does not constitute Agency policy.
1-429 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Female reproductive or developmental toxicity
While studies that evaluated physiological measures of reproductive health in females were
not available, two medium confidence studies reported strong associations of occupational
exposure to formaldehyde with decreased fecundability, increased TTP, and spontaneous abortion
(Taskinen et al.. 1999: Tohn etal.. 1994). A third study also reported an elevated risk of
spontaneous abortion with higher exposure frequency of similar magnitude, but the effect estimate
may have been biased to an unknown degree by confounding from coexposure to xylene (Taskinen
etal.. 19941. Excluding the study would not change the weight-of-evidence conclusion for the
epidemiological evidence. It is recognized that the decreased fecundability and increased TTP
might have resulted from early fetal loss, or be a consequence of alterations in maternal
reproductive function (discussed below). Only one of the occupational studies (in woodworkers)
reported the levels of formaldehyde that resulted in the observed associations (0.27 mg/m3)
(Taskinen et al.. 1999). Studies of hospital, nursing, or medical employees generally did not report
an association with formaldehyde exposure, although these low confidence studies tended to use
less informative exposure-assessment methods, a major limitation that reduced the sensitivity of
these studies. An association of uncharacterized birth defects with maternal exposure fZhu et al..
2006: Saurel-Cubizolles etal.. 1994: Hemminki et al.. 19851 was suggested in some occupational
epidemiological studies; the precision of the ORs was quite low, as indicated by the wide CIs, which
limited the sensitivity of these analyses. Three studies of pregnancy cohorts indicate an association
with fetal growth including biparietal diameter in the 2nd trimester and birthweight, although there
are questions about the interpretation of the results overall given the strength of associations
observed in a population with very low exposures (Franklin et al.. 2019) and a relatively weak
association with potential confounding by TVOCs in a population with higher exposure fChang et
al.. 20171. Preterm birth and low birth weight were not associated with higher formaldehyde
exposure among a cohort of male woodworkers in China (Wang etal.. 2012).
Animal studies evaluated several endpoints relevant to developmental toxicity
(i.e., decreased survival, decreased growth, or increased evidence of structural anomalies) or
female reproductive toxicity (i.e., ovarian and uterine pathology, ovarian weight, or hormonal
changes). All available studies were of low confidence, primarily due to exposure-quality concerns
(i.e., the use of formalin, or an uncharacterized test substance). In addition, there was considerable
heterogeneity in both of these data sets, and consistent evidence supporting manifestations of
toxicity after formaldehyde exposure was not reported. However, as several of these studies did
identify potential findings of concern, these outcomes are deserving of additional study. In
addition, several studies examining effects on the nervous system after formaldehyde exposure in
rats during development suggest that formaldehyde inhalation might have the potential to affect
the developing nervous system (see Section 1.3.1), however, additional studies are needed to clarify
these preliminary findings. Studies on developmental immunotoxicity were considered not
This document is a draft for review purposes only and does not constitute Agency policy.
1-430 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
informative (see Section 1.2.3 and Appendix A.5.4) and no epidemiological studies of children were
identified.
Overall, the evidence indicates that inhalation of formaldehyde likely causes increased risk
of developmental or female reproductive toxicity in humans, given the appropriate exposure
circumstances. This conclusion is based on moderate evidence in observational studies finding
increases in TTP and spontaneous abortion risk among women exposed to occupational
formaldehyde levels; the evidence in animals is indeterminate, and a plausible, experimentally
verified MOA explaining such effects without systemic distribution of formaldehyde is lacking. The
primary basis for this conclusion is from studies of women with occupational exposures to
formaldehyde concentrations as high as 1.2 mg/m3.
Male reproductive toxicity
Few epidemiological studies evaluated effects on the male reproductive system. Two
studies of male woodworkers in China from one research group reported associations with lower
total and progressive sperm motility, and delayed fertility and spontaneous abortion (Wang etal..
2015: Wang etal.. 20121. The investigators used a well-designed exposure assessment to evaluate
associations in this highly exposed occupational population (0.22-2.91 mg/m3). Two other studies
with low sensitivity to detect associations (due to concerns with low precision and exposure
misclassification) did not observe effects on sperm counts and morphology or spontaneous
abortion among exposed men (Lindbohm etal.. 1991: Ward et al.. 1984).
Animal studies were available that evaluated several effects from formaldehyde inhalation
exposure on the male reproductive system. A coherent set of high and medium confidence studies
in mice and rats that tested formaldehyde exposures >6 mg/m3 reported effects on multiple
endpoints, although interpretations could not be drawn regarding the potential for these effects in
experimental animals at lower formaldehyde exposure levels. Qualitative and quantitative
histopathological effects were observed in the testes and epididymides of a high confidence study in
rats (Ozen etal.. 2005) and another in mice (Vosoughi etal.. 2013: Vosoughi etal.. 2012) and in a
medium confidence rat study (Sarsilmaz etal.. 1999). Histopathological findings in testes were also
observed by (Sapmaz etal.. 2018). a medium confidence study in rats. These observations were
supported by similar findings in a number of low confidence studies. Decreased serum testosterone
(T) was also observed in the high confidence studies in rats and mice (Vosoughi etal.. 2013:
Vosoughi etal.. 2012: Ozen etal.. 20051. as well as in a low confidence rat study fZhou etal.. 2011bl.
The decreased serum T is biologically consistent with testicular Leydig cell damage observed in the
histopathological evaluations reported in well-conducted studies (Vosoughi etal.. 2013: Vosoughi
etal.. 2012: Sarsilmaz etal.. 1999). Downstream effects of disruptions in spermatogenesis
observed in the histopathology data included decreased sperm count and motility, and increased
sperm morphological abnormalities in a high confidence study in mice fVosoughi etal.. 2013:
Vosoughi etal.. 20121 and several low confidence studies in rats. Testes and epididymides weight
alterations are often correlated to some degree with histopathology in those organs; however,
This document is a draft for review purposes only and does not constitute Agency policy.
1-431 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
while significantly decreased dose- and duration-dependent testes weights were observed in the
high confidence study in rats by Ozen et al. (2002.), organ weight alterations were not observed in
the high confidence study in mice by Vosoughi et al. (2013.; 20121 or the medium confidence study
in rats by Sarsilmaz et al. (1999), and results in low confidence studies were mixed, preventing
interpretations.
Overall, the evidence indicates that inhalation of formaldehyde likely causes increased risk
of reproductive toxicity in men, given the appropriate exposure circumstances, based on robust
evidence in animals that presents a coherent array of adverse effects in two species, and slight
evidence from observational studies of occupational formaldehyde exposure. No plausible,
experimentally verified MOA exists to explain such effects without systemic distribution of
formaldehyde; however, some support for indirect effects in rodents is provided by relevant
mechanistic changes in male reproductive organs. The primary basis for this conclusion is based on
bioassays in rodents testing formaldehyde concentrations above 6 mg/mg3 (no medium or high
confidence studies tested lower exposure levels).
Data gaps
While reduced fecundity observed in exposed women may be due to reproductive toxicity
or toxicity to the developing fetus, no studies are available in exposed humans or animal
experiments that provide more complete assessments of reproductive organ endpoints. This also is
true for the evaluation of postnatal developmental toxicity. The anthropomorphic findings by a
single study of low residential exposures are concerning and additional studies are needed of these
endpoints. The findings by Wangetal. (2015) suggesting formaldehyde-related toxicity to sperm
and possible resulting effects on fecundity and fetal survival, and which may be supported by a low
confidence study in mice (Xingetal.. 2007a). provide evidence of male-mediated decreases in fetal
viability, and should be investigated further. Ideally, such investigations would include additional
human studies of different populations using similarly detailed exposure assessments, as well as
single or multigeneration reproductive toxicity studies in animals (which were not identified in the
current database). Such studies would also assess female reproductive outcomes, which are not
extensively evaluated in the current database. Ideally, any future toxicology experiments would
generate formaldehyde exposures using paraformaldehyde to eliminate the uncertainties
pertaining to potential confounding by methanol that limit the majority of currently available
studies on developmental and reproductive toxicity.
Importantly, as the hazard conclusion for male reproductive toxicity is based largely on
animal studies that only tested formaldehyde exposures >6 mg/m3 (one study) or >12 mg/m3,
which introduces uncertainties regarding potential irritation-related effects (e.g., reflex bradypnea,
which is not experienced by humans and is expected to be operant at these levels; see
Appendix A.3), well-conducted, detailed animal studies testing these endpoints at lower
formaldehyde concentrations are warranted.
This document is a draft for review purposes only and does not constitute Agency policy.
1-432 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-58. Evidence integration summary for effects of formaldehyde
inhalation on reproduction and development
Evidence
Evidence judgment
Hazard determination
Female reproductive or developmental toxicity
Human
Moderate for female reproductive or developmental toxicity, based on:
Human health effect studies:
• Two medium confidence studies in two independent populations
(woodworkers, cosmetologists): decreased fecundability and increased
spontaneous abortion risk. Supporting evidence of association with
spontaneous abortion from one low confidence study among laboratory
workers. All studies evaluated multiple exposure categories with highest
risk at highest exposure level.
• Two low confidence studies of maternal exposure among health workers
with low precision: small increased risk of malformations (all combined).
• Two medium confidence studies of pregnancy cohorts indicating
decreased birth weight and head circumference.
• Null evidence from five low confidence studies with low sensitivity:
fecundability, spontaneous abortion.
Biological plausibility: No direct evidence. However, evidence of elevated
oxidative stress in the blood of exposed adults (see Section 1.2.3) might
provide a potential indirect linkage (see explanation at right).
The evidence indicates that
inhalation of formaldehyde
likely causes increased risk of
developmental or female
reproductive toxicity in humans,
given the appropriate exposure
circumstances3
Primarily based on studies of
women with occupational
exposures to formaldehyde
concentrations as high as 1.2
mg/m3.
Potential susceptibilities: no
specific data were available to
inform potential differences in
susceptibility.
Animal
Indeterminate for developmental toxicity, based on:
Animal health effect studies:
• Mixed findings for evidence of decreased fetal survival (pre- or
postimplantation loss) across multiple low confidence studies
• Mixed findings for evidence of altered fetal or postnatal growth across
multiple low confidence studies. Variations in study design and reporting
deficiencies inhibit interpretation.
• Mixed findings for evidence of structural anomalies across multiple low
confidence studies.
Biological plausibility: No direct evidence. However, evidence of elevated
oxidative stress and hormonal alterations in the blood of adult rodents
(see Section 1.2.3) might provide a potential indirect linkage, as it is
recognized that both oxidative stress and the HPG axis have potential roles
in developmental toxicity.
Indeterminate for female reproductive toxicity, based on:
Animal health effect studies:
• Two low confidence studies in rats: decreased ovarian weight, ovarian
histopathology, and hormonal alterations
• One low confidence study in mice: Ovarian and uterine histopathology
(hypoplasia)
Biological plausibility. Neuroendocrine-mediated mechanisms, particularly
involving disruption of the hypothalamic-pituitary-gonadal axis, are
consistent with alterations of female reproductive hormones observed in
low confidence rodent studies following formaldehyde exposures.
Other
inferences
• Relevance to humans: Relevant health effects observed in humans are the
primary basis for the hazard determination.
• MOA: No experimentally established MOA exists, and any potential
mechanisms have not been well studied.
Male Reproductive Toxicity
This document is a draft for review purposes only and does not constitute Agency policy.
1-433 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Humans
Sliaht for male reproductive toxicity, based on:
Human health effect studies:
• One medium confidence study of exposure among male woodworkers:
inverse association with sperm motility measures, increased prevalence of
time to pregnancy, spontaneous abortion and birth defects.
• Null evidence for effects on sperm counts and morphology in one low
confidence study (because of low power).
Biological plausibility: No directly relevant studies were identified.
The evidence indicates that
inhalation of formaldehyde
likely causes increased risk of
reproductive toxicity in men,
given appropriate exposure
circumstances3
Primarily based on bioassays in
rats and mice testing
formaldehyde concentrations
above 6 mg/mg3 (no medium or
high confidence studies tested
lower exposure levels).
Potential susceptibilities: No
specific data were available to
inform potential differences in
susceptibility.
Animals
Robust for male reproductive toxicity, based on:
Animal health effect studies:
• One high confidence study in mice, three high or medium confidence
studies in rats, and five low confidence studies in rats: dose-related
qualitative or quantitative histopathological lesions of the testes or
epididymides.
• Null evidence for testes histopathology in one low confidence study in
mice.
• One high confidence study in mice and four low confidence studies in rats:
dose-related effects on epididymal sperm.
• One high confidence study in mice, one high confidence study in rats, and
one low confidence study in rats: dose-related decreased serum
testosterone (and decreased serum luteinizing hormone in the high
confidence study in mice).
• Mixed results for organ weight changes (i.e., testes; epididymis) across
multiple high, medium, and low confidence studies.
• One low confidence study in mice with evidence of male-mediated
decreases in fetal survival.
• Note: No multigeneration study was conducted.
Biological plausibility: Multiple biomarkers of oxidative stress, as well as heat
shock protein induction, have been observed in the testes or epididymides of
exposed rats in well-conducted studies. Heat shock protein
immunoreactivity and oxidative stress resulting in hypomethylated sperm
(no studies were identified that evaluated sperm methylation changes) were
linked to human male infertility.
Other
inferences
• Relevance to humans: Some uncertainty regarding the relevance of the
animal evidence exists, as the studies only tested extremely high
concentrations expected to cause strong irritant effects that may not
occur in humans; however, in light of the concordant findings in a well-
conducted study of humans and an absence of other evidence to the
contrary, the relevance of animal male reproductive toxicity outcomes to
humans is presumed.
• MOA: No experimentally established MOA exists, and any potential
mechanisms have not been well-studied; however, mechanistic data
provide some support for indirect effects on the male reproductive
system.
2
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2.
1.3.3. Lymphohematopoietic Cancers
3 The specific endpoints considered in this section include diagnoses of Hodgkin lymphoma,
4 multiple myeloma, myeloid leukemia, or lymphatic leukemia in exposed humans (note: diagnosis of
5 non-Hodgkin lymphoma, a nonspecific grouping of dozens of different lymphomas, was not
6 formally evaluated; see Appendix A.5.9), as well as experimental animal and mechanistic studies
This document is a draft for review purposes only and does not constitute Agency policy.
1-434 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
relevant to the interpretation of potential effects on the lymphohematopoietic (LHP) system. For
these subtypes, there have been different interpretations of the weight of evidence for whether
formaldehyde inhalation causes LHP cancers. Expert review panels have determined that there is
sufficient evidence to conclude that formaldehyde inhalation increases the risk for myeloid
leukemia based on the results of epidemiological studies alone fNTP. 20111. or additionally
supported by mechanistic research (NRC. 2014b: IARC. 2012). Two European Union scientific
bodies were not in agreement with those conclusions, noting that although there is evidence of
associations between formaldehyde exposure and LHP cancers in the epidemiological literature, the
observations are not biologically plausible since formaldehyde is not distributed to distal tissues
preventing direct interactions in the bone marrow fSCOEL. 2017: ECHA. 20121. Health Canada did
not draw a hazard conclusion for LHP cancer subtypes in their assessment of carcinogenesis and
other health effects for formaldehyde, which was finalized prior to the publication of several
epidemiological studies that reported associations (Health Canada. 2006. 2001). An independent
review of the evidence was conducted and is presented in this section.
Human studies provided robust evidence for myeloid leukemia and slight evidence for
multiple myeloma based on epidemiology studies of occupational formaldehyde levels either in
specific work settings (e.g., cohort studies) or in case-control studies. Aneuploidy in chromosomes
1, 5, and 7 in circulating myeloid progenitor cells, considered a potential primary target for LHP
carcinogenesis, was associated with occupational formaldehyde exposure. The type of aneuploidies
observed in the formaldehyde-exposed asymptomatic human workers are also found in patients
with leukemia, as well as in other worker cohorts at increased risk of developing leukemias, which
provides support for the plausibility of an association between chronic formaldehyde exposure and
leukemogenesis. Moreover, the strong and consistent evidence from a large set of studies that
observed mutagenicity in circulating leukocytes of formaldehyde-exposed humans, specifically
chromosomal aberrations (CA), and micronucleus (MN) formation, provides additional evidence of
biological plausibility for these cancer types. Further support is provided by studies that observed
perturbations to immune cell populations in peripheral blood associated with formaldehyde
exposure. In particular, decreases in red blood cells (RBCs), white blood cells (WBCs), and
platelets, along with a 20% decrease in colony-forming units that arose in vitro as descendants
from dedicated progenitors of granulocytes and macrophages (CFU-GMs) were observed in the
same exposed group, suggesting both a decrease in the circulating numbers of mature RBCs and
WBCs as well as possible decreases in the replicative capacity of myeloblasts.
Increased LHP cancers have not been observed in a well-reported chronic rodent bioassay
involving inhalation exposure of both rats and mice to formaldehyde, nor in another rat bioassay
that failed to report the incidence of non-nasal neoplastic lesions. Further, positive associations
with leukemia have not been reported in rodent studies. Thus, there appears to be a lack of
concordance between evidence from chronic rodent bioassays and human epidemiological
evidence, although such concordance is not necessarily expected fU.S. EPA. 2005al.
This document is a draft for review purposes only and does not constitute Agency policy.
1-435 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Taken together, based on the robust human evidence for these cancers from studies that
reported increased risk in groups exposed to occupational formaldehyde levels, the evidence
demonstrates that formaldehyde inhalation causes myeloid leukemia in humans, given
appropriate exposure circumstances. Separately, based on a limited number of epidemiological
studies and potentially relevant mechanistic evidence in exposed humans, the evidence suggests,
but is not sufficient to infer, that formaldehyde inhalation might cause multiple myeloma and
Hodgkin lymphoma, given appropriate exposure circumstances. While mechanisms for the
induction of myeloid leukemia are yet to be elucidated, they do not appear to require direct
interactions between formaldehyde and bone marrow constituents, and either are different in
animals or the existing animal models tested thus far do not characterize the complex process
leading to cancers in exposed humans.
Literature Search and Screening Strategy
The primary databases used for the literature searches were PubMed, Web of Science, and
Toxline, with the last update of the search completed in September 2016 (see Appendix A.4.7, A. 5.9
and A.5.6), and a systematic evidence map updating the literature through 2021 (see Appendix F).
The occurrences of lymphohematopoietic cancers in humans have been described and grouped
according to the International Classification of Disease (ICD) coding rubrics. Epidemiological
reviews were restricted to those specific cancer diagnoses available in the epidemiological
literature. The primary focus of this review was the specific lymphohematopoietic cancers that are
most commonly reported, myeloid leukemia, lymphatic leukemia, multiple myeloma, and Hodgkin
lymphoma. Published results for nonspecific aggregations of lymphomas, "all leukemias," and "all
lymphohematopoietic cancers" were not reviewed. Only primary epidemiological studies of
specific cancer endpoints with identified or inferred formaldehyde exposure were included.
Additional studies were identified from review articles and government documents. Studies of
non-Hodgkin lymphoma were not formally reviewed (see Appendix A.5.9). In addition, three
pertinent primary research articles and an unpublished Battelle-Columbus report (Battelle. 1982)
were considered relevant to investigations of leukemias following formaldehyde exposure in
experimental animals; these four studies were evaluated. Literature searches pertaining to
potential mechanisms relevant to LHP carcinogenicity, including genotoxicity (Appendix A.4) and
inflammation- and immune-related changes (Appendix A.5.6) also were conducted.
The bibliographic databases, search terms, and specific strategies used to search them are
provided in Appendix A.4, A.5.6, and A.5.9, as are the specific PECO criteria. Literature flow
diagrams summarize the results of the sorting process using these criteria and indicate the number
of studies that were selected for consideration in the assessment through 2016 (see Appendix F for
the identification of newer studies through 2021). The relevant human and animal health effect
studies (i.e., meeting the requirements outlined above), and mechanistic data informative to LHP
cancers were evaluated to ascertain the level of confidence in the study results for hazard
identification (see Appendix A.4.7, A.5.6 and A.5.9).
This document is a draft for review purposes only and does not constitute Agency policy.
1-436 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Overview of Lymphohematopoietic Cancer Biology
LHP neoplasias describe a broad group of cancers of the blood, bone marrow, and lymph
nodes, which includes leukemia, lymphoma, and myeloma. The various LHPs originate through a
multistep process in different stages of the hematopoietic pathway (the process through which
blood cells are formed). In normal human adults, this process occurs primarily in the bone marrow,
with the exception of lymphocytes, which continue to mature in the thymus, spleen, and peripheral
tissues. Therefore, LHPs may derive from discrete precursor or stem cells, as well as mature
lymphoid cells. Figure 1-36 illustrates the hematopoietic pathway, the location of each
differentiation (bone marrow or peripheral tissues), and the likely site of occurrence for
transformation in each subtype of LHP. Briefly, normal hematopoietic stem cells differentiate into
one of two lineages: myeloid or lymphoid progenitor cells. Normal myeloid progenitor cells may
then differentiate into mature RBCs, platelets, or granulocytes; lymphoid progenitor cells derive T
and B lymphocytes as well as natural killer (NK) cells and dendritic cells (see Figure 1-36).
LHP neoplasias arise from abnormal hematopoietic and lymphoid cells that are unable to
differentiate normally to form mature blood cells. Neoplasias following the myeloid lineage are
designated as chronic or acute leukemias, depending on the rate of expansion and the dominant
stage of cell differentiation. Acute leukemias are characterized by a rapid onset, whereas chronic
leukemias develop slower and progress over a period of months or years. Lymphoid neoplasias
may either reside in the blood as chronic or acute lymphoblastic leukemias or develop within
peripheral lymphoid sites such as the lymph nodes, spleen, or thymus—these are designated as
lymphomas. Some rare leukemias exhibit both myeloid and lymphoid characteristics and are
known as biphenotypic leukemias (Russell. 1997).
The majority of leukemias originate in the bone marrow at the hematopoietic stem cell
stage or at a later, lineage-restricted stage. Specifically, adult leukemias of myeloid origin such as
acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and chronic myeloid leukemia
(CML) as well as adult acute lymphoblastic leukemia (ALL) are thought to originate at the stem or
progenitor cell stage (Warner et al.. 2004).
Lymphomas primarily derive from mature lymphoid cells in peripheral tissues such as the
spleen, lymph nodes, and thymus, and are generally classified as either Hodgkin or non-Hodgkin
lymphomas (NHLs) depending on the appearance of a specific cancer cell type found in Hodgkin
lymphomas. Within the larger groupings of NHLs are numerous subtypes with unique
characteristics and origins. Myeloma (also called multiple myeloma) is a cancer of the plasma cells
that forms a mass or tumor located in the bone marrow. Most lymphomas and all myelomas, as
well as some rare leukemias/lymphomas (adult T cell leukemia [TCL], adult chronic lymphocytic
leukemia [CLL], prolymphocytic leukemia [PLL], and hairy cell leukemia [HCL]) originate in mature
lymphoid cells (Harris etal.. 2001: Greaves. 1999).
While hematopoietic stem cells are normally located in the bone marrow, they do
spontaneously mobilize into the peripheral blood at low levels, or in response to chemical insult,
This document is a draft for review purposes only and does not constitute Agency policy.
1-437 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
mobilize in large numbers (Schulz et al.. 2009: Levesque etal.. 20071. These mobilized cells remain
in circulation for very short periods of time (minutes to hours) and then localize to peripheral
tissues or in some cases, return to the bone marrow. Consequently, there may be a recirculation of
hematopoietic stem cells between the bone marrow and other peripheral tissues. Therefore, the
potential exists for DNA damage or other types of leukemogenic alteration during this mobilization
between tissues. Cells confined to the bone marrow are less vulnerable to environmental insult
than cells that enter the general circulation. Therefore, knowledge of the location of origin of
discrete LHPs is important to understanding the potential targets of carcinogenic compounds.
g
O
£
QJ
tz
o
CO
Embryonic
Stem cells
Germ line
i
&>j
i
Blood stem cell C3
X \
Myeloid precursor cell O
Rare AL
AML, CIVIL, MDS
ALL
Lymphoid precursor cell
Childhood AML
3
E
CJ _
¦B S
— Q.
ro w
7
~V \ l\\
©
©
© © ©
NHL, HL, MM
Adult TCL and CLL
PLL, HCL
o>
Red blood Platelets Granulocytes
cells
a)
CL "O
*C O
<11 C
JZ
Q-
E
B & T lymphocytes
NK cells
White Blood Cells
Figure 1-36. The hematopoietic pathway and likely sites of neoplastic
transformation for LHPs.
Abbreviations: AML = acute myeloid leukemia; CML = chronic myeloid leukemia; MDS = myelodysplastic
syndrome; ALL = acute lymphoblastic leukemia; NHL = non-Hodgkin lymphoma; HL = Hodgkin lymphoma;
MM = multiple myeloma; TCL = T cell lymphoma; CLL = chronic lymphocytic leukemia;
PLL = prolymphocytic leukemia; HCL = hairy cell leukemia (adapted from:
https://www.seattlecca.org/diseases/chronic-myeloid-leukemia-cml/cml-facts-0).
Lymphohematopoietic Cancers in Human Studies
Each specific type of LHP cancer (myeloid leukemia, lymphatic leukemia, multiple myeloma,
and Hodgkin lymphoma) is reviewed and evaluated independently in the sections below. For each
type of LHP cancer, the evidence is organized by considerations that inform the strength of
evidence (e.g., consistency, exposure-response) and evaluation of the potential for bias and
insensitivity in individual studies to affect the estimates of relative risk (RR). Evidence tables for
This document is a draft for review purposes only and does not constitute Agency policy.
1-438 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
each type of LHP cancer (Tables 1-59 through 1-62) are included that are organized first by the
study evaluation conclusions (i.e., high, medium, low) and then by publication year.
Methodological issues and approaches for evaluation
The epidemiology studies generally examined occupational exposure to formaldehyde
either in specific work settings (e.g., cohort studies) or in case-control studies. The considerations
with respect to design, exposure assessment, outcome assessment, potential bias and confounding,
and analysis differ for these different types of studies, and are discussed in more detail in
Appendix A.5.9. Because a single epidemiology study may report on several different cancer
endpoints, the confidence classifications are for the specific cancer results and are not judgments
on the study as a whole except when a study has only a single cancer endpoint. The distinction here
is important in that a study of adequate quality overall may still report an effect estimate judged to
be of low confidence due to the rarity of the cancer outcome, the rarity of the exposure, or
noncritical biases that are expected to yield effect estimates that underestimate any true effect.
The diagnosis of cancers in epidemiological studies has historically been ascertained from
death certificates according to the version of the International Classification of Diseases (ICD) in
effect at the time of study subjects'deaths [i.e., ICD-8 and ICD-9: (WHO. 1977.1967)]. The most
specific classification of diagnoses commonly reported across the epidemiological literature was
based on the first three digits of the ICD code (i.e., myeloid leukemia ICD-8/9: 205) without further
differentiation—although a few studies reported results at finer levels (i.e., Acute Myeloid
Leukemia ICD-8/9: 205.0), and these are discussed.
For some cancers, the reliance of cohort studies on death certificates to detect cancers with
relatively high survival may have underestimated the actual incidence of those cancers, especially
when the follow-up time may have been insufficient to capture all cancers that may have been
related to exposure. The potential for bias may depend upon the specific survival rates for each
cancer. Five-year survival rates vary among the selected cancers, from 86% for Hodgkin lymphoma
to less than 50% for multiple myeloma (MM) and myeloid leukemia (ML). EPA considered the
likelihood of underreporting of incident cases to be higher for mortality-based studies of Hodgkin
lymphoma and LL, which may result in undercounting of incident cases and underestimates of
effect estimates compared to general populations (e.g., Mavr etal.. 2010: Hansen and Olsen. 1995:
Hansen etal.. 1994: Hayes etal.. 1990: Soletetal.. 1989).
The overwhelming majority of information bias in epidemiological studies of formaldehyde
stems from the use of occupational records to gauge exposures with some degree of random
exposure misclassification or exposure measurement error considered to be commonplace. A
primary consideration in the evaluation of these studies is the ability of the exposure assessment to
reliably distinguish among levels of exposure within the study population, or between the study
population and the referent population. A large variety of occupations were included within the
studies; some represented work settings with a high likelihood of exposure to high levels of
formaldehyde, and some represented work settings with variable exposures and in which the
This document is a draft for review purposes only and does not constitute Agency policy.
1-439 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
proportion of people exposed was quite small. In the latter case, the potential effect of
formaldehyde would be "diluted" by the random exposure measurement error within the larger
study population, limiting the sensitivity of the study. The exposure-assessment methods of the
identified studies were classified into four groups (A through D), reflecting greater or lesser degree
of reliability and sensitivity of the measures (see Appendix A.5.9). Outcome-specific associations
based on Group A exposures were considered to be without appreciable information bias due to
exposure measurement error while those based on Groups B-D were considered to be somewhat
biased toward the null.
Additional exposure measurement error may arise in circumstances when the time period
of exposure assessment is not well aligned with the time period when formaldehyde exposure
could induce carcinogenesis that develops to a detectable stage (incident cancer) or result in death
from a specific cancer. The cohort studies were evaluated to assure that they analyzed the analytic
impact of different lengths of "latency periods" (i.e., excluded from the analyses the formaldehyde
exposure most proximal to each individual's cancer incidence or cancer mortality). Analyses that
did not evaluate latency were considered to be somewhat biased toward the null because irrelevant
exposure periods were included.
Studies with small case counts may have little statistical power to detect divergences from
the null but are not necessarily expected to be biased and no study was excluded solely on the basis
of case counts as this methodology would exclude any study that saw no effect of exposure.
Therefore, cohort studies with extensive follow-up that reported outcome-specific results on a
number of different cancers, including very rare cancers, were evaluated even when few or even no
cases were observed—if information on the expected number of cases in the study population was
provided so that Cis could be presented to show the statistical uncertainty in the associated effect
estimated.
In addition to potential bias, study sensitivity was specifically evaluated; study results with
low sensitivity could result in effect estimates that underestimated a "true" association if it existed.
For example, an outcome-specific effect estimate based on fewer than five observed cases of a
particular cancer would be classified as low confidence based on a lack of sensitivity—even if there
were no appreciable biases. Another example would be a study that might have relied on exposure-
assessment methodologies that were unbiased, but were nonspecific in nature, so as to yield effect
estimates that were likely biased toward the null and thus underestimated any true effect Finally,
cohort studies should have a sufficiently long follow-up period to allow for any exposure-related
cancer cases to develop and be detected and, ideally, allow for analyses of potential cancer latency.
Outcome-specific effect estimates from cohort studies with short follow-up could be considered
uninformative depending on the size of the study population and the baseline frequency of the
cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
1-440 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Myeloid leukemia
Epidemiological evidence
The most specific classification of myeloid leukemia diagnosis that is commonly reported
across the epidemiological literature has been based on the first three digits of the Eighth or Ninth
Revision of the ICD code (i.e., myeloid leukemia ICD-8/9: 205) —although the smaller sets of
studies that reported specific results for AML (ICD-8/9: 205.0) and CML (ICD-8/9: 205.1) are
discussed. For the purposes of this evaluation, cancer cases reported as monocytic leukemia or
nonlymphocytic leukemia were included as myeloid leukemia. Evidence describing the association
between formaldehyde exposure and the risk of myeloid leukemia was available from 13
epidemiological papers reporting on 10 different study populations—three case-control studies
(Talibov etal.. 2014: Hauptmann et al.. 2009: Blair etal.. 2001) and nine cohort studies (Coggon et
al.. 2014: Pira etal.. 2014: Meyers etal.. 2013: Saberi Hosnijeh etal.. 2013: Beane Freeman et al..
2009: Hayes etal.. 1990: Ottetal.. 1989: Stroup etal.. 1986: Walrath and Fraumeni. 1984.1983).
Hauptmann et al. (2009) combined the study populations from Hayes et al. (1990) with those from
Walrath and Fraumeni (1984,19831 and reconstructed individual exposure estimates. Checkoway
et al. (2015.) reanalyzed Beane Freeman et al. (2009) with a different definition of the exposure
categories and presented results for specific subtypes of myeloid leukemia. These are the only
formaldehyde studies that specifically evaluated the risk of myeloid leukemia. The outcome-
specific evaluations of confidence in the reported effect estimate of an association from each study
are provided in Appendix A.5.9, and the confidence conclusions are provided in the evidence table
for myeloid leukemia (see Table 1-60) following the causal evaluation.
Consistency of the observed association
The majority of studies of the 10 populations reported elevated risks of myeloid leukemia
(or a specific subtype) associated with exposure to formaldehyde for at least one metric of
exposure, although four low confidence studies reported results based on fewer than 10 cases and
two other low confidence studies reported relative effect estimates of RR = 1.02 and OR = 1.17.
These studies examined different populations, in different locations and exposure settings, and
using different study designs. The study results presented in Table 1-60 (by confidence level and
publication date) detail all of the reported associations between exposures to formaldehyde and the
risks of developing or dying from myeloid leukemia along with a summary graphic of any limitation
and the confidence classification of the available effect estimates. Results for all studies are plotted
in Figure 1-37 and grouped by the exposure-assessment methodology (e.g., population-level versus
individual-level) and by the type of occupation of the exposed workers (e.g., anatomist/embalmers,
industrial workers, garment workers). The same results for the high and medium confidence
studies are plotted in Figure 1-38, and exposure-response trends describing the effect estimates of
association between formaldehyde exposure and risk of myeloid leukemia in high confidence
studies are shown in Table 1-59.
This document is a draft for review purposes only and does not constitute Agency policy.
1-441 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
The first five studies in Figure 1-37 (Pira etal.. 2014: Hayes et al.. 1990: Stroup etal.. 1986:
Walrath and Fraumeni. 1984.19831 shown at the left, under the header "Population-level exposure
assessment" followed the health of a group of workers exposed to formaldehyde in a plastics
manufacturing facility and four sets of anatomists and embalmers—professions known to be
exposed to formaldehyde. These studies compared the risk of death from myeloid leukemia among
those workers to the risk of death from myeloid leukemia among the general population. All five
studies showed elevated RRs of myeloid leukemia mortality as measured by the mortality ratios,
including two studies with 95% CIs that excluded the null, thereby decreasing the likelihood of
chance as an alternative explanation for these findings. One study f Stroup etal.. 19861 observed a
much higher RR (standardized mortality ratio [SMR] 8.8) compared with the others (SMR -1.4 to
2.0); this higher estimate was based on one subtype (CML), and was relatively imprecise (95% CI:
1.8, 22.5). The results from Pira et al. (2014) and Stroup et al. (1986) were classified with low
confidence. The results from the other three studies fHaves etal.. 1990: Walrath and Fraumeni.
1984.1983) were classified with medium confidence and are shown in Figure 1-37 to document
their findings while acknowledging that these three studies populations were combined in
(Hauptmann et al.. 20091.
The second set of eight studies (Coggon etal.. 2014: Talibov etal.. 2014: Meyers etal.. 2013:
Saberi Hosniieh etal.. 2013: Beane Freeman et al.. 2009: Hauptmann etal.. 2009: Blair etal.. 2001:
Ottetal.. 19891 is displayed in Figure 1-37 beneath the header of "Individual-level exposure
assessment." A general strength of this second set of eight studies was their use of individualized
exposure data, which, for six of the studies, allowed for the evaluation of exposure-response
relationships with increased formaldehyde exposures using multiple metrics of exposure;
additional detail of this consideration is included below under the exposure-response relationships
section below. A further strength is that three of these studies had their effect estimates classified
with high confidence (Meyers etal.. 2013: Beane Freeman et al.. 2009: Hauptmann et al.. 20091 and
were able to evaluate the impact of the timing of initial exposure relative to mortality; further detail
of this consideration is included below under the temporal relationship section below. One study's
results that were classified with medium confidence due to exposure measurement error (Coggon et
al.. 2014) showed a slightly elevated risk for those workers with the highest job exposures, but also
slightly decreased risk for those with the highest duration of exposure. The results from the other
four studies with individual-level exposure assessment were classified with low confidence due to
the lower quality exposure assessment methods (T alibov et al.. 2 014: Saberi Hosniieh etal.. 2013:
Blair etal.. 2001: Ottetal.. 19891. Additional findings from each of the studies are provided in
Table 1-60. Different measures of exposure reflected different risks and this was true within
studies and across studies but all provided some evidence of increased risk of dying from myeloid
leukemia associated with formaldehyde exposure. One study showed the strongest relationship of
myeloid leukemia mortality with duration of formaldehyde exposure (Hauptmann etal.. 2009).
Another showed increased risks for peak exposure and average exposure but not for cumulative
This document is a draft for review purposes only and does not constitute Agency policy.
1-442 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
exposure or "any" exposure (Beane Freeman et al.. 2009). The Checkoway et al. (20151 reanalysis
of Beane Freeman et al. (2009) reported nonsignificant increased risks of AML and CML after
redefining the referent group to include all workers with peak exposures of less than 2 ppm as well
as some originally classified as having peak exposures of greater than 4 ppm because those
worker's peak exposures were thought to be either too frequent or too rare fBeane Freeman et al..
2009). The result of this change in exposure assessment shifted nine cases of myeloid leukemia
from the highest exposure category to the lowest exposure category (Checkoway etal.. 2015).28
Because this change in methodology for exposure assessment blends the highly exposed people
with the low and unexposed people and thereby induces bias toward the null reducing study
sensitivity, these results were classified with low confidence. A third study showed increased risk
in the study population as a whole that was stronger among workers with the longest duration of
exposure and workers with the greatest length of time since first occupational exposure to
formaldehyde (Meyers etal.. 2013).
The pattern of increased risk of myeloid leukemia (ICD-8/9: '204') associated with exposure
to formaldehyde reflects the associations seen within two subtypes, AML and CML. Among the
studies with separate estimates by subtype, risks were elevated for both AML and CML, with the
associations for CML appearing to be as strong as or stronger than the associations with AML. Four
studies reported specific results for CML fChecko way etal.. 2015: Saberi Hosniieh etal.. 2013: Blair
etal.. 2001: Stroup etal.. 1986). All four studies reported elevated risks of CML. Six studies
reported specific results for AML; two were classified with high confidence (Meyers etal.. 2013:
Hauptmann etal.. 2009). and four with low confidence (Checkoway etal.. 2015: Talibov etal.. 2014:
Saberi Hosniieh etal.. 2013: Blair etal.. 2001). Both of the high confidence results showed
nonsignificantly elevated risks of AML associated with formaldehyde, as did three of four of the low
confidence results—although substantially higher risks were reported in the high confidence
results. One low confidence result showed a slight decrease in risk of AML fBlair etal.. 20011.
Results specific to AML are plotted in Figure 1-39. Four of these six studies reported effect
estimates for both ML and AML (Checkoway etal.. 2015: Meyers etal.. 2013: Saberi Hosnijeh et al..
2013: Hauptmann etal.. 2009) on a total of 14 specific metrics of exposure. To assess whether the
results for AML were comparable to those for ML, the pair-wise effect estimates were evaluated.29
The correlation between the AML results and the ML results was 0.72 (p < 0.0001) and the slope
28In Beane Freeman et al. (20091. for peak exposure there were four cases of ML who were unexposed, 14
cases with peak exposure from >0 to <2 ppm, 11 cases with peak exposure from 2 to <4 ppm, and 19 cases
with peak exposure >4 ppm. In Checkoway et al. (20151 the new definition of peak exposure and the
recategorization resulted in 27 cases of ML with peak exposures from 0 to <2 ppm, 11 cases with peak
exposure from 2 to <4 ppm, and 10 cases with peak exposure >4 ppm. The Checkoway et al. (20151 results
were classified with low confidence due to information bias and low sensitivity.
29 Based on six paired effect estimates from Hauptmann et al. (2009), five paired estimates from Meyers et al.
(2013), two paired effect estimates from Checkoway et al. (2015) and one pair of effect estimates from Saberi
Hosnijeh et al. (2013).
This document is a draft for review purposes only and does not constitute Agency policy.
1-443 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
was 0.97 indicating a very strong alignment among these studies and strongly suggesting that the
collective results for the broader group of ML cases may be inferred to represent AML as well.
Strength of the observed association
While reported relative effect estimates were consistently elevated above the null value of
one across the 10 study populations, the magnitude of the relative effect estimates varied with the
quality of the exposure assessment. Studies with higher quality exposure data based on individual-
level exposure assessment generally reported higher relative effect estimates (stronger
associations). The Hauptmann (2009) study reported the strongest association based on 34 cases
of myeloid leukemia of whom 33 had ever performed an embalming (OR = 11.2, 95% CI 1.3, 95.6; p
= .027); however, with just 1 case subject who had never embalmed in the reference group, the
effect estimate, while statistically significant, is imprecise. The investigators conducted additional
analyses that defined the reference group as having performed fewer than 500 embalmings so as to
include five cases of myeloid leukemia in the reference group and those results are discussed
below.
The results at the highest levels of formaldehyde exposure showed an approximately two-
to three-fold relative increase in risk of mortality from myeloid leukemia (Meyers etal.. 2013:
Beane Freeman et al.. 2009: Hauptmann etal.. 2009: Blair etal.. 20011 with one exception, which
reported no increase in risk among those who had ever had a job in the highest category of
exposure (Coggon etal.. 2014). This may have been due to the choice in (Coggon et al.. 2014) to
classify as highly exposed all workers who ever worked in a highly exposed job, even if just for one
year out of 20, a methodology that mixes workers with many years of high exposure together with
workers with just a single year of high exposure, thereby potentially diluting the strength of the
association. Results from other studies using a cruder exposure classification (i.e., exposed versus
not exposed), and low to medium confidence, generally showed elevated risks in the 1.02- to 2-fold
range (Pira etal.. 2014: Talibovetal.. 2014: Saberi Hosnijeh etal.. 2013: Ottetal.. 1989). Results
from the studies with higher quality exposure data were judged with greater confidence.
Temporal relationship of the observed association
Two related aspects of time are encompassed in the consideration of temporality. One
aspect is the necessity for the exposure to precede the onset of the disease. In each of the studies,
the formaldehyde exposures among the study participants started prior to their diagnoses of
myeloid leukemia or deaths from myeloid leukemia and in the studies that ascertained individual-
level exposures, the estimation of formaldehyde exposures was based on job titles and was done in
a blinded fashion with respect to outcome status. The second aspect involves the time course of
formaldehyde exposures in relation to the incidence of myeloid leukemia and death from myeloid
leukemia; this aspect of time is defined as the etiologically relevant window of time when exposure
to a causal factor is relevant to the causation of disease. From the epidemiological literature of
benzene-related leukemia, it is known that there can be an induction/latency period for some
This document is a draft for review purposes only and does not constitute Agency policy.
1-444 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
environmental agents and that the induction period may exceed 10 years (Rinskv et al.. 19871. The
epidemiological literature for formaldehyde and myeloid leukemia describes three studies that
evaluated the impact of the TSFE fMevers etal.. 2013: Beane Freeman et al.. 2009: Hauptmann et
al.. 20091. All three studies show some indication of an increase in risk at about 15-20 years of
time since first exposure (TSFE) to formaldehyde that is consistent with a biologically relevant
induction/latency period. However, the Hauptmann et al. (2009) study clearly shows increased
risk at 20+ years of time since first exposure. (Beane Freeman etal.. 2009) reported that the best
fitting exposure lag length of time to potentially account for cancer latency was 18 years. While
those three studies support the estimation of the beginning of the potentially relevant window of
time, the window may also have an ending when exposures that have occurred a very long time
before may no longer be relevant to the causation of disease.
In the mortality follow-up of this cohort through 1980, the High peak exposure had RR =
3.92 (95% CI 0.78,19.67; p-trend = 0.12) (Blair etal.. 1986): in the follow-up through 1994, the
High peak exposure had RR = 2.79 (95% CI 1.08, 7.21; p-trend = 0.02) (Hauptmann et al.. 2003):
and in the follow-up through 2004 the RR = 1.78 (95% CI 0.87, 3.64; p-trend = 0.07). Beane
Freeman (2009) reported the effect estimates for follow-up through every individual calendar year
starting with 1965 and ending with 2004. Figure 1 of fBeane Freeman et al.. 20091 shows the
association between peak formaldehyde exposure and the risk of myeloid leukemia; risks of High
exposures were compared against the lowest exposed category. Risks were significantly elevated in
each year of follow-up during the 1990's before losing significance in the 2000's. Such a pattern
may reflect the closing of the potentially relevant window of time when exposures are relevant to
disease causation. With very long follow-up of a cohort, those workers who were highly exposed
may have experienced a window of increased risks of myeloid leukemia associated with exposure
to formaldehyde that tapered off or closed. This phenomenon may occur as additional background
cases of myeloid leukemia - unrelated to formaldehyde exposure, were added to both the High and
the Low exposures groups thereby bringing the relative risks of these groups toward the null value
of 1.00.
As formaldehyde exposure had ceased by 1980 for all but 3.5% of person-time and latency
analyses showed higher risks in the period 15 to 25 years after first exposure with the best fitting
exposure lag of 18 years fBeane Freeman et al.. 20091. the 1994 follow-up of the NCI formaldehyde
cohort fHauptmann et al.. 20031 which reported that High peak exposure had RR = 2.79 (95% CI
1.08, 7.21; p-trend = 0.02) may be a more informative estimate of the association between
formaldehyde exposure and risks of myeloid leukemia. There is some indication that a similar
phenomenon may have occurred in the study of garment worker and the mortality follow-up
through 1988 (Pinkerton et al.. 2004) which reports somewhat stronger results for workers with
20+ years TSFE than was reported in the 2008 follow-up (Meyers etal.. 2013)(SMR = 1.91; p < 0.05
vs. SMR = 1.49 (95% CI 0.90, 2.32); and for duration longer than 10 years (SMR = 2.19 vs. SMR
=1.84). If the follow-up of these two cohorts has exceeded the window of time when exposures are
This document is a draft for review purposes only and does not constitute Agency policy.
1-445 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
relevant to disease causation, then the evidence may be somewhat stronger than is evident in the
reports from the most recent follow-ups.
Exposure-response relationship
Of the studies that provided evidence to evaluate the association between exposure to
formaldehyde and the risk of myeloid leukemia, four studies (Hayes etal.. 1990: Stroup etal.. 1986:
Walrath and Fraumeni. 1984.1983) followed the health of anatomists and embalmers and did not
have specific individual-level exposure data to assess an exposure-response relationship. One
study fOttetal.. 19891 did assess individual-level exposures but did not report differentiated risks
by exposure levels of formaldehyde. One study, Saberi Hosnijeh et al. (20131. which had risk
analyses on three levels of exposure for other health endpoints, did not identify any people with
high exposures to formaldehyde and thus could only compare risks of low exposures with risks of
no exposures.
The remaining studies did present distinct risk estimates differentiated by formaldehyde
exposure levels. Meyers et al. (20131 reported results by workers' year of first exposure, their time
since first occupational exposure, and by their duration of exposure. Data on cumulative exposure
was not available. The investigators considered that the initial study years (prior to 1963) had the
highest formaldehyde exposures as ongoing industrial hygiene practices were thought to have
decreased exposures over time. For first employment in the earliest period (before 1963), the
overall SMRwas 1.37 (95% CI 0.75, 2.30) while first employment in the middle (1963-1970) and
late time periods (after 1970) had ORs of 1.13 and 1.15. There was an extensive investigation of
exposure-response by duration of exposure with external and internal comparisons by strata of
duration as well as multivariate Poisson modeling of exposure duration, all of which showed
increasing risk with longer duration (see Table 1-60). Multiple models all showed positive trends
of increasing rate ratios with increasing exposure duration (see Figure IB in fMevers etal.. 2013
but the continuous model with duration was not statistically significant with rate ratio of 1.04 per
one year increase in duration (95% CI 0.097.1.12) provides only modest evidence of an exposure-
response relationship based on duration of exposure.
Beane Freeman et al. (2009) evaluated results by each worker's highest formaldehyde
concentration during a "peak" exposure event, by average intensity of exposure, by cumulative
exposure, and by duration of exposure. "Peak" exposure events were defined as short-term
exposures (<15 minutes) that exceed the TWA formaldehyde intensity fBeane Freeman et al..
2009). Workers' "peak" exposures were defined as the highest concentration among their "peak"
exposure events. Among only those workers with some "peak" exposure, the RR in the highest
category compared to the lowest category was 1.78 (95% CI 0.87, 3.64) with a trend p-value of 0.13
for the continuous values of the peak exposure data. While the investigators considered the lowest
group of exposed workers to be the most appropriate reference group (possibly due to a potential
for selection bias between exposed and unexposed workers), had the unexposed group been used
as the referent group, the RR would have been higher (~ RR of 2.17). This relationship between
This document is a draft for review purposes only and does not constitute Agency policy.
1-446 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
myeloid leukemia and high peak formaldehyde exposure is not only seen for the complete 2004
follow-up when the average length of follow-up was 42 years, but throughout the cohort experience
fsee Beane Freeman et al.. 2009. Figure 11. These plots show that during the 1970s and 1980s, the
RR >10 until about 1970 and then remained elevated between RR = 4 and RR = 6 until about 1980
and then between about RR = 2 and RR = 3 through the end of follow-up in 2004. Such a consistent
finding of a strong effect over many years of follow-up reduces the possibility that the results for
the full follow-up period could be due to chance. Beane Freeman et al. (20091 reported that among
all workers there was an exposure-response trend through follow-up in 2004 with p-value of 0.07
for the continuous values of the peak exposure data; and there was an exposure-response trend
through follow-up in 1994 with p-value of 0.0087.
Beane Freeman et al. (2009) also reported that among those with any formaldehyde
exposure in the 2004 follow-up, the RR in the highest category of average intensity of exposure was
1.61 (95% CI 0.76, 3.39) with little evidence of any trend for the continuous exposure data at nearly
40 years of follow-up (p = 0.40). However, the supplementary tables from Beane Freeman et al.
(20091 reported that for follow-up through 1994, the exposure-response trend value for all
workers was p = 0.11. No trend in RR was found for cumulative exposure (see Table 1-60). Overall,
the evidence from Beane Freeman et al. (2009) provides limited evidence of an exposure-response
relationship based on "peak" exposures.
Hauptmann et al. (2009) evaluated results by multiple metrics of exposure including
exposure duration, number of embalmings, cumulative exposure, average formaldehyde intensity
while embalming, time-weighted formaldehyde intensity, and peak exposure. Peak exposure levels
were defined as the maximum of moving averages of any series of measurements covering 15
minutes. Results for two different reference groups were reported, the first set from the authors'
Table 3 used unexposed people as the "a priori" reference group but as there was only one case of
myeloid leukemia in this group, the results were statistically unstable with wide Cis. Those results
showed an OR of 13.6 (95% CI 1.6,119.7) for the highest category of duration with a statistically
significant trend p-value of 0.020; and an OR of 9.5 (95% CI 1.1, 86.0) for the highest category of
average exposure; and an OR of 13.0 (95% CI 1.4,116.9) for the highest category of peak exposure.
The second set of results redefined the reference category as those people with fewer than 500
lifetime embalmings. Thus, this referent group includes some exposed individuals, which mutes the
categorical comparisons (i.e., this methodology causes bias toward the null and underestimates the
effect estimates) but allows for more statistically stable effect estimates as there were five cases of
myeloid leukemia in this reference group. Those results showed an OR of 3.9 (95% CI 1.2,12.5) for
the highest category of exposure duration, an OR of 2.3 (95% CI 0.7, 7.5) for the highest category of
average exposure, and an OR of 2.9 (95% CI 0.9, 9.5) for the highest category of peak exposure.
Hauptmann et al. (20091 assessed two methodologies to measure potential exposure-
response trends: (1) trends based on the complete range of continuous exposure metric data and
(2) trends based on the ordinal levels of the categories of the difference exposure metrics, with the
This document is a draft for review purposes only and does not constitute Agency policy.
1-447 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
former method selected a priori. There was a statistically significant positive exposure-response
trend for duration of formaldehyde exposure (p = 0.020) as well as a statistically significant positive
trend for peak exposures (p = 0.036) and the trend p-value for average formaldehyde exposure was
0.058. For the other metrics of exposure, the continuous exposure metric data trend p-values were
greater than 0.10. However, analyses using the ordinal levels of the exposure metrics also showed
trends for the TWA8 intensity (p = 0.021), the number of embalmings (p = 0.012) and for
cumulative exposure (p = 0.023). Table 1-59 provides a summary of the exposure-response trends
reported by Hauptmann et al. (2009). Beane Freeman et al. (2009). and Meyers et al. (2013)—all
three of which reported results that were judged to be of high confidence (see Table 1-60 and
Appendix A.5.9).
Table 1-59. Summary high confidence studies of reported exposure-response
trends describing the effect estimates of association between formaldehyde
exposure and risk of myeloid leukemia
High confidence studies reporting exposure-response trend assessments
Hauptmann et al. (2009)a
Beane Freeman et al. (2009)a
Meyers et al. (2013)a
Exposure metric
Continuous
Categorical
Continuous
2004 follow-up
Continuous
1994 follow-up
Continuous
Categorical
Duration
p = 0.020
NR
NR
NR
p = 0.30
NR
# of
Embalmings
p = 0.314
p = 0.012
NR
NR
NR
NR
Cumulative
p = 0.192
p = 0.023
p = 0.44
p = 0.171
NR
NR
Average
p = 0.058
NR
p = 0.40
p = 0.110
NR
NR
TWA8
p = 0.396
p = 0.021
NR
NR
NR
NR
Peak
p = 0.036
NR
p = 0.07
p = 0.0087
NR
NR
Abbreviations: TWA8 = 8-hour time-weighted average; NR = not reported.
formaldehyde exposure measured as a continuous variable among unexposed and exposed persons.
Coggon etal. (2014) classified workers' exposures according to the highest level of
exposure ever experienced, which can be interpreted as an indicator of peak occupational exposure
because each worker was assigned the highest exposure classification ever experienced, and
reported exposure-level specific results with an OR of 1.10 (95% CI 0.51, 2.38) for workers with
peak occupational exposure of low/moderate and an OR of 1.26 (95% CI 0.39, 4.08) for those
workers who had ever worked in a job with high exposures. Among the group with high exposures,
those with less than one year of employment at high exposure had an OR of 1.77 (95% CI 0.45, 7.03;
9 exposed cases) while those with 1 year or more at high exposure had an OR of 0.96 (95: CI: 0.24,
3.82; 4 exposed cases). The limitation of this study was the likelihood of nondifferential exposure
misclassification due to the quality of the exposure assessment and the lack of any latency analysis.
The expected impact is of a downward bias toward the null thereby muting any potential exposure-
This document is a draft for review purposes only and does not constitute Agency policy.
1-448 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
response. The evidence from Coggon et al. (20141. while potentially biased toward the null and
statistically unstable within the "high" exposure category (nine exposed cases), provided only weak
evidence of an exposure-response relationship with "peak exposure."
Blair et al. (2001) reported separate results for AML and CML by low and high intensity of
exposure although data were only available to examine exposure-response for CML. Blair et al.
(2001) reported an OR = 1.3 (95% CI 0.6, 3.1) for low exposure based on seven cases and an
OR = 2.9 (95% CI 0.3, 24.5) for high exposure based on one case. Given that that the OR in the high
exposure group was based on only one case, these results provided only weak evidence of an
exposure-response relationship.
Talibov et al. (2014) reported results across three levels of cumulative formaldehyde
exposure and showed some increasing risk with each increasing level of exposure from HR = 0.89
(95% CI: 0.81, 0.97) in the lowest group to HR = 0.92 (95% CI: 0.83,1.03) in the middle group and
HR = 1.17 (95% CI: 0.91,1.51) in the highest exposure group. The test for trend showing an
exposure-response had a p-value of 0.07. As with the other results classified with low confidence,
the limitation of this study was the likelihood of nondifferential exposure misclassification due to
the quality of the exposure assessment, which was based on decennial census records. The
expected impact is of a downward bias toward the null thereby muting any potential exposure-
response.
The evidence for an exposure-response relationship is most strongly supported by the
study of embalmers by Hauptmann et al. (2009). which reported statistically significant trends for
five of the six exposure metrics evaluated including duration of exposure, the number of
embalmings, cumulative exposure, average intensity of exposure, TWA8 exposure, and "peak"
exposure; and a borderline significant trend for the sixth exposure metric (average intensity of
exposure). Beane Freeman et al. (2009) reported a borderline significant exposure-response trend
for the measure of "peak" exposure that was shown to be statistically significant over the course of
more than 30 years of annual follow-up but which faded somewhat as the maturity of the cohort
approached 40 years of follow-up—a span of time that far exceeds the latency of all but a few
cancers such as mesothelioma. Meyers et al. (2013) also provided solid evidence of an exposure-
response relationship based on duration of exposure. Coggon et al. (2014). a medium confidence
study, found little evidence for an exposure-response relationship.
While it is not known which of these exposure metrics is of greatest biological relevance for
myeloid leukemia, all of the exposure metrics reflect different aspects of increased exposure to
formaldehyde and associations with increased risks of myeloid leukemia. As the different measures
of exposure are all likely to be correlated with each other, it may not be possible at this time to
single out one exposure metric as more biologically meaningful than another. It appears that these
various trend results reflect some true underlying exposure-response relationship.
Observations of exposure-response relationships are strong evidence in support of an
association consistent with causation (Hill. 19651 and against a spurious association because it
This document is a draft for review purposes only and does not constitute Agency policy.
1-449 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
would necessitate a third (uncontrolled) factor, which changes in the same manner (direction and
magnitude) as the exposure of interest fCDC. 20041 to explain away each of the reported exposure-
response relationships.
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias is an unlikely alternative explanation for the consistent evidence of increased
risk of myeloid leukemia in people exposed to formaldehyde. Selection bias is unlikely in the case-
control studies of myeloid leukemia as the case-control (Blair etal.. 2001) and nested case-control
studies fCoggon etal.. 2014: Hauptmann etal.. 20091 evaluated exposure status without regard to
outcome status and had participation levels of 77-99%. Each of the cohort studies (Coggonetal..
2014: Pira etal.. 2014: Talibovetal.. 2014: Meyers etal.. 2013: Saberi Hosniieh etal.. 2013: Beane
Freeman etal.. 2009: Hayes etal.. 1990: Ottetal.. 1989: Stroup etal.. 1986: Walrath and Fraumeni.
1984.1983) included at least 75% of eligible participants and lost fewer than 3% of participants
over the course of mortality follow-up.
Selection bias due to the comparison of a generally healthier group of workers to those in
the general population (called the healthy worker effect) could have obscured a truly larger effect
of formaldehyde exposure in analyses based on "external" comparisons with mortality in the
general population in one study with an SMR = 0.64 for "all cancers" fStroup etal.. 19861. but would
not influence analyses using "internal" or matched comparison groups fCoggon etal.. 2014: Meyers
etal.. 2013: Beane Freeman et al.. 2009: Hauptmann etal.. 2009: Blair etal.. 2001). The clearest
example of the potential influence of the healthy worker effect is shown in the comparison on
results from the study of garment workers (Meyers etal.. 2013). That study compared SMRs using
an external referent group based on the general population alongside standardized rate ratios
(SRR) using an internal referent group of workers in the lowest category of duration of exposure.
Compared to the general population (matched on sex, race, age, and calendar time), garment
workers with less than a 3-year duration of exposure had an SMR of 0.65 (95% CI 0.18,1.65), which
is a 35% lower risk of dying from myeloid leukemia than people in the general population. For
workers with a 3- to 9-year duration, the SMR was 1.46, and for workers with 10 or more years of
exposure, the SMR was 1.84. Internal comparisons were made by comparing the risk of dying from
myeloid leukemia in workers with 3-9 years of exposure to the risk among those with less than
3 years of exposure for an SRR of 2.12. The SRR for workers with 10 or more years of exposure was
3.25. Selection bias may explain why results based on comparisons of mortality of workers with
the general population are lower than comparisons of workers to workers. Selection bias does not
explain increased risks in exposed workers.
Information bias is an unlikely alternative explanation for the consistent evidence of
increased risk of myeloid leukemia in people exposed to formaldehyde. Information bias may
distort epidemiological findings when subjects' true exposures are inaccurately assigned at the
individual or group level. A differential misclassification, in which exposure status influences
disease classification by the investigator (or disease status influences exposure classification), can
This document is a draft for review purposes only and does not constitute Agency policy.
1-450 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
lead to spurious (i.e., "false positive") associations. However, information bias is considered
unlikely among these studies of myeloid leukemia mortality because the likelihood of differential
misclassification based on these study designs is low. The assignment of exposure status or
calculation of exposure measures in the cohort studies was done independent of knowledge of the
cause of death. In the nested case-control studies by Coggon et al. (2014) and Hauptmann et al.
(2009) the ascertainment of individual-level exposure levels was independent of the cause of death.
In the case-control study by Blair et al. (2001). many different occupational exposures were
evaluated based on interview data and subjects were unlikely to be aware of specific chemical
exposure of interest in the study. Therefore, an exposure-related recall bias of their occupational
histories is unlikely. The exposure assignments in Blair et al. (2001) were based on typical
exposure characteristics of the individual's job and were made blinded to case/control status.
There does not appear to be any evidence of confounding that would provide an alternative
explanation for the observed association of formaldehyde exposure with increased risk of myeloid
leukemia seen in these studies. Chemicals and other coexposures that have not been independently
associated with myeloid leukemia are not expected to confound results. However, other known
risk factors for myeloid leukemia include exposure to benzene, ionizing radiation, and smoking.
Benzene is not used in the embalming process fStewartetal.. 1992: Hayes etal.. 19901 and was not
a chemical coexposure in the garment plants fStavner et al.. 19851. and consequently, could not be a
confounder of those results. Benzene was evaluated by Ott et al. (1989) and not found to be a risk
factor (OR = 1.0), and thus, could not be a confounder. Benzene was specifically assessed as a
potential confounder among the U.S. industrial workers (Beane Freeman et al.. 20091 and found not
to be a confounder. Ionizing radiation can be a coexposure for embalmers but the limited extent of
such radiation exposure is unlikely to explain the observed association in embalmers (Hauptmann
etal.. 20091. Exposures to ionizing radiation were not mentioned as coexposures for the industrial
workers or the garment workers, and would not be expected to be correlated with their
formaldehyde exposures. Smoking was controlled for in the analyses of the embalmers
(Hauptmann et al.. 2009). which demonstrated a statistically significant exposure-response relation
between both duration of formaldehyde exposure and peak exposures with increased risk of death
from myeloid leukemia. Blair et al. (2001) also controlled for smoking in their analyses thereby
reducing the likelihood of confounding by smoking. Smoking was not evaluated as a potential
confounder in the industrial or garment worker cohorts f Coggon etal.. 2014: Meyers etal.. 2013:
Beane Freeman et al.. 20091. However, there is no evidence that smoking rates in the industrial or
garment worker cohorts (Meyers etal.. 2013: Beane Freeman etal.. 2009) were correlated with
formaldehyde exposures—a necessary condition for potential confounding. Moreover, the internal
comparisons used in the analyses of the industrial cohort should mitigate any potential
confounding effects of smoking because smoking rates within a cohort are likely to be more similar
than compared to the general population.
This document is a draft for review purposes only and does not constitute Agency policy.
1-451 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
Consistency across multiple studies is demonstrated by a pattern of increased risk in
different populations, exposure scenarios, and time periods. Such consistency makes unmeasured
confounding an unlikely alternative explanation for the observed associations. This consistency
also reduces the likelihood of chance as an alternative explanation. The observations of
exposure-response trends similarly reduce the likelihood that chance, confounding, or other biases
can explain the observed association.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
myeloid leukemia placed the greatest weight on five particular considerations: (1) the generally
consistent increases in risk observed across a set of high and medium confidence independent
results from epidemiology studies of occupational formaldehyde levels using varied study designs
and populations; (2) the strength of the association showing a 1.5- to 3-fold increase in risk in
studies with higher quality exposure assessment; (3) the reported exposure-response relationships
showing that increased exposure to formaldehyde were associated with increased risk of dying
from myeloid leukemia; (4) a biologically coherent temporal relationship consistent with a pattern
of exposure to formaldehyde and subsequent death from myeloid leukemia allowing time for
cancer induction, latency, and mortality; and (5) reasonable confidence that alternative
explanations are ruled out, including chance, bias, and confounding within individual studies or
across studies. Consistent observations of genotoxicity in peripheral blood lymphocytes across
several occupational studies involving diverse exposure settings further supports the evidence in
humans, as does evidence of perturbations to immune cell populations in peripheral blood with
formaldehyde exposure.
Conclusion
• The available epidemiological studies provide robust evidence of an association consistent
with causation between formaldehyde exposure and increased risk of myeloid leukemia.
This document is a draft for review purposes only and does not constitute Agency policy.
1-452 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Population-
level exposure
assessment
Individual-level exposure assessment
100-i
a>
*-»
TO
E
o
a>
a:
10-
3-
2-
1-
0.1-
"(35
5
a l
Hauptmann
Beane Freeman/
Checkoway
s *
± cr
i i
I 1
c o g
I
Q
1
5 2> 5>
O X if X
Meyers
a a
¦a ^
Blair
Coggon
+
4
Studies of Anatomists
and Embalmers
J L
Industrial Workers
Garment
Workers
00 (N ,—. .—. .—.
_ „ 8 „ „
General
Population
- 8
CO CO CO CM
Figure 1-37. All epidemiological studies reporting myeloid leukemia risk
estimates.
Results specifically for acute myeloid leukemia (AML) or chronic myeloid leukemia (CML) are noted by
these abbreviations: SMR = standardized mortality ratio; PMR = proportionate mortality ratio;
RR = relative risk; OR = odds ratio. For each measure of association, the number of exposed cases is
provided in brackets (e.g., [n = 3]). For studies reporting results on multiple metrics of exposure, each
metric is included; however, only the highest category of each exposure metric is presented in the figure.
*The dotted line extending from Hauptmann et al. (2009) reflects that study's inclusion of the original
cohorts from Walrath and Fraumeni (1984,1983) and Hayes et al. (1990), which were combined with
extended follow-up in Hauptmann etal. (2009) in a nested case-control study with internal referents.
This document is a draft for review purposes only and does not constitute Agency policy.
1-453 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
High confidence results
Medium
confidence
results
100n
a>
*¦»
to
£
tf)
LU
o
£
LU
Q)
C£
10-
3-
2-
1-
Hauptmann
Beane Freeman
Meyers
Coggon
0.1-
— ^ 0> T- o
o
o
O)
8
c
c
c
c
CM
to
CT)
T
&
o
(N
en
II
on
II
oz
en
CO
CO
CO
CO
**
~
Figure 1-38. High and medium confidence epidemiological studies reporting
myeloid leukemia risk estimates.
For each measure of association, the number of exposed cases is provided in brackets (e.g., [n = 14]). For
studies reporting results on multiple metrics of exposure, each metric is included; however, only the
highest category of each exposure metric is presented in the figure. Abbreviations: OR = odds ratio;
RR = relative risk; SMR = standardized mortality ratio; HR = hazard ratio.
This document is a draft for review purposes only and does not constitute Agency policy.
1-454 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
High confidence results
100 -i
0
•*->
TO
£
if)
LU
4-»
o
o
0
>
"5
re
0
a:
10-
3
2
1H
0.1-
i2
I
15
E
O
CD
8
<
l
Meyers
Hauptmann
o
—r
Low confidence results
i2
Industrial
General
Embalmers
J L
Garment Workers
14]
—
T
II
c
II
c
II
C
a
IT)
K
8
ii
UL
CO
ii
cr
s
cn
II
CC
S
C/5
~
~
r~
/ .j* j*
^ <$? #
Figure 1-39. Epidemiological studies reporting acute myeloid leukemia risk
estimates.
For each measure of association, the number of exposed cases is provided in brackets (e.g., [n = 8]). For
studies reporting results on multiple metrics of exposure, each metric is included; however, only the
highest category of each exposure metric is presented in the figure. Abbreviations: OR = odds ratio;
RR = relative risk; SMR = standardized mortality ratio; HR = hazard ratio.
Table 1-60. Epidemiological studies of formaldehyde exposure and risk of
myeloid leukemia
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Beane Freeman et al.
(2009) with supplemental online tables.
Exposure assessment: Individual-level
exposure estimates based on job
titles, tasks, visits to plants by study
Internal comparisons:
Peak exposure:
1980 follow-up:
This document is a draft for review purposes only and does not constitute Agency policy.
1-455 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Population: 25,619 workers employed
at 10 formaldehyde-using or
formaldehyde-producing plants in the
U.S., followed from either the plant
start-up or first employment through
2004. Deaths were identified from the
National Death Index with remainder
assumed to be living. Vital status was
97.4% complete and only 2.6% lost to
follow-up.
Outcome definition: Death certificates
used to determine UCOD from myeloid
leukemia (ICD-8: 205).
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in lowest
exposed category. Lagged exposures
were evaluated to account for cancer
latency.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Blair etal. (1986)
Hauptmann et al. (2003)
Confidence in effect estimates:3
SB IB €f Oth
Overall
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A
Exposures
industrial hygienists who took 2,000
air samples from representative jobs,
and monitoring data from 1960
through 1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3).
Median cumulative
exposure = 0.6 ppm-years (range 0-
107.4).
Multiple exposure metrics including
peak, average, and cumulative
exposures were evaluated using
categorical and continuous data.
Duration and timing: Exposure period
from before 1946 through 1980.
Median length of follow-up: 42 years.
Duration and timing since first
exposure were evaluated.
Variation in exposure:
For all variations in exposure:
Level 1 (unexposed)
Peak exposure:
Level 2 (>0 to <2.0 ppm)
Level 3 (2.0 to <4.0 ppm)
Level 4 (>4.0 ppm)
Average intensity:
Level 2 (>0 to <0.5 ppm)
Level 3 (0.5 to <1.0 ppm)
Level 4 (>1.0 ppm)
Cumulative exposure:
Level 2 (>0 to <1.5 ppm-yrs)
Level 3 (1.5 to <5.5 ppm-yrs)
Level 4 (>5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders
and found not be confounders.
[As noted in Appendix A.5.9: There
was no information on smoking;
however, according to Blair et al.
(1986). "The lack of a consistent
elevation for tobacco-related causes
of death, however, suggests that the
smoking habits among this cohort did
not differ substantially from those of
the general population."
Results: effect estimate (95% CI)
[# of cases]
Highest peak RR = 3.92 (0.78-19.67)
(p-trend = 0.12)
1994 follow-up:
Highest peak RR = 2.79 (1.08-7.21)
(p-trend = 0.02)
2004 follow-up:
Level 1 RR = 0.82 (0.25-2.67) [4]
Level 2 RR = 1.00 (Ref. value) [14]
Level 3 RR = 1.30 (0.58-2.92) [11]
Level 4 RR = 1.78 (0.87-3.64) [19]
p-trend (exposed) = 0.13;
p-trend (all) = 0.07
[4]
[24]
[9]
[11]
Cumulative exposure:
Level 1 RR = 0.61 (0.20-1.91) [4]
Level 2 RR = 1.00 (Ref. value) [26]
Level 3 RR = 0.82 (0.36-1.83) [8]
Level 4 RR = 1.02 (0.48-2.16) [10]
p-trend (exposed) >0.50;
p-trend (all) = 0.44
Duration of exposure:
No evidence of association (data not
shown).
Time since first exposure:
>0-15 yrs RR = 1.00 (Ref. value) [3]
>15-25 yrs RR = 2.44 (0.45-13.25) [11]
>25-35 yrs RR = 0.77 (0.11-5.24) [8]
>35 yrs RR = 0.67 (0.09-4.88) [24]
External comparisons:
SMRunexposed =0.65(0.25-1.74) [4]
SMRexposed =0.90(0.67-1.21) [44]
Average intensity:
Level 1 RR = 0.70 (0.23-2.16)
Level 2 RR = 1.00 (Ref. value)
Level 3 RR = 1.21 (0.56-2.62)
Level 4 RR = 1.61 (0.76-3.39)
p-trend (exposed) = 0.43;
p-trend (all) = 0.40
This document is a draft for review purposes only and does not constitute Agency policy.
1-456 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Beane Freeman et al.
(2009) as re-analyzed by
Checkoway et al. (2015) with
differences noted.
Population: No differences.
Outcome definition: Death certificates
used to determine UCOD from acute
and chronic myeloid leukemia (ICD-8:
205.0 and 205.1).
Design: No differences.
Analysis: HRs estimated using Cox
proportional hazards models controlling
for age, sex, and race; adjusted for pay
category compared to workers in the
redefined lowest exposed category. Did
not control for calendar year as did
Beane Freeman et al. (2009). Lagged
exposures were evaluated to account
for cancer latency.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Blair et al. (1986)
Hauptmann et al. (2003)
Checkoway et al. (2015)
[reviewed here]
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW • (Potential bias \|/)
IB: Exposure Group A (from Beane
Freeman et al., 2009) downgraded to
Group D based on authors' decision to
reclassify all peak exposures <2 ppm as
unexposed and to reclassify peak
Exposures
Beane Freeman et al. (2013)
reported that among a sample of 379
cohort members, they "found no
differences in prevalence of smoking
by level of formaldehyde exposure."]
Exposure assessment: No differences
in measurements; however, the
exposure metrics we redefined.
Redefined peak exposures as having
"at least one continuous month of
employment in jobs identified in the
original exposure characterization as
likely having short-term exposure
excursions of 2 ppm or more to less
than 4 ppm or 4 ppm or more on a
weekly or daily basis."
Redefinition of peak exposures
excluded "employment in jobs likely
experiencing (1) short-term
excursions more than 0 ppm and less
than 2 ppm; (2) short-term excursions
identified as occurring as frequently
as hourly; and (3) short-term
excursions identified as occurring as
infrequently as monthly."
Duration and timing: No differences.
Variation in exposure:
For all variations in exposure:
Peak exposure:
Level 1 (exposed to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Did not evaluate
Cumulative exposure:
Level 1 (exposed to <0.5 ppm-yrs)
Level 2 (>0.5 to <2.5 ppm-yrs)
Level 3 (>2.5 to <5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders by
Beane Freeman et al. (2009) and
found not be confounders.
Checkoway et al. (2015) did
not re-evaluate potential
confounding.
Results: effect estimate (95% CI)
[# of cases]
Internal comparisons:
Myeloid Leukemia
Peak exposure:
Level 1 HR=1.00 (Ref. value) [27]
Level 2 HR=2.09 (1.03-4.26) [11]
Level 3 HR=1.80 (0.85-3.79) [10]
p-trend = 0.06
Cumulative exposure:
Level 1 HR=1.00(Ref. value) [23]
Level 2 HR=0.98 (0.47-2.03) [11]
Level 3 HR=0.94 (0.47-1.86) [14]
p-trend = 0.90
AML
Peak exposure:
Level 1 HR=1.00(Ref. value) [21]
Level 2 HR=1.71 (0.72-4.07) [7]
Level 3 HR=1.43 (0.56-3.63) [6]
p-trend = 0.31
Cumulative exposure:
Level 1 HR=1.00(Ref. value) [17]
Level 2 HR=0.87 (0.36-2.12) [7]
Level 3 HR=0.96 (0.43-2.16) [10]
p-trend = 0.90
CML
Peak exposure:
Level 1 HR=1.00 (Ref. value) [6]
Level 2 HR=2.62 (0.64-10.66) [3]
Level 3 HR=3.07 (0.83-11.40) [4]
p-trend = 0.07
Cumulative exposure:
Level 1 HR=1.00 (Ref. value) [6]
Level 2 HR=0.97 (0.24-3.93) [3]
Level 3 HR=0.92 (0.25-3.36) [4]
p-trend = 0.90
This document is a draft for review purposes only and does not constitute Agency policy.
1-457 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
exposures >2 ppm as unexposed—if
they were either very rare or very
common.
Reference: Hauptmann et al. (2009)
Population: 6,808 embalmers and
funeral directors who died during
1960-1986. Identified from registries of
the National Funeral Directors'
Association, licensing boards and state
funeral directors' associations, NY State
Bureau of Funeral Directors, and CA
Funeral Directors and Embalmers.
Deaths were identified from the
National Death Index. Next of kin
interviews conducted for 96% of cases
and 94% of controls.
Outcome definition: Death certificates
used to determine UCOD from myeloid
leukemia (ICD-8: 205).
Design: Nested case-control study
within a prospective cohort mortality
study using two internal comparison
groups; the first composed of those
who had never embalmed (1 case and
55 controls) and the second composed
of those who had fewer than 500
embalmings (five cases and 83 controls).
Analysis: ORs calculated using
unconditional logistic regression
adjusted for date of birth, age at death,
sex, data source, and smoking. Lagged
exposures were evaluated to account
for cancer latency. These results are
shown in table 3 of Hauptmann et al.
(2009).
Results from the second internal
comparison group with <500
embalmings were selected to increase
statistical stability. These results are
shown in table 4 of Hauptmann et al.
(2009)
Related studies:
Haves et al. (1990)
Walrath and Fraumeni (1983)
Walrath and Fraumeni (1984)
Note: The original cohorts from these
three original studies were combined in
Hauptmann et al. (2009) and follow-
Exposures
Exposure assessment: Occupational
history obtained by interviews with
next of kin and coworkers using
detailed questionnaires. Exposure
was assessed by linking questionnaire
responses to an exposure assessment
experiment providing measured
exposure data. Exposure levels (peak,
intensity, and cumulative) were
assigned to each individual using a
predictive model based on the
exposure data. The model explained
74% of the observed variability in
exposure measurements.
Multiple exposure metrics including
duration (mean = 33.1 yrs in cases), #
of embalming, peak, average, and
cumulative exposures were evaluated
using categorical and continuous
data.
Duration and timing: Exposure period
from <1932 through 1986. Duration
of exposure was evaluated. Duration
is also a surrogate for time because
first exposure since dates of death
was closely related to cessation of
workplace exposures.
Variation in exposure:
For variations in exposure from
table 3 of the publication:
Level 1 (no exposure to
embalming)
For variations in exposure from
table 4 of the publication:
Level 1 (<500 embalming)
Duration of exposure:
Level 2 (<20 years)
Level 3 (20-34 years)
Level 4 (>34 years)
Number of embalming:
Level 2 (500-1,422)
Level 3 (1,423-3,068)
Level 4 (>3,068)
Cumulative exposure:
Level 2 (<4,058 ppm-hrs)
Level 3 (4,059-9,253 ppm-hrs)
Results: effect estimate (95% CI)
[# of cases]
Internal comparisons {from table 3 in the
paper):
Never embalming: OR = 1.00 (Ref. value)
[1]
Ever embalming: OR = 11.2 (1.3-95.6)
[33]
Duration of exposure:
Level 1 OR = 1.00 (Ref. value)
[1]
Level 2 OR = 5.0 (0.5-51.6)
[6]
Level 3 OR = 12.9 (1.4-117.1)
[13]
Level 4 OR = 13.6 (1.6-119.7)
[14]
Number of embalming:
Level 1 OR = 1.0 (Ref. value) [1]
Level 2 OR = 7.6 (0.8-73.5) [7]
Level 3 OR = 12.7 (1.4-116.7) [12]
Level 4 OR = 12.7 (1.4-112.8) [14]
Cumulative exposure:
Level 1 OR = 1.0 (Ref. value) [1]
Level 2 OR = 10.2 (1.1-95.6) [9]
Level 3 OR = 9.4 (1.0-85.7) [10]
Level 4 OR = 13.2 (1.5-115.4) [14]
Average intensity (while embalming):
Level 1 OR = 1.0 (Ref. value) [1]
Level 2 OR = 11.1 (1.2-106.3) [10]
Level 3 OR = 14.8 (1.6-136.9) [13]
Level 4 OR = 9.5 (1.1-86.0)
[10]
TWA8 formaldehyde intensity:
Level 1 OR = 1.0 (Ref. value) [1]
Level 2 OR = 8.4 (0.8-79.3) [8]
Level 3 OR = 13.6 (1.5-125.8) [13]
Level 4 OR = 12.0 (1.3-107.4) [12]
Peak exposure:
Level 1 OR = 1.0 (Ref. value) [1]
Level 2 OR = 15.2 (1.6-141.6) [12]
Level 3 OR = 8.0 (0.9-74.0)
[9]
Level 4 OR = 13.0 (1.4-116.9) [12]
Internal comparisons (from table 4):
Duration of exposure:
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 0.5 (0.1-2.9) [2]
Level 3 OR = 3.2 (1.0-10.1) [13]
Level 4 OR = 3.9 (1.2-12.5) [14]
Number of embalming:
This document is a draft for review purposes only and does not constitute Agency policy.
1-458 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
up was extended so the case-series
overlap and are not independent.
However, the three original cohorts
used external reference groups for
comparison while Hauptmann et al.
(2009) selected internal controls, which
were independent of the reference
groups used in the original studies.
Confidence in effect estimates:3
SB IB Cf Oth
Ove rail
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A
Exposures
Level 4 (>9253 ppm-hrs)
Average intensity (while embalming):
Level 2 (<1.4 ppm)
Level 3 (>1.4-1.9 ppm)
Level 4 (>1.9 ppm)
TWA8 formaldehyde intensity:
Level 2 (<0.10 ppm)
Level 3 (>0.10-0.18 ppm)
Level 4 (>0.18 ppm)
Peak exposure:
Level 2 (<7.0 ppm)
Level 3 (7.0 to <9.3 ppm)
Level 4 (>9.3 ppm)
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Chemical coexposures are not known
risk factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Results: effect estimate (95% CI)
[# of cases]
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 1.2 (0.3-5.5) [3]
Level 3 OR = 2.9 (0.9-9.1) [12]
Level 4 OR = 3.0 (1.0-9.2) [14]
Cumulative exposure:
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 2.1 (0.5-8.1) [5]
Level 3 OR = 2.2 (0.7-7.1) [10]
Level 4 OR = 3.1 (1.0-9.6) [14]
Average intensity (while embalming):
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 2.6 (0.8-8.7) [10]
Level 3 OR = 2.8 (0.8-9.1) [10]
Level 4 OR = 2.3 (0.7-7.5) [9]
TWA8 formaldehyde intensity:
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 2.4 (0.7-8.2) [8]
Level 3 OR = 2.6 (0.8-8.7) [10]
Level 4 OR = 2.6 (0.8-8.3) [11]
Internal comparisons (from table 4):
Peak exposure:
Level 1 OR = 1.0 (Ref. value) [5]
Level 2 OR = 2.9 (0.9-9.8) [9]
Level 3 OR = 2.0 (0.6-6.6) [9]
Level 4 OR = 2.9 (0.9-9.5) [11]
Additional: Acute ML (ICD-8: 205.0)
Internal comparisons (from table 4):
Duration of exposure:
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 0.4 (0.04^.9) [1]
Level 3 OR = 2.9 (0.7-12.2) [8]
Level 4 OR = 3.1 (0.7-13.7) [8]
Number of embalming:
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 no cases
Level 3 OR = 2.9 (0.7-12.0) [8]
Level 4 OR = 2.9 (0.7-11.6) [9]
Cumulative exposure:
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 1.3 (0.2-9.4) [2]
Level 3 OR = 1.9 (0.4-8.2) [6]
Level 4 OR = 3.2 (0.8-13.1) [9]
Average intensity (while embalming):
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 2.5 (0.6-10.9) [6]
Level 3 OR = 2.0 (0.4-9.4) [5]
Level 4 OR = 2.3 (0.5-10.3) [6]
TWA8 formaldehyde intensity:
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 1.4 (0.3-7.8) [3]
Level 3 OR = 2.6 (0.6-11.4) [7]
This document is a draft for review purposes only and does not constitute Agency policy.
1-459 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Meyers et al. (2013)
Population: 11,043 workers in three
U.S. garment plants exposed for at least
3 months. Women comprised 82% of
the cohort. Vital status was followed
through 2008 with 99.7% completion.
Outcome definition: Death certificates
used to determine both the UCOD from
myeloid leukemia (ICD code in use at
time of death).
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates. SRRs calculated
using LTAS.NET. Rate ratios calculated
using Poisson regression analysis based
on internal referents.
Related studies:
Stavner et al. (1985)
Stavner et al. (1988)
Pinkerton et al. (2004)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A
Exposures
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
1984 with 12-73 within each
department. Formaldehyde levels
across all departments and facilities
were similar. Geometric TWA8
exposures ranged from 0.09-
0.20 ppm. Overall geometric mean
concentration of formaldehyde was
0.15 ppm, (GSD 1.90 ppm). Area
measures showed constant levels
without peaks. Historically earlier
exposures may have been
substantially higher.
Duration and timing: Exposure period
from 1955 through 1983. Median
duration of exposure was 3.3 years.
More than 40% exposures <1963.
Median time since first exposure was
39.4 years. Duration and timing since
first exposure were evaluated.
Variation in exposure:
Duration of exposure:
Level 1 (<3 years)
Level 2 (3-9 years)
Level 3 (10 + years)
Time since first exposure:
Level 1 (<10 years)
Level 2 (10-19 years)
Level 3 (20 + years)
Duration of exposure (Poisson
modeling-lagged 2 years):
Level 1 (<1.6 years)
Level 2 (1.6 to <6.5 years)
Level 3 (6.5 to <16 years)
Level 4 (16 to <19 years)
Level 5 (19 + years)
Results: effect estimate (95% CI)
[# of cases]
2.6 (0.6-11.3) [7]
Level 1 OR = 1.0 (Ref. value) [3]
Level 2 OR = 1.8 (0.4-9.3) [4]
Level 3 OR = 2.1 (0.5-9.2) [5]
Level 4 OR = 2.9 (0.7-12.5) [7]
External comparisons:
SMR = 1.28 (0.79-1.96) [21]
[4]
[7]
[10]
Level 1 SMR = 0.90 (0.02-4.99) [1]
Level 2 SMR = 0.40 (0.01-2.21) [1]
Level 3 SMR = 1.49 (0.90-2.32) [19]
Year of first exposure:
<1963 SMR = 1.37 (0.75-2.30) [14]
1963-1970 SMR = 1.13 (0.37-2.63) [5]
1971+ SMR = 1.15 (0.14-4.17) [2]
Internal comparisons:
Duration of exposure:
Level 1 SRR = 1.00 (Ref. value) [4]
Level 2 SRR = 2.12 (0.57-7.85) [7]
Level 3 SRR = 3.25 (0.84-12.63) [10]
Duration of exposure (Poisson modeling-
lagged 2 years) [# of cases not given]:
Level 1 rate ratio = 1.00 (Ref. value)
Level 2 rate ratio = 1.38 (0.39-5.51)
Level 3 rate ratio = 0.43 (0.06-2.39)
Level 4 rate ratio = 6.42 (1.40-32.2)
Level 5 rate ratio = 1.71 (0.25-11.0)
SMR = 1.22 (0.67-2.05) [14]
SMR = 1.35 (0.44-3.15) [5]
Acute myeloid leukemia (ICD: 205.0)
Coexposures: Study population
specifically selected because
industrial hygiene surveys at the
plants did not identify any chemical
exposures other than formaldehyde
that were likely to influence findings.
Internal comparisons:
Duration of exposure:
Level 1 SMR = 0.46 (0.06-1.68) [2]
Level 2 SMR = 1.52 (0.49-3.56) [5]
Level 3 SMR = 1.81 (0.73-3.73) [7]
Time since first exposure:
Level 1 SMR = 0(0.00-6.66) [0]
Additional:
Acute myeloid leukemia (ICD: 205.0)
Chronic myeloid leukemia (ICD: 205.1)
Level 4 OR =
Peak exposure:
Within-study external comparisons:
Duration of exposure:
Level 1 SMR = 0.65 (0.18-1.65)
Level 2 SMR = 1.46 (0.59-3.02)
Level 3 SMR = 1.84 (0.88-3.28)
TSFE:
This document is a draft for review purposes only and does not constitute Agency policy.
1-460 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Coggon et al. (2014)
Population: 14,008 British men
employed in six chemical industry
factories which produced
formaldehyde. Cohort mortality
followed from 1941 through 2012.
Cause of deaths was known for 99% of
5,185 deaths through 2000. Similar
cause of death information not
provided on 7,378 deaths through 2012.
Vital status was 98.9% complete and
only 1.1% lost to follow-up through
2003. Similar information not provided
on deaths through 2012.
Outcome definition: Death certificates
used to determine cause of deaths from
myeloid leukemia (ICD-9: 205).
Design: Cohort mortality study with
external comparison group with a
nested case-control study.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM -i,
(Potential bias toward the nullvU)
IB: Exposure is Group B; lack of latency
analysis
Reference: Haves et al. (1990)
Exposures
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982.
Duration was evaluated as more, or
less, than one year only among the
high exposure group. Timing since
first exposure was not evaluated.
Variation in exposure:
Highest exposure level attained
Level 1 (Background)
Level 2 (low/moderate)
Level 3 (High)
Duration of "High" exposures
Level 1 (Background)
Level 2 (<1 year)
Level 3 (1 year or more)
Coexposures: Not evaluated as
potential confounders. Potential low-
level exposure to stvrene. ethylene
oxide, epichlorhydrin, solvents,
asbestos, chromium salts, and
cadmium; explanation for
underlining:
[As noted in Appendix A.5.9: Stvrene
is associated with LHP cancers.
Asbestos is associated with URT
cancers, but not with LHP cancers.
Other coexposures are not known risk
factors for this outcome.
Authors stated that the extent of
coexposures was expected to be low.
Potential for confounding may be
mitigated by low coexposures.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
Results: effect estimate (95% CI)
[# of cases]
Level 2 SMR = 0 (0.00-2.32) [0]
Level 3 SMR = 1.50 (0.82-2.52) [14]
Year of first exposure:
<1963 SMR = 1.55 (0.77-2.77) [11]
1963-1970 SMR = 0.64 (0.08-2.30) [2]
1971+ SMR = 0.83 (0.02-4.60) [1]
[36]
[12]
[16]
[8]
[17]
[19]
[9]
[17]
[5]
[4]
External comparisons:
PMR = 1.57 (1.01-2.34) [24]
External comparisons:
SMR = 1.20 (0.84-1.66)
Within-study external comparisons:
Highest exposure level attained
Level 1 SMR = 1.16 (0.60-2.02
Level 2 SMR = 1.46 (0.84-2.3J
Level 3 SMR = 0.93 (0.40-1.82
Internal comparisons:
Highest exposure level attained
Level 1 OR = 1.00 (Ref. value)
Level 2 OR = 1.10 (0.51-2.38)
Level 3 OR = 1.26 (0.39-4.08)
Duration of high exposures
Level 1 OR = 1.00 (Ref. value)
Level 1 OR = 1.77 (0.45-7.03)
Level 2 OR = 0.96 (0.24-3.82)
This document is a draft for review purposes only and does not constitute Agency policy.
1-461 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Population: 4,046 deceased U.S. male
embalmers and funeral directors,
derived from licensing boards and
funeral director associations in 32 states
and the District of Columbia who died
during 1975-1985. Death certificates
obtained for 79% of potential study
subjects (n = 6,651) with vital status
unknown for 21%.
Outcome definition: Death certificates
and licensing boards used to determine
cause of death from myeloid leukemia
(ICD-8: 205).
Design: Proportionate mortality cohort
study with external comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM -i,
(Potential bias toward the nullvU)
SB: Missing death certificates
considered to missing at random
IB: Exposure: Group A; latency not
evaluated
Reference: Walrath and Fraumeni
(1984)
Population: 1,007 deceased white male
embalmers from the California Bureau
of Funeral Directing and Embalming
who died during 1925-1980. Death
certificates obtained for all.
Outcome definition: Myeloid leukemia
(ICD-8: 205) listed as cause of death on
death certificates.
Design: Proportionate mortality cohort
study with external comparison group.
Exposures
fixative. Exposure based on
occupation which was confirmed on
death certificate. Authors
subsequently measured personal
embalming exposures ranging from
0.98 ppm (high ventilation) to
3.99 ppm (low ventilation) with peaks
up to 20 ppm.
Authors state that major exposures
are to formaldehyde and possibly
glutaraldehyde and phenol.
Duration and timing: Occupational
exposure preceding death during
1975-1985. Of 115 deaths from LHP
cancer, 66 (57%) were aged 60-
74 years. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Chemical coexposures are not known
risk factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure preceding death during
1916-1978. Birth year ranged from
1847 through 1959. Median age of
death was 62 years. Most deaths
were among embalmers with active
licenses. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
Results: effect estimate (95% CI)
[# of cases]
PMR = 1.52 (0.85-2.52) [# not given]
Chronic myeloid leukemia (ICD-8: 205.1)
PMR = 1.84 (0.79-3.62) [# not given]
External comparisons:
Observed: 8 myeloid leukemia deaths
(including 2 acute monocytic leukemia)
Expected: 4.3 myeloid leukemia deaths
(including 0.3 acute monocytic leukemia)
PMR = 1.86 (0.86-3.53)+ [8]
Additional:
Observed: 6 acute myeloid leukemia deaths
(including 2 acute monocytic leukemia)
Expected: With 4.3 myeloid leukemia
deaths expected, EPA used data from
Selvin et al. (1983) on the expected
ratio of AMLCML (2.2:1) among males ages
Additional:
Acute myeloid leukemia (ICD-8: 205.0)
This document is a draft for review purposes only and does not constitute Agency policy.
1-462 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Ove rail
Confidence
Medium
MEDIUM -i,
(Potential bias toward the nullvU)
IB: Exposure Group A; latency was not
evaluated
Reference: Walrath and Fraumeni
(1983)
Population: 1,132 deceased white male
embalmers licensed to practice during
1902-1980 in New York who died
during 1925-1980 identified from
registration files. Death certificates
obtained for 75% of potential study
subjects (n = 1,678).
Outcome definition: Myeloid leukemia
(ICD-8: 205) listed as cause of death on
death certificates.
Design: Proportionate mortality cohort
study with external comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
numbers of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Ove rail
Confidence
Medium
MEDIUM -i,
(Potential bias toward the nullvU)
SB: Missing death certificates
considered to missing at random
IB: Exposure Group A; latency was not
evaluated
Reference: Talibov et al. (2014)
Population: Individuals from Finland,
Iceland, Norway, and Sweden who were
Exposures
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing:
Occupational exposure preceding
death during 1902-1980. Median
year of birth was 1901. Median year
of initial license was 1931. Median
age at death was 1968. Expected
median duration of exposure was
37 years. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Exposure assessment: Occupational
history from census records were
linked to the Nordic Occupational
Cancer Study (NOCCA) JEM to code
Results: effect estimate (95% CI)
[# of cases]
25+ to estimate 2.96 expected cases of AML
out of the 4.3 expected myeloid leukemia
deaths.
PMR = 2.03 (0.82-4.22)+ [6]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice, 1979)
External comparisons:
Observed: 7 myeloid leukemia deaths
(including 1 acute monocytic leukemia)
Expected: 4.4 myeloid leukemia deaths
(including 0.3 acute monocytic leukemia)
PMR = 1.59 (0.70-3.15)+ [7]
Additional:
Observed: 6 acute myeloid leukemia deaths
(including 1 acute monocytic leukemia)
Expected: With 4.4 myeloid leukemia
deaths expected, EPA used data from
Selvin et al. (1983) on the expected ratio
of AMLCML (2.2:1) among males ages 25+
to estimate 3.03 expected cases of AML out
of the 4.4 expected myeloid leukemia
deaths.
PMR = 1.98 (0.80-4.12)+ [6]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice, 1979)
Internal comparisons:
Acute Myeloid Leukemia (ICD-9: 205.0)
Level 1 OR = 1.00 (ref value) [13781]
Level 2 OR = 0.89 (0.81-0.97) [580]
Acute myeloid leukemia (ICD-8: 205.0)
Acute myeloid leukemia (ICD-8: 205.0)
This document is a draft for review purposes only and does not constitute Agency policy.
1-463 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
recorded in various censuses from 1960
to 1990. Acute myeloid leukemia cases
identified by national registries up until
2003-2005 depending on the country.
Outcome definition: Diagnosis of
incident cancer reported to the National
Cancer Registries.
Design: Multicountry case-control study.
Analysis: HRs calculated for categories
of cumulative formaldehyde exposure
using conditional logistic regression
controlling for year of birth, sex,
country, solvents and other
coexposures. A 10-year latency period
was assumed.
Confidence in effect estimates:3
SB IE Cf Oth
Overall
Confidence
Low
LOW -i,
(Potential bii
IB: Exposure
Exposures
each cohort member as exposed to
formaldehyde. Exposures were
quantified based on the proportion of
people in each occupation considered
to be exposed and the mean level of
exposure during specific periods.
Coexposures to solvents was
evaluated.
Duration and timing: Exposure period
based on occupational histories prior
to 1983. Duration and timing since
first exposure were considered in the
exposure metric but were not
evaluated separated.
Variation in exposure:
Cumulative exposure:
Level 1 (unexposed)
Level 2 (low): <0.171 ppm-yrs
Level 3 (moderate): 0.171-1.6 ppm-
yrs
Level 4 (high): >1.6 ppm-yrs
Results: effect estimate (95% CI)
[# of cases]
Level 3 OR = 0.92 (0.83-1.03) [485]
Level 4 OR = 1.17 (0.91-1.51) [136]
p-trend = 0.07
s toward the nullvU)
Sroup D
Coexposures: Solvents and
coexposures controlled for in
multivariate models included:
aliphatic and alicyclic hydrocarbons,
aromatic hydrocarbons, benzene.
toluene, trichloroethylene, 111-
trichloroethane, methylene chloride,
perchloroethvlene. other organic
solvents, and ionizing radiation.
This document is a draft for review purposes only and does not constitute Agency policy.
1-464 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Pira et al. (2014)
Population: 2,750 workers employed at
a laminated plastic factory in Italy for at
least 180 days between 1947 and 2011
followed until May 2011. Deaths were
identified from population registries.
Vital status was 96.9% complete and
only 3.1% lost to follow-up.
Outcome definition: Death certificates
used to determine UCOD from myeloid
leukemia (ICD-9: 205).
Design: Prospective cohort mortality
study with external comparison group.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in lowest
exposed category. Lagged exposures
were evaluated to account for cancer
latency.
SMRs calculated using sex, age, and 5-
year calendar periods using mortality
rates from the Piedmont region.
Confidence in effect estimates:3
Overall
SB IB
a
Oth
Confidence
LL
Low
Low >1/ (Potential bias toward the null,
low sensitivity)
SB: Healthy worker effect possible.
IB: Exposure Group B (Appendix A.5.9)
Oth: Low power
Reference: Saberi Hosnijeh et al.
(2013)
Population: 241,465 men and women
recruited from 10 European countries
during 1992-2000. Participants were
predominantly ages 35-70 at
recruitment and were followed up
through 2010.
Outcome definition: Incident primary
leukemias.
Exposures
Exposure assessment: Formaldehyde
is a byproduct from the resins used in
production process and all workers
were presumed to have been
exposed.
Duration and timing: Exposure period
from 1947 through 2011. Median
length of follow-up: 23.6 years.
Duration and timing since first
exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
Observed: 3 myeloid leukemia deaths
Expected: 2.16 myeloid leukemia deaths
based on authors' assumption that 40% of
leukemia deaths are from myeloid leukemia
and 5.3 leukemia deaths were expected.
Myeloid Leukemia (ICD-9: 205)
SMR = 1.39 (0.35-3.78)+ [3]
+Note: EPA derived CIs using the Mid-P
Method [See Rothman and Boice (1979)1
Exposure assessment: Individual
occupational histories obtained by
questionnaire about ever working in
any of 52 occupations considered to
be at high risk of developing cancer.
Occupational exposures estimated as
"high," "low," and no exposure by
linking to a JEM.
Duration and timing: Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Internal comparisons:
Exposure to formaldehyde:
Level 1 RR = 1.00 (Ref. value) [130]
Level 2 RR = 1.02 (0.73-1.42) [49]
Level 3 RR = Nodata [0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-465 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Design: Prospective multinational
cohort incidence study with internal
comparison groups.
Analysis: HRs calculated controlling for
age, sex, smoking, alcohol, physical
activity, education, BMI, family history
of cancer, country, other occupational
exposures, and radiation.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
Exposure to formaldehyde:
Level 1 (none)
Level 2 (low)
Level 3 (high)
Coexposures: Coexposure included
pesticides, herbicides, insecticides,
aromatic solvents, benzene.
chlorinated solvents,
trichloroethvlene. metals, and contact
with animals or animal products,
ionizing radiation.
LOW (Potential bias toward the null;
low sensitivity)
IB: Exposure Group C; latency was not
evaluated
Cf: Confounding possible
Oth: Low power
[As noted in Appendix A.5.9:
Coexposures were not controlled for.
Potential for confounding is unknown
but could have inflated the observed
effect.
Potential for confounding may be
mitigated by low correlation between
exposures in the general population.]
Reference: Blair et al. (2001)
Population: White men, 30 years of age
or older, identified from the Iowa
cancer registry and the Minnesota
hospital surveillance network during
1980-1983. Participation of eligible
cases was 86% and approximately 77-
79% for controls including 77% for
surrogate respondents for deceased
subjects.
Outcome definition: Diagnosis of
leukemia was confirmed by pathology
review for all cases.
Design: Population-based case-control
study of 513 white men with leukemia
from Iowa and Minnesota cancer
surveillance networks. 1,087 controls
were frequency matched on 5-yr age
groups, vital status, and state.
Analysis: ORs calculated for job titles,
employment duration, and exposure
intensity using unconditional logistic
regression controlling for age, state,
direct/surrogate response, and
coexposures, including smoking.
Analyses by year of first exposure were
also conducted to evaluate latency.
Exposure assessment: Individual-level
exposure estimates developed based
on a JEM for each job held for more
than 1 year, the industry where
employed, and starting and ending
year the job was held.
Exposure intensity and probability
assessed for formaldehyde and other
exposures. Exposure intensity refers
to the level likely experienced and
considered a TWA8 over a year.
Duration and timing: Exposure period
based on occupational histories prior
to 1983. Duration and timing since
first exposure were evaluated.
Variation in exposure:
Intensity of exposure:
Level 1 (unexposed)
Level 2 (low)
Level 3 (high)
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9: Other
exposures evaluated included
benzene, other organic solvents,
petroleum-based oils and greases,
cooking oils, ionizing radiation, paper
Internal comparisons:
Acute myeloid leukemia (ICD-9: 205.0)
Level 1 OR=1.0(Ref. value) [118]
Level 2 OR = 0.9 (0.5-1.6 ) [14]
Level 3 no cases
Chronic myeloid leukemia (ICD-9: 205.1)
Level 1 OR = 1.0 (Ref. value) [38]
Level 2 OR = 1.3 (0.6-3.1) [7]
Level 3 OR = 2.9 (0.3-24.5 ) [1]
No notable findings were reported for
duration of time since first exposure to
formaldehyde.
This document is a draft for review purposes only and does not constitute Agency policy.
1-466 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Confidence in effect estimates:3
Overall
SB
IK
Ct
Oth
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
IB: Exposure Group C; lack of latency
analysis
Cf: Potential confounding although
relationship between formaldehyde and
coexposures is unknown.
Reference: Ott et al. (1989)
Population: 29,139 men employed at
two large chemical manufacturing
facilities and a research and
development center who worked during
1940-1978. Vital status was known for
96.4%. Death certificates were
available for 5,785 known descendants
(95.4%).
Outcome definition: Death certificates
used to determine UCOD from
lymphatic leukemia based on the ICD
code in used at the time of death.
Design: Nested case-control study
within a prospective cohort mortality
study. Twenty cases of lymphatic
leukemia were frequency matched to
100 controls on time from hire to death.
Analysis: ORs calculated using
unconditional logistic regression.
Related studies:
Rinskv etal. (1988)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
IB: Exposure Group B; latency
evaluation likely to be underpowered to
detect any effects beyond a 5-year
period
Cf: Benzene is a potential confounder
Exposures
dusts, gasoline and exhaust vapors,
paints, metals, wood dust, asbestos,
asphalt, cattle, meat, solder fumes.
However, analyses of formaldehyde
exposures did not control for other
exposures.]
Exposure assessment: Individual-level
exposure ascertained from
employee's work assignments linked
to records on departmental usage of
formaldehyde.
Duration and timing: Occupational
exposures during 1940-1978. Timing
of formaldehyde exposure not
evaluated.
Variation in exposure: Ever/never
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
21 different chemicals were evaluated
including benzene with much cross
exposure.
Benzene was not evaluated as a
potential confounder and may be
positively correlated with
formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.
Potential for confounding may be
mitigated by rarity of coexposures
among cases.]
Results: effect estimate (95% CI)
[# of cases]
[2]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice, 1979)
Internal comparisons:
OR = 2.6 (0.44-8.59)+
This document is a draft for review purposes only and does not constitute Agency policy.
1-467 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Oth: Low power due to the rarity of
exposure
Reference: Stroup et al. (1986)
Population: 2,239 white male members
of the American Association of
Anatomists from 1888 to 1969 who died
during 1925-1979. Death certificates
obtained for 91% with 9% lost to follow-
up.
Outcome definition: Myeloid leukemia
(ICD-8: 205) listed as cause of death on
death certificates.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex,
race, age, and calendar-year-expected
number of deaths from the U.S.
population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
SB: Health worker effect
IB: Exposure Group A; latency not
evaluated
Cf: Potential confounding
Exposures
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative.
Duration and timing: Occupational
exposure preceding death during
1925-1979. Median birth year was
1912. By 1979, 33% of anatomists
had died. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.
Anatomists may also be coexposed to
stains, benzene, toluene, xylene,
chlorinated hydrocarbons, dioxane,
and osmium tetroxide.
Benzene was not evaluated as a
potential confounder and may be
positively correlated with
formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.]
Results: effect estimate (95% CI)
[# of cases]
Leukemias:
10 total reported
1 lymphatic
5 myeloid (3 chronic, 1 acute, 1
unspecified)
1 acute monocytic
3 leukemia not otherwise specified
External comparisons:
Chronic myeloid leukemia (ICD-8: 205.1)
SMR = 8.8 (1.8-25.5) [3]
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: RR = relative risk; SMR = standardized mortality ratio; UCOD = underlying cause of death; OR = odds ratio;
SRR = summary relative risk; SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or
analysis; TSFE = time since first exposure; URT = upper respiratory tract; LHP = lymphohematopoietic; HR = hazard ratio;
PMR = proportionate mortality ratio; BMI = body mass index; JEM = job-exposure matrix.
This document is a draft for review purposes only and does not constitute Agency policy.
1-468 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Lymphatic leukemia
Epidemiological evidence
The most specific level of lymphatic leukemia diagnosis that is commonly reported across
the epidemiological literature has been based on the first three digits of the Eighth or Ninth
Revision of the ICD code (i.e., Lymphatic leukemia ICD-8: 204 and Lymphoid leukemia ICD-9: 204).
Evidence describing the association between formaldehyde exposure and the specific risk of
lymphatic leukemia was available from nine epidemiological studies—two case-control studies
fHauptmann et al.. 2009: Blair etal.. 20011 and seven cohort studies fMeyers etal.. 2013: Saberi
Hosniieh etal.. 2013: Beane Freeman et al.. 2009: Hayes etal.. 1990: Ottetal.. 1989: Walrath and
Fraumeni. 1984.19831. Six of the cohort studies all ascertained lymphatic leukemia diagnoses from
death certificates and one examined incident cases (Saberi Hosnijeh etal.. 2013). All studies
reported lymphatic leukemia outcomes based on the ICD-8 or ICD-9 diagnostic code 204 without
separate results for acute lymphocytic leukemia and CLL. One case-control study (Hauptmann et
al.. 2009) ascertained lymphatic leukemia diagnoses from death certificates whereas the other
ascertained incident cases of lymphatic leukemia from a cancer registry and a hospital network
fBlair etal.. 20011. Both studies reported specific results for CLL; however, while diagnoses of
lymphatic leukemia reviewed here are those identified according to the ICD codes used at the time
of diagnoses, in the ICD-10 coding rubric, CLL would be included as NHL. Study details are
provided in the evidence table for lymphatic leukemia (see Table 1-61). Study results for ICD-7
code 204 were not included because this code includes all leukemias. The outcome-specific
evaluations of confidence in the reported effect estimate of an association from each study are
provided in Appendix A.5.9 and the confidence conclusions are provided in the evidence table for
lymphatic leukemia (see Table 1-61) following the causal evaluation.
Consistency of the observed association
The point estimates and CIs of all eight informative studies were consistently around the
null, which does not provide evidence of an association between formaldehyde exposure and the
risk of developing or dying from lymphatic leukemia. The range of central relative effect estimates
(selecting the highest exposure level results when there was more than one result) was from zero
ffWalrath and Fraumeni. 19841: [zero cases]) to 2.6 ffOttetal.. 19891: [1 case]) and both of these
results were classified with low confidence. The three results classified with high or medium
confidence were SMR = 0.71 in Meyers et al. (2013). OR = 1.0 in Hauptmann et al. (2009). and
SMR = 1.15 in Beane Freeman etal. (2009). The study results presented in Table 1-61 (by
confidence level and publication date) detail all of the reported associations between exposures to
formaldehyde and the risks of developing or dying from lymphatic leukemia along with a summary
graphic of any major limitation and the confidence classification of the effect estimate. Results are
plotted in Figure 1-40.
This document is a draft for review purposes only and does not constitute Agency policy.
1-469 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
Strength of the observed association
Summary effect estimates for the association between formaldehyde exposure and the risk
of mortality from lymphatic leukemia ranged from zero to 2.6 and clustered around the null.
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants occurred
before their lymphatic leukemia was detected and in the studies that ascertained individual-level
exposures, the estimation of formaldehyde exposures was based on job titles and was done in a
blinded fashion with respect to outcome status. None of the eight studies provided analyses of a
temporal relationship between the timing of exposure and the diagnoses of lymphatic leukemia or
deaths from lymphatic leukemia.
Exposure-response relationship
None of the studies evaluated the effect of duration of formaldehyde exposure on the
mortality risk of lymphatic leukemia. There were only two sets of results, one classified with
medium confidence and one with low confidence, which evaluated any form of exposure-response
for increasing measures of formaldehyde exposure fBeane Freeman et al.. 2009: Blair etal.. 20011
and neither showed a pattern of increasing risk with increasing formaldehyde exposure.
Potential impact of selection bias; information bias; confounding bias, and chance
There was potential for selection bias in two studies that were only able to ascertain death
certificated for 75-79% of the decedents fOttetal.. 1989: Walrath and Fraumeni. 19831. but there
was no evidence that inclusion rates may have been related to either exposure or outcome, and
thus, there is little concern about selection bias. Among the studies reporting on the risk of
lymphatic leukemia, which only indicated the equivalent of ever/never exposure to formaldehyde,
there was little potential for information bias. In fact, results consistently showed no evidence of an
association—regardless of the quality of exposure assessment further. Confounding is another
potential bias that could arise if another cause of lymphatic leukemia was statistically associated
with formaldehyde exposure. However, there does not appear to be any evidence of negative
confounding, which could have obscured a real but unobserved effect. While there did not appear
to be an association between exposure to formaldehyde and the risk of lymphatic leukemia, given
the limited database of specific results, and the possibility of biases that could obscure any true
effect, the available epidemiological data are inadequate to conclude that formaldehyde is not likely
to be carcinogenic to humans.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
lymphatic leukemia placed the greatest weight on four particular considerations: (1) the generally
consistent pattern of null results across high, medium, and low confidence studies; (2) the absence
This document is a draft for review purposes only and does not constitute Agency policy.
1-470 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Toxicological Review of Formaldehyde—Inhalation
of exposure-response relationships showing that increased exposure to formaldehyde was
associated with increased risk of lymphatic leukemia; (3) the limited database from which to
evaluate the association; and (4) the absence of evidence to evaluate the potential risk to sensitive
populations or lifestages. Although consistent observations of genotoxicity in peripheral blood
lymphocytes, exfoliated buccal cells or nasal mucosal cells have been observed across several
occupational studies, these data were not interpreted as sufficient to further strengthen the
judgment on the human evidence of lymphatic leukemia.
Conclusion
The available epidemiological studies provide indeterminate evidence to assess the
carcinogenic potential evidence of an association between formaldehyde exposure and an
increased risk of lymphatic leukemia.
a
<0
£
+¦»
(/)
LU
o
a)
«4-
*+-
LU
a)
a
OH
100
r
10
0.1
Population-level
exposure
assessment
J?
Individual-level exposure assessment
£ .1
E
D
O oj
Blair
a3
&
£
1
CL
w
£
TO
c
>2
$
-------
Toxicological Review of Formaldehyde—Inhalation
Results specifically for chronic lymphatic leukemia (CLL) are noted by these abbreviations:
SMR = standardized mortality ratio; PMR = proportionate mortality ratio; RR = relative risk; OR = odds
ratio. For each measure of association, the number of exposed cases is provided in brackets (e.g., [n = 4])
For studies reporting results on multiple metrics of exposure, each metric is included; however, only the
highest category of each exposure metric is presented in the figure.
Table 1-61. Epidemiological studies of formaldehyde exposure and risk of
lymphatic leukemia
Study
Reference: Meyers et al. (2013)
Population: 11,043 workers in three
U.S. garment plants exposed for at
least 3 months. Women comprised
82% of the cohort. Vital status was
followed through 2008 with 99.7%
completion.
Outcome definition: Death
certificates used to determine both
the UCOD from lymphocytic leukemia
(ICD code in use at time of death).
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: SMRs calculated using sex,
age, race, and calendar-year-specific
U.S. mortality rates. Poisson
regression analysis based on internal
referents.
Related studies:
Stavner et al. (1985)
Stavner et al. (1988)
Pinkerton et al. (2004)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
High
HIGH •
IB: Exposure Group A
Reference: Beane Freeman et al.
(2009) with supplemental online
tables.
Exposures
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and 1984.
Geometric TWA8 exposures ranged from
0.09 to 0.20 ppm. Overall geometric
mean concentration of formaldehyde was
0.15 ppm (GSD 1.90 ppm). Area
measures showed constant levels without
peaks. Historically earlier exposures may
have been substantially higher.
Exposure assessment: Individual-level
exposure estimates based on job titles,
tasks, visits to plants by study industrial
hygienists, and monitoring data through
1980.
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
SMR = 0.71 (0.26-1.56) [6]
Internal comparisons:
Peak exposure
Unexposed RR = 0.27 (0.03-2.13) [1]
Level 1 RR = 1.00(Ref. value) [14]
Level 2 RR = 0.81 (0.33-1.96) [8]
Duration and timing: Exposure period
from 1955 through 1983. Median
duration of exposure was 3.3 years.
More than 40% exposures <1963.
Median time since first exposure was
39.4 years. Duration and timing since
first exposure were not evaluated.
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
This document is a draft for review purposes only and does not constitute Agency policy.
1-472 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3). Median cumulative
exposure = 0.6 ppm-years (range 0-
107.4).
Study
Population: 25,619 workers
employed at 10 formaldehyde-using
or formaldehyde-producing plants in
the U.S. followed from either the
plant start-up or first employment
through 2004. Deaths were identified
from the National Death Index with
remainder assumed to be living. Vital
status was 97.4% complete and only
2.6% lost to follow-up.
Outcome definition: Death
certificates used to determine UCOD
from lymphatic leukemia (ICD-8: 204).
Design: Prospective cohort mortality
study with external and internal
comparison groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year,
age, sex, and race; adjusted for pay
category compared to workers in
lowest exposed category. Lagged
exposures were evaluated to account
for cancer latency.
SMRs calculated using sex, age, race,
and calendar-year-specific U.S.
mortality rates.
Related studies:
Blair etal. (1986)
Hauptmann et al. (2003)
Confidence in effect estimates:3
SB IB €f Oth
Overall
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A
Reference: Hauptmann et al.
(2009)
Population: 6,808 embalmers and
funeral directors who died during
1960-1986. Identified from registries
of the National Funeral Directors'
Association, licensing boards, and
state funeral directors' associations,
NY State Bureau of Funeral Directors,
Multiple exposure metrics including peak,
average, and cumulative exposures were
evaluated using categorical and
continuous data.
Duration and timing: Exposure period
from <1946 through 1980. Median length
of follow-up: 42 years. Duration and
timing since first exposure were not
evaluated.
Exposure assessment: Occupational
history obtained by interviews with next
of kin and coworkers using detailed
questionnaires. Exposure was assessed
by linking questionnaire responses to an
exposure assessment experiment
providing measured exposure data.
Exposure levels (peak, intensity, and
cumulative) were assigned to each
individual using a predictive model based
Results: effect estimate (95% CI)
[# of cases]
Level 3 RR = 1.15 (0.54-2.47) [14]
p-trend (exposed) >0.50;
p-trend (all) = 0.30
Average intensity
Unexposed RR = 0.26 (0.03-2.01) [1]
Level 1 RR = 1.00 (Ref. value) [22]
Level 2 RR = 0.92 (0.39-2.16) [7]
Level 3 RR = 0.88 (0.37-2.11) [7]
p-trend (exposed) >0.50;
p-trend (all) >0.50
Cumulative exposure
Unexposed RR = 0.24 (0.03-1.88) [1]
Level 1 RR = 1.00 (Ref. value) [21]
Level 2 RR = 0.57 (0.21-1.54) [5]
Level 3 RR = 1.02 (0.47-2.21) [10]
p-trend (exposed) = 0.46;
p-trend (all) = 0.41
External comparisons:
SMRunexposed =0.26(0.04-1.82) [1]
SMRexposed = 1.15(0.83-1.59) [36]
Internal comparisons:
Embalming:
Never: OR = 1.0 (Ref. value)
[# not given]
Ever: OR = 1.0 (0.5-1.9)
[# not given]
Variation in exposure:
Peak exposure:
Level 1 (>0 to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Level 1 (>0 to <0.5 ppm)
Level 2 (0.5 to <1.0 ppm)
Level 3 (>1.0 ppm)
Cumulative exposure:
Level 1 (>0 to <1.5 ppm-yrs)
Level 2 (1.5 to <5.5 ppm-yrs)
Level 3 (>5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders.
This document is a draft for review purposes only and does not constitute Agency policy.
1-473 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
and CA Funeral Directors and
Embalmers. Deaths were identified
from the National Death Index. Next
of kin interviews conducted for 96%
of cases and 94% of controls.
Outcome definition: Death
certificates used to determine UCOD
from CLL (ICD-8: 204.1).
[Note that while CLL was classified as
lymphocytic leukemia in ICD-8, in ICD-
10, it is included as non-Hodgkin
lymphoma]
Design: Nested case-control study
within a prospective cohort study.
Analysis: ORs calculated using
unconditional logistic regression
adjusted for date of birth, age at
death, sex, data source, and smoking.
Lagged exposures were evaluated to
account for cancer latency.
Related studies: Haves et al. (1990)
Walrath and Fraumeni (1983)
Walrath and Fraumeni (1984)
Note: The original cohorts from these
three related studies were combined
in Hauptmann et al. (2009) and
follow-up was extended so the case-
series overlap and are not
independent. However, the three
related cohorts used external
reference groups for comparison
while Hauptmann et al. (2009)
select internal controls, which were
independent of the reference groups
used in the other studies.
Confidence in effect estimate:3
SB IE Cf Oth
Overall
Confidence
Medium
MEDIUM J,
(Potential bias toward the nullvU)
IB: Exposure Group A
on the exposure data. The model
explained 74% of the observed variability
in exposure measurements.
Multiple exposure metrics including
duration (mean = 33.1 yrs in cases), # of
embalming, peak, average, and
cumulative exposures were evaluated
using categorical and continuous data.
Duration and timing: Exposure period
from <1932 through 1986. Duration of
exposure was evaluated. Duration is also
a surrogate for time because first
exposure since dates of death were
closely related to cessation of workplace
exposures
Variation in exposure:
For variations in exposure from table 3 in
the publication:
Level 1 (no exposure to embalming)
For variations in exposure from table 4 in
the publication:
Level 1 (<500 embalming)
Duration of exposure:
Level 2 (<20 years)
Level 3 (20-34 years)
Level 4 (>34 years)
Number of embalming:
Level 2 (500-1,422)
Level 3 (1,423-3,068)
Level 4 (>3,068)
Cumulative exposure:
Level 2 (<4,058 ppm-hrs)
Level 3 (4,059-9,253 ppm-hrs)
Level 4 (>9,253 ppm-hrs)
Average intensity (while embalming):
Level 2 (<1.4 ppm)
Level 3 (>1.4-1.9 ppm)
Level 4 (>1.9 ppm)
TWA8 formaldehyde intensity:
Level 2 (<0.10 ppm)
Level 3 (>0.10-0.18 ppm)
Level 4 (>0.18 ppm)
Peak Exposure:
Level 2 (<7.0 ppm)
Level 3 (7.0 to <9.3 ppm)
Level 4 (>9.3 ppm)
Coexposures: None evaluated.
This document is a draft for review purposes only and does not constitute Agency policy.
1-474 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Reference: Hayes et al. (1990)
Population: 4,046 deceased U.S. male
embalmers and funeral directors,
derived from licensing boards and
funeral director associations in 32
states and the District of Columbia
who died during 1975-1985. Death
certificates obtained for 79% of
potential study subjects (n = 6,651)
with vital status unknown for 21%.
Outcome definition: Death
certificates and licensing boards used
to determine cause of death from
lymphatic leukemia (ICD-8: 204).
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
deaths from the U.S. population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM -i,
(Potential bias toward the null>|/)
SB: Missing death certificates
considered to missing at random
IB: Exposure Group A; latency not
evaluated. Possible undercounting of
cases due to abbreviated death
certificate
Reference: Saberi Hosnijeh et
al. (2013)
Exposures
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl
alcohol, glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Chemical coexposures are not known risk
factors for this outcome.
Authors state that major exposures are to
formaldehyde and possibly
glutaraldehyde and phenol.
Duration and timing: Occupational
exposure preceding death during 1975-
1985. Of 115 deaths from LHP cancer, 66
(57%) were aged 60-74 years. Duration
and timing since first exposure were not
evaluated.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl
alcohol, glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Chemical coexposures are not known risk
factors for this outcome.
Exposure assessment: Individual
occupational histories obtained by
questionnaire about ever working in any
Results: effect estimate (95% CI)
[# of cases]
[7]
Internal comparisons:
Exposure to formaldehyde:
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Exposure assessment: Presumed
exposure to formaldehyde tissue fixative.
Exposure based on occupation, which was
confirmed on death certificate. Authors
subsequently measured personal
embalming exposures ranging from
0.98 ppm (high ventilation) to 3.99 ppm
(low ventilation) with peaks up to
20 ppm.
External comparisons:
PMR = 0.74 (0.29-1.53)
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
This document is a draft for review purposes only and does not constitute Agency policy.
1-475 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Population: 241,465 men and women
recruited from 10 European countries
during 1992-2000. Participants were
predominantly aged 35-70 at
recruitment and were followed up
through 2010.
Outcome definition: Incident primary
leukemias.
Design: Prospective multinational
cohort incidence study with internal
comparison groups.
Analysis: HRs calculated controlling
for age, sex, smoking, alcohol,
physical activity, education, BMI,
family history of cancer, country,
other occupational exposures, and
radiation.
Confidence in effect estimates:3
of 52 occupations considered to be at
high risk of developing cancer.
Occupational exposures estimated as
"high," "low," and no exposure by linking
to a JEM.
Duration and timing: Duration and timing
since first exposure were not evaluated.
Variation in exposure:
Exposure to formaldehyde:
Level 1 (none)
Level 2 (low)
Level 3 (high)
Coexposures: Coexposure included
pesticides, herbicides, insecticides,
aromatic solvents, benzene, chlorinated
solvents, trichloroethvlene. metals,
contact with animals or animal products,
ionizing radiation.
Level
1
RR = 1.00
(Ref. value)
[130]
Level
2
RR = 1.08
(0.81-1.45)
[64]
Level
3
RR = 1.38
(0.44-4.35)
[3]
SB IB Cf Oth
Overall
Confidence
Low
LOW (Potential bias toward the
null; low sensitivity)
IB: Exposure Group C; Latency was
not evaluated
Cf: Confounding possible
Oth: Low power
[As noted in Appendix A.5.9: Coexposures
were not controlled for.
Potential for confounding is unknown but
could have inflated the observed effect.
Potential for confounding may be
mitigated by low correlation between
exposures in the general population.]
Reference: Blair et al. (2001)
Population: White men, 30 years of
age or older, identified from the Iowa
cancer registry and the Minnesota
hospital surveillance network during
1980-1983. Participation of eligible
cases was 86% and approximately 77-
79% for controls including 77% for
surrogate respondents for deceased
subjects.
Outcome definition: Diagnosis of
leukemia was confirmed by pathology
review for all cases.
Design: Population-based case-
control study of 513 white men with
leukemia from Iowa and Minnesota
cancer surveillance networks. 1,087
controls were frequency matched on
Exposure assessment: Individual-level
exposure estimates developed based on a
JEM for each job held for more than
1 year, the industry where employed, and
starting and ending year the job was held.
Exposure intensity and probability
assessed for formaldehyde and other
exposures. Exposure intensity refers to
the level likely experienced and
considered a TWA8 over a year.
Duration and timing: Exposure period
based on occupational histories prior to
1983. Duration and timing since first
exposure were evaluated.
Variation in exposure:
Intensity of exposure:
Level 1 (unexposed)
Level 2 (low)
Internal comparisons:
Acute lymphatic leukemia (ICD-9:204.0)
No exposed cases
Chronic lymphatic leukemia (ICD-9: 204.1)
Level 1 OR = 1.0 (Ref. value) [483]
Level 2 OR = 1.2 (0.7-1.8 ) [29]
Level 3 OR = 0.6 (0.1-5.3) [1]
No notable findings were reported for
duration of time since first exposure to
formaldehyde.
This document is a draft for review purposes only and does not constitute Agency policy.
1-476 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Exposures
Level 3 (high)
Coexposures: None evaluated as
potential confounders.
Study
5-yr age groups, vital status, and
state.
Analysis: ORs calculated for job titles,
employment duration and exposure
intensity using unconditional logistic
regression controlling for age, state,
direct/surrogate response and
coexposures, including smoking.
Analyses by year of first exposure
conducted.
Confidence in effect estimates:3
Ove rail
SB
IK
Ct
Oth
Confidence
Low
LOW -i,
(Potential bias toward the nullvU)
IB: Exposure Group C; lack of latency
analysis
Cf: Potential confounding
Reference: Ott et al. (1989)
Population: 29,139 men employed at
two large chemical manufacturing
facilities and a research and
development center who worked
during 1940-1978. Vital status was
known for 96.4%. Death certificates
were available for 5,785 known
descendants (95.4%).
Outcome definition: Death
certificates used to determine UCOD
from lymphatic leukemia based on
the ICD code in used at the time of
death.
Design: Nested case-control study
within a prospective cohort mortality
study. Twenty cases of lymphatic
leukemia were frequency matched to
100 controls on time from hire to
death.
Analysis: ORs calculated using
unconditional logistic regression.
Related studies:
Rinskv etal. (1988)
Confidence in effect estimates:3
[As noted in Appendix A.5.9: Other
exposures evaluated included benzene.
other organic solvents, petroleum-based
oils and greases, cooking oils, ionizing
radiation, paper dusts, gasoline and
exhaust vapors, paints, metals, wood
dust, asbestos, asphalt, cattle, meat,
solder fumes. However, analyses of
formaldehyde exposures did not control
for other exposures.]
Duration and timing: Occupational
exposures during 1940-1978. Timing of
formaldehyde exposure not evaluated.
[As noted in Appendix A.5.9: 21 different
chemicals were evaluated including
benzene with much cross exposure.
Benzene was not evaluated as a potential
confounder and may be positively
correlated with formaldehyde exposure.
Potential for confounding is unknown but
could have inflated the observed effect.
Potential for confounding may be
mitigated by rarity of coexposures among
cases.]
Results: effect estimate (95% CI)
[# of cases]
Internal comparisons:
OR = 2.6 (0.13-13.0)+ [1]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
Variation in exposure: Ever/never
Coexposures: None evaluated as
potential confounders.
Exposure assessment: Individual-level
exposure ascertained from employee's
work assignments linked to records on
departmental usage of formaldehyde.
This document is a draft for review purposes only and does not constitute Agency policy.
1-477 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
SB IB Cf Oth
Overall
Confidence
Low
LOW -i,
(Potential bias toward the nullvU)
Low power due to the rarity of
exposure.
IB: Exposure Group B; latency
evaluation likely to be underpowered
to detect any effects beyond a 5-year
period
Cf: Benzene is a potential confounder
Oth: Low power due to the rarity of
exposure
Reference: Walrath and Fraumeni
(1984)
Population: 1,007 deceased white
male embalmers from California who
died during 1925-1980. Death
certificates obtained for all.
Outcome definition: Lymphatic
leukemia (ICD-8: 204) listed as cause
of death on death certificate.
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
deaths from the U.S. population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
IB: Exposure Group A; latency was not
evaluated
Oth: Low power for lymphatic
leukemia
Reference: Walrath and Fraumeni
(1983)
Exposures
Exposure assessment: Presumed
exposure to formaldehyde tissue fixative.
Duration and timing: Occupational
exposure preceding death during 1916-
1978. Birth year ranged from 1847
through 1959. Median age of death was
62 years. Most deaths were among
embalmers with active licenses. Duration
and timing since first exposure were not
evaluated.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl
alcohol, glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Exposure assessment: Presumed
exposure to formaldehyde tissue fixative.
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
Observed: 0 lymphatic leukemia deaths
Expected: 2.2 lymphatic leukemia
deaths
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
External comparisons:
Observed: 4 lymphatic leukemia deaths
Expected: 2.6 lymphatic leukemia
deaths
PMR = 1.54 (0.49-3.71)+ [4]
Population: 1,132 deceased white
male embalmers licensed to practice
during 1902-1980 in New York who
Duration and timing:
Occupational exposure preceding death
during 1902-1980. Median year of birth
PMR = 0(0-1.36)+ [0 vs. 2.2
expected]
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
This document is a draft for review purposes only and does not constitute Agency policy.
1-478 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
died during 1925-1980 identified
from registration files. Death
certificates obtained for 75% of
potential study subjects (n = 1,678).
Outcome definition: Lymphatic
leukemia (ICD-8: 204) listed as cause
of death on death certificate.
Design: Proportionate mortality
cohort study with external
comparison group.
Analysis: PMRs calculated using sex,
race, age, and calendar-year-expected
deaths from the U.S. population.
Confidence in effect estimates:3
SB IB Cf Oth
Ove rail
Confidence
Low
LOW
(Potential bias toward the nullvU)
SB: Missing death certificates
considered to missing at random
IB: Exposure Group A; latency was
not evaluated
Oth: Low power for lymphatic
leukemia
was 1901. Median year of initial license
was 1931. Median age at death was
1968. Expected median duration of
exposure was 37 years. Duration and
timing since first exposure were not
evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl
alcohol, glutaraldehyde, mercury, arsenic,
zinc, and ionizing radiation.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or analysis;
UCOD = underlying cause of death; GSD = geometric standard deviation; SMR = standardized mortality ratio; RR = relative risk;
TWA8 = 8-hour time-weighted average; LHP = lymphohematopoietic; PMR = proportionate mortality ratio; BMI = body mass
index; JEM = job-exposure matrix; OR = odds ratio.
1 Multiple myeloma
2 Epidemiological evidence
3 The most specific classification of multiple myeloma diagnosis that is commonly reported
4 across the epidemiological literature has been based on the first three digits of the Eighth or Ninth
5 Revision of the ICD code without further differentiation (i.e., Multiple myeloma ICD-8/9: 203).
6 Evidence describing the association between formaldehyde exposure and the risk of developing or
7 dying from multiple myeloma was available from 14 epidemiological studies—five case-control
This document is a draft for review purposes only and does not constitute Agency policy.
1-479 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
studies (Hauptmann et al.. 2009: Heineman etal.. 1992: Pottern etal.. 1992: Boffettaetal.. 1989: Ott
etal.. 19891 and nine cohort studies fCoggon et al.. 2014: Pira etal.. 2014: Meyers etal.. 2013:
Beane Freeman et al.. 2009: Stellman etal.. 1998: Band etal.. 1997: Dell andTeta. 1995: Hayes et al..
1990: Edling et al.. 1987bl. Study details are provided in the evidence table for multiple myeloma
(see Table 1-62). The outcome-specific evaluations of confidence in the reported effect estimate of
an association from each study are provided in Appendix A.5.9 and the confidence conclusions are
provided in the evidence table for multiple myeloma (see Table 1-62) following the causal
evaluation. Details of the reported results of high, medium, and low confidence are provided in the
evidence table for multiple myeloma (see Table 1-62) following the causal evaluation.
Consistency of the observed association
The results of these studies appear to be mixed with some showing non-significant
increases in risk and other showing non-significant decreases in risk. Nine of the 14 studies were
low confidence (Pira etal.. 2014: Stellman et al.. 1998: Dell and Teta. 1995: Pottern etal.. 1992:
Boffetta et al.. 1989: Ott etal.. 1989: Edling etal.. 1987b) with many results based on fewer than
five cases.626235 However, only the study by Beane Freeman et al. (2009) reported a result with
high confidence showing an association between peak formaldehyde exposure and risk of multiple
myeloma. The study results presented in Table 1-62 (by confidence level and publication date) and
plotted in Figure 1-41 detail all of the reported associations between exposures to formaldehyde
and the risks of developing or dying from multiple myeloma
The first four studies shown at the left in Figure 1-41 followed the health of groups of
occupationally exposed workers in three different industries and did not have individual-level
exposure estimates fDell and Teta. 1995: Hayes etal.. 1990: Edling etal.. 1987bl. Respectively,
these were: (1) workers making grinding wheels bound with formaldehyde resins, (2) embalmers,
and (3) workers manufacturing plastics—professions known to be exposed to formaldehyde.
Importantly, all of these professions were exposed to high peak concentrations of formaldehyde.
Edling et al. (1987b) reported that the workers making grinding wheels bound with formaldehyde
resins were exposed to peak formaldehyde levels of up to 20-30 mg/m3 (15-23 ppm). Embalmers
(Haves etal.. 1990) were also exposed to high peak formaldehyde concentrations with mean
exposures of more than 2 ppm and peaks as high as 8.7 ppm f Stewart etal.. 19921. Workers at the
plastics manufacturing facilities studied by Dell and Teta (1995) were exposed to formaldehyde,
formaldehyde resins, and formaldehyde molding compounds. An independent occupational
hygiene survey of facilities producing similar products reported peak exposure for these activities
of 1.88 ppm, 30.45 ppm, and 60.77 ppm, respectively (Stewart etal.. 1987). The results of these
three studies are displayed beneath the header of "Population-level exposure assessment." All
three studies showed elevated RRs of multiple myeloma mortality as measured by the mortality
ratios; although, none of the three was statistically robust enough to decrease the likelihood of
chance as an alternative explanation. The Hayes et al. (1990) result (PMR = 1.37; 95% CI 0.84-2.12;
n = 20) was classified with medium confidence but the other two results from Edling et al. (1987b)
This document is a draft for review purposes only and does not constitute Agency policy.
1-480 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
(SMR = 4.0; 95% CI 0.45-14.44; n = 2) and Dell and Teta (19951 (SMR = 2.62; 95% CI 0.85-6.11;
n = 8) were classified with low confidence.
The second set of studies [n = 10) is displayed in Figure 1-41 fCoggon et al.. 2014: Meyers et
al.. 2013: Beane Freeman et al.. 2009: Hauptmann etal.. 2009: Stellman et al.. 1998: Band etal..
1997: Heineman et al.. 1992: Pottern etal.. 1992: Boffetta et al.. 1989: Ottetal.. 19891 beneath the
header of "Individual-level exposure assessment." In principle, a general strength of this second set
of studies was their use of individualized exposure data; however, the quality of the exposure
assessment for each individual varied considerably across this set of studies. These 10 studies with
individual-level exposure assessment can be divided into two groups based on the methods of
individual exposure assessment. The first grouping gathered minimal information
(e.g., questionnaire data on "ever" exposure to formaldehyde) on formaldehyde exposure (Stellman
etal.. 1998: Heineman etal.. 1992: Pottern etal.. 1992: Boffetta et al.. 19891. The second grouping
focused on workers who were occupationally exposed to formaldehyde and used work assignments
or job histories matched to exposure data to assess workers' formaldehyde exposures (Coggon et
al.. 2014: Meyers etal.. 2013: Beane Freeman et al.. 2009: Hauptmann et al.. 2009: Band etal.. 1997:
Ottetal.. 19891.
The exposure assessment methodology for the first grouping of four studies with
individual-level exposures was especially crude. Exposure assessment was limited to either a one-
time questionnaire asking participants to check off a box if they were "ever" exposed to
formaldehyde in the workplace or in daily life (Stellman et al.. 1998: Boffetta etal.. 1989) or using
the occupation listed on individuals' most recent annual tax records to estimate previous
occupational formaldehyde exposure as "none," "possible," or "probable" (Heineman et al.. 1992:
Pottern et al.. 19921. While the large size of these studies was considered to be a strength, the
weaknesses of their relatively low-quality exposure assessment outweighed that strength. It is well
known that the use of low-quality exposure data in epidemiological studies may preclude the ability
to detect all but the strongest association.
The second grouping of studies, with relatively higher quality individual-level exposure to
formaldehyde, examined occupational histories at different points in time and linked this to
measured or estimated exposures (Coggon etal.. 2014: Meyers etal.. 2013: Beane Freeman et al..
2009: Hauptmann etal.. 2009: Band etal.. 1997: Ottetal.. 19891. While the relative effect estimates
for multiple myeloma mortality in each of these cohorts compared to the general population did not
show elevated risks (relative effect estimates of: 0.8,1.4,1.0, 0.94,1.24, 0.99), two studies (Coggon
etal.. 2014: Beane Freeman et al.. 2009) showed somewhat higher risks when analyses focused on
the workers with highest peak exposure. Beane Freeman et al. (2009) evaluated results by each
worker's highest formaldehyde concentration during a "peak" exposure event, by average intensity
of exposure, by cumulative exposure, and by duration of exposure. Peak exposure events were
defined as short-term exposures (<15 minutes) that exceeded the TWA formaldehyde intensity
fBeane Freeman etal.. 20091. Workers' peak exposures were defined as the highest concentration
This document is a draft for review purposes only and does not constitute Agency policy.
1-481 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
among their peak exposure events. In Beane Freeman et al. (20091. the highest peak exposure
category represents the workers who had ever experienced short-term peak exposure to >4.0 ppm.
The Beane Freeman et al. (2009) results in the high category of peak exposures were RR = 2.04
(95% CI 1.01-4.12). In Coggon et al. (2014), the "high" category of exposure represented workers
who ever had a job in the highest formaldehyde exposure category (>2 ppm). The Coggon et al.
(2014) results in the high exposure category were, however, relatively weak SMR = 1.18 versus
0.99 for all workers.
Hauptmann et al. (2009) and Ott et al. (1989) assessed individual-level exposure but only
presented results specific to formaldehyde exposures for the study population as a whole.
Similarly, the study of garment workers f Meyers et al.. 2 0131 relied on individual measures of the
timing of exposure but did not have formaldehyde concentration data beyond the industrial
hygiene data used to plan the study fStavner et al.. 19881. Continuous area monitoring showed that
formaldehyde levels were relatively constant with no substantial peak levels over the work shift
(Stavner et al.. 1988). The results from Meyers et al. (2013) are mixed, with the strongest evidence
showing a statistically significant decreased risk among workers with the longest duration of
formaldehyde exposure in analyses compared to internal referents with less than a 3-year exposure
duration (SRR = 0.28; 95% CI 0.08-0.99).
In summary, among all the studies that used individual-level exposure assessment, the
study with the highest quality exposure assessment methodology was the National Cancer Institute
study (Beane Freeman et al.. 2009) among industrial workers at facilities either using
formaldehyde or producing formaldehyde. Beane Freeman et al. (2009) reported on three
different, but related, measures of exposure to formaldehyde based on different exposure
assessment techniques highlighting peak, cumulative and average exposures and showed elevated
risk across all three measures; the most pronounced effects showed a two-fold increased risk of
mortality from multiple myeloma associated with the highest level of peak exposure to
formaldehyde (RR = 2.04; 95% CI 1.01-4.12).
The three studies with population-level exposure assessment, (Dell and Teta. 1995: Hayes
etal.. 1990: Edling etal.. 1987b). all had very high peak exposure and were consistent with Beane
Freeman et al. (2013) in showing an elevated risk although none was able to rule out chance. The
large population studies with only crude measures of formaldehyde exposure reported mixed
results with only a slightly higher risk for those judged to be "Probably" exposed (see Figure 1-41).
The studies of industrial workers did not show increased risks in their populations as a whole but
did report somewhat higher risks among the workers with highest exposure when individual-level
exposures were considered (Coggon etal.. 2014: Beane Freeman et al.. 2009).
A better understanding of the etiologic progression of multiple myeloma may be needed to
interpret these findings but there is some consistent epidemiological evidence suggesting an
association between peak formaldehyde exposures and increased risk of multiple myeloma and
possibly an increased risk at shorter durations, which could select out the responsive individuals
This document is a draft for review purposes only and does not constitute Agency policy.
1-482 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
leaving the nonresponsive individuals without additional risks. However, it could also be the case
from these data that only peak exposures are associated with multiple myeloma.
Strength of the observed association
While reported relative effect estimates were consistently elevated above the null value of
one across the studies, the magnitude of the relative effect estimates varied with the quality of the
exposure assessment. Studies with higher quality exposure data based on individual-level
exposure assessment generally reported higher relative effect estimates (stronger associations)
Setting aside the large population-based studies with crude exposure assessment fStellman
etal.. 1998: Heineman etal.. 1992: Pottern etal.. 1992: Boffetta et al.. 19891 and focusing on
individual-level exposure results where possible, the strength of the associations ranged from 1.2 to
4.0, but the upper end of that range was based on two studies with very few formaldehyde-exposed
cases. The results at the highest levels of peak formaldehyde exposure showed an approximately
two-fold relative increase in risk of death from multiple myeloma (Beane Freeman etal.. 20091.
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants started
prior to their multiple myeloma diagnosis and in the studies that ascertained individual-level
exposures, the estimation of formaldehyde exposures was based on job titles and was done in a
blinded fashion with respect to outcome status. The epidemiological literature for formaldehyde
and multiple myeloma describes only one study that evaluates the impact of TSFE (Meyers etal..
20131: however, while those results showed what appeared to be a slight downward trend toward
lower risks at shorter times since first exposure, the CIs around those estimated risks were wide
and overlapped substantially. Such findings do not add much additional information.
Exposure-response relationship
There was limited evidence of exposure-response relationships in three multiple myeloma
studies. The study by Beane Freeman et al. (2009) reported on three different measures of
exposure to formaldehyde and showed elevated risk across all three measures, most strongly for
peak exposure (RR = 2.04; 95% CI 1.01-4.12) for the highest category (trend p = 0.08). There was
also a finding of greater risks of multiple myeloma at shorter durations of exposure compared to
longer durations; in two analyses of duration using both internal and external comparison groups,
those workers with the longest duration of exposure (10+ years) were at lower risk than those with
3-9 years of exposure. This would be inconsistent with an exposure-response pattern for duration
of exposure or cumulative exposure but is not necessarily inconsistent with the finding of an
exposure-response for higher levels of peak exposure. Coggon et al. (20141 reported a very modest
increase in risk among those workers in the high exposure category (SMR = 1.18; 95% CI 0.57-
2.18); however, the risk among workers in the low/moderate category was even higher
(SMR = 1.47; 95% CI 0.82-2.43). Pottern et al. (1992) reported increasing relative risks with the
This document is a draft for review purposes only and does not constitute Agency policy.
1-483 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
qualitative likelihood of exposure with "possible" exposure having RR = 1.1 (95% CI 0.8-1.6) and
"probable" exposure having RR = 1.6 (95% CI 0.4-5.3).
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias is an unlikely bias in the epidemiological studies of multiple myeloma as the
case-control studies evaluated exposure status without regard to outcome status and had
participation levels of 77-100% and each of the cohort studies included at least 79% of eligible
participants and lost fewer than 6% of participants over the course of mortality follow-up. The
healthy worker effect and the healthy worker survivor effect could obscure a truly larger effect of
formaldehyde exposure in analyses based on "external" comparisons with mortality in the general
population fCoggon etal.. 2014: Meyers etal.. 2013: Beane Freeman etal.. 2009: Dell and Teta.
1995: Hayes etal.. 1990: Ottetal.. 1989: Edlingetal.. 1987bl. but would not influence analyses
using "internal" or matched comparison groups (Beane Freeman et al.. 2009: Hauptmann etal..
2009: Stellman et al.. 1998: Heineman et al.. 1992: Pottern etal.. 1992: Boffettaetal.. 1989).
Differential exposure misclassification is considered unlikely among these studies of
multiple myeloma mortality. Random measurement error or nondifferential misclassification has
the effect of causing bias toward the null, thereby obscuring potentially real effects by
underestimating their magnitude. This may explain the generally null findings of the four large
studies that relied on very crude assessments of exposure fStellman et al.. 1998: Heineman etal..
1992: Pottern etal.. 1992: Boffetta et al.. 1989).
Confounding is a potential bias that could arise if another cause of multiple myeloma was
also associated with formaldehyde exposure. There does not appear to be any evidence of
confounding that would provide an alternative explanation for the observed association of
formaldehyde exposure with increased risk of multiple myeloma seen in these studies. Known risk
factors for multiple myeloma include age, sex, race, and exposure to benzene fVlaanderen etal..
2011). Chemical, and other coexposures that have not been independently associated with multiple
myeloma are not expected to confound results. Pentachlorophenol is considered to be a likely
carcinogen (U.S. EPA. 2010) and the only study with likely coexposure to pentachlorophenol was
classified as uninformative due to the likelihood of confounding (Robinson et al.. 1987). Risks of
multiple myeloma are higher with advancing age, among men, and the age-adjusted mortality rate
in black Americans was more than twice as high as among white Americans in 2008 (NCI. 2012).
All of the epidemiological studies controlled for age and sex. Only one study reported results
according to race (Hayes etal.. 1990) who reported statistically significant increased risks among
"nonwhites" showing a PMR = 3.69 (95% CI 1.59-7.26).
Benzene was not noted as a coexposure in the studies of workers making grinding wheels
(Edling etal.. 1987b). garment plant workers (Meyers etal.. 2013). or embalmers (Haves etal..
19901 and consequently, would not be expected to be a confounder of those results. In the study of
workers manufacturing plastics, Dell and Teta (1995) examined possible coexposures with benzene
but concluded that there were no obvious common exposures. Benzene exposures were not
This document is a draft for review purposes only and does not constitute Agency policy.
1-484 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
reported in the study of British industrial workers (Coggon etal.. 20031: although, it is a possible
coexposure. However, in a cohort of U.S. industrial workers with similar occupational activities,
benzene was specifically assessed as a potential confounder among the U.S. industrial workers
fBeane Freeman et al.. 20091 and found not to be a confounder.
A single high confidence result supports an association between peak formaldehyde
exposures and increased risks of multiple myeloma (Beane Freeman et al.. 2009) with support from
three results of studies of high peak formaldehyde exposure settings with low to medium
confidence (Dell and Teta. 1995: Hayes etal.. 1990: Edling etal.. 1987bl. However, risk estimates
using other exposure metrics from the same study with the high confidence result fBeane Freeman
etal.. 20091 did not find increased risks and it is not known which metric of exposure is likely to be
the most biologically relevant Bias is unlikely to explain these findings but chance could be an
alternative explanation.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
multiple myeloma placed the greatest weight on five particular considerations: (1) the observations
of increases in risk across one high, one medium, and two low confidence studies of occupational
formaldehyde levels, but limited to groups of people who experienced high peak exposures;
analyses based on other exposure metrics did not report associations in several populations; (2)
the strength of the association showing an approximate 1.2- to 4-fold increase in risk with the
highest quality evidence showing a two-fold increase in risk with high peak exposures; (3) the
limited evidence of an exposure-response trend from a single high confidence study showing that
increased exposure to formaldehyde was associated with increased risk of multiple myeloma; (4)
reasonable confidence that alternative explanations are ruled out, including bias and confounding
within individual studies or across studies, but chance could be an alternative explanation; and (5)
confidence was diminished by reports of inverse relationships with duration of exposure and TSFE.
Given the uncertainties outlined above, and although formaldehyde is genotoxic, the consistent
observations of genotoxicity in peripheral blood lymphocytes observed across several occupational
studies were not interpreted as sufficient to further strengthen the judgment on the human
evidence of multiple myeloma beyond slight.
Conclusion
• The available epidemiological studies provide slight evidence of an association consistent
with causation between formaldehyde exposure and an increased risk of multiple
myeloma—primarily with respect to peak exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
1-485 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
100n
0
4-»
rc
E
(/>
LU
o
£
LU
0
>
_ro
o
o:
i 1
Population-level
exposure
assessment
Individual-level exposure assessment
10-
3-
2-
1-
0.1-
? §
TO £ <5 ££
92 £
5$
&
U)
c
1
0
I?
>
<
V,
fc f
> 13
o cn
si =
High Peak
Exposures
a
a
I £
oj 4!
61
r
Beane Freeman
8 |
P
15
ACS
|studies|
Danish
studies
Meyers
li I
sa J
>
>
>
co
c_
¦ i CO
.
O
1
CO
, ,
, ,
, ,
CO
—
,—,
O)
LO
CO
o
T
T—
CM
LO
CM
T
CM
II
CM
II
ii
II
II
II
II
C
II
II
II
c
c
C
c
c
c
c
i*
a
C7>
00
s
s
CO
CM
CO
CO
CM
$
CO
o
CM
o
d
T—
o
II
c\i
T"
T—
II
II
II
II
II
II
II
(Z
ii
II
II
Cn
cn
a.:
cn
cn
a:
cr
on
Dd
cn
£
2
2
<$>
c59 Sy .cy> ^ <#> qp .cP Kcy> JQP K$>
V ^ <> ^ Vi? ,$p
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-62. Epidemiological studies of formaldehyde exposure and risk of
multiple myeloma
Study
Reference: Beane Freeman et al. (2009)
with supplemental online tables
Population: 25,619 workers employed at
10 formaldehyde-using or formaldehyde-
producing plants in the U.S. followed from
either the plant start-up or first
employment through 2004. Deaths were
identified from the National Death Index
with remainder assumed to be living. 676
workers (3%) were lost to follow-up. Vital
status was 97.4% complete and only 2.6%
lost to follow-up.
Outcome definition: Death certificates
used to determine UCOD from multiple
myeloma (ICD-8: 203).
Design: Prospective cohort mortality study
with external and internal comparison
groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year, age,
sex, and race; adjusted for pay category
compared to workers in lowest exposed
category. Lagged exposures were
evaluated to account for cancer latency.
SMRs calculated using sex, age, race, and
calendar-year-specific U.S. mortality rates.
Related studies:
Blair etal. (1986)
Hauptmann et al. (2003)
Confidence in effect estimates:3
SB IB a Oth
Ove ra 11
Confidence
High
HIGH
IB: Exposure Group A
Exposures
Exposure assessment: Individual-level
exposure estimates based on job titles,
tasks, visits to plants by study industrial
hygienists, and monitoring data
through 1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3). Median cumulative
exposure = 0.6 ppm-years (range 0-
107.4).
Multiple exposure metrics including
peak, average, and cumulative
exposures were evaluated using
categorical and continuous data.
Duration and timing: Exposure period
from <1946 to 1980. Median length of
follow-up: 42 years. Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
Peak exposure:
Level 1 (>0 to <2.0 ppm)
Level 2 (2.0 to <4.0 ppm)
Level 3 (>4.0 ppm)
Average intensity:
Level 1 (>0 to <0.5 ppm)
Level 2 (0.5 to <1.0 ppm)
Level 3 (>1.0 ppm)
Cumulative exposure:
Level 1 (>0 to <1.5 ppm-yrs)
Level 2 (1.5 to <5.5 ppm-yrs)
Level 3 (>5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and
evaluated as potential confounders and
found not be confounders.
[As noted in Appendix A.5.9: There was
no information on smoking; however,
according to Blair et al. (1986). "The
lack of a consistent elevation for
tobacco-related causes of death,
however, suggests that the smoking
habits among this cohort did not differ
substantially from those of the general
population."
Results: effect estimate (95% CI)
[# of cases]
Internal comparisons:
Peak exposure
Unexposed RR = 2.74 (1.18-6.37) [11]
Level 1 RR = 1.00 (Ref. value) [14]
Level 2 RR = 1.65 (0.76-3.61) [13]
Level 3 RR = 2.04 (1.01-4.12) [21]
p-trend (exposed) = 0.08;
p-trend (all) >0.50
Average intensity
Unexposed RR = 2.18 (1.01^.70) [11]
Level 1 RR = 1.00 (Ref. value) [25]
Level 2 RR = 1.40 (0.68-2.86) [11]
Level 3 RR = 1.49 (0.73-3.04) [12]
p-trend (exposed) >0.50;
p-trend (all) >0.50
Cumulative exposure
Unexposed RR = 1.79 (0.83-3.89) [11]
Level 1 RR = 1.00 (Ref. value) [28]
Level 2 RR = 0.46 (0.18-1.20) [5]
Level 3 RR = 1.28 (0.67-2.44) [15]
p-trend (exposed) >0.50;
p-trend (all) >0.50
External comparisons:
SMRunexposed = 1.78(0.99-3.22) [11]
SMRexposed = 0.94 (0.71-1.25) [48]
[NB: Checkoway et al. (2015)
below]
This document is a draft for review purposes only and does not constitute Agency policy.
1-487 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Beane Freeman et al.
(2009) as re-analyzed by Checkoway
et al. (2015) with differences noted
Population: No differences.
Outcome definition: Death certificates
used to determine UCOD from acute and
chronic myeloid leukemia (ICD-8: 205.0
and 205.1).
Design: No differences.
Analysis: HRs estimated using Cox
proportional hazards models controlling
for age, sex, and race; adjusted for pay
category compared to workers in the
redefined lowest exposed category. Did
not control for calendar year as did
Beane Freeman et al. (2009).
Lagged exposures were evaluated to
account for cancer latency.
SMRs calculated using sex, age, race, and
calendar-year-specific U.S. mortality rates.
Related studies:
Blair et al. (1986)
Hauptmann et al. (2003)
Checkoway et al. (2015) [reviewed
here]
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW • (Potential bias
IB: Exposure Group A [from Beane
Freeman et al. (2009)1 (Appendix A.5.9)
downgraded to Group D based on authors'
decision to reclassify all peak exposures
<2 ppm as unexposed and to reclassify
peak exposures >2 ppm as unexposed—if
they were either very rare or very
common.
Checkoway
External comparisons:
SMRunexposed = 1.82(1.01-3.29) [11]
SMRexposed = 0.93 (0.70-1.24) [48]
Reference: Coggon et al. (2014)
Population: 14,008 British men employed
in six chemical industry factories that
Exposure assessment: Exposure
assessment based on data abstracted
from company records. Jobs
External comparisons:
SMR = 0.99 (0.66-1.43) [28]
Within-study external comparisons:
This document is a draft for review purposes only and does not constitute Agency policy.
1-488 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
produced formaldehyde. Cohort mortality
followed from 1941 through 2012. Cause
of deaths was known for 99% of 5,185
deaths through 2000. Similar cause of
death information not provided on 7,378
deaths through 2012. Vital status was
98.9% complete and only 1.1% lost to
follow-up through 2003. Similar
information not provided on deaths
through 2012.
Outcome definition: Death certificates
used to determine cause of deaths from
multiple myeloma (ICD-9: 203).
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2003)
Confidence in effect estimates:3
SB IE a Oth
Overall
Co nfi d e n c e
Medium
MEDIUM >1/
(Potential bias toward the nullvU)
IB: Exposure Group B; latency was not
evaluated.
categorized as background, low,
moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration
was evaluated as more, or less, than
1 year only among the high exposure
group. Timing since first exposure was
not evaluated.
Variation in exposure:
Highest exposure level attained
Level 1 (Background)
Level 2 (low/moderate)
Level 3 (High)
Duration of high exposures
Level 1 (<1 year)
Level 2 (1 year or more)
Coexposures: Not evaluated as
potential confounders. Potential low-
level exposure to stvrene. ethylene
oxide, epichlorhydrin, solvents,
asbestos, chromium salts, and
cadmium.
[As noted in Appendix A.5.9: Stvrene is
associated with LHP cancers.
Asbestos is associated with URT
cancers, but not with LHP cancers.
Other coexposures are not known risk
factors for this outcome.
Authors stated that the extent of
coexposures was expected to be low.
Potential for confounding may be
mitigated by low coexposures.]
Highest exposure level attained
Level 1 SMR = 0.31 (0.06-0.91) [3]
Level 2 SMR = 1.47 (0.82-2.43) [15]
Level 3 SMR = 1.18 (0.57-2.18) [10]
Reference: Meyers et al. (2013)
Population: 11,043 workers in three U.S.
garment plants exposed for at least
3 months. Women comprised 82% of the
cohort. Vital status was followed through
2008 with 99.7% completion
Outcome definition: Death certificates
used to determine both the UCOD from
myeloid leukemia (ICD code in use at time
of death).
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and
1984. Geometric TWA8 exposures
ranged from 0.09 to 0.20 ppm. Overall
geometric mean concentration of
formaldehyde was 0.15 ppm (GSD
1.90 ppm). Area measures showed
constant levels without peaks.
Historically earlier exposures may have
been substantially higher.
Duration and timing: Exposure period
from 1955 through 1983. Median
External comparisons:
SMR = 1.24 (0.79-1.86)
[23]
Within-study external comparisons:
Duration of exposure:
Level 1 SMR = 1.16 (0.50-2.29) [8]
Level 2 SMR = 2.03 (1.01-3.64) [11]
Level 3 SMR = 0.64 (0.17-1.64) [4]
Time since first exposure (TSFE):
Level 1 SMR = 1.73 (0.04-9.61) [1]
Level 2 SMR = 1.63 (0.34-4.76) [3]
Level 3 SMR = 1.18 (0.71-1.84) [19]
This document is a draft for review purposes only and does not constitute Agency policy.
1-489 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Design: Prospective cohort mortality study
with external and internal comparison
groups.
Analysis: SMRs calculated using sex, age,
race, and calendar-year-specific U.S.
mortality rates.
Related studies:
Stayner et al. (1985)
Stavner et al. (1988)
Pinkerton et al. (2004)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM >1/
(Potential bias toward the nullvU)
IB: Exposure Group A; latency was not
evaluated.
duration of exposure was 3.3 years.
More than 40% exposures <1963.
Median time since first exposure was
39.4 years. Duration and timing since
first exposure were evaluated.
Variation in exposure:
Duration of exposure:
Level 1 (<3 years)
Level 2 (3-9 years)
Level 3 (10+ years)
Time since first exposure:
Level 1 (<10 years)
Level 2 (10-19 years)
Level 3 (20+ years)
Coexposures: Study population
specifically selected because industrial
hygiene surveys at the plants did not
identify any chemical exposures other
than formaldehyde that were likely to
influence findings.
Year of first exposure:
<1963 SMR= 1.28 (0.71-2.11) [15]
1963-70 SMR = 0.81 (0.22-2.08) [4]
1971+ SMR = 2.16 (0.59-5.52) [4]
Internal comparisons:
Duration of exposure:
Level 1 SRR = 1.00 (Ref. value) [8]
Level 2 SRR = 1.22 (0.46-3.26) [11]
Level 3 SRR = 0.28 (0.08-0.99) [4]
Reference: Hauptmann et al. (2009)
Population: 6,808 embalmers and funeral
directors who died during 1960-1986.
Identified from registries of the National
Funeral Directors' Association, licensing
boards and state funeral directors'
associations, NY State Bureau of Funeral
Directors, and CA Funeral Directors and
Embalmers. Deaths were identified from
the National Death Index. Next of kin
interviews conducted for 96% of cases and
94% of controls.
Outcome definition: Death certificates
used to determine UCOD from multiple
myeloma (ICD-8: 203).
Design: Nested case-control study within a
prospective cohort mortality study using
two internal comparison groups; the first
composed of those who had never
embalmed (one case and 55 controls) and
the second composed of those who had
fewer than 500 embalmings (5 cases and
83 controls).
Analysis: ORs calculated using
unconditional logistic regression adjusted
for date of birth, age at death, sex, data
source, and smoking. Lagged exposures
Exposure assessment: Occupational
history obtained by interviews with
next of kin and coworkers using
detailed questionnaires.
Exposure was assessed by linking
questionnaire responses to an
exposure assessment experiment
providing measured exposure data.
Exposure levels (peak, intensity, and
cumulative) were assigned to each
individual using a predictive model
based on the exposure data. The
model explained 74% of the observed
variability in exposure measurements.
Multiple exposure metrics including
duration (mean = 33.1 yrs in cases), # of
embalming, peak, average, and
cumulative exposures were evaluated
using categorical and continuous data.
Duration and timing: Exposure period
from <1932 through 1986. Year of
birth ranged from 1876 through 1959.
Year of deaths ranged from 1960
through 1986. Duration of exposure
was evaluated. Duration is also a
surrogate for time since first exposure
since dates of death were closely
External comparisons:
Ever embalming: OR = 1.4 (0.4-5.6)
[# not given]
This document is a draft for review purposes only and does not constitute Agency policy.
1-490 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
were evaluated to account for cancer
latency.
Results from the second internal
comparison group with <500 embalmings
were selected to increase statistical
stability.
Related studies:
Hayes et al. (1990)
Walrath and Fraumeni (1983)
Walrath and Fraumeni (1984)
Note: The original cohorts from these
three related studies were combined in
Hauptmann et al. (2009) and follow-up
was extended so the case-series overlap
and are not independent.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM -i,
(Potential bias toward the nullvU)
IB: Exposure Group A; latency was not
evaluated.
Reference: Haves et al. (1990)
Population: 4,046 deceased U.S. male
embalmers and funeral directors, derived
from licensing boards and funeral director
associations in 32 states and the District of
Columbia who died during 1975-1985.
Death certificates obtained for 79% of
potential study subjects (n = 6,651) with
vital status unknown for 21%.
Outcome definition: Death certificates and
licensing boards used to determine cause
of death from multiple myeloma (ICD-8:
205).
Design: Proportionate mortality cohort
study with external comparison group.
Analysis: PMRs calculated using sex, race,
age, and calendar-year-expected numbers
of deaths from the U.S. population.
Exposures
related to cessation of workplace
exposures
Variation in exposure:
Ever/never
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol, glutaraldehyde,
mercury, arsenic, zinc, and ionizing
radiation.
Chemical coexposures are not known
risk factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Exposure assessment: Presumed
exposure to formaldehyde tissue
fixative. Exposure based on
occupation, which was confirmed on
death certificate. Authors
subsequently measured personal
embalming exposures ranging from
0.98 ppm (high ventilation) to 3.99 ppm
(low ventilation) with peaks up to
20 ppm.
Authors state that major exposures are
to formaldehyde and possibly
glutaraldehyde and phenol.
Duration and timing: Occupational
exposure preceding death during
1975-1985. Of 115 deaths from LHP
cancer, 66 (57%) were aged 60-
74 years. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as
potential confounders.
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
PMR= 1.37(0.84-2.12) [20]
Additional:
By Race
White PMR = 0.97 (0.50-1.69) [12]
Nonwhite PMR = 3.69 (1.59-7.26) [8]
This document is a draft for review purposes only and does not constitute Agency policy.
1-491 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
[As noted in Appendix A.5.9:
Coexposures may have included:
phenol, methyl alcohol, glutaraldehyde,
mercury, arsenic, zinc, and ionizing
radiation.
MEDIUM >1/
(Potential bias toward the nullvU)
SB: Missing death certificates considered
to missing at random.
IB: Exposure: Group A; latency not
evaluated.
Chemical coexposures are not known
risk factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Reference: Pira et al. (2014)
Population: 2,750 workers employed at a
laminated plastic factory in Italy for at
least 180 days between 1947 and 2011
followed until May 2011. Deaths were
identified from population registries. Vital
status was 96.9% complete and only 3.1%
lost to follow-up.
Outcome definition: Death certificates
used to determine UCOD from multiple
myeloma (ICD-9: 203).
Design: Prospective cohort mortality study
with external comparison group.
Analysis: RRs estimated using Poisson
regression stratified by calendar year, age,
sex, and race; adjusted for pay category
compared to workers in lowest exposed
category. Lagged exposures were
evaluated to account for cancer latency.
SMRs calculated using sex, age, and 5-year
calendar periods using mortality rates
from the Piedmont region.
Confidence in effect estimates:3
SE IB Cf Oth
Overall
Confidence
Low
Low (Potential bias toward the null, low
sensitivity)
SB: Healthy worker effect possible
IB: Exposure Group B (Appendix A.5.9)
Oth: Low power
Exposure assessment: Formaldehyde is
a byproduct from the resins used in
production process and all workers
were presumed to have been exposed.
Duration and timing: Exposure period
from 1947 through 2011. Median
length of follow-up: 23.6 years.
Duration and timing since first
exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated
External comparisons:
Observed: 0 multiple myeloma deaths
Expected: 2 multiple myeloma deaths
Myeloid Leukemia (ICD-9: 205)
SMR = 0(0-1.50)+ [0]
+Note: EPA derived CIs using the Mid-P
Method [See Rothman and Boice
(1979)1
This document is a draft for review purposes only and does not constitute Agency policy.
1-492 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Stellman et al. (1998)
Population: 317,424 U.S. men enrolled in
the American Cancer Society's Cancer
Prevention Study II during 1982 with
sufficient data on occupation. Cohort
mortality followed until August 1988 with
98% complete follow-up.
Outcome definition: Death certificates
used to determine cause of deaths from
multiple myeloma (ICD-9: 203).
Design: Prospective cohort study with
internal comparison group.
Analysis: RR calculated using Poisson
regression controlling for sex, age,
smoking.
Confidence in effect estimates:3
Exposure assessment: Individual-level
exposure ascertained from
questionnaire on occupation with
specific exposure to formaldehyde
based on checkbox. Formaldehyde
analyses limited to workers not in
wood-related occupations.
Duration and timing: Occupational
exposures prior to 1982. Timing of
formaldehyde exposure not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Wood dust excluded.
[As noted in Appendix A.5.9:
Coexposures included: asbestos and
wood dust.
Internal comparisons:
RR = 0.74 (0.27-2.02)
[4]
SB IB Cf Oth
Overall
Confidence
Low
However, these coexposures are not
associated with LHP endpoints so
confounding is unlikely.]
LOW >1/
(Potential bias toward the nullvU)
IB: Exposure Group C; latency was not
evaluated.
Oth: Low power
Reference: Band et al. (1997)
Population: 30,157 male workers with at
least 1 year of employment accrued by
January 1950. Followed through
December 1982. Loss to follow-up was
less than 6.5% for workers exposed to the
sulfate process (67% of original cohort of
30,157) and less than 20% for workers
exposed to the sulfite process.
Outcome definition: Cause of death
obtained from the National Mortality
Database based on ICD version in effect at
time of death and standardize to ICD-9
version; multiple myeloma (ICD-9 203).
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex, race,
age, and calendar-year-expected numbers
of deaths from the Canadian population.
Exposure assessment: Occupational
data limited to hire and termination
dates for all workers and type of
chemical process of pulping (sulfate vs.
sulfite). No job-specific data available.
Presumed exposure to formaldehyde
known to be used in the plant.
Formaldehyde is known to be an
exposure for pulp and paper mill
workers: job-specific median exposures
ranging from 0.04 to 0.4 ppm with
peaks as high as 50 ppm (Korhonen et
al., 2004).
External comparisons:
All workers
SMR = 0.80 (90% CI 0.48-1.29) [12]
Duration and timing: Duration and
timinge since first exposure were not
evaluated.
Variation in exposure:
No variation in formaldehyde exposure
was reported. Results presented by
pulping process (sulfate vs. sulfite) but
This document is a draft for review purposes only and does not constitute Agency policy.
1-493 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Confidence in effect estimates:3
SB IE Cf Oth
Overall
Confidence
Low
LOW
(Potential bias toward the null >|,)
IB: Exposure Group C
Cf: Potential confounding
there is no information on differential
exposures between the two processes.
Coexposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9: Potential
confounders for these outcomes
include chlorophenols, acid mists,
dioxin. and perchloroethvlene and
would likely be positively correlated
with formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.]
Reference: Dell and Teta (1995)
Population: 5,932 men employed at a New
Jersey plastics manufacturing plant for at
least 7 months during 1946-1967. Cohort
mortality followed through 1988.
Vital status was 94% complete and only 6%
lost to follow-up. Death certificates
obtained for 98%.
Outcome definition: Death certificates
used to determine UCOD from multiple
myeloma based on ICD code at time of
death.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex, race,
age, and calendar-year-expected numbers
of deaths from the U.S. and local
populations.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW (low sensitivity)
IB: Exposure Group C
Cf: Potential confounding
Oth: Low power due to rarity of exposure
Exposure assessment: Presumed
exposure to formaldehyde known to be
used in the plant. Only 111 men had
assignments involving formaldehyde.
Duration and timing: Exposures during
1946-1967. Duration and timing since
first exposure were not evaluated.
Variation in exposure:
By department: Plant Services and
Research and Development.
By pay status: salaried and hourly.
Coexposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9
coexposures include: acrylonitrile,
asbestos, benzene, carbon black,
epichlorohydrin, PVC (vinyl chloride),
stvrene. and toluene and would likely
be positively correlated with
formaldehyde exposure.
Asbestos is not associated with LHP
cancers.
Benzene and styrene were not
evaluated as potential confounders and
would likely be positively correlated
with formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.]
External comparisons:
All salaried workers
SMR = 2.62 (0.85-6.11) [5]
Research and Development: Hourly
workers
SMR = 2.73 (0.55-7.97) [3]
This document is a draft for review purposes only and does not constitute Agency policy.
1-494 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Reference: Pottern et al. (1992)
Population: Danish women registered in
both the National Cancer Registry and
pension fund. All women with a specific
occupational history other than
"homemaker" were included.
Outcome definition: Incident cases of
multiple myeloma reported to the Danish
Cancer Registry during 1970-1984.
Design: Population-based case-control
study of 363 women with 1,517 age- and
sex-matched controls alive at time of case
diagnosis.
Analysis: ORs calculated for occupation,
industry, and likelihood of exposure using
logistic regression controlling for age.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW >1/
(Potential bias toward the nullvU)
IB: Exposure Group D; latency not
evaluated
Exposure assessment: Individual-level
exposure estimated by industrial
hygienists based on occupation listed
on most recent annual income tax
documents and the industry associated
with that occupation.
Duration and timing: Exposure period
preceding cancer incidence (<1984).
Duration and timing since first
exposure were not evaluated.
Variation in exposure:
Likelihood of exposure:
Level 1 (unexposed)
Level 2 (possible)
Level 3 (probable)
Coexposures: Many other compounds
were identified and evaluated as
independent risk factors.
[As noted in Appendix A.5.9: Other
exposures evaluated included 19
categories grouping 47 substances.
Coexposures were not evaluated for
confounding but exposure to organic
solvents (including benzene) and
radiation were not risk factors for
multiple myeloma so confounding is
unlikely.]
Internal comparisons:
Likelihood of exposure
Level
1
RR= 1.0
(Ref. value)
[303]
Level
2
RR= 1.1
(0.8-1.6)
[56]
Level
3
RR= 1.6
(0.4-5.3)
[4]
Reference: Heineman et al. (1992)
Population: Danish men registered in both
the National Cancer Registry and pension
fund. All men with a specific occupational
history were included.
Outcome definition: Incident cases of
multiple myeloma reported to the Danish
Cancer Registry during 1970-1984. 92% of
cases were histologically confirmed.
Design: Population-based case-control
study of 1,098 men with 4,169 age- and
sex-matched controls alive at time of case
diagnosis.
Analysis: ORs calculated for occupation,
industry, and likelihood of exposure using
logistic regression controlling for age.
Exposure assessment: Individual-level
exposure estimated by industrial
hygienists based on occupation listed
on most recent tax documents.
Duration and timing: Exposure period
preceding cancer incidence (<1984).
Duration and timing since first
exposure were not evaluated.
Variation in exposure:
Likelihood of exposure:
Level 1 (unexposed)
Level 2 (possible)
Level 3 (probable)
Coexposures: Other compounds were
identified and evaluated as
independent risk factors including:
gasoline, oil products, engine exhausts,
Internal comparisons:
Likelihood of exposure
1.0 (Ref. value) [913]
Level
1
RR= 1.0
Level
2
RR= 1.0
Level
3
RR= 1.1
1.8-1.3)
[144]
[41]
This document is a draft for review purposes only and does not constitute Agency policy.
1-495 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW -i,
(Potential bias toward the nullvU)
IB: Exposure Group D; latency not
evaluated
Reference: Boffetta et al. (1989)
Population: 508,637 U.S. men and 676,613
women enrolled in the American Cancer
Society's Cancer Prevention Study II during
1982 with sufficient data on occupation.
Cohort mortality followed until August
1986 with 98.5% complete follow-up.
Outcome definition: Death certificates
used to determine cause of deaths from
incident cases of multiple myeloma (ICD-9:
203) since follow-up began.
Design: Population-based matched nested
case-control within prospective cohort
study.
Analysis: RR calculated using Poisson
regression controlling for sex, age,
smoking, education, diabetes, X-ray
treatment, farming, pesticide, and
herbicide exposure.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
IB: Exposure: Group C; lack of latency
analysis
Oth: Low power (few exposed cases)
Reference: Ott et al. (1989)
Population: 29,139 men employed at two
large chemical manufacturing facilities and
a research and development center who
Exposures
benzene, dyes, phthalates, vinyl
chloride, asbestos, and pesticides.
[As noted in Appendix A.5.9: Other
exposures evaluated included 19
categories grouping 47 substances.
Asbestos is not a risk factor for LHP.
"Possible" benzene exposure was
associated with MM but not "probable"
benzene exposure, so confounding is
considered to be unlikely.]
Exposure assessment: Individual-level
exposure ascertained from
questionnaire on occupation with
specific exposure to formaldehyde
based on checkbox.
Duration and timing: Occupational
exposures prior to 1982. Timing of
formaldehyde exposure not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Various coexposures
were controlled for in the analyses.
[As noted in Appendix A.5.9: Matching
controlled for sex, age, ethnic group,
residence, smoking, education,
diabetes, X-ray treatment, farming,
pesticide, and herbicide exposure.
Other coexposures were not associated
with LHP cancers.]
Exposure assessment: Individual-level
exposure ascertained from employee's
work assignments linked to records on
departmental usage of formaldehyde.
Results: effect estimate (95% CI)
[# of cases]
Internal comparisons:
OR = 1.8 (0.6-5.7) [4]
Internal comparisons:
OR = 1.0 (0.05-4.9) [1]
This document is a draft for review purposes only and does not constitute Agency policy.
1-496 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
worked during 1940-1978. Vital status
was known for 96.4%. Death certificates
were available for 5,785 known
descendants (95.4%).
Outcome definition: Death certificates
used to determine UCOD from multiple
myeloma based on the ICD code in used at
the time of death.
Design: Nested case-control study within a
prospective cohort mortality study.
Twenty cases of multiple myeloma were
frequency matched to 100 controls on
time from hire to death.
Analysis: ORs calculated using
unconditional logistic regression.
Related studies:
Rinskv etal. (1988)
Confidence in effect estimates:3
SB IE Cf Oth
Overall
Confidence
Low
Duration and timing: Occupational
exposures during 1940-1978. Timing
of formaldehyde exposure not
evaluated.
Variation in exposure: Ever/never
Coexposures: None evaluated as
potential confounders.
[As noted in Appendix A.5.9: 21
different chemicals were evaluated
including benzene with much cross
exposure.
Benzene was not evaluated as a
potential confounder and may be
positively correlated with
formaldehyde exposure.
Potential for confounding is unknown
but could have inflated the observed
effect.
Potential for confounding may be
mitigated by rarity of coexposures
among cases.]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
LOW
(Potential bias toward the nullvU)
IB: Exposure Group B; latency evaluation
likely to be underpowered to detect any
effects beyond a 5-year period
Cf: Benzene is a potential confounder
IB: Low power due to the rarity of
exposure
Reference: Edling et al. (1987b)
Population: 521 Swedish male blue collar
workers in an abrasive production plant
with at least 5 years of employment
between 1955 and 1983. Cohort mortality
followed through 1983 with 97% known
vital status.
Outcome definition: Cancer mortality
ascertained using UCOD from the National
Death Registry. Cancer incidence
ascertained from the National Cancer
Registry. Mortality and incidence of
multiple myeloma based on ICD-8:203.
Design: Cohort mortality and incidence
study with external comparison group.
Exposure assessment: Manufacture of
grinding wheels bound by
formaldehyde resins exposed workers
to 0.1-1 mg/m3 formaldehyde; 59
workers manufacturing abrasive belts
had low exposure to abrasives with
intermittent exposures with peaks up
to 20-30 mg/m3 formaldehyde.
Duration and timing: Exposures during
1955-1983. Duration and timing since
first exposure were evaluated.
Variation in exposure: Not evaluated.
Coexposures: Aluminum oxide and
silicon carbide were coexposures but
were not evaluated as confounders.
External comparisons:
Cancer mortality
No increase reported
Cancer Incidence
SMR = 4.0 (0.67-13.2)+ [2]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
This document is a draft for review purposes only and does not constitute Agency policy.
1-497 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Analysis: SMRs calculated using sex, age,
and calendar-year-specific Swedish
mortality rates.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW -i,
(Potential bias toward the nullvU)
IB: Exposure: Group B; latency not
evaluated
Oth: Low power
[As noted in Appendix A.5.9:
Coexposures are not known risk factors
for this outcomes.]
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Results from low confidence studies are shaded; these findings are considered less reliable.
Abbreviations: SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or analysis;
UCOD = underlying cause of death; GSD = geometric standard deviation; SMR = standardized mortality ratio; RR = relative risk;
TWA8 = 8-hour time-weighted average; URT = upper respiratory tract; LHP = lymphohematopoietic; PMR = proportionate
mortality ratio; BMI = body mass index; JEM = job-exposure matrix; OR = odds ratio.
Hodgkin lymphoma
Epidemiological evidence
The most specific level of Hodgkin lymphoma diagnosis that is commonly reported across
the epidemiological literature has been based on the first three digits of the Eighth or Ninth
Revision of the ICD code (i.e., Hodgkin disease ICD-8/9: 201). Evidence describing the association
between formaldehyde exposure and the specific risk of Hodgkin lymphoma was available from 15
epidemiological studies—one case-control study fGerin etal.. 19891 and 14 cohort studies fMevers
etal.. 2013: Beane Freeman etal.. 2009: Coggon etal.. 2003: Band etal.. 1997: Andjelkovich etal..
1995: Hansen and Olsen. 1995: Hall etal.. 1991: Hayes etal.. 1990: Matanoski. 1989: Soletetal..
1989: Robinson etal.. 1987: Stroup etal.. 1986: Walrath and Fraumeni. 1984.19831. Study details
are provided in the evidence table for Hodgkin lymphoma (see Table 1-63). The outcome-specific
evaluations of confidence in the reported effect estimate of an association from each study are
provided in Appendix A.5.9 and the confidence conclusions are provided in the evidence table for
Hodgkin lymphoma (see Table 1-63) following the causal evaluation.
Note that the confidence judgments are for the confidence in the reported effect estimate of
an association from each study and not a confidence judgment in the overall study. Three sets of
This document is a draft for review purposes only and does not constitute Agency policy.
1-498 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
reported results from Hall et al. (1991). Solet et al. (1989). and Matanoski (19891 were classified as
uninformative due to multiple biases and uncertainties; for details see Appendix A.5.9.
Consistency of the observed association
The results of the 12 informative studies were not consistent. The study of the largest
cohort of formaldehyde-exposed workers (Beane Freeman et al.. 2009) reported an elevated risk of
dying from Hodgkin lymphoma for the cohort as a whole (SMR = 1.42; 95% CI 0.96-2.1; 27 cases)
and a pronounced increase in risk among those workers with the highest peak formaldehyde
exposures (RR = 3.96; 95% CI; 1.31-12.02; 11 cases)—results that were classified with medium
confidence. However, the other medium confidence result from Gerin et al. (1989) was an OR = 0.5
(95% CI 0.2-1.2; 8 cases). The results of the other 10 studies (all low confidence) were largely
based on small numbers of cases and yielded generally unstable CIs surrounding the RR (see
Figure 1-42).
Compared with other LHP cancers, the 5-year survival rate for Hodgkin lymphoma is
relatively high at 86% and mortality is rare. In contrast, the survival rate for myeloid leukemia is
38%. The high survival rate for Hodgkin lymphoma may indicate that mortality data are not as
good a proxy for incidence data for this LHP cancer subtype. In this instance, these mortality data
are potentially inadequate to evaluate causation. The low mortality rate for Hodgkin lymphoma
results in few exposed cases and very low statistical power, which may have contributed to the
apparently discordant results. Aside from the Beane Freeman et al. (2009) result (medium
confidence), which reported 25 exposed deaths from Hodgkin lymphoma, only one other cohort
study observed more than 10 deaths from Hodgkin lymphoma among exposed subjects (Hansen
and Olsen. 19951: this study reported 12 observed deaths against 12 expected deaths—a result
classified with low confidence.
The study results presented in Table 1-63 (by confidence level and publication date) detail
all of the reported associations between exposures to formaldehyde and the risks of developing or
dying from Hodgkin lymphoma along with a summary graphic of any major limitation and the
confidence classification of the effect estimate. Results are plotted in Figure 1-42.
Strength of the observed association
Summary effect estimates for the association between formaldehyde exposure and Hodgkin
lymphoma were highly variable and the risk of developing or dying from Hodgkin lymphoma were
predominantly less than one (unity) and ranged from zero to 4.0 (Edlinget al.. 1987b). While the
summary effect estimate from the study by Beane Freeman et al. (2009) was RR = 1.42 (95% CI
0.96-2.10), the strength of the association was substantially higher among those workers exposed
to the highest peak levels (RR = 3.96). Beane Freeman et al. (2009) further showed plots
presenting the RR from the internal analyses for each endpoint and for each year of follow-up. The
association of Hodgkin lymphoma with formaldehyde exposure is not only seen for the complete
2004 follow-up when the average length of follow-up was 42 years, but throughout the cohort
This document is a draft for review purposes only and does not constitute Agency policy.
1-499 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
experience (see Beane Freeman et al.. 20091 (Figure 1). These plots show that during the 1970s
and 1980s, the RR * 8 and remained elevated at about RR = 4 through the end of follow-up in 2004.
Such a consistent finding of a strong effect over many years of follow-up reduces the possibility that
the results for the full follow-up period could be due to chance.
Temporal relationship of the observed association
In each of the studies, the formaldehyde exposures among the study participants occurred
before their Hodgkin lymphoma was detected and in the studies that ascertained individual-level
exposures, the estimation of formaldehyde exposures was based on job titles and was done in a
blinded fashion with respect to outcome status. Only one study (Band etal.. 19971 reported on
analyses of the temporal relationship showing that risks were highest in workers with 15 or more
years since first formaldehyde exposure and 15 or more years of exposure duration (SMR = 1.62;
95% CI 0.55-3.71). However, this finding is without corroboration for Hodgkin lymphoma.
Exposure-response relationship
Only two studies evaluated any other form of exposure-response for increasing measures of
formaldehyde exposure fBeane Freeman et al.. 2009: Coggon etal.. 20031. Coggon et al. (2003.)
reported a lower risk of dying from Hodgkin lymphoma among "highly" exposed workers based on
a single death. Beane Freeman et al. (2009) reported a clear exposure-response relationship
between increasing levels of peak formaldehyde and increased risk of dying from Hodgkin
lymphoma among exposed workers (p = 0.01). Compared to exposed workers in the lowest
exposure category of peak exposure, those in the middle category were at more than two-fold
higher risk (RR = 3.30; 95% CI 1.04-10.50) while those workers in the highest category were at
four-fold higher risk (RR = 3.96; 95% CI 1.31-12.02). Beane Freeman etal. (2009) also reported
exposure-response relationships between increased risk of dying from Hodgkin lymphoma among
exposed workers based on average formaldehyde intensity (OR range: 1.61-2.48; p = 0.05) and
cumulative exposure (OR range: 1.30-1.71; p = 0.08).
Potential impact of selection bias; information bias; confounding bias, and chance
Selection bias is an unlikely bias in the epidemiological studies of Hodgkin lymphoma as the
one case-control study was population-based and used other cancer cases as controls with
exposure status evaluated without regard to outcome status and had a participation level of 83%.
Each of the cohort studies included at least 72% of eligible participants and lost fewer than 9% of
participants over the course of mortality follow-up.
The healthy worker effect including the healthy worker survivor effect could obscure a truly
larger effect of formaldehyde exposure in analyses based on "external" comparisons with mortality
in the general population fMevers etal.. 2013: Beane Freeman et al.. 2009: Coggon etal.. 2003:
Band etal.. 1997: Andielkovich et al.. 1995: Hansen and Olsen. 1995: Hayes etal.. 1990: Robinson et
This document is a draft for review purposes only and does not constitute Agency policy.
1-500 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
al.. 1987: Stroup etal.. 1986: Walrath and Fraumeni. 1984.19831. but would not influence analyses
using "internal" or matched comparison groups fBeane Freeman et al.. 2009: Gerin etal.. 19891.
Information bias is unlikely to have resulted in bias away from the null—especially as the
exposure assessment in these studies were generally of high quality; however, random
measurement error or nondifferential misclassification is almost certain to have resulted in some
bias toward the null among these studies of Hodgkin lymphoma.
Chemical exposures that have not been independently associated with Hodgkin lymphoma
are not expected to confound results. The main support for concluding there is a slight association
of formaldehyde exposure with increased risk of Hodgkin lymphoma is from the results for peak
exposures reported by Beane Freeman et al. (2009) who specifically examined the potential for
confounding from 11 substances including benzene and found that controlling for these exposures
did not meaningfully change the results. This provides evidence against potential confounding by
these coexposures. There does not appear to be any evidence of confounding that would provide an
alternative explanation for the observed association of formaldehyde exposure with increased risk
of Hodgkin lymphoma reported by Beane Freeman et al. (20091. The evidence of an association
with peak exposures reported by Beane Freeman et al. (2009) suggests an association whose risk
increases with greater exposure.
Causal evaluation
The causal evaluation for formaldehyde exposure and the risk of developing or dying from
Hodgkin lymphoma placed the greatest weight on the following particular considerations: (1) the
statistically robust evidence of increased risk of Hodgkin lymphoma in the highest peak exposure
group among industrial workers, with a clear exposure-response relationship observed in one
medium confidence study; (2) the consistent pattern of null results across 10 other studies, many of
which had fewer than five exposed cases; (3) the high survival rate for Hodgkin lymphomas (86%),
which may indicate that mortality data are not as good a proxy for incidence data for this LHP
cancer subtype; and (4) the absence of evidence to evaluate the potential risk to sensitive
populations or lifestages. Although consistent observations of genotoxicity in peripheral blood
lymphocytes have been observed across several occupational studies, these data were not
interpreted as sufficient to further strengthen the judgment on the human evidence of Hodgkin
lymphoma.
Conclusion
• The available epidemiological studies provide slight evidence of an association consistent
with causation between formaldehyde exposure and an increased risk of Hodgkin
lymphoma.
This document is a draft for review purposes only and does not constitute Agency policy.
1-501 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Individual-level exposure assessment
o
+¦»
<0
E
o
<1)
LU
CD
>
JS
CD
or
100
10
0.1
~i
Population-level exposure assessment
Anatomists. Embalmers.
Pathologists
, . o
£
V
tn
&
$
£
3
|
93
CC
a
s
03
c
LU
W
to
3
"5:
LU
£
<
re
gj
o>
.c
O)
c
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Population: 25,619 workers employed at
10 formaldehyde-using or formaldehyde-
producing plants in the U.S. followed from
either the plant start-up or first
employment through 2004. Deaths were
identified from the National Death Index
with remainder assumed to be living. Vital
status was 97.4% complete and only 2.6%
lost to follow-up.
Outcome definition: Death certificates
used to determine underlying cause of
death from Hodgkin disease (ICD-8: 201).
Design: Prospective cohort mortality study
with external and internal comparison
groups.
Analysis: RRs estimated using Poisson
regression stratified by calendar year, age,
sex, and race; adjusted for pay category
compared to workers in lowest exposed
category. Lagged exposures were
evaluated to account for cancer latency.
SMRs calculated using sex, age, race, and
calendar-year-specific U.S. mortality rates.
Related studies:
Blair etal. (1986)
Hauptmann et al. (2003)
Confidence in effect estimates:3
sb ie a oth
Overall
Confidence
High
HIGH • (No appreciable bias)
IB: Exposure Group A; higher survival rates
tasks, visits to plants by study industrial
hygienists, and monitoring data from 1966
through 1980.
Median TWA (over 8 hours) = 0.3 ppm
(range 0.01-4.3).
Median cumulative exposure = 0.6 ppm-
years (range 0-107.4).
Multiple exposure metrics including peak,
average, and cumulative exposures were
evaluated using categorical and continuous
data.
Duration and timing: Exposure period from
<1946 through 1980. Median length of
follow-up: 42 years. Duration and timing
since first exposure were evaluated.
Variation in exposure:
For all variations in exposure:
Level 1 (unexposed)
Peak exposure:
Level 2 (>0 to <2.0 ppm)
Level 3 (2.0 to <4.0 ppm)
Level 4 (>4.0 ppm)
Average intensity:
Level 2 (>0 to <0.5 ppm)
Level 3 (0.5 to <1.0 ppm)
Level 4 (>1.0 ppm)
Cumulative exposure:
Level 2 (>0 to <1.5 ppm-yrs)
Level 3 (1.5 to <5.5 ppm-yrs)
Level 4 (>5.5 ppm-yrs)
Coexposures: Exposures to 11 other
compounds were identified and evaluated
as potential confounders and found not be
confounders.
[As noted in Appendix A.5.9: There was no
information on smoking, however,
according to Blair et al. (1986), "The lack of
a consistent elevation for tobacco-related
causes of death, however, suggests that the
smoking habits among this cohort did not
differ substantially from those of the
general population."]
1994 Follow-up:
Highest peak RR = 3.30 (0.98-11.10)
(p-trend = 0.04)
2004 Follow-up:
Peak exposure
Level 1 RR = 0.67 (0.12-3.6) [2]
Level 2 RR = 1.00 (Ref. value) [6]
Level 3 RR = 3.30 (1.04-10.50) [8]
Level 4 RR = 3.96 (1.31-12.02) [11]
p-trend (exposed) = 0.01;
p-trend (all) = 0.004
Average intensity
Level 1 RR = 0.53 (0.11-2.66) [2]
Level 2 RR = 1.00 (Ref. value) [10]
Level 3 RR = 2.48 (0.84-7.32) [9]
Level 4 RR = 1.61 (0.73-3.39) [6]
p-trend (exposed) = 0.05;
p-trend (all) = 0.03
Cumulative exposure
Level 1 RR = 0.42 (0.09-2.05) [2]
Level 2 RR = 1.00 (Ref. value) [14]
Level 3 RR = 1.71 (0.66-4.38) [7]
Level 4 RR = 1.30 (0.40-4.19) [4]
p-trend (exposed) = 0.08;
p-trend (all) = 0.06
Duration of exposure
No evidence of association (data not
shown).
Time since first exposure
>0-15 yrs RR = 1.00 (Ref. value)
>15-25 yrs RR = 1.54 (0.42-5.62)
>25-35 yrs RR<1.54
>35 yrs RR < 1.54
External comparisons:
SMRunexposed= 0.70 (0.17-2.80) [2]
SMRexposed = 1.42(0.96-2.10) [25]
Reference: Gerin et al. (1989)
Population: Male residents of Montreal,
Canada aged 35-70 years. 4,510 eligible
Exposure assessment: Individual-level
exposure estimates developed based on a
complete and detailed occupational history
ascertained by interviewers using a
standardized questionnaire. A team of
Internal comparisons:
Compared to other cancers
OR = 0.5 (0.2-1.2) [8]
Compared to population controls
This document is a draft for review purposes only and does not constitute Agency policy.
1-503 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
incident cancer cases were identified
during 1979-1985 from 19 major area
hospitals, which report to the Quebec
Tumor Registry over 97% of all cancer
diagnoses from the Montreal area.
Interviews and questionnaires completed
for 3,726 subjects (83% of eligible cases).
18% of interviews were completed by next
of kin.
Outcome definition: Histologically
confirmed diagnosis of Hodgkin lymphoma
(ICD: 201)
Design: Population-based case-control
study of 53 formaldehyde-exposed men
with Hodgkin lymphoma. Cases were
compared with two groups; first, against
other cancer cases excluding those
diagnosed with lung cancer (n = 2,599),
and second against 533 male population
controls selected from electoral list in the
Montreal area.
Analysis: ORs calculated by levels of a
composite exposure index using logistic
regression controlling for age, ethnic
group, socio-economic status, smoking,
and dirtiness of jobs held (white vs. blue
collar).
Related studies:
Siemiatvcki et al. (1987)
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM ^
(Potential bias toward the nullvU)
IB: Exposure Group B
chemists and hygienists translated each job
into a list of potential formaldehyde
exposures based on their confidence level,
the frequency of exposure, and the duration
of exposure.
Duration and timing: Exposure period
based on occupational histories prior to
cancer diagnosis. Duration of exposure was
evaluated.
Variation in exposure: For cancer sites with
fewer than 30 cases exposed to
formaldehyde, results for the exposure
subgroups were not shown.
Coexposures: Additional occupational and
nonoccupational potential confounders
were included in analyses when the
estimated exposure-disease OR changed by
more than 10%.
OR = 0.5 (0.2-1.4)
[8]
Reference: Meyers et al. (2013)
Population: 11,043 workers in three U.S.
garment plants exposed for at least
3 months. Women comprised 82% of the
cohort. Vital status was followed through
2008 with 99.7% completion
Outcome definition: Death certificates
used to determine both the underlying
Exposure assessment: Individual-level
exposure estimates for 549 randomly
selected workers during 1981 and 1984.
Geometric TWA8 exposures ranged from
0.09 to 0.20 ppm. Overall geometric mean
concentration of formaldehyde was
0.15 ppm (GSD 1.90 ppm). Area measures
showed constant levels without peaks.
Historically earlier exposures may have
been substantially higher.
External comparisons:
SMR = 0.95 (0.26-2.44)
[4]
This document is a draft for review purposes only and does not constitute Agency policy.
1-504 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
cause of death from Hodgkin lymphoma
(ICD code in use at time of death).
Design: Prospective cohort mortality study
with external and internal comparison
groups.
Analysis: SMRs calculated using sex, age,
race, and calendar-year-specific U.S.
mortality rates.
Related studies:
Stavner et al. (1985)
Stayner et al. (1988)
Pinkerton et al. (2004)
Duration and timing: Exposure period from
1955 through 1983. Median duration of
exposure was 3.3 years. More than 40%
exposures <1963. Median time since first
exposure was 39.4 years. Duration and
timing since first exposure were evaluated.
Variation in exposure: Not evaluated.
Coexposures: Study population specifically
selected because industrial hygiene surveys
at the plants did not identify any chemical
exposures other than formaldehyde that
were likely to influence findings.
Confidence in effect estimates:3
SB IB a Oth
Overall
Confidence
Medium
MEDIUM >1/
(Potential bias toward the null)
IB: Exposure Group A; latency not
evaluated.
Oth: Low power
Reference: Coggon et al. (2003)
Population: 14,014 British men employed
in six chemical industry factories that
produced formaldehyde. Cohort mortality
followed from 1941 through 2000. Vital
status was 98.9% complete and only 1.1%
lost to follow-up.
Outcome definition: Death certificates
used to determine cause of deaths from
Hodgkin disease (ICD-9: 201).
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs based on English and
Welsh age- and calendar-year-specific
mortality rates.
Related studies:
Acheson et al. (1984)
Gardner et al. (1993)
Coggon et al. (2014)
Exposure assessment: Exposure assessment
based on data abstracted from company
records. Jobs categorized as background,
low, moderate, high, or unknown levels.
Duration and timing: Occupational
exposure during 1941-1982. Duration and
timing since first exposure were not
evaluated.
Variation in exposure:
TWA exposure
Level 1 (low)
Level 2 (moderate)
Level 3 (high)
Coexposures: Not evaluated as potential
confounders. Potential low-level exposure
to stvrene. ethylene oxide, epichlorhydrin,
solvents, asbestos, chromium salts, and
cadmium.
[As noted in Appendix A.5.9: Stvrene is
associated with LHP cancers.
Asbestos is associated with URT cancers,
but not with LHP cancers.
External comparisons:
SMR = 0.70 (0.26-1.53) [6]
Within-study external comparisons:
Worked in high exposure jobs
SMR = 0.36 (0.01-2.01) [1]
This document is a draft for review purposes only and does not constitute Agency policy.
1-505 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Medium
MEDIUM >1/
(Potential bias toward the nullvU)
IB: Exposure Group B; latency was not
evaluated
Cf: Potential confounding
Other coexposures are not known risk
factors for this outcome.
Authors stated that the extent of
coexposures was expected to be low.
Potential for confounding may be mitigated
by low coexposures.]
Reference: Walrath and Fraumeni
(1983)
Population: 1,132 deceased white male
embalmers licensed to practice during
1902-1980 in New York who died during
1925-1980 identified from registration
files. Death certificates obtained for 75%
of potential study subjects (n = 1,678).
Outcome definition: Hodgkin disease (ICD-
8: 201) listed as cause of death on death
certificates.
Design: Proportionate mortality cohort
study with external comparison group.
Analysis: PMRs calculated using sex, race,
age, and calendar-year-expected numbers
of deaths from the U.S. population.
Confidence in effect estimates:3
Exposure assessment: Presumed exposure
to formaldehyde tissue fixative.
Duration and timing:
Occupational exposure preceding death
during 1902-1980. Median year of birth
was 1901. Median year of initial license
was 1931. Median age at death was 1968.
Expected median duration of exposure was
37 years. Duration and timing since first
exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as potential
confounders.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc, and
ionizing radiation.
External comparisons:
Observed: 2 Hodgkin disease deaths
Expected: 2.3 Hodgkin disease deaths
PMR = 0.87 (0.15-2.87)+
[7]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
SB IB cf oth
Overall
Confidence
Medium
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
MEDIUM >1/
(Potential bias toward the nullvU)
IB: Exposure Group A; latency not
evaluated
Reference: Band et al. (1997)
Population: 30,157 male workers with at
least 1 year of employment accrued by
January 1950. Followed through
December 1982. Loss to follow-up was
less than 6.5% for workers exposed to the
sulfate process (67% of original cohort of
30,157) and less than 20% for workers
exposed to the sulfite process.
Outcome definition: Cause of death
obtained from the National Mortality
Exposure assessment: Occupational data
limited to hire and termination dates for all
workers and type of chemical process of
pulping (sulfate vs. sulfite). No job-specific
data available. Presumed exposure to
formaldehyde known to be used in the
plant. Formaldehyde is known to be an
exposure for pulp and paper mill workers:
job-specific median exposures ranging from
0.04 to 0.4 ppm with peaks as high as
50 ppm (Korhonen et al., 2004)
External comparisons:
All workers
SMR = 0.71 (90% CI 0.33-1.34) [7]
Work duration <15 years
TSFE < 15 years
SMR = 0.53 (90% CI 0.14-1.37) [3]
Work duration >15 years
TSFE > 15 years
SMR = 1.62 (90% CI 0.55-3.71) [4]
This document is a draft for review purposes only and does not constitute Agency policy.
1-506 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Database based on ICD version in effect at
time of death and standardize to ICD-9
version. Hodgkin lymphoma: ICD-9 201
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex, race,
age, and calendar-year-expected numbers
of deaths from the Canadian population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW -i,
(Potential bias toward the nullvU)
IB: Exposure Group C
Cf: Potential confounding
Reference: Andjelkovich et al. (1995)
Population: 3,929 automotive industry
iron foundry workers exposed from 1960
through 1987 and followed through 1989.
Outcome definition: UCOD obtained from
Social Security Administration, Pension
Benefit Informations, and National Death
Index)
Hodgkin lymphoma: ICD 201
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex-, age-,
race-, and calendar-year-specific U.S.
mortality rates.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
IB: Exposure Group B; latency not
evaluated
Oth: Low power.
Reference: Hansen and Olsen (1995)
Exposures
Duration and timing: Duration and timing
since first exposure were evaluated.
Variation in exposure:
No variation in formaldehyde exposure was
reported. Results presented by pulping
process (sulfate vs. sulfite) but there is no
information on differential exposures
between the two processes
Coexposures: Not evaluated as
confounders.
[As noted in Appendix A.5.9: Potential
confounders for these outcomes include
chlorophenols, acid mists, dioxin. and
perchloroethvlene and would likely be
positively correlated with formaldehyde
exposure.
Potential for confounding is unknown but
could have inflated the observed effect.]
Exposure assessment: Individual-level
exposure status (yes/no, quartile) based on
review of work histories by an industrial
hygienist.
Exposure assessment blinded to outcome.
Independent testing of iron foundries by
NIOSH reported a range from 0.02 ppm to
18.3 ppm (cited in WHO (1989) Env.
Health Criteria 89: Formaldehyde).
Duration and timing: Duration and timing
since first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9: Nickel and
chromium are associated with URT but not
LHP.
Other coexposures are not known risk
factors for these outcomes.]
Exposure assessment: Individual
occupational histories including industry
Results: effect estimate (95% CI)
[# of cases]
External comparisons:
SMRunexposed =0.70(0.01—3.88) [1]
SMRexposed =0.72(0.01-4.00) [1]
External comparisons:
This document is a draft for review purposes only and does not constitute Agency policy.
1-507 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Population: 2,041 men with cancer who
were diagnosed during 1970-1984 and
whose longest work experience occurred
at least 10 years before cancer diagnosis.
Identified from the Danish Cancer Registry
and matched with the Danish
Supplementary Pension Fund.
Ascertainment considered complete.
Pension record available for 72% of cancer
Outcome definition: Hodgkin disease (ICD-
7: 201) listed on Danish Cancer Registry
file.
Design: Proportionate incidence study
with external comparison group.
Analysis: Standardized proportionate
incidence ratio calculated as the
proportion of cases for a given cancer in
formaldehyde-associated companies
relative to the proportion of cases for the
same cancer among all employees in
Denmark. Adjusted for age and calendar
time.
Confidence in effect estimates:3
SB IB Cf Oth
a
Overall
Confidence
Low
LOW (Potential bias toward the null
IB: Exposure Group D
and job title established through company
tax records to the national Danish Product
Register.
Subjects were considered to be exposed to
formaldehyde if: (1) they had worked in an
industry known to use more than 1 kg
formaldehyde per employee per year and
(2) subjects longest single work experience
(job) in that industry since 1964 was
>10 years prior to cancer diagnosis.
All subjects were stratified based on job
title as either low exposure (white collar
worker), above background exposure (blue
collar worker), or unknown (job title
unavailable).
Duration and timing: Exposure period not
stated. Based on date of diagnosis during
1970-1984, and the requirement of
exposure more than 10 years prior to
diagnosis, the approximate period was
1960-1974.
Variation in exposure: Not evaluated.
Coexposures: Not evaluated.
[As noted in Appendix A.5.9: While other
coexposures were not evaluated, the
overall correlation between coexposures in
multiple occupational industries is likely to
be low.]
Overall (exposure to formaldehyde
>10 years prior to cancer diagnosis)
SPIR = 1.0(0.5-1.7) [12]
Reference: Haves et al. (1990)
Population: 4,046 deceased U.S. male
embalmers and funeral directors, derived
from licensing boards and funeral director
associations in 32 states and the District of
Columbia who died during 1975-1985.
Death certificates obtained for 79% of
potential study subjects (n = 6,651) with
vital status unknown for 21%.
Outcome definition: Death certificates and
licensing boards used to determine cause
of death from Hodgkin disease (ICD-8:
201).
Design: Proportionate mortality cohort
study with external comparison group.
Exposure assessment: Presumed exposure
to formaldehyde tissue fixative. Exposure
based on occupation, which was confirmed
on death certificate. Authors subsequently
measured personal embalming exposures
ranging from 0.98 ppm (high ventilation) to
3.99 ppm (low ventilation) with peaks up to
20 ppm.
Authors state that major exposures are to
formaldehyde and possibly glutaraldehyde
and phenol.
Duration and timing: Occupational
exposure preceding death during 1975-
1985. Of 115 deaths from LHP cancer, 66
(57%) were aged 60-74 years. Duration and
timing since first exposure were not
evaluated.
External comparisons:
PMR = 0.72 (0.15-2.10)
[3]
This document is a draft for review purposes only and does not constitute Agency policy.
1-508 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Analysis: PMRs calculated using sex, race,
age, and calendar-year-expected deaths
from the U.S. population.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
LOW
(Potential bias toward the nullvU)
SB: Missing death certificates considered
to be missing at random
IB: Exposure: Group A; latency not
evaluated
Oth: Low power
Variation in exposure: Not evaluated.
Coexposures: None evaluated as potential
confounders.
[As noted in Appendix A.5.9:
Coexposures may have included: phenol,
methyl alcohol, glutaraldehyde, mercury,
arsenic, zinc, and ionizing radiation.
Chemical coexposures are not known risk
factors for this outcome.
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Reference: Robinson et al. (1987)
Population: 2,283 plywood mill workers
employed at least one year during 1945-
1955 followed for mortality until 1977 with
vital status for 98% and death certificates
for 97% of deceased.
Outcome definition: Death certificates
used to determine UCOD from Hodgkin
disease as coded by trained nosologist
using ICD-7:201.
Design: Prospective cohort mortality study
with external comparison group. A
subcohort of 818 men coexposed to
formaldehyde and pentachlorophenol
were also evaluated.
Analysis: SMRs calculated using sex, age,
race, and calendar-year-specific U.S.
mortality rates.
Confidence in effect estimates:3
SB IB Cf Oth
Overall
Confidence
Low
Exposure assessment: Presumed exposure
to formaldehyde-based glues used to
manufacture and patch plywood.
Subcohort of 818 men coexposed to
formaldehyde and pentachlorophenol
worked for one year or more in the relevant
exposure categories of veneer pressing and
drying, glue mixing, veneer and panel gluing
and patching.
Duration and timing: Exposures during
1945-1955. Duration and timing since first
exposure were not evaluated.
Variation in exposure:
Duration of exposure
Latency (time since first exposure)
Coexposures: Pentachlorophenol
[As noted in Appendix A.5.9: EPA concluded
that pentachlorophenol is likely to be
carcinogenic based on strong evidence from
epidemiological studies of increased risk of
multiple myeloma.
Pentachlorophenol is not a known risk
factor for Hodgkin lymphoma and thus is
not expected to be a confounder.]
External comparisons:
Whole cohort of mill workers In = 2.283)
SMR = 1.11(0.20-3.50) [2]
Subcohort of highly exposed workers
In = 818)
SMR = 3.33(0.59-10.49) [2]
LOW
(Potential bias toward the nullvU)
SB: Healthy worker effect possible
IB: Exposure Group D; latency not
evaluated
Oth: Low power
Reference: Stroup et al. (1986)
Exposure assessment: Presumed exposure
to formaldehyde tissue fixative.
External comparisons:
SMR = 0(0-2.0)
[0]
This document is a draft for review purposes only and does not constitute Agency policy.
1-509 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Results: effect estimate (95% CI)
[# of cases]
Population: 2,239 white male members of
the American Association of Anatomists
from 1888 through 1969 who died during
1925-1979. Death certificates obtained
for 91% with 9% lost to follow-up.
Outcome definition: Hodgkin disease (ICD-
8: 201) listed as cause of death on death
certificates.
Design: Cohort mortality study with
external comparison group.
Analysis: SMRs calculated using sex, race,
age, and calendar-year-expected number
of deaths from the U.S. population.
Confidence in effect estimates:3
Duration and timing: Occupational
exposure preceding death during 1925-
1979. Median birth year was 1912. By
1979, 33% of anatomists had died.
Duration and timing since first exposure
were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as potential
confounders.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc, and
ionizing radiation.
SB IE Cf Oth
Overall
Confidence
Low
LOW 4,
(Potential bias toward the nullvU)
SB: Health worker effect
IB: Exposure Group A; latency not
evaluated
Cf: Potential confounding
Oth: Low power
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.
Anatomists may also be coexposed to
stains, benzene, toluene xylene, chlorinated
hydrocarbons, dioxane, and osmium
tetroxide.
Benzene was not evaluated as a potential
confounder and may be positively
correlated with formaldehyde exposure.
Potential for confounding is unknown but
could have inflated the observed effect.]
Reference: Walrath and Fraumeni
(1984)
Population: 1,007 deceased white male
embalmers from the California Bureau of
Funeral Directing and Embalming who died
during 1925-1980. Death certificates
obtained for all.
Outcome definition: Hodgkin disease (ICD-
8: 201) listed as cause of death on death
certificates.
Design: Proportionate mortality cohort
study with external comparison group.
Analysis: PMRs calculated using sex, race,
age, and calendar-year-expected number
of deaths from the U.S. population.
Confidence in effect estimates:3
Exposure assessment: Presumed exposure
to formaldehyde tissue fixative.
Duration and timing: Occupational
exposure preceding death during 1916-
1978. Birth year ranged from 1847 through
1959. Median age of death was 62 years.
Most deaths were among embalmers with
active licenses. Duration and timing since
first exposure were not evaluated.
Variation in exposure: Not evaluated.
Coexposures: None evaluated as potential
confounders.
[As noted in Appendix A.5.9: Coexposures
may have included: phenol, methyl alcohol,
glutaraldehyde, mercury, arsenic, zinc, and
ionizing radiation.
External comparisons:
Observed: 0 Hodgkin disease deaths
Expected: 2.5 Hodgkin disease deaths
PMR= 0(0-1.20)+
[0]
+Note: EPA derived CIs using the Mid-P
Method (See Rothman and Boice,
1979)
This document is a draft for review purposes only and does not constitute Agency policy.
1-510 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
Study
Exposures
Radiation exposure likely to be poorly
correlated with formaldehyde so
confounding is unlikely.]
Results: effect estimate (95% CI)
[# of cases]
SB IB Cf Oth
Overall
Confidence
Low
LOW
(Potential bias toward the nullvU)
IB: Exposure Group A; latency not
evaluated
Oth: Low power
Evaluation of sources of bias or study limitations (see details in Appendix A.5.9). SB = selection bias; IB = information bias;
Cf = confounding; Oth = other feature of design or analysis. Extent of column shading reflects degree of limitation. Direction
of anticipated bias indicated by arrows: "\|/' for overall confidence indicates anticipated impact would be likely to be toward
the null (i.e., attenuated effect estimate); "/|v' for overall confidence indicates anticipated impact would be likely to be away
from the null (i.e., spurious or inflated effect estimate).
Abbreviations: SB = selection bias; IB = information bias; Cf = confounding; Oth = other feature of design or analysis;
UCOD = underlying cause of death; GSD = geometric standard deviation; SMR = standardized mortality ratio; RR = relative risk;
TWA8 = 8-hour time-weighted average; URT = upper respiratory tract; LHP = lymphohematopoietic; PMR = proportionate
mortality ratio; OR = odds ratio; SPIR = standardized proportional incidence ratio.
Lymphohematopoietic cancers in animals
Few animal bioassays have adequately evaluated the carcinogenic potential of inhaled
formaldehyde with respect to LHP malignancies. The majority of formaldehyde exposure studies in
animals focused primarily on the respiratory tract and did not provide routine examination of other
tissues, limiting the detection of leukemia and lymphoma. The study conducted by Battelle-
Columbus Laboratories for CUT (Battelle. 19821 is currently the only chronic duration inhalation
study to report detailed information on formaldehyde-induced leukemia or lymphoma in rodents
(results not published). Given the paucity of available information and difficulties interpreting the
Battelle fBattelle. 19821 results, the evidence available from animal studies is considered
indeterminate for drawing conclusions as to whether or not formaldehyde exposure might cause
leukemia or lymphoma.
Methodological issues considered in evaluation of studies
Given the assumed differential distribution of inhaled formaldehyde as compared to
exposure by other routes, only inhalation studies were considered relevant to discussions of LHP
cancers in animals. Detailed study evaluation tables of the four relevant inhalation studies are
available in Appendix A.5.9. This section considers incidence data for histopathological lesions
associated with leukemia or lymphoma; other evidence supportive of the development of these
cancers (e.g., hematological changes) is discussed in the Evidence on Mode of Action for
Lymphohematopoietic Cancers Section.
This document is a draft for review purposes only and does not constitute Agency policy.
1-511 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
Lymphohematopoietic Cancers in Animal Studies
This discussion focuses on the few available studies evaluating tumors of the lympho-
hematopoietic system (leukemia and lymphomas), with the evidence organized by species and
study confidence (see Table 1-65). The largest and most comprehensive cancer bioassay evaluating
formaldehyde inhalation exposure in animals is the high confidence chronic study (Battelle. 1982)
conducted at the Battelle Columbus Laboratory in B6C3F1 mice and F344 rats. This was also the
only study to evaluate the majority of tissues relevant to LHP cancers (e.g., no other study reported
histopathological evaluation of the spleen or thymus). The summary reports of these experiments
in the published literature do not discuss leukemia or lymphoma rates fKerns etal.. 1983:
Swenberg etal.. 1980bl. However, tissue slides were examined histopathologically in all animals
from the control and 17.6 mg/m3 dose groups at each interim and terminal necropsy; the lesions
examined included lymphoma and leukemia (note: increased bone marrow hyperplasia, a
nonmalignant lesion that was significantly increased in exposed rats, is also included in Table 1-65
and further discussed in the Evidence on Mode of Action for Lymphohematopoietic Cancers Section).
At the intermediate dose groups of 2.5 mg/m3 and 6.9 mg/m3 exposure concentrations, only the
target (i.e., the nasal passages) tissues were examined unless unusual tissue masses or gross lesions
were noted, or if the animals died spontaneously, and the study report does not provide incidence
at these doses in their summary findings fBattelle. 19821. As stated in the report, survival rates for
rats were decreased by formaldehyde exposure at the 17.6 mg/m3 exposure for males and females.
For the mice, there was no differential mortality across exposure groups; however, there appeared
to be decreased survival in all exposure groups after 6 months. The cumulative incidences of
lymphoma (in B6C3F1 mice) and leukemia (in F344 rats) as reported by Battelle (see Tables 7-10
in fBattelle. 198211 are shown in Table 1-64. The p-values reported by the authors were based on a
Cox-Tarone test for the comparison that adjusts for reduced survival fBattelle. 19821. There was a
suggestion of a possible increased incidence in lymphoma (p-value, 0.06) in female mice, and a
decreased incidence in leukemia in female rats (p-value, 0.006) at the high dose. The possible
increase in lymphoma incidence in mice is of interest for future study, as low incidences of
lymphomas were also observed in two strains of p53 deficient mice after formaldehyde exposure,
whereas no lymphomas were observed in control groups [(Morgan etal.. 20171: see additional
discussion in the Evidence on Mode of Action for Lymphohematopoietic Cancers Section], It is
problematic to infer from these results because of the lack of information at the intermediate dose
groups and the adverse effect on survival rates. It is also difficult to interpret the apparent slight
increase in lymphoma in mice alongside the slight but statistically significant decrease in leukemia
in female rats. Taken together with the exposure-induced increases in bone marrow hyperplasia in
rats, this represents an area of uncertainty warranting additional study.
This document is a draft for review purposes only and does not constitute Agency policy.
1-512 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
Table 1-64. Cumulative incidence of hematopoietic cancers in B6C3F1 mice
and F344 rats
Endpoint, species
Sex
Incidence or percentage incidence
p-Values
0 ppm
17.6 mg/m3
Lymphoma, mice
Male
0/119 (0%)
0/115 (0%)
Female
19/121 (16%)
27/121 (22%)
0.062
Leukemia, rats
Male
11/120 (9%)
5/120 (4%)
0.690
Female
11/120 (9%)
7/120 (6%)
0.006
A separate, medium confidence study in rats did not report any significant differences in
histopathological evaluations of tissues relevant to leukemia or lymphoma (Kamata etal.. 19971.
although specific incidence data for non-nasal lesions were not provided. Although the two other
available studies also failed to observe statistically significant, treatment-related increases in LHP
cancers in potentially sensitive mice fMorgan et al.. 20171 or rats fSellakumar etal.. 19851. these
results were interpreted with low confidence due primarily to concerns regarding insensitivity due
to a very short exposure duration (8 weeks; (Morgan etal.. 2017)). or histopathological evaluations
of LHP tissues only when gross lesions were noted (Sellakumar etal.. 1985).
Overall, the available data are indeterminate for drawing conclusions regarding the
potential for formaldehyde exposure to induce LHP cancers in rodent bioassays. It should be
emphasized that the detection of leukemia/lymphoma in the available animal studies (i.e., other
than the 0 versus 17.6 mg/m3 group comparisons in the Battelle study) may be limited by study
design due to limited statistical power, a lack of routine evaluation of tissues potentially related to
LHP cancers (studies focused on histopathological evaluation of nasal tissue), or early mortality
from toxicities other than LHP cancer, particularly given the few suggestive changes that were
reported (i.e., bone marrow hyperplasia in rats and slight but uncertain increases in lymphomas in
mice). To make definitive conclusions regarding the development of LHP cancers in formaldehyde-
exposed animals, there is a need for studies specifically designed to target these cancers as the main
endpoint.
This document is a draft for review purposes only and does not constitute Agency policy.
1-513 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 1-65. Summary of animal evidence of lymphohematopoietic cancers and
bone marrow histopathology following inhalation exposure to formaldehyde
Reference and study design
Results
Rats
High confidence
Kerns et al. (1983), Battelle (1982)
Rats: Fischer 344; males and females; 119 to 121/sex/group
Exposure: whole-body 6 hr/d, 5 d/wk for up to 24 months;
recovery examined at 27 and 30 months
Test article: Paraformaldehyde analytic concentrations: 0, 2.5
(± 0.01), 6.9 (± 0.02), and 17.6 (± 0.05) mg/m3
Histopathology: Relevant tissues included femur, mandibular
and mesenteric lymph nodes, spleen, and thymus. Note:
Histopathological examination was carried out only for
unusual tissue masses for 2.5 and 6.9 mg/m3 groups (see
text).
Rats, leukemia (all)
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Female
11/109 (9%)
NA
NA
7/113 (6%)
Male
11/109 (9%)
NA
NA
5/115 (4%)
Rats, bone marrow hyperplasia
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Female
7/106 (6%)
NA
NA
28/87* (24%)
Male
NA = Onl
*p = o.oc
6/108 (5%)
y nasal tissue
01 (see Table
NA
was systemj
1-74 for leu
NA
tically analy
cemia p-valu
26/85* (23%)
zed
es)
Medium confidence
Kamata et al. (1997)
Rats: Fischer 344; male; 32/group
Exposure: nose-only 6 hours/day, 5 days/week for
28 months; interim sacrifices at months 12,18, and 24
Test article: Formalin (and methanol control)
Analytic concentrations: 0, 0.40 (± 0.09), 2.67 (± 0.40), or
18.27 (± 2.73) mg/m3. Methanol in the 0 and 18.27 groups
was estimated at 5.2 mg/m3. A room control served as a no
exposure group.
Histopathology. Relevant tissues included mesenteric lymph
nodes and femur; and other tissues with noted gross lesions.
Main limitations: Formalin (gaseous methanol levels were
not analytically measured in the control and exposed groups,
even though a methanol control was included); limited
histopathological examinations of non-nasal tissues.
No lesions attributable to formaldehyde exposure were
detected in organs other than the nasal cavity.
Low confidence
Sellakumar et al. (1985)
Rats: SD; male; 99-100/group
Test article: Paraformaldehyde (slurry in paraffin oil)
Exposure: 6 hr/d, 5 d/wk for lifetime at 0 or 18.2 mg/m3 (note:
prior reporting of levels during first 588 days at 17.5 mg/m3
(Albert et al., 1982)
Histopathology: Histopathology conducted for LHP-relevant
tissues (not specified) only when gross lesions were noted
Related study: Albert et al. (1982)
Main limitations: LHP tissues were only evaluated if gross
lesions were noted.
No differences in tumors outside of the respiratory tract were
noted between treated and control groups.
This document is a draft for review purposes only and does not constitute Agency policy.
1-514 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference and study design
Results
Mice
High confidence
Battelle (1982)
Mice: B6C3F1 mice; males and females; 119 to
121/sex/group
Exposure: whole-body 6 hr/d, 5 d/wk for up to 24 months;
recovery examined at 27 and 30 months
Test article: Paraformaldehyde
Analytic concentrations: 0, 2.5 (± 0.01), 6.9 (± 0.02), and 17.6
(± 0.05) mg/m3
Histopathology. Relevant tissues included femur, mandibular
and mesenteric lymph nodes, spleen, and thymus. Note:
Histopathological examination was carried out only for
unusual tissue masses at 2.5 and 6.9 mg/m3 (see text).
Note: Somewhat limited sampling for potential LHP cancers
and high mortality.
Mice, lymphoma (all)
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Female
19/102
(16%)
NA*
NA
27/121
(22%)
Male
0/119 (0%)
NA
NA
0/115 (0%)
Mice, lymphoid hyperplasia (mandibular lymph node)
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Female
19/59 (24%)
NA
NA
24/63 (28%)
Male
7/58 (11%)
NA
NA
14/49 (22%)
Mice, lymphoid hyperplasia (spleen)
0 mg/m3
2.5 mg/m3
6.9 mg/m3
17.6 mg/m3
Female
*NA = Or
25/90 (22%)
lly nasal tissue
NA
was systema
NA
tically analyz
22/97 (18%)
ed.
Low confidence
Morgan et al. (2017)
Mice: C3B6.129Fl-Trp53tmlBrd (C3B6 TP53±) and B6.129-
Trp53tmlBrd (B6 TP53±); males; 24-35/group
Exposure: Mice were exposed to FA in dynamic whole-body
chambers 6 hours/day, 5 day/week for 8 weeks.
Test article: Paraformaldehyde
Nominal concentrations were 0, 9.23, or 18.45 mg/m3.a
Histopathology: Routine evaluations of relevant tissues
included frontal plane sections of the femur (including bone
marrow), and mesenteric, mandibular, mediastinal, and
bronchial lymph nodes. Tissues with gross lesions were also
evaluated.
Main limitations: Short duration and short follow-up period
to allow for cancer development (note: authors based
exposure duration, in part, on HSPC doubling).
The incidences of leukemia or lymphohematopoietic
neoplasms were not statistically significantly increased by
formaldehyde exposure in either strain.
Lymphomas were observed in several mice exposed to
formaldehyde in both strains (i.e., in "B6" mice: 1/31 at
9.23 mg/m3 and 1/35 at 18.45 mg/m3; in "C3B6" mice: 1/24 at
9.23 mg/m3 and 2/25 at 18.45 mg/m3), while lymphomas were
absent from control groups in both strains (the study authors
determined these lesions were unrelated to treatment).
Abbreviations: LHP = lymphohematopoietic; FA = formaldehyde-specific antibody; HSPC = hematopoietic stem and
progenitor cell.
1 Evidence on Mode of Action for Lymphohematopoietic Cancers
2 Introduction
3 This section evaluates evidence supporting plausible mechanisms of LHP carcinogenesis
4 following inhalation exposure to formaldehyde. As previously discussed, the strength of the
5 evidence in humans was determined to be robust for myeloid leukemia and slight for multiple
6 myeloma, although evidence in experimental animals is considered indeterminate. As a mode(s)-of-
7 action has not been established for how formaldehyde inhalation may result in LHP cancers, the
8 available evidence relevant to interpreting the biological plausibility of the observed associations in
9 humans is presented in this section. This discussion includes consideration of how ge no toxicity
This document is a draft for review purposes only and does not constitute Agency policy.
1-515 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
and other potential molecular and cellular events resulting from formaldehyde interactions in
upper respiratory tract (URT) tissues might result in LHP cancers. Genotoxicity of formaldehyde in
different experimental systems and in human populations is evaluated and described in detail in
Appendix A.4; in this section, conclusions from these data are interpreted specifically as pertaining
to LHP carcinogenesis. Additional evidence relevant to interpreting the biological plausibility of
formaldehyde exposure-induced LHP carcinogenesis has been previously discussed, including DNA
damage in peripheral blood cells, impacts on immune cell populations and inflammation in
peripheral blood in human populations, systemic oxidative stress, and other health effects outside
of the respiratory system, including developmental and reproductive toxicity, hazards for which the
evidence indicates that effects in humans are likely. These data are discussed in Sections 1.2.3,
1.2.5, and 1.3.2.
Approach: consideration of mechanistic events plausibly relevant to LHP cancer induction following
inhaled formaldehyde exposure
This section considers conclusions derived from the analyses of pertinent types of evidence
as they relate to LHP cancer (discussed in detail elsewhere in this Toxicological Review), and
further examines facets of the genotoxicity database and other mechanistic events specifically
relevant to the potential cellular origins of LHP cancer. Rather than a single, linear MOA hypothesis
to which formaldehyde-specific data can be applied and evaluated, a network of mechanistic events
or pathways may be a more appropriate conceptual framework within which to consider the
biological plausibility for many cancers, including LHP carcinogenesis potentially caused by
formaldehyde inhalation. These plausible mechanistic events involve specific aspects of
genotoxicity and mutagenicity, hematologic effects, and changes in gene expression or regulation,
consistent with previous analytical frameworks employed in the evaluation of LHP carcinogenesis
fNRC. 2014bl Additionally, this discussion includes consideration of mechanistic effects which
have been previously described as hallmarks or enabling characteristics of cancer, as well as key
characteristics of carcinogens [e.g., genomic instability and mutation, oxidative stress,
inflammation, and avoidance of immune destruction; (Smith etal.. 2016: Hanahan and Weinberg.
201111.
Although there is evidence that exposure to formaldehyde is associated with changes in cell
populations that are relevant to LHP cancer mechanisms, a number of studies have demonstrated
that direct interactions of formaldehyde with cells in the bone marrow are not likely
(see Appendix A.2). In the bone marrow of monkeys (Moeller etal.. 2011). and in the bone marrow,
liver, lung, spleen, thymus, and blood of rats (Lu etal.. 2010a). DNA monoadducts were formed by
interactions with endogenous formaldehyde, but adducts formed from exogenous formaldehyde
were not found using highly sensitive detection methodology. Recently Lai et al. (2016) described
an ultrasensitive mass spectrometry method, which distinguishes unlabeled DPX from 13CD2-
labeled DPXs induced from endogenous and exogenous formaldehyde, respectively. The authors
demonstrated that inhalation exposure of stable isotope labeled (13CD2) formaldehyde to rats
This document is a draft for review purposes only and does not constitute Agency policy.
1-516 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
(18.45 mg/m3; 6 hours/day; 1, 2, or 4 days) and monkeys (7.38 mg/m3; 6 hours/day; 2 days)
induced DPXs linked to exogenous formaldehyde in nasal passages in both species, but not in distal
tissues, such as bone marrow and peripheral blood monocytes (rats and monkeys) and liver
(monkeys), although DPXs linked to endogenous formaldehyde were detectable in all tissues. In
light of this evidence, in vitro studies of direct administration of formaldehyde to cells from distal
tissues, such as bone marrow and blood, were considered less relevant to the evaluation of
hazard.).
The approach taken in this section was to identify mechanistic events possibly linking
inhaled formaldehyde-induced effects to LHP cancer risk in humans, and then to evaluate the
supporting evidence for these events and relationships. The primary focus was on evidence from
mechanistic studies of exposed humans where available, incorporating results from in vivo animal
studies and in vitro experiments when such information was particularly instructive. The studies
most informative to LHP mechanisms were those that examined changes in leukocyte populations
or function along with genotoxicity in potential target cells (e.g., hematopoietic stem and progenitor
cells [HSPCs], discussed below) or surrogate cell populations (e.g., peripheral blood lymphocytes
[PBLs]) from the same human cohorts. Measuring genotoxicity in mature PBLs as surrogates for
target cells of concern for LHP carcinogenesis (i.e., HSPCs) is a commonly adopted and reasonable
experimental approach fKirsch-Volders etal.. 20141 because PBLs are much more abundant than
HSPCs, which constitute only a fraction of a percentage of circulating leukocytes (de Kruij fetal..
2014: Massberg et al.. 2007). Other studies selectively reporting hematotoxicity, altered immune
function, or genotoxicity in circulating WBCs from formaldehyde-exposed humans or animals also
provided useful information.
The mechanistic events specifically evaluated include:
1) Evidence of formaldehyde-induced DNA damage to peripheral blood leukocytes
a. Genotoxicity in circulating myeloid progenitor cells (possible cancer target population)
b. Genotoxicity in circulating lymphocytes (surrogate population)
2) Evidence of formaldehyde-induced impacts other than genotoxicity on circulating blood cell
populations, including inflammatory changes or immune system dysfunction
3) Evidence of formaldehyde-induced systemic oxidative stress
4) Evidence of formaldehyde-induced changes in the bone marrow niche
5) Evidence of formaldehyde-induced changes in gene expression or posttranscriptional
regulation in peripheral blood leukocytes or bone marrow
In each of the following sections, the formaldehyde-specific mechanistic evidence is briefly
reviewed, then the relevance to LHP carcinogenesis is described alongside a discussion of the
This document is a draft for review purposes only and does not constitute Agency policy.
1-517 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Toxicological Review of Formaldehyde—Inhalation
evidence (or lack thereof) addressing how formaldehyde exposure might cause the observed
effects.
To frame the discussion of the plausible mechanistic events related to LHP carcinogenesis,
relevant elements of HSPC physiology are briefly reviewed. Hematopoietic stem cells (HSCs) are
cells residing in the blood or bone marrow that are functionally defined by their ability to replenish
their own numbers as well as divide asymmetrically into less plastic progenitor cells. The HSCs
reside in localized microenvironments within the bone marrow called "niches," which control their
survival, mobilization, proliferation, self-renewal, and differentiation (Wilson et al.. 20091. For
example, a single HSC can give rise to common myeloid or lymphoid progenitor cells, which can in
turn yield blast cells with dedicated differentiation into specific cell lineages, with a fraction
becoming myeloblasts and lymphoblasts, respectively (see Figure 1-43). HSCs and progenitor cells
(e.g., myeloblasts, common myeloid or lymphoid progenitors, etc.) are described together as HSPCs
(Granick et al.. 2012: Massberg et al.. 2007) (see Figure 1-43). As previously described (see
Section 1.3.3, Overview of Lymphohematopoietic Cancer Biology), LHP cancers are a heterogeneous
group. Most LHP cancers, including acute and chronic myeloid leukemias as well as multiple
myeloma (i.e., LHP cancers best associated with formaldehyde exposure in epidemiology studies)
are thought to arise from damage to HSPCs during hematopoietic and lymphopoietic development,
or as a result of environmental exposure, often in a specific HSPC-type and lifestage-dependent
manner (Greaves. 2004). However, some LHP cancer subtypes, including CLL and some lymphomas,
may arise from mature leukocytes (Eastmond et al.. 2014). Thus, this section discusses HSPCs as
the most likely proximal target for LHP cancers (i.e., those of primary interest in the context of
formaldehyde exposure), while mature leukocytes are discussed as surrogate populations for
cancer target cells.
This document is a draft for review purposes only and does not constitute Agency policy.
1-518 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Hematopoietic stem cell
(HSC; Hemocytoblast)
Hematopoietic
stem and
progenitor cells
(HSPCs)
Differentiated
leukocytes
Figure 1-43. Simplified hematopoiesis.
Hematopoietic stem cells (HSC) are capable of self-renewal, and can asymmetrically divide to create
progenitors committed to either myeloid or lymphoid lineages; together, the HSCs and more committed
progenitors comprise hematopoietic stem and progenitor cells (HSPCs; (Granick et al., 2012; Massberg et
al., 2007)). The progenitors then supply the precursor cells responsible for maintaining the population of
more differentiated cell types within the committed lineage, as depicted. The likely candidate cellular
targets for lymphohematopoietic (LHP) cancers are the varied progenitors associated with the monocyte
and lymphocyte lineages (a few examples illustrated), as well as HSCs themselves.
Evidence of formaldehyde-induced DNA damage to peripheral blood leukocytes
The most pertinent and direct available evidence of formaldehyde-induced effects on target
cells relevant to LHP carcinogenesis (i.e., those that may ultimately become neoplastic) is from two
studies of the same cohort reporting genotoxicity in myeloid progenitor cells in the peripheral
blood of exposed human workers (Appendix A.4). In addition, several studies have been conducted
documenting several measures of DNA and chromosomal damage and instability in PBLs of
workers exposed to formaldehyde. As these exposures occurred in vivo and the effects are not
formaldehyde-specific, no assumptions can be made regarding whether or not formaldehyde must
directly interact with the HSPCs or PBLs (e.g., potentially while migrating through URT tissues) to
induce the observed changes, or, alternatively, if these represent indirect effects. In vitro
formaldehyde exposure of isolated PBLs may also provide some minimal supportive information,
although substantially lower confidence exists regarding the relevance of these data, given the
limited distribution of inhaled formaldehyde beyond the URT and the assumption that the inhaled
formaldehyde concentrations these cells might encounter in URT tissues, if any, would be much
lower than the in vitro levels applied. Notably, human PBLs may be less sensitive to potential in
This document is a draft for review purposes only and does not constitute Agency policy.
1-519 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
vivo genotoxicity compared with HSPCs, as murine HSPCs are more susceptible to aldehyde-
induced DNA damage than mature, differentiated leukocytes fOberbeck et al.. 2014: Garavcoechea
et al.. 20121.
Genotoxic effects on circulating myeloid progenitor cells
Among the human occupational studies with formaldehyde exposure, two studies of the
same cohort reported effects on myeloid progenitor cells cultured from peripheral blood of exposed
workers (Lan etal.. 2015: Zhang etal.. 20101: (see Appendix A.4) compared to cells cultured from
controls without occupational formaldehyde exposure. The specific hematopoietic progenitor cells
assessed were identified as CFU-GMs, but not lymphocytes (i.e., myeloblasts in Figure 1-43). CFUs
of less committed HSPC colonies (e.g., CFU-GEMMS which can give rise to granulocytes,
erythrocytes, macrophages, and megakaryocytes) could not be directly assessed for technical
reasons (Lan etal.. 2015: Zhang etal.. 2010). No information is available to determine if either
progenitor cell type would be more or less susceptible to formaldehyde-induced genotoxicity.
In an initial pilot study, increased monosomy of chromosome 7 and trisomy of chromosome
8 was reported in CFU-GMs cultured from a group of 10 highly exposed subjects and 12 controls (8
hr TWA 2.6 versus 0.032 mg/m3, respectively) evaluated only for aneuploidy in chromosomes 7
and 8. Decreased WBC counts and a 20% decrease in CFU-GM colony formation was also noted,
suggesting hematotoxicity fZhang et al.. 20101. The initial finding of chromosome 7 monosomy was
confirmed in a larger, more comprehensive analysis of the same cohort with 29 occupationally
exposed subjects and 23 referents (1.7 versus 0.032 mg/m3) wherein chromosome-wide
aneuploidy and structural aberrations of all 24 chromosomes were examined (Lan etal.. 20151.
This follow-up study also reported significantly: (a) increased frequencies of monosomy in
numerous chromosomes, with the greatest response for chromosomes 1, 5, and 7; (b) increased
polysomy in several chromosomes including 1 and 5; and (c) increased tetrasomy in various other
chromosomes. In addition to aneuploidy, increased breaks, deletions, and translocations of
chromosome 5 were also reported, while trisomy of chromosome 8 was not significantly elevated
(Lan etal.. 20151. Although the pilot study methods were criticized for not adhering to the assay
protocol (Gentry etal.. 20131. a clarification of the assay protocol was provided by the investigators
with a description of how the study adhered to it fRothman etal.. 20171. Additional findings of
monosomy, trisomy, tetrasomy, and structural aberrations of multiple chromosomes that were
increased in formaldehyde-exposed workers in comparison to the unexposed referent group
indicate that formaldehyde exposure is associated with a potential tendency toward cytotoxicity in
CFU-GM cells that may arise either in vivo or during the in vitro cell culture period.
A more recent study in mice from the same researchers similarly suggests that in vivo
formaldehyde exposure (3 mg/m3 for 2 weeks) might affect the viability of progenitor cells of the
granulocyte/monocyte (CFU-GM) or erythroid (BFU-E) lineage based on the ability to generate
colonies of these cells in culture fZhao etal.. 2020al. Although they did not specifically examine
changes in the blood, the authors reported consistent decrements (across two independent
This document is a draft for review purposes only and does not constitute Agency policy.
1-520 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
experiments) in BFU-E from the nose; BFU-E and CFU-GM from the bone marrow; and CFU-GM
from the spleen. The authors also reported mixed evidence of decrements (across experiments) for
CFU-GM from the nose; BFU-E and CFU-GM from the lung; and BFU-E from the spleen. However,
the study results cannot be reliably interpreted as clear evidence of formaldehyde-induced effects
due to use of formalin as the test article and small sample sizes.
In vitro formaldehyde exposure of cells isolated from healthy, unexposed humans provided
mixed results. Formaldehyde exposure-induced aneuploidy in cultured human erythroid
progenitor cells fli etal.. 2014). but not in cultured myeloid progenitor cells (Kuehner etal.. 2012).
These results suggest either a more complex biological basis for susceptibility to chromosomal
damage, or an inability of in vitro test conditions to detect or replicate formaldehyde-associated
effects observed in the in vivo studies.
Of interest in the context of susceptibility, in mice, knockout of the genes encoding enzymes
responsible for removal of endogenous formaldehyde, namely Aldh2 and AIdh5, results in a
phenotype of severely disrupted hematopoiesis and leukemia, including mutated and abnormal
HSPCs, which is presumably linked to elevated formaldehyde levels (Dingier etal.. 2020: Burgos-
Barragan etal.. 2017b: Pontel etal.. 20151. Likewise, direct treatment of AIdh5-/- bone marrow
cells with formaldehdye causes genotoxic effects and reduces HSPC formation, effects which are
further exacerbated by loss of Fancd2 (this latter deficiency is associated with increased sensitivity
to DNA damage) (Garcia-Calderon etal.. 2018: Burgos-Barragan etal.. 2017b). As reviewed and
tested by Dingier et al. (2020). genetic deficiencies in these Aldh family genes has been linked to
bone marrow failure and related diseases in humans, including specifically in children. Other
changes in these mouse models and humans with reduced ALDH2 or ALDH5 activity that may be
caused, at least in part, by uncontrolled endogenous formaldehyde include postnatal lethality,
stunted growth, cognitive effects (see Section 1.3.1) and various cancers arising from DNA damage
or deficient repair fDingler etal.. 2020: Nakamura etal.. 20201. While formaldehyde inhalation
does not seem to cause appreciable changes in formaldehyde levels in nonrespiratory regions (see
Appendix A.2), HSPCs expressing these enzymes are known to exist in many tissues. However, no
studies in any species have specifically examined these possible linkages in relation to inhaled
formaldehyde, limiting the use of the currently available studies in hazard identification to the
identification of factors of interest to future studies on susceptibility.
Relevance to LHP carcinogenesis and mode of action interpretation
As described above, the cells used in these experiments represent a potential primary target
for LHP carcinogenesis. The aneuploidy observed in chromosomes 5 and 7 is of particular
relevance for chemically induced LHP carcinogenesis because the loss of whole or part of
chromosomes 5 or 7 are common aberrations in therapy-related myelodysplastic syndrome (MDS)
and acute myelogenous leukemia fLessard etal.l. particularly those resulting from alkylation drug
therapy fLan etal.. 2015: Pedersen-Biergaard etal.. 2006: Smith etal.. 20031. Therefore, the
observations of similar cytogenetic effects in asymptomatic formaldehyde-exposed workers
This document is a draft for review purposes only and does not constitute Agency policy.
1-521 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
supports the biological plausibility of the association between chronic formaldehyde exposure and
elevated incidence of LHP cancers in other human cohorts (see Section 1.2.5, Evidence on Mode of
Action for URT Cancers). Although exogenous formaldehyde may not be transported to or
specifically affect the bone marrow in a fashion akin to other well-studied human leukemogens
fe.g.. benzene, chemotherapeutics. ionizing radiationEastmond etal.. 20141. and may therefore not
act via a similar MOA, similar aneuploidies in CFU-GMs from formaldehyde-exposed and benzene-
exposed workers have been observed (i.e., monosomy and trisomy in chromosomes 5 and 7; (Zhang
etal.. 20111. Thus, the presence and type of aneuploidies observed in circulating myeloid
progenitor cells from formaldehyde-exposed asymptomatic human workers are consistent with
those reported in patients with leukemia, specifically MDS and AML, as well as those effects
reported in other worker cohorts at increased risk of developing leukemias, providing further
support for the plausibility of an association between chronic formaldehyde exposure and
leukemogenesis.
While this evidence links formaldehyde exposure to chromosomal toxicity relevant to
leukemogenesis, mechanistic evidence is lacking for how these events may occur. Although no
evidence exists to evaluate the following potential scenarios, there are at least three ways in which
formaldehyde exposure (with distribution limited to the URT) might cause these genotoxic effects:
(1) direct interaction of formaldehyde with HSPCs in the URT; (2) indirect effects on circulating or
bone marrow HSPCs due to secondary, systemic effects following formaldehyde-induced changes in
the URT; and (3) modification and mobilization of precursor-type cells residing in the URT.
As part of their physiological function, HSPCs migrate via the vasculature to extramedullary
tissues (outside medullary bone) such as the liver, lung, small intestine, skin, and kidneys, and
return via lymphatics to the bone marrow by a process termed "homing," which is mediated by
cytokines, growth factors, and hormones f Granick et al.. 2 012: Schulz etal.. 2009: Massberg etal..
20071. Although their numbers in the peripheral blood at any one time constitute a small fraction
of the total circulating leukocyte population in both mice (Massberg et al.. 2007) and humans (de
Kruijfetal.. 2014: Zhang etal.. 2010). these cells can completely replenish bone marrow stem cell
populations (Massberg et al.. 2007). Unlike mature lymphocytes, HSPCs do not necessarily
accumulate in lymphatic tissues (e.g., nasopharynx-associated lymphoid tissue or NALT), but travel
primarily through the lymphatic vasculature fMassberg et al.. 20071. HSPCs accumulate to some
extent in peripheral nonlymphoid tissues and are replenished every few days; alternatively, HSPCs
can divide locally and replenish populations of long-lived resident myeloid cells (e.g., macrophages,
dendritic cells). In addition to triggering local differentiation, inflammatory stimuli can induce
HSPC mobilization from the bone marrow (Wilson et al.. 2009). and may increase recruitment of
mobilized HSPCs to nonlymphoid epithelial tissues (Massberg et al.. 2007). Such inducible
migration to and from sites of inflammation (e.g., formaldehyde-induced URT inflammation, see
Section 1.2.3) could be a mechanism by which HSPCs become more frequent targets of
formaldehyde-induced toxicity. The available data suggest that very little, if any, inhaled
This document is a draft for review purposes only and does not constitute Agency policy.
1-522 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
formaldehyde penetrates beyond the URT (the portal of entry; POE), although it is likely that small
amounts of formaldehyde are able to reach the superficial capillary layer of the URT in some
exposure contexts (see Appendix A.2). In addition, whereas formaldehyde appears to preferentially
target the respiratory and transitional epithelium of the nasal cavity, it is unclear which specific
URT compartments (e.g., respiratory, transitional, or olfactory epithelium; stromal tissue layers)
HSPCs may circulate through. Finally, although HSPCs may be more sensitive to genotoxic effects
than other cell types, even if inhaled formaldehyde did directly encounter HSPCs, no data exist to
draw inferences regarding theoretical concentrations of inhaled formaldehyde that might be
required for genotoxicity. Despite these important uncertainties, it is possible that formaldehyde
may be able to directly interact with potential target cell types present at the POE.
Alternatively, secondary effects resulting from toxicity, irritation, or other processes
disrupted in the affected URT might be capable of causing genotoxicity in HSPCs at sites distal to the
URT or in vascular regions proximal to the URT. Such secondary effects might include increased
production of mediators of inflammation and oxidative stress, which have been reported after
formaldehyde exposure in some studies (see Section 1.2.3), and which may result, indirectly, in
cytotoxicity, genotoxicity, or other perturbations at distal sites containing HSPCs, resulting in
genotoxicity in these cells. However, no data exist to evaluate this hypothesis, including the
potential secondary mediators or what levels of these mediators might be required at target sites.
Lastly, some URT (i.e., rat nasal olfactory epithelium) cells have been shown to be
"multipotent" in nature, in that they can repopulate rat hematopoietic tissues and differentiate into
various leukocyte lineages in irradiated hosts; although, these cells act more similar to neural stem
cells than to bone marrow stem cells (Murrell etal.. 2005). While it might be possible that
formaldehyde could interact with such a cell population, cause genotoxicity, and modify it in such a
way that it becomes more HSPC-like and migrates to the bone marrow, this theory is somewhat
implausible and without supportive evidence.
Overall, the evidence largely does not exist to determine whether any of the proposed
processes explain how formaldehyde exposure might cause genotoxicity in HSPCs.
Genotoxic effects on circulating lymphocytes
Consistent with formaldehyde-induced genotoxicity in circulating myeloid precursor cells,
formaldehyde exposure is associated with DNA and chromosomal damage in PBLs (see
Appendix A.4 for detailed discussions). The studies in which we had more confidence based on
evaluations of study methods reported consistent associations of formaldehyde exposure with DNA
strand breaks or alkali-labile sites visualized using the comet assay, CAs, MN formation, and sister
chromatid exchange (SCE). Formaldehyde was associated with a higher prevalence of chromosomal
aberrations among workers in pathology laboratories (Costa etal.. 2015: Musak etal.. 2013:
Santovito etal.. 2011: Takab etal.. 20101: these effects includedchromatid-type aberrations,
chromosome-type aberrations, chromosomal exchange, and premature centromere division. Costa
et al. (2015.) also reported an increase in aneuploidies and in the number of aberrant and
This document is a draft for review purposes only and does not constitute Agency policy.
1-523 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
multiaberrant cells. Micronuclei frequency in PBLs was higher in exposed compared to referent
workers by 40-50% with a concentration-related response beginning at concentrations of 0.1-
0.2 mg/m3 and above fCostaetal.. 2019: Wangetal.. 2019: liang etal.. 20101. Micronuclei
frequency (and centromeric micronuclei) increased with cumulative exposure fWang etal.. 2019:
Suruda etal.. 19931. A 1.5 to 3-fold difference in measures of DNA damage using the Comet assay
was observed comparing exposed workers to their referent groups at average concentrations as
low as 0.09 mg/m3 f Zendehdel et al.. 2 0171. 0.14 mg/m3 (liang etal.. 20101 or 0.04-0.11 mg/m3
(Peteffi etal.. 20151 and a clear concentration-related response was observed in plywood plant
workers fLin etal.. 2013: liang etal.. 20101. Costa et al. (2019) reported that the frequency of
micronuclei in PBL and EBC were correlated in their study population. In addition, increased DPXs
were observed in circulating WBCs from human workers exposed to formaldehyde concentrations
>0.5 mg/m3. In experimental animals, inhalation studies at relatively high formaldehyde
concentrations (i.e., 12.3 and 18.45 mg/m3) using paraformaldehyde as the test article have not
observed genotoxicity including DNA adducts, chromosome aberrations, or SCEs in PBLs of rats (Lu
etal.. 2010a: Kligerman et al.. 19841. Results of other studies using formalin as the inhalation
source were mixed fSpeitetal.. 2009: Im etal.. 20061. although these data are less reliable. While
evidence from in vitro formaldehyde exposures is likely of minimal value in relation to LHP
carcinogenesis, such evaluations also report increased mutations, DPX, and other DNA damage in
human PBLs, whole blood cells or cultured human lymphoblast cell lines (i.e., TK6 cells) (see
Appendix A.4).
Relevance to LHP carcinogenesis
Genotoxicity in PBLs may reflect formaldehyde-induced effects in HSPCs; because PBLs are
more amenable to experimentation, primarily because they are far more abundant, they can allow
for far more robust analyses (e.g., in terms of sample size), and possibly better detect changes.
Formaldehyde-induced chromosome damage may result from some combination of direct DNA
reactivity in the URT, including downstream sequelae, and numerous indirect mechanisms such as
deficiencies in DNA repair, chromosome segregation, DNA methylation and increased oxidative
stress (see Section 1.2.5 Evidence on MOA for URT Cancers; (Kirsch-Volders etal.. 20141. Similar to
the discussion of the HSPC-specific evidence, direct interactions of formaldehyde with DNA of
lymphocytes and less committed progenitor cells could occur in URT tissue regions, although this
has not been documented experimentally, or through indirect mechanisms occurring systemically
(e.g., as a result of increased oxidative stress). Evidence exists supporting both aneuploidy in PBLs
and clastogenicity in URT tissues; notably, the aneuploidy reported in PBLs is consistent with that
observed in DNA of CFU-GM cells studied by Zhang et al. (20101 and Lan et al. (20151. and observed
in relation to therapy-related MDS and AML as discussed above.
This document is a draft for review purposes only and does not constitute Agency policy.
1-524 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Evidence of formaldehyde-induced impacts other than genotoxicitv on circulating blood cell
populations, including inflammatory changes or immune system dysfunction
A number of studies indicate that formaldehyde exposure causes changes in hematopoietic
cell constituents in blood (see Section 1.2.3); however, an understanding of the observed pattern of
these changes in specific immune cell subtypes across studies, as well as how any of these changes
might be induced, remains incomplete. While there are inconsistencies in the database that
introduce uncertainty, the overall evidence indicates that it is probable that formaldehyde
inhalation causes blood cell changes including decreased total WBCs, CD8 + lymphocytes, and RBCs,
particularly at higher formaldehyde concentrations (e.g., >1 mg/m3; see Section 1.2.3). Relating to
formaldehyde-induced decreases in CD 8+ lymphocytes, one of the mouse studies discussed in
Section 1.2.3 (Ma etal.. 2020) provided evidence consistent with the possibility that formaldehyde
exposure inhibits commitment to the CD8 lineage at early stages of cell development. Perhaps most
relevant to LHP cancers, evidence of pancytopenia (i.e., decrease in RBCs, WBCs, and platelets in the
same exposed population) was reported in peripheral blood samples from formaldehyde-melamine
workers exposed to median formaldehyde concentrations of 1.6 mg/m3, along with a 20% decrease
in CFU-GM colony formation in vitro fZhang etal.. 20101. suggesting both a decrease in the
circulating numbers of mature RBCs and WBCs as well as possible decreases in the replicative
capacity of myeloblasts. This potential for formadehyde to selectively impact immature cells or
progenitors is consistent with observations in mice by Liu et al. (2017) and Zhao et al. (2020b).
although the use of formalin in these studies prevents reliable interpretation. Perhaps relatedly, a
decrease in HSPC colony formation was reported for various CFU populations, including both CFU-
GMs and CFU-GEMMs, cultured from human whole blood and exposed in vitro to 100-200 [J.M
formaldehyde fZhang etal.. 20101: however, these experiments carry the same uncertainties as
other in vitro assays (see above) including coexposure of the cells to methanol, which prevents
reliable interpretation of these findings. In addition, a study of two strains of p53 deficient mice
exposed to high levels of formaldehyde (>9 mg/m3) for 8 weeks (a duration selected based on the
HSPC pool turning over every 8 weeks) did not observe any significant increases in LHP cancers,
including leukemia fMorgan etal.. 20171. Although studies other than Zhang et al. (2010) do not
identify pancytopenia specifically, some report decreases in one or two of these cell types, but not
all three fZhang etal.. 2013b: Lvapina et al.. 2004: Kuo etal.. 19971. or in one or more of these cell
populations without examining all three (Ye etal.. 2005: Thrasher etal.. 1990): while other studies
reported no changes or significant increases for specific cell subsets (Avdin etal.. 2013: Costa et al..
2013: Erdei etal.. 2003). these latter studies tested formaldehyde concentrations of approximately
<0.36 mg/m3. Interestingly, some effects (e.g., changes in T cell populations) tended to increase at
lower formaldehyde concentrations (~ <0.5 mg/m3), while decreases were observed at higher
levels (~1 mg/m3). While the data suggest biologic complexity, pancytopenia such as that reported
by Zhang et al. (2010), is known to be associated with MDS and AML development fPaiva and
This document is a draft for review purposes only and does not constitute Agency policy.
1-525 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Calado. 20141 and may be one of the hematotoxic consequences of exposure to formaldehyde,
possibly only at concentrations >1 mg/m3.
In an effort to examine potential linkages between effects observed in AML patients and
those induced by formaldehyde, several studies have evaluated genotoxicity measures along with
immune system effects in the same cohort of occupationally exposed human workers. These
studies are considered highly informative to understanding the potential relationship between
formaldehyde exposure and systemic toxicity pertaining to LHP carcinogenesis. In several analyses
of the same occupationally exposed cohort in China with median exposures of 1.6 mg/m3
formaldehyde, lower total peripheral blood cell counts fHosgood etal.. 2013: Zhang etal.. 20101.
including CTL memory cells, and changes in cytokine levels fSeow etal.. 20151 were observed
concurrently with genotoxicity in myeloid precursor cells [fLan etal.. 20151 and discussed above].
Findings in this cohort were consistent with findings from Chinese workers and students evaluated
by another research group following short-term average formaldehyde exposures of approximately
0.51-0.99 mg/m3, which observed decreases in various T lymphocyte populations, including CTLs
(Ye etal.. 2005: Ying etal.. 19991. with a corresponding higher incidence of SCEs in worker
lymphocytes at approximately 0.99 mg/m3 fYe etal.. 20051. While CTLs were unchanged in several
other studies testing lower formaldehyde concentrations (0.2-0.8 mg/m3; flia etal.. 2014: Avdin et
al.. 2013: Costa etal.. 20131. one of these studies did report increased CD4 + T cells alongside
evidence of genotoxicity at 0.36 mg/m3 (Costa etal.. 2013). While CTLs were generally decreased
(increasing the ratio of CD4 + T cells to CTLs) in the blood of individuals exposed to formaldehyde
concentrations >0.5 mg/m3 (see Section 1.2.3), an understanding of how the observed cell number
changes might relate to genotoxicity remains unclear.
A reanalysis of data from Zhang et al. (2010) reaffirmed the lower levels of specific immune
cell populations, specifically WBCs, lymphocytes, RBCs and platelets in the exposed participants
with respect to the unexposed group fMundtetal.. 20171. However, when immune cell population
levels were compared within the exposed group using a cutpoint at the median of 1.6 mg/m3
(1.3 ppm), no difference was observed between the higher and lower exposed groups. Likewise, no
association with formaldehyde modeled as a continuous variable and cell population levels was
observed in regression analyses adjusted for sex and smoking. The 43 exposed participants were
highly exposed, ranging from a TWA8 of 0.5 to 3.3 mg/m3 (0.4 to 2.7 ppm) with one outlier at
6.9 mg/m3 (5.6 ppm). Fifty percent of the exposed group was exposed to aTWA8 from 1.1 to
2.5 mg/m3 (interquartile range). Therefore, the exposure levels in the study group did not include
the breadth of exposure levels needed at lower formaldehyde levels to evaluate an exposure-
response trend. The high formaldehyde exposure and the inadequate range of the concentrations
limited the power of the study to detect a trend with exposure level of the expected magnitude
based on those previously detected for benzene exposure (Rothman etal.. 20171.
Changes in serum NK cells and B cells were not entirely consistent across studies, although
the available data suggest that formaldehyde concentration may strongly influence the results,
This document is a draft for review purposes only and does not constitute Agency policy.
1-526 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
similar to findings for CTLs (see Section 1.2.3). For example, while NK cell numbers were
decreased at 0.36 and 1.6 mg/m3 fCostaetal.. 2013: Hosgood etal.. 20131 NK cells were actually
increased at 0.2 and 0.25 mg/m3 flia etal.. 2014: Avdin etal.. 20131 and unchanged at 0.8 mg/m3
flia etal.. 20141. Although changes in B cell counts were supported by moderate evidence across
several medium or high confidence studies conducted after several months of exposure, for
example at0.99 mg/m3 (Ye etal.. 2005) and 0.2-0.8 mg/m3 (Tia etal.. 2014). other medium or high
confidence studies testing formaldehyde exposures for several years, for example at 0.25 mg/m3
(Avdin etal.. 2013) and 1.6 mg/m3 (Hosgood etal.. 2013) did not report B cell changes, or reported
B cell decreases at lower formaldehyde levels (0.36-0.47 mg/m3) fCosta etal.. 2019: Costa etal..
20131. Looking across studies, the overall pattern of these responses across exposure levels and
exposure durations is difficult to interpret.
Although infrequently studied, some limited mechanistic information suggests the potential
for stimulation of the immune system at lower formaldehyde exposures, and decreases in blood cell
numbers at higher exposure concentrations. In one study evaluating immunological markers in a
cohort of plywood workers, exposure to 0.2-0.8 mg/m3 formaldehyde was positively correlated
with increased serum interleukin (IL)-10 and IL-4, alongside decreased IL-8 and interferon-gamma
(IFN-y); no significant changes in total lymphocyte or T cell numbers were observed in this study
flia etal.. 20141. These cytokine changes are consistent with observations of increased plasma IL-4
and decreased IFN-y in a short-term rat study at >6.2 mg/m3 that reported corresponding
lymphocyte genotoxicity (Im etal.. 2006). Workers with higher formaldehyde exposure
(i.e., 1.8 mg/m3) exhibited formaldehyde-associated aneuploidy and had decreased peripheral
blood levels of various chemokines and cytokines, including IL-10 (Seow etal.. 2015). These
observations suggest the possibility of a shift in the functional activation of immune effector cells
such as T lymphocytes and macrophages at formaldehyde concentrations below which overt
changes in cell number become observable; however, studies specifically testing this possibility
have not been performed.
While changes in subpopulations of peripheral leukocytes and circulating levels of
cytokines may indicate the potential for some manner of dysfunction in the host immune system,
direct observations of dysfunction would be most informative; however, only a few studies
specifically examined the potential for events such as immunosuppression in either humans or
experimental animals following formaldehyde exposure. In addition, while studies of immune
function in the affected airways indicate a probable effect of formaldehyde exposure, studies
evaluating immunosuppression at distal sites are inadequate (see Section 1.2.3). In the airways of
exposed humans, indirect evidence of decreased immune capacity exists, including decreased
resistance to URT infection at 0.9 mg/m3 formaldehyde with chronic exposure (Lvapina etal..
2004). and an increased rate of LRT infection in infants exposed to 0.02 mg/m3 during their first
year of life fRoda etal.. 20111. These observations in humans are consistent with the decreased
bactericidal activity of leukocytes from the lungs of mice acutely exposed to >1 mg/m3
This document is a draft for review purposes only and does not constitute Agency policy.
1-527 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
formaldehyde (Takab etal.. 19921. and the enhanced malignancy and growth of lung tumors, in
association with decreases in NK cell numbers and activity, formed by an injection of syngeneic
melanoma cells in mice following exposure to 12 mg/m3 fKim etal.. 2013al. Observations related
to systemic immune dysfunction, including increased survival to Listeria monocytogenes infections
and reduced melanoma tumor mass in B6C3F1 mice fDean et al.. 19841. and increased
autoantibodies in exposed adults (Thrasher et al.. 1990) are mixed and inconclusive. Thus, while it
appears that formaldehyde exposure can suppress immune function in the airways, the pattern of
effects across tissue compartments (i.e., URT, LRT, blood and lymphoid tissues) remains unclear.
Together, the evidence supports a decrease in peripheral blood WBC counts in
formaldehyde-exposed humans (see Section 1.2.3), although some heterogeneity across studies has
been reported in terms of the directionality and magnitude of changes in specific leukocyte subsets
and in levels of soluble immunomodulatory molecules (see Section 1.2.3). Considerable
heterogeneity has also been observed in relation to the formaldehyde concentration or exposure
duration reported for the different observations, further complicating interpretation. Despite this
variability, the available data suggest that formaldehyde exposure modifies immune system
function across a range of concentrations and durations, with changes in specific leukocyte
subpopulations becoming more robust and consistent following exposure to >0.5 mg/m3 (see
Section 1.2.3).
Relevance to LHP carcinogenesis
While many of the changes reported following formaldehyde exposure could create a more
permissive environment favoring tumor growth and progression, evidence does not exist to
determine whether these changes in immune cell populations or cytokine profiles significantly
impact tumor immunosurveillance or cause chronic inflammation; therefore, any specific role for
altered immune function in formaldehyde-associated leukemogenesis remains unclear. Changes in
immune cell subpopulations, distribution, and activation have a complex relationship with
carcinogenesis in terms of tumor suppressing or enhancing activity (Hanahan and Weinberg. 2011).
For example, immune suppression is associated with a greater risk of hematopoietic cancers
(Bassigetal.. 2012). and chronically immunosuppressed human transplant recipients are at
increased risk for developing myeloid neoplasms fMorton et al.. 20141. Together, this evidence
shows that the immune system can operate as a significant barrier to LHP carcinogenesis (Corthav.
20141. In addition, impaired tumor immunosurveillance could result from deficiencies in the
development or function of cytotoxic T lymphocytes (CTLs), type 1 T-helper (ThI) cells, or NK cells,
which might lead to demonstrable increases in tumor incidence (Hanahan and Weinberg. 2011).
Conversely, inflammatory immune effector cells (i.e., neutrophils, macrophages, type 2 T-helper
[TH2] cells, and T and B lymphocytes) can release growth factors and other tolerogenic signaling
mediators, which permit tumor growth. The release of reactive oxygen species (ROS) from such
cells can be actively mutagenic for nearby cancer cells and accelerate their genetic evolution toward
heightened malignancy fCoussens and Werb. 20021. While NK cells play a prominent role in
This document is a draft for review purposes only and does not constitute Agency policy.
1-528 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
infection and carcinogenesis in the airways (and likely elsewhere in the body), the studies and
evidence reporting effects on these cells in any tissue system following formaldehyde exposure are
considered weak. Overall, despite the potential for these associations, cell type-specific changes
indicative of impaired immunosurveillance or enhanced tumor growth have not been conclusively
demonstrated following formaldehyde exposure, particularly at lower levels.
The observed changes in soluble immune factors are similarly difficult to interpret. In
addition to the evidence of increased IL-4 in the blood, multiple observations, primarily from
allergen sensitization studies in rodents, suggest that IL-4 production in the lower respiratory tract
(LRT) in response to antigen stimulation is further exacerbated by formaldehyde
exposures >0.3 mg/m3 (see Sections 1.2.2-1.2.3). Although the specific implications of cytokine
changes for tumor development and progression is still emerging, IL-10 and IL-4 in particular are
important cytokines in tumor immunology fLi etal.. 20091. and the tendency of IL-4 and IL-10 to
increase while IFN-y decreases (see Section 1.2.3) is a pattern commonly observed in human cancer
patients, including those diagnosed with some LHP cancer subtypes (Shurin et al.. 1999). However,
the relationships between cell signaling molecules and affected components of the immune system
are complex, and an understanding of how these molecular changes might relate specifically to
immune cell dysfunction, and further, to LHP carcinogenesis, is incomplete.
Evidence does not exist to describe how formaldehyde exposure might cause the observed
systemic changes in immune system-related responses. While it is possible that these changes
might result from disturbed bone marrow hematopoiesis resulting indirectly from formaldehyde
exposure, studies specifically testing this possibility were not identified. Alternatively, it is possible
that altered immune system responses are related to formaldehyde-induced toxicity at the URT.
Interestingly, while peripheral blood CTL levels were generally decreased in individuals exposed to
formaldehyde concentrations >0.5 mg/m3, respiratory tract CTL levels (and total WBC counts)
tended to increase in rodent studies, although the latter data are limited to short-term exposure at
much higher formaldehyde levels (see Appendix A.5.6). It is possible that CTLs were preferentially
recruited from the peripheral blood into the URT, thus explaining their depletion from the former
and accumulation in the latter tissue; however, none of the identified human studies report WBC
counts from both peripheral blood and POE tissue compartments, and the available animal data
likewise cannot adequately inform this hypothesis.
Overall, while several studies indicate effects on hematopoietic cell populations and
secreted factors, for which exposure concentration may be an important determinant, the impact of
these changes on leukemogenesis cannot be clearly discerned.
Evidence of formaldehyde-induced oxidative stress
Similar to observations in the airways, inhaled formaldehyde has been associated with
biomarkers of oxidative stress in distal tissues (see Section 1.2.3 and Appendix A.5.6).
Some human studies have evaluated changes in markers of oxidative stress in blood or
urine in relation to formaldehyde exposure, and also have attempted to determine whether the
This document is a draft for review purposes only and does not constitute Agency policy.
1-529 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
oxidation of lipid membrane components might be associated with the presence of formaldehyde-
induced DNA damage. Two studies provide evidence of oxidative stress-related ge no toxicity or
mutagenicity, including elevations in malondialdehyde-deoxyguanosine (MldG) adducts
(i.e., exocyclic DNA adducts formed as byproducts of lipid peroxidation) in WBC DNA with exposure
to an average formaldehyde concentration as low as 0.07 mg/m3 fBono etal.. 20101. This finding is
indirectly supported by an observed association between increases in malondialdehyde and p53
protein (a potential biomarker of carcinogenicity; see discussion of the potential for p53 to
contribute to URT carcinogenesis in Section 1.2.5) in plasma with urinary formate levels (which
may serve as an imprecise marker of formaldehyde exposure) among cosmetic workers fAttia et al..
20141. Additional evidence that formaldehyde exposure is associated with oxidative stress is
provided by a study that reported increased urine levels of 15-F2t isoprostane (a sensitive, but
nonspecific marker of oxidative stress) from formaldehyde-exposed workers fRomanazzi etal..
2013): although this marker is not specific to changes in a particular tissue, strong correlations
between measurements from urine and plasma (Rodrigo etal.. 2007: Morrow etal.. 19951 suggest
similarly elevated isoprostanes in the workers' blood. Somewhat in support of the observations in
humans, several animal studies in two species observed increases in markers of oxidative stress
following acute or short-term formaldehyde exposure to a range of formaldehyde concentrations
including <1 mg/m3; however, these studies had notable methodological limitations, and it is not
clear whether these changes persist with long-term exposure (see Section 1.2.3). Suggestive
evidence of elevated indicators of formaldehyde-induced oxidative stress and inflammation have
been reported in bone marrow from exposed mice at >0.5 mg/m3 formaldehyde; however, these
animals were coexposed to methanol, drawing into question the validity of these findings (formalin
was the formaldehyde source; fYu etal.. 2014: Ye etal.. 2013b: Zhang et al.. 2013bll. These limited
studies also observed higher rates of DNA damage in bone marrow. Overall, together with the
genotoxicity data, this evidence indicates the likely presence of DNA damage and, possibly
coincidentally, the likely presence of elevated oxidative stress in circulating leukocytes, although
the data are insufficient to describe this potential relationship in terms of duration or concentration
of exposure.
Studies of susceptibility to DNA damage conferred by polymorphisms in genes coding for
enzymes with activity that either increases or decreases oxidative damage observed greater
genotoxicity associated with formaldehyde exposure and polymorphic variation in genes encoding
the ROS-inducer, CYP2E1 (more damage associated with wildtype), and the detoxifying enzyme,
GSTP1 (more damage associated with variant) (Costa etal.. 2015). although another study using a
different measure of DNA damage found a marginal increase in susceptibility among exposed with
the wildtype GSTP1 allele compared to the variant genotype fliang etal.. 20101. However, DNA
damage in human PBLs was not increased to a greater degree in formaldehyde-exposed human
cohorts with increased susceptibility to oxidative damage due to glutathione-S transferase (GSTM1
This document is a draft for review purposes only and does not constitute Agency policy.
1-530 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
or GSTT1) null genotype (Santovito etal.. 2011: Jiang etal.. 2010: Costa etal.. 20081: therefore,
these results remain inconclusive.
Relevance to LHP carcinogenesis
Together, the available data suggest that oxidative stress may be elevated at distal sites
following formaldehyde exposure in humans, rats, and mice; however, available studies of genetic
susceptibility in exposed workers are not adequate to draw conclusions. Considered alongside the
evidence of oxidative stress in the airways (Sections 1.2.1-1.2.2), the data reporting oxidative stress
at distal sites suggest that formaldehyde exposure might increase the production of potentially
harmful factors throughout the body. If sufficiently severe or sustained for a prolonged duration,
oxidative stress could perturb the function of circulating leukocyte populations including HSPCs,
increasing lipid, protein, and DNA oxidation, causing DNA strand breakage, as well as altering
cellular energetics and signaling pathways (Mikhedetal.. 2015). Regarding any potential role in
LHP carcinogenesis, the impact of oxidative stress-induced DNA damage on gene or chromosomal
changes could be similar to the damage caused by a variety of directly DNA-reactive compounds
fMchale etal.. 2012: DeMarini etal.. 20001. The available evidence is inadequate to determine what
role formaldehyde-associated oxidative stress may play in LHP carcinogenesis, although impacts on
leukocyte genotoxicity, increased HSPC mobilization, or immunomodulation are all plausible
consequences of systemically elevated oxidative stress.
Data are not available to describe how formaldehyde might cause oxidative stress outside of
the airways. Similar to changes in leukocyte cell numbers, this may be secondarily due to sustained
airway inflammation, which could cause the release of factors from the inflamed tissue (s) into the
circulation that result in increased oxidative stress; however, no studies have examined this
possibility. In summary, the potential relationship of increased systemic oxidative stress to LHP
carcinogenesis is unknown.
Evidence of formaldehyde-induced changes in the bone marrow niche
As noted above, there is some evidence of pancytopenia in formaldehyde-exposed humans
that may indicate disturbance of or cytotoxicity in the bone marrow niche at higher environmental
exposures. In F344 rats, bone marrow hyperplasia was elevated following chronic exposure to
18 mg/m3 formaldehyde fBattelle. 19821. In two chronic ratbioassays fKamataetal.. 1997:
Sellakumar et al.. 19851 and a short-term (8-week) study of p53 deficient mice fMorgan etal..
2017). the authors evaluating nonrespiratory tissues did not provide details regarding
nonneoplastic histopathology in tissues outside the URT, and the incidence of hematopoietic
neoplasms did not appear to be elevated in any of these studies. In female B6C3F1 mice exposed
similarly to the F344 rats above, hyperplasia was not observed in the bone marrow, spleen or
lymph nodes (Battelle. 1982). Evaluations of changes in numbers of bone marrow megakaryocytes
were likewise fairly equivocal in mice exposed to 0.5-20 mg/m3 formaldehyde (see
Appendix A.5.6).
This document is a draft for review purposes only and does not constitute Agency policy.
1-531 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Two studies in mice suggest that cell subpopulations in the bone marrow niche might be
differentially affected by formaldehyde exposure. Specifically, in a 20-week study, a dose-
dependent decrease in the ratio of immature to mature RBCs (PCE/NCE ratio) in the bone marrow
was observed after exposure to 1 and 10 mg/m3 formaldehyde for 2 hours per day fLiu etal..
20171: however, there was no corresponding change in micronucleus rate. A short-term, 2-week
study indicated that in vivo formaldehyde exposure of 3 mg/m3 caused a decreased formation of
BFU-E (erythroid progenitor) and CFU-GM (granulocyte/monocyte progenitor) colonies in cultures
from bone marrow or spleen (Zhao etal.. 2020b). However, in both of these studies the
formaldehyde source is presumed to have been formalin, which prevents interpretation of these
results at systemic sites as reliable and highlights this as an area deserving of additional research.
As noted above, a dose-related increase in bone marrow DPXs was observed in BALB/c
mice exposed to 0.5-3.0 mg/m3 formaldehyde generated from evaporating formalin fYe etal..
2013a). However, the presence of methanol in the formalin confounds interpretation of the
potential for systemic formaldehyde effects, as the co-administered methanol could be rapidly
absorbed, distributed to the bone marrow, and locally metabolized to formaldehyde (see
Appendix A.2, A.4). Consistent with this hypothesized contribution of methanol, neither DPXs nor
DNA mono adducts were elevated in rodent bone marrow exposed via paraformaldehyde fLenget
al.. 2019: Lu etal.. 2010a: Heck and Casanova. 2004: Casanova and Heck. 1987: Casanova-Schmitz
etal.. 1984a). While bone marrow has not been evaluated in exposed human cohorts, elevations in
WBC DPX levels have been reported in some human workers chronically exposed to concentrations
>0.5 mg/m3 (Shaham etal.. 2003: Shaham et al.. 1997). but not consistently in others (Lin etal..
20131.
In general, the data relevant to potential formaldehyde-induced changes in the bone
marrow niche were fairly weak and inconsistent across the available studies, although the minimal
data available indicate that additional studies are warranted.
Relevance to LHP carcinogenesis
Bone marrow niches consist of bone marrow mesenchymal stem cells (BM-MSCs) and HSPC
pairings under tight regulation by local input from the surrounding microenvironment, as well as
long-distance cues from soluble signaling mediators (e.g., hormones, cytokines, eicosanoids) and
the autonomic nervous system (Cristina Lo Celsol. 2011). Aberrant bone marrow stroma can lead
to HSPC dysfunction including MDS fCristina Lo Celsol. 20111. a precursor to AML. Therefore,
altered stromal behavior could affect HSPC quiescence and mobilization as well as directly induce
the expansion of leukemic clones over normal cells.
Although inhaled formaldehyde does not likely reach the bone marrow to elicit direct
effects analogous to exposure in the URT (see Appendix A.2), formaldehyde-induced effects in the
URT could indirectly affect the bone marrow microenvironment or "niche" in several ways,
including inflammation or induction of systemic immune responses (see Section 1.2.3), oxidative
This document is a draft for review purposes only and does not constitute Agency policy.
1-532 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
stress (see Sections 1.2.3), hormonal or cytokine changes that affect BM-MSC and HSPC
interactions, and disrupted regulation of HSPC mobilization from the niche. However, evaluations
of bone marrow following formaldehyde inhalation have been limited to histological or genotoxic
endpoints in experimental systems, with no information available regarding either molecular
changes in stromal cell function or HSPC activation, differentiation, or mobilization.
The sympathetic nervous system has some control over the mobilization and circulation
rate of bone marrow progenitor cells including HSPCs (Elenkov etal.. 20001. While formaldehyde
exposure has been shown to activate the trigeminal nerve in the rodent URT via transient receptor
potential channel stimulation at low concentrations ffMcnamara et al.. 20071: see Section 1.2.1), no
studies have examined whether or how this might be indirectly related to regulation of HSPC
mobilization or hematopoiesis; however, it is considered unrealistic that activation of neural
pathways relaying irritant and pain information would convey excitatory or inhibitory signals to
networks responsible for HSPC regulatory functions.
It is difficult to reconcile these disparate observations across the available data streams: the
general lack of bone marrow toxicity in experimental model systems corresponds with no excess
leukemia reported in chronic rodent bioassays, while the varied fluctuations in immune cell
subpopulations, including some evidence of pancytopenia in the peripheral blood of chronically
exposed humans (Section 1.2.3), is consistent with the evidence of leukemia induction in humans.
It is possible that humans are more sensitive to the hematotoxic effects of formaldehyde than either
rodents or nonhuman primates (Goldstein. 2011). as has been noted in the context of chromosomal
damage resulting from direct leukemogens (e.g., benzene; (French etal.. 2015: IARC. 2012: Mchale
etal.. 2012)1. However, mechanism(s) responsible for any potential differential sensitivity remain
to be elucidated. Based on the currently available data, no conclusions can be drawn regarding the
potential involvement of formaldehyde exposure-induced indirect effects on the bone marrow
niche in LHP carcinogenesis.
Evidence of formaldehyde-induced changes in gene expression or posttranscriptional regulation in
peripheral blood leukocytes or bone marrow
Few studies have evaluated the effect of formaldehyde exposure on microRNA (miRNA) or
messenger RNA (mRNA) levels from non-POE tissues in vivo, and none evaluated chronic
exposures. In a small study where human volunteers [N = 21) were variably exposed to <1 mg/m3
formaldehyde for 5 days, statistically significant changes in mRNA expression were observed in
cells from either nasal biopsies or whole blood samples; however, study limitations prevent
interpretation of the changes to be a result of formaldehyde exposure (Zeller etal.. 2011). In F344
rats, significant changes in both miRNA and mRNA expression were reported in the nasal
epithelium and circulating white cells following inhalation exposure to 2.5 mg/m3 formaldehyde
for <4 weeks, primarily involving pathways related to immune/inflammatory response, apoptosis,
and proliferation; no significant changes were observed in miRNA samples from the bone marrow,
and mRNA transcript levels were not evaluated (Rager etal.. 2014). A majority of the reported
This document is a draft for review purposes only and does not constitute Agency policy.
1-533 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
changes appeared to be tissue- and exposure duration-specific, and only expression of one
transcript was consistently affected (miR-326 levels increased) in the WBCs across exposure
conditions fRager etal.. 20141. As these endpoints have not been well-studied, conclusions cannot
be made regarding the consistency and reproducibility of these data across studies.
Relevance to LHP carcinogenesis
Epigenetic mechanisms such as miRNA-mediated regulation of mRNA may play a role in the
pathogenesis of LHP malignancies (Yendamuri and Calin. 20091. For example, differential miRNA
expression profiles have been reported between normal and leukemia cells, and among LHP cancer
subtypes such as AML and ALL (Marcucci etal.. 2009: Mi etal.. 20071. However, the bone marrow
represents a heterogeneous population of cells, and in the context of variable and temporal
responses induced following formaldehyde exposure, such gene expression array results can be
difficult to assimilate and interpret (Weinberg. 2014).
Although the potential role of miR-326 in LHP carcinogenesis is unknown, increased serum
miR-326 expression was associated with bone matrix turnover and positively correlated with lung
cancer bone marrow metastasis fValencia etal.. 20131. Considering that WBCs are a highly
heterogeneous population, of which only a small fraction is likely to contain target cells of interest
in LHP carcinogenesis (i.e., HSPCs), the observation of altered miRNA and mRNA levels in WBCs
from rats provides very limited evidence that supports the biological plausibility for other
formaldehyde-induced effects, such as genotoxicity (Appendix A.4) in the peripheral blood cells of
occupationally exposed humans. Additional studies examining potential epigenetic and
transcriptional mechanisms related to LHP carcinogenesis in non-POE tissues following
formaldehyde exposure are needed to confirm and expand the observations from this limited set of
studies.
Discussion of mechanistic evidence relevant to LHP carcinogenesis.
While the mechanistic events evaluated in the context of formaldehyde-associated LHP
cancer are similar to those described for well-described human leukemogens (IARC. 2012: Mchale
etal.. 2012). the specific mechanism(s) of LHP cancer induction are not understood, which
complicates the construction of any simple, linear MOA (Mchale etal.. 2012). Therefore, a network
of plausible mechanistic events or pathways was discussed, including specific aspects of
genotoxicity and mutagenicity, hematologic effects, oxidative stress, and changes in gene
expression or regulation, consistent with previous analytical frameworks employed in the
evaluation of LHP carcinogenesis (NRC. 2014b). The most pertinent evidence and conclusions for
potential mechanistic events associated with formaldehyde induction of LHP cancers are
summarized inTable 1-66.
It is possible that potential LHP target cells (e.g., HSPCs) are affected in the URT tissue, via
direct interactions with formaldehyde, given observations that stem cell precursors can traverse
between the URT and bone marrow. However, the concentrations of inhaled formaldehyde
This document is a draft for review purposes only and does not constitute Agency policy.
1-534 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
reaching sites through which HSPCs might traverse (e.g., lymphatic URT tissue), as well as the
population of HSPCs present in the URT at any one time, would both be expected to be quite low,
although no specific data address these unknowns. Indirect toxicity to HSPCs in the URT also might
result from inflammation or oxidative stress in these tissues. Furthermore, genotoxic effects on
HSPCs, as well as immune cell toxicity and dysfunction, may occur in peripheral blood or bone
marrow via indirect effects of formaldehyde-associated inflammation in the URT resulting in
systemic oxidative stress and changes in gene expression or regulation. However, no studies of
formaldehyde exposure investigating these hypotheses have been conducted.
Evidence from evaluation of respiratory tract and oral cells (nasal and buccal epithelium),
and circulating leukocytes (e.g., HSPCs and PBLs) consistently demonstrates increased levels of
Comet assay-detectable DNA damage, as well as MN, CAs, and SCEs associated with formaldehyde
exposure from a variety of occupational cohorts. Some of the genotoxic endpoints observed in
circulating blood cell progenitors from formaldehyde-exposed workers have also been specifically
observed in patients with AML (Mchale etal.. 2012: Bowen and Hannigan. 2006). while other
endpoints observed in PBLs, such as MN and CA, are generally regarded as biomarkers associated
with increased human risk for a variety of cancers, including LHP malignancies fKirsch-Volders et
al.. 2014: Fenech etal.. 2011: Bonassi etal.. 2008: Bonassi etal.. 2007: Bonassi etal.. 2004bl: see
Section 1.2.5, Evidence on Mode of Action for URT Cancers). Genotoxicity to circulating PBLs may
also serve as a surrogate biomarker of genotoxicity in HSPCs, which may play a more direct role in
LHP carcinogenesis. No information from the available formaldehyde studies exists to evaluate this
potential association.
Following formaldehyde exposure, the available evidence supports the following
observations: (a) elevated levels or severity of DNA or chromosomal damage in circulating human
blood cells, including in both myeloblasts and mature lymphocyte populations; (b) the specific
nature of DNA damage in circulating human leukocytes exhibits aneugenic characteristics similar to
damage reported in humans with or at increased risk for AML; and (c) that the human immune
system is impacted, possibly as a function of formaldehyde concentration, in a complex manner.
Formaldehyde exposure is associated with reductions in immune cell populations, although other
lines of evidence indicate stimulation of some immune cell populations, which might reflect a
complex concentration or duration dependence in the pattern of effects. The observations of DNA
or chromosomal damage in exposed humans, including aneuploidy, and reductions in immune cell
populations associated with comparable formaldehyde levels (>0.5 mg/m3) provide coherent
evidence suggesting that these effects may be related.
Despite the internal consistency of many of the individual effects described above regarding
formaldehyde-induced damage to target cells and biomarkers of genotoxicity in circulating mature
PBLs in humans, there is a general lack of understanding regarding both how formaldehyde
exposure might cause these changes, as well as how these mechanistic events may lead to LHP
cancer. Regarding the latter, for example, any specific effects on the bone marrow niche have not
This document is a draft for review purposes only and does not constitute Agency policy.
1-535 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
been studied in exposed humans, and the evidence from the available animal studies is generally
inconclusive.
The relationships between leukocyte responses in peripheral blood and formaldehyde
exposure are complex; studies observed changes in different cell populations, which were both
increased and decreased across studies, although some tentative patterns could be discerned,
particularly at exposure concentrations >0.5 mg/m3. The mechanisms responsible for these
observations are unclear, as is any specific contribution of these mechanistic events to LHP
carcinogenesis. Likewise, although some evidence exists to support increased systemic oxidative
stress associated with formaldehyde exposure, its role in targets of LHP cancers is also unclear, and
any specific impacts on immune function or tumor immunosurveillance remain to be determined.
Alternative hypotheses
A hypothesized scenario that does not require bone marrow cytotoxicity is that HSPCs
damaged in the URT tissues do not return to the bone marrow but form local neoplastic foci.
However, there is no evidence supporting this possibility. Collections of neoplastic myeloid cells
localized in extramedullary tissues (myeloid or granulocytic sarcomas occurring outside of the
medulla of the bone), are associated with MDS and AML but are not commonly reported in human
nasal tissue fYamamoto etal.. 2010b: Pavdas etal.. 2006: Prades etal.. 20021. Myeloid sarcomas
have not been specifically associated with formaldehyde exposure, although these lesions are
frequently misclassified as NHLs in patients without concurrent MDS or AML (Yamamoto etal..
2010a). However, HSPCs do not travel through the nasopharynx-associated lymphoid tissue
(Massberg et al.. 2007). and may not be the target cell population responsible for nasal myeloid
sarcoma. This observation could suggest that the nasal tissue does not provide a suitable niche
microenvironment for sustaining neoplastic myeloid cell expansion (Granick etal.. 2012: Wilson et
al.. 20091.
Inferences can be made by extending the proposed hypothesis of circulating or nasal-
resident HSPCs as LHP cancer target cells to the spectrum of effects commonly associated with
leukemias induced by exposure to other agents (U.S. EPA. 2005al. Although the results of this
exercise cannot dismiss the biological plausibility of the events evaluated with specific data from
the formaldehyde exposure database, it may illustrate that the identified set of mechanistic events
are incomplete. For example, if HSPCs are exposed to the genotoxic activity of formaldehyde as
they transit through the URT tissues, and then proceed back to the bone marrow to progressively
become leukemogenic, then other genotoxic URT carcinogens could potentially have a similar effect
and be associated with both URT and bone marrow cancers. The agents in which both
nasopharyngeal cancer and leukemias have been associated with human exposures are tobacco
smoke (IARC. 2012). which contains formaldehyde, and formaldehyde itself (IARC. 2012). Most
agents associated with nasal cancer in humans have not also been associated with leukemia
induction, despite displaying variable genotoxic activity, except for those agents that are also
systemically available and hematotoxic flARC. 20121. This suggests that genotoxicity and
This document is a draft for review purposes only and does not constitute Agency policy.
1-536 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
distribution to the URT alone may not be sufficient to induce LHP carcinogenesis. It has been
proposed flARC. 20121 that well-studied human leukemogens (e.g., ionizing radiation, benzene,
chemotherapeutics) induce hematotoxicity more frequently or to a greater extent than neoplasia,
which would be consistent with DNA damage more frequently resulting in bone marrow cell death
than progenitor transformation. However, this observation cannot rule outleukemogenesis driven
by mechanisms other than ge no toxicity-induced bone marrow cytotoxicity.
Gaps in understanding of formaldehyde exposure-related LHP carcinogenesis
As discussed in this section, there appears to be a lack of concordance between evidence
from chronic rodent bioassays and human epidemiological evidence regarding incidence of LHP
cancers. Moreover, contrary to the consistent evidence supporting genetic damage to circulating
leukocytes in formaldehyde-exposed humans, few positive associations have been reported in
rodent bioassays. This MOA discussion evaluated the mechanistic database pertinent to
leukemogenesis based on the fundamental assumption that exogenous formaldehyde is not
distributed appreciably beyond POE tissues. Differences in physiology between humans and
rodents, as well as the relative insensitivity of rodent models to reflect the human pathogenesis of
AML, may together contribute to the potential lack of concordance between the abundant human
epidemiological data and the more limited results (e.g., most bioassays did not examine tissues
relevant to LHP cancers in detail) from rodent bioassay data.
Conclusion
The available evidence supports some events that could contribute to plausible mechanistic
pathways relating formaldehyde exposure to LHP carcinogenesis. However, the database was
insufficient to support the evaluation or development of any specific MOA. Although this analysis
represents an independent evaluation of all identified, pertinent, primary information, it is
informative to note that the conclusions reached herein are consistent with those reported
following previous reviews by authoritative scientific organizations, including IARC (2012). NTP
(2014). and the NRC (2014b). Notably, there was widespread, general agreement that the available
evidence is largely consistent and strong, particularly for genotoxicity in circulating blood cells.
Both temporal and exposure-response relationships have been demonstrated in studies of humans,
and mechanistic pathways exist that support a biologically plausible relationship between
formaldehyde exposure and cancer, even though the mechanistic pathways explaining such
systemic effects are unclear (NRC. 2014b). It is important to note that systemic delivery of
formaldehyde is not a prerequisite for the observed mechanistic changes, as some of the reported
systemic effects might result from direct interactions with formaldehyde in the URT, while others
could plausibly result indirectly from events such as URT irritation, cytotoxicity, oxidative stress,
and inflammation locally initiated at the POE. Further, the evidence for other effects at distal sites
was compelling. This evidence included increased female reproductive and developmental toxicity
and male reproductive toxicity, based on studies of experimental animals and workers exposed to
This document is a draft for review purposes only and does not constitute Agency policy.
1-537 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 high formaldehyde levels, as well as LRT disease (i.e., current asthma symptoms and decreased
2 asthma control in population-based epidemiology studies). It is plausible that these effects could
3 result indirectly from events occurring in the URT. While the available mechanistic database has
4 limitations, this does not detract from the strength of the association between formaldehyde
5 exposure and myeloid leukemia in epidemiology studies.
6 Conclusions from MOA evaluation
7 Support for the hypothesized mode of action in experimental animal models
8 While evidence for the several identified mechanistic events ranges from strong and
9 consistent to inadequate (see Table 1-66), the supporting evidence was drawn primarily from
10 studies of exposed humans; no single MOA could be assembled and evaluated from the limited
11 relevant experimental animal data available.
12 Relevance of the hypothesized mode of action to humans
13 Due to the paucity of pertinent mechanistic information, no single, stochastic MOA was
14 identified for LHP cancers associated with formaldehyde exposure. However, evidence supporting
15 the identified mechanistic events was obtained primarily from studies of exposed human cohorts,
16 and thus the mechanistic events are all relevant or of presumed relevance to human LHP cancer
17 risk (see Table 1-66).
Table 1-66. Summary conclusions regarding plausible mechanistic events
associated with formaldehyde induction of lymphohematopoietic cancers
Hypothesized
mechanistic
event
Evidence informing mechanistic event
Human
relevance
Weight-of-evidence
conclusion and
biological plausibility
2.1 Formaldehyde-
induced DNA
damage to
peripheral blood
leukocytes
HSPC aneuploidy and structural chromosome damage in
myeloid progenitors (CFU-GMs) from human workers
occupational^ exposed to median levels of 1.6 mg/m3 (Lan
et al., 2015; Zhang et al., 2010).
• 'T* Monosomy and polysomy in multiple chromosomes
(especially monosomy 1, 5, 7) consistent with damage
observed in patients with MDS or AML (Lan et al.,
2015)
• 'T* Breaks, deletions, and translocations in chromosome
5
'T* genotoxicity in circulating PBLs from inhalation-exposed
humans, including increases in strand breaks, MN, CA (see
Appendix A.4: (Kirsch-Volders et al., 2014) NBUDs, or SCE
induction at >0.14 mg/m3 (Jiang et al., 2010), and DPXs at
higher exposures (Lin et al., 2013; Shaham et al., 2003).
• 'T* DPXs in PBLs from mice after inhalation of
formaldehyde generated from formalin (Ye et al.,
Yes.
Evidence
comes
primarily
from
exposed
humans.
Strong and consistent
human data exist
associating formaldehyde
exposure with various
genotoxic outcomes in
myeloid progenitors and
PBLs, and exposure-
response relationships
demonstrated.
Genotoxicity in circulating
leukocytes shows
concordance with similar
endpoints in POE tissues.
Aneugenic damage
observed in CFU-GMs
from formaldehyde-
exposed human workers is
associated with MDS or
AML in humans. Together
this evidence constitutes
This document is a draft for review purposes only and does not constitute Agency policy.
1-538 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Hypothesized
mechanistic
event
Evidence informing mechanistic event
Human
relevance
Weight-of-evidence
conclusion and
biological plausibility
2013b), although results mav be confounded bv
methanol coexposure
• No increase in DPXs in peripheral blood or bone marrow
of monkeys or rats exposed via paraformaldehyde (Lai
et al., 2016; Casanova and Heck, 1987)
• DNA damage in human PBLs is consistently associated
with genotoxicity in human POE tissues (e.g., exfoliated
buccal and nasal epithelial cells) in studies evaluating both
tissues after longer-term exposures (see Appendix A.4;
see Section 1.2.5)
the strongest support for
the biological plausibility
for LHP induction resulting
from formaldehyde
exposure.
2.2 Evidence of
formaldehyde-
induced impacts
other than
genotoxicity on
circulating blood
cell populations,
including
inflammatory
changes and/or
immune system
dysfunction
\1/ CFU-GM colony formation in human workers
occupational^ exposed to median levels 1.6 mg/m3 (Zhang
et al., 2010), which mav reflect not only altered bone
marrow progenitor cell viability, but also immune dysfunction
or altered activation.
• Numerous published studies reporting divergent changes
in various peripheral blood cell populations from
formaldehyde-exposed humans (see Section 1.2.3;
Appendix A.5.6), including:
'T* Pancytopenia and consistent decreases in total
WBCs
\1/ or ^ in some lymphocyte populations, with decreased
CD8 T cells likely at concentration >0.5 mg/m3. Fluctuations
in immune cell numbers and immune/inflammation markers
show a complex pattern with concentration, with decreases
in blood cell number and decreased cytotoxic response
generally at higher concentrations, some of which are
consistent with observations in AML patients (Kim et al.,
2015). Other studies indicate immune cell activation
generally observed at lower concentrations <0.36 mg/m3.
Yes. Most of
the available
data comes
from human
studies.
The evidence supporting
changes in populations or
function of circulating
blood leukocytes following
human exposure to
formaldehyde is strong in
terms of a frequency of
alterations, but different
patterns in changes are
reported (e.g., specific
direction of changes in
various lymphocyte
subpopulations, or in
blood levels of soluble
signaling mediators). LHP
cancer risk increases with
loss of normal immune
function.
This document is a draft for review purposes only and does not constitute Agency policy.
1-539 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Hypothesized
mechanistic
event
Evidence informing mechanistic event
Human
relevance
Weight-of-evidence
conclusion and
biological plausibility
2.3 Formaldehyde-
induced systemic
oxidative stress
• 'T* MldG adducts in whole blood DNA from pathologists,
compared to workers and students in other science labs
(Bono et al., 2010), elevated plasma MDA and
plasma p53 associated with each other and with urinary
formate concentrations (an imprecise marker of
formaldehyde exposure) among cosmetics workers
(Attia et al., 2014), and 'T* 15-F2t isoprostane levels
in the urine of formaldehyde-exposed workers
(Romanazzi et al., 2013)
• Inconclusive evidence for and against involvement by
genes that regulate oxidative stress in formaldehyde
associations with DNA damage risk in PBL in humans (see
Appendix A.4)
• \1/ GSH, 'T* ROS, 'T* MDA in bone marrow, peripheral
blood mononuclear cells, liver, spleen, and testes (Ye et
al., 2013b), although markers of oxidative stress were
not correlated with DPXs and results may be confounded
by methanol coexposure.
Yes. Some
human data
available,
and results
from
experimental
models are
presumed
relevant to
humans.
Limited human and rodent
evidence supports the
association between
formaldehyde exposure
and induction of oxidative
stress beyond the POE.
While biologically
plausible, the available
evidence is inadequate to
determine what role such
oxidative stress may play
in LHP carcinogenesis.
2.4 Formaldehyde-
induced changes in
the bone marrow
niche
• 'T* Bone marrow hyperplasia in rats from one study
(Kerns et al., 1983; Battelle, 1981), but unclear
if other results were negative or null (Kamata et al.,
1997; Sellakumar et al., 1985) due to imprecise
reporting
• Dose-related 'T* DPXs in the bone marrow of formalin-
exposed mice (Ye et al., 2013b), although results
may be confounded by methanol coexposure
• HSPC mobilization and the BM-MSC niche is regulated by
cytokines, hormones, and signals, which may be
distributed through circulation as a result of inflammation
although these effects have not been directly evaluated
following formaldehyde exposure
Yes.
Available
data are
from
experimental
models
presumed
relevant to
humans.
The limited evidence
available is currently
inadequate to evaluate
any effect on bone
marrow or stromal cells
following formaldehyde
exposure, although such
an effect appears
consistent with current
understanding of
hematopoiesis.
2.5 Evidence of
formaldehyde-
induced changes in
gene expression or
posttranscriptional
regulation in
peripheral blood
leukocytes or bone
marrow
Limited study reported some statistically significant
differences in mRNA expression in either nasal or whole
blood samples from human volunteers associated with 5-day
exposures up to 1 mg/m3 formaldehyde; however, study
limitations prevent interpretation that results were related to
formaldehyde exposure (Zeller et al., 2011). In F344
rats, significant changes in both miRNA and mRNA expression
were reported in the nasal epithelium and circulating WBCs
following inhalation exposure to 2.5 mg/m3 formaldehyde for
1 or 4 weeks; no changes were observed in miRNA expression
in the bone marrow, and mRNA was not evaluated (Rager
et al.. 2014).
• "Immune system/inflammation" markers were enriched
in both nasal tissue and WBCs at both time points
Yes.
Available
data are
from
experimental
models
presumed
relevant to
humans.
Limited rodent evidence
supports the association
between formaldehyde
exposure and epigenetic
effects in circulating
leukocytes; the available
human evidence is
inadequate. Insufficient
evidence is available to
determine what role
epigenetics may play in
LHP carcinogenesis.
This document is a draft for review purposes only and does not constitute Agency policy.
1-540 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Hypothesized
mechanistic
event
Evidence informing mechanistic event
Human
relevance
Weight-of-evidence
conclusion and
biological plausibility
• 'T* WBC miR-326 expression, associated with bone
marrow metastasis in other models (Valencia et al.,
2013)
Abbreviations: HSPC = hematopoietic stem and progenitor cell; MN = micronuclei; CA = chromosomal aberration;
CFU-GM = colony-forming unit, granulocytes and macrophages; MDS = myelodysplastic syndrome; AML = acute
myeloid leukemia; PBL = peripheral blood lymphocytes; NBUD = nuclear budding; SCE = sister chromatid
exchange; DPX = DNA-protein crosslink; GSH = glutathione; ROS = reactive oxygen species;
MDA= malondialdehyde.
Integrated Summary of Evidence for Lymphohematopoietic Cancers
The strength of the evidence from human studies is robust for myeloid leukemia (see
Lymphohematopoietic cancers in humans above). The assessment of LHP cancers was based on
epidemiology studies of groups with occupational formaldehyde levels either in specific work
settings (e.g., cohort studies) or in case-control studies. Aneuploidy in chromosomes 1, 5, and 7 in
circulating myeloid progenitor cells, considered a potential primary target for LHP carcinogenesis
was associated with occupational formaldehyde exposure. The type of aneuploidies observed in
the formaldehyde-exposed asymptomatic human workers are also found in patients with leukemia,
specifically MDS and AML, as well as other worker cohorts at increased risk of developing
leukemias, which provides support for the plausibility of an association between chronic
formaldehyde exposure and leukemogenesis. Moreover, the strong and consistent evidence from a
large set of studies that observed mutagenicity in circulating leukocytes of formaldehyde-exposed
humans, specifically CAs, and MN formation, provides additional evidence of biological plausibility
for these cancer types. Further support is provided by studies that observed perturbations to
immune cell populations in peripheral blood associated with formaldehyde exposure. In particular,
decreases in RBCs, WBCs, and platelets, along with a 20% decrease in CFU-GM colony formation in
vitro were observed in the same exposed group (Zhang etal.. 2010). suggesting both a decrease in
the circulating numbers of mature RBCs and WBCs as well as possible decreases in the replicative
capacity of myeloblasts.
Increased LHP cancers have not been observed in a well-reported chronic rodent bioassay
involving inhalation exposure of both rats and mice to formaldehyde, nor in another rat bioassay
that failed to report the incidence of non-nasal neoplastic lesions, although there are notable
uncertainties in the available data (i.e., increased bone marrow hyperplasia in rats; slight but
uncertain increases in lymphoma in mice; and a general lack of rigorous evaluation of non-
respiratory tissues). Further, mechanistic changes related to leukemia have not been consistently
reported in well-conducted rodent studies. Thus, there appears to be a lack of support for the
human epidemiological evidence from rodent bioassays, although concordance across species is not
necessarily expected fU.S. EPA. 2005al The apparent lack of consistency in results raises
This document is a draft for review purposes only and does not constitute Agency policy.
1-541 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
uncertainties about the currently available research results on these diseases, including how
formaldehyde exposure-induced LHP cancers might arise without substantial distribution to target
sites. Notably, the available animal evidence was judged as indeterminate and not compelling
evidence of no effect (see assessment Preface), as there are important uncertainties that prevent
such an interpretation. Thus, the animal evidence does not detract from the strength of the
association between formaldehyde exposure and myeloid leukemia (and related mechanistic
changes) in epidemiology studies (NRC. 2014b). Differences in physiology between humans and
rodents, as well as the relative insensitivity of rodent models to reflect the human pathogenesis of
myeloid leukemia, in particular, may together contribute to the potential lack of concordance
between the abundant human epidemiological data and the limited results available from rodent
bioassay data.
Taken together, based on the robust human evidence from studies of groups with
occupational formaldehyde levels, the evidence demonstrates that formaldehyde inhalation
causes myeloid leukemia in humans, given appropriate exposure circumstances. Separately, based
on a limited number of epidemiological studies and potentially relevant mechanistic evidence in
exposed humans, the integration of the evidence results in a judgment that the evidence suggests,
but is not sufficient to infer, that formaldehyde inhalation might cause multiple myeloma and
Hodgkin lymphoma, given appropriate exposure circumstances. While mechanisms for the
induction of myeloid leukemia are yet to be elucidated, they do not appear to require direct
interactions between formaldehyde and bone marrow constituents, and either are different in
animals or the existing animal models tested thus far do not characterize the complex process
leading to cancers in exposed humans.
Table 1-67. Evidence integration summary for effects of formaldehyde
inhalation on LHP cancers
Evidence
Evidence judgment
Hazard determination
Myeloid Leukemia
Human
evidence
Robust for myeloid leukemia based on:
Human health effect studies:
• Consistent increases in risk across a set of high and medium confidence,
independent studies with varied study designs and populations
• Several of these studies demonstrated strong associations (1.5- to 3-fold
increase in risk) and clear exposure-response relationships across multiple
measures of increasing exposure
• The studies possessed a temporal relationship consistent with causality
(e.g., allowing time for induction, latency, mortality)
Biological plausibility (also of potential relevance to LHP cancer types below):
Evidence from high and medium confidence studies of exposed humans
identifies relevant mechanistic changes for cancers of the blood such as myeloid
leukemia, including impacts on peripheral immune cell populations (which seem
to be affected in a complex manner), and elevated levels or severity of DNA or
chromosomal damage in circulating myeloblasts and mature lymphocyte
populations. The DNA damage exhibits aneugenic characteristics similar to that
found in humans with, or at increased risk for, AML.
The evidence demonstrates
that formaldehyde inhalation
causes myeloid leukemia in
humans, given appropriate
exposure circumstances3
This conclusion was primarily
based on epidemiology studies
of groups with occupational
formaldehyde exposure.
While evidence exists to
suggest a lack of concordance
between chronic rodent
bioassays and human
epidemiological evidence,
notable uncertainties prevent
an animal evidence judgment
This document is a draft for review purposes only and does not constitute Agency policy.
1-542 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Animal
evidence
Indeterminate for any LHP cancer type, based on:
Animal health effect studies:
Overall, the available data do not provide evidence supporting the development
of LHP cancers in a high confidence chronic bioassay of rats and mice, a second
medium confidence rat bioassay, and two other low confidence, long-term
exposure studies.
Biological plausibility:
Although some potentially relevant changes have been observed in mechanistic
studies of exposed animals (e.g., inflammatory and immune changes in systemic
tissues and bone marrow hyperplasia in rats), the evidence related to
genotoxicity (i.e., in systemic tissues) or other more directly relevant changes
were weak (e.g., only in low confidence studies) or not observed and, overall,
the mechanistic data do not suggest a judgment other than indeterminate for
LHP cancers in animals.
of compelling evidence of no
effect
Po ten ti al suscep tibili ties:
There is no evidence to
evaluate the potential risk to
sensitive populations or
lifestages
Other
Inferences
• Relevance to humans: The evidence is from studies in humans.
• MOA: No MOA exists to explain how formaldehyde might cause LHP cancers
without systemic distribution (i.e., without direct interactions of inhaled
formaldehyde with constituents in bone marrow tissue); however, given the
mechanistic changes in exposed humans, it is reasonable to infer that an
undefined MOA is likely to involve modulatory effects on circulating immune
cells.
Multiple myeloma
Human
evidence
Sliaht for multiple myeloma, based on:
Human health effect studies:
• Increases in risk associated with peak exposure metrics across one high, one
medium, and two low confidence studies; no associations with other
exposure metrics
• Increases spanned an approximate 1.2- to 4-fold increase in risk, with the
highest confidence evidence showing a 2-fold increase
• Very limited evidence of an exposure-response relationship in one high
confidence study
• However, risks may have been driven by peak exposures as increases were
limited to groups of people who experienced high peak exposures, and two
low confidence studies reported inverse relationships with duration of
exposure
The evidence suggests, but is
not sufficient to infer, that
formaldehyde inhalation
might cause multiple myeloma
given appropriate exposure
circumstances'5
Animal
evidence
Indeterminate (for any LHP cancer type): see explanation for myeloid leukemia
Other
Inferences
• Relevance to humans: The evidence is from studies in humans.
• MOA: No MOA exists to explain how formaldehyde might cause LHP
cancers without systemic distribution
Hodgkin lymphoma
Human
evidence
Sliaht for Hodgkin lymphoma, based on:
Human health effect studies:
• Significantly increased risk in the highest peak exposure group alongside an
exposure-response relationship in one medium confidence study of
industrial workers
• An inconsistent pattern of risks across 1 medium and the low confidence
studies, many with <5 exposed cases
• The high survival rate for Hodgkin lymphoma may indicate that mortality
data are not a good proxy for incidence.
The evidence suggests, but is
not sufficient to infer, that
formaldehyde inhalation
might cause Hodgkin
lymphoma given appropriate
exposure circumstances'5
Animal
Indeterminate (for any LHP cancer type): see explanation for myeloid leukemia
This document is a draft for review purposes only and does not constitute Agency policy.
1-543 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Toxicological Review of Formaldehyde—Inhalation
evidence
Other
inferences
• Relevance to humans: The evidence is from studies in humans.
• MOA: No MOA exists to explain how formaldehyde might cause LHP
cancers without systemic distribution
Lymphatic leukemia
Human
evidence
Indeterminate for Ivmohatic leukemia, based on:
Human health effect studies:
• A consistent pattern of null results across eight high, medium, and low
confidence studies
• The high survival rate for lymphatic leukemia may indicate that mortality
data are not a good proxy for incidence.
There is inadequate evidence
to determine whether
formaldehyde inhalation may
be capable of causing
lymphatic leukemia in humans
Animal
evidence
Indeterminate (for any LHP cancer type): see explanation for myeloid leukemia
Other
inferences
N/A: no signal exists across lines of evidence
aThe "appropriate exposure circumstances" are more fully evaluated and defined through dose-response analysis in Section 2.
bGiven the uncertainty in this judgment and the available evidence, this assessment does not attempt to define what might be
the "appropriate exposure circumstances" for developing this outcome.
1.4. SUMMARY AND EVALUATION
This section provides summaries of the available evidence on susceptible populations and
life stages and on populations that may have heightened formaldehyde exposures compared to the
general population (Section 1.4.1), the weight of evidence for effects other than cancer
(Section 1.4.2), and the weight of evidence for carcinogenicity (Section 1.4.3).
1.4.1. Susceptible Populations and Lifestages
Susceptible populations and lifestages refers to groups of people who may be at increased
risk for adverse health consequences following chemical exposures due to factors such as age,
genetics, health status and disease, sex, lifestyle, and other coexposures. This discussion of
susceptibility focuses on factors for which there are available formaldehyde exposure-specific data
and on factors hypothesized to be important to formaldehyde. Vulnerable populations, defined as
groups that may be at increased risk for adverse health consequences due to heightened
formaldehyde exposures, are also discussed.
Lifestage
Embryos, fetuses, infants, children, and the elderly may have differing levels of maturity and
functioning of cellular and organ systems, and metabolizing enzymes, as well as unique activity
patterns that may influence the toxicodynamics of chemicals in the body. Embryonic, fetal,
neonatal, and juvenile periods, as well as reproductive lifestages in both men and women, are often
periods of increased susceptibility to negative health consequences following chemical exposures.
This document is a draft for review purposes only and does not constitute Agency policy.
1-544 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Developmental and reproductive effects
The Developmental and Reproductive Toxicity (Section 1.3.2) provides a detailed analysis
of human and animal studies evaluating susceptibility to formaldehyde toxicity while in utero and
during infancy, childhood, and reproductive lifestages. Overall, it was judged that the available
evidence indicates that formaldehyde inhalation exposure likely causes developmental or
reproductive toxicity in humans. This hazard conclusion was primarily based on moderate
evidence from epidemiological studies of women that reported decreased fecundity and increased
spontaneous abortion risk at occupational exposure levels as high as 1.2 mg/m3 (Taskinenetal..
1999: Tohn etal.. 19941 as well as effects on fetal growth among three pregnancy cohorts observed
at indoor formaldehyde concentrations >0.04 mg/m3, and possibly lower fFranklin et al.. 2019:
Amiri and Turner-Henson. 2017: Chang etal.. 20171.
Further research is needed to determine if the increased spontaneous abortion risk and
decreased fecundity in occupationally exposed women is due to toxicity to the reproductive system
or to the developing fetus. Additionally, there is a need for more targeted evaluation of the female
reproductive system following inhalation exposure to formaldehyde, including an assessment of
female reproductive function, such as would be assessed in a two-generation reproductive study in
animals. Further assessment of both female reproductive toxicity and developmental toxicity
would benefit from the use of paraformaldehyde instead of formalin to avoid possible confounding
exposures to methanol.
Several animal studies raise the possibility that formaldehyde exposure might also cause
toxicity to the developing nervous system; however, due to methodological limitations, these data
were considered inconclusive (i.e., evidence suggests). Three publications from one laboratory
fSarsilmaz etal.. 2007: Asian etal.. 2006: Songur etal.. 20031 reported changes in brain structure
and neuron numbers following developmental exposure to formaldehyde. However, two of these
studies were evaluations of the same animals, and all three studies possessed notable
methodological limitations and tested formaldehyde levels >7 mg/m3, which introduces
uncertainties (e.g., differences in toxicokinetics; irritant effects not experienced by humans) in
relating these data to the potential for effects in exposed humans. The changes in brain structure
and neuron number were not tested using similarly sensitive protocols in adult animals, although
less rigorous evaluations failed to observe effects, highlighting additional data gaps. Only low
confidence studies evaluated other potential neurodevelopmental effects (i.e., the evidence is
inadequate).
Children
Lungs in children are underdeveloped at birth and are not fully functional until about 6 to
8 years of age fBateson and Schwartz. 20081: therefore, children may be more susceptible to the
respiratory effects of formaldehyde, compared to adults. In addition, formaldehyde exposure has
been associated with airway inflammation (see Section 1.2.3), which could have a greater impact on
This document is a draft for review purposes only and does not constitute Agency policy.
1-545 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
children's airways because they are narrower than adult airways (OEHHA. 20031. This is
supported by studies of other chemicals suggesting that human sensitivity to sensory irritation may
also be dependent on age fShusterman. 2007: Hummel et al.. 20031. The distribution of inhaled
formaldehyde may be different for children compared with adults as well. For example, population
variability in distribution is influenced by differences in physical characteristics of the URT,
breathing patterns (e.g., oral versus nasal), and ventilation rate. However, studies suggest that
extrathoracic absorption of highly reactive and soluble gases, such as formaldehyde, is similar
between children and adults (Ginsberg etal.. 2010: Ginsberg etal.. 20051. as is overall uptake
efficiency, average flux, and maximum flux levels over the entire nasal lining fGarcia etal.. 20091.
Garcia et al. (2009) did find that local flux between the seven individuals (five adults and two
children) in his study varied by a factor of three to five, which is important as formaldehyde toxicity
is likely to be mediated by its point-of-contact effects along the URT. Because this study only
evaluated seven individuals who had normally shaped nasal cavities, it may not be generalizable to
the entire population, including susceptible individuals. Notably, formaldehyde distribution to
more distal parts of the airways could be substantial under conditions of higher activity and oral
breathing, both of which occur with children.30
The expression of formaldehyde metabolizing enzymes may also be different in infants and
children. The metabolism of formaldehyde is described in more detail in Appendix A2. Briefly,
expression of glutathione-dependent formaldehyde dehydrogenase, also called alcohol
dehydrogenase class III, ADH3, or ADH5, the primary enzyme in formaldehyde metabolism, is
developmentally regulated and thus may alter the toxicokinetics of formaldehyde in early life
(Reviewed in (Thompson etal.. 2009: Hines and McCarver. 200211. ADH3 is critical to the
detoxification of formaldehyde, as it is involved in the pathway leading to formaldehyde's
conversion to formate, a metabolite that is excreted from the body. Therefore, if the concentration
or activity of ADH3 is reduced, more formaldehyde is likely to remain in the body to react with
cellular macromolecules. ADH3 mRNA expression levels are significantly lower in premature
neonates and infants up to 5 months old compared with adults. Benedetti et al. (2007) reported
that ADH activity reached adultlevels by 2.5 to 5 years of age. Thus, neonates and very young
children, in particular, may have a decreased ability to metabolize formaldehyde, increasing their
susceptibility to formaldehyde toxicity; however, enzyme activity levels for ADH3 specifically, and
the potential for alternate metabolic pathways in children, are not known.
Some epidemiological studies have found that children have an increased sensitivity to
formaldehyde exposure-induced respiratory effects. One study reported a relationship between
increased residential formaldehyde exposure and decreased PEFR (both bedtime and morning)
among children exposed to levels averaging 0.032 mg/m3 fKrzvzanowski etal.. 19901. In adults, an
30For example, in the case of ozone concentrations of 0.1 ppm, a moderately reactive gas, Ginsberg (20081
predicted a five-fold variation in the dose to the deep lung between quiet and heavy breathing conditions for
an 8-year-old child.
This document is a draft for review purposes only and does not constitute Agency policy.
1-546 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
association of smaller magnitude was observed, but only among smokers. Krzyzanowski et al.
(1990) also reported an increase in the prevalence of physician-diagnosed asthma in children, but
not in adults, who lived in homes with formaldehyde levels that were higher than 60 ppb
(0.074 mg/m3). Similarly, a study by Zhai et al. (2013.) reported a higher prevalence of current
asthma in children compared with adults at the same exposure levels in their homes. Although
prevalence of current asthma (i.e., symptoms or use of medications in the past 12 months) does not
appear to be increased among adults or children below exposure levels of approximately
0.05 mg/m3, studies of the exacerbation of asthma symptoms (asthma control) among children
suggest their greater susceptibility at lower average formaldehyde concentrations (e.g., 0.04
mg/m3; fDannemiller etal.. 2013: Venn etal.. 20031. Children younger than five years of age also
may experience symptoms consistent with lower respiratory infections in association with
residential formaldehyde levels lower than those at which older individuals experience these
symptoms (Roda etal.. 2011: Rumchev et al.. 2002).
Children are also likely to be more susceptible than adults to the mutagenic effects of
formaldehyde. EPA has concluded that early-life exposure to chemicals that are carcinogenic
through a mutagenic MOA might present a higher risk of cancer than exposure during adulthood
fU.S. EPA. 2005bl. Because formaldehyde-induced carcinogenicity of the URT is attributable, at
least in part, to a mutagenic MOA (see Section 1.2.5), it is expected that children are at heightened
risk of URT cancers following formaldehyde exposure. In contrast, because it is unknown whether
myeloid leukemia resulting from formaldehyde exposure involves a mutagenic MOA, no assumption
about increased early-life susceptibility is made for this type of cancer.
Pregnant women
Because pregnant women have increased sensitivity to the development and exacerbation
of atopic eczema fKar etal.. 2012: Cho etal.. 2010: Weatherhead et al.. 20071. it is likely that they
also have heightened susceptibility to this form of dermatitis following exposure to formaldehyde.
To date, however, no studies are available that specifically evaluate the prevalence of atopic eczema
in pregnant women compared to other populations following exposure to formaldehyde. In one
study, Matsunaga et al. (20081 found a two-fold higher risk for atopic eczema in pregnant women
with formaldehyde exposures of approximately 0.06 mg/m3 measured in their homes.
Later lifestages
In general, older adults may have greater susceptibility than younger adults to chemical
exposures due to slower metabolisms and an increased incidence of altered health status
(Benedetti etal.. 2007: Ginsberg et al.. 20051. One study (Bentaveb etal.. 20151 indicated possible
differential effects of formaldehyde exposure for elderly adults (>65 years old) compared with
other age groups. Bentayeb et al. (20151 observed an elevated risk of decreased pulmonary
function in nursing home residents at lower formaldehyde exposure levels than have been seen to
cause effects in younger adults.
This document is a draft for review purposes only and does not constitute Agency policy.
1-547 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Health Status and Disease
Preexisting health conditions and diseases may predispose individuals to toxic effects
following exposure to formaldehyde. Some epidemiological studies have suggested that asthmatics
are more susceptible than nonasthmatics to declines in respiratory function following
formaldehyde exposure. Krzyzanowski et al. (1990) found that asthmatic children showed a
steeper decline in morning peak expiratory flow rate (PEFR) compared with nonasthmatic children
at formaldehyde concentrations below 0.05 mg/m3. Similarly, a study by Kriebel et al. (1993)
reported a greater decrease in peak expiratory flow (PEF) in asthmatic, compared with
nonasthmatic, medical students after formaldehyde exposures in an anatomy lab. However, this
study fKriebel et al.. 19931 had a small sample size and the effect was not statistically significant
Studies evaluating effect modification by existing allergies are inconsistent. Acute and
short-term studies in two animal species demonstrate that formaldehyde increases responsiveness
to allergens and bronchoconstrictors, particularly with prior sensitization, indicating that allergy
status may modify an individual's sensitivity to bronchial hyperreactivity and other asthma
symptoms due to formaldehyde exposure (Larsen etal.. 2013: Riedel etal.. 1996: Swiecichowski et
al.. 1993: Leikauf. 19921. However, studies of associations with eczema, prevalence of asthma or
asthma control were inconsistent, reporting either an increased or decreased prevalence among
groups with a positive atopy status in adults or children fAnnesi-Maesano etal.. 2012: Matsunaga et
al.. 2008: Venn etal.. 2003: Smedje and Norback. 2001). The evidence, therefore, is inconclusive
and additional research is needed to address the question of potential effect modification by atopy
status. Separately, the swelling of the mucus membrane, which has been observed in humans
exposed to <1 mg/m3 formaldehyde (see Section 1.2.4), is expected to be highly influenced by the
underlying respiratory status of the exposed individuals, such as allergy status or previous or
current respiratory infections. Supporting this assumption, nasal lesions have been found to be
more severe in formaldehyde-exposed rodents with prior nasal damage (Woutersen et al.. 1989:
Appelman etal.. 1988). and similar observations have been made in exposed humans (Falk etal..
1994). Therefore, individuals with prior nasal damage might also have heightened subsceptibililty
to the development of nasal cancer following formaldehyde exposure.
As discussed in Section 1.1.3, nasal anatomy and soluble factors in the URT play a major role
in the uptake of a highly reactive gas like formaldehyde. There are considerable interindividual
variations in nasal anatomy fICRP. 19941. For example, the nasal volumes of 10 adult nonsmoking
subjects between 18 and 50 years of age in a study in the United States varied between 15 and 60
mL (Santiago etal.. 2001). and disease states can result in further variation (Singh etal.. 1998).
Therefore, population variability in the distribution of inhaled formaldehyde, and in the
susceptibility to its health effects, could potentially be large.
To date, many other factors related to health, such as obesity, have not been investigated to
determine if they affect susceptibility to formaldehyde-related adverse effects.
This document is a draft for review purposes only and does not constitute Agency policy.
1-548 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Sex
Males and females can differ greatly in body composition, organ function, and many other
physiological parameters that may influence the toxicokinetics of chemicals and their metabolites
in the body fGochfeld. 2007: Gandhi et al.. 20041. The human epidemiology data set does not
support many specific sex susceptibilities for noncancer effects due to formaldehyde
exposure. However, in general, data suggest that nonpregnant women, on a per kilogram body
weight basis, may have slightly lower air intake than men, which would suggest that women may be
less susceptible than men to inhaled pollutants like formaldehyde, but this has not been
investigated to date.
Similar to age and allergy and respiratory infection status, studies of related chemicals
suggest that human sensitivity to sensory irritation may also be dependent on sex fShusterman.
2007: Hummel etal.. 2003). It is likely that women may be more sensitive than men to the eye and
URT irritant properties of formaldehyde. For example, a higher prevalence of burning or tearing
eyes was observed among women compared to men in a study of residential exposure (Liu etal..
19911.
In contrast, several animal studies suggest that males may be more susceptible than females
to histopathological lesions of the URT, although most studies only examined male animals. For
example, one study in rats reported that males generally had more severe damage, including
metaplasia, to the nasal respiratory and olfactory epithelium and larynx following formaldehyde
exposure (Woutersen etal.. 19871. Supportive findings of increased incidence or severity of lesions
in males as compared to females was also reported in a second study of rats fZwartetal.. 19881.
and in mouse studies of fMaronpotetal.. 1986: Kerns etal.. 19831. Male rats have a higher
metabolic rate and oxygen demand than do female rats; therefore, these findings might reflect a
greater inhaled dose of formaldehyde in males compared to females at similar exposure
concentrations.
It is also concluded that the evidence indicates formaldehyde exposure likely causes sex-
specific health effects related to reproduction, given the relevant exposure circumstances.
Specifically, a coherent spectrum of male reproductive effects was observed in experimental animal
studies following exposure to high levels of formaldehyde, with supporting evidence in a well-
conducted human study. In addition, epidemiological studies identified decreased fecundity and
increased spontaneous abortion risk in women occupationally exposed to formaldehyde. This
evidence could reflect developmental effects, or changes in the female reproductive system.
Race
Race may be a modifying factor of formaldehyde toxicity, for example, if specific
polymorphisms in metabolizing enzymes affecting chemical toxicokinetics are more prevalent in
specific races. Additionally, lifestyle factors that modify toxicity may be more or less prevalent in
specific races. The only study to evaluate the potential role of race in carcinogenicity fHaves etal..
This document is a draft for review purposes only and does not constitute Agency policy.
1-549 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
19901 found significantly increased death rates from nasopharyngeal cancer and multiple myeloma
in nonwhite embalmers and funeral directors; whereas no changes in death rates from
nasopharyngeal cancer or in cases of multiple myeloma were found in white embalmers and
funeral directors. Very few other studies have explored the role of race in formaldehyde
susceptibility, preventing the interpretation and generalizability of this observation.
A more detailed description of the role of polymorphisms in susceptibility is provided
below. Additional research is needed to confirm the findings in Hayes et al. (19901.
Genetic Polymorphisms
Genetic polymorphisms may affect the expression level of genes and resulting activity of
important metabolizing enzymes, and this may lead to differential toxicity following chemical
exposures. As discussed in Appendix A.2, the primary metabolizing enzyme of formaldehyde is
ADH3 (referred to as ADH5 in recent papers). A secondary pathway involves mitochondrial
aldehyde dehydrogenase 2 (ALDH2). Both ADH3 and ALDH2 are important in the detoxification of
formaldehyde, converting it to formate, which is readily excreted from the body. ADH3 is also
known to catalyze the NADP-dependent reduction of the endogenous nitrosylating agent S-
nitrosoglutathione (GSNO) and, in this capacity, is referred to as S-nitrosoglutathione reductase
(GSNOR) flensen etal.. 19981. GSNOR participates in the oxidation of retinol and long-chain
primary alcohols. It also contributes to nitric oxide (NO) signaling through its role in metabolizing
GSNO an endogenous bronchodilator and reservoir of NO (Staab etal.. 2008: Hess etal.. 2005:
Tensen etal.. 19981. indicating ADH3's involvement in bronchial tone allergen-induced
hyperresponsiveness (Gerard. 2005: Hess etal.. 2005: Que etal.. 20051.
Wu et al. (2007) found that carrying one or two copies of the minor allele rsl 154404 for a
single nucleotide polymorphism (SNP) of ADH3 resulted in a decreased risk of asthma in Mexican
children. For a different SNP (rs28730619), homozygotes for the minor allele had an increased risk
of asthma. Although only speculative as their functional characteristics are unknown, these SNPs
may affect the response of individuals to formaldehyde exposure by altering their metabolism. One
study (Hedberg et al.. 20011 identified four polymorphisms in the human ADH3 gene promoter that
resulted in reduced transcriptional activity. Because this would likely result in reduced levels of the
ADH3 protein, individuals with this polymorphism may be at greater risk for formaldehyde toxicity
compared with people with the wild-type gene. This is supported by a study in which deletion of
the ADH3 gene increased the sensitivity of mice to formaldehyde toxicity fDeltour et al.. 19991.
Some studies have also suggested that CNS toxicity can result from reduced activity of the
metabolizing enzymes responsible for clearing formaldehyde from relevant tissues
(e.g., downregulated ALDH2 in Tan et al. (201811. Therefore, it is plausible that individuals with
polymorphisms in ALDH2 or in other genes encoding detoxifying enzymes may be more susceptible
to CNS toxicity caused by formaldehyde exposure compared to those with wild type alleles. This
highlights another area of interest for future studies on potential susceptibility to inhaled
formaldehyde exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
1-550 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
A few studies of genotoxicity among formaldehyde-exposed groups evaluated differences in
response based on polymorphisms in genes coding for proteins involved in the metabolism of
xenobiotics, including CYP2E1, glutathione-S-transferases (GSTs), and ADH3. The X-ray repair
cross-complementing gene 3 (XRCC3), which codes for a protein involved in DNA repair and
chromosome stabilization, also was evaluated fCosta etal.. 2015: Ladeira etal.. 2013: Santovito et
al.. 2011: liang etal.. 2010). The results of these studies were inconsistent and no conclusions
regarding the impact of these genetic polymorphisms on susceptibility can be drawn, (e.g., Shen et
al.. 2016: Rager etal.. 20141
Studies of mice with knocked out AIdh2 and AIdh5, which encode for enzymes that remove
endogenous formaldehyde, have suggested that polymorphisms in Aldh2 and AIdh5, may increase
susceptibility to genotoxicity. These knockouts resulted in severely disrupted hematopoiesis and
leukemia, including mutated and abnormal HSPCs, which is presumably linked to increased
accumulation of endogenous formaldehyde (Dingier etal.. 2020: Burgos-Barragan etal.. 2017b:
Pontel etal.. 20151. Likewise, direct treatment of AIdh5-/- bone marrow cells with formaldehdye
caused genotoxicity and reduced HSPC formation, effects which are further exacerbated by loss of
Fancd2 (this latter deficiency is associated with increased sensitivity to DNA damage) fGarcia-
Calderon etal.. 2018: Burgos-Barragan etal.. 2017bl. As reviewed and tested by Dingier et al.
(2020), genetic deficiencies in these Aldh family genes have been linked to bone marrow failure and
related diseases in humans, including in children. Reduced ALDH2 or ALDH5 activity resulting in
increased endogenous formaldeheyde in mice and humans might also contribute to postnatal
lethality, stunted growth, cognitive effects (see Section 1.3.1) and various cancers arising from DNA
damage or deficient repair (Dingier etal.. 2020: Nakamura etal.. 2020). While formaldehyde
inhalation does not seem to cause appreciable changes in formaldehyde levels in nonrespiratory
regions (see Appendix A.2), HSPCs expressing these enzymes are known to exist in many tissues.
However, no studies in any species have specifically examined these possible linkages in relation to
inhaled formaldehyde. Therefore, while genetic differences may alter susceptibility to the
cytogenetic effects of formaldehyde, more definitive research is needed. A few in vitro studies have
suggested that epigenetic changes or loss of function of important genes might increase
susceptibility to formaldehyde toxicity (e.g., Shen etal.. 2016: Rager etal.. 20141. However,
additional studies are needed to clarify these preliminary observations.
Lifestyle Factors
Lifestyle factors may increase or decrease exposure to formaldehyde and may also affect the
resulting health effects following formaldehyde exposure. These lifestyle factors may vary by race,
ethnicity, socio-economic status, or geographic location. To date, specific studies do not exist to
address the role of lifestyle factors on formaldehyde toxicity.
This document is a draft for review purposes only and does not constitute Agency policy.
1-551 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Nutritional status
Because formaldehyde appears to cause inflammation, particularly in the airways, it is
plausible that a diet rich in antioxidants would protect against inflammation and one that lacks
sufficient antioxidants would result in greater inflammation. Additional research is needed to
specifically evaluate possible modification of formaldehyde toxicity by nutritional status.
Smoking
Smoking is considered a lifestyle factor, but it also introduces coexposures to the many
chemicals in cigarette smoke, including additional formaldehyde. Thus, it is difficult to disentangle
potential indirect contributions of smoking to the health effects of formaldehyde exposure from the
possible direct effects of the formaldehyde in tobacco smoke (see additional discussion below
under "coexposures").
Exercise
The possibility that more extensive distribution of formaldehyde (e.g., to the LRT) may
occur when people are breathing through the mouth during exercise has not been investigated.
However, some controlled human exposure studies observed pulmonary function deficits when a
longer exercise component (15 minutes) was included that were not observed by other studies
with shorter periods or no exercise (Green etal.. 1989: Green etal.. 1987). and another study
observed an increase in bronchial hyperresponsiveness with an exposure protocol using nose clips
necessitating mouth-only breathing (Cassetetal.. 2006). Clearly, further research is warranted to
understand the role of exercise in formaldehyde susceptibility.
Coexposures
Coexposures to other pollutants, such as those that produce similar metabolites and health
effects to formaldehyde and those that are mutagens, may exacerbate the effects of formaldehyde
exposure. In addition, constituents in the diet, such as methanol and caffeine, contribute to the
generation of endogenous formaldehyde in nonrespiratory tissues (Summers etal.. 2012: Riess et
al.. 2010: Hohnloser etal.. 1980). which are promptly detoxified (Burgos-Barragan etal.. 2017a).
Yet, it is not expected that variation in endogenous formaldehyde levels at sites distal to the URT
would affect relative sensitivity to the effects of inhaled formaldehyde. These findings are
inconclusive, however, so additional research is needed to investigate the role of these coexposures.
As described in Section 1.2.3, tobacco smoke may increase the incidence of hypersensitivity
responses in formaldehyde-exposed individuals. Effect modification by environmental tobacco
smoke (i.e., stronger associations, or associations seen at lower formaldehyde exposures, with this
coexposure) were reported in two studies that examined asthma prevalence stratified by
environmental tobacco smoke exposure among children and adults (nonsmokers) fPalczvnski et al..
1999: Krzvzanowski etal.. 19901. Additional studies are needed to establish if this interaction is
seen only in children, in adults and children, or in neither group. One residential study by
This document is a draft for review purposes only and does not constitute Agency policy.
1-552 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Krzyzanowski et al. (19901 indicated that smokers experienced a greater decline in morning PEFR
compared to nonsmokers at formaldehyde concentrations above 0.050 mg/m3. Smokers were not
more responsive to formaldehyde exposures in most occupational studies that stratified by
smoking behavior. Nonsmokers experienced 2- to 3.5-fold larger annual decreases in FEVi,
FEVi/FVC, and FEF25-75 over 5 years fAlexandersson and Hedenstierna. 19891. as well as larger
declines during a work shift (Alexandersson and Hedenstierna. 1989: Alexandersson etal.. 1982).
In contrast, current smokers had an approximately two-fold larger OR for airway obstruction,
defined as an FEVi/FVC <75%, compared with nonsmokers (Herbert etal.. 19941. The magnitude
of the difference associated with formaldehyde exposure may have reflected the existing difference
in baseline pulmonary function values between smokers and nonsmokers.
Although not a chemical coexposure, humidity also appears to modify the effects of
formaldehyde exposure. For example, formaldehyde exposure-induced bronchoconstriction in
mice housed only in humid, but not dry, environments indicating that the bronchoconstrictive
effects of formaldehyde may be impacted by humidity (Larsen et al.. 2 0131. The effects of
formaldehyde on mucus flow patterns also appear to vary based on humidity.
In addition, it is possible that exposure to nochemical stressors, such as poverty, violence,
and other social factors, might make some populations more susceptible to formaldehyde-related
health effects. However, at this time, studies evaluating the contribution of nonchemical stressors to
formaldehyde susceptibility have not been published.
Additional research is needed to investigate whether coexposures to pollutants other than
tobacco smoke and to nonchemical stressors confer additional susceptibility to formaldehyde
toxicity.
Summary of Susceptible Populations and Lifestages
Epidemiological and toxicological studies, as a whole, identify reproductive or
developmental toxicity as a human health hazard of formaldehyde exposure. At this time, it is not
clear whether increased time-to-pregnancy (TTP) and spontaneous abortion rates seen in
occupationally exposed women are due to reproductive system toxicity or to toxicity to the
developing fetus.
Children also appear to be a susceptible population. Studies have indicated that they have
an increased sensitivity to respiratory and immunological effects following formaldehyde exposure.
In addition, younger age is likely to be associated with a higher risk of mutagenic effects and,
therefore, to a higher risk of URT cancers. As age may be a modifying factor of the sensory irritant
properties of formaldehyde, both children and the elderly may be at an either increased or
decreased risk for sensory irritation.
Health status and disease are likely to be modifying factors of formaldehyde toxicity as well.
Studies suggest that asthmatics are more susceptible than nonasthmatics to declines in respiratory
function following formaldehyde exposure. Whether atopy and allergies can also influence the
health effects of formaldehyde exposure remains to be determined; additional studies are needed to
This document is a draft for review purposes only and does not constitute Agency policy.
1-553 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
confirm this relationship. Individuals with prior nasal damage might also have heightened
subsceptibililty to the development of nasal cancer following formaldehyde exposure.
Study findings on the role of genetic susceptibility in formaldehyde toxicity are
inconclusive. Therefore, gene-environment interaction studies are needed to investigate the effects
of polymorphisms in genes that encode formaldehyde metabolizing enzymes, as well as receptors
(e.g., TRPA1) or other proteins that appear to be key components of the MOA for certain human
health effects of formaldehyde exposure.
Coexposures appear to increase susceptibility to health effects following formaldehyde
exposure as well. There is some evidence that cigarette smoking increases sensitivity to
formaldehyde toxicity; however, it is not clear if this increased sensitivity is due to the additional
formaldehyde to which smokers are exposed, to exposures to other chemicals that are present in
cigarette smoke, or to compromised respiratory systems.
Although other factors are hypothesized to confer increased susceptibility to formaldehyde
toxicity, the available data are limited. Overall, the most extensive research on the health effects of
inhaled formaldehyde and susceptible groups indicates a greater susceptibility among children to
respiratory disease, manifested as reduced pulmonary function, increased prevalence of current
asthma, and greater asthma severity (reduced asthma control). More research is needed to
investigate the role of sex, race, nutrition, exercise, and other coexposures that may modulate
susceptibility to formaldehyde toxicity. In addition, these susceptibility factors might interact with
one another. For example, lifestage, pre-existing health conditions, genetic polymoprhisms and co-
exposures to both chemical and nonchemical stressors could all contribute to heightened
susceptibility to formaldehyde toxicity for some individuals.
Summary of Vulnerable Population
Groups that may receive disproportionally high levels of exposure to formaldehyde, and
therefore might experience more frequent or severe formaldehyde-related health consequences,
include people in occupations with workplace exposures. Some industries with the greatest
potential for exposure include health services, business services, printing and publishing, chemical
manufacturing, garment production, beauty salons, and furniture manufacturing (IARC. 19951.
People who spend a significant amount of time in mobile homes and trailers, either as primary
residences, classrooms, job sites or for other reasons, might also be vulnerable because these
structures can have high formaldehyde levels fMurphv et al.. 2 0131. Lastly, in addition to the
potential of cigarette smoking to increase susceptibility to formaldeyde, it also can increase
exposure to it (Fishbein. 1992). It should be noted that individuals who are both susceptible and
highly exposed to formaldeyde are at the highest risk of suffering from formaldehyde-related health
effects.
This document is a draft for review purposes only and does not constitute Agency policy.
1-554 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Toxicological Review of Formaldehyde—Inhalation
1.4.2. Summary of Evidence Integration Conclusions for Effects Other Than Cancer
Overall, the evidence demonstrates that inhalation of formaldehyde causes sensory irritation and
respiratory pathology in humans, given appropriate exposure circumstances, based on studies of
the general population with residential exposure, controlled human exposure studies, and
occupational studies. The evidence indicates that inhalation of formaldehyde likely causes
decrements in pulmonary function, and an increased frequency of current asthma symptoms and
allergic responses, given appropriate exposure circumstances, based on studies of adults and
children exposed in their homes or at school. In addition, the evidence indicates that inhalation of
formaldehyde likely causes female reproductive or developmental toxicity, and reproductive
toxicity in males, given appropriate exposure circumstances, based on studies involving residential
and occupational exposure and toxicological studies. Lastly, while a number of studies reporting
evidence of potential neurotoxic effects were available, including developmental neurotoxicity,
multiple manifestations of behavioral toxicity, and an increased incidence of, or mortality from, the
motor neuron disease, amyotrophic lateral sclerosis (ALS), due to limitations identified in the
database (e.g., poor methodology; lack of consistency), the evidence integration analyses for these
outcomes determined that the evidence suggests, but is not sufficient to infer, a human health
hazard(s). The data on potential nervous system effects were considered insufficient for developing
quantitative estimates of risk. Context on these decisions is provided below:
• Sensory Irritation:
o The evidence demonstrates that inhalation of formaldehyde causes sensory
irritation in humans, given appropriate exposure circumstances, based on robust
human evidence from controlled human exposure studies testing responses to
concentrations 0.1 mg/m3 and above and observational epidemiology studies of
residential populations with mean formaldehyde concentrations >0.05 mg/m3
(range of 0.01 to approximately 1.0 mg/m3), robust evidence for an effect in animals
(this phenomenon is well described and accepted across a range of experimental
species), as well as an established MOA based on mechanistic evidence in animals
(the identified MOA is interpreted to be operant in humans). The irritant response
occurs within minutes to hours depending on concentration, and severity is
concentration dependent Potentially large variations in sensitivity are expected,
depending primarily on differences in nasal health (including allergy or
inflammatory status) and physiology.
• Pulmonary Function:
o The evidence indicates that long-term (chronic) inhalation of formaldehyde likely
causes decrements in pulmonary function, given appropriate exposure
circumstances, based on moderate human evidence primarily from observational
epidemiology studies among occupational cohorts with long-term exposure to
>0.2 mg/m3 and a study of children and adults with residential exposure (mean,
0.03 mg/m3, maximum 0.17 mg/m3), as well as slight evidence for an effect in
animals involving inflammatory airway changes in mechanistic studies (it is
This document is a draft for review purposes only and does not constitute Agency policy.
1-555 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Toxicological Review of Formaldehyde—Inhalation
expected that related mechanistic changes can occur in exposed humans, and some
indirect confirmatory evidence from exposed humans exists). The evidence is
inadequate to interpret whether acute or intermediate-term (hour-weeks)
formaldehyde exposure might cause this effect. Variation in sensitivity is
anticipated to depend on age and respiratory health.
• Respiratory T ract Pathology:
o The evidence demonstrates that inhalation of formaldehyde causes increased
respiratory tract pathology in humans, including hyperplasia and squamous
metaplasia, given appropriate exposure circumstances, based on robust evidence
from animal studies involving multiple species with increases in severity and
frequency of lesions with increasing concentration or longer exposure duration.
The primary support for this conclusion is based on ratbioassays of chronic
exposure which consistently observed squamous metaplasia at formaldehyde
exposure levels >2.5 mg/m3. There is moderate human evidence from occupational
epidemiology studies supported by more limited findings in mechanistic studies of
exposed humans, and strong support for a plausible MOA based largely on
mechanistic evidence in animals (supported by coherent findings in human studies).
Variation in sensitivity may depend on differences in URT immunity and nasal
structure or past injury, but few studies exist that specifically evaluate these
possibilities.
• Immune-mediated Conditions, including Allergies and Asthma:
o The evidence indicates that inhalation of formaldehyde likely causes increases in
the prevalence of allergic conditions in humans, given appropriate exposure
circumstances, based on moderate evidence of an enhanced immune
hypersensitivity response to allergens (i.e., allergic rhinitis or rhinoconjunctivitis;
eczema) in general population studies of adults and children at average exposures
between 0.03 and <0.1 mg/m3 formaldehyde, and slight evidence of effects relevant
to immune-mediated respiratory conditions in animals from mechanistic studies of
airway hyperresponsiveness and some more limited data relevant to systemic
inflammatory changes in both human and animal mechanistic studies; however, the
proposed, incomplete MOA(s) are not established and have not been experimentally
verified.
o The evidence indicates that inhalation of formaldehyde also likely causes increases
in the prevalence of asthma symptoms in humans, given appropriate exposure
circumstances, based on moderate evidence of an increased risk of prevalent current
asthma in occupational settings (>0.1 mg/m3) and population studies in adults and
children, or poor asthma control in children at exposures above 0.05 mg/m3
formaldehyde and slight evidence for effects in animals from mechanistic studies;
however, an MOA explaining this association is not available. Specifically, regarding
the animal evidence, although several events typically associated with asthma are
not well supported by the available data, the animal mechanistic data support that
formaldehyde inhalation induces bronchoconstriction with and without allergen
sensitization and stimulates a number of immunological and neurological processes
that would be expected to augment or drive asthmatic responses. Variation in
This document is a draft for review purposes only and does not constitute Agency policy.
1-556 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Toxicological Review of Formaldehyde—Inhalation
sensitivity is anticipated depending on respiratory health, age, and exposure to
tobacco smoke.
• Developmental and Reproductive Toxicity:
o The evidence indicates that inhalation of formaldehyde likely causes
developmental or female reproductive toxicity in humans, based on moderate
evidence in observational studies finding effects on fetal growth among pregnancy
cohorts observed at indoor formaldehyde concentrations >0.04 mg/m3, and
possibly lower, as well as increases in TTP and spontaneous abortion risk among
occupationally exposed women (average formaldehyde concentrations
>0.1 mg/m3); the evidence in animals is indeterminate, and a plausible,
experimentally verified MOA explaining such effects without systemic distribution
of formaldehyde is lacking.
o The evidence indicates that inhalation of formaldehyde also likely causes
reproductive toxicity in men, given appropriate exposure circumstances, based on
robust evidence in animals that presents a coherent array of adverse effects in two
species, and slight evidence from observational studies of occupational exposure.
Uncertainties include a lack of well-conducted animal studies testing formaldehyde
exposure levels below 6 mg/m3 and no plausible, experimentally verified MOA
explaining such effects without systemic distribution of formaldehyde; however,
some support for indirect effects in rodents is provided by relevant mechanistic
changes in male reproductive organs.
• Nervous System Effects
o The evidence suggests, but is not sufficient to infer, that formaldehyde inhalation
might cause an increase in incidence or mortality from the motor neuron disease,
ALS, given the appropriate exposure circumstances, based on slight epidemiological
evidence. No relevant animal studies (i.e., indeterminate evidence) or mechanistic
information were identified, and additional studies are warranted.
o Likewise, the evidence suggests, but is not sufficient to infer, that formaldehyde
inhalation might cause increases in multiple manifestations of neurobehavioral
toxicity, given appropriate exposure circumstances, based primarily on slight
evidence of effects in animals of two species across several behavioral domains
(i.e., neural sensitization; tests of learning and memory; and tests of motor-related
behaviors), and supported by slight evidence in human observational and controlled
exposure studies. An experimentally verified MOA explaining such effects without
systemic distribution of formaldehyde is lacking; however, some mechanistic
findings support the potential for indirect effects on relevant brain regions. Well-
conducted studies of these potential effects are currently unavailable.
o The evidence suggests, but is not sufficient to infer, that formaldehyde inhalation
might cause developmental neurotoxicity, given appropriate exposure
circumstances, based on slight evidence in animals for neuropathology and
potentially supportive mechanistic findings in relevant brain regions. However, as
neither an experimentally verified MOA nor relevant studies in children were
identified, this is an area in need of further research.
This document is a draft for review purposes only and does not constitute Agency policy.
1-557 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1.4.3. Summary of Evidence Integration Conclusions for Carcinogenicity
1 "Formaldehyde Is Carcinogenic to Humans by the Inhalation Route of Exposure"31
2 Several lines of evidence support this conclusion. Specifically, the hazard descriptor
3 carcinogenic to humans is independently substantiated by three lines of evidence, namely
4 evidence integration judgments that the evidence demonstrates that formaldehyde inhalation
5 causes nasopharyngeal cancer, sinonasal cancer and, myeloid leukemia, in exposed humans, given
6 appropriate exposure circumstances.
7 These overall confidence conclusions, as well as the strength of the human and animal
8 evidence (i.e., robust, moderate, slight, indeterminate), were based on the currently available
9 evidence using the approaches described in the description of methods in the Preface of this report,
10 which included a consideration of mechanistic evidence when drawing each conclusion. Note that,
11 as the site-specific relationship of the animal data to the specific human cancer types involved
12 additional considerations, the inference regarding the relevance of the animal data to each specific
13 human cancer is presented herein as a component of the animal evidence judgments.
14 Evidence Integration Conclusion: Carcinogenic to Humans
15 Three separate evidence integration judgments independently substantiate this conclusion:
16 • Nasopharyngeal cancer—The evidence demonstrates that formaldehyde inhalation
17 causes nasopharyngeal cancer (NPC) in humans. This is based primarily on observations of
18 increased risk of NPC in groups exposed to occupational formaldehyde levels and nasal
19 cancers in mice and several strains of rats, with strong, reliable, and consistent mechanistic
20 evidence in both animals and humans (i.e., robust evidence for both the human and animal
21 evidence, and strong mechanistic support for the human relevance of the animal data). The
22 nasopharynx, although not typically specified in animal studies, is the region adjacent to the
23 nasal cavity, where the animal evidence was predominantly observed (thus, the animal
24 evidence is judged as robust). In addition, the evidence is sufficient to conclude that a
25 mutagenic MOA of formaldehyde is operative in formaldehyde-induced nasopharyngeal
26 carcinogenicity.
27 • Sinonasal cancer—The evidence demonstrates that formaldehyde inhalation causes
28 sinonasal cancer (SNC) in humans. This is based primarily on observations of increased risk
29 of SNC in groups exposed to occupational formaldehyde levels (i.e., robust human evidence)
30 and supported by apical and mechanistic evidence for nasal cancers across multiple animal
31 species. Some uncertainties remain in the interpretation of the animal nasal cavity data as
32 wholly applicable to interpreting human sinonasal cancer (thus, the animal evidence is
31The hazard conclusion for cancer is consistent with those drawn by other expert review panels.
Formaldehyde was classified as a known carcinogen by the NTP (20111 and a Group 1 carcinogen by IARC
(2012. 20061. both based on evidence for nasal cancers in humans and animals and myeloid leukemia in
humans, with supporting data on mechanisms of carcinogenesis. In addition, an expert committee convened
by the NAS NRC confirmed the conclusions of the NTP 12th RoC and conducted an independent review of the
literature through 2013, concluding that formaldehyde is a known carcinogen. The European Union and
Health Canada concluded that formaldehyde is a genotoxic carcinogen with a cytotoxic MOA based on nasal
cancer evidence fSCOEL. 2017: ECHA. 2012: Health Canada. 2006. 20011.
This document is a draft for review purposes only and does not constitute Agency policy.
1-558 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Toxicological Review of Formaldehyde—Inhalation
judged as moderate). While uncertainties remain, the evidence is sufficient to conclude that
a mutagenic MOA of formaldehyde is operative in formaldehyde-induced sinonasal
carcinogenicity.
• Myeloid leukemia—The evidence demonstrates that formaldehyde inhalation causes
myeloid leukemia in humans, given appropriate exposure circumstances. This is based
primarily on robust human evidence of an increased risk of the occurrence of myeloid
leukemia in epidemiological studies among different populations exposed to occupational
formaldehyde levels representing diverse exposure settings. The findings from the
occupational cohorts are further supported by other studies of human occupational
exposure providing strong and coherent mechanistic evidence that formaldehyde exposure
is associated with the detection of additional endpoints relevant to LHP cancers, including
an increased prevalence of multiple markers of genotoxicity in peripheral blood and
myeloid progenitors. Indirect support is also provided by evidence of other systemic health
effects (e.g., reproductive or developmental toxicity) and mechanistic evidence indicating
changes in immune cell populations and markers of inflammation (e.g., oxidative stress) in
the peripheral blood of exposed humans and animals, although the exact pattern of
immune-related changes across studies and species was difficult to interpret Notably,
leukemia has not been observed in the two available rodent bioassays of chronic exposure,
including one testing both sexes of rats and mice, and the evidence for genotoxicity in the
peripheral tissues of exposed rodents is weak, providing indeterminate evidence of LHP
cancers in animals. Taken together, it appears that mechanisms yet to be elucidated that do
not involve direct interactions of formaldehyde in the bone marrow need to be considered,
and that either the mechanistic pathways stimulated by formaldehyde are different in
animals or that the existing animal models tested thus far do not characterize the disease
process in humans for these cancers. The exact mechanism(s) leading to cancer formation
outside of the respiratory tract are unknown.
The remaining evidence relevant to evaluating the potential for formaldehyde inhalation to
cause cancer did not contribute to the overall hazard conclusion above, including formal
evaluations of the following cancer types:
• Oropharyngeal/hypopharvngeal cancer—The available evidence suggests, but is not
sufficient to infer, that formaldehyde inhalation might cause oropharyngeal/
hypopharyngeal cancer in humans, given appropriate exposure circumstances. This is
based primarily on slight human evidence from epidemiological findings and potentially
relevant mechanistic changes (e.g., in buccal cells) and supporting slight animal evidence of
preneoplastic lesions and mechanistic changes. While cancer site concordance is not
required for hazard determination (U.S. EPA. 2005al given the known reactivity and
distribution of inhaled formaldehyde, a lesser level of confidence in the applicability of the
animal nasal findings is inferred for this cancer type as compared to NPC or SNC and the
evidence overall is not interpreted to provide reasonable support for a MOA.
• Multiple myeloma—The available evidence suggests, but is not sufficient to infer, that
formaldehyde inhalation might cause multiple myeloma, given the appropriate exposure
circumstances. This is primarily based on slight human evidence from epidemiological
findings. The animal evidence is indeterminate, and the available mechanistic information
was not interpreted to be influential, indicating a need for additional study.
This document is a draft for review purposes only and does not constitute Agency policy.
1-559 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Hodgkin lymphoma— The available evidence suggests, but is not sufficient to infer, that
formaldehyde inhalation might cause Hodgkin lymphoma, given the appropriate exposure
circumstances. This is primarily based on slight human evidence from epidemiological
findings. The animal evidence is indeterminate, and the available mechanistic information
was not interpreted to be influential, indicating a need for additional study.
Laryngeal cancer— All the evidence related to laryngeal cancer was judged as
indeterminate; thus, the evidence was inadequate to determine whether formaldehyde
inhalation exposure may be capable of causing this cancer type.
Lymphatic leukemia—All the evidence related to lymphatic leukemia was judged as
indeterminate; thus, the evidence was inadequate to determine whether formaldehyde
inhalation exposure may be capable of causing this cancer type.
This document is a draft for review purposes only and does not constitute Agency policy.
1-560 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
2. DOSE-RESPONSE ANALYSIS
2.1. INHALATION REFERENCE CONCENTRATION FOR EFFECTS OTHER
THAN CANCER
The reference concentration RfC (expressed in units of mg/m3) is defined as an estimate
(with uncertainty spanning perhaps an order of magnitude) of a continuous inhalation exposure to
the human population (including sensitive subgroups) that is likely to be without an appreciable
risk of deleterious effects during a lifetime. It can be derived from a no-observed-adverse-effect
level (NOAEL), lowest-observed-adverse-effect level (LOAEL), or the 95% lower bound on the
benchmark concentration (BMCL), with uncertainty factors (UFs) generally applied to reflect
limitations of the data used. The approach for deriving an overall RfC involves the following steps,
the specific methods and considerations for which are outlined within each of the subsequent
sections:
1) Identify studies and endpoints for each health effect that are sufficient (i.e., with one of the
two strongest evidence integration judgments for hazard, namely of evidence
demonstrates or evidence indicates, and high or medium confidence in the study
methodological conduct, as well as data amenable for dose-response analysis), and calculate
points of departure (PODs)
2) Derive candidate RfCs (cRfCs) by applying UFs to the PODs
3) Select organ- or system-specific RfCs (osRfCs) based on the cRfCs
4) Select an overall RfC based on the osRfCs
Candidate RfCs were derived from studies supporting several health hazards, including
sensory irritation (eye irritation), pulmonary function (peak expiratory flow rate), allergies
(rhinoconjunctivitis, atopic eczema), current asthma (i.e., symptoms or medication in the previous
12 months), degree of asthma control, respiratory tract pathology (squamous metaplasia),
developmental toxicity (delayed pregnancy), and male reproductive toxicity (testes weight, serum
testosterone). The rationale for the prioritization of specific endpoints selected for use in dose-
response evaluation (e.g., squamous metaplasia rather than hyperplasia for respiratory tract
pathology) is discussed in Chapter 1. The cRfCs for sensory irritation, pulmonary function, immune
effects including allergies and current asthma, and female and developmental toxicity were derived
using data from epidemiology studies, while the cRfCs for respiratory tract pathology and male
reproductive toxicity were derived using data from experimental animals. cRfCs were not derived
for nervous system effects, as the available evidence was deemed to be too uncertain, and thus
insufficient, to support quantitative dose-response assessment. In this case, the primary sources of
This document is a draft for review purposes only and does not constitute Agency policy.
2-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
uncertainty in the data included study-specific methodological limitations32 and a lack of
reproducibility across well-conducted studies within the databases for the individual outcomes
evaluated, all in the context of an incomplete evidence base.
The studies most applicable to formaldehyde exposure settings in the general population
were preferred, and the level of confidence in cRfCs was incorporated in the derivation of the
osRfCs. An overall RfC for formaldehyde of 0.007 mg/m3 was selected. This value is within the
narrow range (0.006-0.009 mg/m3) of the group of respiratory system-related RfCs (i.e., sensory
irritation, pulmonary function, allergy-related conditions, and current asthma prevalence or degree
of control), which together are interpreted with high confidence based on the confidence
considerations outlined below. These osRfCs are based on PODs that are the lowest of those
identified in population studies for formaldehyde hazards, and with the lowest composite
uncertainty. Uncertainties in the overall RfC are discussed with the rationale for the RfC selection
in Section 2.1.4.
While the RfC is interpreted to be a concentration associated with minimal risk over a
lifetime of exposure, a few of the hazards or outcomes, including sensory irritation symptoms, or
the degree of asthma control, could be relevant to a shorter exposure time frame. The applicability
of the osRfC to shorter exposure periods is noted for the relevant hazards.
2.1.1. Choice of Studies and Endpoints and Calculation of PODs
Data sufficient to support dose-response analyses were available for all of the health
systems for which the integration of all the evidence resulted in judgments of evidence
demonstrates or evidence indicates that inhalation of formaldehyde can cause adverse human
health effects. Rationales for study selection and the specifics of cRfC calculations, as well as the
determination of confidence in the PODs, are detailed in this section.
Methods of Analysis
From among the body of evidence used for the hazard identification assessment, selection
of the studies for dose-response assessment used information from the study confidence
evaluations, with particular emphasis on conclusions regarding the characteristics of the study
population and the accuracy of formaldehyde exposure, the severity of the observed effects, and the
exposure levels analyzed (see Table 2-1 and Appendix A.5.1). Human studies were preferred over
laboratory animal studies if quantitative measures of exposure were analyzed in relation to health
endpoints. Epidemiological studies that evaluated groups most representative of the general
32For example, the reported formaldehyde exposure data in epidemiology studies demonstrating associations
were generally not amenable to use in quantitative dose-response analysis. In the available animal studies,
there were prominent methodological limitations including poor exposure quality; an inability to rule out
nonspecific effects due to irritant or odorant responses, or due to conditions unlikely to be relevant to human
exposure scenarios; and deficiencies in the reporting of quantitative results important to quantitative
analyses (e.g., litter information).
This document is a draft for review purposes only and does not constitute Agency policy.
2-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
population (i.e., residential or school-based study populations) were preferred if exposure-
response analyses were presented. These criteria emphasize the use of high or medium confidence
studies with appropriate study designs, complete reporting of results, and results that would not be
reasonably explained by selection bias or information bias or altered by adjustment for
confounding. Studies with risk estimates for multiple exposure levels or regression coefficients per
unit of formaldehyde concentration were preferred because they provided information about the
concentration-response trend. The presence of an exposure-response gradient and analyses of data
at lower exposure levels were considered. In the absence of such information, a LOAEL or NOAEL
was identified using a rationale specific to the exposure data presented in the study.
If there were no adequate studies of human exposure for exposure-response analysis, then
studies of experimental animals were evaluated. Using similar criteria as described for human
studies (above), the overall quality of the experimental animal studies was considered
(e.g., preference was given to studies with less likelihood of bias, confounding, etc.). To a large
extent, this comparison of studies within a given health domain was facilitated using the study
evaluation categories described in the Preface on assessment methods and organization (e.g., high or
medium confidence). In addition, experimental animal studies were preferred if they were from
models that respond most like humans; tested the effects of formaldehyde inhalation exposure
using paraformaldehyde as the test article; were of longer exposure duration and follow-up,
evaluated across multiple exposure levels; and were adequately powered to detect effects at lower
exposure levels. Table 2-1 shows the high and medium confidence studies for each hazard that
included information possibly suitable to evaluate dose-response relationships and indicates for
each study whether the study was used to develop a POD or the rationale for why the study was not
suitable.
Once the preferred studies and effect(s) were identified within each health domain, PODs
were derived for each chosen endpoint using a NOAEL, LOAEL, or BMCL. These PODs were then
adjusted (PODadj), if appropriate, to extrapolate from the estimated or measured exposures to a
continuous exposure scenario. For laboratory animal studies, as applicable (U.S. EPA. 1994). this
PODadj was then converted to a human equivalent concentration (PODhec) using a mathematical
calibration. Each of the following organ/health system discussions includes a description of
confidence in the PODs derived from the individual studies.
This document is a draft for review purposes only and does not constitute Agency policy.
2-3 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-1. Eligible studies for POD derivation and rationale for decisions to
not select specific studies
Reference
Endpoint
POD
derived?
Rationale for decisions to
not select
Sensory irritation
Hanrahan et al. (1984)
Eye irritation: Prevalence
Yes
Kulleetal. (1987)
Eye irritation: Prevalence
Yes
Andersen and Molhave (1983)
Eye irritation: Prevalence
Yes
Liu et al. (1991)
Eye irritation: Prevalence
No
Incomplete reporting of modeling
results. Provided support for use of
Hanrahan et al. (1984)
Mueller et al. (2013)
Eye irritation: Tear film break-
up time, symptom score using
visual analogue scale (VAS)
No
An exposure-response trend was not
observed for either endpoint. Difficult
to define an adverse response level
cutoff for these endpoints
Lang et al. (2008)
Eye irritation: Conjunctival
redness, blinking frequency,
symptom score
No
Difficult to define an adverse response
level cutoff for these endpoints and
appeared to be less sensitive than
symptom score
Pulmonary function
Krzvzanowski et al. (1990)
PEFR
Yes
Malaka and Kodama (1990)
FEV1( FEF25-75
No
Incomplete reporting of modeling
results
Kriebel et al. (2001)
PEFR
No
Difficult to use modeling results
because of covariance in model
coefficients
Wallner et al. (2012)
FEF25-75
No
Incomplete reporting of modeling
results
Immune-mediated conditions: allergic conditions
Annesi-Maesano et al. (2012)
Rhinoconjunctivitis prevalence:
Children
Yes
Matsunaga et al. (2008)
Allergic rhinitis, atopic eczema
Yes
Yon et al. (2019)
Rhinitis prevalence
No
Minimal details provided on
formaldehyde distribution
Neamtiu et al. (2019)
Allergy-like symptoms (eyes,
nose and skin)
No
Provided support for use of Annesi-
Maesano et al. (2012)
Garrett et al. (1999)
Atopy prevalence (SPTs):
Children
No
Uncertain window of exposure with
respect to test results
Palczvnski et al. (1999)
Atopy prevalence (SPTs):
Children
No
Uncertain window of exposure with
respect to test results; too few
individuals in third tertile
Immune-mediated conditions: current asthma
This document is a draft for review purposes only and does not constitute Agency policy.
2-4 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference
Endpoint
POD
derived?
Rationale for decisions to
not select
Krzvzanowski et al. (1990)
Current asthma prevalence:
Children
Yes
Annesi-Maesano et al. (2012)
Current asthma prevalence:
Children
Yes
Matsunaga et al. (2008)
Current asthma prevalence:
Adults
No
Definition of current asthma was
narrow and resulted in ascertainment
of fewer cases than would be
expected
Palczvnski et al. (1999)
Current asthma prevalence:
Children and adults
No
Uncertainty regarding asthma
definition (current, ever?); few cases
in third tertile (n < 5)
Kim et al. (2011)
Current asthma prevalence:
Children
No
Provided support for use of Annesi-
Maesano et al. (2012)
Mi et al. (2006)
Current asthma prevalence:
Children
No
Provided support for use of Annesi-
Maesano et al. (2012)
Respiratory and immune-related conditions: asthma control
Venn et al. (2003)
Asthma control: Children
Yes
Dannemiller et al. (2013)
Asthma control: Children
Yes
Respiratory pathology0 in animal studies (exposure duration >52 weeks)
Kerns et al. (1983)
Squamous metaplasia: Nasal
turbinates, Fischer 344 rats
Yes
Kerns et al. (1983)
Squamous metaplasia: Nasal
turbinates, B6C3F1 mice
No
Compared to rats, mice are less
susceptible to formaldehyde
exposure-induced nasal pathology
Woutersen et al. (1989)
Squamous metaplasia: Nasal
turbinates, Wistar rats
Yes
Appelman et al. (1988)
Squamous metaplasia: Nasal
turbinates, Wistar rats
No
Limited sample size (n = 10/group) and
exposure duration (1 year), as
compared to Kerns et al. (1983)
(n = up to ~100/group; 24 months)
and Woutersen etal. (1989)
(n = 30/group; 28 months)
Kamata et al. (1997)
Squamous metaplasia: Nose
and trachea, Fischer 344 rats
No
Uncertainty associated with methanol
coexposure from formalin exposure,
although a control group received
methanol; small sample size at
28 months (i.e., no animals in the high
exposure group survived; only n = 7 at
2.43 mg/m3); metaplasia results
pooled across scheduled sacrifices (12,
18, 24, and 28 months) and dead
animals includes exposure durations
that are less likely to reveal effects
Developmental toxicity (occupational cohort)
This document is a draft for review purposes only and does not constitute Agency policy.
2-5 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Reference
Endpoint
POD
derived?
Rationale for decisions to
not select
Taskinen et al. (1999)
Time to pregnancy
Yes
Taskinen et al. (1999)
Spontaneous abortion
No
Uncertain applicability of temporal
window for exposure data with
respect to reported spontaneous
abortions
Franklin et al. (2019)
Birth weight, head
circumference
No
Uncertainties in exposure distribution
due to large % < LOD and impact on
quantitative results
Chang et al. (2017)
Birth weight
No
Evidence of confounding by co-
exposure; Log transformed
formaldehyde concentration
Male reproductive toxicity in animal studies
Ozen et al. (2002)
Relative testes weight,
13-week exposure
Yes
Ozen et al. (2005)
Serum testosterone, Wistar rat,
13-week exposure
Yes
Ozen et al. (2005)
Seminiferous tubule diameter,
Wistar rats, 13-week exposure
No
Unclear usefulness of data for
quantification: for example, as the
results reflect randomly selected
tubules, the tubules could be
oversampled from individual animals
within a group, and the mean and
variability across the group of animals
when using the animal as the
experimental unit is unknown
Vosoughi et al. (2013);
Vosoughi et al. (2012)
Seminiferous tubule diameter,
NMRI mice, 10-day exposure
No
Short exposure duration
Vosoughi et al. (2013):
Vosoughi et al. (2012)
Sperm abnormalities, NMRI
mice, 10-day exposure
No
Short exposure duration
Vosoughi et al. (2013):
Vosoughi et al. (2012)
Serum testosterone, NMRI
mice, 10-day exposure
No
Short exposure duration
Vosoughi et al. (2013):
Vosoughi et al. (2012)
Testes weight, NMRI mice,
10-day exposure
No
Short exposure duration
Sarsilmaz et al. (1999)
Leydig cell quantity or nuclear
damage, Wistar rat, 4-week
exposure
No
Short exposure duration
Sarsilmaz et al. (1999)
Testes weight (relative), Wistar
rats, 4-week exposure
No
Short exposure duration; non-
preferred metric (absolute testes
weight preferred)
Sapmaz et al. (2018)
Seminiferous tubule measures,
Sprague-Dawley rats, 4- and
13-week exposure
No
Short exposure duration (for 4-week
experiment); single exposure level
Abbreviations: PEFR = peak expiratory flow rate; FEF = forced expiratory flow; FEV = forced expiratory volume;
SPT = skin prick test; IIIR = inhalation unit risk.
This document is a draft for review purposes only and does not constitute Agency policy.
2-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
aNote: squamous metaplasia was the preferred endpoint for RfC derivation (see Section 1.2.4 for explanation).
Hyperplasia and cell proliferation are considered in the context of the cancer III R.
Sensory Irritation
The effects of formaldehyde on sensory irritation are thought to occur via direct
interactions of formaldehyde with cellular macromolecules in the nasal mucosa and stimulation of
the trigeminal nerve, mediated through cation channels, resulting in the rapid detection of a
burning sensation. It is not clear if desensitization occurs over time or the concentrations or
timeframes over which this might occur. Because of the rapid nature of the irritant response
generated by inhalation of formaldehyde, the studies that were considered to be the most
informative for derivation of a cRfC were those where the exposure assessment was concurrent
with the outcome assessment
Data from studies in humans involving residential populations with continuous exposure, as
well as controlled human exposure studies evaluating acute effects were determined to be
pertinent to the derivation of a cRfC. The studies of anatomy students and formaldehyde-exposed
workers assessed exposure settings with high formaldehyde concentrations and with frequent
peaks. Thus, average formaldehyde concentrations or TWAs, the exposure metrics used by these
studies, could not capture the variation inherent in these types of settings. Therefore, prevalence of
irritation symptoms might not necessarily have corresponded to the time frame of the exposure
measurements.
Hanrahan et al. (19841 used 1-hour average formaldehyde measurements taken in two
rooms in the mobile homes of a group including teenagers and adults and presented the predicted
concentration-response for prevalence of "burning eyes" experienced by the participants since
moving into the homes from a logistic regression model that adjusted for age, sex, and smoking.
These data were used to derive a POD of 0.09 mg/m3, the concentration corresponding to a
benchmark response (BMR) of 10%. The mathematical expression for the exposure-response
pattern and a BMCLio was determined from a graph of the predicted prevalence and upper and
lower 95% confidence bounds for several concentrations between 100 and 800 ppb (0.12-
0.98 mg/m3).33 The concentration corresponding to a 13% prevalence of "burning eyes" was
calculated from the model (for model details see Appendix B. 1.2). The 13% prevalence represents
a 10% increase in irritation as a result of formaldehyde exposure in addition to an assumed
background prevalence of 3% (in the absence of formaldehyde exposure). The background
33EPA estimates that 44% of the average measured concentrations were below 100 ppb. While it is not clear
from the published report what the distribution of exposures below 100 ppb was, if it can reasonably be
assumed that the formaldehyde concentrations were log-normally distributed with median of 160 ppb and a
standard deviation of 30 ppb (based on the reported standard deviation from the outdoor measurements),
then it would be expected that about 44% of the measured indoor samples were below 100 ppb, with 36%
below 50 ppb. Given that the measured indoor levels were likely to have been more variable than the
reported outdoor levels, the true indoor standard deviation would likely have been higher than 30 ppb and,
consequently, the percentages below 100 ppb and below 50 ppb would have been greater.
This document is a draft for review purposes only and does not constitute Agency policy.
2-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
prevalence of 3% was considered to be a reasonable estimate, but the impact of using alternative
estimates (1 and 2%) was evaluated.
Liu et al. (1991) collected data on symptoms for a period during and 1 week prior to the
exposure assessment using a sampling protocol that captured average formaldehyde
concentrations in the (mobile) home (7-day mean concentration from two rooms). Although Liu et
al. (1991) estimated an exposure-response relation using logistic regression, the regression
coefficients estimated by the model were not reported. The range of 7-day average formaldehyde
concentrations measured by Liu et al. (1991) was comparable to the air concentrations in the
homes studied by Hanrahan et al. (1984) (10-460 ppb [0.012-0.57 mg/m3]). Although a cRfC was
not derived from Liu et al. (1991), the data could be used to check the estimated POD based on
Hanrahan et al. f!984I The prevalence of 10% during the winter and 13.3% during the summer in
the lowest exposure category (<7 ppm-hr/week) is close to the best estimate of 13% benchmark
response estimated from Hanrahan et al. (1984). which occurred at a concentration of 0.19 mg/m3.
A cumulative exposure of 7 ppm-hr/week is approximately equal to 0.07 ppm (0.086 mg/m3)
assuming that participants were in their homes 60% of a 24-hour day, supporting the selection of
the lower confidence limit of the BMCLio (0.087 mg/m3) from the Hanrahan et al. (1984) results as
the POD.
PODs were determined using two controlled human exposure studies of formaldehyde for
which there was medium confidence that evaluated multiple levels of exposure (see study
descriptions in Table 2-2). Kulle et al. (1993) evaluated results for participants exposed for 3 hours
once a week to five concentration levels (including a clean air exposure), while Andersen and
Molhave (1983) exposed subjects for 5-hour periods to four concentration levels with a 2-hour
clean air exposure prior to each trial. The occurrence of irritation symptoms during the clean air
exposure was not reported. The results of these studies were evaluated in BMD models to identify
the concentration at which a 10% increase in symptoms at concentrations above the clean air
exposure was observed (see Appendix B.1.2 for details of the models). Two sets of models were
evaluated using the data from Andersen (1983) and estimates of 0 and 3% for prevalence of
irritation during the clean air exposure. The benchmark concentration (BMC) of 0.37 mg/m3
derived from the model using a baseline prevalence of 3% was selected.
The results from two other controlled human exposure studies were considered, but PODs
were not derived. Blinking frequency, an objective measure of irritation evaluated by Lang et al.
(2008) and Mueller et al. (2013.), was highly variable in all exposure groups, and it was difficult to
define a meaningful magnitude of change in these measures that would be considered to be
minimally adverse for the selection of a POD. Further, increased blinking frequency was observed at
a higher exposure level compared to eye irritation symptoms.
Table 2-2 presents the studies used to calculate a POD with the epidemiology data and
sequence of calculations leading to the derivation of a POD for each data set with effects relating to
sensory irritation.
This document is a draft for review purposes only and does not constitute Agency policy.
2-8 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-2. Summary of derivation of PODs for sensory irritation
Endpoint and
reference
Population
Observed effects by exposure level3
PODadj
(mg/m3)
Residential exposure
Symptom prevalence
Hanrahan et al. (1984)
Teenage and adult
(M and F), n = 61
Third-degree polynomial model fit to In prevalence
odds using presented results of logistic regression
analysis: upper 95% confidence bound for predicted
prevalence between <0.123 and 0.98 mg/m3, BMCi0:
concentration where an increased prevalence of 10%
over a 3% background prevalence is anticipated
BMCiob 0.19
BMCLio 0.09c
Controlled human exposure
Symptom prevalence
Kulleetal. (1987)
Nonsmoking,
healthy, n = 10-19,
mean age 26.3 yr
(M and F)
Exposure and proportion responding
mg/m3
0
0.62
1.2
2.5
3.7
%
5
0
26
53
100
trend, p < 0.05
Probit model BMC = 0.69 ppm
BMCio 0.85c
BMC/2d 0.42
Symptom prevalence
Andersen and Molhave
(1983)
Healthy students,
n = 16, age
30-33 years, 31.2%
smokers (M and F)
Exposure and percentage responding (prevalence at
the end of exposure)
mg/m3
0.3
0.5
1.0
2.0
%
19
31
94
94
BMCio 0.37 c
BMC/2d 0.19
Assuming prevalence for clean air dose
0% Log-logistic model BMC = 0.26 mg/m3
3% Log-logistic model BMC = 0.37 mg/m3
Concentrations reported in publication converted to mg/m3.
bBMCio benchmark concentration at 10% increase in prevalence overestimated 3% background prevalence. An
increase of 10% was selected consistent with EPA guidelines (U.S. EPA, 2012) because the endpoint, burning eyes
with mild to moderate severity, was considered a minimally adverse outcome.
cThe POD was not adjusted for a 24-hour equivalent concentration because the timing of formaldehyde
measurements was concluded to be appropriate to the time frame of reported symptoms.
dThe BMD models did not account for the correlated measures between concentration levels (each participant was
exposed to each concentration). Therefore, the 95% confidence limit for the BMC estimated by the model is too
narrow to use as the POD. A factor of 2 was used to adjust the BMC to identify a lower estimate that
approximates the BMDL
1 Conclusion
2 The POD derived using the exposure-response model using prevalence data from the
3 residential population in Hanrahan et al. (1984) is 0.09 mg/m3. EPA placed medium confidence in
4 the results of this study. The study by Hanrahan et al. (1984) is pertinent to the U.S. general
5 population because: (1) the population was randomly selected from the general population in the
6 study area; (2) the exposure levels were concluded to reflect the usual, relatively constant
7 formaldehyde concentrations in the residences; and (3) exposed individuals included a range of
8 ages (teenagers and adults), men and women, and some with chronic disease. Moreover, a
9 significant proportion of the study population was estimated to be exposed to average
This document is a draft for review purposes only and does not constitute Agency policy.
2-9 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
formaldehyde concentrations below 0.05 mg/m3. The impact of potential confounding by the
presence of coexposures is likely to be minimal. The regression model adjusted for age, sex and
smoking, and the presence of smokers or gas appliances in the home, sources that might contribute
to variability in concentrations, were not associated with indoor formaldehyde concentrations.
Other emissions released from the same sources as formaldehyde that also might contribute to eye
irritation, such as phenols from resins in floor or wall coverings or pinene and terpenes from wood
products, were not analyzed. However, a strong exposure-response relationship with
formaldehyde concentration was observed by this study, which argues against a large effect by
residual confounding by other coexposures.
The PODs based on the two controlled human exposure studies were 0.19 and 0.42 mg/m3
fKulle etal.. 1987: Andersen and Molhave. 19831. less than an order of magnitude greater than the
BMCL estimated from residential exposure. There is less confidence in the PODs based on these
studies because: (1) the study participants were young, healthy volunteers, not representative of
the age distribution and health status in the general population; (2) the PODs are based on small
sample size, more subject to random variation; and (3) formaldehyde concentrations were high,
imposing substantial uncertainty regarding responses at the low tail of the exposure distribution.
The utility of the PODs from these two controlled exposure studies may be greater for other, less
than chronic, exposure durations (e.g., derivation of an acute RfC).
The exposure-response pattern presented in Hanrahan et al. (1984) is consistent with the
overall pattern exhibited when all of the studies of exposure in mobile homes and controlled human
exposure studies with dose-response data less than 1 mg/m3 are graphed together (see Figure 1-3).
Therefore, the POD estimated from Hanrahan et al. (1984) is supported by the set of epidemiology
studies describing formaldehyde-related irritation in humans. Confidence in the POD is medium,
reflecting uncertainty in the temporal relationship of the exposure measurements with respect to
the assessment of irritation symptoms.
Pulmonary Function
The studies that estimated an exposure-response relation with formaldehyde concentration
for effects on pulmonary function involved exposures to anatomy students (Kriebel et al.. 2001). an
occupational population fMalaka and Kodama. 19901. school children fWallner etal.. 20121. and a
residential population (Krzyzanowski etal.. 1990). A POD was derived from the analyses reported
by Krzyzanowski et al. (1990), but not from the other studies that analyzed exposure-response
relationships because important data were not available (see Table 2-1).
Declines in peak expiratory flow rate (PEFR) were associated with increases in 2-week
average indoor residential formaldehyde concentrations, with greater declines observed in children
(5-15 years of age) compared to adults (Krzyzanowski et al.. 1990). This study of effects in a
residential population used the most thorough exposure assessment protocol and repeated
measurements of PEFR, thus enhancing the ability to detect an association at the lower
concentrations found in the homes. Mean formaldehyde levels were 26 ppb (0.032 mg/m3), and
This document is a draft for review purposes only and does not constitute Agency policy.
2-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
more than 84% of the homes had concentrations 40 ppb (0.049 mg/m3) and lower. A BMCio of
0.033 mg/m3 and BMCLio of 0.021 mg/m3 were determined from the regression coefficient from a
random effects model of PEFR among children reported by the study authors (for details, see
Appendix B.1.2). Table 2-3 presents the study used to calculate a POD with the epidemiology data
and sequence of calculations leading to the derivation of a POD relating to pulmonary function.
Table 2-3. Summary of derivation of PODs for pulmonary function
Endpoint and
reference
Population
Results by exposure
level3
BMC and BMCL
(mg/m3)
PODadj"
(mg/m3)
PEFR
Krzvzanowski et
al. (1990)
Residential,
prevalence
202 households, 298
children aged
5-15 years, current
asthma prevalence
15.8%;
613 adults and
adolescents >15 yr,
24.4% current smokers,
current asthma
prevalence 12.9%
Random effects model;
decreased PEFR, children
-1.28 ± 0.46 L/minute-ppb
(95% upper bound -2.04
L/minute-ppb)
Formaldehyde
concentrations: Mean 0.032
mg/m3, maximum 0.172
mg/m3
BMC10c 0.033
BMCLio 0.021
0.02
Concentrations reported in publication converted to mg/m3.
bThe POD was not adjusted for a 24-hour equivalent concentration because formaldehyde is present in all indoor
environments and time-activity information for participants was not reported.
cBMCio benchmark concentration associated with a 10% decrease in pulmonary function. A BMR of 10% reduction
in PEFR was selected as a cut-off point for adversity, based on rationales articulated by the American Thoracic
Society (ATS, 2000). The American Thoracic Society (ATS, 2000) recommended that "a small, transient loss of lung
function, by itself, should not automatically be designated as adverse" and ATS cited EPA's 1989 review of ozone,
which offered a graded classification of lung function changes in persons with asthma as "mild," "moderate," or
"severe" for reductions of less than 10,10-20, and more than 20%, respectively (U.S. EPA, 1989). ATS (2000)
concluded that, in evaluating the adverse health effects of air pollution at the level of population health
(compared to individual risk), "[a]ssuming that the relationship between the risk factor and the disease is causal,
the committee considered that such a shift in the risk factor distribution, and hence the risk profile of the exposed
population, should be considered adverse." This was specifically considered by ATS (2000) even when
"[e]xposure to air pollution could shift the distribution toward lower levels without bringing any individual child to
a level that is associated with clinically relevant consequences." A moderate adverse effect at functional
decrements of 10-20% was considered the best indicator of adverse effects in the study population.
Conclusion
The adjusted POD estimated using the results of Krzyzanowski etal. (1990) (0.021 mg/m3)
was derived from the responses of a randomly selected population of adults and children
continuously exposed to formaldehyde in their homes. In this large, population-based sample, the
investigators observed a linear relationship between increased formaldehyde exposure and
decreased peak expiratory flow rate (PEFR) among children exposed to average concentrations of
0.032 mg/m3 (26 ppb), and a stronger response was observed among children with asthma.
Krzyzanowski et al. (1990) adjusted for smoking and NO2 levels in their analyses; thus, confounding
This document is a draft for review purposes only and does not constitute Agency policy.
2-11 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
by these coexposures can be ruled out. Further, a strong exposure-response relationship with
formaldehyde concentration was observed by this study, which argues against a large effect by
residual confounding by other coexposures. This study was able to evaluate associations with
relatively constant, low formaldehyde concentrations and used a high-quality exposure
measurement protocol, thus, reducing uncertainties for low-dose extrapolation (0.012 to
0.172 mg/m3 (Ouackenboss etal.. 1989c). Average formaldehyde concentrations in these studies
were pertinent to those experienced by the general population (the authors reported that more
than 84% of the homes had concentrations 40 ppb [0.049 mg/m3] and lower). The POD is based on
the findings among children and was derived from a regression model that adjusted for important
potential confounders including asthma status, smoking status, socioeconomic status, NO2 levels,
episodes of acute respiratory illness, and the time of day. Thus, confidence in the POD is high.
Immune-mediated Conditions, Focusing on Allergies and Current Asthma
Allergic conditions and sensitization
Three high or medium confidence epidemiology studies in children or adults provide data
on measures of allergy-related conditions needed to conduct an exposure-response analysis
fAnnesi-Maesano etal.. 2012: Billionnet et al.. 2 011: Matsunaga et al.. 20081. As discussed in
Section 1.2.3 and depicted in Figure 1-8, the results for the studies of rhinoconjunctivitis and
rhinitis are similar, with a stronger effect estimate seen in the only study examining atopic eczema.
Because Billionnet et al. (2011). presented only a dichotomized exposure-response analysis, it is
not considered further as a basis for quantitation; the other studies presented an
exposure-response analysis using formaldehyde as three fAnnesi-Maesano etal.. 20121 or four
groups (Matsunaga et al.. 2008). NOAELs and LOAELs were identified in each of these studies
based on the pattern of risk seen across the exposure groups; the PODs were based on NOAELs.
The study by Annesi-Maesano et al. (2012) uses a relatively long exposure period (5 days), and is a
very large study in a school-based sample of children in France (n = 6,683) with analysis presented
by tertile. Matsunaga et al. (2008) used 24-hour personal samples in a study of 998 pregnant
women in Japan. The primary limitation of the Matsunaga et al. (2008) study is that it is conducted
only among adults, and so is less able to address the variability in susceptibility that would be
anticipated within a population. Given their attributes, the confidence in both studies was
considered high.
Two medium confidence epidemiology studies in children provide data on exposure and
SPTs needed to conduct a quantitative analysis (Garrett etal.. 1999: Palczvnski etal.. 1999).
However, because of the limitations with respect to the timing of the exposure measure and the
interpretation of SPTs, these studies are not considered further as a basis for quantitation.
This document is a draft for review purposes only and does not constitute Agency policy.
2-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Conclusion
For allergy-related conditions (rhinoconjunctivitis), EPA selected NOAEL and LOAEL values
of 0.024 and 0.040 mg/m3, respectively, in the Annesi-Maesano et al. (2012.) study. Higher values
(NOAEL = 0.046, LOAEL = 0.062) were selected based on the study in adults by Matsunaga et al.
(2008). The classification of rhinoconjunctivitis by Annesi-Maesano et al. (2012) was the most
sensitive and specific of the measures, and the narrower confidence intervals in this study reflected
the larger sample size. No other pollutants (e.g., NOx, PM2.5, acetaldehyde, acrolein, ETS) analyzed
by this study were associated with rhinoconjunctivitis.
Current asthma
Several residential and school-based exposure studies examined prevalence of current
asthma in relation to formaldehyde exposure in adults and children in relatively low exposure
settings (see Tables 1-15 and 1-16). As discussed in Section 1.2.3 and seen in Figure 1-9, the six
medium or high confidence studies at exposures of <0.050 mg/m3 do not indicate risk at these
lower exposure levels. Several of the RR estimates from these individual studies at these exposure
levels were limited by low statistical power. However, the consistency of the results, and the
absence of an increased risk in the study by Annesi-Maesano et al. (2012.), a large school-based
study (n = 6,683) that used a 5-day sampling period for formaldehyde measurement, strengthens
the basis for interpreting this set of studies as indicating an absence of risk of current asthma below
0.05 mg/m3. Based on the study by Annesi-Maesano et al. (2012) and this collection of studies, EPA
selected a NOAEL of 0.042 mg/m3 for risk of current asthma.
Two medium confidence studies examined prevalence of current asthma in children in
higher exposure residential settings (>0.05 mg/m3) fZhai etal.. 2013: Krzvzanowski et al.. 19901.
Because Zhai et al. (2013.) presented only a dichotomized exposure-response analysis, it is not
considered further as a basis for quantitation. The Krzyzanowski et al. (1990) results for children
(5-15 years of age) are based on a relatively large sample size, with a comprehensive exposure
assessment protocol (i.e., three locations in the home; two 1-week periods covering two seasons).
An increased prevalence of current asthma was seen in the highest exposure group in a categorical
analysis. The exposure range in this group was 0.075-0.172 mg/m3, but the study also notes that
few values were above 0.11 mg/m3. Based on this information, EPA selected a LOAEL based on the
midpoint of this exposure category using a range estimated as 0.075 to 0.11 mg/m3 (midpoint of
0.092 mg/m3). The estimate for the middle category of exposure was selected as a NOAEL,
although confidence in this NOAEL is lower, given the imprecision of the estimate (n with
asthma =1).
Two of the four medium confidence studies of prevalence of current asthma in adults in
higher exposure residential settings (>0.05 mg/m3) did not provide quantitative results fZhai etal..
2013: Krzyzanowski etal.. 19901. Of the remaining two studies, Billionnet et al. f20111. presented
only a dichotomized exposure-response analysis, and so was not used for quantitation. The four-
This document is a draft for review purposes only and does not constitute Agency policy.
2-13 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
level categorical analysis from Matsunaga et al. (20081 contributed to the evaluation of the NOAEL
for studies with exposures <0.05 mg/m3, but the width of the confidence interval for the highest
exposure group (OR = 2.15; 95% CI 0.41-11.3, for exposures of 0.058-0.161 compared to
<0.022 mg/m3) precludes its interpretation as a LOAEL. Thus, none of the asthma studies in adults
provide a basis for developing a POD.
The collection of occupational studies (see Table 1-17) provides a strong basis for
inferences regarding asthma risk at relatively high exposures (e.g., 0.1 to >0.5 mg/m3) (Fransman et
al.. 2003: Herbert etal.. 1994: Malaka and Kodama. 1990). However, there would be considerable
uncertainty in a POD derived from these studies, identified as a LOAEL, given the dichotomous
analyses used to examine associations and the wide variability in exposure measures within each of
these studies. Therefore, PODs were not determined using the occupational studies.
EPA identified two studies that examined degree of asthma control in children with asthma
in relation to formaldehyde measures in the home (Dannemiller etal.. 2013: Venn etal.. 2003).
Analysis was conducted using four categories of exposure in Venn et al. (2003). based on 3-day
exposure measures taken in the home and daily symptom diaries kept for one month among
children with persistent wheeze. Dannemiller et al. (2013) compared mean exposure levels (based
on 30-minute samples) in two groups (those with very poor control and all others, based on a
five-question survey about symptom control in the past 4 weeks). The larger sample size, longer
sampling period, and more detailed exposure-response analysis makes Venn et al. (2003) a
stronger basis for providing a POD. Additional adjustment of regression models for dampness or
other exposures including visible mold, total VOCs, or N02, did not affect formaldehyde results,
reducing the likelihood of residual confounding by coexposures. EPA selected a NOAEL of
0.027 mg/m3 (median exposure in the third quartile; no or weak RRs seen below this value) and a
LOAEL of 0.041 mg/m3 (median exposure in top quartile, for which a two- to three-fold increased
risk of symptoms was seen). Venn et al. (2003) did identify an exposure-response relationship for
both nighttime symptoms of poor asthma control as OR = 1.40 (95% CI 1.06-1.98) and for daytime
symptoms of poor asthma control as OR = 1.45 (95% CI 1.00-1.94). Using the reported OR per
quartile exposure from the regression results, and the median exposure values for each quartile
(personal communication to EPA (Venn. 2012)). EPA calculated the concentration associated with a
5% increase in prevalence of symptoms above the prevalence observed in the referent group (for
details of BMCL calculations, see Appendix B.1.2). A BMR of 5% was selected because asthma
attacks are overt effects, generally requiring the use of drugs to control symptoms (i.e., a frank or
adverse effect) (U.S. EPA. 2012).
Table 2-4 presents the studies with the epidemiology data and sequence of calculations
leading to the derivation of a POD for each data set with effects relating to allergies and asthma.
This document is a draft for review purposes only and does not constitute Agency policy.
2-14 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-4. Summary of derivation of PODs for allergies and current asthma
based on observational epidemiology studies
Endpoint and
reference
Population
Observed effects by exposure level
PODadj
(mg/m3)
Allergy-related conditions
Rhinoconjunctivitis
(prevalence); school-
based exposure
(5 days)
Annesi-Maesano et
al. (2012)
Children
(M and F)
N = 6,683
Prevalence 12.1%,
OR (95% CI) (adjusted)
<0.0191 mg/m3 1.0 (referent)
>0.0191-0.0284 1.11 (0.94, 1.37)
>0.0284- ~0.055 1.19 (1.03,1.39)
NOAEL selection: 0.024 mg/m3, midpoint of second exposure category
(corresponding to RR 1.11)
LOAEL selection: 0.040 mg/m3, midpoint of third exposure category
(corresponding to RR 1.19)
NOAEL:
0.024
LOAEL:
0.040
Atopic eczema
(prevalence); personal
monitor-based
exposure (24 hours)
Matsunaga et al.
Adult women
(pregnancy
cohort)
N = 998
Atopic eczema Allergic rhinitis
(5.7% prevalence) (14.0% prevalence)
Atopic
eczema
NOAEL:
0.046
LOAEL:
0.062
mg/m3 n OR (95% CI) OR (95% CI)
<0.022 298 1.0 (referent) 1.0 (referent)
0.023-0.033 299 1.03 (0.47,2.29) 1.06 (0.65, 1.73)
0.034-0.057 301 1.11 (0.50,2.42) 0.85 (0.51, 1.40)
0.058-0.161 100 2.36 (0.92,6.09) 1.17 (0.60,2.28)
(trend p-value) (0.08) (0.91)
0.058 to 0.161 vs. 2.25 (1.01,5.01) 1.22 (0.68,2.20)
<0.058
per 0.0123 mg/m3 1.16 (0.99,1.35)
[Stronger associations seen for atopic eczema in women with no family
history of atopy]
For atopic eczema NOAEL selection: 0.046 mg/m3, midpoint for third
exposure category (corresponding to RR 1.11); LOAEL selection: 0.062
mg/m3, estimated median of fourth category (personal communication
to EPA (Matsunaga, 2012)) (corresponding to RR 2.25)
For rhinitis NOAEL selection: 0.062 mg/m3, based on median of fourth
exposure category
(2008)
Current asthma/degree of asthma control
Current asthma
(prevalence);
school-based
exposure (5 days)
Annesi-Maesano et
Children
(M and F)
N = 6,683
Exposure (mg/m3) na OR (95% CI)
NOAEL:
0.042
<0.0191 2,200 1.0 (referent)
>0.0191-0.0284 2,200 1.10 (0.85, 1.39)
>0.0284-~0.055 2,200 0.90 (0.78,1.07)
Approximation, based on tertiles, with total n = 6,590
NOAEL selection: 0.042 mg/m3, midpoint of third exposure category
(corresponding to RR 0.90)
al. (2012)
Current asthma
(prevalence);
residence-based
exposure (two 1-week
periods)
Krzvzanowski et al.
Children
(M and F)
N = 298
Exposure (mg/m3) N Proportion with asthma
NOAEL:
0.062
LOAEL:
0.092
<0.049 248 0.12
0.049-0.074 24 0.04
0.075-0.172 21 0.24
(trend p-value) (0.03)
Only a few values were reported to be above 0.11 mg/m3.
NOAEL selection: 0.062 mg/m3, midpoint of second exposure category
LOAEL selection: 0.092 mg/m3, based on report that only a few values
were above 0.11 mg/m3, so estimated midpoint of third category was
based on range from 0.075 to 0.11, with midpoint of 0.092 mg/m3
(1990)
This document is a draft for review purposes only and does not constitute Agency policy.
2-15 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
Endpoint and
PODadj
reference
Population
Observed effects by exposure level
(mg/m3)
Asthma control among
Children
Exposure (mg/m3) N
Proportion
OR
(95% CI)
NOAEL:
people with asthma,
(M and F)
Frequent nighttime symptoms
0.027
residence-based
N= 194
<0.016 39
0.41
1.0
(referent)
LOAEL:
exposure (3 days)
0.016-0.022 35
0.49
1.40
(0.54, 3.62)
0.041
Venn et al. (2003)
0.022-0.032 36
0.53
1.61
(0.62, 4.19
0.032-0.083 33
0.67
3.33
(1.23, 9.01)
From
(trend p-value)
(0.02)
regression
per quartile increase
1.45
(1.06,1.98)
results:
Frequent daytime symptoms
BMCL5:
<0.016 37
0.62
1.0
(referent)
0.013
0.020-0.022 34
0.47
0.47
(0.47,1.25)
0.022-0.032 37
0.73
2.00
(0.71, 5.65)
0.032-0.083 32
0.73
2.08
(0.71, 6.11)
(trend p-value)
(0.05)
per quartile increase
1.40
(1.00,1.94)
NOAEL selection: 0.027 mg/m3, median of third exposure category
LOAEL selection: 0.041 mg/m3,
median of fourth exposure category
(based on correspondence with Dr. Venn)
Asthma control among
Children
Geometric mean formaldehyde (mg/m3)
NOAEL:
people with asthma,
(M and F)
Very poor control (score <12, n
6) 0.066 mg/m3
0.042
residence-based
N = 37
All others (score >12, n = 31)
0.042 mg/m3 p
= 0.078
exposure (30 minutes)
Venn et al. (2003)
Conclusion
For the analysis of prevalence of current asthma, EPA selected a NOAEL of 0.042 mg/m3
using the data from Annesi-Maesano et al. (20121 (and supported by other studies examining
exposures at <0.05 mg/m3), and a NOAEL of 0.062 mg/m3 based on the data for children in the
study by Krzyzanowski et al. (1990). The NOAEL identified from Krzyzanowski et al. (1990) is
considered to be less reliable because it was based on only one case and a small number of
participants in the exposure group. A BMCLs of 0.013 mg/m3 was also selected based on the data
for degree of asthma control among children with asthma (Venn etal.. 2003). All three studies were
well conducted and are interpreted with high or medium confidence. The study by Annesi-Maesano
et al. (2012) is a large study with a relatively long exposure measurement period, and is supported
by a collection of several other smaller studies (with more imprecise effect estimates) at exposures
of <0.050 mg/m3, which also indicate no increased risk of current asthma at these lower levels
(see Figure 1-9A). The analyses by Annesi-Maesano et al. (2012) were adjusted for age, gender,
passive smoking, and paternal or maternal history of asthma or allergic disease; thus, minimal
impact by confounding is likely. Therefore, both the study and the POD based on the NOAEL in
Annesi-Maesano et al. (2012) is viewed with high confidence. In contrast, only two studies
examined the outcome defined as degree of asthma control among people with asthma
fDannemiller etal.. 2013: Venn etal.. 20031. so the POD derivation based on that specific outcome
(However, Venn et al. (2003) used a strong study design, observed an exposure-related trend in
This document is a draft for review purposes only and does not constitute Agency policy.
2-16 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
response and adjusted the statistical analyses for key confounders, including other indoor
exposures (e.g., visible mold, total VOCs, N02, cotinine levels). Based on these considerations,
confidence in the POD calculations is medium. The lower NOAEL for degree of asthma control in
children with asthma compared with the NOAEL for increased prevalence of current asthma
indicates a greater sensitivity of this more susceptible population.
Respiratory Tract Pathology
The PODs derived were based on exposure-response data from two studies on
histopathological changes (squamous metaplasia34) observed in the nasal passages of F344 rats
(Kerns etal.. 1983) and Wistar rats (Woutersen et al.. 1989). The four medium confidence
occupational studies provide support for the larger evidence base from the experimental studies in
animals fBallarin etal.. 1992: Bovsen etal.. 1990: Holmstrom et al.. 1989c: Edling etal.. 19881.
However, there would be considerable uncertainty in a POD derived from these studies, identified
as a LOAEL, given the dichotomous analyses used to examine associations and the wide variability
in exposure concentrations within each of these studies (e.g., 0.1 to >0.5 mg/m3). Therefore, PODs
were not determined using the occupational studies.
Squamous metaplasia in F344 rat fKerns et al.. 19831
The result of a 2-year bioassay in F344 rats was reported in Kerns et al. (1983) and the
supporting Battelle report (Battelle. 1982). In this study male and female rats, with at least
20/sex/group, were exposed to 2.5, 6.9, and 17.6 mg/m3 with interim sacrifices at 6,12, and
18 months. While Kerns etal. (1983) reported squamous cell metaplasia after inhaled
formaldehyde exposure, detailed information on lesion incidence by concentration, duration, and
cross-section level was provided in the report fBattelle. 19821. The lesions occurred only in the
most anterior region (cross-section Level I) at low concentrations but progressed to more distal
parts of the nose (cross-section Levels II—V) at higher concentrations. Additionally, the incidence of
squamous metaplasia increased with exposure duration. Section 1.2.4 discusses the incidence of
squamous metaplasia in the first five nasal sagittal cross sections of the F344 rat, as reported by
Kerns etal. (1983) and Battelle (1982).35
The POD presented below is based on Level 1. Extrapolation of the rat BMCL to the human
is based on the available dosimetric simulations of formaldehyde flux36 to the nasal lining in rats
34Although a cRfC for hyperplasia was not estimated (see Section 1.2.4 for rationale), a human PODadj that can
be estimated based on the basal cell hyperplasia end point is roughly two-fold greater than that obtained
from the squamous metaplasia data from Woutersen et al. (19891 study. This estimate of hyperplasia
provides context to the development of unit risk estimates for nasal cancer (see Section 2.2.1)
35The data for 27 and 30 mos represent incidence after 3 and 6 mos of nonexposure, respectively, following
24 mos of exposure.
36Flux (in units of mass/area-time) expresses the net transport of formaldehyde from the inspired air to the
air-mucus interface of the nasal lining (prior to disposition within the tissue).
This document is a draft for review purposes only and does not constitute Agency policy.
2-17 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
and humans. This assessment uses dosimetry derived from Kimbell et al. (2001b: 20011 and
Overton et al. (2001) when extrapolating risk-related dose from the rat to the human (discussed in
detail in Appendix B.1.3), and estimates the impact on the dosimetry modeling using Schroeter et al.
(2014).37 A POD based on lesions reported at Level 2 in Battelle (1982) can also be modeled.
However, formaldehyde flux to the nasal lining on Level 2 was not available to EPA and could only
be crudely estimated based on the locations of the nasal regions tabulated in Kimbell et al. (2001a).
as elaborated further in Appendix B.1.3. For this reason, only the Level 1 data were used in
calculating a cRfC.
In determining the BMR level for the POD, severity scores for the squamous metaplasia data
in Battelle (1982) were examined, where provided.38 The average severity score was in the range
of minimal-to-mild at the lowest dose for both the 18- and 24-month durations for Level 1. This
finding supports a BMR of 0.1 extra risk, representing a minimal level of adversity. The 24-month
data for Level 1 cannot be modeled because the dose-response relationship rises too steeply (for
example, the Weibull model fit rises so steeply that the error on the Weibull model power cannot be
bounded). Therefore, the 18-month data, for which incidence rises more gradually, were chosen
even though these data would be less preferred over the 24-month exposure data. To address the
fact that the lesion incidences in Table 1-26 are substantially higher with the longer duration
(i.e., 24-month) data, which suggest a lower POD associated with the 24-month exposure, a UFs will
be applied to the POD derived from the 18-month data.
Interspecies extrapolation of the rat BMCL level to humans was carried out in two steps.
First, average flux values in the Level 1 region of the rat corresponding to the rat BMCL derived
from the incidence of squamous metaplasia were estimated. Next, the exposure concentration at
which any region in the human nose (see Appendix B.1.3) is exposed to this same level of
formaldehyde flux at the inspiratory rate of 15 L/min was estimated from the flux tabulations in
Kimbell et al. (2001a), table 3). These estimates are provided in the Table 2-5 below. The flux-
based extrapolation results in a value similar to that obtained by applying the principle of ppm
37As discussed in the Appendix A.2, Schroeter et al. (2014) revised the dosimetry model of Kimbell et al.
(2001b; 2001) used for the flux estimates presented in Table 2-5, to include endogenous formaldehyde
production and to explicitly model formaldehyde pharmacokinetics in the respiratory mucosa. EPA
estimated the extent to which the results in Table 2-5 change if flux estimates from Schroeter et al. (2014) are
used. The average flux over nonsquamous regions of the rat nose is roughly one-third of that in the human
based on the dosimetry in Schroeter et al. (20141 in which endogenous formaldehyde is taken into account
compared to a ratio of roughly one-half based on the dosimetry in Kimbell et al. (2001b: 20011. Thus the POD
is not altered appreciably (changing only by roughly a factor of 1.4) if the revised dosimetry model by
Schroeter et al. (20141 is applied.
38The individual rat data generally allowed for assigning average severity scores for a given nasal level,
concentration, and time point. In several cases (as with the 24-month, Level 2), the nasal level was not clear
(i.e., the individual rat data could have come from Level 1, 2, or 3).
This document is a draft for review purposes only and does not constitute Agency policy.
2-18 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Toxicological Review of Formaldehyde—Inhalation
equivalence39 (see table footnote). The benchmark dose model fits and such details and further
elaboration of the human extrapolation are provided in Appendix B.1.3.
Table 2-5. Summary of derivation of POD for squamous metaplasia based on
observations in F344 rats (Kerns et al.. 1983)
Rat sagittal
section
BMR
Rat BMCLio
(mg/m3)
Flux3
(pmol/mm2-h)
Human
exposure cone
(mg/m3)
Adjusted13 human
exposure cone
(mg/m3)
Level 1
0.10
0.448
685
0.484
0.086°
Approximate average flux over nasal lining at this level corresponding to the BMCL
bAdjusted for continuous exposure, (6 hours/24 hours) x (5 days/7 days).
clf extrapolation is based on ppm equivalence instead, value increases by 1.14-fold.
Squamous metaplasia Wistar rats (Woutersen et al.. 19891
Woutersen et al. (1989) reported on the nasal histopathology for male Wistar rats exposed
to 0.1,1.2, and 12.1 mg/m3 for 28 months. Incidence of squamous metaplasia was reported by
concentration and cross-section level (i.e., Level 1-2, 3, 4, and 5-6), with Level 1 as the most
anterior region. The dose-response data for this effect is provided in Table 1-26 and can be
modeled.
Following the determination for squamous metaplasia in F344 rats (Kerns etal.. 1983). the
same minimal adversity was considered for this effect in Wistar rats and a BMR of 0.10 extra risk
was used. A dosimetry model for flux to the nasal lining of the Wistar rat is not available. EPA fU.S.
EPA. 20121 concluded that internal dose equivalency in the extrathoracic region for rats and
humans is in general achieved through similar external exposure concentrations (i.e., even for
highly soluble and reactive gases ppm equivalence is a more appropriate default method for
extrapolation than an approach based on adjustment by the ratio of surface area to minute volume).
This concept is supported by the analysis described above of data from the squamous metaplasia
occurring at Level 1 of the F344 rat nose. In that analysis, the extrapolation was based on site-
specific flux in the rat and human and differs from an extrapolation based on ppm equivalence by
only a factor of 1.14. Level 1 in that study was in the anterior portion of the nose, and the section
levels in the Woutersen et al. (1989) study (see Table 2-6) are even more anteriorly located in the
nose; therefore, there is even stronger support in this case for using ppm equivalence as the basis
for extrapolation across species. The benchmark dose model fits and such details are provided in
the appendix; the summary results are in Table 2-6.
39Also, see further discussion below in the analysis of squamous metaplasia in Wistar rats. "PPM equivalence"
refers to toxicological equivalence across species when exposures are expressed in "ppm" and are suffered
over equal durations expressed in units of the species lifetime. This originates from general allometric
principles, wherein tissue exposure is equivalent when scaled by BW3/4 while inhalation rates scale as BW3/4;
these factors cancel each other out when exposure is expressed in ppm.
This document is a draft for review purposes only and does not constitute Agency policy.
2-19 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
Table 2-6. Summary of derivation of PODs for squamous metaplasia based on
studies in F344 and Wistar rats (Woutersen et al.. 1989: Kerns et al.. 1983)
Endpoint and reference
Species/
sex
Model
BMR
Rat BMCa
(mg/m3)
Rat BMCLa
(mg/m3)
Human
PODaADJ
(mg/m3)
Squamous metaplasia
Kerns et al. (1983); Battelle
(1982)
F344 rat, M and
F
Log-probit
0.10b
0.576
0.448
0.086°
Squamous metaplasia
Woutersen et al. (1989)
Wistar rat, M
Log-logistic
0.10b
1.00
0.526
0.094d
aPODADj is the human equivalent of the rat BMCL duration adjusted (6/24) x (5/7) for continuous daily exposure.
bBMR = 0.10 because the severity of squamous metaplasia, as indicated by the severity scores, was considered
minimally adverse.
cHuman extrapolation was based on modeled estimates of regional formaldehyde tissue flux.
dHuman extrapolation was based on ppm equivalence derived from pharmacokinetic principles.
Conclusion
Confidence is high in the two studies used to derive PODs, as both studies were well
designed and executed with adequate reporting of data. Kerns et al. (1983: Battelle. 1982) was
conducted under Good Laboratory Practice conditions, and the inhalation exposure protocols in
both studies were adequately documented and well conducted. Confidence in the POD calculations
based on Wouterson et al. (19891 is medium, while confidence based on Kerns et al. (19831 is low.
Confidence is lower in the POD from Kerns et al. (1983) because the calculation involved an
extrapolation well below the tested formaldehyde concentrations, the BMCL was based on the 18-
month exposure although the response was greater in magnitude after 24 months, and the
incidence at Level 1 in the nose was modeled rather than the incidence at Level 2 where
concentrations were lower. Studies with various durations and in multiple species/strains have
consistently reported histopathological effects after inhaled formaldehyde exposure. Squamous
metaplasia was also observed in humans exposed to formaldehyde levels between 0.1 and
2.5 mg/m3 (see Section 1.2.4).
Reproductive and Developmental Toxicity
Female reproductive or developmental toxicity
Of the epidemiology studies that evaluated effects on fecundity or spontaneous abortion,
one study developed individual exposure estimates suitable for dose-response evaluation.
Taskinen et al. (1999) presented risk estimates for increased TTP for index pregnancies of women
in three exposure categories. The exposure assignments were made for jobs held beginning at least
6 months prior to the index pregnancy to evaluate TTP, the primary endpoint of interest. Taskinen
et al. (1999) calculated a fecundity density ratio for the three exposure categories based on 8-hour
This document is a draft for review purposes only and does not constitute Agency policy.
2-20 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
(time-weighted average) TWA (TWA8) formaldehyde concentrations composed of measured
concentrations associated with specific work tasks and reported time spent conducting those tasks
in the workplace. TTP was elevated in the high exposure group relative to the unexposed group.
EPA selected the middle TWA8 exposure level as a NOAEL.
The mean TWA concentrations for each exposure category needed to be adjusted for
background formaldehyde exposures experienced by the employees when they were not
conducting work tasks with identified formaldehyde exposure. Notably, the mean exposure (18
ppb TWA8) and lowest reported concentration measured in a work area (10 ppb) in the "low
exposed" category were less than the reported average ambient exposures for Finland (21.4 ppb)
flurvelin et al.. 20011. The investigators in Taskinen et al. (1999) appear to have assumed that,
while the women were away from their "exposed" work area, their exposure to formaldehyde was
zero, not accounting for background occupational exposures and ambient levels of formaldehyde.
Therefore, EPA recalculated the mean TWA8 concentrations. These calculations are presented in
Table 2-7.
Normally, exposures from occupational studies are adjusted to account for the daily
breathing volume appropriate to an environmental (versus occupational) setting and for exposure
every day of the year fU.S. EPA. 19941. However, with formaldehyde, there is potential for exposure
outside of work from in-home and environmental sources of formaldehyde. Therefore, the POD
represents exposure during an 8-hour workday.
This document is a draft for review purposes only and does not constitute Agency policy.
2-21 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-7. Adjusted time-weighted average formaldehyde exposures for
Taskinen et al. (1999)
(A) Proportion of work shift corresponding to the exposure group mean task-
level formaldehyde exposure (ppb) and the exposure group daily exposure
index (8-hour time-weighted average, TWA8). (B) Recalculation of daily
exposure index (TWA8) where background formaldehyde exposure is
estimated for work time spent on tasks considered unrelated to occupational
use of formaldehyde.
A
Exposure
group (n)
Reported mean
exposure
(TWA8)
Measured average
task-level
concentrations (ppb)
Estimate of work time for
formaldehyde-related tasks assuming
mean exposure levels
Mean
(ppb)
Range
Mean
Range
Percentage of
work time3
Hours per 8-hr
work shift
Low (119)
18
1-39
70
10-300
26%
2
Medium (77)
76
40-129
140
50-400
54%
4.3
High (39)
219
130-630
330
150-1,000
66%
5.3
Calculated as mean exposure (ppb, TWA8) divided by mean task-level exposures for the exposure group.
B
Exposure
group (n)
Estimate of formaldehyde
exposure during formaldehyde-
related
work tasks
Estimate of formaldehyde xposure
from background levels during the
work shift
Alternative
daily
eExposure
index (ppb,
TWA8)
Mean
(mg/m3)a
Percentage of
work time in
formaldehyde task
Background
formaldehyd
e (ppb)
Percentage of time
in tasks unrelated
to formaldehyde
Low (119)
0.086
26%
0.026
74%
0.042
Medium (77)
0.172
54%
0.026
46%
0.106
High (39)
0.406
66%
0.026
34%
0.278
Converted from units of ppb reported in paper.
1 Taskinen et al. (1999) also presented ORs for previous spontaneous abortion by multiple
2 exposure categories based on the work experience relevant to the index pregnancy. Although
3 spontaneous abortion risk was estimated only for events that occurred at the same workplace as
4 the index pregnancy, there is more uncertainty regarding the relevant time window of the exposure
5 characterization for this outcome. A POD for spontaneous abortion was not identified from this
6 data set or any of the other studies.
This document is a draft for review purposes only and does not constitute Agency policy.
2-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
Table 2-8. Summary of derivation of PODs for reproductive toxicity in females
Endpoint and
reference
Population
Observed effects by exposure level
POD (mg/m3)
Time-to-Pregnancy in Females
Occupational
prevalence
Taskinen et al. (1999)
Adult
women,
a? = 602
Time-to-Pregnancy by Formaldehyde Category
Fecundability density ratio (FDR)a
Mean TWA8 # FDRb 95% CI
(mg/m3)
NOAEL = 0.106
LOAEL = 0.278
Not exposed 367 1.00
0.042 119 1.09 0.86-1.37
0.106 77 0.96 0.72-1.26
0.278 39 0.64 0.43-0.92
FDR = ratio of average incidence densities of pregnancies in
exposed compared to employed unexposed women
Discrete proportional hazards regression; adjusted for
employment, smoking, alcohol consumption, irregular
menstrual cycles and # children
Comparison: index pregnancies that occurred when
participants were not employed in exposed workplace
Abbreviations: TWA8 = 8-hour time-weighted average; FDR = false discovery rate; NOAEL = no-observed-adverse-
effect level; LOAEL = lowest-observed-adverse-effect level.
Concentrations converted to mg/m3.
bTWA8 reported by authors was recalculated by EPA to account for background formaldehyde exposure while
working in "nonexposed" work areas.
Conclusion
A POD was identified based on the findings of Taskinen et al. (1999). The study was well-
conducted, a robust exposure assessment was used, and the data analysis was adjusted for other
risk factors and workplace exposures that could be associated with developmental toxicity.
However, because the study evaluated an occupational cohort, generalization to the entire general
population is more uncertain; EPA places medium confidence in the study. Confidence in the
candidate RfC derivation is low. Stratification by use of gloves (yes/no) indicated that women who
did not use gloves had a lower FDR. The stronger association among this group implies that dermal
absorption might have resulted in a greater response. Therefore, the level of certainty concerning
the value of the NOAEL associated solely with inhalation exposure is lessened.
Male reproductive toxicity
Two studies reporting effects on the male reproductive system in rats were considered to
be of sufficient quality for candidate reference value derivation (Ozen etal.. 2005: Ozenetal..
20021. Both studies exposed the animals to paraformaldehyde via inhalation; thus, the
interpretation of the results from these studies was not compromised by possible methanol
coexposure as with the other studies that evaluated male reproductive toxicity endpoints. In Ozen
et al. (2002). statistically significant and dose-dependent decreases in testis weight (relative to
This document is a draft for review purposes only and does not constitute Agency policy.
2-23 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Toxicological Review of Formaldehyde—Inhalation
body weight) were observed after 4 and 13 weeks of formaldehyde exposure. Although absolute
organ weights are preferred for this measure because testis weights are generally conserved when
body weight is decreased, mean body weights were also significantly decreased with exposure;
thus, this response pattern suggests that the organ weight decreases were likely due to a direct
effect on the testis (note: in this case, decreased relative testis weight is likely an underestimate of
the more appropriate decrease in absolute testis weight). Also of note, the effects increased with
duration of treatment (to 8 and 10% of control at 13 weeks) and were associated with alterations
in testicular zinc, copper, and iron levels (measured in the same study), thus, increasing confidence
in the study results. Although the decreased testis weight data at 4 weeks were successfully
modeled40 (see AppendixB.1.3) to derive aBMDLisD of 2.60 mg/m3, this endpointwas notusedto
calculate a cRfC because a subacute endpointwas not considered an appropriate basis for a chronic
RfC when data from longer-term exposure were available from the same study. For the decreased
testis weight at week 13 (Ozen etal.. 2002). a LOAEL of 12.3 mg/m3 was adjusted for continuous
exposure based upon the experimental paradigm to yield a PODadj of 2.93 mg/m3
(PODadj = 12.3 mg/m3 x 8 hr exposed per day/24 hrs per day x 5 days exposed per week/7 days per
week).
In Ozen et al. (2005), statistically significant dose-dependent decreases in serum
testosterone levels (6 to 9% decreases from control values) were observed following 91 days of
inhalation exposure. At the same exposure levels, significant decreases of 23 to 26% from control
were noted in mean seminiferous tubule diameters, an effect that could have been directly related
to testosterone decreases. For the decreased serum testosterone at day 91 (Ozen etal.. 2005). a
BMCLisd of 0.208 mg/m3 was calculated. This value was adjusted for continuous exposure based
upon the experimental paradigm to yield a PODadj of 0.050 mg/m3 (PODadj = 0.208 mg/m3 x 8 hr
exposed per day/24 hrs per day x 5 days exposed per week/7 days per week). EPA fU.S. EPA.
20121 indicates that for highly soluble and reactive gases that interact with tissue at the portal of
entry or for gases with systemic penetration ppm equivalence is likely to be the most appropriate
default method for extrapolation. Accordingly, the human equivalent concentration (HEC) was
derived by adjusting the POD derived for the rat by the duration adjustment of (6/24) x (5/7) for
continuous daily exposure.
Although the Ozen et al. (2005; 20021 studies evaluated a small number of animals (seven
and six male rats per group, respectively), the sample sizes were adequate to detect statistically
significant effects and did not demonstrate excessive variability.
40Using this BMR, a BMC of 3.81 mg/m3 was derived, and a PODadj of 0.619 mg/m3 was calculated, while this
is lower than the PODadj at 13 weeks of 2.93 mg/m3, the uncertainty in extrapolating the 13-week LOAEL to a
NOAEL would be expected to result in a comparably lower cRfC.
This document is a draft for review purposes only and does not constitute Agency policy.
2-24 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Toxicological Review of Formaldehyde—Inhalation
Conclusion
The confidence in the PODs derived from these studies is low, as the lowest formaldehyde
concentration tested in Ozen et al. (2002) was 12.2 mg/m3, and in Ozen et al. (2005) was
6.2 mg/m3. Both Ozen et al. (2005; 20021 studies were well conducted and interpreted with high
confidence that exposed the animals to paraformaldehyde via inhalation, and the observed
responses in each study were statistically significant, dose-dependent, and supported by the larger
body of animal study data for formaldehyde. Nevertheless, the magnitude of the testis weight
response in Ozen et al. (2002) was greater than that of the testosterone decreases observed in Ozen
et al. (2005), and a number of other rodent studies in the formaldehyde database demonstrated
similar testis (and epididymal) weight deficits, while specific evidence of treatment-related serum
testosterone decreases is quite limited. Uncertainties associated with the Ozen et al. (2002) study
include the small sample size (7 male rats per test group), lack of reported information on absolute
organ weight values, and no indication in the study report that exposure levels were confirmed
analytically. Additionally, the data could not successfully be modeled, and thus it was necessary to
use the study LOAEL to derive the RfC.
Table 2-9. Summary of derivation of PODs for reproductive toxicity in males
Endpoint and
reference
Species/
sex
Model
BMR
(mg/m3)
BMC
(mg/m3)
BMCL
(mg/m3)
PODadj
(mg/m3)
Ozen et al. (2002)
Decreased relative
testis weight (13 wk)
Rat/M
LOAEL
N/A
N/A
N/A
2.91
Ozen et al. (2005)
Decreased serum
testosterone (13 wk)
Rat/M
Exponential
(M2)
1SD
0.284
0.208
0.050
2.1.2. Derivation of Candidate Reference Concentrations
In this section, the PODs (either PODadj or PODhec) calculated in Section 2.1.1 were used to
derive candidate reference concentrations (cRfCs). These derivations are presented according to
the specific uncertainty factors (UFs) applied (to reduce redundancy for similar decisions across
health effects); the resultant cRfCs are then organized in a table and figure according to health
effect The text below explains the rationale for the UFs that are applied for each candidate RfC; the
implementation of those decisions is most easily seen by looking at Table 2-10 that immediately
follows the explanatory text.
Methods of Analysis
A series of five UFs were applied to each of the PODs developed for each endpoint/study,
specifically addressing the following areas of uncertainty: interspecies uncertainty (UFa) to account
for animal-to-human extrapolation, and consisting of equal parts representing toxicokinetic and
toxicodynamic differences; intraspecies uncertainty (UFh) to account for variation in susceptibility
This document is a draft for review purposes only and does not constitute Agency policy.
2-25 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Toxicological Review of Formaldehyde—Inhalation
across the human population (see Section 1.4.1), and the possibility that the available data may not
be representative of individuals who are most susceptible to the effect; LOAEL-to-NOAEL
uncertainty (UFl) to estimate an exposure level where effects are not expected when a POD is based
on a LOAEL; subchronic-to-chronic uncertainty (UFs) to account for the uncertainty in using
subchronic studies to make inferences about lifetime exposure, and to consider whether lifetime
exposure would have effects at lower levels (e.g., for studies other than subchronic studies); and
database uncertainty (UFd) to account for database deficiencies if an incomplete database raises
concern that further studies might identify a more sensitive effect, organ system, or lifestage. The
application of these UFs (i.e., assigning a value) was based on EPA's Review of the Reference Dose
and Reference Concentration Processes fU.S. EPA. 20021 (Section 4.4.5).
UFa interspecies uncertainty: animal-to-human variation
• For the 10 candidate RfCs derived from human epidemiology studies, an interspecies
uncertainty factor (UFa) was not applied.
• For the candidate RfCs for respiratory tract pathology (squamous metaplasia) and male
reproductive toxicity from rat data, an HEC was estimated using either dosimetry modeling
(Kerns etal.. 1983. metaplasia) or an assumption of ppm equivalence derived from
pharmacokinetic principles (Woutersenetal.. 1989. respiratory pathology): (Ozen etal..
2005: Ozen etal.. 2002. male reproductive toxicity).
o A factor of 3 was then applied to account for residual uncertainties in interspecies
extrapolation from the two candidate RfCs for respiratory pathology and the two
cRfCs for reproductive toxicity in males derived from rat studies.
UFh intraspecies uncertainty: Human variation
• As summarized in Section 1.4.1, populations or lifestages demonstrated to have potentially
increased susceptibility to the health effects of inhaled formaldehyde exposure include
pregnant women and children, persons with pre-existing health conditions (particularly
respiratory conditions such as asthma), and smokers. The UFH selections below explicitly
considered the ability of the selected studies to quantitatively address these potential
susceptibilities. This resulted in reduced UFhS for several endpoints with quantitative
analyses for several potentially susceptible groups, namely children, pregnant women, and
asthmatics. In addition, co-exposure to tobacco smoke was considered during the
evaluation of the individual studies. Section 1.4.1 discusses several other possible scenarios
that might result in increased susceptibility to inhaled formaldehyde but for which the
currently available information is inconclusive. While they may have an impact, these
potential susceptibility factors without specific experimental support were not considered
quantitatively.
• For four candidate RfCs derived from human epidemiology studies, an intraspecies
uncertainty factor (UFh) of 3 (i.e., 101/2) was used.
o For Venn et al. (2003), a UFh of 3 was used because the POD was based on the
degree of asthma control in children with asthma, a highly sensitive group. (A UFh of
This document is a draft for review purposes only and does not constitute Agency policy.
2-26 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Toxicological Review of Formaldehyde—Inhalation
1 was considered but not used because the number of individuals in the two higher
exposure groups was relatively low [n = 31-35), and likely did not characterize all
possible human variability.)
o For the POD for decreased peak expiratory flow rates (PEFRs) among children from
Krzyzanowski et al. (1990). a UFH of 3 was used with support from the model results
reported by the authors. The authors of this study evaluated a model of the
association of formaldehyde with PEFR that assessed differences between asthmatic
and nonasthmatic children. Multiple observations in the study indicate that a UFH of
3 applied to the endpoint can be expected to be protective of asthmatic children and
other susceptible individuals. EPA used the published regression coefficients from
the random effects model to calculate the predicted decrease in PEFR from the
baseline level (i.e., formaldehyde concentration equal to zero) for each group (for
details of the analysis see Appendix B.1.2). At the BMC (i.e., PEFR change of 10% in
the entire group), the asthmatic children experienced a decrement in PEFR that was
1.5-fold greater than that of the nonasthmatic children. Further, at the BMCL
(0.021 mg/m3), which was selected as the POD, the decrease in PEFR among
asthmatic children was 10.5% while that in nonasthmatic children was 7.2%. The
authors also stated that other characteristics that could affect variability such as
acute respiratory illness episodes during the observation period, environmental
tobacco smoke in the home, or socioeconomic status (education level of head of
household) did not increase sensitivity. All of these observations indicate that a UFh
of 3 can be expected to be protective of asthmatic children and other susceptible
individuals.
o For rhinoconjunctivitis and current asthma prevalence among children (school
exposure) from Annesi-Maesano et al. (2012.), a UFh of 3 was used for the POD.
Although Annesi-Maesano et al. (2012) did not select the study population based on
characteristics that increased susceptibility to formaldehyde's respiratory effects,
childhood is a susceptible lifestage for asthma and allergy, and the sample size of
6,600 children was large enough to have characterized an adequate spectrum of
human variability. However, a UFh of 1 was not used because susceptibility among
subsets of the study population was not specifically assessed.
o Matsunaga et al. (2008) was a study of pregnant women, a sensitive population for
eczema prevalence and an UFh of 3 was used for the POD. An UFh of 1 was not
applied because the study participants were adult women and no information was
available for other sensitive lifestages, including children, a subgroup with a higher
prevalence of eczema compared to adults.
• A UFh of 10 was used for the POD for current asthma prevalence in children (Krzyzanowski
et al.. 19901. the five cRfCs derived from epidemiology studies of adults, and the four cRfCs
derived from animal studies.
o For current asthma prevalence among children with residential exposure
fKrzvzanowski et al.. 19901. a UFh of 10 was used because susceptibility among
subsets of the population was not specifically assessed, and the precision of the
NOAEL was lower compared to Annesi-Maesano et al. (2012).
This document is a draft for review purposes only and does not constitute Agency policy.
2-27 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Toxicological Review of Formaldehyde—Inhalation
o For the cRfC for sensory irritation in adult (and teenage) populations (residential
exposures) in Hanrahan et al. (1984). a UFh of 10 was used. Although the study
population in Hanrahan et al. (1984) comprised randomly selected households in
mobile homes with individuals representing a range of age, sex, health behavior,
occupational status, and health status, the identified PODs were not based on
evaluation of differential susceptibility among subgroups with conditions or
characteristics that may contribute to variation in response.
o For the two sensory irritation PODs derived from short-term controlled human
exposure studies (Kulle etal.. 1987: Andersen and Molhave. 1983). as well as the
developmental toxicity POD based on reduced fecundity in reproductive-age women
in an occupational cohort studied by Taskinen et al. (1999). a factor of 10 was
applied to account for variation in the broader human population not represented
by occupationally exposed groups or participants in controlled human exposure
studies who met the eligibility criteria. Physiological differences that affect
sensitivity may become less of a concern for exposure to acute, high concentrations
of direct-acting irritants (such as formaldehyde) for the derivation of an acute RfC,
which could justify application of a lessor UFh as noted by the NRC (2001).
o For the four cRfCs based on studies in animals, a factor of 10 was applied to account
for the limited variability in susceptibility factors encompassed by these typical
studies of inbred laboratory animal populations.
UFt LOAEL uncertainty: LOAEL-to-NOAEL extrapolation
• A LOAEL-to-NOAEL UF was not applied to the five PODs based on a NOAEL.
• For the eight PODs derived from BMD modeling, a factor was not applied in keeping with
EPA guidelines (U.S. EPA. 2012). EPA selected a BMR of 10% to identify a POD based on
specific studies for several effects: sensory irritation, pulmonary function, and respiratory
pathology. A BMR of 5% was selected for the POD identified using the Venn et al. (2003)
study for effects on degree of asthma control. A BMR of 1 standard deviation from the
control mean was selected for male reproductive toxicity.
UFs subchronic uncertainty: extrapolation to chronic exposure
• Three experimental studies in animals evaluated exposures of durations less than a lifetime
fOzen etal.. 2005: Ozenetal.. 2002: Kerns etal.. 19831.
o A factor of 10 was applied to the two PODs for male reproductive toxicity to
approximate the potential effect of lifetime exposure, as these effects are not
necessarily dependent on a specific exposure window and they are expected to
worsen with continued exposure fOzen etal.. 2005: Ozen etal.. 20021.
o A factor of 3 was applied to the respiratory tract pathology POD from Kerns et al.
(1983) because it was based on 18-month exposure data from that rodent study in
lieu of the 24-month exposure data available in the same study. As discussed in
Section 1.2.4, there are data to suggest that exposure concentration would be more
important to the development of this lesion than duration, although the specifics of
this relationship have not been defined. However, the lesion incidences for this
This document is a draft for review purposes only and does not constitute Agency policy.
2-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Toxicological Review of Formaldehyde—Inhalation
particular study were substantially higher with the longer duration data
(i.e., 24-month versus 18-month), and thus a lower POD would be expected if the
24-month data could have been modeled. Thus, while use of the 18-month exposure
duration is expected to reduce the uncertainty associated with extrapolating to
lifetime exposure compared with a shorter duration such as 90 days, this reduction
in extrapolation to lifetime was considered incomplete (see text in 2.1.1) and a
factor of 3 was applied, consistent with EPA guidelines [a factor other than 10 may
be used, depending on the duration of the studies and the nature of the response
riJ.S. EPA. 2002. 1998.1994)].
• For one study in a human population, a UFS of 3 was applied to the POD. Matsunaga et al.
(2008) evaluated the occurrence of atopic eczema during the past 12 months in a group of
pregnant women and analyzed this outcome in relation to formaldehyde concentrations
measured in their homes, which is a less-than-lifetime window of vulnerability. However,
this outcome may have been pre-existing in a portion of the study sample and the window
of susceptibility may not have been sufficiently represented by the shorter exposure period
fCho etal.. 20101. Therefore, a UFs of 1 was not applied.
• For the remaining seven PODs derived from human studies, a UFs of 1 was applied. Three
studies were of sensory irritation, which is considered to be predominantly an acute
response (Kulle etal.. 1987: Hanrahan et al.. 1984: Andersen and Molhave. 1983). Notably,
the controlled exposure studies by Kulle et al. (1987) and Andersen and Molhave (1983)
demonstrate formaldehyde-induced sensory irritation after only brief periods of exposure;
thus, these studies would be relevant for estimating the sensory irritant effects resulting
from acute formaldehyde exposure. Three studies that were used for PODs for pulmonary
function, allergic conditions, current asthma, and asthma control evaluated these outcomes
in children and considered an appropriate window of exposure (Annesi-Maesano etal..
2012: Venn etal.. 2003: Krzvzanowski et al.. 1990). The study of Taskinen et al. (1999)
evaluated TTP, which in this review is categorized as a female reproductive or
developmental endpoint and the exposure window was considered to be appropriate.
Matsunaga et al. (2008) evaluated the occurrence of atopic eczema during the past
12 months in a group of pregnant women and analyzed this outcome in relation to
formaldehyde concentrations measured in their homes, which is a less-than-lifetime
window of vulnerability.
UFn database uncertainty
• A factor to account for database deficiencies was not applied to any of the PODs (i.e., UFd =
1). The formaldehyde database is not considered complete, as important questions remain
regarding the potential for formaldehyde inhalation exposure to cause reproductive and
developmental toxicity and nervous system effects (both of which demonstrate an
incomplete evidence base with methodological limitations). An incomplete database can
raise concern that further studies might identify a more sensitive effect, organ system, or
lifestage (U.S. EPA. 2002.1998.1996.1994.1991). However, given the breadth of the
literature on formaldehyde toxicity, and given the poor distribution of inhaled
formaldehyde to distal sites, an expectation that additional data are unlikely to reveal
systemic effects (i.e., by indirect MOAs) at lower exposure levels than those eliciting adverse
respiratory system changes seems unlikely; thus, this assessment uses a database
uncertainty factor (UFd) of 1.
This document is a draft for review purposes only and does not constitute Agency policy.
2-29 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 Summary of Candidate Reference Concentrations
2 Table 2-10 summarizes the application of UFs to each POD from the medium or high
3 confidence studies identified in Section 2.1.1 to derive one or more cRfC(s) in each health effect
4 system. Figure 2-1 presents graphically these cRfCs, UFs, and PODs, with each bar corresponding to
5 one data set described in Table 2-10.
Table 2-10. Health effects and corresponding derivation of candidate RfCs
Endpoint (reference; population)
PODa
POD
basis
UFa
UF
H
UFl
UFs
UF
D
UFcomposite
cRfC
(mg/m3)
Sensory Irritation
Eve irritation symptoms (Hanrahan et al.,
1984); adult M + F, n = 61, residential,
prevalence at POD 13%
0.087
BMCLin
1
10
1
1
1
10
0.009
Eve irritation symptoms (Kulle et al.,
1987); adult M + F, n = 10, controlled
exposure
0.42
BMC/2
1
10
1
1
1
10
0.04
Eve irritation symptoms (Andersen and
Molhave, 1983): adult M + F, n = 16,
controlled exposure
0.19
BMC/2
1
10
1
1
1
10
0.02
Pulmonary Function
Peak expiratory flow rate (Krzvzanowski
et al., 1990): Children M + F, n = 298,
residential
0.021
BMCL10
1
3
1
1
1
3
0.007
Allergy-related Conditions
Rhinoconjunctivitis prevalence
(Annesi-Maesano et al., 2012): children
M + F, n = 2,200 at POD, school-based
exposure
0.024
NOAEL
1
3
1
1
1
3
0.008
Atopic eczema prevalence (Matsunaga et
al., 2008): adult F (pregnant), n = 301 at
POD, personal monitor-based exposure
0.046
NOAEL
1
3
1
3
1
10
0.005
Asthma
Current asthma prevalence (Annesi-
Maesano et al., 2012): children M + F,
n = 2,200 at POD, school-based exposure
0.042
NOAEL
1
3
1
1
1
3
0.01
Current asthma prevalence
(Krzvzanowski et al., 1990): children
M + F, n = 24 at POD, residential
0.06
NOAEL
1
10
1
1
1
10
0.006
Asthma control (Venn et al., 2003):
children with asthma M + F, n = 35 at POD,
residential
0.013
BMCLs
1
3
1
1
1
3
0.004
This document is a draft for review purposes only and does not constitute Agency policy.
2-30 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Endpoint (reference; population)
PODa
POD
basis
UFa
UF
H
UFl
UFs
UF
D
UFcomposite
cRfC
(mg/m3)
Respiratory Tract Pathology
Sauamous metaplasia: (Kerns et al.,
1983: Battelle, 1982): adult F344 rat
M + F, 18-month exposure
0.086
BMCLio
3
10
1
3
1
100
0.0009
Sauamous metaplasia: (Woutersen et
al., 1989); adult Wistar rat, M + F,
28-month exposure
0.094
BMCLio
3
10
1
1
1
30
0.003
Female Reproductive and/or Developmental Toxicity
Delayed pregnancy (Taskinen et al.,
1999); pregnant F, n = 11 at POD
0.106
NOAEL
1
10
1
1
1
10
0.01
Male Reproductive Toxicity
Relative testis weight (Ozen et al., 2002);
adult rat, M, 13-week exposure
2.91
LOAEL
3
10
10
10
1
3,000
0.001
Serum testosterone (Ozen et al., 2005);
adult rat, M, 13-week exposure
0.05
BMCL1SD
3
10
1
10
1
300
0.0002
Abbreviations: cRfC = candidate reference concentration; UF = uncertainty factor; POD = point of departure;
BMC = benchmark concentration; BMCL = benchmark concentration, lower confidence bound; NOAEL = no-
observed-adverse-effect level; LOAEL= lowest-observed-adverse-effect level.
aPOD may be adjusted (e.g., to continuous exposure; to a human equivalent concentration) (see Section 2.1.1).
This document is a draft for review purposes only and does not constitute Agency policy.
2-31 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Eye irritation symptoms (humans)
Hanrahan etal, 1984
Eye irritation symptoms (humans)
Ku lie et al,, 1987
Eye irritation symptoms (humans)
Andersen, 1983
Peak expiratory flow rate (humans)
Krjyzanowski et al., 1990
Rhinoconjunctivitis prevalence (children)
Annesi-Maesana et al., 2012
Atopic excema prevalence (preganant women)
Matsunaga et al, 2008
Asthma control (children with asthma)
Venn et al, 2003
Current asthma prevalence (children)
Annesi-Maesano et al, 2012
Current asthma prevalence (children)
Krzyzanowski et al, 1990
Squamous metaplasia (male Wistar rats)
Woutersen et al, 1989
iS ™ Squamous metaplasia (F344 rats of both sexes)
Kerns et al, 1983
-5 io o
<
E
-C
<
E
QJ £ ^
U. C- "O
t= >c aj c
"O ^
Delayed pregnancy (pregnant women)
(Taskinen et al., 1999)
Relative testes weight (male Wistar rats)
(Ozen et al., 2002)
Serum testosterone (male Wistar rats)
(Ozen et al., 2005)
Composite UF
~ Candidate RfC
• POD(HEC)
.
A •
~ •
~ •
~ •
A •
A •
0,0001 0/1010 0.0100 o.iooo
log formaldehyde concentration (mg/ m3)
Figure 2-1. Candidate RfCs with corresponding POD and composite UF.
As the PODs reflect exact values, and the cRfCs are rounded to one significant figure, the UFcomposite
extrapolation between the two is not always exact.
2.1.3. Selection of Organ- or System-specific Reference Concentrations
1 This section distills the candidate values from Table 2-10 (i.e., the clusters of health effect-
2 specific cRfCs) into a single value representing a level without an appreciable risk of deleterious
3 effects on each particular organ or system during a lifetime. These organ- or system-specific RfCs
4 (osRfCs) may be useful for subsequent cumulative risk assessments that consider the combined
5 effect of multiple agents acting at a common site. In addition to the UFs applied, a set of three
6 confidence descriptors are included with each osRfC to reflect confidence in the health hazard, in
This document is a draft for review purposes only and does not constitute Agency policy.
2-32 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
the ability of the study to provide an accurate quantitative estimate, and in the completeness of the
database of studies available to evaluate each hazard.
Methods of Analysis
EPA selected the osRfC for each specific organ or system using rationales specific to the data
and studies for that health area, as described below. In general, studies of human populations with
exposures that best represent that of the general population, and human or animal studies that
evaluated long-term exposure were preferred, when available, unless a shorter window of
susceptibility was appropriate. In addition, cRfCs with lower composite UFs were generally
preferred. An osRfC was typically selected from cRfCs from higher confidence studies and higher
confidence in the POD estimate used to derive the cRfC. osRfCs were sometimes derived using a
method that combined two or more cRfCs.
Because the studies that are the basis of each of the osRfCs are interpreted to be
representative of the sets of studies available for each of the health outcomes evaluated, the overall
hazard descriptor for each database is presented. These descriptors represent the overall
confidence in the findings from the sets of individual studies, as compared to the confidence in the
individual medium or high confidence studies most amenable to estimating a cRfC.
An overall confidence level of high, medium, or low was also assigned to each osRfC based
on the reliability of the associated POD. Confidence in the POD included considerations of the
quality and variability of the exposure assessment in an epidemiology study or the exposure
protocols in an animal study. Moreover, higher confidence was placed in the osRfC when the POD
was identified close to the range of the observed data and the magnitude of exposure was relevant
to those experienced in the general U.S. population.
In addition, a descriptor was included to describe the coverage and quality of studies that
informed the hazard conclusion for that specific organ/system. The evidence base for different
health effects varies in size, coverage of critical endpoints, and quality of the studies; this
confidence level reflects database completeness for each of the organ systems.
Sensory Irritation
The osRfC for sensory irritation of 0.009 mg/m3 is based on the cRfC for eye irritation
derived using the results of Hanrahan et al. fl9841. As described previously, the study population
was more representative of the general population in terms of demographic characteristics and
exposure levels, and the cRfC reflects more certainty compared to the cRfCs calculated from the two
controlled human exposure studies. The POD is based on formaldehyde measurements in the
participants' homes (1-hour sampling period in two rooms). The confidence in the POD and cRfC
derivations is medium because of uncertainty related to the precise correspondence of the window
of exposure with the period symptoms were experienced.
This document is a draft for review purposes only and does not constitute Agency policy.
2-33 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
There is an extensive literature on this response to formaldehyde and the completeness of
the database is considered to be high. Because sensory irritation is an immediate response to
exposure, the osRfC is applicable to short-term as well as long-term exposure scenarios.
Pulmonary Function
Data from a study in a residential population exposed over multiple years was used to
calculate a cRfC for pulmonary function of 0.007 mg/m3 (Krzvzanowski et al.. 1990). This value
was chosen as the osRfC. The results from this study are generalizable to the general population,
and a robust exposure assessment based on 2-week average measurements in multiple rooms and
two different seasons. A strong exposure-response relationship with formaldehyde concentration
was observed by this study, which reduces concern that residual confounding by unmeasured
coexposures (smoking and NO2 were controlled for) strongly influenced the association. Hence,
confidence in the POD value is high. There is extensive information on this response to
formaldehyde from multiple studies in diverse exposure settings, and the completeness of the
database is considered to be high.
Allergy-related Conditions
The osRfC for allergy-related conditions is based on one study in children fAnnesi-Maesano
etal.. 20121 and one study in adults fMatsunaga et al.. 20081. Both PODs were based on NOAELs
and are interpreted with high confidence. In particular, the large study of children [n = 6,683) by
Annesi-Maesano et al. (2012) was able to address the variability in susceptibility that would be
anticipated within a population. No other pollutants (e.g., NOx, PM2.5, acetaldehyde, acrolein, ETS)
analyzed by this study were associated with rhinoconjunctivitis; thus confounding by coexposures
is unlikely. EPA selected an osRfC of 0.008 mg/m3, based on the overall greater strength of Annesi-
Maesano et al. (2012.). The completeness of the database relating formaldehyde exposure to
allergic sensitization is considered to be high, based on the variety of endpoints, populations, and
exposure scenarios considered in these studies.
Current Asthma/Degree of Asthma Control
There were three cRfCs developed for asthma based on the endpoints, current asthma, and
degree of asthma control fAnnesi-Maesano etal.. 2012: Venn etal.. 2003: Krzvzanowski et al..
19901. The POD based on Annesi-Maesano et al. (2012.) was derived from a NOAEL using a large
study with a relatively long exposure measurement period, supported by a collection of several
other smaller studies. Although the effect estimates derived by Venn et al. (2003) were less precise
because of relatively small group sizes, the POD derived from Venn et al. (2003) reflects the
response among a susceptible population, asthmatic children. Venn et al. (2003) used a strong
study design, observed an exposure-related trend in response and adjusted the statistical analyses
for key confounders, including other indoor exposures (e.g., visible mold, total VOCs, N02, cotinine
levels). To account for the different uncertainties in the PODs from the three studies, the median of
This document is a draft for review purposes only and does not constitute Agency policy.
2-34 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
the three PODs, 0.006 mg/m3, was selected for the osRfC. The overall confidence in the PODs is
medium. Two factors contribute to the determination that the completeness of the database
relating formaldehyde exposure to prevalence of current asthma is medium. One factor is the
relatively small number of studies examining asthma risk in relation to exposures between 0.05 and
0.1 mg/m3, and limitations of these studies (e.g., low statistical power, incomplete reporting of
study results and exposure measures). The second factor is the scarcity of data pertaining to
asthma control among people with asthma.
Respiratory Tract Pathology
The osRfC for respiratory tract pathology is based on squamous metaplasia observed in
anterior rodent nasal passages in two studies of long-term exposure. EPA could discern no
particular basis to select either the Woutersen et al. (1989) study or Kerns et al. (1983; Battelle.
1982) study over the other on grounds of confidence in the study methods, or known differences in
sensitivity between Wistar and F344 rats. In addition, the PODs were nearly identical and the cRfCs
are very similar for the two data sets [i.e., cRfCs of 0.0009 for Kerns et al. (1983) and 0.003 for
Woutersen et al. (1989), which are comparable given the limited precision of the calculations].
However, there was lower confidence in the derivation of the POD from Kerns et al. (1983). which
involved an extrapolation well below the tested formaldehyde concentrations. In addition, the cRfC
for Kerns et al. (1983) involved the application of a UF for exposure duration. While exposure
duration is important to the development of this lesion, such effects appear to be more dependent
on exposure concentration (see MOA discussion in Section 1.2.4). Thus, if a factor describing the
concentration-duration relationship41 were available for formaldehyde (and interpretable in the
context of metaplasia), a data-defined UF could have been applied. Considering these uncertainties
and the comparability of the cRfCs, to represent the results of both studies, the cRfC from
Woutersen et al. (1989) was used to derive an osRfC of 0.003 mg/m3 for the respiratory pathology
endpoint. Because the POD basis for this value is from Woutersen et al. (1989). the confidence in
the POD is considered medium. Completeness of the database for respiratory tract pathology is
considered high, based primarily on the numerous well-conducted, long-term studies in
experimental animals.
Female or Developmental Toxicity
Data from one study of women exposed to formaldehyde in the Finnish woodworking
industry are available to derive a cRfC for effects on delayed pregnancy (Taskinen et al.. 1999). This
value was chosen as the osRfC. Although TTP is a sensitive measure of effects on the reproductive
system, confidence in the POD is judged to be low because the outcome was evaluated in a healthy
41Studies of other irritants have, on average, identified a factor of ~1.8-1.9 for relationships between acute
exposure and mortality (i.e., the observed mortality is more attributable to concentration, by 1.8- to 1.9-fold,
than duration; see Section 1.2.4). A value for formaldehyde was not identified, nor were values for long-term
exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
2-35 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
working population with relatively high exposure, and thus required substantial extrapolation.
More complete assessments of developmental endpoints by epidemiology or toxicology studies
were not available. Thus, the completeness of the database is considered low. The relevant period
for exposure effects on TTP through unrecognized fetal losses or factors controlling the ability to
conceive could range from the weeks just prior and after conception, to the entire period of prior
exposure during the life of the individual because the mechanisms and events through which
formaldehyde may cause this outcome are not known.
Male Reproductive Toxicity
The cRfC derived from Ozen et al. (20021 was considered the stronger of the two candidates
for male reproductive toxicity, and thus was chosen to represent the osRfC. The magnitude of the
testes weight response in Ozen et al. (2002) was greater than that of the testosterone decreases
observed in Ozen et al. (2005). and a number of other rodent studies in the formaldehyde database
demonstrated similar testes (and epididymal) weight deficits, while specific evidence of treatment-
related serum testosterone decreases is quite limited. The LOAEL from Ozen et al. (2002) was used
to derive the POD. The confidence in the POD derived from its results is low, given that the lowest
formaldehyde concentration tested in this study was 12 mg/m3. Confidence in the database is also
considered low because while there are a number of published studies that evaluated reproductive
toxicity in males, the interpretation of study results is complicated by their methodological
limitations and exclusive use of formaldehyde concentrations above 6 mg/m3, and data are lacking
regarding functional endpoints.
2.1.4. Summary of Organ- or System-specific RfCs and RfC Selection
Table 2-11. Organ- or system-specific RfCs for formaldehyde inhalation
Health effect
Basis
reference(s) [species]
UFC
osRfC
(mg/m3)
Integrated
hazard
judgment
Confidence
in POD
estimate(s)3
Database
completeness13
Sensory
irritation
Hanrahan et al. (1984)
[human]
10
0.009
evidence
demonstrates
medium
high
Pulmonary
function
Krzvzanowski et al.
(1990) fhumanl
3
0.007
evidence
indicates
(likely)
high
high
Allergy-related
conditions
Annesi-Maesano et al.
(2012) [humanl
3
0.008
evidence
indicates
(likely)
high
high
This document is a draft for review purposes only and does not constitute Agency policy.
2-36 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Health effect
Basis
reference(s) [species]
UFC
osRfC
(mg/m3)
Integrated
hazard
judgment
Confidence
in POD
estimate(s)a
Database
completeness13
Asthma
(prevalence of
current
asthma/degree
of asthma
control)
Annesi-Maesano et al.
(2012); Venn et al.
(2003); Krzvzanowski et
al. (1990) [humanl
10°
0.006
evidence
indicates
(likely)
medium
medium
Respiratory
pathology
Woutersen et al. (1989);
Kerns et al. (1983) (rati
30°
0.003
evidence
demonstrates
medium
high
Female or
developmental
toxicity
Taskinen et al. (1999)
[human]
10
0.01
evidence
indicates
(likely)
low
low
Male
reproductive
toxicity
Ozen et al. (2002) [rati
3,000
0.001
evidence
indicates
(likely)
low
low
Abbreviations: osRfC = organ- or system-specific reference concentration; UF = uncertainty factor; POD = point of
departure.
aThis reflects a judgment regarding how well the study-specific data are able to estimate a no-effect- or minimal-
effect-level of response (e.g., a lower level of confidence would be applied to high-concentration studies that
required extrapolation far below the lowest tested concentration to estimate a POD). A descriptor of low means
that the POD derived is expected to be less accurate.
bAlthough no UFD was applied to any RfC, it is recognized that the evidence databases for the various health effects
are not equal. This descriptor was added to emphasize the health areas where additional research could reduce
existing uncertainties. A descriptor of low means the degree of certainty regarding the RfC is lower.
These two osRFCs are based on multiple studies and candidate values, sometimes with different UFcs applied.
The UFc values shown in this table and Figure 2-2 reflect the candidate values selected to represent each osRfC
[i.e., the UFC applied to the POD from Krzyzanowski et al. (Krzvzanowski et al., 1990) for asthma and from
Woutersen et al. (Woutersen et al., 1989) for respiratory pathology],
1 Selection of the Proposed Overall Reference Concentration
2 The following discussion outlines the selection of an overall RfC from among the osRfCs
3 presented in Table 2-11. The overall RfC was chosen to reflect an estimate of continuous inhalation
4 exposure to the human population (including sensitive subgroups) that is likely to be without an
5 appreciable risk of deleterious effects during a lifetime. The amount of risk between the RfC and
6 the PODs from which the RfC is derived is not known.
7 Methods of Analysis
8 Choice of the overall RfC involves consideration of both the level of certainty in the
9 estimated organ- or system-specific values, as well as the level of confidence in the observed
10 effect(s) (see Figure 2-2). An overall confidence level is assigned to the RfC to reflect an
11 interpretation regarding confidence in the collection of study/studies used to determine the
This document is a draft for review purposes only and does not constitute Agency policy.
2-37 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 hazard(s) and derive the RfC, the RfC calculation itself, as well as the overall completeness of the
2 database on the potential health effects of formaldehyde exposure.
3 Comparison
osRfCs and RfC (formaldehyde mg/ m3)
A
¦
Pulmonary function (Krzyzanowski etal., 1990)
CD
u
~
Allergy-related Conditions (Annesi-Maesano et al., 2012)
CD
"O
•
Sensory Irritation (Hanrahan et al., 1984)
4—
c
o
o
Respiratory Tract Pathology (Kerns et al., 1983; Woutersen et al., 1989)
u
I—
fl)
T
Asthma (Venn et al., 2003; Annesi-Maesano et al., 2012; Kryzanowski et al., 1990)
_c
txo
~
Female Reproductive and/ or Developmental Toxicity (Taskinen et al., 1999)
X
o
Male Reproductive Toxicity (Ozen et al., 2002)
Figure 2-2. Organ- or system-specific RfC scatterplot.
Organ/system RfCs (osRfCs) that are represented by larger shapes and that are closer to the top of the
graph are interpreted with higher confidence regarding the basis from which the value was derived (see
Table-2-11), and with less uncertainty (i.e., lower UFs were applied). Size of the shape represents
confidence in the study(ies) and health hazard (i.e., hazards with evidence demonstrates judgments are
larger than those with evidence indicates [likely] judgments), POD estimate(s) (for the purposes of this
graphic, confidence in the POD was given slightly greater weight than the others), and completeness of
the available evidence database for each health outcome: larger shapes indicate higher confidence; solid
shapes indicate studies in humans; hollow shapes indicate animal studies. For composite UF, if multiple
studies served as the basis for an osRfC, the composite UF associated with the candidate value selected to
This document is a draft for review purposes only and does not constitute Agency policy.
2-38 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
represent the osRfC was used (see Table 2-11). The dashed line represents the proposed overall RfC of
0.007 mg/m3; the circled osRfCs indicate the cluster of effects selected as the basis for this value.
Choice of the Proposed Overall RfC
An overall RfC for formaldehyde of 0.007 mg/m3 was selected. This value is within the
narrow range (0.006-0.009 mg/m3) of the group of respiratory system-related RfCs, which
together are interpreted with high confidence (sensory irritation, pulmonary function, allergy-
related conditions, and current asthma prevalence or degree of control) (see Figure 2-2). These
osRfCs are based on PODs that are the lowest of those identified in population studies for
formaldehyde hazards, and with the lowest composite uncertainty. The RfC for developmental
toxicity, although only slightly higher than the range observed for the selected respiratory effects, is
associated with less confidence in the POD. Likewise, the osRfCs for respiratory pathology and
male reproductive effects were associated with a larger degree of uncertainty, as reflected by their
position along the y-axis.
The RfC is an estimate of exposure that is likely to be without an appreciable risk of adverse
health effects over a lifetime. As illustrated in Figure 2-3, the selected RfC is at the upper end of the
range of outdoor formaldehyde levels recorded in some locations, and it would be expected that
levels in indoor air would exceed this concentration in most situations. However, it is important to
reiterate that this level is interpreted to be without appreciable risk. It is also important to note
that the RfC does not provide information about the magnitude of the risk of respiratory-related
effects that might occur at different concentrations above the RfC (e.g., at 0.02 or 0.03 mg/m3). As
illustrated in Figure 2-3, nearly all the study-specific findings of effects (e.g., LOAELs, BMCs) were
not observed until formaldehyde levels were in the upper end of the range of average indoor air
concentrations, with effects generally being observed at or above ~35-40 |ig/m3. One study that
contributed to the RfC derivation involved an analysis of the degree of asthma control in children
with current asthma, and the RfC is expected to apply to this susceptible subgroup in the
population. Although current asthma symptoms and allergic conditions were not observed in
studies of children with exposures less than the range of 0.02-0.05 mg/m3, at 0.021 mg/m3, a
10.5% decrease in PEFR among asthmatic children could be estimated (the regression model
included a term for asthma status), based on a model using results of Krzyzanowski et al. (1990)
(see Table 2-12). Thus, attributes that increase susceptibility in individuals are expected to play a
role in increasing the advent of adverse responses to formaldehyde levels above the RfC
(e.g., somewhere between 0.007 and 0.04 mg/m3).
This document is a draft for review purposes only and does not constitute Agency policy.
2-39 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Hanrahan (1984);
Burning eyes
Krzyzanowski (1990);
PEFR measures
Annesi-Maesano (2012);
Rhinoconjunctivitis
Krzyzanowski (1990);
asthma prevalence
Annesi-Maesano (2012);
asthma prevalence
Venn (2003);
asthma control
Woutersen (1989);
squamous metaplasia
Kerns (1983);
squamous metaplasia
Tasktnen (1999);
time-to- pregn a ncy
Ozen (2002);
testes weight
Outdoor
Indoor Air (normal conditions)
Indoor Air (atypical; e.g., some sealed mobile homes)
5H
E) [}--
a
E-
Sh
Eh r
S-
x] 4-
H-
- - -m
-Q-
¦O
4
O
o
¦c
G—
Effects continue
at higher levels
Effects not observed
until higher levels
15 20 40 60 80 1(1)0 200 4(5)0 600 800 10t)0
Formaldehyde Concentration [\xg/ m3)
2000 4000 60D0
: RfC; E cRfC; No appreciable risk O NOAEL, POD, N/C: No adverse change (study) 0 LOAEL or above: adverse effect in study (size = effect magnitude) — - UFs
<^> POD*,: Negligible risk (adjusted study data) ~ 8MCL POD: Negligible risk in study ¦ SMC: 5-10% change (study data) — POD adjustment • - - BMC to BMCL
Figure 2-3. Illustration of noncancer toxicity value estimations.
This figure provides a representation of the estimates from studies supporting the osRfCs, including a
summary of formaldehyde exposure data. Formaldehyde exposure estimates reflect approximates of the
range (boxes), medians or means (black vertical bars), and more commonly reported estimates
(gradations), based on the data discussed in Appendix A,1.2. Florizontal lines in the figure reflect the
extrapolation process for arriving at points of departure (PODs) and toxicity values (unfilled symbols) in
the context of the study-specific evidence for effects (filled symbols; effect magnitude estimated based on
study figures, tables, or reported regressions; see previous sections). Note: The x-axis is intentionally not
on a linear or log scale so as not to convey a false level of precision. Abbreviations: cRfC = candidate RfC;
N/LOAEL = no-/lowest-observed-adverse-effect level; UFs = uncertainty factors; BMCL = benchmark
concentration, lower confidence bound.
1 Although the RfC is designed to apply to exposures over a lifetime, the relevant window of
2 exposure for some of the effects observed in the contributing studies may be less than lifetime.
3 Sensory irritation is an immediate response to reactive compounds such as formaldehyde. The
4 relevant window of exposure for effects on asthma outcomes also is less than lifetime, although the
5 time frame for the control of asthma symptoms (i.e., a few weeks] is expected to be different than
6 that for the prevalence of current asthma symptoms or a decrease in pulmonary function (i.e., the
7 last 12 months). In addition, the relevant window of exposure for the osRfC for female
8 reproductive or developmental outcomes is from conception to the end of the pregnancy.
This document is a draft for review purposes only and does not constitute Agency policy,
2-40 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
The exposure paradigm used by controlled human exposure studies evaluates an immediate
response (i.e., on the order of minutes to hours) to acute formaldehyde exposure and it may be
appropriate to use the results from these studies to derive an acute RfC. The evidence base for
formaldehyde included results from controlled human exposure studies of formaldehyde inhalation
and sensory irritation endpoints, pulmonary function response among healthy or asthmatic
individuals and hyperbronchoreactivity among allergic asthmatics in response to an allergen
challenge. Two cRfCs for sensory irritation were derived from short-term controlled human
exposure studies (Kulle etal.. 1987: Andersen and Molhave. 19831. Generally, pulmonary function
measures were not changed by acute exposure in several controlled human exposure studies of
healthy or asthmatic volunteers, although small decrements were observed after longer exercise
components (15 minutes). Two additional studies did not observe pulmonary function changes in
response to acute formaldehyde inhalation, but did observe an early phase increase in airway
reactivity in response to an allergen challenge indicating a potential exacerbation effect by
formaldehyde inhalation on asthma symptoms (Ezratty etal.. 2007: Cassetetal.. 2006). Cassetetal.
(2006) observed a statistically significant response at lower dust mite amounts with formaldehyde
levels of 0.092 mg/m3 and mouth breathing only, while Ezratty et al. (2007) observed an increase in
a reactivity index in response to a grass allergen challenge (p = 0.06) using a higher formaldehyde
concentration (0.5 mg/m3).
Table 2-12. Proposed overall RfC for formaldehyde inhalation
Health effect(s) basis
RfC (mg/m3)
Overall confidence
Sensory irritation, pulmonary function, allergy-related
conditions, and degree of asthma control/prevalence of
current asthma in human studies3
0.007
High
aBased on the following studies: Annesi-Maesano et al. (2012); Matsunaga et al. (2008); Venn et al. (2003);
Krzyzanowski et al. (1990); Hanrahan et al. (1984).
Uncertainties in the Derivation of the Proposed Overall Reference Concentration
Research in experimental animals with regard to two health effects, respiratory tract
pathology and male reproductive toxicity, indicates that the proposed overall RfC may not be
protective against these hazards. Based on these effects, an alternative RfC of 0.001-0.003 mg/m3
would be derived. However, the confidence in this alternative RfC would be low because
uncertainties regarding these osRfCs are greater and the extrapolation from concentrations at
which effects were observed in these experimental animal studies was much larger.
The potential for formaldehyde to adversely affect the nervous system, female and male
reproduction, as well as development are not well studied, and the systemic effects of inhaled
formaldehyde are not well understood. The potential for a localized, immunosuppressive effect in
the respiratory tract, with implications for infectious diseases spread through inhalation, is another
This document is a draft for review purposes only and does not constitute Agency policy.
2-41 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
understudied issue. Additional research in these areas would increase understanding of the
spectrum of effects seen with formaldehyde exposure, formaldehyde concentrations that pose a
hazard for specific types of effects, and MOAs for these effects.
Confidence Statement Regarding the Proposed Overall Reference Concentration
An overall confidence level of high, medium, or low is assigned to reflect the level of
confidence in the study(ies) and hazard(s) used to derive the RfC, the overall database, and the RfC
itself, as described in Section 4.3.9.2 of EPA's Methods for Derivation of Inhalation Reference
Concentrations and Application of Inhalation Dosimetry fU.S. EPA. 19941. Overall confidence in the
RfC is high; the RfC is based on a spectrum of adverse effects reported in multiple well-conducted
studies involving different populations of exposed humans. Most of the study populations were
exposed to formaldehyde levels in a residential or school setting, and some of the studies focused
on sensitive individuals. An extensive literature database supports the hazard conclusions.
2.1.5. Previous IRIS Assessment: Reference Value
An inhalation RfC for formaldehyde has not previously been derived. In 1990, an oral
reference dose (RfD) of 0.2 mg/kg-day was developed. This value was based on reduced weight
gain and histopathology (primarily of the gastrointestinal system) in Wistar rats during a 2-year
bioassay in which formaldehyde was administered in the drinking water fTil etal.. 19891. A UFc of
100 was applied to the NOAEL to account for inter- and intraspecies differences. This RfD was
interpreted with medium confidence, based on high confidence in the principal study and medium
confidence in the database.
2.2. INHALATION UNIT RISK ESTIMATE FOR CANCER
Unit risk estimates for cancer were derived from different data sets available from both
epidemiological and experimental animal studies. Unit risk estimates could be derived for two of
three cancer types for which the evidence supporting a human health hazard was sufficiently
strong (evidence demonstrates): nasal cancers (i.e., nasopharyngeal cancer in human studies;
nasal SCC in experimental animal studies) and myeloid leukemia. While the evidence supporting a
human health hazard from sinonasal cancer from studies in occupational cohorts and experimental
animals also was sufficiently strong to support the derivation of unit risk estimates, no adequate
exposure-response data sets were available to derive unit risk estimates.
Section 2.2.1 focuses on the derivation of unit risk estimates for nasal cancers with an
examination of sources of uncertainty, and Section 2.2.2 discusses the derivation of unit risk
estimates for myeloid leukemia and examines sources of uncertainty. Section 2.2.3 presents a
summary of the unit risk estimates obtained from the different data sets and selection of the
preferred estimate. Section 2.2.4 describes adjustments to the preferred estimate for assumed
early-life susceptibility for cancers with a mutagenic MOA. In addition, an approach to bound low-
This document is a draft for review purposes only and does not constitute Agency policy.
2-42 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
dose cancer risks from formaldehyde exposure using DNA adduct concentrations in nasal
epithelium and bone marrow from animal experiments and U.S. cancer incidence statistics (a
"bottom-up" approach) is summarized to provide some perspective on the uncertainty in
extrapolating from high-dose animal toxicology or human occupational data (Section 2.2.5). Finally,
Section 2.2.6 provides a summary of the final adjusted unit risk estimate and uncertainties.
EPA concluded that the evidence for increased risks of NPC and myeloid leukemia was
sufficiently strong to support the derivation of unit risk estimates. A judgment that the evidence
demonstrates that formaldehyde inhalation causes NPC cancer was based on robust human
evidence of increased risk in groups exposed to occupational formaldehyde levels, and robust
animal evidence of nasal cancers in rats and mice that exhibits steeply increasing incidence at high
formaldehyde levels. Strong mechanistic support is provided across species (primarily rats, but
also mice, monkeys, and humans), including genotoxicity (sometimes at low formaldehyde levels in
rats), epithelial damage or remodeling, and cellular proliferation that are consistent with neoplastic
development in a regional, temporal, and dose-related fashion. A judgment that the evidence
demonstrates that formaldehyde inhalation also causes myeloid leukemia was based on robust
human evidence of increased risk in groups exposed to occupational formaldehyde levels.
Supporting mechanistic evidence consistent with leukemia development is provided across
numerous studies of peripheral blood isolated from exposed workers, including evidence of
mutagenicity and other genotoxic damage in lymphocytes and myeloid progenitors, and
perturbations to immune cell populations. The animal evidence is inadequate and, although notable
uncertainties remain (see Section 1.3.3), the findings to date suggest either a lack of concordance
across species or a lack of long-term studies in animal models that characterize the disease process
in humans for leukemia. Leukemia was not increased in two well-conducted chronic bioassays of
rats or mice, and the available animal data provide weak mechanistic support for LHP cancers. No
MOA has been established to explain how formaldehyde inhalation can cause myeloid leukemia
without systemic distribution (inhaled formaldehyde does not appear to be distributed to an
appreciable extent beyond the URT to distal tissues).
EPA's standard approach for deriving an inhalation unit risk (IUR) estimate using results
from epidemiology studies involves using a regression coefficient that describes the relationship
between increases in cancer risk and increases in cumulative exposure, and estimating a (upper-
bound) lifetime extra risk-per-unit exposure concentration through a life-table analysis.
Cumulative exposure, which incorporates both average concentration and the duration of time over
which exposure occurred, is generally the preferred metric for quantitative estimates of lifetime
risk from environmental exposure to carcinogens, and thus cumulative exposure was chosen as the
exposure metric for calculations in this assessment. The "true" exposure metric best describing the
biologically relevant delivered dose of formaldehyde is unknown. Few epidemiological studies
presented dose-response analyses based on cumulative measures of formaldehyde concentration
that could support the derivation of unit risk estimates. A unit risk estimate was derived based on
This document is a draft for review purposes only and does not constitute Agency policy.
2-43 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
dose-response modeling of mortality and cumulative formaldehyde exposure for nasopharyngeal
cancer (NPC) in a human occupational cohort. Upper respiratory tract (URT) cancer risk was also
extrapolated from the incidence of nasal squamous cell carcinoma (SCC) in experimental studies on
F344 rats. Results from several approaches used to model these data are evaluated and compared,
including biologically based dose-response (BBDR) modeling, statistical time-to-tumor modeling,
and statistical benchmark dose modeling using data on DNA-protein crosslinks (DPXs) and
formaldehyde flux as dose measures. Additional analyses and comparisons were conducted based
on mechanistic hypotheses, including derivation of RfCs based solely on estimates of cell
proliferation (i.e., one contributing MOAto formaldehyde exposure-induced nasal cancers; see MOA
discussion in Section 1.2.5), and assessing impacts of endogenous formaldehyde concentration on
dosimetric estimates.
Results from the follow-up of mortality from LHP cancer in the same occupational cohort
were used to derive a unit risk estimate for myeloid leukemia. In this study (see Section 2.2.2),
however, there is no apparent association between myeloid leukemia mortality and cumulative
exposure. A clearer association is observed with peak exposure, though it is not statistically
significant in the latest follow-up (in an earlier 1994 follow-up of that study, myeloid leukemia
mortality was statistically significantly associated with peak exposure; see Section 1.3.3). Although
multiple approaches for deriving a unit risk estimate for myeloid leukemia were explored, EPA did
not develop an approach based on the peak exposure metric because EPA deemed the uncertainty
associated with the peak exposure metric and the difficulties in translating risk from peak exposure
to risk from chronic low-level exposure to be prohibitive.
Instead, EPA explored alternative approaches for deriving a unit risk estimate for myeloid
leukemia based on cumulative exposure. Although an association between myeloid leukemia and
cumulative formaldehyde exposure was not apparent in the key exposure-response study, there are
indications that this may, at least in part, reflect a misclassification of myeloid leukemia deaths on
death certificates. Percy et al. (1990: 1981) have reported that myeloid leukemia is often recorded
as "leukemia" (not otherwise specified) on death certificates and hence underreported]. The
approach described in the Toxicological Review is to estimate a unit risk for myeloid leukemia
using the regression coefficient for myeloid and other/unspecified leukemias combined; this cancer
grouping had a stronger association with cumulative exposure in the key exposure-response study
than did myeloid leukemia alone and it captures the unclassified myeloid leukemias with the least
inclusion of nonmyeloid leukemias. A comparison of the use of the different cancer groupings
shows that they yield similar unit risk estimates (see Table 2-34).
An IUR estimate for cancer was estimated based on the unit risk estimate for NPC using the
results from the occupational study for cumulative exposure. While the estimates for NPC and
myeloid leukemia could be combined to derive an inhalation unit risk (IUR) for formaldehyde, there
is considerable scientific uncertainty in the data used to estimate a unit risk for myeloid leukemia.
Therefore, the unit risk estimate for myeloid leukemia is not included in the IUR calculation in this
This document is a draft for review purposes only and does not constitute Agency policy.
2-44 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
draft assessment A charge question will be provided for peer review asking for advice regarding
the development of a unit risk estimate for myeloid leukemia and how, if at all, the unit risk
estimate might inform the quantification of risk for cancer. Section 2.2.6 provides a summary and
conclusions from the cancer exposure-response modeling, presenting the preferred unit risk
estimate based on the extra risk of NPC associated with lifetime exposure to formaldehyde,
calculated from the epidemiology studies. Because the MOA for formaldehyde's effect on nasal
cancer risk was concluded to involve mutagenicity, the unit risk estimate was adjusted for assumed
increased early-life susceptibility.
2.2.1. Unit Risk Estimates for Nasal Cancer
Derivation of Cancer Unit Risk Estimates Based on Human Data
Choice of epidemiology study
While several studies of cancer in workers exposed to formaldehyde evaluated
exposure-response relationships, only a few reported risk estimates in relation to changes in
formaldehyde concentration rather than duration of exposure, TSFE, probability of exposure, or
exposure intensity score, measures which are not generally adequate for the derivation of cancer
unit risk estimates. Beane Freeman et al. (2013) presented results of the follow-up of the large
National Cancer Institute (NCI) retrospective cohort mortality study [originally described by Blair
et al. (1986)] of workers at 10 U.S. plants producing or using formaldehyde. Marsh et al. (2007b:
2002) focused on pharyngeal cancer and, in particular, NPC mortality in sequential follow-up
analyses of the Marsh et al. (1996) cohort study, which examined one of the 10 plants studied by
NCI.
The quantitative analyses presented in this Toxicological Review are based on the NPC
fBeane Freeman etal.. 20131 results from the latest follow-up of the NCI cohort of industrial
workers exposed to formaldehyde. The NCI cohort study is the largest of the three independent
industrial worker cohort studies [the other two being Meyers et al. (2013) and Coggon et al.
(2014)] and, more importantly, it is the only one with sufficient individual exposure data for
exposure-response modeling. In addition, the NCI study is the only one of the three studies that
used internal comparisons rather than standardized mortality ratios (SMRs), thus minimizing the
potential impact of the healthy worker effect by addressing unmeasured confounding, which can
bias effect estimates.
The NCI cohort consists of 25,619 workers (88% male) employed in any of the 10 plants
prior to 1966. The most recent follow-up, based on 998,239 person-years of observation (through
2004) reported a total of 13,951 deaths (Beane Freeman etal.. 2013). Beane Freeman et al. (2013)
analyzed 10 deaths from NPC as well as deaths from other solid tumors. Some demographic details
about the cohort are summarized in Table 2-13.
This document is a draft for review purposes only and does not constitute Agency policy.
2-45 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
Table 2-13. Demographic details about the NCI industrial workers cohort3
Factor
Quantity
Number of workers
25,619
Person-years of follow-up
998,239
Percentage male
87.8%
Percentage white
92.7%
Percentage hourly workers
78.5%
Median duration of follow-up
42 yrs
Median (range) length of employment
2.6 yrs (<1 day-47.7 yrs)
Number of deaths
13,951
Number of cancer deaths
3,703
aFollow-up through December 31, 2004 (Beane Freeman et al., 2013).
A detailed exposure assessment was conducted for each worker in the NCI cohort, based on
exposure estimates for different jobs held and tasks performed (Stewart etal.. 1986). Exposure
estimates were made using several different metrics—peak exposure, average intensity, cumulative
exposure, and duration of exposure. Respirator use and exposures to formaldehyde-containing
particulates and other chemicals were also considered. Some exposure details about the cohort are
summarized in Table 2-14.
Table 2-14. Exposure details about the NCI industrial workers cohort3
Factor
Quantity
Percentage workers never exposed
10.5%
Median (range) formaldehyde TWA8 for exposed workers
0.3 (0.01-4.3) ppm
Median (range) cumulative exposure for exposed workers
0.6 (0.0-107.4) ppm x yrs
Number of workers who experienced peak exposures
>4 ppm
6,255
aFollow-up through December 31, 2004 (Beane Freeman et al., 2013).
For NPC, RR estimates were increased in the highest exposure category for each of the
exposure metrics fBeane Freeman et al.. 20131. although these increases were generally not
statistically significant, given the small number of deaths involved. A statistically significant trend
was observed only for the peak exposure metric and only among the exposed person-years [two of
the 10 deaths from this rare cancer were in the unexposed workers fBeane Freeman etal.. 20131],
The (log-linear) trend for cumulative exposure (as a continuous variable) approached statistical
significance (p = 0.06 among exposed person-years only and p = 0.07 among all person-years).
This document is a draft for review purposes only and does not constitute Agency policy.
2-46 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
With respect to the other solid cancers of interest, while Beane Freeman et al. (20131 reported
results for cancers of the nose and nasal sinus, there were just five deaths for that endpoint Marsh
et al. (2002) reported some exposure-response results from their case-control study of all
pharyngeal cancers in one of the industrial plants studied by the NCI, but they did not observe
positive trends for cumulative or average exposure.
Exposure assessment and choice of exposure metric from the National Cancer Institute cohort
A detailed exposure assessment was conducted for the NCI cohort of industrial workers
exposed to formaldehyde, and quantitative exposure estimates were generated for each worker
(Stewart etal.. 19861. Formaldehyde exposure estimates, including TWA8 concentration and
categories of peak concentrations, were derived for each job, work area, and calendar year
combination. A peak was defined as a short-duration exposure (typically <15 minutes) above the
TWA, which could be related to either routine or nonroutine tasks (Beane Freeman et al.. 2009).
The frequency of peak exposures was also estimated, but these estimates were based on
assumptions made by the assessors rather than direct measures or observations, making this
metric highly uncertain. Cumulative exposures (in ppm x years) were estimated by multiplying the
time a worker spent in a specific job by the TWA exposure for that job and summing over all the
jobs held by the worker. Duration was the total time spent in jobs with formaldehyde exposure,
and average intensity was the ratio of cumulative exposure to duration. Formaldehyde exposures
after 1980 were not taken into account in the follow-up study, but this was considered to have a
generally minimal impact on the results (Beane Freeman et al.. 2013).
Some of the strongest exposure-response relationships in the NCI cohort studies (Beane
Freeman et al.. 2013) (e.g., for NPC) were observed for the peak exposure metric. It is not clear how
to extrapolate RR estimates based on peak exposure estimates to meaningful estimates of lifetime
extra risk of cancer from continuous exposure to low environmental levels. In addition, peak
exposure level is a more subjective measure than the other metrics, it is not based on formaldehyde
concentration measurements, and it is a categorical rather than continuous measure. Individual
workers were assigned to peak exposure level categories based on their work histories and a
matrix of job-, work area-, and calendar time-specific TWA8 formaldehyde measurements.
Historical sampling records and sampling conducted by the investigators contributed to the
development of this matrix. If a short-term (<15 minute) excursion above the TWA8 concentration
for a job was observed, or expected based on industrial hygiene expertise, then that job was
assigned to a peak exposure category: none, >0 to <0.5 ppm (>0 to 0.62 mg/m3), 0.5 to <2.0 ppm
(0.62 to <2.46 mg/m3), 2.0 to <4.0 ppm (2.46 to 4.92 mg/m3), or >4.0 ppm (>4.92 mg/m3).
Individual workers may have experienced these peak concentrations rarely, intermittently, or
routinely, and in jobs they held for a long time or only briefly. At a given time point, a worker's
peak exposure estimate is the highest peak exposure category ever attained by the worker. As
such, this exposure metric is not interpretable in terms of a lifetime exposure risk.
This document is a draft for review purposes only and does not constitute Agency policy.
2-47 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
Similarly, the average exposure metric is not a measure of long-term exposure for chronic
effects because it does not account for duration of exposure (e.g., exposure to a given exposure level
for 1 year conveys the same amount of risk as exposure to the same level for 70 years). Likewise,
duration of exposure does not account for the level of exposure and is not a useful metric for the
calculation of risk estimates as a function of exposure level, such as the cancer unit risk estimate.
Cumulative exposure, which incorporates both average concentration and the duration of
time over which exposure occurred, is generally the preferred metric for quantitative risk
assessment of lifetime risk from environmental exposure to carcinogens, and cumulative exposure
was chosen as the exposure metric for the risk estimate calculations for the cancer endpoints in this
assessment. The "true" exposure metric best describing the biologically relevant delivered dose of
formaldehyde is unknown.
Dose-response modeling of the National Cancer Institute cohort
The results of the internal analyses (i.e., comparing exposed workers to an internal referent
group of other workers in the cohort) of Beane Freeman et al. (2013) for NPC using the cumulative
exposure metric, with comparisons to the results using the peak exposure and average intensity
metrics, are presented in Table 2-15. The relative risks (RRs; in this case, rate ratios) were
estimated using log-linear Poisson regression models stratified by calendar year, age (in 5-year
intervals), sex, and race (black/white) and adjusted for pay category (salary/wage). As shown by
Callas et al. (1998). when age is well characterized and adjusted for, as it was in the Beane Freeman
et al. (2013) study, the Poisson regression and Cox proportional hazards models yield essentially
the same results. Beane Freeman et al. (2013) used a 15-year lag interval in estimating exposures
to account for a latency period for the development of solid cancers, including NPCs. Lag intervals
of 2-20 years were evaluated, and changing the interval had little impact on the RR estimates; thus,
the interval of 15 years that was used in the previous follow-up analyses fHauptmann et al.. 20041
was retained. For all cancer types, the NCI investigators used the low-exposure category as the
reference category to "minimize the impact of any unmeasured confounding variables since
nonexposed workers may differ from exposed workers with respect to socioeconomic
characteristics" (Hauptmann et al.. 2004). Table 2-15 also presents the p-value for the (log-linear)
trend of risk changing with exposure level for all workers and for only those workers exposed to
formaldehyde. The strongest exposure-response relationship for NPC is observed for the peak
exposure metric among exposed workers.
The log-linear trend analyses for the cumulative exposure metric approach statistical
significance (p-trend = 0.07 for all person-years; p-trend = 0.06 for exposed person-years only).
The fact that the two-sided p-values are not strictly <0.05 is not critical here, given that the hazard
for NPC was established a priori in Chapter 1. The nonexposed person-years were included in the
primary cancer risk analyses to use all the available exposure-response data. Furthermore, the data
were stratified by pay category, which provided at least partial adjustment for socioeconomic
characteristics. Final results for the exposed person-years only are also presented for comparison.
This document is a draft for review purposes only and does not constitute Agency policy.
2-48 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 The log-linear trend tests conducted by Beane Freeman et al. (20131 used exposure as a
2 continuous variable (except for peak exposure, for which categorical ranks were used) (general
3 model form: RR = ePx, where (3 represents the regression coefficient and X is exposure). Dr. Beane
4 Freeman provided EPA with the p estimates (and their standard errors) from the trend tests for
5 NPC and the cumulative exposure metric for all person-years and for exposed person-years only
6 (personal communication to EPA from Laura Beane Freeman, NCI, to Jennifer Jinot, EPA, February
7 22,2013). These estimates are presented in Table 2-16.
Table-2-15. Relative risk estimates for mortality from nasopharyngeal
malignancies (ICD-8 code 147) by level of formaldehyde exposure for different
exposure metrics
Rate ratio (number of deaths)
p-Trend
>111 person-years3
Exposed person-
years'5
Peak exposure (ppm)
0
>0 to <2.0C
2.0 to <4.0
>4.0
4.39 (2)
1.0(1)
-(0)
7.66 (7)
0.10
0.005
Average intensity (ppm)
0
>0 to <0.5C
0.5 to <1.0
>1.0
6.79 (2)
1.0(1)
2.44 (1)
11.54 (6)
0.16
0.09
Cumulative exposure (ppm x years)
0
>0 to <1.5C
1.5 to <5.5
>5.5
1.87 (2)
1.0 (4)
0.86(1)
2.94(3)
0.07
0.06
aLikelihood ratio test (1 degree of freedom) of zero slope for formaldehyde exposure (continuous variable, except
for peak exposure metric) among all (nonexposed and exposed) person-years.
bLikelihood ratio test (1 degree of freedom) of zero slope for formaldehyde exposure (continuous variable, except
for peak exposure metric) among exposed person-years only.
Reference category for all categories with the same exposure metric.
8 Source: Beane Freeman et al. (2013).
Table-2-16. Regression coefficients from NCI log-linear trend test models for
NPC mortality from cumulative exposure to formaldehyde3
Person-years
P (per ppm x year)
Standard error (per ppm x year)
All
0.04311
0.01865
Exposed only
0.0439
0.01852
aModels stratified by calendar year, age, sex, and race and adjusted for pay category; cumulative exposures
calculated using a 15-year lag interval.
9 Source: Personal communication to EPA from Laura Beane Freeman to Jennifer Jinot (February 22, 2013).
This document is a draft for review purposes only and does not constitute Agency policy.
2-49 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Toxicological Review of Formaldehyde—Inhalation
Prediction of lifetime extra risk of nasopharyngeal cancer mortality
The regression coefficients presented in Table 2-16 were used to predict the extra risk of
NPC mortality from environmental exposure to formaldehyde.
Extra risk = (Rx - R0)(1 - R0), (2-1)
where Rx is the lifetime risk in the exposed population and R0 is the lifetime risk in an unexposed
population (i.e., the background risk). Extra risk estimates were calculated using the p regression
coefficients and a life-table program that accounts for competing causes of death.42 U.S. age-specific
2010 all-cause mortality rates and 2000-201043 NPC (ICD-10 C11.0-C11.9) mortality rates for all
race and sex groups combined44 were used to specify the all-cause and cause-specific background
mortality rates in the life-table program. Risks were computed up to age 85 because cause-specific
mortality (and incidence) rates for ages above 85 years are less reliable. Conversions between
occupational formaldehyde exposures and continuous environmental exposures were made to
account for differences in the number of days exposed per year (240 versus 365) and in the amount
of air inhaled per day (10 versus 20 m3). An adjustment was also made for the 15-year lag period.
The reported standard errors for the regression coefficients were used to compute the one-sided
95% upper confidence limits (UCLs) for the extra risks based on a normal approximation.
Point estimates and one-sided 95% UCLs for the extra risk of NPC mortality associated with
varying levels of continuous exposure to formaldehyde are presented in Table 2-17. The model
predicts extra risk estimates that are fairly linear for exposures below about 0.001 to 0.01 ppm but
not for exposures above 0.01 ppm.
Table 2-17. Extra risk estimates for nasopharyngeal cancer mortality from
various levels of continuous exposure to formaldehyde
Exposure concentration (ppm)
Extra risk
95% UCL on extra risk
0.0001
1.24 x 10"7
2.12 x 10"7
0.001
1.24 x 10"6
2.13 x 10"6
0.01
1.28 x 10"5
2.25 x 10"5
0.1
1.79 x 10"4
4.12 x 10"4
1
2.67 x 10"1
8.74 x 10"1
10
9.83 x 10"1
9.87 x 10"1
42This program is an adaptation of the approach that was previously used in BEIRIV, "Health Risks of Radon
and Other Internally Deposited Alpha Emitters." National Academy Press, Washington, DC, 1988, pp. 131-
134. A spreadsheet illustrating the life table used for the extra risk calculation for the derivation of the
LECooos for NPC incidence is presented in Appendix B.2.1.
43Typically, 5-year ranges are used as the basis for population cause-specific disease and mortality rates; a
larger range is used here to get better stability in the rates because NPC is a rare cancer.
44Centers for Disease Control and Prevention, National Center for Health Statistics. Underlying Cause of Death
on CDC WONDER Online Database. Accessed at http://wonder.cdc.gov/ucd-icdlO.html on September 19, 2013.
This document is a draft for review purposes only and does not constitute Agency policy.
2-50 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
Consistent with EPA's Guidelines for Carcinogen Risk Assessment fU.S. EPA. 2005a). the same
data and methodology were also used to estimate the exposure level (effective concentration [ECx])
and the associated (one-sided) 95% lower confidence limit (LECx) corresponding to an extra risk of
0.05% (x = 0.0005). Although EPA guidelines emphasize the use of exposure levels associated with
a 10% extra risk level for the POD for low-dose extrapolation, that would not be appropriate in this
instance. A 10% extra risk level is very high for responses generally observed in epidemiology
studies; thus, a 1% extra risk level is typically used for epidemiological data to avoid upward
extrapolation. However, NPC has a very low background mortality rate (e.g., lifetime background
risk is about 0.00019); therefore, even a 1% extra risk (i.e., 0.01) would be a large increase relative
to the background risk. This is consistent with the fact that, even with a large cohort followed for a
long time, only 10 NPC deaths were observed in the NCI follow-up through 2004.45 The 1% level of
risk is associated with RR estimates that are substantially higher than those observed in the
epidemiology study. Based on the life-table program, the RR estimate for an extra risk of 1% for
NPC mortality is 53, an upward extrapolation. Even 0.1% yields an RR estimate on the high end of
the observable range of the epidemiology study (RR = 6.2). A 0.05% extra risk level yields an RR
estimate of 3.6, which better corresponds to the RRs in the range of the data. Thus, 0.05% extra
risk was selected for determination of the POD, and, consistent with EPA's Guidelines for Carcinogen
Risk Assessment fU.S. EPA. 2005al. the LEC value corresponding to that risk level was used as the
POD.
Because formaldehyde is a mutagenic carcinogen and the weight of evidence supports the
conclusion that formaldehyde carcinogenicity for URT cancers can be attributed, at least in part, to
a mutagenic MOA (see Section 1.2.5), a linear low-dose extrapolation was performed in accordance
with EPA's carcinogen risk assessment guidelines fU.S. EPA. 2005al. The ECooos, LECooos, and IUR
estimates for NPC mortality are presented in Table 2-18.
Table 2-18. ECooos, LECooos, and inhalation unit risk estimates for
nasopharyngeal cancer mortality from formaldehyde exposure based on the
Beane Freeman et al. (2013) log-linear trend analyses for cumulative
exposure
ECooos
LECooos
Unit risk3
Unit risk
Person-years
(PPm)
(PPm)
(per ppm)
(per mg/m3)
All
0.191
0.112
4.5 x 10"3
3.7 x 10"3
Exposed only
0.187
0.111
4.5 x 10"3
3.7 x 10"3
aUnit risk = 0.0005/LEC0oos-
45Eleven NPCs were reported on death certificates and included in NCI's SMR analyses, but one of these cases
was apparently misclassified on the death certificate, so only 10 cases were used to estimate the RRs in the
internal comparison analyses (Beane Freeman et al.. 2013).
This document is a draft for review purposes only and does not constitute Agency policy.
2-51 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Prediction of lifetime extra risk of nasopharyngeal cancer incidence
EPA cancer risk estimates are typically derived to represent a plausible upper bound on
increased risk of cancer incidence, as from experimental animal incidence data. Cancer data from
epidemiology studies are more often mortality data, as is the case in the NCI study. For cancers
with low survival rates, mortality-based estimates are reasonable approximations of cancer
incidence risk. However, for NPC, the survival rate is substantial (51% at 5 years in the 1990s in
the United States, according to Lee and Ko (20051 and incidence-based risks are preferred because
EPA is concerned with cancer occurrence, not just cancer mortality.
Therefore, an additional calculation was done using the same regression coefficients
provided by Dr. Beane Freeman (see Table 2-16) but with age-specific NPC incidence rates from
NCI's Surveillance, Epidemiology, and End Results (SEER) Program in place of the NPC mortality
rates in the life-table program. SEER collects cancer incidence data from a variety of geographical
areas in the United States. The incidence data used here are from SEER-18, a registry covering
about 27.8% of the U.S. population, which was the most current SEER registry at the time this
analysis was done. SEER-18 age-specific background incidence rates for NPC (ICD-10 C11.0-C11.9)
for 2000-2010 were obtained from the SEER public-use database (www.seer.cancer.gov) using
NCI's SEER*Stat software (www.seer.cancer.gov/seerstat). The incidence-based calculation relies
on the reasonable assumptions that NPC incidence and mortality have the same exposure-response
relationship for formaldehyde exposure and that the incidence data are for first occurrences of NPC
or that relapses provide a negligible contribution. The calculation, as presented in the life-table
spreadsheet in Appendix B.2.1, also takes advantage of the fact that NPC incidence rates are
negligible compared with the all-cause mortality rates and thus no special adjustment to the
population at risk to account for live individuals who have been diagnosed with NPC is necessary.
The resulting ECooos, LECooos, and IUR estimates for NPC incidence are presented in
Table 2-19. The unit risk estimate for cancer incidence is two-fold higher than the corresponding
mortality-based estimate, for all person-years, reflecting the high survival rates for NPC.
Table 2-19. ECooos, LECooos, and inhalation unit risk estimates for
nasopharyngeal cancer incidence from formaldehyde exposure based on the
Beane Freeman et al. (2013) log-linear trend analyses for cumulative
exposure
Person-years
ECooos (ppm)
LECooos (ppm)
Unit risk3 (per ppm)
Unit risk (per mg/m3)
All
0.0942
0.0550
9.1 x 10"3
7.4 x 10"3
Exposed only
0.0925
0.0546
9.2 x 10"3
7.5 x 10"3
aUnit risk = 0.0005/LECooos.
The preferred estimate for the inhalation cancer unit risk for NPC is the estimate of
9.1 x 10"3 per ppm (7.4 x 103 per mg/m3) derived using incidence rates for the cause-specific
This document is a draft for review purposes only and does not constitute Agency policy.
2-52 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
background rates, for all person-years. The results from the exposed person-years are essentially
identical.
Because NPC is a rare cancer in the United States, with a relatively low number of cases
occurring per year, a rough calculation was done to ensure that the unit risk estimate derived for
NPC incidence is not implausible in comparison to actual case numbers. For example, assuming an
average constant lifetime formaldehyde exposure level of 5 ppb for the U.S. population, the IUR
estimate for NPC equates to a lifetime extra risk estimate of 4.6 x 10-5. Assuming an average
lifetime of 75 years (this is not EPA's default average lifetime of 70 years but rather a value more
representative of actual demographic data) and a U.S. population of 300,000,000, this lifetime extra
risk estimate suggests a crude upper-bound estimate of 180 incident cases of NPC attributable to
formaldehyde exposure per year. Alternatively, assuming an average constant lifetime
formaldehyde exposure level of 20 ppb, the calculation suggests a crude upper-bound estimate of
730 incident cases of NPC per year. Both upper-bound estimates, using different assumed lifetime
exposure levels, are well below the estimated 2,300 total incident NPC cases per year calculated
from the SEER NPC incidence rate of 0.75/100,000.46 47
Derivation of Cancer Unit Risk Estimates Based on Squamous Cell Carcinoma in the
Respiratory Tract Using Animal Data
In this section, dose-response analyses of cancer risk based on nasal tumor data from
laboratory bioassays using F344 rats are presented. The Agency takes the position that human
data, if adequate data are available, provide a more appropriate basis for estimating human cancer
risk than do rodent data (U.S. EPA. 2005a). primarily because uncertainties in extrapolating
quantitative risks from rodents to humans are avoided; therefore, the epidemiology-derived
estimates presented in the previous section are the preferred unit risk estimates for nasal cancers.
Nonetheless, it is useful to compare human health risk estimates from available
epidemiology data with estimates extrapolated from animal studies. Furthermore, a large body of
mechanistic data on cell replication, DPX and DNA monoadduct formation, and dosimetry modeling
of formaldehyde flux to local tissue exist for formaldehyde that can potentially inform the shape of
the dose-response curve. This information, as well as data on the incidence of hyperplasia,
facilitates the interpretation and extrapolation of nasal squamous cell carcinoma (SCC) incidence
46The crude NPC (ICD-10 C11.0-C11.9) incidence rate from 2000-2010 SEER-18 data was obtained from the
SEER public-use database (www.seer.cancer.gov) using NCI's SEER*Stat software
(www.seer.cancer.gov/seerstat). This value is similar to a published NPC incidence rate for the United States
of 0.7/100,000 person-years (Lee and Ko. 20051. The age-adjusted NPC incidence rate from SEER was also
0.75/100,000.
47With the application of age-dependent adjustment factors (see Section 2.2.4), the lifetime unit risk estimate
for NPC would increase by a factor of 1.42, and the crude upper-bound estimates of the incident cases per
year attributable to formaldehyde exposure would similarly increase by a factor of 1.42. The resulting
adjusted upper-bound estimates of 260 and 1,030 for 5- and 20-ppb exposure levels, respectively, are still
well below the estimated total number of 2,300 incident cases per year in the United States.
This document is a draft for review purposes only and does not constitute Agency policy.
2-53 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
results from F344 rat bioassays within the context of formaldehyde's reactivity and MOAs. The
estimates derived from animal data incorporate this information into the modeling.
This section describes the data and modeling approaches available; presents PODs from the
considered models at benchmark response rates in the range of the available data; presents results
from a biologically based model for extrapolation to human exposure scenarios; evaluates
uncertainties in the dose-response models and discusses the use of any of the models for
extrapolating below the POD, including implications for low-dose risk; and presents candidate IURs
and RfCs derivable from the modeled PODs.
Multiple approaches, including conventional multistage Weibull time-to-tumor modeling
and a biologically based clonal expansion model of cancer, are used to model the incidence of nasal
SCC in F344 rats. Use of the biologically based modeling allowed the use of various data, including
mechanistic information, in an integrated manner. For a given benchmark response level, PODs
and their corresponding HECs are remarkably similar across multiple models and dose metrics
(formaldehyde inhaled exposure concentrations, formaldehyde inhaled flux to tissue, DNA-protein
crosslink [DPX] concentrations).
A clonal expansion model fConollv et al.. 20041. as well as possible variations of this model,
developed for extrapolation of the rat nasal cancer risk to human exposure scenarios are carefully
evaluated. Predictions using these models for humans are found to be not robust at any exposure
concentration. Furthermore, a key model inference in Conolly et al. (2003). (Conollv et al.. 2004)
that formaldehyde-induced mutagenicity, modeled as proportional to DPX concentration, is not
relevant to its carcinogenicity is found to be extremely uncertain. Accordingly, the clonal expansion
modeling of the rat data is employed to derive multiple PODs and corresponding HECs but it is not
used for extrapolating to human exposure scenarios. Unit risks derived by straight line
extrapolation from a POD as well as candidate RfCs (cRfCs) derived from benchmark modeling of
data on cell proliferation and basal hyperplasia observed in F344 rats and Wistar rats, respectively,
are also presented, with the cRfC interpreted as the concentration below which nasal cancers
arising from increased cell proliferation due to formaldehyde-induced cytotoxicity are unlikely to
occur. The assessment presents arguments from the literature that protection against these
putative precursor events is sufficient to prevent a cancer response. However, the proven
genotoxicity and mutagenicity of the chemical and the observation of human cytogenetic effects in
human occupational exposures provide strong support for preferring the linear extrapolation from
the POD to the origin. An additional contribution to uncertainty in the low dose-response comes
from the potential for endogenous formaldehyde levels in respiratory tissue to reduce the uptake of
the inhaled gas at low doses, as demonstrated in modeling efforts by Schroeter et al. (2014) and
Campbell Jr et al. (2020) (discussed further in the context of toxicokinetics in Section 1.1.3).
Candidate unit risks based on a point of departure at the 0.005 extra risk are found to be
comparable to that derived from analysis of the NCI occupational epidemiology data on
nasopharyngeal cancers (NPCs).
This document is a draft for review purposes only and does not constitute Agency policy.
2-54 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Toxicological Review of Formaldehyde—Inhalation
Animal nasal tumor incidence
An increased incidence of nasal SCC was seen in two long-term bioassays using F344 rats
fBermudez. 2004: Monticello etal.. 1996: Kerns etal.. 19831. As shown in Table 2-20 and in
Chapter 1 (SCC incidence in rats exposed to formaldehyde in long-term studies), the incidences are
similar between the two studies even though they were conducted 13 years apart, and the
incidence is similar between males and females in Kerns et al. (1983) (there were only male rats in
Monticello et al. (1996)). Therefore, it appears appropriate to combine these studies for greater
power in dose-response analysis. No other long-term studies have been conducted on F344 rats
(see Table 2-20). These two studies, when combined (see Table 2-20), provide a well-defined
spread of concentrations with at least 90 animals per group from each study, whereas other chronic
rat studies were either single concentration or had a relatively small number of animals per group.
Thus, although other studies in laboratory animals exist, the two studies (Monticello etal.. 1996:
Kerns etal.. 1983) combined provide the most robust data for analyses. The table shows only the
grouped incidence; however, the individual animal incidence data were available and used in the
assessment.
Table -2-20. F344 rat nasal cancer data
Formaldehyde exposure levels
Incidence of SCC tumors
References
0, 0.7, 2.0, 6.01, 9.93, and 14.96 ppm (0,
0.86, 2.5, 7.4,12.2, and 18.4 mg/m3)
0/341, 0/107, 0/353, 3/343,
22/103,162/386 (time-to-tumor
characteristics shown in Fig. 1)
Bermudez (2004); Monticello et
al. (1996); Kerns etal. (1983)
(combined bioassays)
Mechanistic information
In addition, two types of mechanistic data are used in the dose-response modeling. These
include site-specific measurements of DNA-protein crosslinks (DPX) formed by formaldehyde in the
F344 rat and rhesus monkey, and site-specific measurements of changes in cell labeling induced by
inhalation exposure to formaldehyde in the F344 rat.
Formaldehyde is a direct-acting mutagen, and DPXs serve as a surrogate marker for the
tissue dose associated with this mutagenic potential. The modeling uses physiologically based
pharmacokinetic (PBPK) models that have been developed based on the DPX data in Table 2-21 to
calculate DPX levels as a function of local formaldehyde flux, and to predict DPX levels in the
human. As discussed in Chapter 1 and Appendix C.3, exposure to inhaled formaldehyde induces
dose-related changes in rates of cell division as inferred from cell labeling studies in the
formaldehyde-exposed F344 rat In turn, regenerative increases in cell proliferation increase the
probability of errors in DNA replication.
This document is a draft for review purposes only and does not constitute Agency policy.
2-55 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Toxicological Review of Formaldehyde—Inhalation
Computational fluid dynamic modeling
The ability to use mechanistic data in dose-response modeling is further facilitated by the
availability of computational fluid dynamic (CFD) modeling of airflow in the rat, monkey, and
human respiratory passages. The CFD modeling is useful on multiple accounts.
Formaldehyde-induced squamous cell carcinomas (SCCs) and other lesions that occur in the
rat and monkey nasal passages and in the monkey LRT are seen to be distributed in localized
patterns that differ across species. The anatomy of the respiratory tract, in particular the nasal
passages, and the pattern of airflow, show large regional differences across species
(see Appendix A). On this basis, several authors have argued that regional dose would be the main
determinant of interspecies differences in tumor incidence for a highly reactive and water soluble
chemical such as formaldehyde fBogdanffv etal.. 1999: Monticello etal.. 1996: Monticello and
Morgan. 1994: Morgan et al.. 1991). thus motivating the use of modeling local formaldehyde flux in
the nasal region of each species.
Kimbell et al. (19931. Kepler et al. (19981. and Subramaniam et al. (19981 developed
anatomically realistic finite-element representations of the noses of F344 rats, rhesus monkeys, and
humans, and used them in physical and computational models fKimbell et al.. 2001a: Kimbell etal..
2001b). The nasal dosimetry modeling by Kimbell et al. (2001a; 2001b) was revised by Schroeter
et al. (2014) to include air:tissue partitioning and air and tissue phase diffusivity; production of
endogenous formaldehyde in the respiratory mucosa as a zero-order process; clearance of
formaldehyde in the form of a saturable pathway for enzymatic metabolism, a first-order pathway
for nonenzymatic reactions, and a pseudo first-order pathway to include its binding to DNA to form
DPX.
This assessment uses dosimetry derived from fKimbell etal.. 2001a: Kimbell etal.. 2001bl
when extrapolating risk-related dose from the rat to the human (discussed in detail in
Appendix B.2), and estimates the impact on the point of departure of using an alternate dosimetry
model developed by Schroeter et al. (2014). Furthermore, DPX levels and cell labeling data are
characterized as a function of regional formaldehyde flux to further inform the interpretation of
cancer incidence results. These are tabulated in Table 2-21 and used in the different dose-response
methods presented in this assessment (see Appendix B.2 for additional details).
This document is a draft for review purposes only and does not constitute Agency policy.
2-56 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table-2-21. Dosimetric and mechanistic information supporting dose-
response assessment based on rat nasal tumors
Data or information
Formaldehyde
exposure
Notes
Study references
FA dosimetry in
anatomically realistic
representations of the
F344 rat and human
nasal passages and in
an idealized
representation of the
human lower
respiratory tract
Inhaled concentrations of
0,0.7, 2, 6,10, or 15 ppm
(0, 0.9, 2.5, 7.4, 12.3, or
18.5 mg/m3) at various
steady-state inhalation
rates
Fluid dynamic models of
local FA flux to tissue
Subramaniam et al. (2008);
Kimbell et al. (2001a); Kimbell et
al. (2001b); Overton et al.
(2001); Kimbell etal. (1997b);
Kimbell et al. (1993). See
Appendix B.2
DPXa in F344 rat (2
studies) and in rhesus
monkey
Rat study 1(1989): 0.3,
0.7, 2.0, 6.0, 10.0 ppm
(0.4,0.9, 2.5, 7.4, 12.3
mg/m3) for 6 hours. Rat
study 2 (1994): 0.7, 2.0,
6.0, 15.0 ppm (0.9, 2.5,
7.4,18.5 mg/m3) for
3 hours. DPX measured
over whole nose in study
1, and over two regions
("low" and "high" tumor
sites) in study 2. Monkey
study: 0.7, 2.0, 6.0 ppm
(0.9, 2.5, 7.4 mg/m3) for
6 hours
DPX lesions observed at all
exposure concentrations
(0.3 ppm-15 ppm/0.37
mg/m3-18.5 mg/m3). DPX
tracheal and lung lesions
in monkeys at 6.0 ppm
(7.4 mg/m3). Data used in
PBPK model for FA and
DPX
Conollv et al. (2000); Casanova
et al. (1994); Casanova et al.
(1991); Casanova et al. (1989)
Cell labeling indexb;
F344 rats. Labeling
study with two phases
0,0.7, 2, 6,10, or 15 ppm
(0, 0.9, 2.5, 7.4, 12.3, or
18.5 mg/m3). Phase 1
exposure duration: 1, 4,
and 9 days and 6 weeks.
Phase 2 exposure
duration: 13, 26, 52, and
78 weeks
Phase 1 used injection
labeling with a 2-hour
pulse of tritiated
thymidine; Phase 2 used
osmotic mini pump
tritiated thymidine
labeling with a 120-hour
release time
Phase 1: Monticello et al.
(1991). Phase 2: Monticello et
al. (1996); Data analyzed in
Appendix B.
Abbreviations: FA = formaldehyde exposure; DPX = DNA-protein crosslink; PBPK = physiologically based
pharmacokinetic.
aNote that these studies do not present DPX measurements on control animals.
bThese data were used as input for modeling the nasal tumors observed in F344 rats and for benchmark modeling
of cell proliferation as a precursor response by authors from the same laboratory as this study (Conollv et al.,
2003; Schlosser et al., 2003). Many other studies (see below on "uncertainty in dose-response estimates" and
Appendix B.2) inform the effect of formaldehyde on cell proliferation and are brought to bear upon the discussion
of uncertainties in modeling the dose-response. However, Monticello et al. (1996) is the only study that followed
long-term exposure to formaldehyde.
This document is a draft for review purposes only and does not constitute Agency policy.
2-57 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
Dose-response modeling of nasal SCC incidence in the rat
EPA used multiple dose-response models of the observed tumor incidence in F344 rats
fMonticello etal.. 1996: Kerns etal.. 19831. These are briefly described below. Dose metrics
derived from PBPK modeling or CFD modeling are included in all these approaches.
Time-to-tumor modeling without using mechanistic data
Because higher exposures were associated with both earlier tumor occurrence and
increased mortality in the rats, methods that can reflect the influence of competing risks and
intercurrent mortality on site-specific tumor incidence rates are preferred. For this reason, EPA
used the multistage Weibull time-to-tumor model fPortier and Bailer. 1989: Krewski et al.. 19831.
which (a) models the replicate animal data, (b) includes the exact time of observation of the tumors
and therefore gives appropriate weight to the amount of time each animal was on study without a
tumor, and (c) acknowledges earlier tumor incidence with increasing dose level.
The model has the following form: P(d) = 1 - exp[-(q0 + qid + q2d2 + ... + qkdk) x (t - to)z],
where p(d) represents the lifetime risk (probability) of cancer at dose d (i.e., human equivalent
exposure in this case); parameters > 0, for i = 0,1,..., k; t is the time at which the tumor was
observed; and z is a parameter estimated in fitting the model, which characterizes the change in
response with age. The parameter to represents the time between when a potentially fatal tumor
becomes observable and when it causes death.
A further consideration is the distinction between tumor types as being either fatal or
incidental to adjust for competing risks. Incidental tumors are those tumors thought not to have
caused the death of an animal (such as those observed during interim or terminal sacrifices), while
fatal tumors are thought to have resulted in animal death. For these data, nasal tumors observed
with early deaths other than interim sacrifices were considered to be fatal.
The data used in this analysis were obtained from the appendix in Conolly et al. (2003.) and
combined the nasal squamous carcinoma data of Kerns et al. (1983) and Monticello et al. (1996)
along with results from an additional 94 animals not previously examined in the Monticello et al.
(1996) study. The dose-response analyses, estimation of parameters, plots of model fits for fatal
and incidental tumors, and model selection criteria are detailed in Appendix B.2.2.
Modeling of the grouped incidence data
This assessment also presents results from statistical modeling of the same data by
Schlosser et al. (2003) in Table 2-22. These authors did not carry out a time-to-tumor analysis of
the individual animal data but applied a Kaplan-Meier survival adjustment of the grouped incidence
data. The best fit in Schlosser et al. (2003) was obtained with the polynomial and Weibull models
for the tumor incidence data with a nonzero intercept (threshold) on the dose axis. See Schlosser et
al. (2003) for further details.
This document is a draft for review purposes only and does not constitute Agency policy.
2-58 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Biologically based dose-response modeling
A biologically based time-to-tumor dose-response (BBDR) model for modeling the
formaldehyde-induced rat nasal tumors is available fConollv etal.. 2003: CUT. 19991. This model
consists of interfacing dosimetry models for formaldehyde and formaldehyde-induced DPX in the
rat nasal passages (Kimbell et al.. 2001a: Kimbell et al.. 2001b: Conollv etal.. 2000) with a two-stage
clonal expansion (TSCE) model for predicting the probability of occurrence of nasal SCC (Conollv et
al.. 20031. The term "BBDR modeling" is used here to collectively refer to various toxicokinetic and
toxicodynamic dose-response modeling components.
The cancer modeling in the BBDR approach is based on an approximation of the
Moolgavkar, Venzon, and Knudson stochastic TSCE model of cancer fMoolgavkar et al.. 1988:
Moolgavkar and Knudson. 1981: Moolgavkar and Venzon. 19791. which accounts for growth of a
pool of normal cells, mutation of normal cells to initiated cells, clonal expansion of initiated cells,
and mutation of initiated cells to fully malignant cells. The molecular dose associated with
formaldehyde's direct mutagenic action was represented in this approach by the DPX formed by
formaldehyde. Formaldehyde-induced changes in cell replication and DPX concentrations, derived
from the data indicated in Table 2-21, were considered a function of local formaldehyde flux
(pmol/mm2-hour) to each region of nasal tissue as predicted by CFD modeling on anatomically
accurate representations of the nasal passages of a single F344 rat (see Appendix B.2). The TSCE
model was calibrated with the observed tumor incidence data to estimate various unknown
parameters as indicated below. DPX tissue concentrations in Conolly et al. (20031 were calculated
using a physiologically based pharmacokinetic model developed in Conolly et al. (20001.
Conolly et al. (20031 characterized the dose-response for cell replication rates as a J-shaped
curve, indicating that cell division rates decreased below that determined for the unexposed case at
low-exposure concentrations. In addition, these authors also used a hockey stick-shaped curve
such that the dose-response for cell division rates remained unchanged from the baseline, rising
only at 6 ppm (7.4 mg/m3) and higher exposure concentrations. This resulted in more conservative
estimates of risk when used in the clonal expansion model for cancer.
In addition to the data from the two tumor bioassays, Conolly et al. (20031 included
historical control data on 7,684 animals obtained from the National Toxicology Program (NTP)
F344 rat inhalation and oral bioassays. The resulting model predicts the probability of a nasal SCC
in the F344 rat as a function of age and exposure to formaldehyde. The fit to the tumor incidence
data is shown in Figure 2-4 as indicated by the long, dashed line. (For later reference in
Appendices B.2, this figure compares the fit to the data obtained by the modeling in Conolly et al.
(20031 with that obtained by EPA's reimplementation of this model under identical conditions,
inputs, and assumptions in Subramaniam et al. (20071. as indicated by the dash-dot line). The
reader is referred to the original papers for further details regarding the methodology.
This document is a draft for review purposes only and does not constitute Agency policy.
2-59 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
0.0020
0.0015
J 0.0010
0.0005
0.0000
0
200
0 ppm
400 600 800 1000
0.20 -i
0.15
0.10
0.05
0.00
0
6 ppm
t /
/
~ /
' /
200
400 600 800 1000
l.O-i
0.8
0.6
0.4
0.2
0.0
10 ppm
0 200 400 600 800 1000
Age (days)
l.O-i
0.8
0.6
0.4
0.2
0.0
0
200
Kap lan-M eier C on oily fit Our fit
Figure 2-4. Fit to the rat tumor incidence data using the model and
assumptions in Conolly et al. (2003).
Fitted curves obtained by Conolly et al. (2003) is compared with EPA reproduction of these results under
identical conditions, inputs, and assumptions; there were minor residual differences among the
implementations (see Subramaniam et al., 2007). The tumor incidence data are shown here by the
Kaplan-Meier adjusted probabilities.
1 The BBDR modeling approach affords a convenient way to integrate multiple types of
2 mechanistic information in modeling the time-to-tumor data, and visually it appears to fit these
3 data well (as shown in Figure 2-4). Further clarification pertaining to the structure and calibration
4 of the models in Conolly et al. (2004. 20031 that are key to understanding model assumptions is
5 provided in Appendix B.2.
6 Benchmark modeling of cancer incidence and human equivalents within the range of the data
7 Benchmark concentrations (BMCs) and the corresponding 95% lower confidence bounds
8 (BMCLs) were calculated at a benchmark response level (BMR) at the lowest end of the range of the
9 observed data (U.S. EPA. 20121. BMCs and BMCLs at the BMRs of 0.005 and 0.01 extra risk were
400 600 800 1000
Age (days)
This document is a draft for review purposes only and does not constitute Agency policy.
2-60 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
determined with the BBDR models. These were compared with values determined at the BMRs of
0.05 or 0.1 extra risk level to facilitate comparison with other chemicals. A BMR of 0.005 is lower
than the lowest observed tumor response (0.0085), when corrected for survival, from the combined
data from the Kerns et al. (1983) and Monticello et al. (1996) bioassays. Using this lower value is
considered appropriate because the BBDR modeling incorporates information on regenerative cell
proliferation, derived from cell labeling data, which may be considered a precursor response. The
BBDR models (model 1 & model 2 below) used for this purpose provided good fits to the time-to-
tumor incidence data, similar to the fit shown in Figure 2-4, and are based on the Conolly et al.
(2003) model with the following modifications.
Model 1 is based on the more conservative model in Conolly et al. (2003), where the
parameters governing the kinetics of normal and initiated cells were derived as hockey stick-
shaped functions of flux, with a critical modification. Conolly et al. (2003) added historical control
animals from all NTP studies to the data from the concurrent controls, whereas model 1 includes
NTP historical data from only the inhalation route of exposure. This is because the incidence rate of
nasal SCC is very different between these two categories of NTP historical studies, and the generally
accepted practice is to not include studies from other routes of exposure when using historical
controls (see Subramaniam et al. (2008; 20071 for an explanation of this issue). Model 1 is the
same as Model E in Table III of Subramaniam et al. (2007).
Model 2 makes major modifications to Conolly et al. (2003) in regards model structure as
well as values for input parameters. First, the shape of the dose-response for the division rates of
normal (N) cells as a function of formaldehyde flux, aN(flux) [an input to the TSCE model], was
monotone increasing without a threshold in dose, and obtained by fitting the 13-week cell
replication data in Monticello et al. (1996). (See modeling of cell replication data in Appendix B.2.2
for a discussion pertaining to using the 13-week data.) The raw replicate animal data from this
study was provided to EPA by the Hamner Institutes for Health Research. Second, the dose-
response for the division rates of initiated (I) cells, ai(flux), was assumed to be a sigmoidal-shaped
curve, increasing monotonically with flux from a background value up to an asymptotic value, and
constrained by afflux = 0) > aN(flux = 0). The death rate of initiated cells was given by the
assumption, Pi(flux) = K-ai(flux), where k is an estimated constant. This model is discussed in detail
as "model 15" in Appendix B.2.2. Furthermore, as in model 1, only the historical controls from
inhalation studies were added to the concurrent controls.
Weekly averaged DPX concentrations as calculated by the PBPK model described in
Subramaniam et al. (2007). a variant of the PBPK model in Conolly et al. (2000). were used. The
model fits to the observed tumor incidence data, parameter values, and respective comparisons
with Conolly et al. (2003) are provided in Appendix B.2.2. The results based on these models are
included in Table 2-22, and details pertaining to the model structure are provided in
Appendix B.2.2.
This document is a draft for review purposes only and does not constitute Agency policy.
2-61 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Toxicological Review of Formaldehyde—Inhalation
The BMCs mentioned above and their corresponding BMCLs were then converted to their
equivalent concentrations in humans (HECs) based on formaldehyde flux to the nasal tissue
obtained using CFD modeling in the rat and human fKimbell et al.. 2001bl The average mass flux of
formaldehyde (pmol/mm2-hour) to the entire surface of the airway lining, but excluding surface
lined by nonmucus-coated squamous tissue which is thought not to absorb formaldehyde, was used
for the extrapolation (see the Section, Computational fluid dynamic modeling, above, also in
Section 2.2.1). The HEC corresponding to a particular benchmark level in the rat was then
calculated by assuming that continuous lifetime exposure to a given steady-state flux of
formaldehyde, expressed in pmol/mm2-hour, leads to equivalent risk of nasal cancer across species.
This extrapolation included a multiplication by (6/24) x (5/7) to adjust the laboratory exposure
regimen for continuous exposure.
Schlosser et al. (2003) also calculated benchmark levels and corresponding HECs using DPX
as the relevant dose metric expressed as pmol of formaldehyde equivalents covalently bound to
DNA per unit volume of nasal tissue. These calculations used CFD and PBPK models to calculate
formaldehyde flux and DPX concentrations in the rat and human. The assumption in using DPX
data to calculate the HEC was that lifetime exposure to the same DPX concentration for a given
duration each day leads to equivalent risk across species. These were exposures that resulted in
the same steady-state DPX concentrations as the weekly time-weighted averaged DPX values in rats
at the rat benchmark exposure concentrations.
The benchmark levels in the rat and the HECs obtained using the above methods and dose
metrics are shown in Table 2-22.
This document is a draft for review purposes only and does not constitute Agency policy.
2-62 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Table 2-22. Benchmark concentrations and human equivalents using
formaldehyde flux and DPX as dose metrics
Models
Rat benchmark cone (ppm)
Human equivalent conca (ppm)
Extra risk
Extra risk
0.005b
0.01
0.05
0.1
Dose
metricc
0.005b
0.01
0.05
0.1
Weibulld
with threshold (Schlosser et al.,
2003)
BMC
BMCL
5.91
5.58
6.12
5.94
6.40
6.22
Flux
BMC
BMCL
0.75
0.71
0.78
0.76
0.82
0.79
DPX
BMC
BMCL
0.76
0.71
0.79
0.76
0.84
0.81
Multistage Weibull time-to-
tumor0'5
BMC
BMCL
4.28
3.57
5.93
5.52
6.84
6.41
Flux
BMC
BMCL
0.35
0.30
0.49
0.46
0.57
0.53
Rat BBDR model 1
BMC
BMCL
4.99f
4.95
5.37f
5.19
Flux
BMC
BMCL
0.42
0.41
0.45
0.43
Rat BBDR model 2
BMC
BMCL
5.41
5.25
5.75
5.59
Flux
BMC
BMCL
0.45
0.44
0.48
0.46
Abbreviations: BMC = benchmark concentration; BMCL = benchmark concentration; BBDR = biologically based
dose-response; TWA = time=weighted average; DPX = DNA-protein crosslink; CFD = computational fluid dynamic;
PBPK = physiologically based pharmacokinetic.
aHuman benchmark levels were continuous environmental exposures that would result in steady-state flux (or
DPX) levels in humans equal to the average flux (or weekly TWA DPX) levels in rats at the rat BMCs adjusted for
6 hours/day and 5 days/week. Values derived using flux as dose metric decrease by a factor of 1.4 if flux
estimates based on Schroeter et al. (2014) are used instead of Kimbell et al. (2001a).
bThe BMR of 0.005 was used only with the BBDR modeling because these models incorporate precursor response
data related to cellular proliferation (see discussion in surrounding text). Because benchmark concentrations at
0.005 and 0.010 extra risk levels were reported when BBDR modeling was used, they were not calculated at the
0.05 and 0.1 levels.
cFlux and DPX levels were computed by CFD and PBPK modeling, respectively.
dp-value for Weibull model fit = 0.90, best fit obtained with a positive intercept on dose axis.
eP(d,t) = 1 - exp[-(q0 + qid + q2d2 + ... + qkdk)x tz], q0, qi, q2, q3, q4 = 0, q5 = 2.9 x 10~22, z = 8.1. Curve passes through
origin. Fit was judged by comparing fitted curve to Kaplan-Meir survival estimates since goodness-of-fit p-value
was not provided by software package.
'Roughly similar result was obtained with model in Conolly et al. (2003). BMC0os = 4.84 ppm and BMC0i = 5.48 ppm
for their hockey-stick model as discerned from Figure 5 of their paper. BMCL values could not be estimated since
confidence bounds were not reported.
1
2
3
4
5
6
7
480.33 at 0.1 ppm, 0.32 at 1 ppm.
This document is a draft for review purposes only and does not constitute Agency policy.
2-63 DRAFT-DO NOT CITE OR QUOTE
As discussed in Section 1.1.3, Toxicokinetics of Formaldehyde, Schroeter etal. (20141
revised the dosimetry model of Kimbell et al. (2001b; 20011. used for the flux estimates in the table
above, to include endogenous formaldehyde production and to explicitly model formaldehyde
pharmacokinetics in the respiratory mucosa. EPA estimated the extent to which the results in the
above table change if flux estimates from Schroeter et al. (2014) are used. The average flux over
nonsquamous regions of the rat nose is roughly one-third48 of that in the human, based on the
dosimetry in Schroeter et al. (20141 in which endogenous formaldehyde is taken into account
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Toxicological Review of Formaldehyde—Inhalation
compared to a ratio of roughly one-half based on the dosimetry in Kimbell et al. (2001a). Thus,
wherever flux is used as the dose metric, the benchmark concentrations calculated in the above
table are not altered appreciably if the revised dosimetry model by Schroeter et al. (2014) is
applied, decreasing only by roughly a factor of 1.4.49
Benchmark modeling of precursor lesion data in the rat: cell proliferation and hyperplasia
Benchmark concentrations based on signatures of increased cell proliferation are useful in
that increased regenerative cell proliferation is assumed to be a contributory MOA—a factor that
can lead to a greater likelihood that DNA damage becomes heritable mutations before it is repaired.
Significantly increased cell proliferation as well as hyperplasia (increased cellular proliferation that
is identified to be pathologically "abnormal" in tissues) has been observed in response to exposure
to formaldehyde as described earlier in Section 1.2.4.
Cell proliferation
Schlosser et al. (2003) used cell proliferation to represent an adverse response and
modeled the dose-response for unit length labeling index measurements in F344 rats. They
reported benchmark concentrations and 95% lower confidence bounds corresponding to 1%, 5%,
and 10% increase in this index over the mean level for controls using dose-response functions that
allowed for a threshold in dose.50 The corresponding HECs spanned a tight range of 0.44-0.47 ppm
(0.54-0.58 mg/m3) (see Table 8 of their paper.)
The data used in their modeling were constructed using a cellular labeling index over
several locations on the F344 rat nose, as reported by Monticello et al. (1996). The data from
Monticello et al. (1996) represent the longest duration cell proliferation study available, which
included measurements across a range of study time points and nasal regions. Due to
methodological constraints intrinsic to all the available cellular labeling studies, including
Monticello et al. (1996), these data are based on DNA labeling of actively proliferating cells only
during the last day of exposure (see Appendix A.5.6 for additional discussion). Schlosser et al.
(2003) averaged the data collected from several nasal sites after weighting by exposure time. This
introduces some uncertainty because time-weighted averaging underweights early exposures
(e.g., 12-13 weeks of exposure) that may have contributed significantly to carcinogenesis (see
Section, Uncertainty-variability in cell replication dose-response of normal cells, later in this section
for further discussion); for instance, the few studies that investigated latent effects in rats
(i.e., Wistar) did observe an increased tumor incidence at 1 to >2 years following high-level
formaldehyde exposure lasting only ~13 weeks (Woutersen et al.. 1989: Feron etal.. 1988).
Similarly, additional methodological uncertainties that are difficult to address experimentally
49This is an approximate estimate for resting inspiration. The various components of the BBDR modeling
were not recalibrated or rerun in light of the revised flux estimates for both species.
50They also modeled with functions that were constrained to pass through the origin but do not report BMCL
values.
This document is a draft for review purposes only and does not constitute Agency policy.
2-64 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
include large site-to-site variation in the labeling (i.e., >10-fold); differences in the number of cells
across nasal sites; and the possibility that histologic changes and thickening of epithelium that
occur at later times for the higher doses likely affect the replication rate. These issues are discussed
further and several other plausible dose-response curves for cell replication from Monticello et al.
(1996) are developed (see Appendix B.2).
Other well-conducted studies of cellular proliferation using similar labeling methods help
estimate the potential impact of these uncertainties in the benchmark concentrations calculated by
Schlosser et al. (2003). In general, data from other studies investigating shorter-term
formaldehyde exposure durations, as well as the data for shorter duration exposures in Monticello
et al. (1996), routinely indicate proliferative effects at lower formaldehyde exposure levels within
similar nasal regions51 (see Appendix A.5.6 for comparisons across various durations of exposure).
As discussed in the Appendix, it appears reasonable to assume that all formaldehyde exposures
longer than 12 weeks are equally relevant to potential cancer development The data available
from medium and high confidence studies longer than 12 weeks, including multiple measures in
Monticello et al. (1996). are arrayed in Figure 2-5, below, and point to a two- to-three-fold range of
observed values below the benchmark concentration estimated by Schlosser et al. (2003) as
represented by the dotted vertical lines in the figure. This comparison partly elucidates the
uncertainty in the HEC values derived by Schlosser et al. (2003) to understand the cumulative
effects of chronic formaldehyde exposure on cellular proliferation.
51As the regions analyzed varied across studies, comparisons in Appendix A.5.6 and in Figure 2-5 compare
proliferation observed in locations as near to the anterior lateral meatus as possible, as this region was most
commonly reported across studies and is a region at which tumors have commonly been observed (see
Section 1.2.5, URT cancer in experimental animals).
This document is a draft for review purposes only and does not constitute Agency policy.
2-65 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
3x&2x Schlosser et al.
below (2003) rat BMCL
3000-
2000
1000
0 5 10 15 20
Formaldehyde concentration (mg/m3)
Study
High/Med
Strain
Exposure
Nasal region shown
Labeling
Metric
B-
Andersen et al„ 2010
H
F344
13wk
ALM (L2)
3d BrdU
ULLI
E)
Meng et al., 2010
H
F344
13wk
ALM (note: p<0.01)
3d BrdU
LI
-©-
Wil mer et al., 1989
H
Wistar
13wk
NT/MT
18h thym.
LI
-e-
Zwart et al., 1988
H
Wistar
13wk
NT/MT/ALM (L2; NC in L3)
18h thym.
turnover
Monticello et al., 1996
M
F344
12wk
ALM (note: no statistics)
18h thym.
ULLI
-B-
Monticello et al., 1996
M
F344
6mos
ALM (note: no statistics)
18h thym.
ULLI
¦EF
Monticello et al., 1996
M
F344
lyr
ALM (note: no statistics)
18h thym.
ULLI
Monticello etal., 1996
Mi
F344
18mos
ALM (note: no statistics)
18h thym.
ULLI
-A-
Casanova etal., 1994
M
F344
12wk
LM (less in M/PM)
3h 14C
14„ .
C incorp.
Figure 2-5. Cellular proliferation measured by DNA labeling in
studies 212 weeks.
Data from high and medium confidence studies (High/Med; H/M) exposing rats to formaldehyde for at
least 12 weeks (wk), and up to 18 months (mos), were normalized to percentage change from controls to
compare across the different metrics of proliferation reported (e.g., labeling index [LI]; unit length labeling
index [ULLI]; incorporation of radiolabeled carbon). The regions compared typically included the lateral
meatus (LM) in anterior regions (e.g., LI; L2; anterior LM), although one comparison was in related
structures (i.e., nasoturbinates [NT] and maxiiloturbinates [MT] in Wilmer et al. (1989). The DNA labeling
procedures included bromodeoxyuridine (BrdU), thymidine (thym.), and radiolabel. Filled shapes
represent statistical significance (p < 0.05), as reported by the study authors. The vertical lines represent
the rat BMDL01, as reported by Schlosser et al. (2003) and estimates which are two- and three-fold lower
than the Schlosser et al. (2003) rat BMDL. References: Andersen et al. (2010); Meng et al. (2010);
Monticello et al. (1996); Casanova et al. (1994); Wilmer et al. (1989); Zwart et al. (1988).
This document is a draft for review purposes only and does not constitute Agency policy,
2-66 ' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Hyperplasia
EPA modeled the incidence of basal hyperplasia reported by Woutersen et al. (1989) in a
28-month bioassay using Wistar rats. These animals were exposed to 0, 0.1,1.0, and 9.8 ppm (0,
0.123,1.23, and 12.05 mg/m3) formaldehyde and the observed incidences of hyperplasia were
0/26,1/26, 2/28, and 14/26. The BMC and BMCL at the benchmark response of 0.1 extra risk were
1.68 and 1.108 ppm (2.07, and 1.36 mg/m3), respectively. The HEC corresponding to the BMCL is
0.1609 ppm (0.198 mg/m3) when adjusted for continuous human lifetime exposure, which is
roughly three times lower than the HEC derived from the time-weighted averaged labeling index
by Schlosser et al. (2003). It is useful to note that this value is roughly comparable to the LECooos
derived from EPA's modeling of the NPC risk from the NCI epidemiology data.
Extrapolation using a biologically based dose-response model
In the case of formaldehyde, there are multiple options available for extrapolating to human
exposure scenarios which are typically at lower concentrations than the various HECs calculated
above. Subsequent to their BBDR modeling (Conolly et al. (2003.)) of nasal cancer in the rat, Conolly
et al. (2004) developed a corresponding model for humans, which they used for the purpose of
extrapolating the observed risk in the rat to human exposures. This human extrapolation model is
conceptually similar to the modeling in Conolly et al. (2003) but does not incorporate any data on
human responses to formaldehyde exposure. A particular contribution of this model toward
extrapolation is that it uses, as input, DPX concentrations and values of local formaldehyde flux to
the tissue as obtained from PBPK and fluid dynamic dosimetry models respectively (Conolly etal..
2000: Subramaniam etal.. 19981. The modeling in Conolly et al. (2004, 20031. while still a
statistical model where some key parameters are determined by model fit to the tumor data,
incorporates more detailed biological hypothesis and mechanistic data than is normally employed
in modeling cancer risk. Toxicodynamic models developed on the basis of an agent's MOA, if robust,
are generally preferred over default approaches for extrapolation, with the extent of extrapolation
determined by model uncertainty (U.S. EPA. 2005a).
In this section, we present extrapolations of the rat nasal cancer risk to humans carried out
in Conolly et al. (2004)- Continuous human lifetime extra risk estimates from this model following
inhalation exposure to 1.0 ppb-1.0 ppm (1.23 |a,g/m3-1.23 mg/m3) formaldehyde concentrations
are provided in Table 2-23, and compared with human risk estimates derived from EPA's modeling
of the NPC mortality in the NCI occupational epidemiology data (note: the comparison with
mortality estimates appears appropriate since Conolly et al. (2004) had modeled the tumors as
rapidly fatal). This comparison is provided only for perspective, noting in particular that NPCs are
specific to tumors only in the human nasopharynx (see Section 1.2.5). Conolly et al. (2004)
developed two clonal growth models based on using different representations of the low dose-
response for the cell division rate as input data. The first, denoted as optimal in the table, was
derived from using the best fit, a J-shaped curve, to the dose-response for the TWA of the cell
This document is a draft for review purposes only and does not constitute Agency policy.
2-67 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Toxicological Review of Formaldehyde—Inhalation
labeling data in rats such that values at 0.7 ppm and 2.0 ppm (0.9 mg/m3 and 2.46 mg/m3) were
below the control value; the second, presented as their conservative (in the sense of being more
health protective) approach, was derived from using a hockey-stick shape to replace the J-shape in
the low-dose portion of the optimal case such that values at the two lowest concentrations were the
same as the control. In either case, risk estimates reported in Conolly et al. (2004) were based on
using maximum likelihood estimate (MLE) values for all model parameters except the parameter
kmu associated with formaldehyde's mutagenic potential for which they used an upper-bound
value; [kmu is the constant of proportionality that relates DPX concentrations to the probability of
formaldehyde-induced mutation occurring per-cell generation).
The optimal model in Conolly et al. (2004) indicates lifetime human risk estimates to be
substantially below baseline risk levels (i.e., negative values of extra risk) for formaldehyde
exposures less than roughly 2 ppm (2.46 mg/m3), while their conservative model predicts values
that do not appreciably exceed baseline levels (i.e., extra risk less than 10 5) for exposures less than
0.2 ppm (0.25 mg/m3). Atthe ECooos benchmark concentration of 0.19 ppm (0.23 mg/m3) derived
from the NCI occupational epidemiology data, the conservative model in Conolly et al. (2004)
predicts roughly a 100-fold lower continuous lifetime risk than the central estimate indicated by
EPA's analysis of the epidemiology data. The difference is roughly the same at lower exposure
concentrations, while at 1.0 ppm (1.23 mg/m3) the conservative model predicts a 1,000-fold lower
value than EPA's central estimate based on the epidemiology data (see Appendix B.2.2).
The maximum likelihood value of the parameter kmu was estimated to be zero in the
modeling, leading to the inference by the authors that formaldehyde's direct mutagenic action is not
relevant to carcinogenicity in the rat or human, and that the observed tumor response in the rat can be
explained on the basis of regenerative cellular proliferation in response to cell injury. These results
have been interpreted by some to mean that exposures protective of the effects of cell proliferation are
adequate to protect against formaldehyde-induced nasal cancers fConollv et al.. 2004: Slikker etal..
2004). The uncertainty in these estimates and conclusions are evaluated below.
Table 2-23. BBDR model estimated extra risk of SCC in human respiratory
tract compared with EPA's modeling of extra risk of NPC from the human
occupational epidemiology data
Formaldehyde concentrations
0.001 ppm
0.01 ppm
0.10 ppma
1.0 ppm
Conollv et al. (2004) optimal estimate15
-1.0 X 10"5
-1.0 X 10"4
-9.1 x 10"4
-5.0 x 10 3
Conollv et al. (2004) conservative estimate15
+3.1 x 10"8
+3.2 x 10"7
+3.5 x 10"6
+2.7 x 10"4
EPA analysis-NCI NPC mortality MLE (UCL)C
+1.2 x 10"6
(+2.1 x 10"6)
+1.3 x 10"5
(+2.3 x 10"5)
+1.8 x 10"4
(+4.1 x 10"4)
+2.7 x 101
(+8.7 x 10-1)
aFor reference, the mortality-based LECooos derived from the NCI occupational data is 0.11 ppm (ECooos is 0.19 ppm).
bConollv et al. (2004) risk estimates were based on using MLE values for all model parameters except the
parameter associated with formaldehyde's mutagenic potential for which they used an upper bound.
cSee section 2.2.1; MLE = maximum likelihood estimate; UCL = 95% upper confidence limit.
This document is a draft for review purposes only and does not constitute Agency policy.
2-68 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
Uncertainty in the dose-response estimates
The ratio of the BMCL to the BMC is a convenient way to express the statistical uncertainty
in the benchmark concentration derived by a given model. (This ratio is also dependent on the
value of the benchmark response considered.) Table 2-22 indicates this ratio to be tight, ranging
from 0.83 to 0.96 across the models atthe BMRof 0.1. However, it is well-recognized (U.S. EPA.
2005a) that there is a large uncertainty inherent to using statistical models to extrapolate outside
the range of observed data. For example, in the context of the multistage Weibull model fit to the
formaldehyde time-to-tumor data in Table 2-22, the slope atthe origin, ql, was zero, whereas the
upper bound on this value, ql* was 0.02 ppm1; as shown in Table 2-22, this value is comparable to
that derived using EPA's straight line extrapolation.
The level of confidence in various components of the biologically based modeling approach
and its use for extrapolation is next addressed; the relevant question is whether the BBDR modeling
decreases uncertainty in extrapolating risk or, by explicitly identifying the sources of uncertainty,
points to approaches and data needs that may help reduce the uncertainty.
Uncertainties and confidence in the BBDR modeling and extrapolation
EPA carefully evaluated the level of confidence and sources of uncertainties in different
components of both the rat BBDR model and the corresponding human extrapolation model (Table
2-24). Seven issues that were evaluated are tabulated below, and elaborated in more detail in
Appendix B.2.2 and supporting references. Of these, issue numbers 3, 6 and 7—related to
replication rates of normal and initiated cells and the use of historical control animals—were found
to have major impacts on qualitative and quantitative conclusions drawn from the modeling, and
are briefly discussed below.
This document is a draft for review purposes only and does not constitute Agency policy.
2-69 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Table 2-24. Evaluation of BBDR modeling issues
Issue
Supporting references for evaluation
1
Confidence in FA airflow and flux model,
and assessment of interindividual
variability in FA flux; airway
reconfiguration due to long-term dosing
Kimbell et al. (2001a); Subramaniam et al.
(1998); Kimbell et al. (1997a); Garcia et al.
(2009); Subramaniam et al. (2008); Cohen Hubal
et al. (1997); Morgan (1997); Monticello et al.
(1996); Kimbell etal. (1997b)
2
Uncertainties in FA-DPX PBPK model
Subramaniam et al. (2008); Subramaniam et al.
(2007)
3
Uncertainties and variability in the rat cell
labeling data, the derivation of cell
division rates from these data, and their
applicability to human cell division rates
Subramaniam et al. (2008); Conollv et al. (2004)
4
Use of an approximate method by
Hoogenveen et al. to solve the two-stage
clonal expansion model equations
Subramaniam et al. (2007); Crump et al. (2005)
5
Assumption that all observed SCC in rats
were rapidly fatal; Model assumption of a
time delay from occurrence of malignant
cell to death
Subramaniam et al. (2007); Crump et al. (2005);
Crump et al. (2008)
6
Sensitivity of model results to the use of
historical control animals drawn from all
NTP cancer bioassays
Crump et al. (2008); Subramaniam et al. (2007)
7
Uncertainties in assumed division and
death rates of initiated cells
Crump et al. (2009); Crump et al. (2008);
Subramaniam et al. (2008)
Uncertainty-variability in cell replication dose-response of normal cells
Use of the raw cell labeling data from Monticello et al. (1996: 1991) to calculate replication
rates of normal cells for input to the TSCE models in Conolly et al. (2004. 2003) involved several
steps and assumptions. First, as shown in Table 2-21, the first phase for early exposure periods
Monticello et al. (1991) employed injection labeling with a 2-hour pulse labeling, whereas the
second phase for longer exposure periods Monticello et al. (1996) used osmotic mini-pumps for
labeling with a 120-hour labeling time. These data were pooled by using a normalization procedure
for the injection labeled data. Second, the average values from the labeling (averaged over the
replicate animals and after the above normalization) were weighted by the exposure times in
Monticello et al. (1996: 1991) and averaged over the nasal sites. Thus, the data were combined into
one TWA for each exposure concentration. Third, Monticello et al. (1996: 1991) used unit length
labeling index (ULLI) to quantify cell replication within the respiratory epithelium. ULLI is a ratio
between a count of labeled cells and the corresponding length (in millimeters) of basal membrane
examined. Therefore, ULLI had to be converted to the per-cell labeling index (LI), which is the ratio
of labeled cells to all epithelial cells, in this case, along some length of basal membrane and its
This document is a draft for review purposes only and does not constitute Agency policy.
2-70 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
associated layer of epithelial cells. This was accomplished by using data from a different
experiment fMonticello etal.. 1990al where both quantities had been measured for two sites in the
nose. Fourth, cell division rates were then calculated from the TWA using an approximation
developed by Moolgavkar and Luebeck (1992).
Fifth, the empirical data could be used in Conolly et al. (2003.) to directly calculate cell
replication rates only for approximately the lower one-fourth of the full flux range
(0-39,600 pmol/mm2-h) needed to model the bioassay data. The unknown cell replication rates for
the upper three-fourths of the flux range were determined by linear interpolation to a maximum
cell replication rate that was estimated as a statistical parameter fit to model predictions of the
tumor incidence data (see fSubramaniam etal.. 20081 for further details and biological implications
of this procedure).
Finally, because there are no equivalent labeling index data available for the human
respiratory epithelium, the above dose-response for normal cell replication derived for the rat was
also directly assumed to apply to the human except for different values for the fraction of rat and
human nasal epithelial cells capable of dividing (Conolly et al.. 2004).
The TSCE model is generally sensitive to normal cell division rates, and there are
considerable uncertainties (quantitative and qualitative) and variability in the dose-response for
the replication rates of normal cells (an) as characterized in the above steps. For example,
Figure 2-6, below, shows an as a function of formaldehyde flux to the rat nasal epithelial tissue
[using only values derived from the continuous ULLI data in (Monticello etal.. 1996)].
Corresponding to any particular dose (in terms of formaldehyde flux to tissue) aN varies by one to
two orders of magnitude. As shown in Appendix B.2.2, a variety of cell replication dose-response
curves can be drawn to fit these data, and the use of an exposure TWA of cell labeling data over
sites was found to be problematic on multiple accounts. Furthermore, the formula relating LI to an
was for continuous labeled data and its use for pulse labeled data, as evaluated in the appendix, was
found to be extremely uncertain.
The results in Table 2-23 for the optimal and conservative models in Conolly et al. (2003)
represent a sensitivity analysis of the impact on risk estimates of varying the dose-response for
normal cell replication rates at the low-dose range, and the differences between the two model
results point to large variations in predicted human risk estimates from incorporating some of the
variability and uncertainty in normal cell division rates in inputs to the TSCE model. In the
neighborhood of the POD from the observed occupational epidemiology data, these models
compute extra risk estimates of-9.1 x 10"4and+3.5 x 106 respectively compared to a value of
+4.1 x 10"4 indicated by the epidemiology data.
This document is a draft for review purposes only and does not constitute Agency policy.
2-71 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
o.i q
CD
4—>
ro
~ALM
¦ MMT
A AM5
XPLM
¦X PMS
• A D5
0.00001 -1=
0
A
2000 4000 6000 8000 10000 12000 14000
2
Formaldehyde flux (pmol/mm -h)
Figure 2-6. Dose-response for normal cell division rate, aN, versus
formaldehyde flux to tissue for the F344 rat nasal epithelium.
Values were derived from continuous unit length labeled data by Monticello et al. (1996). Each point
represents a measurement for one rat, at one nasal site, and at a given exposure time. Data shown for six
nasal sites (legend, nasal sites are as denoted in original paper) and four exposure durations (13, 26, 52,
78 weeks). For comparison purposes, the double black bars on the y-axis indicate the extent of difference
between two curves, modO and mod5, for the dose-response for cell division rates of initiated cells.
The assumption in Conolly et al. (2004) that cell division rates exhibit a similar dose-
response across rats and humans appears uncertain (Conolly et al. (20041 did consider different
values for rats and humans for the fractions of cells with replicative potential) (see Appendix B.2.2).
EPA was unable to find a rationale for this assumption in the literature. To the contrary it seems
possible that basal cell division rates may scale allometrically across species, considering that
enzymatic metabolism is likely to play a role in mitosis. [For example, West and Brown (2005)
argue thatDNA nucleotide substitution rates and inverse of lifespan scale as mass to the inverse
one-fourth power.]
Miller et al. (2017) found the modeling in Conolly et al. (2004) [that is, their human
extrapolation model] to be sensitive to the fraction of cells considered to have replicative potential
in the human respiratory tract, a parameter in the human model. For example, added risk over
background increased (by 87%) from -1.0 x 10 3 to -1.3 x 10 4 at 0.4 ppm exposure concentration
but decreased (by 127%) from +7.7xl0 4 to -2.1 x 0 4 at 2.0 ppm, when this parameter was changed
This document is a draft for review purposes only and does not constitute Agency policy.
2-72 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
from that experimentally observed by Mercer et al. (19941 for various cell types to a value of 1.0
(i.e., all cells to have replicative potential) for the nonsmoking population at resting breathing.
Miller et al. (20T7) also reported new results obtained with the Conolly et al. (2004) model
in regards the site distribution of extrapolated human risk estimates over the respiratory tract. At
0.2 ppm and 1.2 ppm (0.25 mg/m3 and 1.48 mg/m3) inhaled exposure concentrations of
formaldehyde, the highest risk was predicted to occur in nasal tissue that received the lowest
formaldehyde flux, but which comprised the largest surface areas. Based on the flux patterns
displayed in Kimbell et al. (2001). this likely overlaps with the human nasopharyngeal region, and
indicates an important role for dosimetry in regards the epidemiological observation of
nasopharyngeal carcinomas. For the high exposure concentrations (3.6 ppm and 4.5 ppm; 4.43
mg/m3 and 0.62 mg/m3), the highest risk region was instead predicted to occur in regions of the
nose that received intermediate levels of formaldehyde flux.
Kinetics of initiated cells
There are no data on initiated (I) cells (the available empirical cell labeling data are for
normal [N] cells). Therefore, Conolly etal. (2004) assumed relationships that linked the division
rate, ai, and death rate, (31, for initiated cells to the division rate for normal cells, an, as a function of
local formaldehyde flux (since local flux was the most sensitive dose metric):
aj(flux) = aN(flux) x {cx- c2 • max [an(flux) - ccNbasai, 0]}- (2-2)
Pi(flux) = aN(flux), for all values of flux. - (2-3)
where Ci and C2 are constants estimated by fitting the clonal expansion model to the tumor
incidence data. No biological rationale was provided for these assumptions; however, these
assumptions allowed for a good fit to the rat tumor incidence data. The TSCE model is known to be
very sensitive to the kinetics of initiated cells, and the authors did not examine whether other
expressions would also fit the rat data but lead to different predictions of human risk. Therefore, to
evaluate the sensitivity of model predictions to the assumed relation (eq 2-2) between ai and aN in
the low flux region, EPA slightly modified this relation for ai(flux) for flux <475 pmol/mm2-h, while
keeping it identical to the values in Conolly et al. (2004) for 475
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
maximum extent of these modifications to the assumed replication rates of initiated cells is overlaid
by the double black bars in Figure 2-6, above, on the rates for normal cells, afflux), that are derived
from empirical data. As seen in the Figure, the extent of the modifications is extremely small in
relation to the empirical variability seen in normal cells. Thus, the modifications considered in the
sensitivity analysis appear biologically reasonable.
EPA's sensitivity analyses retained the same values for (3i (equation 2-3) as considered in
the original analysis. However, the ratio cti:Pi over the flux range in the modeling was closely
monitored. Because this ratio represents the growth advantage of initiated cells in the model, it
was kept close to the value of 1.0, similar to the range of 0.96-1.07 for the values of ai/Pi in (Conolly
etal.. 20041 [modO], In the sensitivity analysis, ai/Pi varied from 0.96-1.07 in mod-1; 0.96-1.08 for
modi, mod2, mod3, mod4; and 0.96-1.10 for mod5. Table 2-25 provides MLEs of continuous
lifetime human extra risk estimates at 0.15 ppm (0.18 mg/m3) exposure concentration for the
original Conolly model (modO) and compares those derived from the above modifications. For
perspective, the table also compares with human risk estimates derived from EPA's modeling of the
NPC mortality52 in the NCI occupational epidemiology data (see Section 2.2.1).
mod 1
modi
mod2
mod3
0.00040
0.00020 1
0 500 1000 1500 2000
formaldehyde flux (pmol/mm2/h)
Figure 2-7. Small variations to ai(flux) for flux <475 pmol/mm2-h carried out
for sensitivity analysis.
ModO is the original model in Conolly et al. (2004); mod-1 decreases ai and modl-5 increase ai in modO
for low flux.
52The comparison with mortality estimates appeared appropriate since the tumors were modeled as rapidly
fatal in Conolly et al. (2004, 20031.
This document is a draft for review purposes only and does not constitute Agency policy.
2-74 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
Table 2-25. Sensitivity of BBDR modeled human SCC risk at 0.15 ppm to small
variations in normal (aN) and initiated (ai) cell replication rates
Model*
Extra risk
modO: Conollv et al. (2004), J-shaped aN, a
-1.0 x 10"3
mod-1: Decrease ai for low flux in modO
-1.5 x 10"3
modi: Increase ai for low flux in modO
-3.0 x 10 4
mod2: Increase ai for low flux in modO
+9.0 x 10"5
mod3: Increase ai for low flux in modO
+3.0 x 10"4
mod4 Increase ai for low flux in modO
+9.0 x 10"4
mod5: Increase ai for low flux in modO
+3.0 x 10"3
Conollv et al. (2004), hockev-stick shaped aN, a
+5.7 x 10"6
EPA analysis of NCI NPC
+5.5 x 10"3
*See Figure 2-7 for depiction of modO, mod-1, mod0-5.
The results in this table indicate that extremely small differences in assumptions for ai
appear to have extremely large effects on the human model predictions. This analysis is continued
in Appendix B.2.2, where similar sensitivity of model predictions is demonstrated over a large
range of exposure concentrations. Larger variations in ai fsee Crump etal.. 20081. while still in
agreement with the model constraint of reproducing the observed tumor incidence data and the
background rate of lung tumors in humans, considerably broaden the range of predicted risk on
either side (below and above) of the baseline risk. Such an extreme sensitivity indicates that the
formaldehyde human TSCE model is unstable in response to the slight perturbations carried out to
the assumed values of ai, and is therefore not robust. It is well known that models are generally
uncertain outside of the range of the data over which they were calibrated f Crump etal.. 20101 and
this is indeed the case with the rat BBDR model. As discussed by Crump et al. (2009: 2008). the
human extrapolation BBDR model, on the other hand, is noteworthy for its extreme uncertainty at
all exposure concentrations, above as well as below the HECs that were calculated in the
benchmark modeling section.
There are currently no data of any kind, even in rats, to inform the effect of formaldehyde
on the kinetics of initiated cells. However, assuming that initiated cells related to tumors in the
respiratory tract can be identified and their division rates measured, it is reasonable to suppose
that these rates would be at least as variable as division rates of normal cells. Based on the normal
variation in such rates observed in normal cells in Figure 2-7, and the extreme sensitivity of the
formaldehyde model to small differences in assumed division rates of initiated cells, EPA concluded
that it would be impossible to measure these accurately enough to lead to any substantive
reduction in the large uncertainty in risk estimated by this model.
This document is a draft for review purposes only and does not constitute Agency policy.
2-75 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
Use of historical control animals
Because SCC in the nose is a rare tumor, Conolly et al. (2004, 20031 included in their model
7,684 control rats from all NTP cancer bioassays in addition to the 347 control animals in the Kerns
et al. (1983) and Monticello et al. (1996) inhalation bioassays used in the dose-response modeling.
In general, the inclusion of all NTP historical control animals regardless of exposure route, time of
study, etc. is problematic because there are legitimate questions regarding comparability of results
in rats from different stocks, studied at different times, in different laboratories, and by different
routes of exposure and evaluated by using somewhat different pathological procedures (Haseman
and Hailev. 1997: Rao et al.. 19871. In particular, the incidence rate in the inhalation historical
controls was found to be an order of magnitude lower than the rate in all historical controls
combined fsee Subramaniam etal.. 20071. Therefore, EPA examined the sensitivity of the BBDR
model predictions to the use of historical NTP control animals by restricting the historical controls
to only inhalation studies or by using only the concurrent controls.
When the NTP control data were restricted to those animals from NTP inhalation studies,
the upper-bound human risk estimate obtained by Conolly et al. (20041 (i.e., with everything else in
their modeling retained unchanged) was increased by 50-fold fCrump etal.. 20081. If only
concurrent controls are used, as is normally the practice in dose-response analysis of animal
bioassays, the Conolly et al. (2004) model for extrapolation of risk to humans becomes numerically
unstable, i.e., the MLE and upper-bound estimates of risk become infinite (Subramaniam et al.
(2007), Crump etal. ("200811.
Overall confidence in the formaldehyde BBDR models
The other issues listed in Table 2-24 are evaluated at length in Appendix B.2.2. Although
CFD model predictions of formaldehyde flux to the respiratory lining have not been verified
experimentally (due to formidable experimental challenges), predictions from other models that
use the calculated formaldehyde flux as input have been shown to agree with various kinds of
available data, and thus project a reasonable, albeit indirect, level of confidence in the formaldehyde
dosimetry modeling in both the rat and human nasal passages (see Appendix B.2.2). The CFD
models of formaldehyde flux are based on data collected from a single individual of each species.
Therefore, interindividual differences in regional dosimetry, particularly in the human, are not
accounted for fGarcia etal.. 2009: Subramaniam etal.. 20081.
Repair of DPX was assumed to be rapid and complete in 18 hours in the PBPK model for
DPX (Conolly et al.. 2000): this assumption was found to be highly uncertain (Subramaniam et al..
20081. While it has no impact on the rat BBDR model predictions (see Appendix B.2.2), the impact
of this assumption on the human extrapolation model, on the other hand, was significant (Crump et
al.. 20081. Furthermore, more recent results by Lai et al. (2016) indicate that in vivo DPX repair
may be slow and that DPX readily accumulates long-term in the nasal respiratory tissue in contrast
to its rapid hydrolysis in vitro.
This document is a draft for review purposes only and does not constitute Agency policy.
2-76 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
In summary, the human extrapolation modeling in Conolly et al. (20041 is extremely
uncertain on two accounts, and does not provide robust measures of human nasal SCC risk at any
exposure concentration. Therefore, the human extrapolation model is not used in this assessment
to directly calculate risk at human exposure scenarios. On the other hand, the rat BBDR modeling
improves the dose-response modeling of the observed nasal cancers in the F344 rat, and multiple
BBDR model implementations provide similar estimates of risk and confidence bounds in the
general range of the observed rat tumor incidence data. Therefore, the rat BBDR models are used
to calculate benchmark concentrations for PODs, and the benchmark response was extended
slightly below the observed. There is reasonable confidence in flux estimates derived from the rat
and human CFD models, which were accordingly used in deriving HECs corresponding to these
PODs. A candidate RfC and candidate unit risk estimates using these values are included in the
following section.
RfC approach for precursor lesion data in the rat: cell proliferation and hyperplasia
The highly curvilinear and steeply increasing dose-responses for DPX formation and cell
proliferation, concomitant with the highly nonlinear observed tumor incidence in the F344 rat,
have led to mechanistic arguments that formaldehyde's nasal carcinogenicity arises only in
response to significant cytotoxicity-induced regenerative cell proliferation fSwenbergetal.. 2011:
Conolly etal.. 2002: Morgan. 19971. In particular, Conolly et al. (2003) and Slikker et al. (2004)
inferred from BBDR modeling results that the direct mutagenicity of formaldehyde is less relevant
compared to the importance of cytotoxicity-induced cell proliferation in explaining the rat tumor
response. Thus, candidate RfCs (cRfCs) derived from available experimental data relevant to this
mechanism are presented and discussed. These cRfCs are interpreted as formaldehyde
concentrations below which it is unlikely that hyperplastic lesions develop or that cancers arising
from cytotoxicity-induced regenerative cell proliferation occur. In this interpretation, cytotoxicity-
induced regenerative cell proliferation, which increases the probability of errors in DNA replication,
and the subsequent development of hyperplastic lesions, are considered to be precursor events
that, if protected against, would prevent these mechanisms from contributing to the cancer
response. Below these cRfCs, formaldehyde may still increase the risk of nasal or upper respiratory
cancer through direct mutagenicity or other mechanisms, but the magnitude of cancer risk may be
significantly lower due to the absence of increased cellular proliferation or hyperplasia.
The following benchmark PODs and corresponding HECs were developed based on
increased cell proliferation as well as hyperplasia: (a) 0.44 ppm (0.54 mg/m3) corresponding to the
BMCLoi in Schlosser et al. (2003). and roughly two- to three-fold lower estimates based on
examining data from other cell labeling studies (as discussed above in the section on modeling
precursor lesion data), resulting in an overall range from 0.18 to 0.54 mg/m3; and (b) 0.16 ppm
(0.20 mg/m3) based on EPA's modeling of the incidence of basal hyperplasia reported by
Kleinnijenhuis et al. (2013) in Wistar rats. To these values, it is necessary to apply a UF = 3 to reflect
other uncertainties in extrapolating from animals to humans and a UF = 10 to account for human
This document is a draft for review purposes only and does not constitute Agency policy.
2-77 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Toxicological Review of Formaldehyde—Inhalation
variability (total UF = 30). This results in cRfCs that range from 0.006 mg/m3 to 0.018 mg/m3
when based on cell proliferation data and a cRfC of 0.007 mg/m3 from the hyperplasia data.
As noted earlier, it has been argued that the rat nasal tumors can be quantitatively
explained based solely on formaldehyde's cytotoxic potential. In accordance with this point of view,
a cRfC estimated from benchmark concentrations derived using the two rat BBDR models may be a
reasonable approximation for the dose at which there is no regenerative cell proliferative
contribution to the nasal or upper respiratory cancer response. A cRfC of 0.017 mg/m3 may be
obtained in this manner corresponding to the average HEC estimated using the two models at a
benchmark response of 0.005 extra risk reduced by a UF of 30. This value is encompassed by the
overall range of 0.006-0.018 mg/m3 obtained as explained above for the cRfCs based on cell
proliferation and hyperplasia.
However, Chapter 1 of this assessment also provides multiple lines of evidence that the
direct mutagenicity of formaldehyde plays a key role in its carcinogenicity. Cytogenetic effects in
occupational studies and the formation of DPXs in experimental animals have been reported at
exposures well below those considered to be cytotoxic (e.g., approximately 0.7-2 ppm or 0.9-2.5
mg/m3 in rats), and as noted earlier, DPX formation was detected in rats at exposures ranging from
0.3 ppm (0.37 mg/m3) to 15 ppm (18.5 mg/m3). The DPX dose-response shows a trend consistent
with an increase over baseline levels at 0.7 ppm (0.86 mg/m3), which becomes statistically
significant at 2 ppm (2.46 mg/m3) and above.
Furthermore, the previously mentioned inference that formaldehyde's direct mutagenic
action is relatively irrelevant to explaining the observed rat tumor response was found to be
extremely uncertain in EPA's uncertainty analysis. A reanalysis presented in Subramaniam et al.
(2007) indicated that, depending on the choice of control animals and alternate model assumptions,
a large contribution from formaldehyde's mutagenic potential may be needed to explain
formaldehyde carcinogenicity at low dose as well as in describing the observed tumor incidence.
Finally, as discussed in Section 1.2.5, Evidence on mode of action for URT cancers, genotoxicity is
itself thought to be one of the mechanisms by which formaldehyde exerts its cytotoxic action. Thus,
it appears difficult to argue for a demarcation along the concentration axis of one MOA relative to
the other. Therefore, because formaldehyde-induced tumors are not explained only by the cell
proliferative MOA at any exposure, and since EPA does not develop an RfC specifically for one MOA
when other MOAs also contribute to the tumor response, the use of an RfC approach is not
preferred.
Low-dose risk without extrapolating models below the observed data
The various arguments presented in the last two paragraphs of the previous section on an
RfC-like approach for cancer, particularly regarding formaldehyde's direct mutagenic potential,
provide greater support for a low-dose linear approach in extrapolating low-dose formaldehyde
cancer risk from the rat data. Following the procedures in EPA's cancer guidelines fU.S. EPA.
2005a) to be applied when knowledge of the MOA does not support an alternative approach or
This document is a draft for review purposes only and does not constitute Agency policy.
2-78 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 when direct mutagenicity does not contribute to the cancer response, this extrapolation was
2 carried out as a straight line drawn to the origin from the HEC corresponding to the BMDL. Unit
3 risks so calculated are shown in Table 2-26 below. The unit risks corresponding to BMRs at the
4 0.005 or 0.01 extra risk levels, spanned a remarkably tight range, 0.01-0.03 per ppm, across the
5 different methods and dose metrics (see Table 2-22). It is useful to contrast the unit risk value at
6 the 0.005 extra risk with that obtained for the statistical upper bound on the coefficient associated
7 with the first-order term in the multistage Weibull model described above in the statistical time-to-
8 tumor modeling (denoted ql* in an earlier EPA approach to low-dose linear extrapolation), ql*
9 was determined to be equal to 0.02 per ppm, and falls within this tight range.
This document is a draft for review purposes only and does not constitute Agency policy.
2-79 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Table 2-26. Unit risk estimates derived from benchmark estimates3
Models
Dose metric
Unit risk estimates from various PODs (1/ppm)
BMCLoos
BMCLoi
BMCLos
BMCLio
Weibull with threshold (Schlosser et al.,
2003)
Flux
0.014
0.066
0.127
DPX
0.014
0.066
0.127
Multistage Weibull time-to-tumor
Flux
0.033
0.109
0.189
Rat BBDR model
Flux
0.012
0.023
Rat BBDR model
Flux
0.011
0.022
aUnit risks derived using flux as dose metric increase by a factor of 1.4 if flux estimates based on Schroeter et al.
(2014) are used instead of Kimbell et al. (2001a). Also, see other footnotes from Table 2-22.
In conclusion, use of biologically based modeling allowed the use of various data, including
mechanistic information, in an integrated manner for modeling the incidence of nasal SCC in F344
rats and for deriving benchmark levels for extrapolation. A conventional multistage Weibull time-
to-tumor modeling was also used to model these data. For a given benchmark response level, PODs
and their corresponding HECs are remarkably similar across multiple models and dose metrics
(formaldehyde inhaled exposure concentrations, formaldehyde inhaled flux to tissue, DPX
concentrations). Biologically based clonal expansion models were carefully evaluated for directly
extrapolating the rat nasal cancer risk to human exposure scenarios. Predictions using these
models for humans were found to be not robust at any exposure concentration. Accordingly, the
clonal expansion modeling of the rat data was employed to derive multiple PODs and
corresponding HECs but not used for extrapolating to human exposure scenarios.
Selection of a Unit risk Estimate for Nasal Cancers
The unit risk estimates derived using the available human and animal data on nasal cancers
are similar (see Table 2-27), with the human estimate being only slightly lower than those values
estimated using ratbioassay and mechanistic data.
This document is a draft for review purposes only and does not constitute Agency policy.
2-80 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Toxicological Review of Formaldehyde—Inhalation
Table 2-27. Comparison and basis of unit risk estimates for nasopharyngeal
cancer in humans and nasal squamous cell carcinomas in rats
Human NPC estimate
Animal nasal cancer estimate
Study/endpoint
Beane Freeman et al. (2013) (NCI
industrial cohort): NPC mortality
Monticello et al. (1996); Kerns et al. (1983):
Incidence of nasal SCC in rats
Model features
Estimation of IIIR using Poisson
regression model and life-table
analysis:
• U.S. national incidence data
• Regression coefficients from log-
linear models of nasopharyngeal
cancer (NPC) mortality (exposed and
unexposed workers)
• Linear low-dose extrapolation from
LEC
Multiple mechanistic and statistical models,
including BBDR modeling, used for modeling
tumor incidence
Mechanistic information included:
• Dosimetric (CFD) modeling of formaldehyde
flux to rat, monkey and human airway lining
• PBPK model for rats incorporating dose-
response data on DPXs
• site-specific cell labeling measurements in nose
A linear low-dose extrapolation from human
equivalent dose at BMCL was employed
POD
95% lower bound on concentration at
0.05% incidence (approx. 0.05 ppm)
95% lower bound on concentration at 0.5%
incidence (approx. 0.2 ppm)
Unit risk estimatea
7.4 x 10"3 per mg/m3
(9.1 x 10"3 per ppm)
8.9 x 10"3 to 1.8 x 10"2 per mg/m3
(1.1 x 10"2 to 2.2 x 10"2 per ppm)
aNote that these estimates are provided for comparison purposes and do not represent ADAF-adjusted values.
ADAF = age-dependent adjustment factor.
A comparison of the preferred unit risk estimates based on human and rodent data
summarized above reveals that the different databases yield similar estimates. When data from
epidemiological studies of sufficient quality are available, these data are generally preferred for
estimating risks (U.S. EPA. 2005a). In the case of formaldehyde, the NCI epidemiological study
(Beane Freeman etal.. 20131 is a high-quality study for the purposes of deriving quantitative risk
estimates, and the estimates based on this study are preferred to the estimates based on the rat
data. Although there are uncertainties inherent in estimates from both the human and rodent
databases, the estimates based on the human data are considered better estimates of the risk to
humans.
Next, given that it was concluded in Section 1.2.5 that a mutagenic MOA was operative for
URT cancers, the unit risk estimate for NPC is adjusted for potential increased early-life
susceptibility, in accordance with EPA guidelines (U.S. EPA. 2005bl (see Section 2.2.4).
Uncertainties and Confidence in the Preferred Unit Risk Estimate for Nasal Cancers
The strengths and uncertainties in the unit risk estimate for NPC incidence are summarized
in Table 2-28. One of the largest sources of uncertainty in the NPC estimate has to do with the
This document is a draft for review purposes only and does not constitute Agency policy.
2-81 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
1 rarity of the cancer and, thus, the small number of exposed cases (n = 8) that informed the dose-
2 response analysis.
Table 2-28. Strengths and uncertainties in the cancer type-specific unit risk
estimate for nasopharyngeal cancer
Strengths
Uncertainties
• III R estimated from data that is
directly relevant to humans.
• Based on the results of a large,
high confidence epidemiology
study involving multiple
industries with detailed,
individual cumulative exposure
estimates and allowance for
cancer latency.
• Low-dose linear extrapolation is
supported by a mutagenic mode
of action (i.e., not a default).
• Similar unit risk estimates derived
using rat bioassay and
mechanistic data on nasal
cancers.
• NPC is a very rare cancer. This study followed more than 25,000 workers
for over 40 years and observed a statistically significant increase in RR
associated with the highest category of average exposure intensity,
however, only 10 cases occurred. The small number of deaths creates
uncertainties for the dose-response modeling (borderline model fit for
cumulative exposure including exposed and unexposed person-years,
p = 0.07).
• Uncertainty about optimal exposure metric(s). Cumulative exposure is the
standard metric used for unit risk estimates. Use of cumulative exposure
assumes equal importance of concentration and duration on cancer
incidence; yet, associations with peak exposure in epidemiological studies
and the nonlinear shape of the dose-response from animal bioassays
suggests greater influence of concentration.
• Although statistically significant increases in risk for NPC were reported by
multiple studies for several metrics of exposure (duration, cumulative,
time since first exposure, peak), the relationship with cumulative exposure
in the study used for IUR derivation was less precise (p-trend = 0.07 based
on the regression coefficient for the continuous model).
• Some uncertainty in the low-dose extrapolation is introduced based on the
potential for endogenous formaldehyde to reduce the uptake of the
inhaled gas at low doses, as demonstrated in modeling efforts by
Schroeter et al. (2014) and Campbell Jr et al. (2020).
3 Based on the attendant strengths and uncertainties outlined above, there is medium
4 confidence in the unit risk estimate for NPC incidence. The greatest uncertainty was related to the
5 small number of cases that contributed to the statistical analysis and resulting imprecision in
6 modeling the shape of the dose-response curve.
2.2.2. Derivation of a Myeloid Leukemia Unit Risk Estimate Based on Human Data
7 Choice of Epidemiology Study
8 Similar to the unit risk estimate for NPC, the estimate for myeloid leukemia is based on
9 results from the latest follow-up of the NCI cohort of industrial workers exposed to formaldehyde
10 (Beane Freeman et al.. 2009). the largest (25,619 workers) of the three independent industrial
11 worker cohort studies and the only one with sufficient individual exposure data for dose-response
12 modeling. Beane Freeman et al. (2009) conducted dose-response analyses of 123 deaths attributed
13 to leukemia and leukemia subtypes, as well as deaths from other LHP malignancies. As previously
This document is a draft for review purposes only and does not constitute Agency policy.
2-82 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
described, this well-conducted study is the only one that used internal comparisons rather than
standardized mortality ratios (reducing the impact of potential unmeasured confounding), and it
included a detailed exposure assessment conducted for each worker based on exposure estimates
for different jobs held and tasks performed fStewartetal.. 19861. and exposure estimates were
made using several different metrics—peak exposure, average intensity, cumulative exposure, and
duration of exposure.
For the LHP cancers, the strongest trends for the subtypes of interest were generally
observed with the peak exposure metric (Beane Freeman et al.. 20091. For myeloid leukemia,
Beane Freeman et al. (2009) reported an increasing trend in mortality risk (p = 0.07 for all person-
years) for peak exposure, but no trend was observed for cumulative exposure. For myeloid
leukemia and other/unspecified leukemias combined, recognizing that a substantial proportion of
the unspecified leukemias are probably myeloid leukemias, there was a nearly significant (log-
linear) trend with cumulative exposure (p = 0.10 for all person-years) (personal communication
from Laura Beane Freeman, NCI, to Jennifer Jinot, U.S. EPA, 21 February 2014). No exposure-
response relationships were indicated for multiple myeloma for any of the exposure metrics.
Another study, Hauptmann et al. (2009), was a case-control study of LHP cancers, with
exposure-response analyses, nested in the cohorts of "professional" workers (funeral industry
workers, in this case) studied by Hayes et al. (1990) and Walrath and Fraumeni (1984.19831.
Hauptmann et al. (2009) estimated exposures for each case and control using multiple exposure
metrics. Because of limitations in the exposure assessment, this study, while useful for hazard
assessment, was not used by EPA to derive quantitative risk estimates. Of primary concern, the
worker histories were obtained from surrogate responders (next of kin who had worked in the
funeral home with the study subject and coworkers). This is a valid approach for general metrics
such as 'ever embalming' or 'years of embalming', and statistically significant associations (for ever
embalming) and trends (for years of embalming) were observed for myeloid leukemia. However,
there is less confidence for more specific variables such as number and duration of embalmings per
calendar period and frequency of spills per calendar period, variables that are needed in the study's
exposure model to estimate cumulative exposure. For example, where information on a particular
variable was obtained from multiple respondents, Hauptmann et al. (20091 reported a substantial
amount of discordance for variables such as number of any embalmings and number of autopsied
embalmings. Furthermore, considerable amounts of data were missing. For example, Hauptmann
et al. (2009) reported that all but 16 of 44 cases of LHP cancer of nonlymphoid origin had 3 0% or
more of their detailed work history missing. Thus, although the results of the Hauptmann et al.
(2009) study were supportive of the hazard assessment, the uncertainty in the quantitative
estimates of cumulative exposure was considered dissuasive for the development of quantitative
cancer risk estimates.
This document is a draft for review purposes only and does not constitute Agency policy.
2-83 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
Exposure-response Modeling of the National Cancer Institute Cohort
The NCI cohort study fBeane Freeman etal.. 20091. was the only study with adequate data
for exposure-response modeling; however, the derivation of a unit risk estimate for myeloid
leukemia from these data is not straightforward, and several quantitative risk assessment
approaches were considered. Beane Freeman et al. (2009) used log-linear Poisson regression
models stratified by calendar year, age, sex, and race and adjusted for pay category
(salary/wage/unknown) to estimate RRs for various categorical exposure groups (see Table 2-29).
The NCI investigators used the low-exposure category as the reference category to "minimize the
impact of any unmeasured confounding variables since nonexposed workers may differ from
exposed workers with respect to socioeconomic characteristics" fHauptmann et al.. 20041. A 2-year
lag interval was used to determine exposures to account for a latency period for LHP cancers.
The log-linear trend tests conducted by Beane Freeman et al. (2009) used exposure as a
continuous variable (except for peak exposure, for which categorical ranks were used) (general
model form: RR = ePx, where (3 represents the regression coefficient and X is exposure). As shown
by Callas et al. (1998). the Poisson regression model converges to the Cox proportional hazards
model as the age strata are made infinitely small, and when age is well characterized and adjusted
for, as it was in the Beane Freeman et al. (2009) Poisson regression model, these two models yield
essentially the same RR point estimates and CIs.
Dr. Beane Freeman provided EPA with the regression coefficient estimates from the
log-linear trend test models for cumulative exposure for several LHP cancer subtype groupings.
These estimates are presented in Table 2-30. As with the NPC calculations, the nonexposed person-
years were included in the primary unit risk estimate derivations and other quantitative
approaches to be more inclusive of all the exposure-response data. Results for the exposed person-
years only are presented for some of the unit risk estimates for comparison.
Table 2-29. Relative risk estimates for mortality from multiple myeloma
(ICD-8 code 203), leukemia (ICD-8 codes 204-207), myeloid leukemia (ICD-8
code 205), and other/unspecified leukemia (ICD-8 code 207) by level of
formaldehyde exposure for different exposure metrics
Cancer type
Relative risk (number of deaths)
p-Trend
All person-
years3
Exposed
onlyb
Peak exposure (ppm)
0
>0 to <2.0C
2.0 to <4.0
>4.0
Multiple myeloma
2.74 (11)
1.0 (14)
1.65 (13)
2.04 (21)
>0.50
0.08
Leukemia
0.59 (7)
1.0 (41)
0.98 (27)
1.42 (48)
0.02
0.12
Myeloid leukemia
0.82 (4)
1.0 (14)
1.30 (11)
1.78 (19)
0.07
0.13
This document is a draft for review purposes only and does not constitute Agency policy.
2-84 DRAFT-DO NOT CITE OR QUOTE
-------
Toxicological Review of Formaldehyde—Inhalation
Cancer type
Relative risk (number of deaths)
p-Trend
All person-
years3
Exposed
onlyb
Other/unspecified
leukemia
0.61(2)
1.0 (13)
0.86 (8)
1.15 (13)
0.50
>0.50
Average Intensity (ppm)
0
>0 to <0.5C
0.5 to <1.0
>1.0
Multiple myeloma
2.18 (11)
1.0 (25)
1.40 (11)
1.49 (12)
>0.50
>0.50
Leukemia
0.54 (7)
1.0 (67)
1.13 (25)
1.10 (24)
0.50
>0.50
Myeloid leukemia
0.70 (4)
1.0 (24)
1.21(9)
1.61 (11)
0.40
0.43
Other/unspecified
leukemia
0.58 (2)
1.0 (21)
0.98 (7)
0.84 (6)
>0.50
>0.50
Cumulative Exposure (ppm x years)
0
>0 to <1.5C
1.5 to <5.5
>5.5
Multiple myeloma
1.79 (11)
1.0 (28)
0.46 (5)
1.28 (15)
>0.50
>0.50
Leukemia
0.53 (7)
1.0 (63)
0.96 (24)
1.11 (29)
0.08
0.12
Myeloid leukemia
0.61(4)
1.0 (26)
0.82 (8)
1.02 (10)
0.44
>0.50
Other/unspecified
leukemia
0.77 (2)
1.0 (15)
1.65 (10)
1.44 (9)
0.13
0.15
aLikelihood ratio test (1 degree of freedom) of zero slope for formaldehyde exposure (continuous variable, except
for peak exposure metric) among all (nonexposed and exposed) person-years.
bLikelihood ratio test (1 degree of freedom) of zero slope for formaldehyde exposure (continuous variable, except
for peak exposure metric) among exposed person-years only.
Reference category for all categories with the same exposure metric.
Source: Beane Freeman et al. (2009)
Table 2-30. Regression coefficients for leukemia, myeloid leukemia, and
myeloid plus other/unspecified leukemias mortality from NCI trend test
models of cumulative exposure3
Cancer type
Person-years
P (per ppm x years)
Standard error
(per ppm x years)
Leukemia
All
0.01246
0.006421
Exposed only
0.01131
0.00661
Myeloid leukemia
All
0.009908
0.01191
Exposed only
0.008182
0.01249
Myeloid leukemia plus
other/unspecified leukemiab
All
0.01408
0.007706
Exposed only
0.01315
0.007914
This document is a draft for review purposes only and does not constitute Agency policy.
2-85 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
aModels were stratified by calendar year, age, sex, and race and adjusted for pay category; exposures included a
2-year lag interval.
bp-trend values for the myeloid and other/unspecified leukemia categories combined are 0.10 for all person-years
and 0.13 for exposed person-years only.
Source: Personal communications from Laura Beane Freeman to John Whalan (August 26, 2009) and to Jennifer
Jinot (February 21, 2014).
Approaches Used for Quantitative Risk Assessment of Myeloid Leukemia
As discussed above, cumulative exposure, which incorporates both exposure intensity and
duration, is the preferred exposure metric for the evaluation of long-term exposure to chemicals
and effects on cancer, and it is the exposure metric of choice for the estimation of cancer risks in
this assessment EPA explored several approaches for deriving a unit risk estimate for myeloid
leukemia based on cumulative exposure.
EPA considered a standard approach for deriving the unit risk estimate using the regression
coefficient for myeloid leukemia and cumulative exposure; however, the p-value (0.44) for that
regression coefficient was far from 0.05, indicating a poor model fit The poor model fit could be
due, at least in part, to inadequate statistical power, likely exacerbated by the underreporting of
myeloid leukemia deaths suggested by the analyses by Percy etal. (1990; 19811. Table 2-30 shows
that the regression coefficient for all person-years for myeloid leukemia is only slightly lower than
that for all leukemia, which had a lower p-value of 0.08 and which should include all the myeloid
leukemia deaths, both specified and unspecified. The "other/unspecified" leukemias comprise a
sizable portion of all leukemia deaths (almost 30%) in the cohort and presumably include a good
proportion of unclassified myeloid leukemias. The results of two NCI studies done at different
times to evaluate the accuracy of death certificates by comparing the underlying cause of death on
death certificates to original hospital diagnoses suggest that a third to a half of leukemias not
otherwise specified on death certificates were diagnosed as myeloid leukemias in the hospital
(Percy etal.. 1990: Percy etal.. 1981).53 Thus, two additional approaches for deriving a unit risk
estimate for myeloid leukemia, which attempted to address the underreporting of myeloid
leukemias, were considered.
One approach involved using the all leukemia grouping.54 Use of the all leukemia
background rates in the life-table calculations (described in more detail below) might inflate the
53In the Percy et al. (1990:1981) studies, only about 10% of leukemia deaths were classified as "other or
unspecified" based on hospital diagnoses [versus 29% from death certificates in the Beane Freeman et al.
(2009) study], and 51% (Percy et al.. 1981) and 53% (Percy et al.. 1990) of leukemia deaths were myeloid
leukemias based on hospital diagnoses [versus 39% from death certificates in the Beane Freeman et al.
(20091 study], suggesting that about a third or more of the "other or unspecified" leukemia deaths in the
Beane Freeman et al. (2009) study were probably myeloid leukemias. Percy et al. (1990) reported in their
study that "Of the nearly 600 deaths from leukemia NOS [other or unspecified] nearly 50% were originally
diagnosed as myeloid... Obviously myeloid leukemia is grossly underreported on death certificates."
54The all leukemia category includes all 123 leukemias observed in the cohort. Of these, 48 (39.0%) were
myeloid, 37 (30.1%) were lymphoid, and 36 (29.3%) were other/unspecified; the remaining 2 (1.6%) were
monocytic leukemias (ICD-8 code 206).
This document is a draft for review purposes only and does not constitute Agency policy.
2-86 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Toxicological Review of Formaldehyde—Inhalation
unit risk estimate by increasing the background risk relative to which the formaldehyde-related
risks are calculated. However, the inclusion of any leukemia subtypes not related to formaldehyde
exposure should theoretically dampen the exposure-response relationship (lowering the regression
coefficient) relative to that for all the myeloid leukemias alone; thus, this should mitigate at least
some of the effect of using the all leukemia background rates.
The preferred approach involved using a combined grouping of the myeloid leukemia and
other/unspecified leukemias subcategories. The myeloid and other/unspecified leukemias
grouping had a stronger association with cumulative exposure (p-trend = 0.10 for all person-years)
in the Beane Freeman et al. (2009) study than did myeloid leukemia alone and it captures the
unclassified myeloid leukemias with the least inclusion of nonmyeloid leukemias. There is likely
more uncertainty associated with the background rates for the other/unspecified leukemias than
for the specified myeloid and lymphocytic leukemia subtypes (discussed further below); however,
the benefits of focusing on the myeloid plus other/unspecified leukemias rather than the broader
"all leukemias" grouping in attempting to be more inclusive of all the myeloid leukemias were
deemed to outweigh any additional uncertainty associated with the background rates.
Although the unit risk estimate based on the preferred approach of using myeloid plus
other/unspecified leukemias inevitably includes some nonmyeloid leukemias, it is considered the
best approach for deriving a unit risk estimate for myeloid leukemia specifically.55 Results for all
the approaches will be presented for comparison, and it will be apparent that the different
approaches yield similar unit risk estimates. Because the purpose in presenting the results from
the various approaches is to compare relative quantitative differences across the different
approaches, not all the sensitivity analyses that would be presented in a final assessment were
performed for each approach (e.g., performing comparison analyses based on exposed person-
years only).
Prediction of Lifetime Extra Risk of Myeloid Leukemia Mortality and Incidence
Lifetime extra risk estimates for myeloid leukemia mortality were calculated from the
regression results using the different approaches discussed above and the same general
methodology described for the NPC mortality estimates. U.S. age-specific 2006 all-cause mortality
rates fNCHS. 20091 were used in the life-table programs. For the cause-specific background
mortality rates, NCHS age-specific 2006-2010 mortality rates for all race and sex groups combined
were used for all leukemia
55Although the inclusion of cancer subtypes not necessarily causally associated with the chemical exposure in
the grouping of cancers represented in the regression coefficient and the corresponding background rates for
the life-table analysis is overt here, it is not uncommon that, due to data limitations, unit risk estimates based
on human data reflect cancer groupings broader than what might be strictly causally associated with the
chemical exposure (e.g., all leukemias, or all lung cancers). As noted in the text, any inclusion of unassociated
cancer subtypes in the derivation of the regression coefficient should theoretically attenuate the coefficient in
a manner that would offset the use of the unassociated subtypes in the background rates in the life-table
analysis.
This document is a draft for review purposes only and does not constitute Agency policy.
2-87 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
(http://seer.cancer.gov/csr/1975_2010/results_merged/sect_13_leukemia.pdf] and NCHS (20061
age-specific mortality rates were used for myeloid leukemia (ICD-10 C92) and for
other/unspecified leukemias (C94-C95) fNCHS. 20061. In addition, a 2-year lag period was used, as
selected by Beane Freeman et al. (2009).
The resulting point estimates and one-sided 95% UCLs for the extra risk of myeloid plus
other/unspecified leukemias are shown in Table 2-31. The model predicts extra risk estimates that
are fairly linear for exposures below about 0.01-0.1 ppm (0.012-0.123 mg/m3) but not for
exposures above 0.1 ppm (0.123 mg/m3).
Table 2-31. Extra risk estimates for myeloid plus other/unspecified leukemia
mortality from various levels of continuous lifetime exposure to formaldehyde
Exposure concentration (ppma)
Extra risk
95% UCL on extra risk
0.0001
1.32 x 10"6
2.51 x 10"6
0.001
1.32 x 10"5
2.51 x 10"5
0.01
1.34 x 10"4
2.58 x 10"4
0.1
1.59 x 10"3
3.38 x 10"3
1
8.40 x 10"2
6.26 x 10"1
10
9.81 x 10"1
9.90 x 10"1
aValues used in the derivation of the unit risk estimate are presented in ppm throughout this section. To convert
from ppm to mg/m3, lppm = 1.23 mg/m3.
Although the background mortality rates of leukemia are higher (lifetime risk of 0.0062
according to the life-table analysis) than those of NPC, the 1% extra risk level typically used as the
basis for the POD for epidemiological data still corresponds to an RR estimate (2.5) that would be
above the highest categorical result reported, even after adjusting the RR estimates upward relative
to the 0-exposure group (because our primary analyses include the nonexposed workers). A 0.5%
extra risk level yields an RR estimate of 1.8, which better corresponds to the RRs in the range of the
data. Thus, the LEC value corresponding to 0.5% extra risk (LECoos) was selected for the POD for all
leukemia and for myeloid leukemia and myeloid plus other/unspecified leukemias, which have
lower background rates than all leukemia (lifetime risks of 0.0031 and 0.0046, respectively).
There are insufficient data to establish the MOA(s) for formaldehyde-induced myeloid
leukemia; thus, linear low-dose extrapolation was performed as the default approach, in
accordance with EPA's Guidelines for Carcinogen Risk Assessment fU.S. EPA. 2005al. The ECoos,
LECoos, and IUR estimates for myeloid plus other/unspecified leukemia mortality are presented in
Table 2-32.
This document is a draft for review purposes only and does not constitute Agency policy.
2-88 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
Table 2-32. ECoos, LECoos, and inhalation unit risk estimates for myeloid plus
other/unspecified leukemia mortality from formaldehyde exposure based on
log-linear trend analyses of cumulative exposure data from the Beane
Freeman et al. (2009) study
Person-years
ECoos (ppm)
LECoos (ppm)
Unit risk3 (per ppm)
Unit risk (per mg/m3)
All
0.253
0.133
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Toxicological Review of Formaldehyde—Inhalation
some age groups the mortality rates are actually larger than the incidence rates. This irregularity is
to be expected for "other/unspecified" classifications because greater attention is given to
diagnosing incident leukemia cases than to accounting for causes of death, so one would anticipate
less underreporting of myeloid leukemias as incident cases than as causes of death on death
certificates.
Table 2-33. ECoos, LECoos, and inhalation unit risk estimates for myeloid plus
other/unspecified leukemia incidence from formaldehyde exposure based on
Beane Freeman et al. (20091 log-linear trend analyses for cumulative
exposure
Person-years
ECoos (ppm)
LECoos (ppm)
Unit risk3 (per ppm)
Unit risk (per mg/m3)
All
0.224
0.118
4.2 x 10~2
3.4 x 10~2
Exposed only
0.239
0.120
4.2 x 10"2
3.4 x 10"2
aUnit risk = 0.005/LECoos.
The ECoos and LECoos estimates for mortality and incidence and incidence unit risk estimates
for all leukemia and for myeloid leukemia using the alternate approaches discussed above are
presented in Table 2-34. The same underlying life-table methodology was used for each of these
approaches—only the regression coefficients and background cancer rates differed. As discussed
above, and consistent with the results just presented, the preferred approach (shaded in Table 2-33
to 2-35) is the life-table analysis using the regression coefficient and background rates for myeloid
plus other/unspecified leukemias because this grouping captures the unclassified myeloid
leukemias with the least inclusion of nonmyeloid leukemias.
Table 2-34. ECoos and LECoos estimates for mortality and incidence and
incidence unit risk estimates for all leukemia and for myeloid leukemia using
alternate approaches (all person-years)
Approach (by cancer type used as
basis for regression coefficient and
cause-specific background rates)
ECoos (ppm)
LECoos (ppm)
Unit risk estimate
(per ppm)a
Unit risk estimate
(per mg/m3)
Incidence
Mortality
(Incidence)
(Incidence)
Myeloid leukemia
0.378
0.127
0.468
0.157
3.9 x 10"2
3.2 x 10"2
All leukemia
0.156
0.0846
0.229
0.124
5.9 x 10"2
4.8 x 10"2
Myeloid + Other/Unspecifiedb
0.224
0.118
0.253
0.133
4.2 x 10"2
3.4 x 10"2
Note: Shaded estimate is preferred.
aUnit risk estimate = 0.005/(LECoo5 for incidence).
incidence background rates also include monocytic leukemia, but that contribution is negligible.
This document is a draft for review purposes only and does not constitute Agency policy.
2-90 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Toxicological Review of Formaldehyde—Inhalation
The preferred unit risk estimate for myeloid leukemia is the estimate of 4.2 x 10~2 per ppm
(3.4 x 10"2 per mg/m3) derived using incidence rates (and regression coefficient) for myeloid plus
other/unspecified leukemias, for all (exposed and nonexposed) person-years.57 The results from
the exposed person-years only are essentially indistinguishable (see Table 2-33). The unit risk
estimates from the other approaches considered are fairly close, with the unit risk estimate based
on the myeloid leukemia category being virtually identical to the preferred estimate based on
myeloid plus other/unspecified leukemias and the estimate based on all leukemia being somewhat
greater (see Table 2-34).
Table 2-35 summarizes some of the key information comparing the different approaches
considered for the derivation of the unit risk estimate for myeloid leukemia.
Table 2-35. Exposure-response modeling (all person-years) and (incidence)
unit risk estimate derivation results for different leukemia groupings
Cancer grouping
Number of
deaths in NCI
cohort
Regression
coefficient
(per
ppm x year)
SE
(per
ppm x year)
p-Value
Unit risk
estimate
(per ppm)
Unit risk
estimate
(per mg/m3)
Myeloid leukemia
48
0.009908
0.01191
0.44
3.9 x 10"2
3.2 x 10"2
All leukemia
123
0.01246
0.006421
0.08
5.9 x 10"2
4.8 x 10"2
Myeloid +
Other/Unspecified
leukemias
84a
0.01408
0.007706
0.10
4.2 x 10"2
3.4 x 10"2
Note: Shaded estimate is preferred.
aThis is the sum of the leukemias classified as myeloid and those classified as "other/unspecified". At least 70-80%
of this number is expected to be myeloid leukemias, assuming that a third to a half of leukemias not otherwise
specified on death certificates are myeloid leukemias, as discussed above.
In summary, as discussed above, EPA explored several approaches for deriving a unit risk
estimate for myeloid leukemia based on cumulative exposure. The first approach involved using
the grouping of leukemias classified as myeloid leukemia on the death certificate. The regression
coefficient for this grouping had a p-value (0.44) indicative of a poor model fit It was reasoned that
"Comparable to calculations done for NPC above, a rough calculation was done to ensure that the unit risk
estimate derived for myeloid leukemia incidence is not implausible in comparison to actual case numbers.
For example, assuming an average constant lifetime formaldehyde exposure level of 20 ppb for the U.S.
population, the inhalation unit risk estimate for myeloid (and other/unspecified) leukemia equates to a
lifetime extra risk estimate of 8.4 x 1CM. Assuming an average lifetime of 75 years (this is not EPA's default
average lifetime of 70 years, but rather a value more representative of actual demographic data) and a U.S.
population of 300,000,000, this lifetime extra risk estimate suggests a crude upper-bound estimate of
3,400 incident cases of myeloid leukemia attributable to formaldehyde exposure per year. This upper-bound
estimate is well below the estimated 17,100 total incident myeloid leukemia (not including other/unspecified
leukemias) cases per year calculated from the SEER myeloid leukemia incidence rate of 5.7/100,000 (age-
adjusted incidence rate for AML and CML combined from 2008-2012 SEER-18 data; www.seer.cancer.gov).
This document is a draft for review purposes only and does not constitute Agency policy.
2-91 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Toxicological Review of Formaldehyde—Inhalation
the poor model fit could be due, at least in part, to the underreporting of myeloid leukemia deaths
discussed above. It can be seen in Table 2-35 that the regression coefficient for myeloid leukemia is
only slightly lower than that for all leukemia, which had a lower p-value of 0.08 and should include
all the myeloid leukemia deaths, both specified and unspecified. Thus, a second approach involved
using the all leukemia grouping, which includes other subtypes likely not associated with
formaldehyde exposure. The preferred approach involved using the combined grouping of the
myeloid leukemia and other/unspecified leukemias subcategories. The myeloid and
other/unspecified leukemias grouping had a stronger association with cumulative exposure
(p = 0.10) in the Beane Freeman et al. (2009) study than did myeloid leukemia alone and it captures
the unclassified myeloid leukemias with the least inclusion of nonmyeloid leukemias. The benefits
of focusing on the myeloid plus other/unspecified leukemias rather than the broader "all leukemia"
grouping in attempting to be more inclusive of all the myeloid leukemias were deemed to outweigh
any additional uncertainty associated with the background rates for the other/unspecified
leukemias (discussed further below). It is reassuring that the unit risk estimates from the three
different approaches are quite similar, with the preferred estimate based on myeloid plus
other/unspecified leukemias being essentially identical to the estimate based on the myeloid
leukemia category and both those estimates being about two-thirds of the estimate for all leukemia.
Uncertainties and Confidence in the Preferred Unit Risk Estimate for Myeloid Leukemia
The strengths and uncertainties in the unit risk estimate for myeloid leukemia incidence are
summarized in Table 2-36. The primary uncertainty in this estimate relates to the complexities in
the study-specific data for cumulative formaldehyde exposure and mortality from myeloid
leukemia.
Table 2-36. Strengths and uncertainties in the cancer type-specific unit risk
estimate for myeloid leukemia
Strengths
Uncertainties
• IIIR estimated from
data that is directly
relevant to humans.
• Based on the results of
a large, high
confidence
epidemiological study
involving multiple
industries with
detailed, individual
cumulative exposure
estimates and
allowance for cancer
latency.
• Uncertainties with a potentially greater impact:
o Although the dose-response relationship with peak exposure was marginally
significant (p = 0.07), and statistically significant associations were reported
for several metrics of exposure in other studies, the reported relationship
with cumulative exposure showed a nonsignificant, small increase in risk for
myeloid leukemia (based on the regression coefficient for the continuous
model), potentially due in part to misclassification of myeloid leukemia cases.
o The association with cumulative exposure was stronger for the
other/unspecified grouping of leukemia diagnoses (N = 36) than for myeloid
leukemia alone (N = 48). Although a sizable proportion of this category is
assumed to include myeloid leukemia cases, the stronger association is
surprising given the more heterogeneous set of leukemia cases in this
category, some presumably not associated with formaldehyde exposure.
Hence, the association would be expected to be attenuated.
This document is a draft for review purposes only and does not constitute Agency policy.
2-92 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Toxicological Review of Formaldehyde—Inhalation
Strengths
Uncertainties
• Moderate number of
deaths to model
(N = 84).
o Uncertainty about optimal exposure metric(s). Use of cumulative exposure
assumes equal importance of concentration and duration on cancer
incidence. The specific metrics analyzed differed across studies, and the
results of the NCI study were not completely consistent with those of other
studies (associated only with peak exposure).
o Uncertainties likely to have a minor impact:
o Grouping of myeloid leukemias used for exposure-response modeling
includes nonmyeloid leukemias.
o Borderline model fit for myeloid plus other/unspecified leukemias (p = 0.1)
and uncertain shape of exposure-response function.
Based on the attendant strengths and uncertainties outlined above, there is low confidence
in the unit risk estimate for myeloid leukemia incidence. However, given the strength of the
evidence integration judgment (i.e., evidence demonstrates formaldehyde inhalation causes
myeloid leukemia in humans), and the associated public health burden that it poses (e.g., myeloid
leukemia is far more prevalent than NPC), EPA thoroughly considered the complexity in the data
and used an innovative approach to derive and present a potential unit risk estimate for myeloid
leukemia. A charge question will be provided for the peer-review panel regarding the development
of a unit risk estimate for myeloid leukemia and asking for advice about how, if at all, the unit risk
estimate might inform the quantification of risk for cancer. This uncertainty is discussed further in
the summary section below.
2.2.3. Summary of Unit Risk Estimates and the Preferred Estimate for Inhalation Unit Risk
Table 2-37. Inhalation unit risk estimates by cancer type based on human
data3
Cancer subtype
Unit risk estimate (per ppm)
Unit risk estimate (per mg/m3)
Mortality
Incidence
Mortality
Incidence
Nasopharyngeal
4.5 x 10"3
9.1 x 10"3
3.7 x 10"3
7.4 x 10"3
Myeloid leukemiab
3.8 x 10"2
4.2 x 10"2
3.1 x 10"2
3.4 x 10"2
aBased on entire cohort (exposed and unexposed).
bBased on myeloid plus other/unspecified leukemias.
The unit risk estimates for NPC and myeloid leukemia derived using data from the NCI
occupational cohort are summarized in Table 2-37. As discussed previously, the NPC unit risk
estimate based on data from the human occupational epidemiology study of the NCI updated by
Beane Freeman et al. (2013.) was preferred over estimates based on rodent cancer bioassay data,
although these estimates were very similar (Table 2-27). The best estimate that could be
This document is a draft for review purposes only and does not constitute Agency policy.
2-93 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Toxicological Review of Formaldehyde—Inhalation
developed for myeloid leukemia was also derived from the human occupational epidemiology study
of the NCI updated by Beane Freeman et al. (2009). However, the data reported for myeloid
leukemia fBeane Freeman et al.. 20091 are complex and there are reasons for and against the use of
these data in the derivation of the IUR. Given the the strength of the evidence integration judgment
(i.e., evidence demonstrates formaldehyde inhalation causes myeloid leukemia in humans), and
the associated public health burden that it poses (e.g., myeloid leukemia is far more prevalent than
NPC), EPA thoroughly considered the complexity in the data and used an innovative approach to
derive and present a potential unit risk estimate for myeloid leukemia. Some important
uncertainties are discussed in greater detail below.
• Despite the quality of the literature base for the formaldehyde assessment and the
confidence in the qualitative hazard information for myeloid leukemia, the only study
suitable for dose-response quantification for myeloid leukemia may be viewed as
insufficient for developing a quantitative estimate of risk with an acceptable level of
confidence.
o The Beane Freeman study (2009) failed to observe an association between
cumulative formaldehyde exposure and myeloid leukemia (p = 0.44), despite a
reasonable number of cases (n = 48) and adequate follow-up. The peak exposure
metric was marginally associated (p = 0.07). This result raises questions about the
relative importance of the intensity of exposure and duration in the association of
myeloid leukemia mortality. On the other hand, myeloid leukemia mortality
increased with TSFE, cumulative exposure, and exposure duration in two other
occupational cohorts (garment workers and embalmers).
o The available animal studies do not provide evidence supporting an association
between formaldehyde inhalation and myeloid leukemia. Thus, there are no animal
data that can be used to support the POD estimate that can be derived from the only
suitable human study.
• Analyses from NCI comparing causes of death recorded on death certificates with original
diagnoses in hospital records suggest a misclassification of myeloid leukemia cases (N = 48),
with a significant proportion reported as "other/unspecified" (N = 36).
o In the Percy et al. (1990; 19811 studies, only about 10% of leukemia deaths were
classified as "other or unspecified" based on hospital diagnoses [versus 29% from
death certificates in the Beane Freeman et al. (2009) study], and 51% fPercy etal..
19811 and 53% fPercv etal.. 19901 of leukemia deaths were myeloid leukemias
based on hospital diagnoses [versus 39% from death certificates in the Beane
Freeman et al. (20091 study], suggesting that about a third or more of the "other or
unspecified" leukemia deaths in the Beane Freeman et al. (2009) study were
probably myeloid leukemias. Percy et al. (1990) reported in their study that "Of the
nearly 600 deaths from leukemia NOS (other or unspecified) nearly 50% were
originally diagnosed as myeloid... Obviously myeloid leukemia is grossly
underreported on death certificates."
This document is a draft for review purposes only and does not constitute Agency policy.
2-94 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Toxicological Review of Formaldehyde—Inhalation
o Because it is likely that a proportion of myeloid leukemia cases were reported as
"other/unspecified," a more complete estimate of the association of cumulative
formaldehyde exposure with myeloid leukemia might be obtained using the
regression results for a combination of myeloid leukemia and other/unspecified
leukemia.
o Although a unit risk estimate that combines myeloid leukemia and
other/unspecified leukemia overtly includes cancer subtypes not necessarily
causally related with the chemical exposure, it is sometimes the case that, due to
data limitations, unit risk estimates are based on less directly causal groupings
(e.g., all leukemias, or all lung cancers). The inclusion of unassociated cancer
subtypes in the derivation of the regression coefficient should theoretically
attenuate the association.
o A comparison of the unit risk estimates for all leukemia, myeloid leukemia plus
other unspecified leukemia, and myeloid leukemia (ICD-8/9: 205) indicates that all
of the estimates are within a factor of 1.5. Unit risk estimates were 3.9 x 10~2,
4.2 x 10"2, and 5.9 x 10 2 for all leukemia, myeloid leukemia plus other unspecified
leukemia, and myeloid leukemia (ICD-8/9: 205), respectively.
• The approach for combining myeloid leukemia and other/unspecified leukemia to estimate
risk, while arguably consistent with the identified misclassification of myeloid leukemia on
death certificates fPercv etal.. 1990: Percy etal.. 19811. is uncommon but retains significant
quantitative uncertainties, including some inconsistencies in statistical results.
o The combination of myeloid leukemia and other/unspecified leukemia in the
regression model yields a p-value of 0.1. While the number of cases is increased by
n = 36, cancers in this category, with the exception of the myeloid leukemia cases,
were not identified to be causally related with formaldehyde exposure during the
hazard evaluation. The inclusion of cancers not causally related with formaldehyde
exposure would be expected to attenuate the association, but in contrast to this
expectation, there was a stronger association for the regression model of
other/unspecified leukemia alone (p = 0.13) compared to the model of myeloid
leukemia alone (p = 0.44). There is not a clear explanation for why the association
would be stronger for the more heterogeneous leukemia category.
o There is likely more uncertainty associated with the background cancer rates in the
U.S. population for the other/unspecified leukemias than for the specified myeloid
and lymphocytic leukemia subtypes. The survival rates of the other/unspecified
cancers had to be estimated by subtracting myeloid and lymphocytic leukemia rates
from the rates for all leukemia.
• Given the completely unknown MOA for myeloid leukemia, it is possible, and perhaps likely,
that there are dose and duration effects for the development of myeloid leukemia following
formaldehyde inhalation that are not fully understood.
o Acknowledging the complexity of the different dose metrics available in the
observational studies, as well as the lack of an association between cumulative
exposure and myeloid leukemia in the Beane Freeman study (2009). it is possible
that the specific, individual exposure metrics in this study failed to fully capture the
This document is a draft for review purposes only and does not constitute Agency policy.
2-95 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
patterns of exposure with which the development of myeloid leukemia is causally
related. Importantly, this concern is independent of the identified hazard for
myeloid leukemia, as myeloid leukemia mortality was increased in association with
the peak exposure metric in this study (industrial workers) and others, as well as
with duration-dependent metrics including TSFE, cumulative exposure, and
exposure duration in two other occupational cohorts (garment workers and
embalmers).
o As information supporting a nonlinear extrapolation from the identified POD is not
available for myeloid leukemia, the current approach uses a default linear
extrapolation. It is possible that additional study on the development of this cancer
after formaldehyde exposure could provide support for the linear extrapolation or,
alternatively, support a nonlinear approach.
2.2.4. Adjustment of Human-based Unit Risk Estimates for Potential Increased Early-life
Susceptibility
When there is sufficient weight of evidence to conclude that a mutagenic MOA is operative
in a chemical's carcinogenicity and there are inadequate chemical-specific data to assess age-
specific susceptibility, as is the case for formaldehyde inhalation exposure-induced NPCs (see
Section 1.2.5), EPA guidelines fU.S. EPA. 2005bl recommend the application of default age-
dependent adjustment factors (ADAFs) to adjust for potential increased susceptibility from early-
life exposure. In brief, the supplemental guidelines establishe ADAFs for three specific age groups.
The current ADAFs and their age groupings are 10 for <2 years, 3 for 2 to <16 years, and 1 for
16 years and above (U.S. EPA. 2005b). For risk assessments based on specific exposure
assessments, the 10-fold and three-fold adjustments to the unit risk estimates are to be combined
with age-specific exposure estimates when estimating cancer risks from early-life (<16 years of
age) exposure.
These ADAFs were formulated based on comparisons of the ratios of cancer potency
estimates from juvenile-only exposures to cancer potency estimates from adult-only exposures
from rodent bioassay data sets with appropriate exposure scenarios, and they are designed to be
applied to cancer potency estimates derived from adult-only exposures. Thus, alternate life-table
analyses were conducted for NPC to derive comparable adult-based unit risk estimates to which
ADAFs would be applied to account for early-life exposure. In the NCI Poisson regression model,
the RR estimates are adjusted for age, for the ages represented in the cohort. In deriving lifetime
unit risk estimates, EPA generally extrapolates that relationship and assumes that RR is
independent of age for all ages, for application of the RR exposure-response model across the full
age range (0-85 years) considered in the life-table analysis. For the alternate life-table analyses, it
was assumed that RR is independent of age for adults, which represent the lifestage for which the
exposure-response data and the Poisson regression modeling results from the NCI cohort study
specifically pertain, but that there is increased early-life susceptibility, based on the weight of
evidence-based conclusion that formaldehyde carcinogenicity for NPC has a mutagenic MOA (see
This document is a draft for review purposes only and does not constitute Agency policy.
2-96 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Toxicological Review of Formaldehyde—Inhalation
Section 1.2.5), which supersedes the more general assumption that RR is independent of age for all
ages including children.
In the alternate analyses, exposure in the lifetable was taken to start at age 16 years, the age
cut-point that was established in EPA's supplemental guidelines fU.S. EPA. 2005bl. to derive an
adult-exposure-only unit risk estimate. The adult-exposure-only unit risk estimate, when rescaled
as described below, yields an adult-based unit risk estimate that is comparable to the unit risk
estimate calculated from a typical (i.e., with adult exposures only) rodent bioassay and to which
ADAFs can be applied in the standard way to account for early-life exposure.58 Other than the age
at which exposure was initiated, the life-table analysis is identical to that conducted for the results
presented in Section 2.2.1. Using this approach yields adult-exposure-only unit risk estimates of
3.15 x 10"3 per ppm (2.56 x 10 6 per ng/m3) for NPC mortality and 6.09 x 10~3 per ppm
(4.95 x 10"6 per ng/m3) for NPC incidence; these results are about 70 and 67%, respectively, of the
unit risk estimates derived for lifetime exposure under the assumption of age independence across
all ages.
When EPA derives unit risk estimates from standard rodent bioassay data, there is a
blurring of the distinction between lifetime and adult-only exposures because the relative amount
of time that a rodent spends as a juvenile is negligible (e.g., 9 of 104 weeks <9%) compared to its
lifespan. [According to the supplemental guidelines, puberty begins around 5-7 weeks of age in
rats and around 4-6 weeks in mice (U.S. EPA. 2005b). and Sengupta (2013) suggests that adulthood
in rats typically begins around postnatal day 63.] Thus, when exposure in a rodent is initiated at 5-
8 weeks (most of the way through the juvenile period), as in the standard rodent bioassay, and the
bioassay is terminated after 104 weeks of exposure, the unit risk estimate derived from the
resulting cancer incidence data is considered a unit risk estimate from lifetime exposure, except
when the ADAFs were formulated and are applied, in which case the same estimate is considered to
reflect adult-only exposure. Yet, when adult exposures are considered in the application of ADAFs,
the adult-exposure-only unit risk estimate is pro-rated over the full default human lifespan of
70 years, presumably because that is how adult exposures are treated when a unit risk estimate
calculated in the same manner from the same bioassay exposure paradigm is taken as a lifetime
unit risk estimate.
However, in humans, a greater proportion of time is spent in childhood (e.g., 16 of
70 years = 23%) (and for the purposes of unit risk estimates, exposure is considered to commence
58In this assessment, adult-exposure-only unit risk estimates refer to estimates derived from the life-table
analysis assuming exposure only for ages >16 years. The adult-exposure-only unit risk estimates are merely
intermediate values in the calculation of adult-based unit risk estimates and should not be used in any risk
calculations. Adult-based unit risk estimates refer to estimates derived after rescalingthe
adult-exposure-only unit risk estimates to a (70-year) lifetime, as described later. The adult-based unit risk
estimates are intended to be used in ADAF calculations (U.S. EPA. 2005b) for the computation of extra risk
estimates for specific exposure scenarios. Note that the unit risk estimates in this section, which are derived
under an assumption of increased early-life susceptibility, supersede those that were derived in Section 2.2.1
under the assumption that RR is independent of age.
This document is a draft for review purposes only and does not constitute Agency policy.
2-97 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Toxicological Review of Formaldehyde—Inhalation
at birth), and the distinction between lifetime exposure and adult-only exposure cannot be ignored
when human data are used as the basis for the unit risk estimates. Thus, adult-exposure-only unit
risk estimates were calculated distinct from the lifetime estimates that were derived in
Section 2.2.1 under the assumption of age independence for all ages. In calculating the adult-
exposure-only unit risk estimates, RR is assumed to be independent of age for adulthood. Next, the
adult-exposure-only unit risk estimates need to be rescaled to a 70-year lifespan to be used in the
ADAF calculations and risk estimate calculations involving less-than-lifetime exposure scenarios in
the standard manner, which includes pro-rating even adult-based unit risk estimates over 70 years.
Thus, the adult-exposure-only unit risk estimates are multiplied by 70/54 to rescale the 54-year
adultperiod ofthe 70-year default lifespan to 70 years. Then, for example, if a risk estimate were
calculated for a less-than-lifetime exposure scenario involving exposure only for the full adult
period of 54 years, the rescaled unit risk estimate would be multiplied by 54/70 in the standard
calculation and the adult-exposure-only unit risk estimate would be appropriately reproduced.
Without rescaling the adult-exposure-only unit risk estimates, the example calculation just
described for exposure only for the full adult period of 54 years would result in a risk estimate 77%
(i.e., 54/70) of that obtained directly from the adult-exposure-only unit risk estimates, which would
be illogical. The rescaled adult-based unit risk estimates for NPC mortality and incidence for use in
ADAF calculations and risk estimate calculations involving less-than-lifetime exposure scenarios
are presented in Table 2-38.
Table 2-38. Adult-based unit risk estimates for nasopharyngeal cancer for use
in ADAF calculations and risk estimate calculations involving less-than-
lifetime exposure scenarios
NPC response
Adult-based unit risk estimate
(per ppm)
(per pg/m3)
Mortality
4.08 x 10"3
3.31 x 10"6
Incidence
7.90 x 10"3
6.42 x 10"6
An example calculation illustrating the application of the ADAFs to the human-data-derived
adult-based (rescaled as discussed above) NPC (incidence) unit risk estimate for formaldehyde for
a lifetime exposure scenario is presented below. For inhalation exposures, assuming ppm
equivalence across age groups, i.e., equivalent risk from equivalent exposure levels, independent of
body size, the ADAF calculation is fairly straightforward. Thus, the ADAF-adjusted lifetime NPC unit
risk estimate is calculated as illustrated in Table 2-39.
This document is a draft for review purposes only and does not constitute Agency policy.
2-98 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Toxicological Review of Formaldehyde—Inhalation
Table 2-39. NPC incidence risk from exposure to constant formaldehyde
exposure level of 1 pg/m3 from ages 0 to 70 years
Age group
ADAF
Unit risk
(per pg/m3)
Concentration
(pg/m3)
Duration
adjustment
Partial risk3
0 to <2 years
10
6.42 x 10"6
1
2 yr/70 yr
1.83 x 10"6
2 to <16 years
3
6.42 x 10"6
1
14 yr/70 yr
3.85 x 10"6
>16 years
1
6.42 x 10"6
1
54 yr/70 yr
4.95 x 10"6
Total Lifetime (70 yr) Risk: 1.06 x 10"5
aThe partial risk for each age group is the product of the values in columns 2-5
[e.g., 10 x (6.42 x 10"6) x 1 x 2/70 = 1.83 x 10"6], and the total risk is the sum of the partial risks.
This 70-year risk estimate for a constant exposure of 1 |J.g/m3 is equivalent to a lifetime
unit risk estimate of 1.1 x 10"5 per pg/m3 (1.3 x 10-2 per ppm) for NPC incidence, adjusted for
potential increased early-life susceptibility, assuming a 70-year lifetime and constant exposure
across age groups. Note that because of the use of the rescaled adult-based unit risk estimate, the
partial risk for the >16 years' age group is the same as would be obtained for a 1 |J.g/m3 constant
exposure directly from the adult-exposure-only unit risk estimate of 4.95 x 10~6 per |J.g/m3 that was
presented above, as it should be. Recall that the adult-based unit risk estimate for NPC incidence
for use in ADAF calculations and risk estimate calculations involving less-than-lifetime exposure
scenarios is 6.42 x 10 6 per |J.g/m3 (7.90 x 10 3 per ppm).
In addition to the uncertainties discussed in Section 2.2.1 for the IUR estimates based on
human data, there are uncertainties in the application of ADAFs to adjust for potential increased
early-life susceptibility. The ADAFs reflect an expectation of increased risk from early-life exposure
to carcinogens with a mutagenic MOA (U.S. EPA. 2005b). but they are general adjustment factors
and are not specific to formaldehyde. Overall, the application of ADAFs to the NPC unit risk
estimate could be overestimating or underestimating the true extent of any increased early-life
susceptibility in the total cancer unit risk estimate, although the quantitative impact of this source
of uncertainty is likely to be small.
2.2.5. Cancer Risk Based on Background Cancer Incidence and Internal Dose of Endogenous
and Exogenous Formaldehyde
EPA has considered estimates derived by Swenberg et al. (20111 and Starr and Swenberg
(2016) that are referred to by the authors as a "bottom-up" approach, to bound low-dose human
cancer risks from formaldehyde exposure in a manner that only uses information regarding
background incidence in the U.S. population of nasopharyngeal cancers (NPC), leukemia, and
Hodgkin lymphoma; background (endogenous) metrics of internal formaldehyde dose in laboratory
animals; and exogenous exposure to formaldehyde expressed in terms of an internal dose. The
results in Starr and Swenberg (2016) are updates, based on newer data, to those presented earlier
This document is a draft for review purposes only and does not constitute Agency policy.
2-99 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Toxicological Review of Formaldehyde—Inhalation
in (Starr and Swenberg. 20131: however, the approach remains unchanged. Estimates using this
approach are presented by the authors as providing a bounding "check" on risk estimates derived
from high-dose data fStarr and Swenberg. 20131.
The data for the internal dose in these calculations were obtained from measurements in
rats and monkeys of formaldehyde-induced DNA adducts experiments based on a highly sensitive
mass spectrometry (MS) method using [13CD2]-formaldehyde (Yu etal.. 2015a: Lu etal.. 2011:
Moeller etal.. 2011: Lu etal.. 2010a). The authors of these experiments conclude that their method
can be used to distinguish whether formaldehyde-induced hydroxymethyl-DNA monoadducts, in
particular the N2-hydroxymethyl-dG (N2-hmdG) adduct, originate from endogenous or exogenous
sources of formaldehyde. The experiments quantified these mono adducts formed from both
sources in various tissues of rats and monkeys: nasal cavity, bone marrow, mononuclear WBCs,
spleen, and thymus (rats); nasal cavity and bone marrow (monkeys). These adduct measurements
and data on the background incidences of NPC, Hodgkin lymphoma, and leukemia in the U.S.
population were then used (Starr and Swenberg. 20161 to develop cancer risk estimates by
attributing all the background incidences to endogenous formaldehyde, using the measured
endogenous N2-hmdG adducts formed by formaldehyde in specific tissues as a biomarker of
exposure. Their risk model assumes a linear relation between cancer incidence and N2-hmdG
adduct levels over the concentration range of endogenous adducts as well as in the low-exposure
range for exogenous adducts.
Risk estimates from this approach are claimed by the authors to produce conservative
upper bounds primarily on the grounds that: (a) the method attributes all of the background risks
of specific cancers to endogenous formaldehyde (based on N2-hmdG adducts); (b) lower confidence
bounds on measured adduct levels are used; and (c) a linear relation is assumed between cancer
incidence and N2-hmdG adduct levels over the endogenous range as well as in the low-exposure
range of interest for exogenous exposure.
Swenberg et al. (2011) and Starr and Swenberg (20161 then compared these values with
the risk estimates in EPA's 2010 draft Toxicological Review, which were obtained by linearly
extrapolating to lower doses from a POD (a lower bound on the concentration associated with the
benchmark response) derived by dose-response modeling of the epidemiological data. When
adduct data from rats were used, the estimates Swenberg and Starr estimated at 1 ppm (1.23
mg/m3) exposure concentration were 2.67 x 10 4 for nasal cancer (based on Yu et al.. 2015a) and
were at most 12.6 x 10 4 for leukemia (based on the limit of detection. LOP, from Lu et al..
2010a), since no exogenous adducts were detected in bone marrow. In monkeys (Yu et al.. 2015a),
the Swenberg and Starr bottom-up estimates were 2.69 x 10 4 for NPC and were less than
1.24 x 10"6 for leukemia. In comparison, the EPA upper-bound risk estimates were higher than the
adduct-based upper-bound estimates by 40-fold for NPC and at least 45-fold (rat adduct data) or
over 45,000-fold (monkey adduct data) for leukemia.
There is considerable uncertainty in extrapolating downward from high-dose animal or
occupational data, particularly in the case of a dose-response that is highly curvilinear; thus, an
This document is a draft for review purposes only and does not constitute Agency policy.
2-100 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
approach that allows an upward linear extrapolation in lieu of the traditional downward
extrapolation is appealing. The bottom-up approach uses cancer incidence in the general
population and is independent of the tumor dose-response data (other than to identify the type of
tumors of concern for analysis); therefore, it can potentially provide a perspective on the likely
contribution of a specific MOA and on the uncertainty in risk estimates derived from higher dose
data where other phenomena such as significant cytotoxicity and impact on DNA repair prior to
mutations may be occurring.
An evaluation of this bottom-up approach identifies scenarios under which this approach
will yield an underestimate of the total (endogenous plus exogenous) risk for a specific cancer type
fCrump etal.. 20141 (and elaborated further in Appendix B.2.3), leading EPA to conclude that the
method does not necessarily provide an upper bound on the slope of the dose-response at low
exogenous exposures, fStarr and Swenberg. 20131 note that the bottom-up approach is based on
applying the concept of additivity to background disease processes (Crump etal.. 1976). However,
this concept of additivity to background only assumes local linearity in the proximity of zero
exogenous dose to be reasonable, while the bottom-up approach assumes linearity over a large
dose range; in particular, the bottom-up approach assumes a linear dose-response below zero
exogenous dose, which is not required in the concept of additivity to background. As a result, it is
unclear if, overall, the bottom-up approach results in a conservative bound on risk, given that
extrapolation upwards in a sublinear dose-response would underestimate risk and underestimate
the slope of the dose-response curve at higher doses. This is further discussed and illustrated in
Crump et al. (2014). Furthermore, the bottom-up approach assumes direct interaction of inhaled
formaldehyde with a particular target tissue; if other sites of interaction and mechanisms are
involved, the measures of DNA adducts in a specific tissue could lead to underestimates of the
cancer potency when utilizing the "bottom-up" approach. In view of these problems, the bottom-up
approach is not carried forward in the candidate unit risks presented in this assessment.
2.2.6. Preferred Inhalation Unit Risk Estimate
The preferred IUR, summarized in Table 2-40, reflects the estimate for NPC incidence alone.
The NPC unit risk estimates are based on the modeling results of the association of cumulative
formaldehyde exposure with NPC mortality in an occupational cohort followed by the NCI (Beane
Freeman et al.. 2013). The regression coefficient from the exposure-response model (log-linear
Poisson regression model) was applied to age-specific cancer incidence rates from the SEER
database using life-table methods to estimate the POD from which to derive the (upper-bound) unit
risk estimate. The IUR estimate is typically expressed as the (upper-bound) increase in cancer risk
expected as a function of a change of 1 |ig/m:i.
EPA has concluded that early-life exposure to chemicals that are carcinogenic through a
mutagenic MOA might present a higher risk of cancer than exposure during adulthood fU.S. EPA.
2005b)- In this document, it was determined that formaldehyde-induced carcinogenicity of the
URT is attributable, at least in part, to a mutagenic MOA (see Section 1.2.5). Therefore, the cancer
This document is a draft for review purposes only and does not constitute Agency policy.
2-101 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Toxicological Review of Formaldehyde—Inhalation
unit risk estimate was adjusted by applying age-dependent adjustment factors (ADAFs). Table 2-40
can be used as a template for incorporating the ADAFs when addressing less-than-lifetime exposure
scenarios. For exposure scenarios comprising primarily adult exposures, it may not be worth the
additional complexity of calculating the ADAF-adjusted risk estimates, and one may choose to use
the unadjusted cancer unit risk estimate presented in Table 2-40 with a "c" superscript, to calculate
risk estimates in the standard way (i.e., without application of ADAFs).
Table 2-40. Inhalation unitriska b
Cancer type
Preferred unit risk
estimate
(ppm-1)
ADAF-adjusted
unit risk estimate
(ppm"1)
Preferred unit risk
estimate
((Hg/m3)"1)
ADAF-adjusted
unit risk estimate
((Hg/m3)"1)
Nasopharyngeal
0.0079°
0.013
6.4 x 10"6 c
1.1 X 10"5
aThe inhalation unit risk estimate is typically expressed as the (upper-bound) increase in cancer risk estimated for
an exposure increase of 1 ng/m3.
bThe unit risk estimate is for cancer incidence.
cAdult-based (rescaled) unit risk estimate for NPC intended for the application of ADAFs.
Benchmark Response /Effective Concentration Estimates
For benefits analyses and certain other situations, "central" estimates of risk-per-unit dose
may be preferred over (upper-bound) unit risk estimates. For nonlinear models, the POD-approach
used by EPA for low-dose extrapolation, which is designed to distinguish between dose-response
modeling in the observable range and inferences made about lower doses (U.S. EPA. 2005a) is not
amenable to providing central estimates of risk at lower doses. Instead, the standard practice for
IRIS assessments is to provide linear extrapolations of risk from the central estimate (here, the
effective concentration [EC] estimate, which is the MLE of the exposure concentration associated
with the benchmark response level of risk) corresponding to the POD, which is the lower bound on
the EC (i.e., the LEC estimate). Table 2-41 presents estimates of risk-per-unit dose linearly
extrapolated from the EC (i.e., BMR/EC estimates).
Table 2-41. Summary of BMR/EC estimates3
Cancer type
BMR/EC estimate
(ppm"1)
ADAF-adjusted
BMR/EC estimate13
(ppm"1)
BMR/EC estimate
((Hg/m3)"1)
ADAF-adjusted
BMR/EC estimate13
((Hg/m3)"1)
Nasopharyngeal
0.0046°
0.0076
3.7 x 10"6c
6.2 x 10"6
aThe BMR/EC estimates based on a longitudinal occupational mortality study (Beane Freeman et al., 2013) are all
for cancer incidence. The BMR is 0.0005 extra risk for NPC. The EC value is the exposure concentration
associated with the BMR based on the Poisson regression model and life-table analysis (see Section 2.2.1). The
This document is a draft for review purposes only and does not constitute Agency policy.
2-102 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Toxicological Review of Formaldehyde—Inhalation
ECooos for NPC was calculated from a life-table analysis of adult-exposure-only and then rescaled as discussed for
the adult-based unit risk estimates in Section 2.2.4.
bSee Section 2.2.4 for a discussion of the ADAF adjustments and how to apply the ADAFs for less-than-lifetime
exposure scenarios.
cAdult-based (rescaled) BMR/EC estimate for NPC intended for the application of ADAFs (see Section 2.2.4).
Sources of Uncertainty Associated with the Preferred Unit Risk Estimate
In general, the major areas of uncertainty in unit risk estimates arise from limitations in the
database, e.g., limitations resulting in the need for interspecies and high- to low-dose extrapolation
and limitations in information on human variability, including especially sensitive populations. The
ideal database would provide sufficient data for the direct calculation of robust cancer (incidence)
estimates for the general population at environmental levels of exposure.
The availability of suitable human data from which to derive unit risk estimates eliminates
one of the major sources of uncertainty inherent in most unit risk estimates—the uncertainty
associated with interspecies extrapolation. The NCI study used as the basis for the preferred unit
risk estimate is considered a well-conducted study for the purposes of deriving unit risk estimates.
The NCI study is a large longitudinal cohort study that developed individual worker exposure
estimates using detailed employment histories and formaldehyde concentration measurements. In
addition to the detailed exposure assessment, the study used internal analyses and carefully
considered potential confounding or modifying variables. Moreover, the NCI study comprises a
large cohort that has been followed for a long time. Nonetheless, uncertainties in derived unit risk
estimates are inevitable. The sources of uncertainty related to these limitations include use of a
single study to derive the unit risk estimate, the inability to derive unit risk estimates for all
potential cancer sites, and the derivation of (incidence) unit risk estimates for the general
population from an occupational mortality study.
Overall confidence in the preferred unit risk estimate is medium. Although substantial
uncertainty exists with respect to the low-exposure extrapolation, the estimate is based on human
data from a large, high-quality epidemiological study. Furthermore, the estimate is similar to the
estimate derived from rodent data.
Use of a single study to derive unit risk
Although several studies contributed to the hazard evaluation and causal conclusion for
myeloid leukemia, a major limitation in the human database for formaldehyde is that there was
only one independent59 epidemiology study, the NCI study (Beane Freeman etal.. 2013: Beane
Freeman et al.. 2009). with adequate exposure estimates for the derivation of unit risk estimates, as
discussed above. Although the unit risk estimation from human data used data from one
epidemiological study, it is a large longitudinal cohort study that included workers from 10
59Another study, by Marsh et al. (2007b; 2002:1996). also derived exposure estimates for the individual
workers; however, it examined one of the 10 plants included in the NCI study, and thus, is not an independent
study.
This document is a draft for review purposes only and does not constitute Agency policy.
2-103 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Toxicological Review of Formaldehyde—Inhalation
different industrial plants, in different states, that produced or used formaldehyde in different
products. These factors decrease the likelihood that the results are overly influenced by
uncontrolled confounding related to either location or production process. The NCI study
developed individual worker exposure estimates using detailed employment histories and
formaldehyde concentration measurements. In addition to the detailed exposure assessment, the
study used internal comparisons of risk from exposure and gave careful consideration to potential
confounding or modifying variables. Thus, although the unit risk estimates are based on a single
study, there is relatively high confidence in that study.
Inability to derive unit risk estimates for all potential cancer sites
The IUR is based on results for NPC from the NCI study; however, the NCI study did not
support the computation of unit risk estimates for all the cancer sites with an evidence integration
judgment of evidence demonstrates based on the totality of the evidence.
With the exception of myeloid leukemia, the contribution by these cancers to the total cancer risk
associated with formaldehyde inhalation is unknown. The impact by myeloid leukemia suggested
by the estimated unit risk estimate (myeloid leukemia plus other/unspecified leukemia) might
increase the ADAF-adjusted IUR by almost four-fold.
Derivation of incidence estimates from mortality data
The NCI study is a retrospective mortality study, and cancer incidence data are unavailable
for the cohort. Using mortality risk would markedly underestimate incidence for NPC because
survival for this cancer type is relatively high. This limitation was addressed quantitatively in the
calculation of cancer incidence risk estimates using the dose-response relationships from the
mortality study, although as discussed above, it was necessary to make certain assumptions. It was
assumed that cancer incidence and mortality have the same exposure-response relationship for
formaldehyde exposure, which is reasonable for NPC at the low induction rates observed. Despite
the uncertainties introduced, the incidence-based estimates are believed to be better estimates of
cancer incidence risk than the mortality-based estimates, given the high survival rates for these
cancers. The estimates may under- or overpredictthe true risk, although the quantitative impact
would be relatively low because the incidence estimates are constrained by the relative
incidence:mortality rates and necessarily bounded by the mortality estimates, which are about 50%
of the incidence estimates (see Tables 2-18 and 2-19).
Generalizabilitv of estimates from a worker population
The NCI data represent an industrial worker cohort that is generally healthier than the U.S.
population at large. Therefore, the unit risk estimates derived from the NCI worker cohort data
could underestimate the cancer risk for the general population to an unknown extent, although the
impact is expected to be relatively low for the majority of the population.
This document is a draft for review purposes only and does not constitute Agency policy.
2-104 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Toxicological Review of Formaldehyde—Inhalation
Industrial workers can also differ from the general population in factors other than health
status. In terms of representing the general population in other ways, the NCI cohort was
somewhat diverse, but the workers were predominantly white males (81%), then white females
(12%), black males (7%), and black females (<1%), and they were all adults. Thus, for example,
cancer risk in the general population could be underestimated if females are more susceptible than
males, or overestimated if males are more susceptible than females. The potential for increased
early-life susceptibility is addressed explicitly in Section 2.2.4.
High- to low-dose extrapolation
The availability of human data from this occupational epidemiology study for the derivation
of quantitative cancer risk estimates removes the need to extrapolate from the findings of rodent
bioassays, a major source of uncertainty in most risk assessments. However, another major source
of uncertainty inherent in most unit risk estimates remains—the uncertainty associated with
extrapolation from high (in this case occupational) exposures to lower (environmental or typical
nonoccupational indoor) exposures. One factor contributing to uncertainty in the low dose-
response comes from the potential for endogenous formaldehyde levels in respiratory tissue to
reduce the uptake of the inhaled gas at low doses, as demonstrated in modeling efforts by Schroeter
et al. (2014) and Campbell Jr et al. (2020). This would be expected to result in an overprediction of
the true risk.
Although the actual exposure-response relationship at low-exposure levels is unknown, the
use of linear low-dose extrapolation is supported by evidence that formaldehyde has a mutagenic
MOA for NPC. The linear low-dose extrapolation from the 95% lower bound on the exposure level
associated with the extra risk level serving as the benchmark response is considered to be a
plausible upper bound on the risk at lower exposure levels. Actual low-dose risks may be lower to
an unknown, but possibly substantial (e.g., over an order of magnitude) extent.
Additional Sources of Uncertainty Stemming from the NCI Study and Its Analysis
Other sources of uncertainty arise from the key epidemiological study and its analysis
(Beane Freeman etal.. 2013). including the retrospective estimation of formaldehyde exposures in
the cohort, the modeling of the epidemiological exposure-response data, the exposure metric for
exposure-response analysis, and potential confounding or modifying factors.
Exposure estimates
With respect to exposure estimation, the NCI investigators (Stewart etal.. 1986) conducted
a detailed retrospective exposure assessment to estimate the individual worker exposures.
Formaldehyde exposures were estimated for specific jobs/tasks based on monitoring data,
discussions with workers and plant managers, and assessment by industrial hygienists. Individual
worker estimates were derived for a variety of exposure metrics based on work histories. This
This document is a draft for review purposes only and does not constitute Agency policy.
2-105 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Toxicological Review of Formaldehyde—Inhalation
exposure assessment was a major undertaking, involving over 100 person-months. Hauptmann et
al. (2004) suggested that employment of such a detailed exposure assessment would tend to
minimize exposure misclassification for average and cumulative exposure and duration of exposure
but that peak exposure estimates could be more susceptible to misclassification because they were
defined more qualitatively. In addition, the follow-up study did not account for exposures after
1980. Beane Freeman et al. (2013) suggest that any underestimation of total exposure resulting
from the 1980 cutoff would be small because only 3.5% of all person-years were contributed by
workers who were 65 years or younger and in exposed jobs in 1980 and because exposure levels
were believed to have been much lower after 1980 than in earlier years.
Marsh et al. (1996) also estimated individual worker exposures at one of the 10 plants
(Wallingford, Connecticut) studied by the NCI team. The Marsh et al. (1996) exposure estimates
were about 10-fold lower than those derived by the NCI for the workers at the Wallingford plant.
Marsh et al. (2002) hypothesized that "the NCI used data from several facilities to estimate
exposures in a single facility." However, the NCI investigators maintain that they estimated
exposures for each plant separately. While the exact reasons for such a large discrepancy are
unclear, some differences in the assessment procedures which could have resulted in substantial
differences in the estimates are apparent First, according to Marsh et al. (1996), 91.7% of the
white male Wallingford plant workers were specified as being exposed to formaldehyde in the NCI
study, while only 83.3% were considered to have been exposed in the Marsh et al. (1996) analysis
(it should be noted that these two cohorts of the Wallingford plant are not identical). Second, the
NCI investigators (Stewart etal.. 1987: Stewart etal.. 1986) did their own exposure monitoring at
all the plants, including the Wallingford facility, to standardize the data provided by the plants as
well as to fill data gaps for certain jobs. There is no indication that Marsh et al. (1996) made any
additional measurements themselves. Third, although the Marsh et al. (2002; 19961 papers are not
entirely consistent on this point, those investigators apparently assumed that the job-specific
exposures at the plant were essentially constant over the history of the plant, whereas the NCI
team, based on interviews with plant personnel knowledgeable about equipment and process
changes, assumed that past exposures were higher.
In any event, despite the discrepancies in the absolute exposure values, the relative
exposures for both the Marsh et al. (2002j 19961 and NCI studies, as reflected in the
exposure-response relationships, are less subject to misclassification and are considered to be
reliable. The Wallingford plant is just one of the 10 plants in the NCI study (representing 4,389 of
the 25,619 workers in the NCI cohort), but if the Marsh et al. (1996) exposure estimates, which are
roughly 10-fold lower than the NCI estimates, are closer to the actual exposures for those workers,
then the true potency of formaldehyde could be greater than that suggested by the unit risk
estimates calculated above based on the NCI data. Furthermore, if the NCI exposure values were
significantly overestimated across all 10 plants, then the actual potency could be higher still.
This document is a draft for review purposes only and does not constitute Agency policy.
2-106 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
In summary, EPA has relatively high confidence in the NCI exposure assessment because of
the large effort and high degree of expertise that NCI devoted to developing their detailed exposure
estimates. Nonetheless, errors in retrospective exposure assignments are inevitable, and as a
result, the unit risk estimates based on the NCI study could overpredict or underpredict the true
risks to an unknown extent, although the discrepancy with the independently derived Marsh et al.
(1996) exposure estimates suggests that the risks might be underestimated.
Exposure-response modeling
With respect to the exposure-response model, the log-linear Poisson regression model used
by the investigators (Beane Freeman etal.. 2013: Beane Freeman et al.. 20091 for their trend tests
(i.e., RR = ePx) is generally an appropriate model for the analyses of epidemiological cancer data.60
As discussed above, when age is well characterized and adjusted for, as it was in the NCI study, the
results of the Poisson regression model should be essentially the same as results from the Cox
proportional hazards model (Callas etal.. 19981. The investigators reported efforts to check for
deviations from log-linearity by adding a quadratic term to their models; none of these additional
terms was statistically significant However, the "true" underlying exposure-response relationships
are unknown.
Even if the correct exposure-response model for NPC was known, there would be
substantial uncertainty in estimating the model parameters because there are only 10 NPC deaths
to model. Additionally, a 15-year lag was used for all the NCI solid cancer models. The actual best
lag interval is unknown; the NCI investigators reported that lag intervals between 2 and 20 years
yielded similar results.
Exposure metrics
Another potentially significant source of uncertainty is associated with the exposure
metrics. With the log-linear model used for modeling the occupational data, the peak exposure
metric gave the strongest exposure-response relationship between formaldehyde exposure and
increased risk of NPCs. However, as discussed above, there are limitations in the peak exposure
metric, and it is unclear how to extrapolate RR estimates based on peak exposure estimates to
meaningful estimates of lifetime extra risk of cancer from environmental exposure (i.e., extra risk
from lifetime continuous low-level environmental exposures). The cumulative exposure metric
also yielded nearly statistically significant exposure-response relationships (p = 0.07) and was used
for the cancer risk calculations in this assessment. The "true" exposure metric best describing the
toxicologically relevant dose of formaldehyde for carcinogenesis is unknown. If a peak-exposure
type of metric is the best representative of the toxicologically relevant dose, this suggests that there
are dose-rate effects in the exposure-response relationship for formaldehyde and cancer. If this is
60EPA relied on the results of the NCI exposure-response analyses and did not investigate other possible
exposure-response models beyond those conducted by NCI.
This document is a draft for review purposes only and does not constitute Agency policy.
2-107 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Toxicological Review of Formaldehyde—Inhalation
the case, the unit risk estimates presented here, which are based on a linear low-dose extrapolation,
may overpredictthe true risks to an unknown, but possibly substantial, extent.
Influence of confounding or effect modification
Beane Freeman et al. (2013.) provided a detailed description of their evaluation of potential
confounding and modifying factors in their analyses. The important factors of age, race, sex,
calendar year, and pay category were taken into account in the Poisson regression and trend
analyses. Furthermore, they used the low-exposure person-years, rather than the unexposed
person-years, as their referent group to minimize any potential confounding effects resulting from
differences in socioeconomic or other characteristics between exposed and unexposed workers.
When the slope estimate (i.e., regression coefficient) for the exposed person-years only was used in
the analyses presented here, the unit risk estimate was essentially identical to that calculated from
the slope estimate for all person-years (see Tables 2-18, 2-19, 2-23, and 2-24).
In addition, these investigators evaluated routine respirator use, exposure to formaldehyde-
containing particulates, durations of exposure to 11 other chemicals/substances in the plants
(antioxidants, asbestos, carbon black, dyes and pigments, hexamethylenetetramine, melamine,
phenol, plasticizers, urea, wood dust, and benzene), and duration of employment as a chemist or
laboratory technician. Only 133 workers ever routinely used a respirator fHauptmann etal.. 20031.
RR estimates reportedly did not change substantially when adjusted for exposure to any of the
other 10 chemicals/substances in the NPC (with cumulative exposure) or leukemia analyses (Beane
Freeman et al.. 2013). Only one of the workers who died of NPC was identified as being exposed to
wood dust, a recognized nasopharynx carcinogen. Adjusting for duration of time spent working as
a chemist or laboratory technician did not substantially alter the results for NPC fBeane Freeman et
al.. 20131.
Beane Freeman et al. (2013.) reported that their analyses showed no evidence of plant
heterogeneity for the solid tumor results. In addition, six of the 10 deaths with NPC on the death
certificate were from the Wallingford plant also studied by Marsh et al. (2007c).61 Marsh et al.
(2007b) hypothesized that the excess NPCs in the Wallingford plant could be due to external
employment in metal-working industries. However, as noted by Beane Freeman et al. (2013). when
Marsh et al. (2007b) adjusted for metal-working, the associations of NPC with formaldehyde for
different metrics of exposure did not decrease.
Although smoking data were not available for the cohort, smoking is unlikely to explain the
excesses in NPCs because there was no consistent increase for tobacco-related diseases, including
lung cancer, across the same exposure metrics. No information was available on Epstein-Barr virus
infections, a major risk factor for NPC, in the cohort
61In the previous follow-up of the NCI cohort by Hauptmann et al. (20041.10 NPCs were reported on death
certificates and included in NCI's SMR analyses, but one of these cases was apparently misclassified on the
death certificate, so only nine cases were used to estimate the RRs in the internal comparison analyses; the
misclassified case was not from the Wallingford plant (Beane Freeman et al.. 20131
This document is a draft for review purposes only and does not constitute Agency policy.
2-108 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Toxicological Review of Formaldehyde—Inhalation
In the reporting of the previous follow-up, Hauptmann et al. (20041 noted that each of the
seven formaldehyde-exposed workers who had died of NPC was also exposed to particulates and
neither of the two workers who died of NPC but were not exposed to formaldehyde was exposed to
particulates. Due to the complete collinearity of formaldehyde and particulate exposures, one
cannot estimate the exposure-response slope in workers exposed only to formaldehyde. The
exposure-response relationships observed for formaldehyde within the NCI cohort and the
associations observed between formaldehyde exposure and NPC in workers not exposed to
particulates indicate that there is a formaldehyde effect independent of particulates; however, one
cannot rule out a possible modifying effect of particulates, which might, for example, enhance
delivery of formaldehyde to the nasopharynx.
In summary, uncontrolled confounding could theoretically result in unit risk estimates that
are either under- or overestimated; nevertheless, given the careful consideration paid to potential
confounding, any quantitative impacts are expected to be minimal. However, a possible modifying
effect of particulate exposure on NPC cannot be ruled out, which could overestimate the risk from
formaldehyde alone to an unknown extent
2.2.7. Previous IRIS Assessment: Inhalation Unit Risk
In the previous assessment (last updated in 1991), an inhalation unit risk of 1.3 x 10"5 per
[ig/m3 was developed based on nasal SCCs in F344 rats from Kerns et al. (1983). The data were
modeled from the estimates of the probability of death with tumor and its variance using a
linearized multistage procedure.
This document is a draft for review purposes only and does not constitute Agency policy.
2-109 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Toxicological Review of Formaldehyde—Inhalation
REFERENCES
Abreu. M. d: Neto. AC: Carvalho. G: Casquillo. NY: Carvalho. N: Okuro. R: Ribeiro. GC: Machado. M:
Cardozo. A: Silva. AS: Barboza. T: Vasconcellos. LR: Rodrigues. DA: Camilo. L: Carneiro. L:
Tandre. F: Pino. AY: Giannella-Neto. A: Zin. WA: Correa. LH: Souza. MN: Carvalho. AR. (2016).
Does acute exposure to aldehydes impair pulmonary function and structure? Respir Physiol
Neurobiol 229: 34-42. http://dx.doi.Org/10.1016/i.resp.2016.04.002
Acheson. ED: Barnes. HR: Gardner. Ml: Osmond. C: Pannett. B: Taylor. CP. (1984). Cohort study of
formaldehyde process workers [Letter], Lancet 2: 403. http://dx.doi.org/10.1016/s0140-
6736C84190568-3
Aglan. MA: Mansour. GN. (2018). Hair straightening products and the risk of occupational
formaldehyde exposure in hairstylists. Drug Chem Toxicol 43: 1-8.
http://dx.doi.Org/10.1080/01480545.2018.1508215
Ahmed. S: Tsukahara. S: Tin-Tin-Win-Shwe: Yamamoto. S: Kunugita. N: Arashidani. K: Fuiimaki. H.
(2007). Effects of low-level formaldehyde exposure on synaptic plasticity-related gene
expression in the hippocampus of immunized mice. J Neuroimmunol 186: 104-111.
http://dx.doi.Org/10.1016/i.ineuroim.2007.03.010
Ai L. T. (2019). Endogenous formaldehyde is a memory-related molecule in mice and humans. 2:
446. http://dx.doi.org/10.1038/s42003-019-0694-x
Akbar-Khanzadeh. F: Mlvnek. IS. (1997). Changes in respiratory function after one and three hours
of exposure to formaldehyde in non-smoking subjects. Occup Environ Med 54: 296-300.
http://dx.doi.Org/10.1136/oem.54.5.296
Akbar-Khanzadeh. F: Vaquerano. MU: Akbar-Khanzadeh. M: Bisesi. MS. (1994). Formaldehyde
exposure, acute pulmonary response, and exposure control options in a gross anatomy
laboratory. Am J Ind Med 26: 61-75. http://dx.doi.org/10.1002/aiim.47002601Q6
Albert. M: Garcia. BC: Kuhnert. C: Peschla. R: Maurer. G. (2000). Vapor-liquid equilibrium of aqueous
solutions of formaldehyde and methanol. AIChE J 46: 1676-1687.
http://dx.doi.org/10.1002/aic.69046Q818
Albert. RE: Sellakumar. AR: Laskin. S: Kuschner. M: Nelson. N: Snyder. CA. (1982). Gaseous
formaldehyde and hydrogen chloride induction of nasal cancer in the rat J Natl Cancer Inst
68: 597-603.
Alexandersson. R: Hedenstierna. G. (1988). Respiratory hazards associated with exposure to
formaldehyde and solvents in acid-curing paints. Arch Environ Health 43: 222-227.
http://dx.doi.Org/10.1080/00039896.1988.9934937
Alexandersson. R: Hedenstierna. G. (1989). Pulmonary function in wood workers exposed to
formaldehyde: A prospective study. Arch Environ Health 44: 5-11.
http://dx.doi.Org/10.1080/00039896.1989.9935865
Alexandersson. R: Hedenstierna. G: Kolmodin-Hedman. B. (1982). Exposure to formaldehyde:
effects on pulmonary function. Arch Environ Health 37: 279-284.
http://dx.doi.Org/10.1080/00039896.1982.10667579
Ames. BN: Gold. LS. (1990). Too many rodent carcinogens: mitogenesis increases mutagenesis
[Review], Science 249: 970-971. http://dx.doi.org/10.1126/science.2136249
This document is a draft for review purposes only and does not constitute Agency policy.
R-l DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Amiri. A: Turner-Henson. A. (2017). The roles of formaldehyde exposure and oxidative stress in
fetal growth in the second trimester. J Obstet Gynecol Neonatal Nurs 46: 51-62.
http://dx.doi.Org/10.1016/i.iogn.2016.08.007
Andersen. I. (1979). Formaldehyde in the indoor environment - health implications and the setting
of standards. In PO Fanger; 0 Valbjorn (Eds.), Indoor climate: Effects on human comfort,
performance, and health in residential, commercial, and light-industry buildings (pp. 65-
87). Copenhagen, Denmark: Danish Building Research Institute. internal-pdf://Andersen I
1979 FA2765-1655166213/Andersen I 1979 FA2765.pdf
Andersen. I: Molhave. L. (1983). Controlled human studies with formaldehyde. In JE Gibson (Ed.),
Formaldehyde toxicity (pp. 154-165). Washington, DC: Hemisphere Publishing.
http://internal-pdf://Andersen and Molhave 1983 in Gibson book FA1781_0CR-
1421690395/Andersen and Molhave 1983 in Gibson book FA1781_0CR.pdf
Andersen. ME: Clewell. HI: Bermudez. E: Dodd. DE: Willson. GA: Campbell. TL: Thomas. RS. (2010).
Formaldehyde: integrating dosimetry, cytotoxicity, and genomics to understand dose-
dependent transitions for an endogenous compound. Toxicol Sci 118: 716-731.
http: / /dx. doi. or g/10.109 3 /toxsci /kfq 3 0 3
Andersen. ME: III. CH: Bermudez. E: Willson. GA: Thomas. RS. (2008). Genomic signatures and dose-
dependent transitions in nasal epithelial responses to inhaled formaldehyde in the rat
Toxicol Sci 105: 368-383. http: / /dx. do i. or g /10.10 9 3/toxsci /kfnO97
Anderson. GP. (2008). Endotyping asthma: new insights into key pathogenic mechanisms in a
complex, heterogeneous disease. Lancet 372: 1107-1119.
http: //dx.doi.org/10.1016/S0140-6736C08161452-X
Andielkovich. DA: Tanszen. DB: Brown. MH: Richardson. RB: Miller. FT. (1995). Mortality of iron
foundry workers: IV. Analysis of a subcohort exposed to formaldehyde. J Occup Environ
Med 37: 826-837. http://dx.doi.Org/10.1097/00043764-199507000-00012
Andrews. LS: Clary. IT: Terrill. IB: Bolte. HF. (1987). Subchronic inhalation toxicity of methanol. J
Toxicol Environ Health 20: 117-124. http://dx.doi.org/10.1080/15287398709530965
Annesi-Maesano. I: Hulin. M: Lavaud. F: Raherison. C: Kopferschmitt. C: de Blav. F: Charpin. DA:
Denis. C. (2012). Poor air quality in classrooms related to asthma and rhinitis in primary
schoolchildren of the French 6 Cities Study. Thorax 67: 682-688.
http: / /dx. doi. or g/10.113 6 /thoraxi nl-2011-200391
Antelman. SM: Eichler. AT: Black. CA: Kocan. D. (1980). Interchangeability of stress and
amphetamine in sensitization. Science 207: 329-331.
http://dx.doi.org/10.1126/science.7188649
Appelman. LM: Woutersen. RA: Zwart. A: Falke. HE: Feron. VI. (1988). One-year inhalation toxicity
study of formaldehyde in male rats with a damaged or undamaged nasal mucosa. J Appl
Toxicol 8: 85-90. http://dx.doi.org/10.1002/iat25500802Q4
Arican. RY: Sahin. Z: Ustunel. I: Sarikcioglu. L: Ozdem. S: Oguz. N. (2009). Effects of formaldehyde
inhalation on the junctional proteins of nasal respiratory mucosa of rats. Exp Toxicol Pathol
61: 297-305. http://dx.doi.org/10.1016/i.etp.2008.09.005
Armitage. TW: Cullis. CF. (1971). STUDIES OF REACTION BETWEEN NITROGEN-DIOXIDE AND
SULFUR-DIOXIDE. Combust Flame 16: 125.
Armon. C. (2009). Smoking may be considered an established risk factor for sporadic ALS [Review],
Neurology 73: 1693-1698. http://dx.doi.org/10.1212/WNL.ObOl 3e3181 cldf48
Armstrong. RW: Imrev. PB: Lye. MS: Armstrong. Ml: Yu. MC: Sani. S. (2000). Nasopharyngeal
carcinoma in Malaysian Chinese: occupational exposures to particles, formaldehyde and
heat Int J Epidemiol 29: 991-998. http://dx.doi.Org/10.1093/iie/29.6.991
Arundel. AY: Sterling. EM: Biggin. TH: Sterling. TP. (1986). Indirect health effects of relative
humidity in indoor environments. Environ Health Perspect 65: 351-361.
This document is a draft for review purposes only and does not constitute Agency policy.
R-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Asgharian. B: Price. OT: Schroeter. TP: Kimbell. IS: Singal. M. (2012). A lung dosimetry model of
vapor uptake and tissue disposition. Inhal Toxicol 24: 182-193.
http://dx.doi.org/10.3109/089S8378.2012.6S48S7
Asian. H: Songur. A: Tunc. AT: Ozen. OA: Bas. 0: Yagmurca. M: Turgut. M: Sarsilmaz. M: Kaplan. S.
(2006). Effects of formaldehyde exposure on granule cell number and volume of dentate
gyrus: a histopathological and stereological study. Brain Res 1122: 191-200.
http://dx.doi.Org/10.1016/i.brainres.2006.09.005
ATS (American Thoracic Society). (2000). What constitutes an adverse health effect of air pollution?
Am J Respir Crit Care Med 161: 665-673. http://dx.doi.Org/10.1164/airccm.161.2.ats4-00
ATSDR (Agency for Toxic Substances and Disease Registry). (1999). Toxicological profile for
formaldehyde [ATSDR Tox Profile], Atlanta, GA: U.S. Department of Health and Human
Services, Public Health Service, http://www.atsdr.cdc.gov/toxprofiles/tplll.pdf
ATSDR (Agency for Toxic Substances and Disease Registry). (2010). Addendum to the toxicological
profile for formaldehyde. Atlanta, GA.
http://www.atsdr.cdc.gov/toxprofiles/formaldehyde addendum.pdf
Attia. D: Mansour. N: Taha. F: El Dein. AS. (2014). Assessment of lipid peroxidation and p53 as a
biomarker of carcinogenesis among workers exposed to formaldehyde in cosmetic industry.
Toxicol Ind Health 32: 1097-1105. http://dx.doi.org/l 0.1177/0748233714S47152
Audrezet. MP: Robaszkiewicz. M: Mercier. B: Nousbaum. IB: Bail. TP: Hardy. E: Volant. A: Lozach. P:
Charles. IF: Gouerou. H: Ferec. C. (1993). TP53 gene mutation profile in esophageal
squamous-cell carcinomas. Cancer Res 53: 5745-5749.
Axelsson. G: Liitz. C: Rvlander. R. (1984). Exposure to solvents and outcome of pregnancy in
university laboratory employees. Br J Ind Med 41: 305-312.
Avdin. S: Canpinar. H: Undeger. U: Gtic. D: Colakoglu. M: Kars. A: Basaran. N. (2013). Assessment of
immunotoxicity and genotoxicity in workers exposed to low concentrations of
formaldehyde. Arch Toxicol 87: 145-153. http://dx.doi.Org/10.1007/s00204-012-0961-9
Avdin. S: Ogeturk. M: Kuloglu. T: Kavakli. A: Avdin. S. (2014). Effect of carnosine supplementation
on apoptosis and irisin, total oxidant and antioxidants levels in the serum, liver and lung
tissues in rats exposed to formaldehyde inhalation. Peptides 64C: 14-23.
http://dx.doi.Org/10.1016/j.peptides.2014.ll.008
Bach. B: Pedersen. OF: M0lhave. L. (1990). Human performance during experimental formaldehyde
exposure. Environ Int 16: 105-113. http://dx.doi.org/10.1016/0160-4120r90190150-5
Bahadori. K: Dovle-Waters. MM: Marra. C: Lvnd. L: Alasalv. K: Swiston. 1: Fitzgerald. TM. (2009).
Economic burden of asthma: a systematic review [Review], BMC Pulm Med 9: 24.
http://dx.doi.org/10.1186/1471-2466-9-24
Baird. DP: Wilcox. AT: Weinberg. CR. (1986). Use of time to pregnancy to study environmental
exposures. Am J Epidemiol 124: 470-480.
Balkwill. FR: Capasso. M: Hagemann. T. (2012). The tumor microenvironment at a glance. J Cell Sci
125: 5591-5596. http://dx.doi.org/10.1242/ics.l 16392
Ballarin. C: Sarto. F: Giacomelli. L: Bartolucci. GB: Clonfero. E. (1992). Micronucleated cells in nasal
mucosa of formaldehyde-exposed workers. MutatRes Genet Toxicol 280: 1-7.
http://dx.doi.org/10.1016/0165-1218r92190012-0
Band. PR: Le. ND: Fang. R: Threlfall. WT: Astrakianakis. G: Anderson. IT: Keefe. A: Krewski. D. (1997).
Cohort mortality study of pulp and paper mill workers in British Columbia, Canada. Am J
Epidemiol 146: 186-194. http://dx.doi.org/10.1093/oxfordjournals.aje.a009250
Barnea. ER: Tal. 1. (1991). Stress-related reproductive failure [Review], 8: 15-23.
http://dx.doi.org/10.1007/bf01131586
Barnes. PI. (2008). Immunology of asthma and chronic obstructive pulmonary disease [Review],
Nat Rev Immunol 8: 183-192. http://dx.doi.org/10.1038/nri2254
This document is a draft for review purposes only and does not constitute Agency policy.
R-3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Barrow. CS: Steinhagen. WH: Chang. TCF. (1983). Formaldehyde sensory irritation. In JE Gibson
(Ed.), Formaldehyde toxicity (pp. 16-25). Washington, DC: Hemisphere Publishing.
Bassig. B: Zhang. L: Cawthon. R: Yin. S: Li. G: Rappaport. S: Hu. W. ei: Smith. MT: Rothman. N:
Vermeulen. R: Lan. 0. (2012). Occupational exposure to benzene and leukocyte telomere
length. Cancer Res 72. http://dx.doi.org/10.1158/1538-7445.AM2012-4474
Bassig. BA: Zhang. L: Vermeulen. R: Tang. X: Li. G: Hu. W. ei: Guo. W: Purdue. MP: Yin. S: Rappaport.
SM: Shen. M. in: Ti. Z: Oiu. C: Ge. Y: Hosgood. HP: Reiss. B: Wu. B: Xie. Y: Li. L: Yue. F. ei:
Freeman. LEB: Blair. A: Hayes. RB: Huang. H: Smith. MT: Rothman. N: Lan. 0. (2016).
Comparison of hematological alterations and markers of B-cell activation in workers
exposed to benzene, formaldehyde and trichloroethylene. Carcinogenesis 37: 692-700.
http: / /dx. doi. or g/10.10 9 3 /carcin/bgwO 5 3
Bates. TH: Rincon. M: Irvin. CG. (2009). Animal models of asthma [Review], Am J Physiol Lung Cell
Mol Physiol 297: L401-L410. http://dx.doi.org/10.1152/aiplung.00027.2009
Bateson. TF: Schwartz. 1. (2008). Children's response to air pollutants [Review], J Toxicol Environ
Health A 71: 238-243. http: //dx.doi.org/10.1080 /152873907Q1598234
Battelle (Battelle Columbus Laboratories). (1981). Final report on a chronic inhalation toxicology
study in rats and mice exposed to formaldehyde to Chemical Industry Institute of
Toxicology: Volume 1. Research Triangle Park, NC: Chemical Industry Institute of
Toxicology.
Battelle (Battelle Columbus Laboratories). (1982). A chronic inhalation toxicology study in rats and
mice exposed to formaldehyde. Research Triangle Park, NC: Chemical Industry Institute of
Toxicology.
Bautista. DM: Tordt. SE: Nikai. T: Tsuruda. PR: Read. AT: Poblete. 1: Yamoah. EN: Basbaum. AI: Tulius.
D. (2006). TRPA1 Mediates the Inflammatory Actions of Environmental Irritants and
Proalgesic Agents. Cell 124: 1269-1282. http://dx.doi.Org/10.1016/i.cell.2006.02.023
Beane Freeman. LE: Blair. A: Lubin. TH: Stewart. PA: Hayes. RB: Hoover. RN: Hauptmann. M. (2013).
Mortality from solid tumors among workers in formaldehyde industries: an update of the
NCI cohort Am J Ind Med 56: 1015-1026. http://dx.doi.org/10.1002/aiim.22214
Beane Freeman. LE: Blair. A: Lubin. TH: Stewart. PA: Hayes. RB: Hoover. RN: M. H. (2009). Mortality
from lymphohematopoietic malignancies among workers in formaldehyde industries: The
National Cancer Institute Cohort. J Natl Cancer Inst 101: 751-761.
http: / / dx. do i. o r g /10.10 9 3 /i nci/di p 0 9 6
Becklake. MR: White. N. (1993). Sources of variation in spirometric measurements. Identifying the
signal and dealing with noise [Review], Occup Med 8: 241-264.
Beland. FA: Fullerton. NF: Heflich. RH. (1984). Rapid isolation, hydrolysis and chromatography of
formaldehyde-modified DNA. J Chromatogr A 308: 121-131.
http: //dx.doi.org/10.1016/0378-4347r84180202-9
Bell. IR: Baldwin. CM: Fernandez. M: Schwartz. GE. (1999). Neural sensitization model for multiple
chemical sensitivity: overview of theory and empirical evidence [Review], T oxicol Ind
Health 15: 295-304. http://dx.doi.org/10.1177/074823379901500303
Bell. IR: Miller. CS: Schwartz. GE. (1992). An olfactory-limbic model of multiple chemical sensitivity
syndrome: Possible relationships to kindling and affective spectrum disorders [Review],
Biol Psychiatry 32: 218-242. http://dx.doi.org/10.1016/0006-3223C92190105-9
Bellavia. A: Dickerson. AS: Rotem. RS: Hansen. 1: Gredal. 0: Weisskopf. MG. (2021). Joint and
interactive effects between health comorbidities and environmental exposures in predicting
amyotrophic lateral sclerosis. Int J Hyg Environ Health 231: 113655.
http://dx.doi.Org/10.1016/i.iiheh.2020.113655
Bellisario. V: Mengozzi. G: Grignani. E: Bugiani. M: Sapino. A: Bussolati. G: Bono. R. (2016). Towards
a formalin-free hospital. Levels of 15-F2t-isoprostane and malondialdehyde to monitor
This document is a draft for review purposes only and does not constitute Agency policy.
R-4 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
exposure to formaldehyde in nurses from operating theatres. Toxicology Research 5: 1122-
1129. http://dx.doi.org/10.1039/c6tx00Q68a
Bender. 1R: Mullin. LS: Grapel. Gl: Wilson. WE. (1983). Eye irritation response of humans to
formaldehyde. Am Ind Hyg Assoc J 44: 463-465.
http://dx.doi.org/10.1080/15298668391405139
Benedetti. MS: Whomslev. R: Canning. M. (2007). Drug metabolism in the paediatric population and
in the elderly [Review], Drug Discov Today 12: 599-610.
http://dx.doi.Org/10.1016/i.drudis.2007.06.011
Bentaveb. M: Norback. D: Bednarek. M: Bernard. A: Cai. G: Cerrai. S: Eleftheriou. KK: Gratziou. C:
Hoist. Gl: Lavaud. F: Nasilowski. 1: Sestini. P: Sarno. G: Sigsgaard. T: Wieslander. G: Zielinski.
1: Viegi. G: Annesi-Maesano. I: Study. G. (2015). Indoor air quality, ventilation and
respiratory health in elderly residents living in nursing homes in Europe. Eur Respir J 45:
1228-1238. http://dx.doi.Org/10.1183/09031936.00082414
Berglund. B: Hoglund. A: Esfandabad. HS. (2012). Abisensory method for odor and irritation
detection of formaldehyde and pyridine. Chemosensory Perception 5: 146-157.
http://dx.doi.Org/10.1007/sl2078-011-9101-9
Berglund. B: Nordin. S. (1992). Detectability and perceived intensity for formaldehyde in smokers
and non-smokers. Chem Senses 17: 291-306. http://dx.doi.Org/10.1093 /chemse/17.3.291
Bermudez. EG. (2004). HCHO studies - Tumor incidence [Memorandum], Available online
Berrino. F: Richiardi. L: Boffetta. P: Esteve. 1: Belletti. I: Raymond. L: Troschel. L: Pisani. P: Zubiri. L:
Ascunce. N: Guberan. E: Tuvns. A: Terracini. B: Merletti. F: Group. MTW. (2003). Occupation
and larynx and hypopharynx cancer: A job-exposure matrix approach in an international
case-control study in France, Italy, Spain and Switzerland. Cancer Causes Control 14: 213-
223. http://dx.doi.Org/10.1023/a:1023661206177
Bertazzi. PA: Pesatori. A: Guercilena. S: Consonni. D: Zocchetti. C. (1989). Cancer risk among
workers producing formaldehyde-based resins: Extension of follow-up. 80: 111-122.
Bertazzi. PA: Pesatori. AC: Radice. L: Zocchetti. C. (1986). Exposure to formaldehyde and cancer
mortality in a cohort of workers producing resins. Scand J Work Environ Health 12: 461-
468. http://dx.doi.org/doi: 10.5271 /siweh.2111
Biagini. RE: Moorman. WT: Knecht. EA: Clark. TC: Bernstein. IL. (1989). Acute airway narrowing in
monkeys from challenge with 2.5 ppm formaldehyde generated from formalin. Arch
Environ Health 44: 12-17. http: //dx.doi.org/10.1080/00039896.1989.9935866
Billionnet. C: Gay. E: Kirchner. S: Levnaert. B: Annesi-Maesano. I. (2011). Quantitative assessments
of indoor air pollution and respiratory health in a population-based sample of French
dwellings. Environ Res 111: 425-434. http://dx.doi.Org/10.1016/i.envres.2011.02.008
Binawara. BK: Raninee. CS: Mathur. KC: Sharma. H: Goval. K. (2010). Acute effect of formalin on
pulmonary function tests in medical students. Pak J Physiol 6: 8-10.
Blair. A: Stewart. P: O'Berg. M: Gaffev. W: Walrath. I: Ward. 1: Bales. R: Kaplan. S: Cubit. D. (1986).
Mortality among industrial workers exposed to formaldehyde. J Natl Cancer Inst 76: 1071-
1084.
Blair. A: Zheng. T: Linos. A: Stewart. PA: Zhang. YW: Cantor. KP. (2001). Occupation and leukemia: A
population-based case-control study in Iowa and Minnesota. Am J Ind Med 40: 3-14.
http: / / dx. do i. o r g /10.10 0 2 /ai im. 10 6 6
Boffetta. P: Stellman. SD: Garfinkel. L. (1989). A case-control study of multiple myeloma nested in
the American Cancer Society prospective study. Int J Cancer 43: 554-559.
http://dx.doi.Org/10.1002/iic.2910430404
Bogdanffv. MS: Morgan. PH: Starr. TB: Morgan. KT. (1987). Binding of formaldehyde to human and
rat nasal mucus and bovine serum albumin. Toxicol Lett 38: 145-154.
http://dx.d0i.0rg/l 0.1016/0378-4274f87190122-6
This document is a draft for review purposes only and does not constitute Agency policy.
R-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Bogdanffv. MS: Randall. HW: Morgan. KT. (1986). Histochemical localization of aldehyde
dehydrogenase in the respiratory tract of the Fischer-344 rat Toxicol Appl Pharmacol 82:
560-567. http://dx.doi.org/10.1016/0041 -008Xf86190291 -7
Bogdanffv. MS: Sarangapani. R: Plowchalk. PR: Tarabek. AM: Andersen. ME. (1999). A biologically
based risk assessment for vinyl acetate-induced cancer and noncancer inhalation toxicity.
Toxicol Sci 51: 19-35. http://dx.doi.org/10.1093/toxsci/51.1.19
Boia. TW: Nielsen. TA: Foldvarv. E: Truitt. EB. Tr. (1985). Acute low-level formaldehyde behavioural
and neurochemical toxicity in the rat. Prog Neuropsychopharmacol Biol Psychiatry 9: 671-
674. http: / /dx.doi.org/10.1016/0278-5846C85190038-7
Bolm-Audorff. U: Vogel. C: Woitowitz. H. (1990). Occupation and smoking as risk factors of nasal
and nasopharyngeal cancer. In RR Monson (Ed.), Occupational epidemiology (2nd ed., pp.
71-74). Boca Raton, FL: CRC Press.
Bonassi. S: Lando. C: Ceppi. M: Landi. S: Rossi. AM: Barale. R. (2004a). No association between
increased levels of high-frequency sister chromatid exchange cells (HFCs) and the risk of
cancer in healthy individuals. Environ Mol Mutagen 43: 134-136.
http://dx.doi.Org/10.1002/em.20006
Bonassi. S: Norppa. H: Ceppi. M: Stromberg. U: Vermeulen. R: Znaor. A: Cebulska-Wasilewska. A:
Fabianova. E: Fucic. A: Gundy. S: Hansteen. IL: Knudsen. LE: Lazutka. 1: Rossner. P: Sram. RT:
Boffetta. P. (2008). Chromosomal aberration frequency in lymphocytes predicts the risk of
cancer: results from a pooled cohort study of 22 358 subjects in 11 countries.
Carcinogenesis 29: 1178-1183. http://dx.doi.org/10.1093/carcin/bgn075
Bonassi. S: Znaor. A: Ceppi. M: Lando. C: Chang. WP: Holland. N: Kirsch-Volders. M: Zeiger. E: Ban. S:
Barale. R: Bigatti. MP: Bolognesi. C: Cebulska-Wasilewska. A: Fabianova. E: Fucic. A: Hagmar.
L: Toksic. G: Martelli. A: Migliore. L: Mirkova. E: Scarfi. MR: Ziino. A: Norppa. H: Fenech. M.
(2007). An increased micronucleus frequency in peripheral blood lymphocytes predicts the
risk of cancer in humans. Carcinogenesis 28: 625-631.
http://dx.doi.org/10.1093/carcin/bgll77
Bonassi. S: Znaor. A: Norppa. H: Hagmar. L. (2004b). Chromosomal aberrations and risk of cancer in
humans: an epidemiologic perspective [Review], Cytogenet Genome Res 104: 376-382.
http://dx.d0i.0rg/l 0.1159/000077519
Bono. R: Munnia. A: Romanazzi. V: Bellisario. V: Cellai. F: Peluso. MEM. (2016). Formaldehyde-
induced toxicity in the nasal epithelia of workers of a plastic laminate plant Toxicology
Research 5: 752-760. http://dx.doi.org/10.1039/c5tx00478k
Bono. R: Romanazzi. V: Munnia. A: Piro. S: Allione. A: Ricceri. F: Guarrera. S: Pignata. C: Matullo. G:
Wang. P: Giese. RW: Peluso. M. (2010). Malondialdehyde-deoxyguanosine adduct formation
in workers of pathology wards: the role of air formaldehyde exposure. Chem Res Toxicol 23:
1342-1348. http://dx.d0i.0rg/l0.1021 /txl00083x
Bono. R: Romanazzi. V: Pirro. V: Degan. R: Pignata. C: Suppo. E: Pazzi. M: Vincenti. M. (2012).
Formaldehyde and tobacco smoke as alkylating agents: The formation of N-methylenvaline
in pathologists and in plastic laminate workers. Sci Total Environ 414: 701-707.
http://dx.doi.0rg/lO.lOl6/j.scitotenv.2Oll.lO.O47
Boreiko. CI: Ragan. PL. (1983). Formaldehyde effects in the C3H/10Ty2 cell transformation assay. In
JE Gibson (Ed.), Formaldehyde toxicity (pp. 63-71). Washington, DC: Hemisphere Publishing
Corporation.
Bowen. SE: Hannigan. TH. (2006). Developmental toxicity of prenatal exposure to toluene [Review],
AAPS J 8: E419-E424. http: / /dx.doi.org/10.1007/BFO2854915
Bovsen. M: Zadig. E: Digernes. V: Abeler. V: Reith. A. (1990). Nasal mucosa in workers exposed to
formaldehyde: a pilot study. Occup Environ Med 47: 116-121.
http://dx.doi.Org/10.1136/oem.47.2.116
This document is a draft for review purposes only and does not constitute Agency policy.
R-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Branco. PTB. S: Alvim-Ferraz. MCM: Martins. FG: Ferraz. C: Vaz. LG: Sousa. SIV. (2020). Impact of
indoor air pollution in nursery and primary schools on childhood asthma. Sci Total Environ
745: 140982. http://dx.doi.Org/10.1016/i.scitotenv.2020.140982
Branzei. D: Foiani. M. (2008). Regulation of DNA repair throughout the cell cycle [Review], Nat Rev
Mol Cell Biol 9: 297-308. http://dx.doi.Org/l0.1038/nrm2351
Brevsse. PA. (1984). Formaldehyde levels and accompanying symptoms associated with individuals
residing in over 1000 conventional and mobile homes in the state of Washington. In B
Berglund; T Lindvall; J Sundell (Eds.), Indoor air: Proceedings of the 3rd International
Conference on Indoor Air Quality and Climate Volume 3: Sensory and hyperreactivity
reactions to sick buildings (pp. 403-408). Stockholm, Sweden: Swedish Council for Building
Research. https://search.proquest.com/docview/14197938?accountid=171501
Brinton. LA: Blot. WT: Becker. TA: Winn. DM: Browder. TP: Farmer. TC. Tr: Fraumeni. IF. Tr. (1984). A
case-control study of cancers of the nasal cavity and paranasal sinuses. Am J Epidemiol 119:
896-906. http://dx.doi.org/10.1093/oxfordiournals.aie.all3812
Brinton. LA: Blot. WT: Fraumeni. IF. Tr. (1985). Nasal cancer in the textile and clothing industries.
Occup Environ Med 42: 469-474.
Broadhurst. PL. (1969). Psychogenetics of emotionality in the rat Ann N Y Acad Sci 159: 806-824.
http://dx.doi.Org/l 0.1111 /i.l 749-6632.1969.tbl 2980.x
Broder. I: Corey. P: Brasher. P: Lipa. M: Cole. P. (1988a). Comparison of health of occupants and
characteristics of houses among control homes and homes insulated with urea
formaldehyde foam: III. Health and house variables following remedial work. Environ Res
45: 179-203. http://dx.d0i.0rg/l0.1016/SOOl3-9351C88180046-X
Broder. I: Corey. P: Cole. P: Lipa. M: Mintz. S: Nethercott. TR. (1988b). Comparison of health of
occupants and characteristics of houses among control homes and homes insulated with
urea formaldehyde foam: I Methodology. Environ Res 45: 141-155.
http://dx.d0i.0rg/l 0.1016/S0013-9351 f88180044-6
Broder. I: Corey. P: Cole. P: Lipa. M: Mintz. S: Nethercott. TR. (1988c). Comparison of health of
occupants and characteristics of houses among control homes and homes insulated with
urea formaldehyde foam: II initial health and house variables and exposure-response
relationships. Environ Res 45: 156-178. http://dx.doi.0rg/lO.lOl6/SOOl3-9351(88)80045-
8
Brondeau. MT: Bonnet. P: Guenier. TP: Simon. P: de Ceaurriz. 1. (1990). Adrenal-dependent
leucopenia after short-term exposure to various airborne irritants in rats. J Appl Toxicol 10:
83-86. http://dx.d0i.0rg/l 0.1002/iat2550100204
Brooks. BR. (1994). El Escorial World Federation of Neurology criteria for the diagnosis of
amyotrophic lateral sclerosis. J Neurol Sci 124: 96-107. http://dx.doi.org/10.1016/0Q22-
510XC94190191-0
Brooks. PI: Zakhari. S. (2014). Acetaldehyde and the genome: Beyond nuclear DNA adducts and
carcinogenesis [Review], Environ Mol Mutagen 55: 77-91.
http ://dx. doi. or g/10.10 0 2 /em. 2182 4
Brown. HR. (1990). Neoplastic and Potentially Preneoplastic Changes in the Upper Respiratory
Tract of Rats and Mice. Environ Health Perspect 85: 291.
http://dx.doi.org/10.2307/3430690
Brown. HR: Monticello. TM: Maronpot. RR: Randall. HW: Hotchkiss. TR: Morgan. KT. (1991).
Proliferative and neoplastic lesions in the rodent nasal cavity. Toxicol Pathol 19: 358-372.
http://dx.doi.Org/10.1177/0192623391019004-105
Buckley. LA: Morgan. KT: Swenberg. TA: Tames. RA: Tr. HT: Barrow. CS. (1985). The toxicity of
dimethylamine in F-344 rats and B6C3F1 mice following a 1-year inhalation exposure.
FundamAppl Toxicol 5: 341-352. http://dx.doi.Org/10.1093/toxsci/5.2.341
This document is a draft for review purposes only and does not constitute Agency policy.
R-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Buckley. TM: Schatzberg. AF. (2005). On the interactions of the hypothalamic-pituitary-adrenal
(HPA) axis and sleep: normal HPA axis activity and circadian rhythm, exemplary sleep
disorders [Review], J Clin Endocrinol Metab 90: 3106-3114.
http://dx.doi.Org/10.1210/jc.2004-1056
Burgaz. S: Cakmak. G: Erdem. 0: Yilmaz. M: Karakava. AE. (2001). Micronuclei frequencies in
exfoliated nasal mucosa cells from pathology and anatomy laboratory workers exposed to
formaldehyde. Neoplasma 48: 144-147.
Burgaz. S: Erdem. 0: Cakmak. G: Erdem. N: Karakava. A: Karakava. AE. (2002). Cytogenetic analysis
of buccal cells from shoe-workers and pathology and anatomy laboratory workers exposed
to n-hexane, toluene, methyl ethyl ketone and formaldehyde. Biomarkers 7: 151-161.
http://dx.d0i.0rg/l 0.1080 /13547500110113242
Burgos-Barragan. G: Wit. N: Meiser. 1: Dingier. FA: Pietzke. M: Mulderrig. L: Pontel. LB: Rosado. IV:
Brewer. TF: Cordell. RL: Monks. PS: Chang. CI: Vazquez. A: Patel. KT. (2017a). Erratum:
Mammals divert endogenous genotoxic formaldehyde into one-carbon metabolism
[Erratum], Nature 548: 612. http://dx.doi.org/10.1038/nature23904
Burgos-Barragan. G: Wit. N: Meiser. 1: Dingier. FA: Pietzke. M: Mulderrig. L: Pontel. LB: Rosado. IV:
Brewer. TF: Cordell. RL: Monks. PS: Chang. CI: Vazquez. A: Patel. KT. (2017b). Mammals
divert endogenous genotoxic formaldehyde into one-carbon metabolism. Nature 548: 549-
554. http://dx.doi.org/10.1038/nature23481
Cadieux. A: Springall. PR: Mulderrv. PK: Rodrigo. 1: Ghatei. MA: Terenghi. G: Bloom. SR: Polak. TM.
(1986). Occurrence, distribution and ontogeny of CGRP immunoreactivity in the rat lower
respiratory tract: effect of capsaicin treatment and surgical denervations. Neuroscience 19:
605-627.
Callas. PW: Pastides. H: Hosmer. DW. (1998). Empirical comparisons of proportional hazards,
poisson, and logistic regression modeling of occupational cohort data. Am J Ind Med 33:33-
47. http://dx.d0i.0rg/l 0.1002/fsicill 097-0274fl 99801133:1 <33::aid-aiim5>3.0.co:2-x
Campbell Tr. TL: Gentry. PR: Clewell III. HI: Andersen. ME. (2020). Akinetic analysis of DNA-deoxy
guanine adducts in the nasal epithelium produced by inhaled formaldehyde in rats-
assessing contributions to adduct production from both endogenous and exogenous sources
of formaldehyde. Toxicol Sci 177: 325-333. http: //dx.doi.org/10.1093/toxsci/kfaal22
Carr. Ml: Undem. BI. (2001). Inflammation-induced plasticity of the afferent innervation of the
airways. Environ Health Perspect4: 567-571.
Casanova-Schmitz. M: David. RM: Heck. H. (1984a). Oxidation of formaldehyde and acetaldehyde by
NAD+-dependent dehydrogenases in rat nasal mucosal homogenates. Biochem Pharmacol
33: 1137-1142. http://dx.doi.org/10.1016/0006-2952r84190526-4
Casanova-Schmitz. M: Heck. H. (1983). Effects of formaldehyde exposure on the extractability of
DNA from proteins in the rat nasal mucosa. Toxicol Appl Pharmacol 70: 121-132.
http://dx.doi.org/10.1016/0041-008xr83190185-0
Casanova-Schmitz. M: Starr. TB: Heck. HP. (1984b). Differentiation between metabolic
incorporation and covalent binding in the labeling of macromolecules in the rat nasal
mucosa and bone marrow by inhaled [14C]- and [3H]formaldehyde. Toxicol Appl Pharmacol
76: 26-44. http://dx.doi.Org/10.1016/0041-008xr84190026-7
Casanova. M: Devo. DF: Heck. H. (1989). Covalent binding of inhaled formaldehyde to DNA in the
nasal mucosa of Fischer 344 rats: Analysis of formaldehyde and DNA by high-performance
liquid chromatography and provisional pharmacokinetic interpretation. Fundam Appl
Toxicol 12: 397-417. http://dx.doi.org/10.1016/0272-0590r89190015-8
Casanova. M: Heck. H. (1987). Further studies of the metabolic incorporation and covalent binding
of inhaled [3H]- and [14C]formaldehyde in Fischer-344 rats: Effects of glutathione
depletion. Toxicol Appl Pharmacol 89: 105-121. http://dx.doi.org/10.1016/0041-
008Xr87190181-5
This document is a draft for review purposes only and does not constitute Agency policy.
R-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Casanova. M: Heck. H. (1997). Lack of evidence for the involvement of formaldehyde in the
hepatocarcinogenicity of methyl tertiary-butyl ether in CD-I mice. Chem Biol Interact 105:
131-143.
Casanova. M: Heck. H: Everitt. II: Harrington. WW. Tr: Popp. TA. (1988). Formaldehyde
concentrations in the blood of rhesus monkeys after inhalation exposure. Food Chem
Toxicol 26: 715-716. http://dx.doi.org/10.1016/0278-6915C88190071 -3
Casanova. M: Morgan. KT: Gross. EA: Moss. OR: Heck. H. (1994). DNA-protein cross-links and cell
replication at specific sites in the nose of F344 rats exposed subchronically to
formaldehyde. Fundam Appl Toxicol 23: 525-536.
http ://dx. doi. or g/10.10 0 6 /faat 199 4.1137
Casanova. M: Morgan. KT: Steinhagen. WH: Everitt. II: Popp. TA: Heck. H. (1991). Covalent binding of
inhaled formaldehyde to DNA in the respiratory tract of rhesus monkeys: pharmacokinetics,
rat-to-monkey interspecies scaling, and extrapolation to man. Toxicol Sci 17: 409-428.
http: //dx.doi.org/10.1016/0272-0590f91190230-2
Cassee. FR: Feron. VI. (1994). Biochemical and histopathological changes in nasal epithelium of rats
after 3-day intermittent exposure to formaldehyde and ozone alone or in combination.
Toxicol Lett 72: 257-268. http://dx.doi.org/10.1016/0378-4274C94190037-X
Cassee. FR: Groten. TP: Feron. VI. (1996). Changes in the nasal epithelium of rats exposed by
inhalation to mixtures of formaldehyde, acetaldehyde, and acrolein. Toxicol Sci 29: 208-218.
http://dx.doi.org/10.1006/faatl996.0Q24
Casset. A: Marchand. C: Purohit. A: le Calve. S: Uring-Lambert. B: Donnav. C: Meyer. P: de Blav. F.
(2006). Inhaled formaldehyde exposure: effect on bronchial response to mite allergen in
sensitized asthma patients. Allergy 61: 1344-1350. http://dx.doi.Org/10.l111/i.1398-
9995.2006.01174.x
Casset. A: Purohit. A: Marchand. C: Le Calve. S: Pauli. G: de Blav. F. (2007). Inhaled formaldehyde
and the bronchial response. Rev Fr Allergol Immunol Clin 47: 80-83.
http://dx.doi.0rg/lO.lOl6/i.allerg.2OO6.lO.OO9
CDC (Centers for Disease Control and Prevention). (2004). The health consequences of smoking: A
report of the Surgeon General. Washington, DC: U.S. Department of Health and Human
Services, http://www.cdc.gov/tobacco/data statistics/sgr/2004/index.htm
Ceppi. M: Biasotti. B: Fenech. M: Bonassi. S. (2010). Human population studies with the exfoliated
buccal micronucleus assay: Statistical and epidemiological issues [Review], MutatRes Rev
MutatRes 705: 11-19. http://dx.doi.Org/10.1016/i.mrrev.2009.ll.001
Chamanza. R: Wright. TA. (2015). A Review of the Comparative Anatomy, Histology, Physiology and
Pathology of the Nasal Cavity of Rats, Mice, Dogs and Non-human Primates. Relevance to
Inhalation Toxicology and Human Health Risk Assessment J Comp Pathol 153: 287-314.
http://dx.doi.Org/10.1016/i.icpa.2015.08.009
Chan-Yeung. M. (2000). Spirometry and tests of bronchial hyperresponsiveness in population
studies [Review], IntJTuberc LungDis4: 633-638.
Chandra. M: Riley. MG: Tohnson. DE. (1992). Spontaneous neoplasms in aged Sprague-Dawley rats.
Arch Toxicol 66: 496-502. http: //dx.d0i.0rg/l 0.1007/BF01970675
Chang. TCF: Barrow. CS. (1984). Sensory irritation tolerance and cross-tolerance in F-344 rats
exposed to chlorine or formaldehyde gas. Toxicol Appl Pharmacol 76: 319-327.
http://dx.d0i.0rg/l 0.1016/0041-008XC84190013-9
Chang. TCF: Gross. EA: Swenberg. TA: Barrow. CS. (1983). Nasal cavity deposition, histopathology,
and cell proliferation after single or repeated formaldehyde exposures in B6C3F1 mice and
F-344 rats. Toxicol Appl Pharmacol 68: 161-176. http: //dx.doi.org/10.1016/0Q41-
008xC83190001-7
This document is a draft for review purposes only and does not constitute Agency policy.
R-9 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Chang. TCF: Steinhagen. WH: Barrow. CS. (1981). Effect of single or repeated formaldehyde
exposure on minute volume of B6C3F1 mice and F-344 rats. Toxicol Appl Pharmacol 61:
451-459. http://dx.doi.org/10.1016/0041 -OOBxCBl 190368-9
Chang. M: Park. H: Ha. M: Hong. YC: Lim. YH: Kim. Y: Kim. YT: Lee. D: Ha. EH. (2017). The effect of
prenatal TVOC exposure on birth and infantile weight: the Mothers and Children's
Environmental Health study. Pediatr Res 82: 423-428.
http://dx.doi.org/10.1038/pr.2017.55
Checkowav. H: Dell. LP: Boffetta. P: Gallagher. AE: Crawford. L: Lees. PS: Mundt. KA. (2015).
Formaldehyde Exposure and Mortality Risks From Acute Myeloid Leukemia and Other
Lymphohematopoietic Malignancies in the US National Cancer Institute Cohort Study of
Workers in Formaldehyde Industries. J Occup Environ Med 57: 785-794.
http://dx.doi.Org/10.1097/TOM.0000000000000466
Chen. CI: You. SL: Lin. LH: Hsu. WL: Yang. YW. (2002). Cancer epidemiology and control in Taiwan:
A brief review. Jpn J Clin Oncol 32: S66-S81. http://dx.doi.org/10.1093/iico/hyel38
Cheng. G: Wang. M: Upadhyava. P: Villalta. PW: Hecht. SS. (2008). Formation of formaldehyde
adducts in the reactions of DNA and deoxyribonucleosides with alpha-acetates of 4-
(methylnitrosamino)-l-(3-pyridyl)-l-butanone (NNK), 4-(methylnitrosamino)-l-(3-
pyridyl)-l-butanol (NNAL), and N-nitrosodimethylamine (NDMA). Chem Res Toxicol 21:
746-751. http://dx.d0i.0rg/l 0.1021 /tx7003823
Cheng. YT: Hildesheim. A: Hsu. MM: Chen. IH: Brinton. LA: Levine. PH: Chen. CI: Yang. CS. (1999).
Cigarette smoking, alcohol consumption and risk of nasopharyngeal carcinoma in Taiwan.
Cancer Causes Control 10: 201-207. http://dx.doi.Org/10.1023/A:1008893109257
Cheng. Z: Li. Y: Liang. B: Wang. C. (2004). [Investigation of formaldehyde level and health of
personnel in clinical pathology], 29: 266-267.
Chia. SE: Ong. CN: Foo. SC: Lee. HP. (1992). Medical students' exposure to formaldehyde in a gross
anatomy dissection laboratory. J Am Coll Health 41: 115-119.
http://dx.doi.Org/10.1080/07448481.1992.9936310
Cho. S: Kim. HI: Oh. SH: Park. CO: Tung. TY: Lee. KH. (2010). The influence of pregnancy and
menstruation on the deterioration of atopic dermatitis symptoms. Annals of Dermatology
22: 180-185. http://dx.d0i.0rg/l0.5021 /ad.2010.22.2.180
Choi. DW: Moon. KW: Bveon. SH: Lee. EI: Sul. DG: Lee. TH: Oh. EH: Kim. YH. (2009). Indoor volatile
organic compounds in atopy patients' houses in South Korea. Indoor Built Environ 18: 144-
154. http://dx.d0i.0rg/l 0.1177/1420326X08101945
CUT (Chemical Industry Institute of Toxicology). (1999). Formaldehyde: hazard characterization
and dose-response assessment for carcinogenicity by the route of inhalation.
Clayton. TC: Meade. TW: Turner. EL: De Stavola. BL. (2014). Peak flow rate and death due to
coronary heart disease: 30-year results from the Northwick Park Heart cohort study. 1:
e000164. http://dx.doi.org/10.1136/openhrt-2014-00Q164
Coggon. D: Harris. EC: Poole. 1: Palmer. KT. (2003). Extended follow-up of a cohort of British
chemical workers exposed to formaldehyde. J Natl Cancer Inst 95: 1608-1615.
http: / /dx. doi. or g/10.10 9 3/) nci/dj g046
Coggon. D: Ntani. G: Harris. EC: Palmer. KT. (2014). Upper Airway Cancer, Myeloid Leukemia, and
Other Cancers in a Cohort of British Chemical Workers Exposed to Formaldehyde. Am J
Epidemiol 179: 1301-1311. http://dx.doi.org/10.1093/aie/kwu049
Cohen Hubal. EA: Schlosser. PM: Conollv. RB: Kimbell. IS. (1997). Comparison of inhaled
formaldehyde dosimetry predictions with DNA-protein cross-link measurements in the rat
nasal passages. Toxicol Appl Pharmacol 143: 47-55.
http://dx.doi.org/10.1006/taap.1996.8076
This document is a draft for review purposes only and does not constitute Agency policy.
R-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Cohn. L: Elias. TA: Chupp. GL. (2004). Asthma: mechanisms of disease persistence and progression
[Review], Annu Rev Immunol 22: 789-815.
http://dx.doi.org/10.1146/annurev.immunol.22.0127Q3.104716
Comba. P: Barbieri. PG: Battista. G: Belli. S: Ponterio. F: Zanetti. D: Axelson. 0. (1992a). Cancer of the
nose and paranasal sinuses in the metal industry: a case-control study. Occup Environ Med
49: 193-196. http://dx.doi.org/10.1136 /oem.49.3.193
Comba. P: Battista. G: Belli. S: Decapua. B: Merler. E: Orsi. D: Rodella. S: Vidigni. C: Axelson. 0.
(1992b). A case-control study of cancer of the nose and paranasal sinuses and occupational
exposures. Am J Ind Med 22: 511-520. http://dx.doi.org/10.1002 /aiim.4700220406
Conollv. RB: Kimbell. IS: Tanszen. D: Schlosser. PM: Kalisak. D: Preston. 1: Miller. FT. (2003).
Biologically motivated computational modeling of formaldehyde carcinogenicity in the F344
rat. Toxicol Sci 75: 432-447. http://dx.doi.org/10.1093/toxsci/kfgl82
Conollv. RB: Kimbell. IS: Tanszen. D: Schlosser. PM: Kalisak. D: Preston. 1: Miller. FT. (2004). Human
respiratory tract cancer risks of inhaled formaldehyde: dose-response predictions derived
from biologically-motivated computational modeling of a combined rodent and human
dataset. Toxicol Sci 82: 279-296. http: / /dx.doi. or g/10.109 3/toxsci/kfh223
Conollv. RB: Kimbell. IS: Tanszen. DB: Miller. FT. (2002). Dose response for formaldehyde-induced
cytotoxicity in the human respiratory tract Regul Toxicol Pharmacol 35: 32-43.
http: / /dx. doi. or g/10.10 0 6 /rtph. 2001.1515
Conollv. RB: Lilly. PD: Kimbell. IS. (2000). Simulation modelling of the tissue disposition of
formaldehyde to predict nasal DNA-protein cross-links in Fischer 344 rats, rhesus monkeys,
and humans. Environ Health Perspect 108: 919-924. http: //dx.doi.org/10.2307/3454325
Corlev. RA: Kabilan. S: Kuprat. AP: Carson. TP: Tacob. RE: Minard. KR: Teeguarden. TG: Timchalk. C:
Pipavath. S: Glennv. R: Einstein. DR. (2015). Comparative risks of aldehyde constituents in
cigarette smoke using transient computational fluid dynamics/physiologically based
pharmacokinetic models of the rat and human respiratory tracts. Toxicol Sci 146: 65-88.
http: / /dx. doi. or g/10.109 3 /toxsci /kfvO 71
Corthav. A. (2014). Does the immune system naturally protect against cancer? [Review], 5: 197.
http://dx.doi.org/10.3389/fimmu.2014.0Q197
Costa. S: Carvalho. S: Costa. C: Coelho. P: Silva. S: Santos. LS: Gaspar. IF: Porto. B: Laffon. B: Teixeira.
TP. (2015). Increased levels of chromosomal aberrations and DNA damage in a group of
workers exposed to formaldehyde. Mutagenesis 30: 463-473.
http: / /dx. doi. or g /10.109 3 /mutage / ge vO 0 2
Costa. S: Coelho. P: Costa. C: Silva. S: Mayan. 0: Santos. LS: Gaspar. I: Teixeira. TP. (2008). Genotoxic
damage in pathology anatomy laboratory workers exposed to formaldehyde. Toxicology
252: 40-48. http://dx.doi.Org/10.1016/i.tox.2008.07.056
Costa. S: Costa. C: Madureira. 1: Valdiglesias. V: Teixeira-Gomes. A: Guedes de Pinho. P: Laffon. B:
Teixeira. TP. (2019). Occupational exposure to formaldehyde and early biomarkers of cancer
risk, immunotoxicity and susceptibility. Environ Res 179: 108740.
http://dx.doi.Org/10.1016/i.envres.2019.108740
Costa. S: Garcia-Leston. I: Coelho. M: Coelho. P: Costa. C: Silva. S: Porto. B: Laffon. B: Teixeira. TP.
(2013). Cytogenetic and immunological effects associated with occupational formaldehyde
exposure. J Toxicol Environ Health A 76: 217-229.
http: //dx.doi.org/10.1080/15287394.2013.757212
Coussens. LM: Werb. Z. (2002). Inflammation and cancer [Review], Nature 420: 860-867.
http://dx.doi.org/10.1038/nature01322
Coussens. LM: Zitvogel. L: Palucka. AK. (2013a). Neutralizing tumor-promoting chronic
inflammation: A magic bullet? [Review], Science 339: 286-291.
http: / /dx. doi. or g/10.112 6 /science. 12 3 2 2 2 7
This document is a draft for review purposes only and does not constitute Agency policy.
R-ll DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Coussens. LM: Zitvogel. L: Palucka. AK. (2013b). Neutralizing tumor-promoting chronic
inflammation: A magic bullet? : Erratum [Erratum], Science 339.
Cox. CP: Ruhl. Dl. (1966). Simplified computation of confidence intervals for relative potencies
using Fieller's theorem. J Pharm Sci 55: 368-371.
http://dx.doi.Org/10.1002/ips.2600550403
Craig. EA: Schlesinger. Ml. (1985). The heat shock response. CRC Crit Rev Biochem 18: 239-280.
http://dx.doi. org/10.3109 /10409238509085135
Cristina Lo Celsol. aDTS. (2011). The haematopoietic stem cell niche at a glance. J Cell Sci 124:
3529-3535. http://dx.doi.org/10.1242/ics.074112
Crump. KS: Bussard. DA: Chen. C: Tinot. 1: Subramaniam. R. (2014). The bottom-up approach does
not necessarily bound low-dose risk [Letter], Regul Toxicol Pharmacol 70: 735-736.
http://dx.doi.Org/10.1016/i.yrtph.2014.10.008
Crump. KS: Chen. C: Chiu. WA: Louis. TA: Portier. CI: Subramaniam. RP: White. PP. (2010). What
role for biologically based dose-response models in estimating low-dose risk? [Review],
Environ Health Perspect 118: 585-588. http://dx.doi.org/10.1289/ehp.0901249
Crump. KS: Chen. C: Fox. IF: Subramaniam. R: van Landingham. C. (2009). Reply to: Letter to the
editor. Formaldehyde risk assessment [Letter], Ann Occup Hyg 53: 181-189.
Crump. KS: Chen. C: Fox. IF: Van Landingham. C: Sumbramaniam. R. (2008). Sensitivity analysis of
biologically motivated model for formaldehyde-induced respiratory cancer in humans. Ann
Occup Hyg 52: 481-495. http://dx.doi.org/10.1093/annhyg/men038
Crump. KS: Hoel. DG: Langlev. CH: Peto. R. (1976). Fundamental carcinogenic processes and their
implications for low dose risk assessment Cancer Res 36: 2973-2979.
Crump. KS: Subramaniam. RP: Van Landingham. CB. (2005). A numerical solution to the
nonhomogeneous two-stage MVK model of cancer. Risk Anal 25: 921-926.
http://dx.doi.Org/10.llll/i.1539-6924.2005.00651.x
Curado. MP: Shin. HR: Storm. H: Ferlav. 1: Heanue. M: Boyle. P. (2007). Cancer incidence in five
continents, Vol. IX [IARC Monograph], In Cancer Incidence in Five Continents Vol IX. (IARC
Scientific Publication No. 160). Lyon, France: International Agency for Research on Cancer
(IARC).
D'Errico. A: Pasian. S: Baratti. A: Zanelli. R: Alfonzo. S: Gilardi. L: Beatrice. F: Bena. A: Costa. G.
(2009). A case-control study on occupational risk factors for sino-nasal cancer. Occup
Environ Med 66: 448-455. http://dx.doi.org/10.1136/oem.2008.041277
Dalbev. WE. (1982). Formaldehyde and tumors in hamster respiratory tract. Toxicology 24: 9-14.
http://dx.doi.Org/10.1016/0300-483Xr82190058-0
Dannemiller. KC: Murphy. IS: Dixon. SL: Pennell. KG: Suuberg. EM: Tacobs. DE: Sandel. M. (2013).
Formaldehyde concentrations in household air of asthma patients determined using
colorimetric detector tubes. Indoor Air 23: 285-294. http://dx.doi.org/10.llll/ina.12024
de Graaf. B: Clore. A: Mccullough. AK. (2009). Cellular pathways for DNA repair and damage
tolerance of formaldehyde-induced DNA-protein crosslinks. DNA Repair 8: 1207-1214.
http://dx.doi.Org/10.1016/i.dnarep.2009.06.007
de Kruijf. El: Alkemade. GM: van Os. R: Fibbe. WE: van Pel. M. (2014). Peripheral blood
hematopoietic stem and progenitor cell frequency is unchanged in patients with alpha-1-
antitrypsin deficiency. Int J Hematol 99: 714-720. http://dx.doi.org/10.1007/sl2185-014-
1581-3
Dean. TH: Lauer. LP: House. RV: Murray. MI: Stillman. WS: Irons. RD: Steinhagen. WH: Phelps. MC:
Adams. DO. (1984). Studies of immune fuction and host resistance in B6C3F1 mice exposed
to formaldehyde. Toxicol Appl Pharmacol 72: 519-529. http://dx.doi.org/10.1016/0Q41-
008XC84190129-7
This document is a draft for review purposes only and does not constitute Agency policy.
R-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Dell. L: Teta. MI. (1995). Mortality among workers at a plastics manufacturing and research and
development facility: 1946-1988. AmJInd Med 28: 373-384.
http://dx.doi.Org/10.1002/aiim.4700280307
Deltour. L: Foglio. MH: Duester. G. (1999). Metabolic deficiencies in alcohol dehydrogenase Adhl,
Adh3, and Adh4 null mutant mice. Overlapping roles of Adhl and Adh4 in ethanol clearance
and metabolism of retinol to retinoic acid. J Biol Chem 274: 16796-16801.
http://dx.doi.org/10.1074/ibc.274.24.16796
DeMarini. DM: Shelton. ML: Kohan. MT: Hudgens. EE: Kleindienst. TE: Ball. LM: Walsh. D: de Boer. TG:
Lewis-Bevan. L: Rabinowitz. 1. R.: Claxton. LP: Lewtas. 1. (2000). Mutagenicity in lung of Big
Blue(R) mice and induction of tandem-base substitutions in Salmonella by the air pollutant
peroxyacetyl nitrate (PAN): Predicted formation of intrastrand cross-links. MutatRes
Fundam Mol Mech Mutagen 457: 41-55. http://dx.doi.Org/10.1016/S0027-5107r00100121-
4
Dingier. FA: Wang. M: Mu. A: Millington. CL: Oberbeck. N: Watcham. S: Pontel. LB: Kamimae-
Lanning. AN: Langevin. F: Nadler. C: Cordell. RL: Monks. PS: Yu. R: Wilson. NK: Hira. A:
Yoshida. K: Mori. M: Okamoto. Y: Okuno. Y: Muramatsu. H: Shiraishi. Y: Kobavashi. M:
Moriguchi. T: Osumi. T: Kato. M: Mivano. S: Ito. E: Koiima. S: Yabe. H: Yabe. M: Matsuo. K:
Ogawa. S: Gottgens. B: Hodskinson. MRG: Takata. M: Patel. KT. (2020). Two aldehyde
clearance systems are essential to prevent lethal formaldehyde accumulation in mice and
humans. Mol Cell 80: 996-1012.el019. http://dx.doi.Org/10.1016/i.molcel.2020.10.012
Dinsdale. D: Riley. RA: Verschovle. RD. (1993). Pulmonary cytochrome P450 in rats exposed to
formaldehyde vapor. Environ Res 62: 19-27. http://dx.doi.org/10.1006/enrs.1993.1085
Dix. DT. (1997). Hsp 70 expression and function during gametogenesis. Cell Stress Chaperones 2: 73.
http://dx.doi.org/10.1379/1466-1268ri9971002<0073:heafdg>2.3.co:2
Dix. DT: Allen. TW: Collins. BW: Poorman-Allen. P: Mori. C: Blizard. PR: Brown. PR: Goulding. EH:
Strong. BP: Eddy. EM. (1997). HSP70-2 is required for desynapsis of synaptonemal
complexes during meiotic prophase in juvenile and adult mouse spermatocytes.
Pevelopment 124: 4595-4603.
Poi. S: Suzuki. S: Morishita. M: Yamada. M: Kanda. Y: Torii. S: Sakamoto. T. (2003). The prevalence of
IgE sensitization to formaldehyde in asthmatic children. Allergy 58: 668-671.
http://dx.doi.Org/10.1034/i.1398-9995.2003.00044.x
Pouglas. SP: Lai. TP: Tuluc. F: Schwartz. L: Kilpatrick. LE. (2008). Neurokinin-1 receptor expression
and function in human macrophages and brain: perspective on the role in HIV
neuropathogenesis. Ann N Y Acad Sci 1144: 90-96.
http://dx.doi.org/10.1196/annals.1418.007
Puong. A: Steinmaus. C: Mchale. CM: Vaughan. CP: Zhang. L. (2011). Reproductive and
developmental toxicity of formaldehyde: a systematic review [Review], Mutat Res 728: 118-
138. http://dx.doi.Org/10.1016/i.mrrev.2011.07.003
Putta. R: Pubal. PM: Svider. PF: Liu. IK: Baredes. S: Elov. TA. (2015). Sinonasal malignancies: A
population-based analysis of site-specific incidence and survival. Laryngoscope 125: 2491-
2497. http: / /dx.doi. or g/10.10 0 2/larv.25465
Pvkewicz. MS: Patterson. R: Cugell. PW: Harris. KE: Wu. AF. (1991). Serum IgE and IgG to
formaldehyde-human serum albumin: Lack of relation to gaseous formaldehyde exposure
and symptoms. J Allergy Clin Immunol 87: 48-57. http: / /dx. doi. or g /10.1016 /0 091 -
6749f91190212-7
Pzirasa. K: Ribeiro. S: Costa. R: Santos. LM: Lin. SC: Grosmark. A: Sotnikova. TP: Gainetdinov. RR:
Caron. MG: Nicolelis. MA. (2006). Popaminergic control of sleep-wake states. J Neurosci 26:
10577-10589. http://dx.doi.org/10.1523/TNEIJROSCT.l 767-06.2006
Eastmond. PA: Keshava. N: Sonawane. B. (2014). Lymphohematopoietic cancers induced by
chemicals and other agents and their implications for risk evaluation: An overview
This document is a draft for review purposes only and does not constitute Agency policy.
R-13 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
[Review], MutatRes Rev MutatRes 761: 40-64.
http://dx.doi.Org/10.1016/i.mrrev.2014.04.001
ECHA (European Chemicals Agency). (2012). Committee for risk assessment RAC. Opinion
proposing harmonized classification and labeling at EU level of formaldehyde.
https://echa.europa.eu/documents/10162/254a73cf-ff8d-4bf4-95dl-109fl3ef0f5a
Edling. C: Hellquist. H: Odkvist. L. (1987a). Occupational formaldehyde exposure and the nasal
mucosa. Rhinology 25: 181-187.
Edling. C: Hellquist. H: Odkvist. L. (1988). Occupational exposure to formaldehyde and
histopathological changes in the nasal mucosa. Br J Ind Med 45: 761-765.
http://dx.doi.org/10.1136/oem.45.ll.761
Edling. C: Tarvholm. B: Andersson. L: Axelson. 0. (1987b). Mortality and cancer incidence among
workers in an abrasive manufacturing industry. Br J Ind Med 44: 57-59.
http://dx.doi.Org/10.1136/oem.44.l.57
Edrissi. B: Taghizadeh. K: Dedon. PC. (2013a). Quantitative analysis of histone modifications:
formaldehyde is a source of pathological n(6)-formyllysine that is refractory to histone
deacetylases. PLoS Genet 9: el003328. http://dx.doi.org/10.1371/iournal.pgen.1003328
Edrissi. B: Taghizadeh. K: Moeller. BC: Kracko. D: Dovle-Eisele. M: Swenberg. TA: Dedon. PC.
(2013b). Dosimetry of N6-formyllysine adducts following [13C2H2]-formaldehyde exposures
in rats. Chem Res Toxicol 26: 1421-1423. http://dx.doi.org/10.1021/tx400320u
Egle. TL. Tr. (1972). Retention of inhaled formaldehyde, propionaldehyde, and acrolein in the dog.
Arch Environ Health 25: 119-124. http://dx.doi.org/10.1080/00039896.1972.10666147
EHS Consultants Ltd. (EHS Consultants Limited). (1999). Consultancy study for indoor air pollution
in offices and public places in Hong Kong (Agreement no. CE 14/95). Hong Kong:
Environmental Protection Department
El-Zein. RA: Schabath. MB: Etzel. CI: Lopez. MS: Franklin. ID: Spitz. MR. (2006). Cytokinesis-blocked
micronucleus assay as a novel biomarker for lung cancer risk. Cancer Res 66: 6449-6456.
http://dx.doi.Org/10.1158/0008-5472-06-0326
Elenkov. IT: Wilder. RL: Chrousos. GP: Vizi. ES. (2000). The sympathetic nerve-an integrative
interface between two supersystems: the brain and the immune system [Review],
Pharmacol Rev 52: 595-638.
Erb. KT: Le Gros. G. (1996). The role of Th2 type CD4+ T cells and Th2 type CD8+ T cells in asthma
[Review], Immunol Cell Biol 74: 206-208. http://dx.doi.org/10.lQ38/icb.1996.29
Erdei. E: Bobvos. 1: Brozik. M: Paldv. A: Farkas. I: Vaskovi. E: Rudnai. P. (2003). Indoor air pollutants
and immune biomarkers among Hungarian asthmatic children. Arch Environ Occup Health
58: 337-347.
Ericson. A: Eriksson. M: Kallen. B: Westerholm. P: Zetterstrom. R. (1984). Delivery outcome of
women working in laboratories during pregnancy. Arch Environ Health 39:5-10.
Escanilla. 0: Yuhas. C: Marzan. D: Linster. C. (2009). Dopaminergic modulation of olfactory bulb
processing affects odor discrimination learning in rats. Behav Neurosci 123: 828-833.
http://dx.doi.org/10.1037/a0Q15855
Ezrattv. V: Bonav. M: Neukirch. C: Orset-Guillossou. G: Dehoux. M: Koscienlnv. S: Cabanes. PA:
Lambrozo. I: Aubier. M. (2007). Effect of formaldehyde on asthmatic response to inhaled
allergen challenge. Environ Health Perspect 115: 210-214.
http://dx.doi.org/10.1289/ehp.9414
Fabbri. M: Garzon. R: Cimmino. A: Liu. Z: Zanesi. N: Callegari. E: Liu. S: Alder. H: Costinean. S:
Fernandez-Cvmering. C: Volinia. S: Guler. G: Morrison. CD: Chan. KK: Marcucci. G: Calin. GA:
Huebner. K: Croce. CM. (2007). MicroRNA-29 family reverts aberrant methylation in lung
cancer by targeting DNA methyltransferases 3A and 3B. Proc Natl Acad Sci USA 104: 15805-
15810. http://dx.doi.org/10.1073/pnas.0707628104
This document is a draft for review purposes only and does not constitute Agency policy.
R-14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Falk. IE: Tuto. IE: Stridh. G: Bvlin. G. (1994). Dose-response study of formaldehyde on nasal mucosa
swelling. A study on residents with nasal distress at home. Am J Rhinol Allergy 8: 143-146.
http://dx.doi. org/10.2500/105065894781874412
Fang. F: Ouinlan. P: Ye. W: Barber. MK: Umbach. DM: Sandler. DP: Kamel. F. (2009). Workplace
exposures and the risk of amyotrophic lateral sclerosis. Environ Health Perspect 117: 1387-
1392. http://dx.doi.org/10.1289/ehp.0900580
Fantel. AG: Macphail.BT. (1982). THE TERATOGENICITY OF COCAINE. Teratology 26: 17-19.
http://dx.doi.Org/10.1002/tera.1420260104
Fenech. M: Holland. N: Zeiger. E: Chang. WP: Burgaz. S: Thomas. P: Bolognesi. C: Knasmueller. S:
Kirsch-Volders. M: Bonassi. S. (2011). The HUMN and HUMNxL international collaboration
projects on human micronucleus assays in lymphocytes and buccal cells--past, present and
future [Review], Mutagenesis 26: 239-245. http://dx.doi.org/10.1093/mutage/geq051
Feron. VI: Bruvnties. TP: Woutersen. RA: Immel. HR: Appelman. LM. (1988). Nasal tumours in rats
after short-term exposure to a cytotoxic concentration of formaldehyde. Cancer Lett 39:
101 -111, http://dx.doi.org/10.1016/0304-3835r88190045-6
Ferris. BG. (1978). Epidemiology standardization project (American Thoracic Society). Am Rev
Respir Dis 118: 1-120.
Fishbein. L. (1992). Exposure from occupational versus other sources [Review], Scand J Work
Environ Health 18: 5-16.
Flamant-Hulin. M: Caillaud. D: Sacco. P: Penard-Morand. C: Annesi-Maesano. I. (2010). Air pollution
and increased levels of fractional exhaled nitric oxide in children with no history of airway
damage. J Toxicol Environ Health A 73: 272-283.
http://dx.doi.Org/10.1080/15287390903249206
Fox. EM. (1985). Urea formaldehyde foam insulation: defusing a timebomb. Am J Law Med 11: 81-
104.
Franklin. P: Dingle. P: Stick. S. (2000). Raised exhaled nitric oxide in healthy children is associated
with domestic formaldehyde levels. Am J Respir Crit Care Med 161: 1575-1759.
http://dx.doi.Org/10.1164/airccm.161.5.9905061
Franklin. P: Tan. M: Hemv. N: Hall. GL. (2019). Maternal Exposure to Indoor Air Pollution and Birth
Outcomes. Int J Environ Res Public Health 16. http: //dx.doi.org/10.3390/ijerphl6081364
Fransman. W: Mclean. D: Douwes. I: Demers. PA: Leung. V: Pearce. N. (2003). Respiratory symptoms
and occupational exposures in New Zealand plywood mill workers. Ann Occup Hyg 47: 287-
295. http://dx.doi.org/10.1093/annhyg/meg046
French. IE: Gatti. DM: Morgan. PL: Kissling. GE: Shocklev. KR: Knudsen. GA: Shepard. KG: Price. HC:
King. D: Witt. KL: Pedersen. LC: Munger. SC: Svenson. KL: Churchill. GA. (2015). Diversity
outbred mice identify population-based exposure thresholds and genetic factors that
influence benzene-induced genotoxicity. Environ Health Perspect 123: 237-245.
http://dx.doi.org/10.1289/ehp.14082Q2
Fuiimaki. H: Kurokawa. Y: Kakevama. M: Kunugita. N: Fueta. Y: Fukuda. T: Hori. H: Arashidani. K.
(2004a). Inhalation of low-level formaldehyde enhances nerve growth factor production in
the hippocampus of mice. Neuroimmunomodulation 11: 373-375.
http://dx.doi.Org/10.1159/000080147
Fuiimaki. H: Kurokawa. Y: Kunugita. N: Kikuchi. M: Sato. F: Arashidani. K. (2004b). Differential
immunogenic and neurogenic inflammatory responses in an allergic mouse model exposed
to low levels of formaldehyde. Toxicology 197: 1-13.
http://dx.doi.Org/10.1016/i.tox.2003.ll.015
Gandhi. M: Aweeka. F: Greenblatt. RM: Blaschke. TF. (2004). Sex differences in pharmacokinetics
and pharmacodynamics [Review], Annu Rev Pharmacol Toxicol 44: 499-523.
http://dx.doi.org/10.1146/annurev.pharmtox.44.101802.121453
This document is a draft for review purposes only and does not constitute Agency policy.
R-15 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Garavcoechea. II: Crossan. GP: Langevin. F: Daly. M: Arends. MI: Patel. KT. (2012). Genotoxic
consequences of endogenous aldehydes on mouse haematopoietic stem cell function.
Nature 489: 571-575. http ://dx.doi. or g/10.10 38/nature 11368
Garcia-Calderon. CB: Bejarano-Garcia. TA: Tinoco-Gago. I: Castro. Ml: Moreno-Gordillo. P: Piruat. II:
Caballero-Velazquez. T: Perez-Simon. TA: Rosado. IV. (2018). Genotoxicity of tetrahydrofolic
acid to hematopoietic stem and progenitor cells. Cell Death Differ 25: 1967-1979.
http: //dx.doi.org/10.1038/s41418-018-0089-4
Garcia. GTM: Schroeter. ID: Segal. RA: Stanek. 1: Foureman. GL: Kimbell. IS. (2009). Dosimetry of
nasal uptake of water-soluble and reactive gases: A first study of interhuman variability.
Inhal Toxicol 21: 607-618. http://dx.doi.org/10.1080/08958370802320186
Gardner. Ml: Pannett. B: Winter. PD: Cruddas. AM. (1993). A cohort study of workers exposed to
formaldehyde in the British chemical industry: An update. Br J Ind Med 50: 827-834.
http://dx.doi.Org/10.1136/oem.50.9.827
Garrett. MH: Hooper. MA: Hooper. BM: Ravment. PR: Abramson. MT. (1999). Increased risk of
allergy in children due to formaldehyde exposure in homes. Allergy 54: 330-337.
http://dx.doi.Org/10.1034/i.1398-9995.1999.00763.x
Garzon. R: Heaphv. CEA: Havelange. V: Fabbri. M: Violinia. S: Tsao. T: Zanesi. H: Kornblau. SM:
Marcucci. G: Calin. GA: Andreeff. M: Croce. CM. (2009). MicroRNA 29b functions in acute
myeloid leukemia. Blood 114: 5331-5341. http://dx.doi.org/10.1182/blood-2009-03-
211938
Gee. IL: Watson. AFR: Tavernier. G: Stewart. LI: Fletcher. G: Niven. RM. (2005). Indoor air quality,
environmental tobacco smoke and asthma: A case control study of asthma in a community
population. Indoor Built Environ 14: 215-219.
http://dx.doi.org/10.1177/1420326X05054288
Gentry. PR: Rodricks. TV: Turnbull. D: Bachand. A: Van Landingham. C: Shipp. AM: Albertini. RT:
Irons. R. (2013). Formaldehyde exposure and leukemia: Critical review and reevaluation of
the results from a study that is the focus for evidence of biological plausibility [Review], Crit
Rev Toxicol 43: 661-670. http: //dx.doi.org/10.3109/10408444.2013.818618
George. L: Brightling. CE. (2016). Eosinophilic airway inflammation: role in asthma and chronic
obstructive pulmonary disease [Review], 7: 34-51.
http://dx.doi.org/10.1177/2040622315609251
Gerard. C. (2005). Biomedicine. Asthmatics breathe easier when it's SNO-ing [Comment], Science
308: 1560-1561. http://dx.doi.org/10.1126/science. 1114163
Gerin. M: Siemiatvcki. 1: Nadon. L: Dewar. R: Krewski. D. (1989). Cancer risks due to occupational
exposure to formaldehyde: Results of a multi-site case-control study in Montreal. Int J
Cancer 44: 53-58. http://dx.doi.org/10.1002/iic.2910440110
Giavina-Bianchi. P: Aun. MV: Takeiima. P: Kalil. I: Agondi. RC. (2016). United airway disease: current
perspectives [Review], Journal of Asthma and Allergy 9: 93-100.
http: / /dx.doi.org/10.2147/TAA.S81541
Ginsberg. G: Guvton. K: Tohns. D: Schimek. T: Angle. K: Sonawane. B. (2010). Genetic polymorphism
in metabolism and host defense enzymes: Implications for human health risk assessment
[Review], Crit Rev Toxicol 40: 575-619. http: //dx.doi.org/10.3109/10408441003742895
Ginsberg. GL: Asgharian. B: Kimbell. IS: Ultman. IS: Tarabek. AM. (2008). Modeling approaches for
estimating the dosimetry of inhaled toxicants in children [Review], J Toxicol Environ Health
A 71: 166-195. http://dx.doi.Org/l 0.1080/15287390701597889
Ginsberg. GL: Foos. BP: Firestone. MP. (2005). Review and analysis of inhalation dosimetry methods
for application to children's risk assessment [Review], J Toxicol Environ Health A 68: 573-
615. http:/ /dx.doi.org/10.1080/15287390590921793
Gochfeld. M. (2007). Framework for gender differences in human and animal toxicology [Review],
Environ Res 104: 4-21. http://dx.doi.Org/10.1016/i.envres.2005.12.005
This document is a draft for review purposes only and does not constitute Agency policy.
R-16 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Gofmekler. VA: Pushkina. NN: Klevtsova. GN. (1968). [Various biochemical shifts during a study of
the embryotropic effect of benzene and formaldehyde]. Gig Sanit 33: 96-98.
Golalipour. Ml: Azarhoush. R: Ghafari. S: Gharravi. AM: Fazeli. SA: Davarian. A. (2007).
Formaldehyde exposure induces histopathological and morphometric changes in the rat
testis. Folia Morphol (Warsz) 66: 167-171.
Goldstein. BP. (2011). Hematological and toxicological evaluation of formaldehyde as a potential
cause of human leukemia [Review], Hum Exp Toxicol 30: 725-735.
http://dx.doi.org/10.1177/0960327110381682
Goodson. WH: Lowe. L: Carpenter. DO: Gilbertson. M: Manaf Ali. A: Lopez de Cerain Salsamendi. A:
Lasfar. A: Carnero. A: Azqueta. A: Amedei. A: Charles. AK: Collins. AR: Ward. A: Salzberg. AC:
Colacci. A: Olsen. AK: Berg. A: Barclay. BT: Zhou. BP: Blanco-Aparicio. C: Baglole. CI: Dong. C:
Mondello. C: Hsu. CW: Naus. CC: Yediou. C: Curran. CS: Laird. DW: Koch. DC: Carlin. DT:
Felsher. DW: Roy. D: Brown. DG: Ratovitski. E: Ryan. EP: Corsini. E: Roias. E: Moon. EY:
Laconi. E: Marongiu. F: Al-Mulla. F: Chiaradonna. F: Darroudi. F: Martin. FL: Van Schooten.
FT: Goldberg. GS: Wagemaker. G: Nangami. GN: Calaf. GM: Williams. G: Wolf. GT: Koppen. G:
Brunborg. G: Lverlv. HK: Krishnan. H: Ab Hamid. H: Yasaei. H: Sone. H: Kondoh. H: Salem.
HK: Hsu. HY: Park. HH: Koturbash. I: Miousse. IR: Scovassi. AI: Klaunig. IE: Vondracek. 1:
Raiu. 1: Roman. 1: Wise. TP: Whitfield. 1. R.: Woodrick. 1: Christopher. TA: Ochieng. 1: Martinez-
Leal. IF: Weisz. 1: Kravchenko. 1: Sun. 1: Prudhomme. KR: Narayanan. KB: Cohen-Solal. KA:
Moorwood. K: Gonzalez. L: Soucek. L: Tian. L: D'Abronzo. LS: Lin. LT: Li. L: Gulliver. L:
Mccawlev. LI: Memeo. L: Vermeulen. L: Levns. L: Zhang. L: Valverde. M: Khatami. M:
Romano. MF: Chapellier. M: Williams. MA: Wade. M: Maniili. MH: Lleonart. ME: Xia. M:
Gonzalez. Ml: Karamouzis. MV: Kirsch-Volders. M: Vaccari. M: Kuemmerle. NB: Singh. N:
Cruickshanks. N: Kleinstreuer. N: van Larebeke. N: Ahmed. N: Ogunkua. 0: Krishnakumar.
PK: Vadgama. P: Marignani. PA: Ghosh. PM: Ostroskv-Wegman. P: Thompson. PA: Dent. P:
Heneberg. P: Darbre. P: Sing Leung. P: Nangia-Makker. P: Cheng. OS: Robev. RB: Al-Temaimi.
R: Roy. R: Andrade-Vieira. R: Sinha. RK: Mehta. R: Vento. R: Pi Fiore. R: Ponce-Cusi. R:
Dornetshuber-Fleiss. R: Nahta. R: Castellino. RC: Palorini. R: Abd Hamid. R: Langie. SA:
Eltom. SE: Brooks. SA: Rveom. S: Wise. SS: Bay. SN: Harris. SA: Papagerakis. S: Romano. S:
Pavanello. S: Eriksson. S: Forte. S: Casey. SC: Luanpitpong. S: Lee. TT: Otsuki. T: Chen. T:
Massfelder. T: Sanderson. T: Guarnieri. T: Hultman. T: Dormov. V: Odero-Marah. V:
Sabbisetti. V: Maguer-Satta. V: Rathmell. WK: Engstrom. W: Decker. WK: Bisson. WH:
Roianasakul. Y: Luqmani. Y: Chen. Z: Hu. Z. (2015). Assessing the carcinogenic potential of
low-dose exposures to chemical mixtures in the environment: the challenge ahead [Review],
Carcinogenesis 36 Suppl 1: S254-S296. http://dx.doi.org/10.1093/carcin/bgv039
Gorski. P: Krakowiak. A. (1991). Formaldehyde-induced bronchial asthma-does it really exist? Pol
J Occup Med Environ Health 4: 317-320.
Grammer. LC: Harris. KE: Shaughnessv. MA: Sparks. P: Avars. GH: Altman. LC: Patterson. R. (1990).
Clinical and immunologic evaluation of 37 workers exposed to gaseous formaldehyde. J
Allergy Clin Immunol 86: 177-181. http: //dx.doi.org/10.1016/S0091-6749C05180063-6
Granick. TL: Simon. SI: Borjesson. PL. (2012). Hematopoietic stem and progenitor cells as effectors
in innate immunity. 2012: 165107. http://dx.doi.org/10.1155/2012/1651Q7
Greaves. M. (1999). Molecular genetics, natural history and the demise of childhood leukaemia. Eur
J Cancer 35: 173-185. http://dx.doi.org/10.1016/S0959-8049C98100433-X
Greaves. MF. (2004). Biological models for leukemia and lymphoma. In P Buffler; JM Rice; R Baan; M
Bird; P Boffetta (Eds.), Mechanisms of carcinogenesis: Contributions of molecular
epidemiology (pp. 351-372). Lyon, France: International Agency for Research on Cancer.
Green. PI: Bascom. R: Healev. EM: Hebel. TR: Sauder. LR: Kulle. TT. (1989). Acute pulmonary
response in healthy, nonsmoking adults to inhalation of formaldehyde and carbon. J Toxicol
Environ Health 28: 261-275. http://dx.doi.org/10.1080/152873989Q9531347
This document is a draft for review purposes only and does not constitute Agency policy.
R-17 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Green. DT: Sauder. LR: Kulle. TT: Bascom. R. (1987). Acute response to 3.0 ppm formaldehyde in
exercising healthy nonsmokers and asthmatics. Am Rev Respir Dis 135: 1261-1266.
http://dx.doi.Org/10.1164/arrd.1987.135.6.1261
Gu. Y: Fujimiva. Y: Kunugita. N. (2008). Long-term exposure to gaseous formaldehyde promotes
allergen-specific IgE-mediated immune responses in a murine model. Hum Exp Toxicol 27:
37-43. http://dx.doi.org/10.1177/0960327108088973
Gustafson. P: Barregard. L: Lindahl. R: Sallsten. G. (2005). Formaldehyde levels in Sweden: Personal
exposure, indoor, and outdoor concentrations. J Expo Anal Environ Epidemiol 15: 252-260.
http ://dx. doi. or g/10.10 3 8 /si .i ea. 75 0 0 3 9 9
Gustavsson. P: Takobsson. R: Tohansson. H: Lewin. F: Norell. S: Rutkvist. LE. (1998). Occupational
exposures and squamous cell carcinoma of the oral cavity, pharynx, larynx, and oesophagus:
A case-control study in Sweden. Occup Environ Med 55: 393-400.
Hall. A: Harrington. TM: Aw. TC. (1991). Mortality study of British pathologists. Am J Ind Med 20: 83-
89. http://dx.doi.Org/10.1002/aiim.4700200108
Hamelmann. E: Vella. AT: Oshiba. A: Kappler. TW: Marrack. P: Gelfand. EW. (1997). Allergic airway
sensitization induces T cell activation but not airway hyperresponsiveness in B cell-
deficient mice. Proc Natl Acad Sci USA 94: 1350-1355.
Han. P: Zhao. 1: Liu. SB: Yang. CI: Wang. YO: Wu. GC: Xu. DM: Mi. WL. (2013). Interleukin-33
mediates formalin-induced inflammatory pain in mice. Neuroscience 241: 59-66.
http://dx.doi.Org/10.1016/i.neuroscience.2013.03.019
Han. SP: Zhou. DX: Lin. P: Oin. Z: An. L: Zheng. LR: Lei. L. (2015). Formaldehyde exposure induces
autophagy in testicular tissues of adult male rats. Environ Toxicol 30: 323-331.
http ://dx. doi. or g/10.10 0 2 /tox. 21910
Hanahan. D: Weinberg. RA. (2000). The hallmarks of cancer [Review], Cell 100: 57-70.
http: //dx.doi.org/10.1016/S0092-8674r00181683-9
Hanahan. D: Weinberg. RA. (2011). Hallmarks of cancer: The next generation [Review], Cell 144:
646-674. http://dx.doi.Org/10.1016/i.cell.2011.02.013
Hankinson. TL: Odencrantz. TR: Fedan. KB. (1999). Spirometric reference values from a sample of the
general US population. Am J Respir Crit Care Med 159: 179-187.
http://dx.doi.Org/10.1164/ajrccm.159.l.9712108
Hanrahan. LP: Dally. KA: Anderson. HA: Kanarek. MS: Rankin. 1. (1984). Formaldehyde vapor in
mobile homes: A cross sectional survey of concentrations and irritant effects. Am J Public
Health 74: 1026-1027. http://dx.doi.Org/10.2105/aiph.74.9.1026
Hansen. 1: Olsen. TH. (1995). Formaldehyde and cancer morbidity among male employees in
Denmark. Cancer Causes Control 6: 354-360. http://dx.doi.org/10.1007/BF00Q51411
Hansen. 1: Olsen. TH: Larsen. AI. (1994). Cancer morbidity among employees in a Danish
pharmaceutical plant Int J Epidemiol 23: 891-898. http://dx.doi.Org/10.1093/iie/23.5.891
Hardell. L: Tohansson. B: Axelson. 0. (1982). Epidemiological study of nasal and nasopharyngeal
cancer and their relation to phenoxy acid or chlorophenol exposure. Am J Ind Med 3: 247-
257. http:/ /dx.doi.org/10.1002/aiim.4700030304
Harkema. TR: Carey. SA: Wagner. TG. (2006). The nose revisited: A brief overview of the comparative
structure, function, and toxicologic pathology of the nasal epithelium [Review], Toxicol
Pathol 34: 252-269. http://dx.doi.org/10.1080/01926230600713475
Harkema. TR: Nikula. KT: Haschek. WM. (2013). Respiratory system. In W Haschek; C Rousseaux; M
Wallig (Eds.), Haschek and Rousseaux's handbook of toxicologic pathology (3rd ed., pp.
1935-2003). Waltham, MA: Academic Press. http://dx.doi.org/10.1016/B978-0-12-
415759-0.00051-0
Harrington. TM: Oakes. D. (1984). Mortality study of British pathologists 1974-80. Occup Environ
Med 41: 188-191. http://dx.doi.Org/10.1136/oem.41.2.188
This document is a draft for review purposes only and does not constitute Agency policy.
R-18 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Harris. NL: Stein. H: Coupland. SE: Hummel. M: Favera. RD: Pasqualucci. L: Chan. WC. (2001). New
approaches to lymphoma diagnosis [Review], Hematology Am Soc Hematol Educ
Programl94-220.
Harving. H: Korsgaard. 1: Dahl. R: Pedersen. OF: Molhave. L. (1986). Low concentrations of
formaldehyde in bronchial asthma: a study of exposure under controlled conditions. Br Med
J 293: 310.
Harving. H: Korsgaard. 1: Pedersen. OF: Malhave. L: Dahl. R. (1990). Pulmonary function and
bronchial reactivity in asthmatics during low-level formaldehyde exposure. Lung 168: 15-
21. http://dx.doi.org/10.1007/BF02719669
Haseman. IK: Hailev. TR. (1997). An update of the National Toxicology Program database on nasal
carcinogens. MutatRes 380: 3-11. http://dx.doi.org/10.1016/S0027-5107(97)00121-8
Hauptmann. M: Lubin lav. H: Stewart. PA: Hayes. RB: Blair. A. (2003). Mortality from
lymphohematopoietic malignancies among workers in formaldehyde industries. J Natl
Cancer Inst 95: 1615-1623. http://dx.doi.org/10.1093/inci/digO83
Hauptmann. M: Lubin. TH: Stewart. PA: Hayes. RB: Blair. A. (2004). Mortality from solid cancers
among workers in formaldehyde industries. Am J Epidemiol 159: 1117-1130.
http://dx.doi.org/10.1093/aie/kwhl74
Hauptmann. M: Stewart. PA: Lubin. TH: Beane Freeman. LE: Hornung. RW: Herrick. RF: Hoover. RN:
Fraumeni. IF. Tr: Blair. A: Hayes. RB. (2009). Mortality from lymphohematopoietic
malignancies and brain cancer among embalmers exposed to formaldehyde. J Natl Cancer
Inst 101: 1696-1708. http://dx.doi.org/10.1093/inci/dip416
Havashi. H: Kunugita. N: Arashidani. K: Fuiimaki. H: Ichikawa. M. (2004). Long-term exposure to
low levels of formaldehyde increases the number of tyrosine hydroxylase-immunopositive
periglomerular cells in mouse main olfactory bulb. Brain Res 1007: 192-197.
http://dx.doi.Org/10.1016/i.brainres.2003.12.052
Hayes. RB: Blair. A: Stewart. PA: Herrick. RF: Mahar. H. (1990). Mortality of U.S. embalmers and
funeral directors. Am J Ind Med 18: 641-652. http://dx.doi.org/10.1002/aiim.47001806Q3
Hayes. RB: Gerin. M: Raatgever. TW: de Bruvn. A. (1986a). Wood-related occupations, wood dust
exposure, and sinonasal cancer. Am J Epidemiol 124: 569-577.
http://dx.doi.org/10.1093/oxfordjournals.aje.all4429
Hayes. RB: Raatgever. TW: de Bruvn. A: Gerin. M. (1986b). Cancer of the nasal cavity and paranasal
sinuses, and formaldehyde exposure. Int J Cancer 37: 487-492.
http: //dx.doi.org/10.1002 /iic.2910370403
He. HI: Liu. HL: Wu. I: Lu. ZS: Yan. Y: Yang. X: Li. CM. (2005). A study on the acute irritation
responses and molecular mechanism of gaseous formaldehyde. In X Yang; B Zhao; R Zhao
(Eds.), Indoor Air 2005: Proceedings ofthe 10th International Conference on Indoor Air
Quality and Climate, vol 5 (pp. 3691-3695). Beijing, China: Tsinghua University Press.
https://www.isiaq.org/docs/PDFs/3691.pdf
Health Canada. f20011. Priority substances list assessment report Formaldehyde. Hull, Quebec,
Canada: Environment Canada and Health Canada.
Health Canada. (2006). Residential indoor air quality guideline. Formaldehyde.
http://healthycanadians.gc.ca/publications/healthy-living-vie-
saine/formaldehyde/alt/formaldehyde-eng.pdf
Heck. H: Casanova-Schmitz. M: Dodd. PB: Schachter. EN: Witek. TT: Tosun. T. (1985). Formaldehyde
(CH20) concentrations in the blood of humans and Fischer-344 rats exposed to CH20 under
controlled conditions. AIHA J 46: 1-3. http://dx.doi.org/10.1080/15298668591394275
Heck. H: Casanova. M. (1999). Pharmacodynamics of formaldehyde: Applications of a model for the
arrest of DNA replication by DNA-protein cross-links. Toxicol Appl Pharmacol 160: 86-100.
http://dx.doi.org/10.1006/taap.1999.8764
This document is a draft for review purposes only and does not constitute Agency policy.
R-19 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Heck. H: Casanova. M. (2004). The implausibility of leukemia induction by formaldehyde: A critical
review of the biological evidence on distant-site toxicity [Review], Regul Toxicol Pharmacol
40: 92-106. http://dx.doi.Org/10.1016/i.yrtph.2004.05.001
Heck. H: Chin. TY: Schmitz. MC. (1983). Distribution of [14C] formaldehyde in rats after inhalation
exposure. In JE Gibson (Ed.), Formaldehyde toxicity (pp. 26-37). Washington, DC:
Hemisphere Publishing.
Hedberg. 11: Grafstrom. RC: Vondracek. M: Sarang. Z: Warngard. L: Hoog. 10. (2001). Micro-array
chip analysis of carbonyl-metabolising enzymes in normal, immortalised and malignant
human oral keratinocytes. Cell Mol Life Sci 58: 1719-1726.
http://dx.doi. org/10.1007/PL00000810
Hedberg. IT: Hoog. TO: Nilsson. TA: Xi. Z: Elfwing. A: Grafstrom. RC. (2000). Expression of alcohol
dehydrogenase 3 in tissue and cultured cells from human oral mucosa. Am J Pathol 157:
1745-1755. http://dx.doi.Org/10.1016/S0002-9440n0164811-0
Heineman. EF: Olsen. TH: Pottern. LM: Gomez. M: Raffn. E: Blair. A. (1992). Occupational risk factors
for multiple myeloma among Danish men. Cancer Causes Control 3: 555-568.
http://dx.doi.Org/10.1007/BF00052753
Hemminki. K: Kyvronen. P: Lindbohm. ML. (1985). Spontaneous abortions and malformations in the
offspring of nurses exposed to anaesthetic gases, cytostatic drugs, and other potential
hazards in hospitals, based on registered information of outcome. J Epidemiol Community
Health 39: 141-147. http://dx.doi.org/10.1136/iech.39.2.141
Hemminki. K: Mutanen. P: Saloniemi. I: Niemi. ML: Vainio. H. (1982). Spontaneous abortions in
hospital staff engaged in sterilizing instruments with chemical agents. J Occup Environ Med
285: 1461-1463.
Herbert. C: Rietschel. RL. (2004). Formaldehyde and formaldehyde releasers: How much avoidance
of cross-reacting agents is required? Contact Derm 50: 371-373.
http: //dx.d0i.0rg/l 0.1111 /i.01 05-1873.2004.00381 ,x
Herbert. FA: Hessel. PA: Melenka. LS: Yoshida. K: Nakaza. M. (1994). Respiratory consequences of
exposure to wood dust and formaldehyde of workers manufacturing oriented strand board.
Arch Environ Health 49: 465-470. http://dx.doi.org/10.1080/00039896.1994.9955002
Hess. DT: Matsumoto. A: Kim. SO: Marshall. HE: Stamler. IS. (2005). Protein S-nitrosylation: Purview
and parameters [Review], Nat Rev Mol Cell Biol 6: 150-166.
http://dx.doi.org/10.lQ38/nrml569
Hester. SD: Barry. WT: Zou. F: Wolf. DC. (2005). Transcriptomic analysis of F344 rat nasal
epithelium suggests that the lack of carcinogenic response to glutaraldehyde is due to its
greater toxicity compared to formaldehyde. Toxicol Pathol 33: 415-424.
http://dx.doi.Org/10.1080/01926230590953105
Hester. SD: Benavides. GB: Yoon. L: Morgan. KT: Zou. F: Barry. W: Wolf. DC. (2003). Formaldehyde-
induced gene expression in F344 rat nasal respiratory epithelium. Toxicology 187: 13-24.
http://dx.d0i.0rg/l 0.1016/S0300-483Xr03100008-8
Hildesheim. A: Anderson. LM: Chen. CI: Cheng. YT: Brinton. LA: Daly. AK: Reed. CD: Chen. IH:
Caporaso. NE: Hsu. MM: Chen. TY: Idle. TR: Hoover. RN: Yang. CS: Chhabra. SK. (1997).
CYP2E1 genetic polymorphisms and risk of nasopharyngeal carcinoma in Taiwan. J Natl
Cancer Inst 89: 1207-1212. http://dx.doi.org/10.1093/inci/89.16.1207
Hildesheim. A: Dosemeci. M: Chan. CC: Chen. CI: Cheng. YT: Hsu. MM: Chen. IH: Mittl. BF: Sun. B:
Levine. PH: Chen. TY: Brinton. LA: Yang. CS. (2001). Occupational exposure to wood,
formaldehyde, and solvents and risk of nasopharyngeal carcinoma. Cancer Epidemiol
Biomarkers Prev 10: 1145-1153.
Hildesheim. A: West. S: Devevra. E: De Guzman. MF: Turado. A: Tones. C: Imai. I: Hinuma. Y. (1992).
Herbal medicine use, Epstein-Barr virus, and risk of nasopharyngeal carcinoma. Cancer Res
52: 3048-3051.
This document is a draft for review purposes only and does not constitute Agency policy.
R-20 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Hildesheim. A: West. S: Dosemeci. M: De Vevra. E: De Guzman. MF: Turado. A. (1993).
Nasopharyngeal carcinoma in the Philippines preliminary results from a case-control study
of multiple factors. In T Tursz; JS Pagano; DV Ablashi; G de The; G Lenoir; GR Pearson (Eds.),
The Epstein-Barr virus and associated diseases (pp. 743-747). Montrouge, France: John
Libbey Eurotext
Hill. AB. (1965). The environment and disease: Association or causation? Proc R Soc Med 58: 295-
300.
Hines. RN: McCarver. DG. (2002). The ontogeny of human drug-metabolizing enzymes: Phase I
oxidative enzymes [Review], J Pharmacol Exp Ther 300: 355-360.
http://dx.doi.Org/10.1124/ipet.300.2.355
Hisamitsu. M: Okamoto. Y: Chazono. H: Yonekura. S: Sakurai. D: Horiguchi. S: Hanazawa. T: Terada.
N: Konno. A: Matsuno. Y: Todaka. E: Mori. C. (2011). The influence of environmental
exposure to formaldehyde in nasal mucosa of medical students during cadaver dissection.
Allergol Int 60: 373-379. http: / /dx. do i. or g /10.2 3 3 2 /alle r golint. 10 - 0 A- 0 210
Ho. KF: Lee. SC: Tsai. WY. (2006a). Carbonyl compounds in the roadside environment of Hong Kong.
J Hazard Mater 133: 24-29. http://dx.doi.Org/10.1016/i.ihazmat2005.09.054
Ho. SSH: Yu. TZ: Chu. KW: Yeung. LL. (2006b). Carbonyl emissions from commercial cooking sources
in Hong Kong. J Air Waste Manag Assoc 56: 1091-1098.
http: //dx.doi.org/10.1080/10473289.2006.10464532
Hohnloser. W: Osswald. B: Lingens. F. (1980). ENZYMOLOGICAL ASPECTS OF CAFFEINE
DEMETHYLATION AND FORMALDEHYDE OXIDATION BY PSEUDOMONAS-PUTIDA-C1.
Hoppe Seylers Z Physiol Chem 361: 1763-1766.
Holland. N: Bolognesi. C: Kirsch-Volders. M: Bonassi. S: Zeiger. E: Knasmueller. S: Fenech. M. (2008).
The micronucleus assay in human buccal cells as a tool for biomonitoring DNA damage: the
HUMN project perspective on current status and knowledge gaps [Review], MutatRes Rev
MutatRes 659: 93-108. http://dx.doi.Org/10.1016/j.mrrev.2008.03.007
Hollstein. MC: Peri. L: Mandard. AM: Welsh. TA: Montesano. R: Metcalf. RA: Bak. M: Harris. CC.
(1991). Genetic-analysis of human esophageal tumors from 2 high-incidence geographic
areas - frequentp53 base substitutions and absence of ras mutations. Cancer Res 51: 4102-
4106.
Holmes. BT: Macarv. PA: Noble. A: Kemenv. DM. (1997). Antigen-specific CD8+ T cells inhibit IgE
responses and interleukin-4 production by CD4+ T cells. Eur J Immunol 27: 2657-2665.
http: //dx.doi.org/10.1002 /eii.l 830271027
Holmstrom. M: Rynnel-Dagoo. B: Wilhelmsson. B. (1989a). Antibody production in rats after long-
term exposure to formaldehyde. Toxicol Appl Pharmacol 100: 328-333.
http://dx.doi.org/10.1016/0041-008Xr89190318-9
Holmstrom. M: Wilhelmsson. B. (1988). Respiratory symptoms and pathophysiological effects of
occupational exposure to formaldehyde and wood dust Scand J Work Environ Health 14:
306-311.
Holmstrom. M: Wilhelmsson. B: Hellquist. H. (1989b). Histological changes in the nasal mucosa in
rats after long-term exposure to formaldehyde and wood dust. Acta Otolaryngol 108: 274-
283. http://dx.doi.Org/10.3109/00016488909125528
Holmstrom. M: Wilhelmsson. B: Hellquist. H: Rosen. G. (1989c). Histological changes in the nasal
mucosa in persons occupationally exposed to formaldehyde alone and in combination with
wood dust Acta Otolaryngol 107: 120-129.
http://dx.doi.Org/10.3109/00016488909127488
Holness. PL: Nethercott. TR. (1989). Health status of funeral service workers exposed to
formaldehyde. Arch Environ Occup Health 44: 222-228.
http://dx.doi.Org/10.1080/00039896.1989.9935887
This document is a draft for review purposes only and does not constitute Agency policy.
R-21 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Holness. PL: Sass-Kortsak. AM: Pilger. CW: Nethercott. TR. (1985). Respiratory Function And
Exposure-Effect Relationships In Wood Dust-Exposed And Control Workers. J Occup Med
27: 501-506.
Horton. AW: Tve. R: Stemmer. KL. (1963). Experimental carcinogenesis of the lung: Inhalation of
gaseous formaldehyde or an aerosol of coal tar by C3H mice. J Natl Cancer Inst 30: 31-40.
http ://dx. doi. or g/10.109 3 /i nci /3 0.1.31
Horvath. EP. Tr: Anderson. H. Tr: Pierce. WE: Hanrahan. L: Wendlick. ID. (1988). Effects of
formaldehyde on the mucous membranes and lungs: A study of an industrial population.
JAMA 259: 701-707. http://dx.doi.org/10.1001/iama.1988.03720050037020
Hosgood. HP. Ill: Zhang. L: Tang. X: Vermeulen. R: Hao. Z: Shen. M. in: Oiu. C: Ge. Y: Hua. M: Ti. Z: Li.
S: Xiong. 1. un: Reiss. B: Liu. S: Xin. KX: Azuma. M: Xie. Y: Freeman. LB: Ruan. X: Guo. W:
Galvan. N. oe: Blair. A: Li. L: Huang. H: Smith. MT: Rothman. N: Lan. 0. (2013). Occupational
exposure to formaldehyde and alterations in lymphocyte subsets. Am J Ind Med 56: 252-
257. http://dx.doi.org/10.1002/aiim.22088
Hsu. NY: Lee. CC: Wang. TY: Li. YC: Chang. HW: Chen. CY: Bornehag. CG: Wu. PC: Sundell. I: Su. HI.
(2012). Predicted risk of childhood allergy, asthma and reported symptoms using measured
phthalate exposure in dust and urine. Indoor Air 22: 186-199.
http: //dx.doi.org/10.1111 /i.l 600-0668.2011.00753.x
Huang. C: Liu. W: Cai. 1: Wang. X: Zou. Z: Sun. CI. (2017). Household formaldehyde exposure and its
associations with dwelling characteristics, lifestyle behaviours, and childhood health
outcomes in Shanghai, China. Build Environ 125: 143-152.
http://dx.doi.Org/10.1016/i.buildenv.2017.08.042
Hulin. M: Caillaud. D: Annesi-Maesano. I. (2010). Indoor air pollution and childhood asthma:
variations between urban and rural areas. Indoor Air 20: 502-514.
http: //dx.doi.org/10.1111 /i.l 600-0668.2010.00673.x
Hulsmann. AR: Dejongste. TC. (1996). Modulation of airway responsiveness by the airway
epithelium in humans: Putative mechanisms. Clin Exp Allergy 26: 1236-1242.
Hummel. T: Futschik. T: Frasnelli. 1: Hiittenbrink. KB. (2003). Effects of olfactory function, age, and
gender on trigeminally mediated sensations: a study based on the lateralization of
chemosensory stimuli. Toxicol Lett 140-141: 273-280. http://dx.doi.org/10.1016/s0378-
4274C03100078-X
Hwang. G: Yoon. C: Choi. 1. (2011). A case-control study: exposure assessment of vocs and
formaldehyde for asthma in children. Aerosol Air Qual Res 11: 908-914.
http://dx.doi.Org/10.4209/aaqr.2011.05.0072
I ARC (International Agency for Research on Cancer). (1995). Wood dust and formaldehyde. Lyon,
France. http: //monographs.iarc.fr/ENG/Monographs/vol62 /index.php
I ARC (International Agency for Research on Cancer). (2006). Formaldehyde, 2-butoxyethanol and
l-tert-butoxypropan-2-ol [IARC Monograph], Lyon, France.
https://publications.iarc.fr/Book-And-Report-Series/Iarc-Monographs-On-The-
Identification-0f-Carcinogenic-Hazards-To-Humans/Formaldehyde-2-Butoxyethanol-And-
l-Em-Tert-Em-Butoxypropan-2-ol-2006
IARC (International Agency for Research on Cancer). (2012). Formaldehyde [IARC Monograph], In
A review of human carcinogens: Chemical agents and related occupations (pp. 401-435).
Lyon, France. http://monographs.iarc.fr/ENG/Monographs/vollOOF/index.php
Ichinose. T: Miller. MG: Shibamoto. T. (1989). Gas chromatographic analysis of free and bound
malonaldehyde in rat liver homogenates. Lipids 24: 895-898.
http://dx.doi.org/10.1007/BF02535765
ICRP (International Commission on Radiological Protection). (1994). Human respiratory tract
model for radiological protection. Ann ICRP 24.
This document is a draft for review purposes only and does not constitute Agency policy.
R-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Tm. H: Oh. E: Mun. T: Khim. TY: Lee. E: Kang. HS: Kim. E: Kim. H: Won. NH: Kim. YH: Tung. WW: Sul. D.
(2006). Evaluation of toxicological monitoring markers using proteomic analysis in rats
exposed to formaldehyde. J Proteome Res 5: 1354-1366.
http://dx.doi.org/10.1021/pr050437b
IPCS (International Programme on Chemical Safety). (2012). Harmonization project document no.
10: Guidance for immunotoxicity risk assessment for chemicals. (Harmonization Project
Document No. 10). Geneva, Switzerland: World Health Organization.
http://www.inchem.org/documents/harmproi/harmproi/harmproilO.pdf
Isa. KNM: Hashim. Z: Talaludin. 1: Norback. D. an: Tabbar. MA: Hashim. TH. (2020). The Impact of
Exposure to Indoor Pollutants on Allergy and Lung Inflammation among School Children in
Selangor, Malaysia: An Evaluation Using Factor Analysis. Aerosol Air Qual Res 20: 2371-
2383. http://dx.doi.Org/10.4209/aaqr.2020.03.0128
Ito. K: Sakamoto. T: Havashi. Y: Morishita. M: Shibata. E: Sakai. K: Takeuchi. Y: Torii. S. (1996). Role
of tachykinin and bradykinin receptors and mast cells in gaseous formaldehyde-induced
airway microvascular leakage in rats. Eur J Pharmacol 307: 291-298.
http://dx.doi.Org/10.1016/0014-2999f96100285-3
Takab. GT. (1992). Relationship between carbon black particulate-bound formaldehyde, pulmonary
antibacterial defenses, and alveolar macrophage phagocytosis. Inhal Toxicol 4: 325-342.
http: //dx.doi.org/10.3109/08958379209145312
Takab. GT: Risbv. TH: Hemenwav. DR. (1992). Use of physical chemistry and in vivo exposure to
investigate the toxicity of formaldehyde bound to carbonaceous particles in the murine lung
(pp. 1-39, discussion 41-39). (ISSN 1041-5505
EISSN 2688-6855
NTIS/02982016_2). Cambridge, MA: Health Effects Institute.
Takab. MG: Klupp. T: Besenvei. K: Biro. A: Maior. I: Tompa. A. (2010). Formaldehyde-induced
chromosomal aberrations and apoptosis in peripheral blood lymphocytes of personnel
working in pathology departments. MutatRes 698: 11-17.
http://dx.doi.Org/10.1016/i.mrgentox.2010.02.015
Takobsson. K: Mikoczv. Z: Skerfving. S. (1997). Deaths and tumours among workers grinding
stainless steel: a follow up. Occup Environ Med 54: 825-829.
http://dx.doi.org/10.1136/oem.54.ll.825
Takopovic. M: Thomas. A: Balasubramaniam. S: Schrump. D: Giaccone. G: Bates. SE. (2013).
Targeting the epigenome in lung cancer: expanding approaches to epigenetic therapy
[Review], 3: 261. http://dx.doi.org/10.3389/fonc.2013.00261
Tensen. DE: Belka. GK: Du Bois. GC. (1998). S-Nitrosoglutathione is a substrate for rat alcohol
dehydrogenase class III isoenzyme. Biochem J 331: 659-668.
http://dx.doi.org/10.1042/bi3310659
Ti. Z: Li. X: Fromowitz. M: Mutter-Rottmaver. E: Tung. I: Smith. MT: Zhang. L. (2014). Formaldehyde
induces micronuclei in mouse erythropoietic cells and suppresses the expansion of human
erythroid progenitor cells. Toxicol Lett 224: 233-239.
http://dx.doi.Org/10.1016/j.toxlet.2013.10.028
Tia. X: Tia. 0: Zhang. Z: Gao. W: Zhang. X: Niu. Y: Meng. T: Feng. B: Duan. H: Ye. M: Dai. Y: Tia. Z: Zheng.
Y. (2014). Effects of formaldehyde on lymphocyte subsets and cytokines in the peripheral
blood of exposed workers. PLoS ONE 9: el04069.
http://dx.doi.org/10.1371/journal.pone.0104069
Tiang. S: Yu. L: Cheng. 1: Leng. S: Dai. Y: Zhang. Y: Niu. Y: Yan. H: Ou. W: Zhang. C: Zhang. K: Yang. R:
Zhou. L: Zheng. Y. (2010). Genomic damages in peripheral blood lymphocytes and
association with polymorphisms of three glutathione S-transferases in workers exposed to
formaldehyde. MutatRes 695: 9-15. http://dx.doi.Org/10.1016/j.mrgentox.2009.09.011
This document is a draft for review purposes only and does not constitute Agency policy.
R-23 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Tohn. EM: Savitz. DA: Shy. CM. (1994). Spontaneous abortions among cosmetologists. Epidemiology
5: 147-155. http://dx.doi.Org/10.1097/00001648-199403000-00004
Tuchau. MR: Lee. OP: Fantel. AG. (1992). Xenobiotic biotransformation/bioactivation in
organogenesis-stage conceptual tissues: implications for embryotoxicity and teratogenesis
[Review], DrugMetab Rev 24: 195-238. http://dx.doi.org/10.3109/03602539208996293
Tung. W: Kim. E: Lee. E: Yun. H: Tu. H: leong. M: Hwang. K: Sul. D: Kang. H. (2007). Formaldehyde
exposure induces airway inflammation by increasing eosinophil infiltrations through the
regulation of reactive oxygen species production. Environ Toxicol Pharmacol 24: 174-182.
http://dx.doi.Org/10.1016/i.etap.2007.05.001
Turvelin. 1: Vartiainen. M: Tantunen. M: Pasanen. P. (2001). Personal exposure levels and
microenvironmental concentrations of formaldehyde and acetaldehyde in the Helsinki
metropolitan area, Finland. J Air Waste Manag Assoc 51: 17-24.
Kamata. E: Nakadate. M: Uchida. 0: Ogawa. Y: Suzuki. S: Kaneko. T: Saito. M: Kurokawa. Y. (1997).
Results of a 28-month chronic inhalation toxicity study of formaldehyde in male Fisher-344
rats. J Toxicol Sci 22: 239-254.
Kane. LE: Alarie. T. (1977). Sensory irritation to formaldehyde during single and repeated
exposures in mice [Abstract], Toxicol Appl Pharmacol 41: 180-181.
Kaplan. S. (2012). [Email to Andrew Kraft regarding formaldehyde neurodevelopmental studies].
Available online
Kaplan. S. (2014). [Email with Andrew Kraft regarding follow-up on Sarsilmaz et al., 2007 and Asian
et al., 2006], Available online
Kar. S: Krishnan. A: Shivkumar. PV. (2012). Pregnancy and skin. Journal of Obstetrics and
Gynaecology of India268-275. http://dx.doi.org/10.1007/sl3224-012-0179-z
Karamouzis. MY: Konstantinopoulos. PA: Papavassiliou. AG. (2007). The activator protein-1
transcription factor in respiratory epithelium carcinogenesis [Review], Mol Cancer Res 5:
109-120. http://dx.d0i.0rg/l 0.1158/1541 -7786.MCR-06-0311
Kashiba. H: Ueda. Y: Senba. E. (1997). Systemic capsaicin in the adult rat differentially affects gene
expression for neuropeptides and neurotrophin receptors in primary sensory neurons.
Neuroscience 76: 299-312.
Katsnelson. BA: Degtyareva. TP: Privalova. LI: Minigaliveva. IA: Slvshkina. TV: Rvzhov. W:
Beresneva. OY. u. (2013). Attenuation of subchronic formaldehyde inhalation toxicity with
oral administration of glutamate, glycine and methionine. Toxicol Lett 220: 181-186.
http: / /dx. doi. or g/10.1016 /i. toxlet. 2013.04.024
Keller. DA: Heck. H: Randall. HW: Morgan. KT. (1990). Histochemical localization of formaldehyde
dehydrogenase in the rat Toxicol Appl Pharmacol 106: 311-326.
http://dx.doi.Org/10.1016/0041-008xf90190250-x
Kepler. GM: Richardson. RB: Morgan. KT: Kimbell. IS. (1998). Computer simulation of inspiratory
nasal airflow and inhaled gas uptake in a rhesus monkey. Toxicol Appl Pharmacol 150: 1-11.
http://dx.doi.org/10.1006/taap.1997.835Q
Kerns. WD: Pavkov. KL: Donofrio. DT: Gralla. ET: Swenberg. TA. (1983). Carcinogenicity of
formaldehyde in rats and mice after long-term inhalation exposure. Cancer Res 43: 4382-
4392.
Keshava. C: Davis. TA: Stanek. 1: Thayer. KA: Galizia. A: Keshava. N: Gift. 1: Vulimiri. SV. Woodall. G.:
Gigot. C: Garcia. K: Greenhalgh. A: Schulz. B: Volkoff. S: Camargo. K: Persad. AS. (2020).
Application of systematic evidence mapping to assess the impact of new research when
updating health reference values: A case example using acrolein. Environ Int 143: 105956.
http://dx.doi.Org/10.1016/i.envint2020.105956
Khaliq. F: Tripathi. P. (2009). Acute effects of formalin on pulmonary functions in gross anatomy
laboratory. Indian J Physiol Pharmacol 53: 93-96.
This document is a draft for review purposes only and does not constitute Agency policy.
R-24 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Khamgaonkar. MB: Fulare. MB. (1991). Pulmonary effects of formaldehyde exposure--an
environmental-epidemiological study. Indian J Chest Dis Allied Sci 33: 9-13.
Kilburn. KH: Warshaw. R: Thornton. TC. (1987). Formaldehyde impairs memory, equilibrium, and
dexterity in histology technicians: Effects which persist for days after exposure. Arch
Environ Occup Health 42: 117-120. http://dx.doi.org/10.1080/00039896.1987.9935806
Kilburn. KH: Warshaw. R: Thornton. TC: Husmark. I. (1989). An examination of factors that could
affect choice reaction time in histology technicians. Am J Ind Med 15: 679-686.
http://dx.doi.Org/10.1002/aiim.4700150607
Kilburn. KH: Warshaw. RH. (1992). Neurobehavioral effects of formaldehyde and solvents on
histology technicians: Repeated testing across time. Environ Res 58: 134-146.
http://dx.doi.org/10.1016/S0013-935ir05180210-5
Kim. CW: Song. IS: Ahn. YS: Park. SH: Park. TW: Noh. TH: Hong. CS. (2001). Occupational asthma due
to formaldehyde. Yonsei Med J 42: 440-445. http://dx.doi.Org/10.3349/ymi.2001.42.4.440
Kim. EM: Lee. HY: Lee. EH: Lee. KM: Park. M: Ti. KY: Tang. TH: Teong. YH: Lee. KH: Yoon. IT: Kim. SM:
Teong. MI: Kim. KD: Kang. HS. (2013a). Formaldehyde exposure impairs the function and
differentiation of NK cells. Toxicol Lett 223: 154-161.
http://dx.doi.Org/10.1016/i.toxlet.2013.09.008
Kim. H: Kim. YD: Cho. SH. (1999). Formaldehyde exposure levels and serum antibodies to
formaldehyde-human serum albumin of Korean medical students. Arch Environ Health 54:
115-118. http://dx.doi.Org/10.1080/00039899909602245
Kim. TL: Elfman. L: Wieslander. G: Ferm. M: Toren. K: Norback. D. (2011). Respiratory health among
Korean pupils in relation to home, school and outdoor environment. J Korean Med Sci 26:
166-173. http://dx.doi.Org/10.3346/ikms.2011.26.2.166
Kim. TY: Teong. MS: Park. KY: Seo. ST. (2013b). Aggravation of atopic dermatitis-like symptoms by
consecutive low concentration of formaldehyde exposure in NC/Nga mice [Letter], Exp
Dermatol 22: 219-221. http://dx.doi.org/10.llll/exd.12092
Kim. Y: Tekarl. DW: Kim. I: Kwon. A: Choi. H: Lee. S: Kim. YT: Kim. HI: Kim. Y: Oh. IH: Kim. M. (2015).
Genetic and epigenetic alterations of bone marrow stromal cells in myelodysplastic
syndrome and acute myeloid leukemia patients. Stem Cell Research 14: 177-184.
http://dx.doi.Org/10.1016/j.scr.2015.01.004
Kimbell. IS: Godo. MN: Gross. EA: Tovner. PR: Richardson. RB: Morgan. KT. (1997a). Computer
simulation of inspiratory airflow in all regions of the F344 rat nasal passages. Toxicol Appl
Pharmacol 145: 388-398. http://dx.doi.org/10.1006/taap.1997.82Q6
Kimbell. IS: Gross. EA: Tovner. PR: Godo. MN: Morgan. KT. (1993). Application of computational fluid
dynamics to regional dosimetry of inhaled chemicals in the upper respiratory tract of the
rat. Toxicol Appl Pharmacol 121: 253-263. http://dx.doi.org/10.10Q6/taap.1993.1152
Kimbell. IS: Gross. EA: Richardson. RB: Conollv. RB: Morgan. KT. (1997b). Correlation of regional
formaldehyde flux predictions with the distribution of formaldehyde-induced squamous
metaplasia in F344 rat nasal passages. MutatRes 380: 143-154.
http://dx.doi. org/10.1016/S0027-5107C97100132-2
Kimbell. IS: Overton. TH: Subramaniam. RP: Schlosser. PM: Morgan. KT: Conollv. RB: Miller. FT.
(2001a). Posimetry modeling of inhaled formaldehyde: Binning nasal flux predictions for
quantitative risk assessment. Toxicol Sci 64: 111-121.
Kimbell. IS: Subramaniam. RP. (2001). Use of computational fluid dynamics models for dosimetry of
inhaled gases in the nasal passages [Review], Inhal Toxicol 13: 325-334.
http://dx.doi.Org/10.1080/08958370120442
Kimbell. IS: Subramaniam. RP: Gross. EA: Schlosser. PM: Morgan. KT. (2001b). Posimetry modeling
of inhaled formaldehyde: comparisons of local flux predictions in the rat, monkey, and
human nasal passages. Toxicol Sci 64: 100-110.
This document is a draft for review purposes only and does not constitute Agency policy.
R-25 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Kimura. R: Kimoto. I: Takeda. M: Mivake. M: Sakamoto. T. (2010). Alteration in airway
microvascular leakage induced by sensorineural stimulation in rats exposed to inhaled
formaldehyde. Toxicol Lett 199: 254-260. http://dx.doi.Org/10.1016/i.toxlet.2010.09.007
Kirsch-Volders. M: Bonassi. S: Knasmueller. S: Holland. N: Bolognesi. C: Fenech. MF. (2014).
Commentary: Critical questions, misconceptions and a road map for improving the use of
the lymphocyte cytokinesis-block micronucleus assay for in vivo biomonitoring of human
exposure to ge no toxic chemicals-A HUMN project perspective. Mutat Res Rev Mutat Res
759: 49-58. http://dx.doi.Org/10.1016/i.mrrev.2013.12.001
Kitaev. EM: Savchenko. ON: Lovchikov. VA: Altukhov. W: Vishnvakov. YS. (1984). Razvitie
zarodyshey i nekotorye pokazateli reproductivnoy funktsii u krys posle ingalyatsionnogo
vozdeystviya formal'degida do oplodotvoreniya [Akush Ginekol 10: 49-52.
Kleinniienhuis. AT: Staal. YC: Duistermaat. E: Engel. R: Woutersen. RA. (2013). The determination of
exogenous formaldehyde in blood of rats during and after inhalation exposure. Food Chem
Toxicol 52: 105-112. http://dx.doi.org/10.1016/i.fct.2012.11.008
Kligerman. AD: Phelps. MC: Erexson. GL. (1984). Cytogenetic analysis of lymphocytes from rats
following formaldehyde inhalation. Toxicol Lett 21: 241-246.
http: //dx.doi.org/10.1016/0378-4274f84190079-l
Klonisch. T: Fowler. PA: Hombach-Klonisch. S. (2004). Molecular and genetic regulation of testis
descent and external genitalia development [Review], Dev Biol 270: 1-18.
http://dx.doi.Org/10.1016/i.ydbio.2004.02.018
Kominskv. TR: Stroman. RE. (1977). Health Hazard Evaluation Determination, Report No. HHE-77-
15-421, Leed and Northrup Corporation, Expendable Devices Division Ellwood City,
Pennsylvania (pp. 77-15). (NIOSH/00074930). Kominsky, JR; Stroman, RE.
https://www.cdc.gov/niosh/hhe/reports/pdfs/77-15-421.pdf
Kopf. M: Le Gros. G: Bachmann. M: Lamers. MC: Bluethmann. H: Kohler. G. (1993). Disruption of the
murine IL-4 gene blocks Th2 cytokine responses. Nature 362: 245-248.
http://dx.doi.org/10.1038/362245a0
Korhonen. K: Liukkonen. T: Ahrens. W: Astrakianakis. G: Boffetta. P: Burdorf. A: Heederik. D:
Kauppinen. T: Kogevinas. M: Osvoll. P: Rix. BA: Saalo. A: Sunver. 1: Szadkowska-Stanczvk. I:
Teschke. K: Westberg. H: Widerkiewicz. K. (2004). Occupational exposure to chemical
agents in the paper industry. Int Arch Occup Environ Health 77: 451-460.
http: //dx.doi.org/10.1007/s00420-004-0530-5
Krakowiak. A: Gorski. P: Pazdrak. K: Ruta. U. (1998). Airway response to formaldehyde inhalation in
asthmatic subjects with suspected respiratory formaldehyde sensitization. Am J Ind Med 33:
274-281. http://dx.doi.org/10.1002/fSICIH097-0274f 199803133:3<274::AID-
AJIM9>3.0.CO;2-W
Krewski. D: Crump. KS: Farmer. 1: Gavlor. DW: Howe. R: Portier. C: Salsburg. D: Sielken. RL: Van
Rvzin. 1. (1983). A comparison of statistical methods for low dose extrapolation utilizing
time-to-tumour data. Fundam Appl Toxicol 3: 140-160. http://dx.doi.org/10.1016/SQ272-
0590C83180075-X
Kriebel. D: Myers. D: Cheng. M: Woskie. S: Cocanour. B. (2001). Short-term effects of formaldehyde
on peak expiratory flow and irritant symptoms. Arch Environ Health 56: 11-18.
http: //dx.doi.org/10.1080/00039890109604049
Kriebel. D: Sama. SR: Cocanour. B. (1993). Reversible pulmonary responses to formaldehyde. A
study of clinical anatomy students. Am Rev Respir Dis 148: 1509-1515.
http://dx.doi.Org/10.1164/airccm/148.6 Pt 1.1509
Krzvzanowski. M: Ouackenboss. 11: Lebowitz. MP. (1990). Chronic respiratory effects of indoor
formaldehyde exposure. Environ Res 52: 117-125. http://dx.doi.org/10.1016/S0Q13-
9351(05180247-6
This document is a draft for review purposes only and does not constitute Agency policy.
R-26 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Ku. RH: Billings. RE. (1984). Relationships between formaldehyde metabolism and toxicity and
glutathione concentrations in isolated rathepatocytes. Chem Biol Interact 51: 25-36.
http://dx.doi. org/10.1016/0009-2797fB4190017-6
Kuehner. S: Holzmann. K: Speit. G. (2013). Characterization of formaldehyde's genotoxic mode of
action by gene expression analysis in TK6 cells. Arch Toxicol 87: 1999-2012.
http://dx.doi. org/10.1007/s00204-013-1060-2
Kuehner. S: Schlaier. M: Schwarz. K: Speit. G. (2012). Analysis of leukemia-specific aneuploidies in
cultured myeloid progenitor cells in the absence and presence of formaldehyde exposure.
Toxicol Sci 128: 72-78. http://dx.doi.org/10.1093/toxsci/kfsl26
Kulle. TT. (1993). Acute odor and irritation response in healthy nonsmokers with formaldehyde
exposure. Inhal Toxicol 5: 323-332. http://dx.doi.org/10.3109/08958379308998389
Kulle. TT: Cooper. GP. (1975). Effects of formaldehyde and ozone on the trigeminal nasal sensory
system. Arch Environ Occup Health 30: 237-243.
Kulle. TT: Sauder. LR: Hebel. TR: Green. DT: Chatham. MP. (1987). Formaldehyde dose-response in
healthy nonsmokers. J Air Pollut Control Assoc 37: 919-924.
http://dx.doi.Org/10.1080/08940630.1987.10466285
Kum. C: Kiral. F: Sekkin. S: Sevrek. K: Bovacioglu. M. (2007). Effects of xylene and formaldehyde
inhalations on oxidative stress in adult and developing rats livers. Exp Anim 56: 35-42.
http://dx.doi.org/10.1538/expanim.56.35
Kumar. H: Tain. R: Douglas. RG: Tawhai. MH. (2016). Airflow in the human nasal passage and sinuses
of chronic rhinosinusitis subjects. PLoS ONE 11: e0156379.
http://dx.doi.org/10.1371/iournal.pone.0156379
Kumari. A: Owen. N: Tuarez. E: Mccullough. AK. (2015). BLM protein mitigates formaldehyde-
induced genomic instability. DNA Repair 28: 73-82.
http://dx.doi.Org/10.1016/i.dnarep.2015.02.010
Kunkler. PE: Ballard. CI: Oxford. GS: Hurley. TH. (2011). TRPA1 receptors mediate environmental
irritant-induced meningeal vasodilatation. Pain 152: 38-44.
http://dx.doi.Org/10.1016/i.pain.2010.08.021
Kuo. HW: Tian. GT: Chen. CL: Liu. CS: Lai. IS. (1997). White blood cell count as an indicator of
formaldehyde exposure. Bull Environ Contam Toxicol 59: 261-267.
http://dx.doi.Org/10.1007/s001289900473
Kuper. CF: van Oostrum. L: Ma-Hock. L: Durrer. S: Woutersen. RA. f2011I Hyperplasia of the
lymphoepithelium of NALT in rats but not in mice upon 28-day exposure to 15 ppm
formaldehyde vapor. Exp Toxicol Pathol 63: 25-32.
http://dx.doi.Org/10.1016/i.etp.2009.09.004
Labreche. F: Forest. 1: Trottier. M: Lalonde. M: Simard. R. (2003). Characterization of chemical
exposures in hairdressing salons. Appl Occup Environ Hyg 18: 1014-1021.
http://dx.doi.Org/10.1080/10473220390244667
Ladeira. C: Viegas. S: Carolino. E: Gomes. MC: Brito. M. (2013). The influence of genetic
polymorphisms in XRCC3 and ADH5 genes on the frequency of genotoxicity biomarkers in
workers exposed to formaldehyde. Environ Mol Mutagen 54: 213-221.
http://dx.doi.org/10.1002/em.21755
Ladeira. C: Viegas. S: Carolino. E: Prista. 1: Gomes. MC: Brito. M. (2011). Genotoxicity biomarkers in
occupational exposure to formaldehyde-the case of histopathology laboratories. MutatRes
721: 15-20. http://dx.doi.Org/10.1016/j.mrgentox.2010.ll.015
Laforest. L: Luce. D: Goldberg. P: Begin. D: Gerin. M: Demers. PA: Brugere. 1: Leclerc. A. (2000).
Laryngeal and hypopharyngeal cancers and occupational exposure to formaldehyde and
various dusts: A case-control study in France. Occup Environ Med 57: 767-773.
http://dx.doi.org/10.1136/oem.57.ll.767
This document is a draft for review purposes only and does not constitute Agency policy.
R-27 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Lai. Y: Yu. R: Hartwell. HI: Moeller. BC: Bodnar. WM: Swenberg. TA. (2016). Measurement of
Endogenous versus Exogenous Formaldehyde-Induced DNA-Protein Crosslinks in Animal
Tissues by Stable Isotope Labeling and Ultrasensitive Mass Spectrometry. Cancer Res 76:
2652-2661. http://dx.doi.org/10.n58/0008-5472.CAN-15-2527
Laioie. P: Aubin. D: Gingras. V: Daigneault. P: Ducharme. F: Gauvin. FD: Fugler. D: Leclerc. TM: Won.
D: Won. D: Courteau. M: Gingras. S: Heroux. ME: Yang. W: Schleibinger. H. (2014). The
IVAIRE Project - A Randomized Controlled Study of the Impact of Ventilation on Indoor Air
Quality and the Respiratory Symptoms of Asthmatic Children in Single Family Homes.
Indoor Air 25: 582-597. http://dx.doi.org/10.llll/ina.12181
Lam. CW: Casanova. M: Heck. H. (1985). Depletion of nasal mucosal glutathione by acrolein and
enhancement of formaldehyde-induced DNA-protein cross-linking by simultaneous
exposure to acrolein. Arch Toxicol 58: 67-71. http://dx.doi.org/10.1007/bf00348311
Lan. 0: Smith. MT: Tang. X: Guo. W: Vermeulen. R: Ti. Z: Hu. W: Hubbard. AE: Min. S: Mchale. CM: Oiu.
C: Liu. S: Reiss. B: Beane Freeman. L: Blair. A: Ge. Y: Xiong. I: Li. L: Rappaport. SM: Huang. H:
Rothman. N: Zhang. L. (2015). Chromosome-wide aneuploidy study (CWAS) of cultured
circulating myeloid progenitor cells from workers occupationally exposed to formaldehyde.
Carcinogenesis 36: 160-167. http://dx.doi.org/10.1093/carcin/bgu229
Lang. I: Bruckner. T: Triebig. G. (2008). Formaldehyde and chemosensory irritation in humans: A
controlled human exposure study. Regul Toxicol Pharmacol 50: 23-36.
http://dx.doi.Org/10.1016/i.yrtph.2007.08.012
Langevin. F: Crossan. GP: Rosado. IV: Arends. MT: Patel. KT. (2011). Fancd2 counteracts the toxic
effects of naturally produced aldehydes in mice. Nature 475: 53-58.
http://dx.doi.org/10.1038/naturel0192
Larsen. ST: Wolkoff. P: Hammer. M: Kofoed-S0rensen. V: Clausen. PA: Nielsen. GD. (2013). Acute
airway effects of airborne formaldehyde in sensitized and non-sensitized mice housed in a
dry or humid environment Toxicol Appl Pharmacol 268: 294-299.
http://dx.doi.Org/10.1016/i.taap.2013.02.006
Lazenbv. V: Hinwood. A: Callan. A: Franklin. P. (2012). Formaldehyde personal exposure
measurements and time weighted exposure estimates in children. Chemosphere 88: 966-
973. http://dx.doi.Org/10.1016/j.chemosphere.2012.03.029
Lebowitz. MP: Krzvzanowski. M: Ouackenboss. II: Orourke. MK. (1997). Diurnal variation of PEF
and its use in epidemiological studies. Eur Respir J10: S49-S56.
Leclerc. A: Luce. D: Demers. PA: Boffetta. P: Kogevinas. M: Belli. S: Bolm-Audorff. U: Brinton. LA:
Colin. D: Comba. P: Gerin. M: Hardell. L: Hayes. RB: Magnani. C: Merler. E: Morcet. IF:
Preston-Martin. S: Vaughan. TL: Zheng. W. (1997). Sinonasal cancer and occupation. Results
from the reanalysis of twelve case-control studies. Am J Ind Med 31: 153-165.
http://dx.doi. org/10.1002 /fsicill 097-0274C199702131:2<153::aid-aiim4>3.0.co:2-0
Leclerc. A: Martinez Cortes. M: Gerin. M: Luce. D: Brugere. I. (1994). Sinonasal cancer and wood dust
exposure: Results from a case-control study. Am J Epidemiol 140: 340-349.
http://dx.doi.org/10.1093/oxfordiournals.aie.all7256
Lee. HK: Alarie. Y: Karol. MH. (1984). Induction of formaldehyde sensitivity in guinea pigs. Toxicol
Appl Pharmacol 75: 147-155. http://dx.doi.org/10.1016/0041-008Xr84190085-l
Lee. IT: Ko. CY. (2005). Has survival improved for nasopharyngeal carcinoma in the United States?
Otolaryngol Head Neck Surg 132: 303-308. http://dx.doi.Org/10.1016/i.otohns.2004.09.018
Lee. PN: Fry. IS. (2010). Systematic review of the evidence relating FEV1 decline to giving up
smoking [Review], BMC Med 8: 84. http://dx.doi.org/10.1186/1741-7Q15-8-84
Leikauf. GD. (1992). Mechanisms of aldehyde-induced bronchial reactivity: role of airway
epithelium. Res Rep Health Eff Instl-35.
Leng. 1: Liu. CW: Hartwell. HI: Yu. R: Lai. Y: Bodnar. WM: Lu. K: Swenberg. TA. (2019). Evaluation of
inhaled low-dose formaldehyde-induced DNA adducts and DNA-protein cross-links by
This document is a draft for review purposes only and does not constitute Agency policy.
R-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
liquid chromatography-tandem mass spectrometry. Arch Toxicol 93: 763-773.
http://dx.doi.Org/10.1007/s00204-019-02393-x
Lessard. M: Helias. C: Struski. S: Perrusson. N: Uettwiller. F: Mozziconacci. Ml: Lafage-Pochitaloff. M:
Dastugue. N: Terre. C: Brizard. F: Cornillet-Lefebvre. P: Mugneret. F: Barin. C: Herrv. A:
Luquet. I: Desangles. F: Michaux. L: Verellen-Dumoulin. C: Perrot. C: Van Den Akker. 1:
Lespinasse. 1: Eclache. V: Berger. R: Hematologique. GFdC. (2007). Fluorescence in situ
hybridization analysis of 110 hematopoietic disorders with chromosome 5 abnormalities:
Do de novo and therapy-related myelodysplastic syndrome-acute myeloid leukemia actually
differ? Cancer Genet Cytogenet 176: 1-21.
http://dx.doi.Org/10.1016/i.cancergencyto.2007.01.013
Levesque. TP: Winkler. IG: Larsen. SR: Rasko. IE. (2007). Mobilization of bone marrow-derived
progenitors [Review], Handb Exp Pharmacol3-36. http://dx.doi.org/10.1007/978-3-540-
68976-8 1
Levine. RT: Andielkovich. DA: Shaw. LK. (1984a). The mortality of Ontario (Canada) undertakers and
a review of a formaldehyde-related mortality studies. J Occup Med 26: 740-746.
Levine. RT: Dalcorso. RD: Blunden. PB: Battigelli. MC. (1984b). The effects of occupational exposure
on the respiratory health of West Virginia morticians. J Occup Med 26: 91-98.
Li. AM: Fung. CK: Yu. IT: Goggins. WB: Chan. GY: Chan. CK: Lau. AP: Leung. 10. (2019). Associations
of wheeze during the first 18 months of life with indoor nitrogen dioxide, formaldehyde,
and family history of asthma: a prospective cohort study. Hong Kong Med J 25 Suppl 3: 20-
23.
Li. B: Huang. G: Zhang. X: Li. R: Wang. T: Dong. Z: He. Z. (2013a). Increased phosphorylation of
histone H3 at serine 10 is involved in Epstein-Barr virus latent membrane protein-1-
induced carcinogenesis of nasopharyngeal carcinoma. BMC Cancer 13: 124.
http://dx.doi. org/10.1186/1471-2407-13-124
Li. 0: Mei. 0: Huvan. T: Xie. L: Che. S: Yang. H: Zhang. M: Huang. 0. (2013b). Effects of formaldehyde
exposure on human NK cells in vitro. Environ Toxicol Pharmacol 36: 948-955.
http://dx.doi.Org/10.1016/i.etap.2013.08.005
Li. T: Chen. IX: Fu. XP: Yang. S: Zhang. Z: Chen. KH: Li. Y. (2011). microRNA expression profiling of
nasopharyngeal carcinoma. Oncol Rep 25: 1353-1363.
http://dx.doi.org/10.3892/or.2011.1204
Li. W: Ray. RM: Gao. PL: Fitzgibbons. ED: Seixas. NS: Camp. IE: Wernli. KT: Astrakianakis. G: Feng. Z:
Thomas. DB: Checkowav. H. (2006). Occupational risk factors for nasopharyngeal cancer
among female textile workers in Shanghai, China. Occup Environ Med 63: 39-44.
http://dx.doi.org/10.1136/oem.2005.0217Q9
Li. Y: Song. Z: Ding. Y: Xin. Y. e: Wu. T: Su. T. ao: He. R: Tai. F: Lian. Z. (2016). Effects of formaldehyde
exposure on anxiety-like and depression-like behavior, cognition, central levels of
glucocorticoid receptor and tyrosine hydroxylase in mice. Chemosphere 144: 2004-2012.
http://dx.doi.Org/10.1016/i.chemosphere.2015.10.102
Li. Z: Chen. L: Oin. Z. (2009). Paradoxical roles of IL-4 in tumor immunity [Review], Cell Mol
Immunol 6: 415-422. http://dx.doi.org/10.1038/cmi.2009.53
Liao. S: Tiang. L: Zhang. X. (2010). [Effects of inhaled formaldehyde on learning and memory and
expression of CaMK II in hippocampus of Wistar rats of different ages], 35: 342-344.
LICM (Lithium Ion Cell Manufacturers' Coalition). (2008). Effect of inhaled formaldehyde on
learning and memory of mice. Indoor Air 18: 77-83. http://dx.doi.Org/10.l 111/i.1600-
0668.2008.00524.x
Lim. SK: Choi. H: Park. MI: Kim. PI: Kim. TC: Kim. GY: Teong. SY: Rodionov. RN: Han. HI: Yoon. KC:
Park. SH. (2013). The ER stress-mediated decrease in PPAH1 expression is involved in
formaldehyde-induced apoptosis in lung epithelial cells. Food Chem Toxicol 62: 763-769.
http://dx.doi.Org/10.1016/i.fct2013.10.014
This document is a draft for review purposes only and does not constitute Agency policy.
R-29 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Lin. D: Guo. Y: Yi. T: Kuang. D. an: Li. X: Deng. H: Huang. K. un: Guan. L. ei: He. Y: Zhang. X: Hu. D. ie:
Zhang. Z: Zheng. H: Zhang. X: Mchale. CM: Zhang. L: Wu. T. (2013). Occupational exposure to
formaldehyde and genetic damage in the peripheral blood lymphocytes of plywood
workers. J Occup Health 55: 284-291. http://dx.doi.org/10.1539/joh.12-0288-OA
Lindahl. T. (1993). INSTABILITY AND DECAY OF THE PRIMARY STRUCTURE OF DNA. Nature 362:
709-715.
Lindbohm. ML: Hemminki. K: Bonhomme. MG: Anttila. A: Rantala. K: Heikkila. P: Rosenberg. MI.
(1991). Effects of paternal occupational exposure on spontaneous abortions. Am J Public
Health 81: 1029-1033. http: //dx.doi.org/10.2105 /aiph.81.8.1029
Lippman. SM: Peters. EI: Wargovich. Ml: Stadnvk. AN: Dixon. DO: Dekmezian. RH: Loewv. TW:
Morice. RC: Cunningham. IE: Hong. WK. (1990). Bronchial micronuclei as a marker of an
early stage of carcinogenesis in the human tracheobronchial epithelium. Int J Cancer 45:
811-815. http://dx.doi.Org/10.1002/iic.2910450503
Liteplo. RG: Meek. ME. (2003). Inhaled formaldehyde: Exposure estimation, hazard
characterization, and exposure-response analysis [Review], J Toxicol Environ Health B Crit
Rev 6: 85-114. http: //dx.doi.org/10.1080/1093740030648Q
Liu. D: Zheng. Y: Li. B: Yao. H: Li. R: Zhang. Y: Yang. X. (2011a). Adjuvant effects of gaseous
formaldehyde on the hyper-responsiveness and inflammation in a mouse asthma model
immunized by ovalbumin. J Immunotoxicol 8: 305-314.
http://dx.doi.org/10.3109/1547691X.2011.600738
Liu. KS: Huang. FY: Havward. SB: Wesolowski. I: Sexton. K. (1991). Irritant effects of formaldehyde
exposure in mobile homes. Environ Health Perspect 94: 91-94.
http://dx.doi.org/10.2307/3431298
Liu. 0: Yang. L: Gong. C: Tao. G: Huang. H: Liu. 1: Zhang. H: Wu. D: Xia. B: Hu. G: Wang. K: Zhuang. Z.
(2011b). Effects of long-term low-dose formaldehyde exposure on global genomic
hypomethylation in 16HBE cells. Toxicol Lett 205: 235-240.
http://dx.doi.Org/10.1016/i.toxlet.2011.05.1039
Liu. OB: Wang. W: Ting. W. (2018). Indoor air pollution aggravates asthma in Chinese children and
induces the changes in serum level of miR-155. Int J Environ Health Res 29: 1-9.
http://dx.doi.org/10.1080/09603123.2018.1506569
Liu. W: Zhang. 1: Hashim. TH: lalaludin. I: Hashim. Z: Goldstein. BP. (2003). Mosquito coil emissions
and health implications. Environ Health Perspect 111: 1454-1460.
http://dx.doi.org/10.1289/ehp.6286
Liu. Y: Yu. D: Xiao. S. (2017). Effects of chronic exposure to Formaldehyde on micronucleus rate of
bone marrow cells in male mice. J Pak Med Assoc 67: 933-935.
Llorente. TL: Lopez. F: Suarez. C: Hermsen. MA. (2014). Sinonasal carcinoma: clinical, pathological,
genetic and therapeutic advances [Review], Nat Rev Clin Oncol 11: 460-472.
http://dx.doi.org/10.1038/nrclinonc.2Q14.97
Lofstedt. H: Westberg. H: Selden. AI: Brvngelsson. IL: Svartengren. M. (2011a). Respiratory
symptoms and lung function in foundry workers using the hot box method: A 4-year follow-
up. J Occup Environ Med 53: 1425-1429.
http://dx.doi.org/10.1097/TOM.0b013e3182363cl7
Lofstedt. H: Westberg. H: Selden. AI: Lundholm. C: Svartengren. M. (2009). Respiratory symptoms
and lung function in foundry workers exposed to low molecular weight isocyanates. Am J
Ind Med 52: 455-463. http://dx.doi.org/10.1002/ajim.20693
Lofstedt. H: Westberg. H: Selden. AI: Rudblad. S: Brvngelsson. IL: Ngo. Y: Svartengren. M. (2011b).
Nasal and ocular effects in foundry workers using the hot box method. J Occup Environ Med
53: 43-48. http://dx.doi.org/10.1097/TQM.ObOl3e318181 ff05cc
This document is a draft for review purposes only and does not constitute Agency policy.
R-30 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Lourenco. 0: Fonseca. AM: Taborda-Barata. L. (2016). Human CD8+ T Cells in Asthma: Possible
Pathways and Roles for NK-Like Subtypes [Review], 7: 638.
http://dx.doi.org/10.3389/fimmu.2016.0Q638
Lu. K: Bovsen. G: Gao. L: Collins. LB: Swenberg. TA. (2008). Formaldehyde-induced histone
modifications in vitro. Chem Res Toxicol 21: 1586-1593.
http ://dx. doi. or g/10.10 21 /tx80 0 0576
Lu. K: Collins. LB: Ru. H: Bermudez. E: Swenberg. TA. (2010a). Distribution of DNA adducts caused
by inhaled formaldehyde is consistent with induction of nasal carcinoma but not leukemia.
Toxicol Sci 116: 441-451. http://dx.doi.org/10.1093/toxsci/kfq061
Lu. K: Craft. S: Nakamura. 1: Moeller. BC: Swenberg. TA. (2012). Use of LC-MS/MS and stable isotopes
to differentiate hydroxymethyl and methyl DNA adducts from formaldehyde and
nitrosodimethylamine. Chem Res Toxicol 25: 664-675.
http ://dx. doi. or g/10.10 21 /tx2 0 042 6b
Lu. K: Moeller. B: Dovle-Eisele. M: Mcdonald. 1: Swenberg. TA. f2011I Molecular dosimetry of N2-
hydroxymethyl-dG DNA adducts in rats exposed to formaldehyde. Chem Res Toxicol 24:
159-161. http://dx.doi.org/10.1021 /txl003886
Lu. K: Ye. W: Zhou. L: Collins. LB: Chen. X: Gold. A: Ball. LM: Swenberg. TA. (2010b). Structural
characterization of formaldehyde-induced cross-links between amino acids and
deoxynucleosides and their oligomers. J Am Chem Soc 132: 3388-3399.
http://dx.doi.org/10.1021/ia908282f
Lu. Z: Li. CM: Qiao. Y: Liu. Y: Yan. Y: Yang. X. (2005). Type II vanilloid receptor signaling system: One
of the possible mechanisms for the rise in asthma cases. Front Biosci 10: 2527-2533.
http://dx.doi.org/10.2741/1717
Luce. D: Gerin. M: Leclerc. A: Morcet. IF: Brugere. 1: Goldberg. M. (1993). Sinonasal cancer and
occupational exposure to formaldehyde and other substances. Int J Cancer 53: 224-231.
http://dx.doi.Org/10.1002/ijc.2910530209
Luce. D: Leclerc. A: Begin. D: Demers. PA: Gerin. M: Orlowski. E: Kogevinas. M: Belli. S: Bugel. I:
Bolm-Audorff. U: Brinton. LA: Comba. P: Hardell. L: Hayes. RB: Magnani. C: Merler. E:
Preston-Martin. S: Vaughan. TL: Zheng. W: Boffetta. P. (2002). Sinonasal cancer and
occupational exposures: a pooled analysis of 12 case-control studies. Cancer Causes Control
13: 147-157. http://dx.doi.Org/10.1023/A:1014350004255
Luce. D: Leclerc. A: Morcet. IF: Casal-Lareo. A: Gerin. M: Brugere. I: Haguenoer. TM: Goldberg. M.
(1992). Occupational risk factors for sinonasal cancer: A case-control study in France. Am J
Ind Med 21: 163-175. http: //dx.doi.org/10.1002 /aiim.47002102Q6
Luch. A: Frev. FC: Meier. R: Fei. 1: Naegeli. H. (2014). Low-dose formaldehyde delays DNA damage
recognition and DNA excision repair in human cells. PLoS ONE 9: e94149.
http://dx.doi.org/10.1371/iournal.pone.0094149
Lundberg. TM: Saria. A. (1983). Capsaicin-induced desensitization of airway mucosa to cigarette
smoke, mechanical and chemical irritants [Letter], Nature 302: 251-253.
http://dx.doi.org/10.1038/3Q2251a0
Luo. YL: Guo. HM: Zhang. YL: Chen. PX: Zhu. YX: Huang. TH: Zhou. WL. (2013). Cellular mechanism
underlying formaldehyde-stimulated CI- secretion in rat airway epithelium. PLoS ONE 8:
e54494. http://dx.doi.org/10.1371/iournal.pone.0054494
Lvapina. M: Zhelezova. G: Petrova. E: Boev. M. (2004). Flow cytometric determination of neutrophil
respiratory burst activity in workers exposed to formaldehyde. Int Arch Occup Environ
Health 77: 335-340. http://dx.doi.Org/10.1007/s00420-004-0516-3
Ma. H: Song. X: Zhang. W: Ling. X: Yang. X: Wu. W: Lou. K: Xu. H. (2020). Formaldehyde inhibits
development of T lymphocytes in mice. Toxicol Environ Chem 102: 473-489.
http://dx.doi.Org/10.1080/02772248.2020.1815202
This document is a draft for review purposes only and does not constitute Agency policy.
R-31 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Ma. L: Weinberg. RA. (2008). MicroRNAs in malignant progression [Review], Cell Cycle 7: 570-572.
http://dx.doi.Org/10.4161/cc.7.5.5547
Madureira. 1: Paciencia. I: Cavaleiro-Rufo. 1: de Oliveira Fernandes. E. (2016). Indoor pollutant
exposure among children with and without asthma in Porto, Portugal, during the cold
season. Environ Sci Pollut Res Int 23: 20539-20552. http://dx.doi.org/10.1007/sll356-
016-7269-x
Magnani. C: Comba. P: Ferraris. F: Ivaldi. C: Meneghin. M: Terracini. B. (1993). A case-control study
of carcinomas of the nose and paranasal sinuses in the woolen textile manufacturing
industry. Arch Environ Health 48: 94-97.
http://dx.doi.org/10.1080/00039896.1993.9938401
Main. DM: Hogan. TT. (1983). Health effects of low level exposure to formaldehyde. J Occup Environ
Med 25: 896-900. http://dx.doi.Org/10.1097/00043764-198312000-00013
Malaka. T: Kodama. AM. (1990). Respiratory health of plywood workers occupationally exposed to
formaldehyde. Arch Environ Health 45: 288-294.
http: //dx.doi.org/10.1080 /00039896.1990.10118748
Malek. FA: Moritz. KU: Fanghanel. 1. (2003a). Formaldehyde inhalation & open field behaviour in
rats. Indian J Med Res 118: 90-96.
Malek. FA: Moritz. KU: Fanghanel. 1. (2003b). A study on specific behavioral effects of formaldehyde
in the rat JExpAnim Sci 42: 160-170. http://dx.d0i.0rg/l 0.1016/S0939-8600r03180009-3
Malek. FA: Moritz. KU: Fanghanel. 1. (2003c). A study on the effect of inhalative formaldehyde
exposure on water labyrinth test performance in rats. Ann Anat 185: 277-285.
http://dx.doi.Org/10.1016/S0940-9602r03180040-7
Malek. FA: Moritz. KU: Fanghanel. I. (2004). Effects of a single inhalative exposure to formaldehyde
on the open field behavior of mice. Int J Hyg Environ Health 207: 151-158.
http://dx.doi.org/10.1078/1438-4639-00268
Malker. HSR: Mclaughlin. IK: Weiner. TA: Silverman. DT: Blot. WT: TLE. E: Fraumeni. I. r. I. F. (1990).
Occupational risk factors for nasopharyngeal cancer in Sweden. Br J Ind Med 47: 213-214.
http://dx.doi.Org/10.1136/oem.47.3.213
Mandrvk. 1: Alwis. KU: Hocking. AD. (2000). Effects of personal exposures on pulmonary function
and work-related symptoms among sawmill workers. Ann Occup Hyg 44: 281-289.
Mantovani. A: Allavena. P: Sica. A: Balkwill. F. (2008). Cancer-related inflammation [Review], Nature
454: 436-444. http://dx.doi.org/10.1038/nature07205
Marcucci. G: Radmacher. MP: Mrozek. K: Bloomfield. CD. (2009). MicroRNA expression in acute
myeloid leukemia [Review], Curr Hematol Malig Rep 4: 83-88.
http://dx.doi.Org/10.1007/sll899-009-0012-7
Maronpot. RR: Miller. RA: Clarke. WT: Westerberg. RB: Decker. TR: Moss. OR. (1986). Toxicity of
formaldehyde vapor in B6C3F1 mice exposed for 13 weeks. Toxicology 41: 253-266.
http://dx.doi.org/10.1016/0300-483Xr86190180-0
Marsh. GM: Stone. RA: Esmen. NA: Henderson. VL. (1994). Mortality patterns among chemical plant
workers exposed to formaldehyde and other substances. J Natl Cancer Inst 86: 384-386.
http://dx.doi.Org/10.1093/jnci/86.5.384
Marsh. GM: Stone. RA: Esmen. NA: Henderson. VL: Lee. KY. (1996). Mortality among chemical
workers in a factory where formaldehyde was used. Occup Environ Med 53: 613-627.
http://dx.doi.Org/10.1136/oem.53.9.613
Marsh. GM: Youk. AO. (2005). Reevaluation of mortality risks from nasopharyngeal cancer in the
formaldehyde cohort study of the National Cancer Institute. Regul Toxicol Pharmacol 42:
275-283. http://dx.doi.Org/10.1016/i.vrtph.2005.05.003
Marsh. GM: Youk. AO: Buchanich. TM: Cassidv. LP: Lucas. LI: Esmen. NA: Gathuru. IM. (2002).
Pharyngeal cancer mortality among chemical plant workers exposed to formaldehyde.
Toxicol Ind Health 18: 257-268. http://dx.doi.org/10.1191/07482337Q2thl49oa
This document is a draft for review purposes only and does not constitute Agency policy.
R-32 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Marsh. GM: Youk. AO: Buchanich. TM: Cunningham. M: Esmen. NA: Hall. TA: Phillips. ML. (2007a).
Mortality patterns among industrial workers exposed to chloroprene and other substances:
II. Mortality in relation to exposure. Chem Biol Interact 166: 301-316.
http://dx.doi.Org/10.1016/j.cbi.2006.08.012
Marsh. GM: Youk. AO: Buchanich. TM: Erdal. S: Esmen. NA. (2007b). Work in the metal industry and
nasopharyngeal cancer mortality among formaldehyde-exposed workers. Regul Toxicol
Pharmacol 48: 308-319. http://dx.doi.Org/10.1016/i.yrtph.2007.04.006
Marsh. GM: Youk. AO: Morfeld. P. (2007c). Mis-specified and non-robust mortality risk models for
nasopharyngeal cancer in the National Cancer Institute formaldehyde worker cohort study.
Regul Toxicol Pharmacol 47: 59-67. http ://dx.doi. or g/10.1016/i.vrtph.2006.07.007
Martin. WT. (1990). A teratology study of inhaled formaldehyde in the rat Reprod Toxicol 4: 237-
239. http://dx.doi.org/10.1016/0890-6238r90190065-4
Mashaghi. A: Marmalidou. A: Tehrani. M: Grace. PM: Pothoulakis. C: Dana. R. (2016). Neuropeptide
substance P and the immune response [Review], Cell Mol Life Sci 73: 4249-4264.
http: //dx.doi.org/10.1007/s00018-016-2293-z
Massberg. S: Schaerli. P: Knezevic-Maramica. I: Kollnberger. M: Tubo. N: Moseman. EA: Huff. IV:
Tunt. T: Wagers. AT: Mazo. IB: von Andrian. UH. (2007). Immunosurveillance by
hematopoietic progenitor cells trafficking through blood, lymph, and peripheral tissues. Cell
131: 994-1008. http://dx.doi.org/10.1016/i.cell.2007.09.047
Matanoski. GM. (1989). Risks of pathologists exposed to formaldehyde (final report). (DHHS Grant
No. 5 R01-OH-01511-03). Baltimore, MD: Johns Hopkins University Department of
Epidemiology.
https://ntrl.ntis. gov/NTRL/dashboard/searchResults.xhtml?search0uery=PB91173682
Matsunaga. I: Mivake. Y: Yoshida. T: Miyamoto. S: Ohva. Y: Sasaki. S: Tanaka. K: Oda. H: Ishiko. 0:
Hirota. Y: Group. OMaCHS. (2008). Ambient formaldehyde levels and allergic disorders
among Japanese pregnant women: Baseline data from the Osaka maternal and child health
study. Ann Epidemiol 18: 78-84. http: //dx.doi.Org/10.1016/i.annepidem.2007.07.095
Matsunaga. M. (2012). [Email to Glinda Cooper regarding follow-up on formaldehyde-allergy-
asthma study]. Available online
Matsuoka. T: Takaki. A: Ohtaki. H: Shioda. S. (2010). Early changes to oxidative stress levels
following exposure to formaldehyde in ICR mice. J Toxicol Sci 35: 721-730.
http ://dx. doi. or g/10.2131 /its. 3 5.721
Mavr. SI: Hafizovic. K: Waldfahrer. F: Iro. H: Kiitting. B. (2010). Characterization of initial clinical
symptoms and risk factors for sinonasal adenocarcinomas: results of a case-control study.
Int Arch Occup Environ Health 83: 631-638. http://dx.doi.Org/10.1007/s00420-009-0479-
5
Mazzone. SB: Undem. BT. (2016). Vagal afferent innervation of the airways in health and disease
[Review], Physiol Rev 96: 975-1024. http://dx.doi.org/10.1152/physrev.00039.2015
McConnell. EE: Solleveld. HA: Swenberg. TA: Boorman. GA. (1986). Guidelines for combining
neoplasms for evaluation of rodent carcinogenesis studies. J Natl Cancer Inst 76: 283-289.
http://dx.doi.Org/10.1093/jnci/76.2.283
Mcgradv. AV. (1984). Effects of psychological stress on male reproduction - A review. Arch Androl
13: 1-7. http://dx.doi.org/10.3109/01485018408987495
Mcgregor. D: Bolt. H: Cogliano. V: Richter-Reichhelm. HB. (2006). Formaldehyde and glutaraldehyde
and nasal cytotoxicity: Case study within the context of the 2006 IPCS Human Framework
for the Analysis of a cancer mode of action for humans [Review], Crit Rev Toxicol 36: 821-
835. http://dx.doi.Org/10.1080/10408440600977669
Mchale. CM: Smith. MT: Zhang. L. (2014). Application of toxicogenomic profiling to evaluate effects
of benzene and formaldehyde: from yeast to human [Review], Ann N Y Acad Sci 1310: 74-
83. http://dx.doi.org/10.1111 /nvas. 12382
This document is a draft for review purposes only and does not constitute Agency policy.
R-33 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Mchale. CM: Zhang. L: Smith. MT. (2012). Current understanding of the mechanism of benzene-
induced leukemia in humans: implications for risk assessment [Review], Carcinogenesis 33:
240-252. http://dx.doi.org/10.1093/carcin/bgr297
Mcnamara. CR: Mandel-Brehm. 1: Bautista. DM: Siemens. 1. an: Deranian. KL: Zhao. M: Havward. NT:
Chong. TA: Tulius. D: Moran. MM: Fanger. CM. (2007). TRPA1 mediates formalin-induced
pain. Proc Natl Acad Sci USA 104: 13525-13530.
http://dx.doi.org/10.1073/pnas.07059241Q4
Meek. ME: Boobis. A: Cote. I: Dellarco. V: Fotakis. G: Munn. S: Seed. 1: Vickers. C. (2014). New
developments in the evolution and application of the WHO/IPCS framework on mode of
action/species concordance analysis [Review], J Appl Toxicol 34: 1-18.
http: / /dx. doi. or g/10.10 0 2 /) at 2 949
Mei. YF: Duan. CL: Li. XX: Zhao. Y: Cao. FH: Shang. S: Ding. SM: Yue. XP: Gao. G: Yang. H: Shen. LX:
Feng. XY: Tia. TP: Tong. ZO: Yang. X. (2016). Reduction of Endogenous Melatonin Accelerates
Cognitive Decline in Mice in a Simulated Occupational Formaldehyde Exposure
Environment Int J Environ Res Public Health 13.
http: / /dx. doi. or g/10.3 3 9 0 /ii erphl 3 0 3 0 2 5 8
Mendell. MT: Mirer. AG: Cheung. K: Tong. M: Douwes. 1. (2011). Respiratory and allergic health
effects of dampness, mold, and dampness-related agents: a review of the epidemiologic
evidence [Review], Environ Health Perspect 119: 748-756.
http://dx.doi.org/10.1289/ehp.1002410
Menezes. AM: Perez-Padilla. R: Wehrmeister. FC: Lopez-Varela. MY: Muino. A: Valdivia. G: Lisboa. C:
Tardim. 1. R: de Oca. MM: Talamo. C: Bielemann. R: Gazzotti. M: Laurenti. R: Celli. B: Victora.
CG: team. P. (2014). FEV1 is a better predictor of mortality than FVC: the PLATINO cohort
study. PLoS ONE 9: el09732. http://dx.doi.org/10.1371/iournal.pone.0109732
Meng. F: Bermudez. E: Mckinzie. PB: Andersen. ME: III. CH: Parsons. BL. (2010). Measurement of
tumor-associated mutations in the nasal mucosa of rats exposed to varying doses of
formaldehyde. Regul Toxicol Pharmacol 57: 274-283.
http://dx.doi.Org/10.1016/i.yrtph.2010.03.007
Mercer. RR: Russell. ML: Roggli. VL: Crapo. ID. (1994). Cell number and distribution in human and
rat airways. Am J Respir Cell Mol Biol 10: 613-624.
http://dx.doi.Org/10.1165/aircmb.10.6.8003339
Merler. E: Baldasseroni. A: laria. R: Faravelli. P: Agostini. R: Pisa. R: Berrino. F. (1986). On the causal
association between exposure to leather dust and nasal cancer: Further evidence from a
case-control study. Br J Ind Med 43: 91-95. http://dx.doi.Org/10.1136/oem.43.2.91
Merv. S: Gross. EA: Tovner. PR: Godo. M: Morgan. KT. (1994). Nasal diagrams: A tool for recording
the distribution of nasal lesions in rats and mice. Toxicol Pathol 22: 353-372.
http: //dx.doi.org/10.1177/019262339402200402
Meyers. AR: Pinkerton. LE: Hein. MT. (2013). Cohort mortality study of garment industry workers
exposed to formaldehyde: Update and internal comparisons. Am J Ind Med 56: 1027-1039.
http: / /dx. doi. or g/10.10 0 2 /ai im. 2 219 9
Mi. S: Lu. I: Sun. M: Li. Z: Zhang. H: Neillv. MB: Wang. Y: Oian. Z: Tin. 1: Zhang. Y: Bohlander. SK: Le
Beau. MM: Larson. RA: Golub. TR: Rowley. ID: Chen. T. (2007). MicroRNA expression
signatures accurately discriminate acute lymphoblastic leukemia from acute myeloid
leukemia. Proc Natl Acad Sci USA 104: 19971-19976.
http://dx.doi.Org/10.1073/pnas.0709313104
Mi. YH: Norback. D: Tao. 1: Mi. YL: Ferm. M. (2006). Current asthma and respiratory symptoms
among pupils in Shanghai, China: Influence of building ventilation, nitrogen dioxide, ozone,
and formaldehyde in classrooms. Indoor Air 16: 454-464.
http://dx.d0i.0rg/l 0.1111 /i.l 600-0668.2006.00439.X
This document is a draft for review purposes only and does not constitute Agency policy.
R-34 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Mikhed. Y: Goerlach. A: Knaus. UG: Daiber. A. (2015). Redox regulation of genome stability by
effects on gene expression, epigenetic pathways and DNA damage/repair [Review], 5: 275-
289. http://dx.doi.Org/10.1016/i.redox.2015.05.008
Miller. FT: Conollv. RB: Kimbell. IS. (2017). An updated analysis of respiratory tract cells at risk for
formaldehyde carcinogenesis. Inhal Toxicol 29: 586-597.
http://dx.doi. org/10.1080/08958378.2018.1430191
Miller. MR: Crapo. R: Hankinson. 1: Brusasco. V: Burgos. F: Casaburi. R: Coates. A: Enright. P: van Per
Grinten. CP: Gustafsson. P: Tensen. R: Tohnson. DC: Macintyre. N: Mckav. R: Navaias. D:
Pedersen. OF: Pellegrino. R: Viegi. G: Wanger. 1: Force. AET. (2005a). General considerations
for lung function testing [Review], Eur Respir J 26: 153-161.
http://dx.doi.Org/10.1183/09031936.05.00034505
Miller. MR: Hankinson. 1: Brusasco. V: Burgos. F: Casaburi. R: Coates. A: Crapo. R: Enright. P: van Per
Grinten. CP: Gustafsson. P: Tensen. R: Tohnson. DC: Macintyre. N: Mckav. R: Navaias. D:
Pedersen. OF: Pellegrino. R: Viegi. G: Wanger. T: Force. AET. (2005b). Standardisation of
spirometry. Eur Respir J 26: 319-338. http://dx.doi.Org/10.1183/09031936.05.00034805
Milton. DK: Walters. MP: Hammond. K: Evans. TS. (1996). Worker exposure to endotoxin, phenolic
compounds, and formaldehyde in a fiberglass insulation manufacturing plant Am Ind Hyg
Assoc J 57: 889-896. http://dx.doi.org/10.1080/15428119691014396
Moeller. BC: Lu. K: Dovle-Eisele. M: Mcdonald. 1: Gigliotti. A: Swenberg. TA. (2011). Determination of
N2-hydroxymethyl-dG adducts in the nasal epithelium and bone marrow of nonhuman
primates following 13CD2-formaldehyde inhalation exposure. Chem Res Toxicol 24: 162-
164. http://dx.doi.org/10.1021/txl004166
Mohner. M: Liu. Y: Marsh. GM. (2019). New insights into the mortality risk from nasopharyngeal
cancer in the national cancer institute formaldehyde worker cohort study. J Occup Med
Toxicol 14: 4. http://dx.doi.org/10.1186/sl 2995-019-0224-2
Monfared. AL. (2012). Histomorphological and ultrastructural changes of the placenta in mice
exposed to formaldehyde. Toxicol Ind Health 30: 174-181.
http: //dx.doi.org/10.1177/0748233712452603
Monteiro-Riviere. NA: Popp. TA. (1986). Ultrastructural evaluation of acute nasal toxicity in the rat
respiratory epithelium in response to formaldehyde gas. Fundam Appl Toxicol 6: 251-262.
http://dx.doi.org/10.1016/0272-0590r86190238-l
Monticello. TM: Gross. EA: Morgan. KT. (1993). Cell Proliferation and Nasal Carcinogenesis. Environ
Health Perspect 101: 121. http://dx.doi.org/10.2307/3431854
Monticello. TM: Miller. FT: Morgan. KT. (1991). Regional increases in rat nasal epithelial cell
proliferation following acute and subchronic inhalation of formaldehyde. Toxicol Appl
Pharmacol 111: 409-421. http://dx.doi.org/10.1016/0041-008Xr91190246-B
Monticello. TM: Morgan. KT. (1994). Cell proliferation and formaldehyde-induced respiratory
carcinogenesis [Review], Risk Anal 14: 313-319. http://dx.doi.org/10.1111 /i. 1539-
6924.1994.tb00246.x
Monticello. TM: Morgan. KT: Everitt. II: Popp. TA. (1989). Effects of formaldehyde gas on the
respiratory tract of rhesus monkeys: Pathology and cell proliferation. Am J Pathol 134: 515-
527.
Monticello. TM: Morgan. KT: Hurtt. ME. (1990a). Unit length as the denominator for quantitation of
cell proliferation in nasal epithelia. Toxicol Pathol 18: 24-31.
http://dx.d0i.0rg/l 0.1177/019262339001800104
Monticello. TM: Morgan. KT: Uraih. L. (1990b). Nonneoplastic nasal lesions in rats and mice
[Review], Environ Health Perspect 85: 249-274.
Monticello. TM: Swenberg. TA: Gross. EA: Leininger. TR: Kimbell. TS: Seilkop. S: Starr. TB: Gibson. IE:
Morgan. KT. (1996). Correlation of regional and nonlinear formaldehyde-induced nasal
cancer with proliferating populations of cells. Cancer Res 56: 1012-1022.
This document is a draft for review purposes only and does not constitute Agency policy.
R-35 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Moolgavkar. SH: Dewanji. A: Venzon. PI. (1988). A stochastic two-stage model for cancer risk
assessment. I. The hazard function and the probability of tumor. Risk Anal 8: 383-392.
http://dx.doi. org/10.1111 /i.l 539-6924.1988.tb00502.x
Moolgavkar. SH: Knudson. AG. Jr. (1981). Mutation and cancer: a model for human carcinogens. J
Natl Cancer Inst 66: 1037-1052.
Moolgavkar. SH: Luebeck. EG. (1992). Interpretation of labeling indices in the presence of cell death.
Carcinogenesis 13: 1007-1010. http://dx.doi.Org/10.1093/carcin/13.6.1007
Moolgavkar. SH: Venzon. DT. (1979). Two-event models for carcinogenesis: Incidence curves for
childhood and adult tumors. Math Biosci 47: 55-77. http: //dx.doi.org/10.1016/0Q25-
5564179190005-1
Moore. KW: de Waal Malefvt. R: Coffman. RL: O'Garra. A. (2001). Interleukin-10 and the interleukin-
10 receptor [Review], Annu Rev Immunol 19: 683-765.
http://dx.doi.Org/10.1146/annurev.immunol.19.l.683
Morgan. PL: Dixon. D: King. DH: Travlos. GS: Herbert. RA: French. IE: Tokar. El: Waalkes. MP:
Tokinen. MP. (2017). NTP research report on absence of formaldehyde-induced neoplasia in
Trp53 haploinsufficient mice exposed by inhalation. (Research Report 3). Research Triangle
Park, NC: National Toxicology Program.
https://ntp.niehs.nih.gov/ntp/results/pubs/rr/reports/rr03 508.pdf
Morgan. KT. (1997). A brief review of formaldehyde carcinogenesis in relation to rat nasal
pathology and human health risk assessment [Review], Toxicol Pathol 25: 291-305.
http://dx.doi. org/10.1177/019262339702500307
Morgan. KT: Gross. EA: Patterson. PL. (1986a). Pistribution, progression, and recovery of acute
formaldehyde-induced inhibition of nasal mucociliary function in F-344 rats. Toxicol Appl
Pharmacol 86: 448-456. http://dx.doi.org/10.1016/0041-008Xr86190372-8
Morgan. KT: Tiang. XZ: Starr. TB: Kerns. WP. (1986b). More precise localization of nasal tumors
associated with chronic exposure of F-344 rats to formaldehyde gas. Toxicol Appl
Pharmacol 82: 264-271. http://dx.doi.Org/10.l016/0041-008X(86)90201-2
Morgan. KT: Kimbell. IS: Monticello. TM: Patra. AL: Fleishman. A. (1991). Studies of inspiratory
airflow patterns in the nasal passages of the F344 rat and rhesus monkey using nasal molds:
Relevance to formaldehyde toxicity. Toxicol Appl Pharmacol 110: 223-240.
http://dx.doi.org/10.1016/S0041-008Xr05180005-5
Morgan. KT: Patterson. PL: Gross. EA. (1984). Frog palate mucociliary apparatus: Structure,
function, and response to formaldehyde gas. Fundam Appl Toxicol 4: 58-68.
http://dx.d0i.0rg/l 0.1016/0272-0590r84190219-7
Morgan. KT: Patterson. PL: Gross. EA. (1986c). Responses of the nasal mucociliary apparatus of F-
344 rats to formaldehyde gas. Toxicol Appl Pharmacol 82: 1-13.
http://dx.d0i.0rg/l 0.1016/0041 -008Xr86190431 -X
Morgan. KT: Patterson. PL: Gross. EA. (1983). Formaldehyde and the nasal mucociliary apparatus.
In JJ Clary; JE Gibson; RS Waritz (Eds.), Formaldehyde: toxicology, epidemiology,
mechanisms (pp. 193-209). New York, NY: Marcel Pekker, Inc.
Morren. MA: Przvbilla. B: Bamelis. M: Hevkants. B: Revnaers. A: Pegreef. H. (1994). Atopic
dermatitis: Triggering factors [Review], J Am Acad Permatol 31: 467-473.
http://dx.d0i.0rg/l 0.1016/S0190-9622 r94170213-6
Morrow. TP: Frei. B: Longmire. AW: Gaziano. TM: Lynch. SM: Shvr. Y: Strauss. WE: Oates. TA: Roberts.
LT. (1995). Increase in circulating products of lipid peroxidation (F2-isoprostanes) in
smokers. Smoking as a cause of oxidative damage. N Engl J Med 332: 1198-1203.
http://dx.doi.org/10.1056/NETM199505043321804
Morton. LM: Gibson. TM: Clarke. CA: Lynch. CF: Anderson. LA: Pfeiffer. R: Landgren. 0:
Weisenburger. PP: Engels. EA. (2014). Risk of myeloid neoplasms after solid organ
transplantation. Leukemia 28: 2317-2323. http://dx.doi.org/10.1038/leu.2014.132
This document is a draft for review purposes only and does not constitute Agency policy.
R-36 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Mueller. TU: Bruckner. T: Triebig. G. (2013). Exposure study to examine chemosensory effects of
formaldehyde on hyposensitive and hypersensitive males. Int Arch Occup Environ Health
86: 107-117. http://dx.doi.org/l 0.1007/s00420-012-0745-9
Mullane. K: Williams. M. (2014). Animal models of asthma: Reprise or reboot? [Review], Biochem
Pharmacol 87: 131-139. http://dx.doi.Org/10.1016/i.bcp.2013.06.026
Mullen. NA: Li. 1: Russell. ML: Spears. M: Less. BP: Singer. BC. (2015). Results of the California
Healthy Homes Indoor Air Quality Study of 2011-2013: impact of natural gas appliances on
air pollutant concentrations. Indoor Air 26: 231-245. http://dx.doi.org/10.llll/ina.12190
Mundt. KA: Gallagher. AE: Dell. LP: Natelson. EA: Boffetta. P: Gentry. PR. (2017). Does occupational
exposure to formaldehyde cause hematotoxicity and leukemia-specific chromosome
changes in cultured myeloid progenitor cells? [Review], Crit Rev Toxicol 47: 1-11.
http://dx.doi.Org/10.1080/10408444.2017.1301878
Murphy. MW: Lando. IF: Kieszak. SM: Sutter. ME: Noonan. GP: Brunkard. TM: Mcgeehin. MA. (2013).
Formaldehyde levels in FEMA-supplied travel trailers, park models, and mobile homes in
Louisiana and Mississippi. Indoor Air 23: 134-141. http://dx.doi.org/10.ll 11/j.1600-
0668.2012.00800.x
Murrell. W: Feron. F: Wetzig. A: Cameron. N: Splatt. K: Bellette. B: Bianco. I: Perry. C: Lee. G: Mackav-
Sim. A. (2005). Multipotent stem cells from adult olfactory mucosa. Dev Dyn 233: 496-515.
http ://dx. doi. or g/10.10 0 2 /dvdy. 20360
Musak. L: Smerhovskv. Z: Halasova. E: Osina. 0: Letkova. L: Vodickova. L: Polakova. V: Buchancova.
1: Hemminki. K: Vodicka. P. (2013). Chromosomal damage among medical staff
occupationally exposed to volatile anesthetics, antineoplastic drugs, and formaldehyde.
Scand J Work Environ Health 39: 618-630. http://dx.doi.org/10.5271/siweh.3358
Nakamura. I: Hollev. DW: Kawamoto. T: Bultman. ST. (2020). The failure of two major formaldehyde
catabolism enzymes (ADH5 and ALDH2) leads to partial synthetic lethality in C57BL/6
mice. Genes Environ 42: 21. http://dx.doi.org/10.1186/s41021-020-00160-4
Nakano. T: Katafuchi. A: Matsubara. M: Terato. H: Tsuboi. T: Masuda. T: Tatsumoto. T: Pack. SP:
Makino. K: Croteau. PL: Van Houten. B: Iiiima. K: Tauchi. H: Ide. H. (2009). Homologous
recombination but not nucleotide excision repair plays a pivotal role in tolerance of PNA-
protein cross-links in mammalian cells. J Biol Chem 284: 27065-27076.
http://dx.doi.org/10.1074/ibc.M109.019174
NASEM (National Academies of Sciences, Engineering, and Medicine). (2021). Review of U.S. EPA's
ORP staff handbook for developing IRIS assessments: 2020 version. Washington, PC:
National Academies Press, http://dx.doi.org/10.17226/26289
Navarro-Costa. P: Nogueira. P: Carvalho. M: Leal. F: Cordeiro. I: Calhaz-Torge. C: Goncalves. I:
Plancha. CE. (2010). Incorrect PNA methylation of the PAZL promoter CpG island associates
with defective human sperm. Hum Reprod 25: 2647-2654.
http://dx.doi.org/10.1093/humrep/deq20Q
NCHS (National Center for Health Statistics). (2006). Chartbook on Trends in the Health of
Americans. Hyattsville, MP: National Center for Health Statistics, Centers for Pisease
Control and Prevention, http://www.cdc.gov/nchs/data/hus/hus06.pdf
NCHS (National Center for Health Statistics). (2009). Peaths: Final data for 2006. Natl Vital Stat Rep
57: 1-134.
NCI (National Cancer Institute). (2012). Mortality - All cause of death, total U.S. (1969-2010), from
the surveillance, epidemiology, and end results program. Available online at
http://seer.cancer.gov/mortalitv/
Neamtiu. IA: Lin. S: Chen. ML: Roba. C: Csobod. E. va: Gurzau. ES. (2019). Assessment of
formaldehyde levels in relation to respiratory and allergic symptoms in children from Alba
County schools, Romania. Environ Monit Assess 191: 591.
http://dx.doi.Org/10.1007/sl0661-019-7768-6
This document is a draft for review purposes only and does not constitute Agency policy.
R-37 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Neghab. M: Soltanzadeh. A: Choobineh. A. (2011). Respiratory morbidity induced by occupational
inhalation exposure to formaldehyde. Ind Health 49: 89-94.
http://dx.doi.org/10.2486/indhealth.MS1197
Negro-Vilar. A. (1993). Stress and other environmental factors affecting fertility in men and women:
Overview. Environ Health Perspect 101 Suppl. 2:59. http://dx.doi.org/10.23Q7/3431377
Neuss. S: Moepps. B: Speit. G. (2010). Exposure of human nasal epithelial cells to formaldehyde does
not lead to DNA damage in lymphocytes after co-cultivation. Mutagenesis 25: 359-364.
http://dx.doi.org/10.1093/mutage/geq013
Nielsen. GD. (1991). Mechanisms of activation of the sensory irritant receptor by airborne
chemicals [Review], Crit Rev Toxicol 21: 183-208.
http://dx.d0i.0rg/l 0.3109 /10408449109089879
Nielsen. GD: Hougaard. KS: Larsen. ST: Hammer. M: Wolkoff. P: Clausen. PA: Wilkins. CK: Alarie. Y.
(1999). Acute airway effects of formaldehyde and ozone in BALB/c mice. Hum Exp Toxicol
18: 400-409. http://dx.d0i.0rg/l0.1191 /096032799678840246
Nilsson. TA: Hedberg. IT: Vondracek. M: Staab. CA: Hansson. A: Hoog. TO: Grafstrom. RC. (2004).
Alcohol dehydrogenase 3 transcription associates with proliferation of human oral
keratinocytes. Cell Mol Life Sci 61: 610-617. http://dx.doi.org/10.1007/sQ0018-003-3433-9
Nilsson. TA: Zheng. X: Sundqvist. K: Liu. Y: Atzori. L: Elfwing. A: Arvidson. K: Grafstrom. RC. (1998).
Toxicity of formaldehyde to human oral fibroblasts and epithelial cells: influences of culture
conditions and role of thiol status. J Dent Res 77: 1896-1903.
http://dx.d0i.0rg/l 0.1177/00220345980770110601
Noda. T: Takahashi. A: Kondo. N: Mori. E: Okamoto. N: Nakagawa. Y: Ohnishi. K: Zdzienicka. MZ:
Thompson. LH: Helledav. T: Asada. H: Ohnishi. T. (2011). Repair pathways independent of
the Fanconi anemia nuclear core complex play a predominant role in mitigating
formaldehyde-induced DNA damage. Biochem Biophys Res Commun 404: 206-210.
http://dx.doi.0rg/lO.lOl6/j.bbrc.2OlO.ll.O94
Norback. D: Biornsson. E: Tanson. C: Widstrom. I: Boman. G. (1995). Asthmatic symptoms and
volatile organic compounds, formaldehyde, and carbon dioxide in dwellings. Occup Environ
Med 52: 388-395. http://dx.d0i.0rg/l 0.1136/oem.52.6.388
Norback. D: Hashim. TH: Hashim. Z: Ali. F. (2017). Volatile organic compounds (VOC), formaldehyde
and nitrogen dioxide (N02) in schools in Johor Bahru, Malaysia: Associations with rhinitis,
ocular, throat and dermal symptoms, headache and fatigue. Sci Total Environ 592: 153-160.
http: / /dx. doi. or g/10.1016/i. scitotenv. 2017.02.215
Norback. D: Walinder. R: Wieslander. G: Smedje. G: Erwall. C: Venge. P. (2000). Indoor air pollutants
in schools: nasal patency and biomarkers in nasal lavage. Allergy 55: 163-170.
http://dx.doi.Org/10.1034/i.1398-9995.2000.00353.x
Nordman. H: Keskinen. H: Tuppurainen. M. (1985). Formaldehyde asthma-rare or overlooked? J
Allergy Clin Immunol 75: 91-99. http://dx.d0i.0rg/l0.1016/0091-6749(85)90018-1
NRC (National Research Council). (2001). Standing operating procedures for developing acute
exposure guideline levels (AEGLs) for hazardous chemicals. Washington, DC: National
Academies Press, http://dx.doi.org/10.17226/10122
NRC (National Research Council). (2011). Review of the Environmental Protection Agency's draft
IRIS assessment of formaldehyde (pp. 1-194). Washington, DC: The National Academies
Press, http://dx.doi.org/10.17226/13142
NRC (National Research Council). (2014a). Review of EPA's Integrated Risk Information System
(IRIS) process. Washington, DC: The National Academies Press.
http://dx.doi.org/10.17226/18764
NRC (National Research Council). (2014b). Review of the Formaldehyde Assessment in the National
Toxicology Program 12th Report on Carcinogens. Washington (DC): National Academies
Press (US), http://dx.doi.org/10.17226/18948
This document is a draft for review purposes only and does not constitute Agency policy.
R-38 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
NTP (National Toxicology Program). (1988). Toxicology and carcinogenesis studies of 1,2-
epoxybutane (CAS no 106-88-7) in F344/N rats and B6C3F1 mice (inhalation studies).
NTP (National Toxicology Program). (2010). Final report on carcinogens. Background document for
formaldehyde [NTP] (pp. i-512).
NTP (National Toxicology Program). (2011). Twelfth Report On Carcinogens, 2011. [CBRNIAC-
1953235] (pp. 499). https://www.dtic.mil/DOAC/document?document=CBRNIAC-
1953235&collection=ac-tems&contentType=PDF&citationFormat=lf
NTP (National Toxicology Program). (2014). 13th Report on carcinogens [NTP], Research Triangle
Park, NC: U.S. Department of Health and Human Services, Public Health Service.
NTP (National Toxicology Program). (2015). Handbook for conducting a literature-based health
assessment using OHAT approach for systematic review and evidence integration. U.S. Dept.
of Health and Human Services, National Toxicology Program.
https://ntp.niehs.nih.gov/ntp/ohat/pubs/handbookian2015 508.pdf
Nunn. AT: Craigen. AA: Darbvshire. TH: Venables. KM: Taylor. ATN. f!990I Six year follow up of lung
function in men occupationally exposed to formaldehyde. Br J Ind Med 47: 747-752.
http://dx.doi.org/10.1136/oem.47.ll.747
O'Connor. TM: O'Halloran. DT: Shanahan. F. (2000). The stress response and the hypothalamic -
pituitary-adrenal axis: from molecule to melancholia [Review], QJM 93: 323-333.
Oberbeck. N: Langevin. F: King. G: de Wind. N: Crossan. GP: Patel. KT. (2014). Maternal aldehyde
elimination during pregnancy preserves the fetal genome. Mol Cell 55: 807-817.
http://dx.doi.Org/10.1016/i.molcel.2014.07.010
Odkvist. LM: Edling. C: Hellquist. H. (1985). Influence of vapours on the nasal mucosa among
industry workers. Rhinology 23: 121-127.
OECD (Organisation for Economic Co-operation and Development). (2013). Guidance document
supporting OECD test guideline 443 on the extended one generation reproductive toxicity
test (No. 151 / ENV/JM/MONO(2013)10). Paris, France: OECD Environment, Health and
Safety Publications.
http://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=ENV/TM/MON
0%282013%2910&doclanguage=en
OEHHA (California Office of Environmental Health Hazard Assessment). (2003). Air pollution and
children's health [Fact Sheet], http: //oehha.ca.gov/public info/facts/airkids.html
OEHHA (California Office of Environmental Health Hazard Assessment). (2008). Draft Technical
Support Document for Noncancer Risk Assessment Appendix G. Value of the Haber's Law
Exponent (n) for various gases and vapors for acute RELs developed using OEHHA (1999)
procedures.
Ohmichi. K: Komivama. M: Matsuno. Y: Sawabe. Y: Mivaso. H: Fukata. H: Ohmichi. M: Kadota. T:
Nomura. F: Moria. C. (2006). Relationship between exposure to formaldehyde and
immunoglobulin E (IgE) production during the gross anatomy laboratory. J Health Sci 52:
642-647. http://dx.doi.org/10.1248/ihs.52.642
Olsen. 1: Tensen. 0. (1984). Case-control study on sinonasal cancer and formaldehyde exposure
based on a national data linkage system for occupation and cancer [Abstract], Am J
Epidemiol 120: 459.
Olsen. TH: Asnaes. S. (1986). Formaldehyde and the risk of squamous cell carcinoma of the sinonasal
cavities. Br J Ind Med 43: 769-744. http://dx.doi.org/10.1136/oem.43.ll.769
Olsen. TH: Dossing. M. (1982). Formaldehyde induced symptoms in day care centers. AIHA J 43:
366-370. http://dx.doi.org/10.1080/15298668291409866
Olsen. TH: Tensen. SP: Hink. M: Faurbo. K: Breum. NO: Tensen. ON. (1984). Occupational
formaldehyde exposure and increased nasal cancer risk in man. Int J Cancer 34: 639-644.
http://dx.doi.Org/10.1002/ijc.2910340509
This document is a draft for review purposes only and does not constitute Agency policy.
R-39 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Ott. MG: Teta. 1: Greenberg. HL. (1989). Lymphatic and hematopoietic tissue cancer in a chemical
manufacturing environment Am J Ind Med 16: 631-644.
http://dx.doi.Org/10.1002/aiim.4700160603
Overton. TH: Kimbell. IS: Miller. FT. (2001). Dosimetry modeling of inhaled formaldehyde: The
human respiratory tract. Toxicol Sci 64: 122-134.
http ://dx. doi. or g/10.109 3 /toxsci /64.1.122
Ozen. OA: Akpolat. N: Songur. A: Kus. I: Zararsiz. I: Ozacmak. VH: Sarsilmaz. M. (2005). Effect of
formaldehyde inhalation on Hsp70 in seminiferous tubules of rat testes: An
immunohistochemical study. Toxicol Ind Health 21: 249-254.
http://dx.doi.org/10.n91/0748233705th235oa
Ozen. OA: Kus. MA: Kus. I: Alkoc. OA: Songur. A. (2008). Protective effects of melatonin against
formaldehyde-induced oxidative damage and apoptosis in rat testes: An
immunohistochemical and biochemical study. Sys Biol Reprod Med 54: 169-176.
http://dx.doi.Org/10.1080/19396360802422402
Ozen. OA: Songue. A: Sars. M: Yaman. M: Kus. I. (2003). Changes of zinc, copper, and iron levels in
the lung of male rats after subacute (4-week) and subchronic (13-week) exposure to
formaldehyde. J Trace Elem Exp Med 16: 67-74. http://dx.doi.org/10.1002/itra. 10026
Ozen. OA: Yaman. M: Sarsilmaz. M: Songur. A: Kus. I. (2002). Testicular zinc, copper and iron
concentrations in male rats exposed to subacute and subchronic formaldehyde gas
inhalation. J Trace Elem Med Biol 16: 119-122. http://dx.doi.org/10.1016/S0946-
672XC02180038-4
Paiva. RM: Calado. RT. (2014). Telomere dysfunction and hematologic disorders [Review], 125:
133-157. http://dx.doi.Org/10.1016/B978-0-12-397898-l.00006-2
Palczvnski. C: Krakowiak. A: Hanke. W: Walusiak. 1: Gorski. P. (1999). Indoor formaldehyde
exposure and airway allergic diseases. Int Rev Allergol Clin Immunol 5: 65-69.
Palma. T. (2018). RE: NATA 2011 data. Available online
Park. I: Yang. H: Song. MK: Kim. D: Lee. K. (2020). Formaldehyde exposure induces regulatory T cell-
mediated immunosuppression via calcineurin-NFAT signalling pathway. Sci Rep 10: 17023.
http://dx.doi.Org/10.1038/s41598-020-72502-9
Pavdas. S: Zorludemir. S: Ergin. M. (2006). Granulocytic sarcoma: 32 cases and review of the
literature. Leuk Lymphoma 47: 2527-2541.
http: //dx.doi.org/10.1080/10428190600967196
Pazdrak. K: Gorski. P: Krakowiak. A: Ruta. U. (1993). Changes in nasal lavage fluid due to
formaldehyde inhalation. Int Arch Occup Environ Health 64: 515-519.
http://dx.doi.Org/10.1007/BF00381101
Pedersen-Biergaard. 1: Christiansen. DH: Desta. F: Andersen. MK. (2006). Alternative genetic
pathways and cooperating genetic abnormalities in the pathogenesis of therapy-related
myelodysplasia and acute myeloid leukemia [Review], Leukemia 20: 1943-1949.
http://dx.doi.org/10.1038/si.leu.24Q4381
Pellegrino. R: Viegi. G: Brusasco. V: Crapo. RO: Burgos. F: Casaburi. R: Coates. A: van Per Grinten. CP:
Gustafsson. P: Hankinson. I: Tensen. R: Tohnson. DC: Macintyre. N: Mckav. R: Miller. MR:
Navaias. D: Pedersen. OF: Wanger. 1. (2005). Interpretative strategies for lung function tests.
Eur Respir J 26: 948-968. http://dx.doi.org/10.1183/09031936.05.00035205
Percy. C: Stanek. E. Ill: Gloeckler. L. (1981). Accuracy of cancer death certificates and its effect on
cancer mortality statistics. Am J Public Health 71: 242-250.
http://dx.doi.Org/10.2105/ATPH.71.3.242
Percy. CL: Miller. BA: Gloeckler Ries. LA. (1990). Effect of changes in cancer classification and the
accuracy of cancer death certificates on trends in cancer mortality. Ann N Y Acad Sci 609:
87-99. http://dx.doi.Org/10.llll /i.1749-6632.1990.tb32059.x
This document is a draft for review purposes only and does not constitute Agency policy.
R-40 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Pesch. B: Pierl. CB: Gebel. M: Gross. I: Becker. D: Tohnen. G: Rihs. HP: Donhuiisen. K: Lepentsiotis. V:
Meier. M: Schulze. T: Briining. T. (2008). Occupational risks for adenocarcinoma of the nasal
cavity and paranasal sinuses in the German wood industry. Occup Environ Med 65: 191-
196. http://dx.doi.org/10.n36/oem.2007.033886
Peteffi. GP: Basso da Silva. L: Antunes. MY: Wilhelm. C: Valandro. ET: Glaeser. 1: Kaefer. D: Linden. R.
(2015). Evaluation of genotoxicity in workers exposed to low levels of formaldehyde in a
furniture manufacturing facility. Toxicol Ind Health 32: 1763-1773.
http://dx.doi.org/10.1177/0748233715584250
Peters. TL: Kamel. F: Lundholm. C: Fevchting. M: Weibull. CE: Sandler. DP: Wiebert. P: Sparen. P: Ye.
W: Fang. F. (2017). Occupational exposures and the risk of amyotrophic lateral sclerosis.
Occup Environ Med 74: 87-92. http://dx.doi.org/10.1136/oemed-2016-103700
Pinkerton. LE: Hein. Ml: Meyers. A: Kamel. F. (2013). Assessment of ALS mortality in a cohortof
formaldehyde-exposed garment workers. 14: 353-355.
http: //dx.doi.org/10.3109/21678421.2013.778284
Pinkerton. LE: Hein. Ml: Stavner. LT. (2004). Mortality among a cohort of garment workers exposed
to formaldehyde: an update. Occup Environ Med 61: 193-200.
http://dx.doi.org/10.1136/oem.2003.0Q7476
Pira. E: Romano. C: Verga. F: La Vecchia. C. (2014). Mortality from lymphohematopoietic neoplasms
and other causes in a cohort of laminated plastic workers exposed to formaldehyde. Cancer
Causes Control 25: 1343-1349. http://dx.doi.org/10.1007/slQ552-014-0440-0
Pitten. FA: Kramer. A: Herrmann. K: Bremer. 1: Koch. S. (2000). Formaldehyde neurotoxicity in
animal experiments. Pathol Res Pract 196: 193-198. http://dx.doi.org/10.1016/SQ344-
0338r00180100-4
Plant. TM: Marshall. GR. (2001). The functional significance of FSH in spermatogenesis and the
control of its secretion in male primates [Review], Endocr Rev 22: 764-786.
http://dx.doi.Org/10.1210/edrv.22.6.0446
Pontel. LB: Rosado. IV: Burgos-Barragan. G: Garavcoechea. II: Yu. R: Arends. Ml: Chandrasekaran. G:
Broecker. V: Wei. W: Liu. L: Swenberg. TA: Crossan. GP: Patel. KT. (2015). Endogenous
Formaldehyde Is a Hematopoietic Stem Cell Genotoxin and Metabolic Carcinogen. Mol Cell
60: 177-188. http://dx.d0i.0rg/l 0.1016/i.molcel.2015.08.020
Poon. R: Chu. I: Biarnason. S: Vincent. R: Potvin. M: Miller. RB: Valli. YE. (1995). Short-term
inhalation toxicity of methanol, gasoline, and methanol/gasoline in the rat Toxicol Ind
Health 11: 343-361. http://dx.d0i.0rg/l0.1177/074823379501100305
Popa. V: Teculescu. D: Stanescu. D: Gavrilescu. N. (1969). Bronchial asthma and asthmatic
bronchitis determined by simple chemicals. Dis Chest 56: 395-402.
Porsolt. RD: Bertin. A: Talfre. M. (1977). Behavioral despair in mice: a primary screening test for
antidepressants. Arch Int Pharmacodyn Ther 229:327-336.
Portier. CI: Bailer. AT. (1989). Testing for increased carcinogenicity using a survival-adjusted
quantal response test FundamAppl Toxicol 12: 731-737. http://dx.doi.org/10.1016/0272-
0590C89190004-3
Poteracki. 1: Walsh. KM. (1998). Spontaneous neoplasms in control Wistar rats: A comparison of
reviews. Toxicol Sci 45: 1-8. http://dx.doi.org/10.1006/toxs.1998.2493
Pottern. LM: Heineman. EF: Olsen. TH: Raffn. E: Blair. A. (1992). Multiple myeloma among Danish
women: Employment history and workplace exposures. Cancer Causes Control 3: 427-432.
http://dx.doi.Org/10.1007/BF00051355
Prades. TM: Alaani. A: Mosnier. IF: Dumollard. TM: Martin. C. (2002). Granulocytic sarcoma of the
nasal cavity. Rhinology40: 159-161.
Priha. E: Pennanen. S: Rantio. T: Uitti. 1: Liesivuori. I. (2004). Exposure to and acute effects of
medium-density fiber board dust J Occup Environ Hyg 1: 738-744.
http://dx.doi.Org/10.1080/15459620490520774
This document is a draft for review purposes only and does not constitute Agency policy.
R-41 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Pushkina. NN: Gofmekler. VA: Klevtsova. GN. (1968). Change in ascorbic acid and nucleic acid levels
upon exposure to benzene and formaldehyde. Biull Eksp Biol Med 66: 51-53.
Qiao. Y: Li. B: Yang. G: Yao. H: Yang. 1: Liu. D: Yan. Y: Sigsgaard. T: Yang. X. (2009). Irritant and
adjuvant effects of gaseous formaldehyde on the ovalbumin-induced hyperresponsiveness
and inflammation in a rat model. Inhal Toxicol 21: 1200-1207.
http://dx.doi.org/10.3109/08958370902806159
Ouackenboss. 11: Lebowitz. MP: Bronnimann. D: Michaud. TP. (1987). Formaldehyde exposure and
acute health effects study. In Indoor air '87: Proceedings of the 4th international conference
on indoor air quality and climate Vol 2: Environmental tobacco smoke, multicomponent
studies, radon, sick buildings, odours and irritants, hyperreactivities and allergies. Berlin:
Institute for Water, Soil and Air Hygiene.
Ouackenboss. IT: Lebowitz. MP: Hayes. C. (1989a). Epidemiological study of respiratory responses
to indoor/outdoor air quality. Environ Int 15: 493-502. http://dx.doi.org/10.1016/016Q-
4120C89190067-6
Ouackenboss. IT: Lebowitz. MP: Hayes. C: Young. CL. (1989b). Respiratory responses to
indoor/outdoor air pollutants: combustion pollutants, formaldehyde, and particulate
matter.
Ouackenboss. II: Lebowitz. MP: Michaud. TP: Bronnimann. P. (1989c). Formaldehyde exposure and
acute health effects study. Environ Int 15: 169-176. http://dx.doi.org/10.1016/0160-
4120C89190023-8
Que. LG: Liu. L: Yan. Y: Whitehead. GS: Gavett. SH: Schwartz. PA: Stamler. IS. (2005). Protection from
experimental asthma by an endogenous bronchodilator. Science 308: 1618-1621.
http: / /dx. doi. or g/10.112 6 /science. 110 82 2 8
Raaschou-Nielsen. 0: Hermansen. MN: Loland. L: Buchvald. F: Pipper. CB: S0rensen. M: Loft. S:
Bisgaard. H. (2010). Long-term exposure to indoor air pollution and wheezing symptoms in
infants. Indoor Air 20: 159-167. http://dx.doi.Org/10.llll/j.1600-0668.2009.00635.x
Ragan. PL: Boreiko. CI. (1981). Initiation of C3H/10T1/2 cell transformation by formaldehyde.
Cancer Lett 13: 325-331. http://dx.doi.org/10.1016/0304-3835r81190061-6
Rager. IE: Moeller. BC: Povle-Eisele. M: Kracko. P: Swenberg. TA: Fry. RC. (2013). Formaldehyde and
epigenetic alterations: microRNA changes in the nasal epithelium of nonhuman primates.
Environ Health Perspect 121: 339-344. http://dx.doi.org/10.1289/ehp.12Q5582
Rager. IE: Moeller. BC: Miller. SK: Kracko. P: Povle-Eisele. M: Swenberg. TA: Fry. RC. (2014).
Formaldehyde-Associated Changes in microRNAs: Tissue and Temporal Specificity in the
Rat Nose, White Blood Cells, and Bone Marrow. Toxicol Sci 138: 36-46.
http: / /dx. doi. or g/10.109 3 /toxsci /kft2 67
Rager. IE: Smeester. L: Taspers. I: Sexton. KG: Fry. RC. (2011). Epigenetic changes induced by air
toxics: formaldehyde exposure alters miRNA expression profiles in human lung cells.
Environ Health Perspect 119: 494-500. http: //dx.doi.org/10.1289/ehp.10Q2614
Ramirez. A: Saldanha. PH. (2002). Micronucleus investigation of alcoholic patients with oral
carcinomas. Genet Mol Res 1: 246-260.
Rao. GN: Piegorsch. WW: Haseman. IK. (1987). Influence of body weight on the incidence of
spontaneous tumors in rats and mice of long-term studies. Am J Clin Nutr 45: 252-260.
Recio. L: Sisk. S: Pluta. L: Bermudez. E: Gross. EA: Chen. Z: Morgan. K: Walker. C. (1992). p53
mutations in formaldehyde-induced nasal squamous cell carcinomas in rats. Cancer Res 52:
6113-6116.
Redlich. CA: Tarlo. SM: Hankinson. TL: Townsend. MC: Eschenbacher. WL: Von Essen. SG: Sigsgaard.
T: Weissman. PN: Setting. ATSCoSitO. (2014). Official American Thoracic Society technical
standards: spirometry in the occupational setting : Supplementary materials [Supplemental
Pata], Am J Respir Crit Care Med 189.
This document is a draft for review purposes only and does not constitute Agency policy.
R-42 PRAFT-PO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Rekhadevi. PV: Mahboob. M: Rahman. MF: Grover. P. (2009). Genetic damage in wood dust-exposed
workers. Mutagenesis 24: 59-65. http://dx.doi.org/10.1093/mutage/gen053
Ren. X: Ti. Z: Mchale. CM: Yuh. T: Bersonda. T: Tang. M: Smith. MT: Zhang. L. (2013). The impact of
FANCD2 deficiency on formaldehyde-induced toxicity in human lymphoblastoid cell lines.
Arch Toxicol 87: 189-196. http: //dx.doi.Org/10.1007/s00204-012-0911-6
Renne. R: Brix. A. my: Harkema. 1: Herbert. R: Kittel. B: Lewis. D: March. T: Nagano. K: Pino. M:
Rittinghausen. S: Rosenbruch. M: Tellier. P: Wohrmann. T. (2009). Proliferative and
nonproliferative lesions of the rat and mouse respiratory tract Toxicol Pathol 37: 5S-73S.
http://dx.doi. org/10.1177/0192623309353423
Renne. RA: Gideon. KM. (2006). Types and patterns of response in the larynx following inhalation.
Toxicol Pathol 34: 281-285. http://dx.doi.org/10.1080/01926230600695631
Reuzel. PGT: Wilmer. TWG. M: Woutersen. RA: Zwart. A: Rombout. PTA: Feron. VI. (1990). Interactive
effects of ozone and formaldehyde on the nasal respiratory lining epithelium in rats. J
Toxicol Environ Health 29: 279-292. http: / /dx.doi.org/10.1080/15287399009531391
Reznik. GK. (1990). Comparative anatomy, physiology, and function of the upper respiratory tract
Environ Health Perspect 85: 171-176. http: //dx.doi.org/10.2307/3430681
Ridpath. TR: Nakamura. A: Tano. K: Luke. AM: Sonoda. E: Arakawa. H: Buerstedde. TM: Gillespie. DA:
Sale. IE: Yamazoe. M: Bishop. DK: Takata. M: Takeda. S: Watanabe. M: Swenberg. TA:
Nakamura. 1. (2007). Cells deficient in the FANC/BRCA pathway are hypersensitive to
plasma levels of formaldehyde. Cancer Res 67: 11117-11122.
http: //dx.doi.org/10.1158/0008-5472.CAN-07-3028
Riedel. F: Hasenauer. E: Barth. PI: Koziorowski. A: Rieger. CHL. (1996). Formaldehyde exposure
enhances inhalative allergic sensitization in the guinea pig. Allergy 51: 94-99.
http://dx.doi.Org/10.llll/i.1398-9995.1996.tb00041.x
Riess. U: Tegtbur. U: Fauck. C: Fuhrmann. F: Markewitz. D: Salthammer. T. (2010). Experimental
setup and analytical methods for the non-invasive determination of volatile organic
compounds, formaldehyde and NOx in exhaled human breath. Anal Chim Acta 669: 53-62.
http://dx.doi.Org/10.1016/i.aca.2010.04.049
Rigbv. M: O'Donnell. R: Rupniak. NM. (2005). Species differences in tachykinin receptor
distribution: further evidence that the substance P (NK1) receptor predominates in human
brain. J Comp Neurol 490: 335-353. http://dx.doi.org/10.1002/cne.20664
Rinskv. RA: Ott. MG: Ward. E: Greenberg. HL: Halperin. W: Leet. T. (1988). Study of mortality among
chemical workers in the Kanawha Valley of West Virginia. Am J Ind Med 13: 429-438.
http://dx.doi.Org/10.1002/ajim.4700130403
Rinskv. RA: Smith. AB: Hornung. R: Filloon. TG: Young. RT: Okun. AH: Landrigan. PI. (1987). Benzene
and leukemia. An epidemiologic risk assessment N Engl J Med 316: 1044-1050.
http://dx.doi.org/10.1056/NETM198704233161702
Roberts. DM: Yates. C: Megarbane. B: Winchester. IF: Maclaren. R: Gosselin. S: Nolin. TP: Lavergne.
V: Hoffman. RS: Ghannoum. M: Group. EW. (2015). Recommendations for the role of
extracorporeal treatments in the management of acute methanol poisoning: a systematic
review and consensus statement [Review], Crit Care Med 43: 461-472.
http://dx.doi.Org/10.1097/CCM.0000000000000708
Robinson. CF: Fowler. D: Brown. DP: Lemen. RA. (1987). Plywood mill workers' mortality patterns
1945 1977 (revised March 1987). (NIOSH/00197140). Cincinnati, OH: NIOSH.
Roda. C: Kousignian. I: Guihenneuc-Touyaux. C: Dassonville. C: Nicolis. I: lust. 1: Momas. I. (2011).
Formaldehyde exposure and lower respiratory infections in infants: findings from the
PARIS cohort study. Environ Health Perspect 119: 1653-1658.
http://dx.doi.org/10.1289/ehp.1003222
This document is a draft for review purposes only and does not constitute Agency policy.
R-43 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Rodrigo. R: Prat. H: Passalacqua. W: Arava. I: Guichard. C: Bachler. TP. (2007). Relationship between
oxidative stress and essential hypertension. Hypertens Res 30: 1159-1167.
http://dx.doi.org/10.1291/hypres.30.1159
Romanazzi. V: Pirro. V: Bellisario. V: Mengozzi. G: Peluso. M: Pazzi. M: Bugiani. M: Verlato. G: Bono.
R. (2013). 15-F2tisoprostane as biomarker of oxidative stress induced by tobacco smoke
and occupational exposure to formaldehyde in workers of plastic laminates. Sci Total
Environ 442: 20-25. http://dx.doi.org/10.1016/i.scitotenv.2012.10.057
Rosado. IV: Langevin. F: Crossan. GP: Takata. M: Patel. KT. (2011). Formaldehyde catabolism is
essential in cells deficient for the Fanconi anemia DNA-repair pathway. Nat Struct Mol Biol
18: 1432-1434. http://dx.doi.org/10.1038/nsmb.2173
Rothman. KT: Boice. TP. (1979). Epidemiologic analysis with a programmable calculator. (NIH
Publication No. 79-1649). Washington, D.C.: National Institutes of Health.
Rothman. N: Lan. 0: Smith. MT: Vermeulen. R: Zhang. L. (2017). Response to letter to the editor of
Carcinogenesis by Pira etal., 2017 [Letter], Carcinogenesis 38: 1253-1255.
http://dx.doi.org/10.1093/carcin/bgxlll
Roush. GC: Schvmura. MT: Stevenson. TM: Holford. TR. (1987a). Time and age trends for sinonasal
cancer in Connecticut incidence and US mortality rates. Cancer 60: 422-428.
http://dx.doi.org/10.1002 /1097-0142 ("19870801 "160:3<422::aid-
cncr2820600324>3.0.co;2-r
Roush. GC: Walrath. 1: Stavner. LT: Kaplan. SA: Flannerv. IT: Blair. A. (1987b). Nasopharyngeal
cancer sinonasal cancer and occupations related to formaldehyde: A case-control study. J
Natl Cancer Inst 79: 1221-1224.
Rozhok. AI: Wahl. GM: Degregori. I. (2015). A critical examination of the "bad luck" explanation of
cancer risk [Comment], 8: 762-764. http://dx.doi.org/10.1158/1940-6207.CAPR-15-0229
Rumchev. K: Spickett. I: Bulsara. M: Phillips. M: Stick. S. (2004). Association of domestic exposure to
volatile organic compounds with asthma in young children. Thorax 59: 746-751.
http://dx.doi.org/10.1136/thx.2003.01368Q
Rumchev. KB: Spickett. IT: Bulsara. MK: Phillips. MR: Stick. SM. (2002). Domestic exposure to
formaldehyde significantly increases the risk of asthma in young children. Eur Respir J 20:
403-408. http://dx.doi.Org/10.1183 /09031936.02.00245002
Rusch. GM: Clary. IT: Rinehart. WE: Bolte. HF. (1983). A 26-week inhalation toxicity study with
formaldehyde in the monkey, rat, and hamster. Toxicol Appl Pharmacol 68: 329-343.
http: //dx.doi.org/10.1016/0041-008XC83190276-4
Russell. NH. (1997). Biology of acute leukaemia. Lancet 349: 118-122.
Russo. 1: Gusterson. BA: Rogers. AE: Russo. IH: Wellings. S. R.: Van Zwieten. MT. (1990). Comparative
study ofhuman and rat mammary tumorigenesis [Review], Lablnvest62: 244-278.
Saberi Hosniieh. F: Christopher. Y: Peeters. P: Romieu. I: Xun. W: Riboli. E: Raaschou-Nielsen. 0:
Ti0nneland. A: Becker. N: Nieters. A: Trichopoulou. A: Bamia. C: Orfanos. P: Oddone. E:
Luian-Barroso. L: Dorronsoro. M: Navarro. C: Barricarte. A: Molina-Montes. E: Wareham. N:
Vineis. P: Vermeulen. R. (2013). Occupation and risk of lymphoid and myeloid leukaemia in
the European Prospective Investigation into Cancer and Nutrition (EPIC). Occup Environ
Med 70: 464-470. http://dx.doi.org/10.1136/oemed-2012-101135
Sadakane. K: Takano. H: Ichinose. T: Yanagisawa. R: Shibamoto. T. (2002). Formaldehyde enhances
mite allergen-induced eosinophilic inflammation in the murine airway. J Environ Pathol
Toxicol Oncol 21: 267-276.
Sahin-Yilmaz. A: Naclerio. RM. (2011). Anatomy and physiology of the upper airway [Review], Proc
Am Thorac Soc 8: 31-39. http://dx.doi.org/10.1513/pats.201007-050RN
Saillenfait. AM: Bonnet. P: de Ceaurriz. 1. (1989). The effects of maternally inhaled formaldehyde on
embryonal and foetal development in rats. Food Chem Toxicol 27: 545-548.
http://dx.doi.org/10.1016/0278-6915r89190051-3
This document is a draft for review purposes only and does not constitute Agency policy.
R-44 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Saito. Y: Nishio. K: Yoshida. Y: Niki. E. (2005). Cytotoxic effect of formaldehyde with free radicals via
increment of cellular reactive oxygen species. Toxicology 210: 235-245.
http://dx.doi.Org/10.1016/i.tox.2005.02.006
Salthammer. T: Mentese. S: Marutzkv. R. (2010). Formaldehyde in the indoor environment Chem
Rev 110: 2536-2572. http://dx.doi.org/10.1021/cr800399g
Sandel. M: Murphy. IS: Dixon. SL: Adgate. TL: Chew. GL: Dorevitch. S: Tacobs. DE. (2014). A side-by-
side comparison of three allergen sampling methods in settled house dust. J Expo Sci
Environ Epidemiol 24: 650-656. http: / /dx. doi. or g/10.10 3 8 /i e s. 2 014.3 0
Sandikci. M: Eren. U: Kum. S. (2007a). Effects of formaldehyde and xylene on alpha-naphthyl acetate
esterase positive T-lymphocytes in bronchus associated lymphoid tissue and peripheral
blood in rats. Rev Med Vet 158: 297-301.
Sandikci. M: Eren. U: Kum. S. (2007b). Effects of formaldehyde and xylene on CD4- and CD8-positive
T cells in bronchus-associated lymphoid tissue in rats. Toxicol Ind Health 23: 471-477.
http: //dx.doi.org/10.1177/0748233708089025
Santiago. LY: Hann. MC: Ben-lebria. A: Ultman. IS. (2001). Ozone absorption in the human nose
during unidirectional airflow. J Appl Physiol (1985) 91: 725-732.
http://dx.doi.Org/10.1152/iappl.2001.91.2.725
Santovito. A: Schiliro. T: Castellano. S: Cervella. P: Bigatti. MP: Gilli. G: Bono. R: Delpero. M. (2011).
Combined analysis of chromosomal aberrations and glutathione S-transferase Ml and T1
polymorphisms in pathologists occupationally exposed to formaldehyde. Arch Toxicol 85:
1295-1302. http:/ /dx.doi.org/10.1007/s00204-011 -0668-3
Sapmaz. HI: Sarsilmaz. M: Godekmerdan. A: Ogetiirk. M: Tas. U: Kose. E. (2015). Effects of
formaldehyde inhalation on humoral immunity and protective effect of Nigella sativa oil: An
experimental study. Toxicol Ind Health 32: 1564-1569.
http://dx.doi.org/10.1177/0748233714566294
Sapmaz. HI: Yildiz. A: Polat. A: Vardi. N: Kose. E: Tanbek. K: Cuglan. S. (2018). Protective efficacy of
Nigella sativa oil against the harmful effects of formaldehyde on rat testicular tissue. 8: 548-
553. http://dx.doi.org/10.4103/2221-1691.245970
Sari. DK: Kuwahara. S: Tsukamoto. Y: Hori. H: Kunugita. N: Arashidani. K: Fuiimaki. H: Sasaki. F.
(2004). Effect of prolonged exposure to low concentrations of formaldehyde on the
corticotropin releasing hormone neurons in the hypothalamus and adrenocorticotropic
hormone cells in the pituitary gland in female mice. Brain Res 1013: 107-116.
http://dx.doi.Org/10.1016/i.brainres.2004.03.070
Sarigiannis. DA: Karakitsios. SP: Gotti. A: Liakos. IL: Katsoviannis. A. (2011). Exposure to major
volatile organic compounds and carbonyls in European indoor environments and associated
health risk [Review], Environ Int 37: 743-765.
http://dx.doi.Org/10.1016/i.envint2011.01.005
Sarsilmaz. M: Kaplan. S: Songur. A: Colakoglu. S: Asian. H: Tunc. AT: Ozen. OA: Turgut. M: Bas. 0.
(2007). Effects of postnatal formaldehyde exposure on pyramidal cell number, volume of
cell layer in hippocampus and hemisphere in the rat: A stereological study. Brain Res 11:
157-167. http://dx.doi.Org/10.1016/j.brainres.2007.01.139
Sarsilmaz. M: Ozen. OA: Akpolat. N: Kus. I: Songur. A. (1999). The histopathologic effects of inhaled
formaldehyde on leydig cells of the rats in subacute period. Firat Univ Med Sci J 13: 37-40.
Sauder. LR: Chatham. MP: Green. DT: Kulle. TT. (1986). Acute pulmonary response to formaldehyde
exposure in healthy nonsmokers. J Occup Environ Med 28: 420-424.
http://dx.doi.Org/10.1097/00043764-198606000-00008
Sauder. LR: Green. DT: Chatham. MP: Kulle. TT. (1987). Acute pulmonary response of asthmatics to
3.0 ppm formaldehyde. Toxicol Ind Health 3: 569-578.
http: //dx.doi.org/10.1177/074823378700300408
This document is a draft for review purposes only and does not constitute Agency policy.
R-45 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Saurel-Cubizolles. Ml: Hays. M: Estrvn-Behar. M. (1994). Work in operating rooms and pregnancy
outcome among nurses. Int Arch Occup Environ Health 66: 235-241.
http://dx.doi.org/10.1007/bf0Q454361
Schachter. EN: Witek T T. T. R.: Tosun. T: Leaderer. BP: Reck. GT. (1986a). A STUDY OF RESPIRATORY
EFFECTS FROM EXPOSURE TO 2 PARTS-PER-MILLION FORMALDEHYDE IN HEALTHY
SUBJECTS. Arch Environ Health 41: 229-239.
Schachter. EN: Witek. TT. Tr: Brodv. DT: Tosun. T: Beck. GT: Leaderer. BP. (1987). A study of
respiratory effects from exposure to 2.0 ppm formaldehyde in occupationally exposed
workers. Environ Res 44: 188-205. http://dx.doi.org/1 0.1016/S0013-9351C87180227-X
Schachter. EN: Witek. TT. Tr: Tosun. T: Leaderer. BP: Beck. GT. (1986b). A study of respiratory effects
from exposure to 2 ppm formaldehyde in healthy subjects. Arch Environ Occup Health 41:
229-239. http://dx.doi.Org/10.1080/00039896.1986.9938338
Schafer. D: Brommer. C: Riechelmann. H: Mann. TW. (1999). In vivo and in vitro effect of ozone and
formaldehyde on human nasal mucociliary transport system. Rhinology 37: 56-60.
Schlosser. PM. (1999). Relative roles of convection and chemical reaction for the disposition of
formaldehyde and ozone in nasal mucus. Inhal Toxicol 11: 967-980.
http://dx.doi. org/10.1080/089583799196736
Schlosser. PM: Lilly. PD: Conollv. RB: Tanszen. DB: Kimbell. IS. (2003). Benchmark dose risk
assessment for formaldehyde using airflow modeling and a single-compartment, DNA-
protein cross-link dosimetry model to estimate human equivalent doses. Risk Anal 23: 473-
487. http://dx.doi.org/10.llll /1539-6924.00328
Schoenberg. IB: Mitchell. CA. (1975). Airway disease caused by phenolic (phenol-formaldehyde)
resin exposure. Arch Environ Health 30: 574-577.
http://dx.doi.Org/10.1080/00039896.1975.10666782
Schroeder. EB: Welch. VL: Couper. D: Nieto. FT: Liao. DP: Rosamond. WD: Heiss. G. (2003). Lung
function and incident coronary heart disease - The atherosclerosis risk in communities
study. Am J Epidemiol 158: 1171-1181. http://dx.doi.org/10.1093/aie /kwg276
Schroeter. ID: Campbell. I: Kimbell. IS: Conollv. RB: Clewell. HI: Andersen. ME. (2014). Effects of
endogenous formaldehyde in nasal tissues on inhaled formaldehyde dosimetry predictions
in the rat, monkey, and human nasal passages. Toxicol Sci 138: 412-424.
http: / /dx. doi. or g/10.109 3 /toxsci /kft3 3 3
Schulz. C: von Andrian. UH: Massberg. S. (2009). Hematopoietic stem and progenitor cells: their
mobilization and homing to bone marrow and peripheral tissue [Review], Immunol Res 44:
160-168. http: //dx.doi.org/10.1007/sl 2026-009-8109-6
Schunemann. HI: Porn. 1: Grant. BIB: Winkelstein. W. Tr: Trevisan. M. (2000). Pulmonary function is
a long-term predictor of mortality in the general population: 29-year follow-up of the
Buffalo health study. Chest 118: 656-664. http://dx.doi.Org/10.1378/chest.118.3.656
SCOEL (Scientific Committee on Occupational Exposure Limits). (2017). SCOEL/REC/125
formaldehyde: recommendation from the scientific committee on occupational exposure
limits.
Seals. RM: Kioumourtzoglou. MA: Gredal. 0: Hansen. I: Weisskopf. MG. (2017). Occupational
formaldehyde and amyotrophic lateral sclerosis. Eur J Epidemiol 32: 893-899.
http: //dx.doi.org/10.1007/sl 0654-017-0249-8
Seitz. T: Baron. S. (1990). Health hazard evaluation report No. HETA-87-349-2022, Rockcastle
Manufacturing, Mount Vernon, Kentucky (pp. 87-349). (HETA-87-349-2022). Cincinnati,
OH: National Institute of Occupational Safety and Health.
http s://www, cdc. gov /niosh /nioshtic- 2/00194425. html
Sellakumar. AR: Snyder. CA: Solomon. II: Albert. RE. (1985). Carcinogenicity of formaldehyde and
hydrogen chloride in rats. Toxicol Appl Pharmacol 81: 401-406.
http://dx.doi.org/10.1016/0041-008Xr85190411-9
This document is a draft for review purposes only and does not constitute Agency policy.
R-46 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Selvin. S: Levin. LI: Merrill. DW: Winkelstein. W. Tr. (1983). Selected epidemiologic observations of
cell-specific leukemia mortality in the United States, 1969-1977. Am J Epidemiol 117: 140-
152. http://dx.doi.org/10.1093/oxfordiournals.aie.all3524
Sengupta. P. (2013). The laboratory rat: Relating its age with human's [Review], IJPM 4: 624-630.
Senichenkova. II. (1991a). Embryotoxic effects of industrial environment pollutants: Formaldehyde
and gasoline. Gig Sanit-: 35-38.
Senichenkova. IN. (1991b). Embryotoxic effect of formaldehyde and gasoline pollutants of
industrial environment. Gig Sanit 0: 35-38.
Senichenkova. IN: Chebotar. NA. (1996a). [The effect of benzine and formaldehyde on the prenatal
development of rats with induced iron trace-element disorder], Ontogenez 27: 108-113.
Senichenkova. IN: Chebotar. NA. (1996b). Effects of gasoline and formaldehyde on prenatal
development of rats with induced iron micronutrient disorder (iron deficiency). Ontogenez
27: 108-113.
Seow. WT: Zhang. L: Vermeulen. R: Tang. X: Hu. W: Bassig. BA: Ti. Z: Shiels. MS: Kemp. TT: Shen. M:
Oiu. C: Reiss. B: Beane Freeman. LE: Blair. A: Kim. C: Guo. W: Wen. C: Li. L: Pinto. LA: Huang.
H: Smith. MT: Hildesheim. A: Rothman. N: Lan. 0. (2015). Circulating immune/inflammation
markers in Chinese workers occupationally exposed to formaldehyde. Carcinogenesis 36:
852-857. http://dx.doi.org/10.1093/carcin/bgv055
Serre. K: Mohr. E: Gaspal. F: Lane. PI: Bird. R: Cunningham. AF: Maclennan. IC. (2010). IL-4 directs
both CD4 and CD8 T cells to produce Th2 cytokines in vitro, but only CD4 T cells produce
these cytokines in response to alum-precipitated protein in vivo. Mol Immunol 47: 1914-
1922. http://dx.doi.Org/10.1016/i.molimm.2010.03.010
Sexton. K: Liu. KS: Petreas. MX. (1986). Formaldehyde concentrations inside private residences: A
mail-out approach to indoor air monitoring. J Air Pollut Control Assoc 36: 698-704.
http://dx.doi.org/10.1080/00022470.1986.10466104
Shaham. I: Bomstein. Y: Gurvich. R: Rashkovskv. M: Kaufman. Z. (2003). DNA-protein crosslinks and
p53 protein expression in relation to occupational exposure to formaldehyde. Occup
Environ Med 60: 403-409. http://dx.doi.Org/10.1136/oem.60.6.403
Shaham. 1: Bomstein. Y: Meltzer. A: Kaufman. Z: Palma. E: Ribak. I. (1996). DNA-protein crosslinks, a
biomarker of exposure to formaldehyde - in vitro and in vivo studies. Carcinogenesis 17:
121-126. http://dx.doi.Org/10.1093/carcin/17.l.121
Shaham. 1: Bomstein. Y: Melzer. A: Ribak. 1. (1997). DNA-protein crosslinks and sister chromatid
exchanges as biomarkers of exposure to formaldehyde. Int J Occup Environ Health 3: 95-
104. http://dx.doi.org/10.1179/107735297800407695
Shangina. 0: Brennan. P: Szeszenia-Dabrowska. N: Mates. D: Fabianova. E: Fletcher. T: T'Mannetie.
A: Boffetta. P: Zaridze. D. (2006). Occupational exposure and laryngeal and hypopharyngeal
cancer risk in central and eastern Europe. Am J Epidemiol 164: 367-375.
http: / /dx. doi. or g/10.10 9 3/ai e/kwi 208
Shebl. FM: Bhatia. K: Engels. EA. (2010). Salivary gland and nasopharyngeal cancers in individuals
with acquired immunodeficiency syndrome in United States [Letter], Int J Cancer 126:
2503-2508. http://dx.doi.org/10.1002/iic.24930
Shen. H. ua: Mchale. CM: Haider. SI: Tung. C: Zhang. S: Smith. MT: Zhang. L. (2016). Identification of
Genes That Modulate Susceptibility to Formaldehyde and Imatinib by Functional Genomic
Screening in Human Haploid KBM7 Cells. Toxicol Sci 151: 10-22.
http: / /dx. doi. or g/10.109 3 /toxsci /kfwO 3 2
Shenolikar. R: Song. X: Anderson. TA: Chu. BC: Cantrell. CR. (2011). Costs of asthma among US
working adults. American Journal of Managed Care 17: 409-416.
Sheppard. D: Eschenbacher. W: Epstein. 1. (1984). Lack of bronchomotor response to up to 3 ppm
formaldehyde in subjects with asthma. Environ Res 35: 133-139.
http://dx.doi.org/10.1016/0013-935ir84190120-8
This document is a draft for review purposes only and does not constitute Agency policy.
R-47 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Sheveleva. G. (1971). Study of the specific effect of formaldehyde on the embryogenesis and
progeny of white rats. Toksikol Nov Prom Khim Veshchestv 12: 78-86.
Shi. YO: Chen. X: Dai. T: Tiang. ZF: Li. N: Zhang. BY: Zhang. ZB. (2014). Selenium pretreatment
attenuates formaldehyde-induced genotoxicity in A549 cell lines. Toxicol Ind Health 30:
901-909. http://dx.doi.org/10.1177/0748233712466129
Shin. YS: Takeda. K: Gelfand. EW. (2009). Understanding asthma using animal models. Allergy
Asthma Immunol Res 1: 10-18. http://dx.doi.Org/10.4168/aair.2009.l.l.10
Shurin. MR: Lu. L: Kalinski. P: Stewart-Akers. AM: Lotze. MT. (1999). Thl/Th2 balance in cancer,
transplantation and pregnancy [Review], 21: 339-359.
Shusterman. D. (2007). Trigeminally-mediated health effects of air pollutants: Sources of inter-
individual variability [Review], Hum Exp Toxicol 26: 149-157.
http://dx.doi.Org/10.1177/0960327107070550
Siemiatycki. 1: Wacholder. S: Richardson. L: Dewar. R: Gerin. M. (1987). Discovering carcinogens in
the occupational environment: Methods of data collection and analysis of a large case-
referent monitoring system. Scand J Work Environ Health 13: 486-492.
Siew. SS: Kauppinen. T: Kyvronen. P: Heikkila. P: Pukkala. E. (2012). Occupational exposure to wood
dust and formaldehyde and risk of nasal, nasopharyngeal, and lung cancer among Finnish
men. Cancer Management and Research 4: 223-232.
http://dx.doi.org/10.2147/CMAR.S30684
Sin. DP: Wu. LL: Man. SFP. (2005). The relationship between reduced lung function and
cardiovascular mortality: a population-based study and a systematic review of the literature
[Review], Chest 127: 1952-1959. http://dx.doi.Org/10.1378/chestl27.6.1952
Singh. I: Raizada. RM: Chaturvedi. VN: Tain. SK. (1998). Nasal mucous ciliary clearance and olfaction
in atrophic rhinitis. 50: 57-59. http://dx.doi.org/10.1007/BF02996772
Slikker. W. Tr: Andersen. ME: Bogdanffv. MS: Bus. IS: Cohen. SD: Conollv. RB: David. RM: Doerrer.
NG: Dorman. DC: Gavlor. DW: Hattis. D: Rogers. TM: Setzer. RW: Swenberg. TA: Wallace. K.
(2004). Dose-dependent transitions in mechanisms of toxicity: Case studies [Review],
Toxicol Appl Pharmacol 201: 226-294. http://dx.doi.Org/10.1016/i.taap.2004.06.027
Smedie. G: Norback. D. (2001). Incidence of asthma diagnosis and self-reported allergy in relation to
the school environment: A four-year follow-up study in schoolchildren. Int J Tuberc Lung
Dis 5: 1059-1066.
Smedie. G: Norback. D: Edling. C. (1997). Asthma among secondary schoolchildren in relation to the
school environment Clin Exp Allergy 27: 1270-1278. http://dx.doi.org/10.1046/i. 1365-
2222.1997.1780977.x
Smerhovskv. Z: Landa. K: Rossner. P: Brabec. M: Zudova. Z: Hola. N: Pokorna. Z: Mareckova. I:
Hurvchova. D. (2001). Risk of cancer in an occupationally exposed cohort with increased
level of chromosomal aberrations. Environ Health Perspect 109: 41-45.
http://dx.doi.org/10.2307/3434919
Smerhovskv. Z: Landa. K: Rossner. P: Tuzova. D: Brabec. M: Zudova. Z: Hola. N: Zarska. H:
Nevsimalova. E. (2002). Increased risk of cancer in radon-exposed miners with elevated
frequency of chromosomal aberrations. Mutat Res Genet Toxicol Environ Mutagen 514:
165-176. http://dx.doi.org/10.1016/S1383-5718r01100328-X
Smith. MT: Guvton. KZ: Gibbons. CF: Fritz. TM: Portier. CI: Rusvn. I: DeMarini. DM: Caldwell. TC:
Kavlock. RT: Lambert. PF: Hecht. SS: Bucher. TR: Stewart. BW: Baan. RA: Cogliano. VI: Straif.
K (2016). Key characteristics of carcinogens as a basis for organizing data on mechanisms
of carcinogenesis [Review], Environ Health Perspect 124: 713-721.
http://dx.doi.org/10.1289/ehp.1509912
Smith. SM: Le Beau. MM: Huo. D: Karrison. T: Sobecks. RM: Anastasi. 1: Vardiman. TW: Rowley. ID:
Larson. RA. (2003). Clinical-cytogenetic associations in 306 patients with therapy-related
This document is a draft for review purposes only and does not constitute Agency policy.
R-48 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
myelodysplasia and myeloid leukemia: the University of Chicago series. Blood 102: 43-52.
http://dx.doi.org/10.1182/blood-2002-ll-3343
Smith. SM: Vale. WW. (2006). The role of the hypothalamic-pituitary-adrenal axis in
neuroendocrine responses to stress. 8: 383-395.
Snedecor. GW: Cochran. WG. (1980). Statistical methods (7th ed.). Ames, IA: Iowa State University
Press.
Solet. D: Zoloth. SR: Sullivan. C: Tewett. 1: Michaels. DM. (1989). Patterns of mortality in pulp and
paper workers. J Occup Med 31: 627-630.
Songur. A: Akpolat. N: Kus. I: Ozen. OA: Zararsiz. I: Sarsilmaz. M. (2003). The effects of the inhaled
formaldehyde during the early postnatal period in the hippocampus of rats: A
morphological and immunohistochemical study. Neurosci Res Commun 33: 168-178.
http: / /dx. doi. or g/10.10 0 2 /nrc. 10093
Songur. A: Sarsilmaz. M: Ozen. OA. (2008). The effects of inhaled formaldehyde on oxidant and
antioxidant systems of rat cerebellum during the postnatal development process. Toxicol
Mech Meth 18: 569-574. http://dx.doi.org/10.1080/15376510701555288
Sonnenschein. C: Soto. A. naM. (2013). The aging of the 2000 and 2011 Hallmarks of Cancer
reviews: A critique. J Biosci 38: 651-663. http://dx.doi.org/10.1007/sl2038-Q13-9335-6
Sorg. BA: Bailie. TM: Tschirgi. ML: Li. N: Wu. WR. (2001a). Exposure to repeated low-level
formaldehyde in rats increases basal corticosterone levels and enhances the corticosterone
response to subsequent formaldehyde. Brain Res 898: 314-320.
http://dx.doi.Org/10.1016/S0006-8993r01102208-9
Sorg. BA: Hochstatter. T. (1999). Behavioral sensitization after repeated formaldehyde exposure in
rats. Toxicol Ind Health 15: 346-355. http://dx.doi.org/10.1177/074823379901500309
Sorg. BA: Swindell. S: Tschirgi. ML. (2004). Repeated low level formaldehyde exposure produces
enhanced fear conditioning to odor in male, but not female, rats. Brain Res 1008: 11-19.
http://dx.doi.Org/10.1016/j.brainres.2004.02.015
Sorg. BA: Tschirgi. ML: Swindell. S: Chen. L: Fang. 1. (2001b). Repeated formaldehyde effects in an
animal model for multiple chemical sensitivity [Review], Ann N Y Acad Sci 933: 57-67.
http://dx.doi.Org/10.llll/i.1749-6632.2001.tb05814.x
Sorg. BA: Willis. TR: Nowatka. TC: Ulibarri. C: See. RE: Westberg. HH. (1996). Proposed animal
neurosensitization model for multiple chemical sensitivity in studies with formalin.
Toxicology 111: 135-145. http://dx.doi.org/l0.1016/0300-483xr96103371 -9
Sorg. BA: Willis. TR: See. RE: Hopkins. B: Westberg. HH. (1998). Repeated low-level formaldehyde
exposure produces cross-sensitization to cocaine: Possible relevance to chemical sensitivity
in humans. Neuropsychopharmacology 18: 385-394.
http://dx.doi.org/10.1038/si.npp.139516Q
Sorlie. PD: Kannel. WB: O'Connor. G. (1989). Mortality associated with respiratory function and
symptoms in advanced age: the Framingham study. Am J Respir Crit Care Med 140: 379-
384. http ://dx. doi.org/10.1164/airccm/140.2.379
Speit. G: Schmid. 0: Frohler-Keller. M: Lang. I: G. T. (2007). Assessment of local genotoxic effects of
formaldehyde in humans measured by the micronucleus test with exfoliated buccal mucosa
cells. MutatRes GenetToxicol Environ Mutagen 627: 129-135.
http://dx.doi.Org/10.1016/i.mrgentox.2006.10.013
Speit. G: Schiitz. P: Weber. I: Ma-Hock. L: Kaufmann. W: Gelbke. HP: Durrer. S. (2011). Analysis of
micronuclei, histopathological changes and cell proliferation in nasal epithelium cells of rats
after exposure to formaldehyde by inhalation. MutatRes 721: 127-135.
http://dx.doi.Org/10.1016/i.mrgentox.2011.01.008
Speit. G: Zeller. 1: Schmid. 0: Elhaiouii. A: Ma-Hock. L: Neuss. S. (2009). Inhalation of formaldehyde
does not induce systemic genotoxic effects in rats. Mutat Res Genet Toxicol Environ
Mutagen 677: 76-85. http://dx.doi.Org/10.1016/i.mrgentox.2009.05.020
This document is a draft for review purposes only and does not constitute Agency policy.
R-49 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Staab. CA: Hellgren. M: Hoog. TO. (2008). Dual functions of alcohol dehydrogenase 3: Implications
with focus on formaldehyde dehydrogenase and S-nitrosoglutathione reductase activities
[Review], Cell Mol Life Sci 65: 3950-3960. http: //dx.doi.org/10.1007/s00018-008-8592-2
Starr. TB: Swenberg. TA. (2013). A novel bottom-up approach to bounding low-dose human cancer
risks from chemical exposures. Regul Toxicol Pharmacol 65: 311-315.
http://dx.doi.Org/10.1016/i.yrtph.2013.01.004
Starr. TB: Swenberg. TA. (2016). The bottom-up approach to bounding potential low-dose cancer
risks from formaldehyde: An update. Regul Toxicol Pharmacol 77: 167-174.
http://dx.doi.Org/10.1016/i.yrtph.2016.01.021
Stavner. L: Smith. AB: Reeve. G: Blade. L: Elliott. L: Keenlvside. R: Halperin. W. (1985).
Proportionate mortality study of workers in the garment industry exposed to
formaldehyde. Am J Ind Med 7: 229-240.
Stavner. LT: Elliott. L: Blade. L: Keenlvside. R: Halperin. W. (1988). A retrospective cohort mortality
study of workers exposed to formaldehyde in the garment industry. Am J Ind Med 13: 667-
681. http://dx.doi.Org/10.1002/ajim.4700130606
Steele. LL: Wilkins. I. R. (1996). Occupational exposures and risks of spontaneous abortion among
female veterinarians. Int J Occup Environ Health 2: 26-36.
Stellman. SD: Demers. PA: Colin. D: Boffetta. P. (1998). Cancer mortality and wood dust exposure
among participants in the American Cancer Society Cancer Prevention Study-II (CPS-II). Am
J Ind Med 34: 229-237. http://dx.doi.org/10.1002/fSICniQ97-
0274fl99809"134:3<229::ATD-AITM4>3.0.CO:2-O
Stewart. PA: Blair. A: Cubit. DA: Bales. RE: Kaplan. SA: Ward. T: Gaffev. W: O'Berg. MT: Walrath. T.
fl986). Estimating historical exposures to formaldehyde in a retrospective mortality study.
Appllnd Hygl: 34-41.
Stewart. PA: Cubit. D: Blair. A. (1987). Formaldehyde levels in seven industries. Appl Ind Hyg 2:
231-236.
Stewart. PA: Herrick. RF: Feiglev. CE: Utterback. DF: Hornung. R: Mahar. H: Hayes. R: Douthit. DE:
Blair. A. (1992). Study design for assessing exposures of embalmers for a case-control study.
Part I. Monitoring results. Appl Occup Environ Hyg 7: 532-540.
Stroup. NE: Blair. A: Erikson. GE. (1986). Brain cancer and other causes of death in anatomists. J
Natl Cancer Inst 77: 1217-1224.
Subramaniam. RP: Chen. C: Crump. KS: Devonev. D: Fox. IF: Portier. CI: Schlosser. PM: Thompson.
CM: White. P. (2008). Uncertainties in biologically-based modeling of formaldehyde-induced
respiratory cancer risk: Identification of key issues. Risk Anal 28: 907-923.
http://dx.doi.Org/10.llll/i.1539-6924.2008.01083.x
Subramaniam. RP: Crump. KS: Van Landingham. C: White. P: Chen. C: Schlosser. PM. (2007).
Uncertainties in the CUT model for formaldehyde-induced carcinogenicity in the rat: A
limited sensitivity analysis-I. Risk Anal 27: 1237-1254. http://dx.doi.Org/10.llll/i. 1539-
6924.2007.00968.x
Subramaniam. RP: Richardson. RB: Morgan. KT: Kimbell. IS: Guilmette. RA. (1998). Computational
fluid dynamics simulations of inspiratory airflow in the human nose and nasopharynx. Inhal
Toxicol 10: 91-120. http://dx.doi.org/10.1080/Q89583798197772
Sul. D: Kim. H: Oh. E: Phark. S: Cho. E: Choi. S: Kang. HS: Kim. EM: Hwang. KW: Tung. WW. (2007).
Gene expression profiling in lung tissues from rats exposed to formaldehyde. Arch Toxicol
81: 589-597. http://dx.doi.org/10.1007/s00204-007-0182-9
Summers. RM: Louie. T: Yu. C: Gakhar. L: Louie. KC: Subramanian. M. (2012). Novel, Highly Specific
N-Demethylases Enable Bacteria To Live on Caffeine and Related Purine Alkaloids. J
Bacteriol 194: 2041-2049. http://dx.d0i.0rg/l0.1128/TB.06637-11
This document is a draft for review purposes only and does not constitute Agency policy.
R-50 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Suruda. A: Schulte. P: Boeniger. M: Hayes. RB: Livingston. GK: Steenland. K: Stewart. P: Herrick. R:
Douthit. D: Fingerhut. MA. (1993). Cytogenetic effects of formaldehyde exposure in students
of mortuary science. Cancer Epidemiol Biomarkers Prev 2: 453-460.
Swenberg. 1: Kerns. W: Pavkov. K: Mitchell. R: Gralla. El. (1980a). Carcinogenicity of formaldehyde
vapor: interim findings in a long-term bioassay of rats and mice. Dev Toxicol Environ Sci 8:
283-286.
Swenberg. TA: Gross. EA: Randall. HW. (1986). Localization and quantitation of cell proliferation
following exposure to nasal irritants. In CS Barrow (Ed.), Toxicology of the nasal passages
(pp. 291-300). New York, NY: Hemisphere Publishing Corp.
Swenberg. TA: Gross. EA: Randall. HW: Barrow. CS. (1983). The effect of formaldehyde exposure on
cytotoxicity and cell proliferation. In JJ Clary; JE Gibson; RS Waritz (Eds.), Formaldehyde,
toxicology, epidemiology, mechanisms (pp. 225-236). New York, NY: Marcel Dekker.
Swenberg. TA: Kerns. WD: Mitchell. RI: Gralla. EI: Pavkov. KL. (1980b). Induction of squamous cell
carcinomas of the rat nasal cavity by inhalation exposure to formaldehyde vapor. Cancer
Res 40: 3398-3402.
Swenberg. TA: Lu. K: Moeller. BC: Gao. L: Upton. PB: Nakamura. I: Starr. TB. (2011). Endogenous
versus exogenous DNA adducts: Their role in carcinogenesis, epidemiology, and risk
assessment [Review], Toxicol Sci 120: S130-S145. http://dx.doi.org/10.1093/toxsci/kfq371
Swenberg. TA: Moeller. BC: Lu. K: Rager. IE: Fry. RC: Starr. TB. (2013). Formaldehyde carcinogenicity
research: 30 years and counting for mode of action, epidemiology, and cancer risk
assessment [Review], Toxicol Pathol 41: 181-189.
http://dx.doi. org/10.1177/0192623312466459
Swiecichowski. AL: Long. KT: Miller. ML: Leikauf. GD. (1993). Formaldehyde-induced airway
hyperreactivity in vivo and ex vivo in guinea pigs. Environ Res 61: 185-199.
http://dx.doi.org/10.1006/enrs.1993.1063
Takahashi. S: Tsuji. K: Fujii. K: Okazaki. F: Takigawa. T: Ohtsuka. A: Iwatsuki. K. (2007). Prospective
study of clinical symptoms and skin test reactions in medical students exposed to
formaldehyde gas. J Dermatol 34: 283-289. http://dx.doi.Org/10.l111/i.1346-
8138.2007.00274.x
Takavama. S: Reed. TC: Homma. S. (2003). Heat-shock proteins as regulators of apoptosis [Review],
Oncogene 22: 9041-9047. http://dx.doi.org/10.1038/si.onc. 1207114
Takigawa. T: Usami. M: Yamasaki. Y: Wang. B: Sakano. N: Horike. T: Kataoka. H: Ohtsuka. A: Kira. S.
(2005). Reduction of indoor formaldehyde concentrations and subjective symptoms in a
gross anatomy laboratory. Bull Environ Contam Toxicol 74: 1027-1033.
http://dx.doi.Org/10.1007/s00128-005-0683-2
Talibov. M: Lehtinen-lacks. S: Martinsen. II: Kiaerheim. K: Lvnge. E: Sparen. P: Trvggvadottir. L:
Weiderpass. E: Kauppinen. T: Kyvronen. P: Pukkala. E. (2014). Occupational exposure to
solvents and acute myeloid leukemia: A population-based, case-control study in four Nordic
countries. Scand J Work Environ Health 40: 511-517.
http://dx.doi.org/10.5271/siweh.3436
Tan. T: Zhang. Y: Luo. W: Lv. 1: Han. C: Hamlin. TNR: Luo. H: Li. H: Wan. Y: Yang. X: Song. W: Tong. Z.
(2018). Formaldehyde induces diabetes-associated cognitive impairments. FASEB J 32:
3669-3679. http://dx.d0i.0rg/l0.1096/fi.201701239R
Tang. LX: Zhang. YS. (2003). [Health investigation on workers exposed to formaldehyde] (pp. 34-
35). Tang, LX; Zhang, YS.
Tarkowski. M: Gorski. P. (1995). Increased IgE antiovalbumin level in mice exposed to
formaldehyde. Int Arch Allergy Immunol 106: 422-424.
http://dx.d0i.0rg/l 0.1159/000236876
Taskinen. H: Kyvronen. P: Hemminki. K. (1994). Laboratory work and pregnancy outcome. J Occup
Med 36: 311-319. http://dx.doi.Org/10.1097/00043764-199403000-00008
This document is a draft for review purposes only and does not constitute Agency policy.
R-51 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
Taskinen. HK: Kvyronen. P: Sallmen. M: Virtanen. SV: Liukkonen. TA: Huida. 0: Lindbohm. ML:
Anttila. A. (1999). Reduced fertility among female wood workers exposed to formaldehyde.
Am J Ind Med 36: 206-212. http://dx.doi.org/10.1002/fsicillQ97-
0274fl 99907136:1 <206::aid-aiim29>3.0.co:2-d
Tavernier. G: Fletcher. G: Gee. I: Watson. A: Blacklock. G: Francis. H: Fletcher. A: Frank. T: Frank. P:
Pickering. CA: Niven. R. (2006). IPEADAM study: Indoor endotoxin exposure, family status,
and some housing characteristics in English children. J Allergy Clin Immunol 117: 656-662.
http://dx.doi.Org/10.1016/i.iaci.2005.12.1311
ten Berge. WF: Zwart. A: Appelman. LM. (1986). Concentration-time mortality response
relationship of irritant and systemically acting vapours and gases. J Hazard Mater 13: 301-
309. http://dx.doi.org/10.1016/0304-3894(86185003-8
Teng. S: Beard. K: Pourahmad. I: Moridani. M: Easson. E: Poon. R: O'Brien. PI. (2001). The
formaldehyde metabolic detoxification enzyme systems and molecular cytotoxic
mechanism in isolated rat hepatocytes. Chem Biol Interact 130-132: 285-296.
http://dx.doi.Org/10.1016/S0009-2797r00100272-6
Tepper. RS: Wise. RS: Covar. R: Irvin. CG: Kercsmar. CM: Kraft. M: Liu. MC: O'Connor. GT: Peters. SP:
Sorkness. R: Togias. A. (2012). Asthma outcomes: Pulmonary physiology [Review], J Allergy
Clin Immunol 129: S65-S87. http://dx.doi.Org/10.1016/i.iaci.2011.12.986
Teschke. K: Morgan. MS: Checkowav. H: Franklin. G: Spinelli. IT: van Belle. G: Weiss. NS. (1997).
Surveillance of nasal and bladder cancer to locate sources of exposure to occupational
carcinogens. Occup Environ Med 54: 443-451. http://dx.doi.Org/10.1136/oem.54.6.443
Thompson. CM: Ceder. R: Grafstrom. RC. (2010). Formaldehyde dehydrogenase: beyond phase I
metabolism. Toxicol Lett 193: 1-3. http://dx.doi.Org/10.1016/i.toxlet2009.ll.023
Thompson. CM: Sonawane. B: Grafstrom. RC. (2009). The ontogeny, distribution, and regulation of
alcohol dehydrogenase 3: Implications for pulmonary physiology [Review], Drug Metab
Dispos 37: 1565-1571. http://dx.d0i.0rg/l 0.1124/dmd.l 09.027904
Thrasher. ID: Broughton. A: Madison. R. (1990). Immune activation and autoantibodies in humans
with long-term inhalation exposure to formaldehyde. Arch Environ Health 45: 217-223.
http://dx.doi.Org/10.1080/00039B96.1990.9940805
Thrasher. TP: Wojdani. A: Cheung. G: Heuser. G. (1987). Evidence for formaldehyde antibodies and
altered cellular immunity in subjects exposed to formaldehyde in mobile homes. Arch
Environ Health 42: 347-350. http://dx.doi.org/10.10B0/00039896.19B7.9934357
Til. HP: Woutersen. RA: Feron. VI: Hollanders. VHM: Falker. HE: Clary. II. (1989). Two-year
drinking-water study of formaldehyde in rats. Food Chem Toxicol 27: 77-87.
http://dx.doi.Org/10.1016/0278-6915f89190001-X
Titenko-Holland. N: Levine. AT: Smith. MT: Ouintana. PI: Boeniger. M: Hayes. R: Suruda. A: Schulte. P.
(1996). Quantification of epithelial cell micronuclei by fluorescence in situ hybridization
(FISH) in mortuary science students exposed to formaldehyde. Mutat Res 371: 237-248.
http://dx.d0i.0rg/l 0.1016/S01 65-1218C96190112-3
Tobe. M: Kaneko. Y: Uchida. Y: al.. e. (1985). Studies of the inhalation toxicity of formaldehyde.
National Sanitary and Medical Laboratory Service, Toxicity, Tokyo, Japan; T-85-0236. Tobe,
M; Kaneko, Y; Uchida, Y; et al.
Togias. A. (1999). Mechanisms of nose-lung interaction [Review], Allergy 54: 94-105.
http://dx.d0i.0rg/l 0.1111 /i.l 398-9995.1999.tb04410.x
Togias. A. (2004). Systemic effects of local allergic disease [Review], J Allergy Clin Immunol 113: S8-
14. http://dx.doi.Org/10.1016/i.iaci.2003.09.051
Tomasetti. C: Vogelstein. B. (2015). Variation in cancer risk among tissues can be explained by the
number of stem cell divisions. Science 347: 78-81.
http://dx.doi.org/10.1126/science.1260825
This document is a draft for review purposes only and does not constitute Agency policy.
R-52 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Tomashefski. I. (2008). Dail and Hammar's pulmonary pathology: Neoplastic lung disease. In J
Tomashefski (Ed.), (3rd ed., pp. 165-182). New York, NY: Springer.
Tong. ZM: Zhu. SX: Shi. 1. (2007). [Effect of formaldehyde on blood component and blood
biochemistry of exposed workers], 20: 409-410.
Toriussen. W: Solberg. LA: Hogetveit. AC. (1979). Histopathological changes of the nasal mucosa in
active and retired nickel workers. Br J Cancer 40: 568-580.
Tsigonia. A: Lagoudi. A: Chandrinou. S: Linos. A: Evlogias. N: Alexopoulos. EC. (2010). Indoor air in
beauty salons and occupational health exposure of cosmetologists to chemical substances.
Int J Environ Res Public Health 7: 314-324. http://dx.doi.org/10.3390/iierph7010314
Tsubone. H: Kawata. M. (1991). Stimulation to the trigeminal afferent nerve of the nose by
formaldehyde, acrolein, and acetaldehyde gases. Inhal Toxicol 3: 211-222.
http://dx.doi.org/10.3109/08958379109145285
Tsukahara. S: Yamamoto. S: Shwe. TTW: Ahmed. S: Kunugita. N: Arashidani. K: Fuiimaki. H. (2006).
Inhalation of low-level formaldehyde increases the Bcl-2/Bax expression ratio in the
hippocampus of immunologically sensitized mice. Neuroimmunomodulation 13: 63-68.
http://dx.doi.Org/10.1159/000094829
Tunc. 0: Tremellen. K. (2009). Oxidative DNA damage impairs global sperm DNA methylation in
infertile men. J Assist Reprod Genet 26: 537-544. http://dx.doi.org/10.1007/slQ815-009-
9346-2
Turner. TH: Reh. DP. (2012). Incidence and survival in patients with sinonasal cancer: A historical
analysis of population-based data. Head Neck 34: 877-885.
http://dx.doi.org/10.1002/hed.21830
Tvihak. E: Bocsi. 1: Timar. F: Racz. G: Szende. B. (2001). Formaldehyde promotes and inhibits the
proliferation of cultured tumour and endothelial cells. Cell Prolif 34: 135-141.
http://dx.doi. org/10.1046/i.l 365-2184.2001.00206.x
U.S. EPA (U.S. Environmental Protection Agency). (1989). Lymphatic and hematopoietic tissue
cancer in a chemcial MFG environment and a mortality study of men assigned to ethylene
oxide production or other related chemical MFG w-letter 081789 [TSCA Submission],
(EPA/OTS; Doc #89-890000225). Danbury, CT: Union Carbide Corporation.
https://ntrl.ntis.gOv/NTRL/dashboard/searchResults/titleDetail/OTS05134143.xhtml
U.S. EPA (U.S. Environmental Protection Agency). (1991). Guidelines for developmental toxicity risk
assessment. Fed Reg 56: 63798-63826.
U.S. EPA (U.S. Environmental Protection Agency). (1994). Methods for derivation of inhalation
reference concentrations and application of inhalation dosimetry [EPA Report],
(EPA600890066F). Research Triangle Park, NC.
https://cfpub.epa. gov/ncea/risk/recordisplay.cfm?deid=71993&CFID=51174829&CFTOKE
N=25006317
U.S. EPA (U.S. Environmental Protection Agency). (1996). Guidelines for reproductive toxicity risk
assessment. Fed Reg 61: 56274-56322.
U.S. EPA (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicity risk
assessment [EPA Report] (pp. 1-89). (ISSN 0097-6326
EISSN 2167-2520
EPA/630/R-95/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment
Forum, http://www.epa.gov/risk/guidelines-neurotoxicity-risk-assessment
U.S. EPA (U.S. Environmental Protection Agency). (2002). A review of the reference dose and
reference concentration processes. (EPA630P02002F). Washington, DC.
https://www.epa.gov/sites/production/files/2014-12/documents/rfd-final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2003). Summary of the toxicological review of
acrolein [EPA Report], Washington, DC.
https://cfpub.epa.gov/ncea/iris2/chemicalLanding.cfm7substance nmbr=364
This document is a draft for review purposes only and does not constitute Agency policy.
R-53 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Toxicological Review of Formaldehyde—Inhalation
U.S. EPA (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk
assessment [EPA Report], (EPA630P03001F). Washington, DC.
https://www.epa.gov/sites/production/files/2013-
09/documents/cancer guidelines final 3-25-05.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2005b). Supplemental guidance for assessing
susceptibility from early-life exposure to carcinogens [EPA Report], (EPA/630/R-03/003F).
Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
https://www.epa.gov/risk/supplemental-guidance-assessing-susceptibility-early-life-
exposure-carcinogens
U.S. EPA (U.S. Environmental Protection Agency). (2010). Toxicological Review of Formaldehyde
(Inhalation) (External Review Draft 2010).
http://cfpub.epa.gov/ncea/iris drafts/recordisplay.cfm?deid=2 23614
U.S. EPA (U.S. Environmental Protection Agency). (2012). Benchmark dose technical guidance [EPA
Report], (EPA100R12001). Washington, DC: U.S. Environmental Protection Agency, Risk
Assessment Forum, https: //www.epa.gov/risk/benchmark-dose-technical-guidance
U.S. EPA (U.S. Environmental Protection Agency). (2013). Toxicological review of Methanol
(Noncancer) (CASRN 67-56-1) in support of summary information on the Integrated Risk
Information System (IRIS) [EPA Report], (EPA/635/R-11-001F). Washington, DC.
U.S. EPA (U.S. Environmental Protection Agency). (2020). 0RD staff handbook for developing IRIS
assessments (public comment draft). (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/recordisplav.cfm?deid=350086
Uba. G: Pachorek. D: Bernstein. I: Garabrant. DH: Balmes. TR: Wright. WE: Amar. RB. (1989).
Prospective study of respiratory effects of formaldehyde among healthy and asthmatic
medical students. Am J Ind Med 15: 91-101. http://dx.doi.org/10.1002/ajim.4700150110
Usanmaz. SE: Akarsu. ES: Vural. N. (2002). Neurotoxic effects of acute and subacute formaldehyde
exposures in mice. Environ Toxicol Pharmacol 11: 93-100.
http://dx.doi. org/10.1016/SI 382-6689C01100109-0
Vaissiere. T: Sawan. C: Herceg. Z. (2008). Epigenetic interplay between histone modifications and
DNA methylation in gene silencing [Review], MutatRes 659: 40-48.
http://dx.doi.Org/10.1016/i.mrrev.2008.02.004
Valencia. K: Martin-Fernandez. M: Zandueta. C: Ormazabal. C: Martinez-Canarias. S: Bandres. E: de
la Piedra. C: Lecanda. F. (2013). miR-326 associates with biochemical markers of bone
turnover in lung cancer bone metastasis. Bone 52: 532-539.
http://dx.doi.Org/10.1016/i.bone.2012.10.033
Vandenplas. 0: Fievez. P: Delwiche. TP: Boulanger. 1: Thimpont. I. (2004). Persistent asthma
following accidental exposure to formaldehyde. Allergy 59: 115-116.
http://dx.doi. org/10.1046/i.l 398-9995.2003.00340.X
Vaughan. TL. (1989). Occupation and squamous cell cancers of the pharynx and sinonasal cavity.
Am J Ind Med 16: 493-510. http://dx.doi.org/10.1002 /aiim.4700160503
Vaughan. TL. (1996). Agents causing other respiratory cancers. In Occupational and environmental
respiratory disease. St. Louis, MO: Mosby-Year Book, Inc.
Vaughan. TL: Davis. S. (1991). Wood dust exposure and squamous cell cancers of the upper
respiratory tract Am J Epidemiol 133: 560-564.
http://dx.doi.org/10.1093/oxfordiournals.aie.all5927
Vaughan. TL: Stewart. PA: Teschke. K: Lynch. CF: Swanson. GM: Lyon. TL: Berwick. M. (2000).
Occupational exposure to formaldehyde and wood dust and nasopharyngeal carcinoma.
Occup Environ Med 57: 376-384. http://dx.doi.Org/10.1136/oem.57.6.376
This document is a draft for review purposes only and does not constitute Agency policy.
R-54 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Vaughan. TL: Strader. C: Davis. S: Paling. TR. (1986a). Formaldehyde and cancers of the pharynx,
sinus and nasal cavity: I. Occupational exposures. Int J Cancer 38: 677-683.
http://dx.doi.Org/10.1002/iic.2910380510
Vaughan. TL: Strader. C: Davis. S: Paling. TR. (1986b). Formaldehyde and cancers of the pharynx,
sinus and nasal cavity: II. Residential exposures. Int J Cancer 38: 685-688.
http://dx.doi.Org/10.1002/iic.2910380511
Venn. A. (2012). [Email to Glinda Cooper concerning FW: Follow-up question regarding 2003
wheeze study]. Available online
Venn. AT: Cooper. M: Antoniak. M: Laughlin. C: Britton. I: Lewis. SA. (2003). Effects of volatile
organic compounds, damp, and other environmental exposures in the home on wheezing
illness in children. Thorax 58: 955-960. http://dx.doi.org/10.1136/thorax.58.ll.955
Veres. TZ: Rochlitzer. S: Braun. A. (2009). The role of neuro-immune cross-talk in the regulation of
inflammation and remodelling in asthma [Review], Pharmacol Ther 122: 203-214.
http://dx.doi.Org/10.1016/i.pharmthera.2009.02.007
Viegas. S: Ladeira. C: Nunes. C: Malta-Vacas. T: Gomes. M: Brito. M: Mendonca. P: Prista. I. (2010).
Genotoxic effects in occupational exposure to formaldehyde: A study in anatomy and
pathology laboratories and formaldehyde-resins production. J Occup Med Toxicol 5: 25.
http://dx.doi. org/10.1186/1745-6673-5-25
Vlaanderen. I: Lan. 0: Kromhout. H: Rothman. N: Vermeulen. R. (2011). Occupational benzene
exposure and the risk of lymphoma subtypes: a meta-analysis of cohort studies
incorporating three study quality dimensions [Review], Environ Health Perspect 119: 159-
167. http://dx.doi.org/10.1289/ehp. 1002318
Vokes. EE: Weichselbaum. RR: Lippman. SM: Hong. WK. (1993). Head and neck cancer [Review], N
Engl J Med 328: 184-194. http://dx.doi.org/10.1056/NETM199301213280306
Vosoughi. S: Khavanin. A: Salehnia. M: Asilian Mahabadi. H: Shahverdi. A: Esmaeili. V. (2013).
Adverse effects of formaldehyde vapor on mouse sperm parameters and testicular tissue.
Int J Fertility Sterility 6: 250-257.
Vosoughi. S: Khavanin. A: Salehnia. M: Mahabadi. HA: Soleimanian. A. (2012). Effects of
simultaneous exposure to formaldehyde vapor and noise on mouse testicular tissue and
sperm parameters. Health Scope 1: 110-117. http: //dx.doi.org/10.17795 /jhealthscope-
7973
Wallner. P: Kundi. M: Moshammer. H: Piegler. K: Hohenblum. P: Scharf. S: Frohlich. M: Damberger.
B: Tapplere. P: Hutter. HP. (2012). Indoor air in schools and lung function of Austrian school
children. J Environ Monit 14: 1976-1982. http://dx.doi.org/10.1039/c2em30059a
Walrath. I: Fraumeni. IF. Tr. (1983). Mortality patterns among embalmers. Int J Cancer 31: 407-411.
http://dx.doi.Org/10.1002/iic.2910310403
Walrath. I: Fraumeni. TF. Tr. (1984). Cancer and other causes of death among embalmers. Cancer Res
44: 4638-4641.
Walsh. DA: Edwards. MI: Smith. MSR. (1997). Heat shock proteins and their role in early
mammalian development. Exp Mol Med 29: 139-150.
http://dx.doi.org/10.1038/emm.1997.21
Wang. H: Li. H. eC: Lv. M: Zhou. D: Bai. L: Du. L: Xue. X. ia: Lin. P. u: Oiu. S. (2015). Associations
between occupation exposure to Formaldehyde and semen quality, a primary study. Sci Rep
5: 15874. http://dx.doi.org/10.1038/srepl5874
Wang. H: O'Reilly. El: Weisskopf. MG: Logroscino. G: Mccullough. ML: Thun. Ml: Schatzkin. A:
Kolonel. LN: Ascherio. A. (2011). Smoking and risk of amyotrophic lateral sclerosis: a pooled
analysis of 5 prospective cohorts. Arch Neurol 68: 207-213.
http://dx.doi.org/10.1001/archneurol.201Q.367
This document is a draft for review purposes only and does not constitute Agency policy.
R-55 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Toxicological Review of Formaldehyde—Inhalation
Wang. HX: Wang. XY: Zhou. DX: Zheng. LR: Zhang. I: Huo. YW: Tian. H. (2013). Effects of low-dose,
long-term formaldehyde exposure on the structure and functions of the ovary in rats.
Toxicol Ind Health 29: 609-615. http://dx.doi.org/10.1177/0748233711430983
Wang. HX: Zhou. DX: Zheng. LR: Zhang. 1: Huo. YW: Tian. H: Han. SP: Zhang. 1: Zhao. WB. (2012).
Effects of paternal occupation exposure to formaldehyde on reproductive outcomes. J Occup
Environ Med 54: 518-524. http://dx.doi.org/10.1097/IOM.ObQl 3e31824e6937
Wang. K: Wang. TW: Xu. 1: Zhu. Y: Tian. L: Au. W: Xia. ZL. (2019). Determination of benchmark dose
based on adductand micronucleus formations in formaldehyde-exposed workers. Int J Hyg
Environ Health 222: 738-743. http://dx.doi.Org/10.1016/i.iiheh.2019.05.008
Wang. R: Zhang. Y: Lan. 0: Holford. TR: Leaderer. B: Zahm. SH: Boyle. P: Dosemeci. M: Rothman. N:
Zhu. Y: Oin. 0: Zheng. T. (2009). Occupational exposure to solvents and risk of non-Hodgkin
lymphoma in Connecticut women. Am J Epidemiol 169: 176-185.
http: / /dx. doi. or g/10.109 3 /ai e /kwn3 0 0
Wang. W: Yan. Y: Li. CW: Xia. HM: Chao. SS: Wang, d: Wang. ZP. (2014). Live human nasal epithelial
cells (hNECs) on chip for in vitro testing of gaseous formaldehyde toxicity via airway
delivery. Lab Chip 14: 677-680. http://dx.doi.org/10.1039/c31c51208h
Wantke. F: Demmer. CM: Tappler. P: Gotz. M: Tarisch. R. (1996a). Exposure to gaseous formaldehyde
induces IgE-mediated sensitization to formaldehyde in school-children. Clin Exp Allergy 26:
276-280. http://dx.doi.Org/10.llll /i.1365-2222.1996.tb00092.x
Wantke. F: Focke. M: Hemmer. W: Bracun. R: Wolf-Abdolvahab. S: Gotz. M: Tarisch. R: Gotz. M:
Tschabitscher. M: Gann. M: Tappler. P. (2000). Exposure to formaldehyde and phenol during
an anatomy dissecting course: Sensitizing potency of formaldehyde in medical students.
Allergy 55: 84-87. http://dx.doi.Org/10.1034/i.1398-9995.2000.00307.x
Wantke. F: Focke. M: Hemmer. W: Tschabitscher. M: Gann. M: Tappler. P: Gotz. M: Tarisch. R.
(1996b). Formaldehyde and phenol exposure during an anatomy dissection course: A
possible source of IgE-mediated sensitization. Allergy 51: 837-841.
http://dx.doi.Org/10.llll/i.1398-9995.1996.tb00031.x
Ward. IB. Tr: Hokanson. TA: Smith. ER: Chang. LW: Pereira. MA: Whorton. EB. Tr: Legator. MS. (1984).
Sperm count, morphology and fluorescent body frequency in autopsy service workers
exposed to formaldehyde. MutatRes Environ Mutagen Relat Subj 130: 417-424.
http://dx.doi.org/10.1016/0165-116ir84190014-l
Warner. IK: Wang. TC: Hope. KT: Tin. L: Dick. IE. (2004). Concepts of human leukemic development
[Review], Oncogene 23: 7164-7177. http://dx.doi.org/10.1038/si.one. 1207933
Weatherhead. S: Robson. SC: Reynolds. NT. (2007). Eczema in pregnancy [Review], BMJ 335: 152-
154. http://dx.doi.org/10.1136/bmi.39227.671227.AE
Weidinger. S: Novak. N. (2016a). Atopic dermatitis [Review], Lancet 387: 1109-1122.
http: //dx.doi.org/10.1016/S0140-6736C15100149-X
Weidinger. S: Novak. N. (2016b). Atopic dermatitis : Supplementary materials [Supplemental Data],
Lancet 387.
Weinberg. RA. (2014). Coming full circle-from endless complexity to simplicity and back again
[Review], Cell 157: 267-271. http://dx.doi.Org/10.1016/i.cell.2014.03.004
Weisel. CP: Zhang. T: Turpin. BT: Morandi. MT: Colome. S: Stock. TH: Spektor. DM: Korn. L: Winer.
AM: Kwon. T: Meng. OY: Zhang. L: Harrington. R: Liu. W: Reff. A: Lee. TH: ATimokhtari. S:
Mohan. K: Shendell. D: Tones. 1: Farrar. L: Maberti. S: Fan. T. (2005). Relationships of indoor,
outdoor, and personal air (RIOPA): Parti. Collection methods and descriptive analyses (pp.
1-107; discussion 109-127). (ISSN 1041-5505
EISSN 2688-6855
HEI Research Report 130-1, NUATRC Research Report 7). Boston, MA: Health Effects Institute.
This document is a draft for review purposes only and does not constitute Agency policy.
R-56 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Weisskopf. M: Morozova. N: O'Reilly. El: Mccullough. ML: Calle. EE: Thun. MI: Ascherio. A. (2009).
Prospective study of chemical exposures and amyotrophic lateral sclerosis mortality. J
Neurol Neurosurg Psychiatry 80: 558-561. http://dx.doi.org/10.1136/innp.2008.156976
Wells. PG: Winn. LM. (1996). Biochemical toxicology of chemical teratogenesis. Crit Rev Biochem
Mol Biol 31: 1-40. http://dx.doi.org/10.3109/10409239609110574
Werner. A: Meinhardt. A: Seitz. 1: Bergmann. M. (1997). Distribution of heat-shock protein 60
immunoreactivity in testes of infertile men. Cell Tissue Res 288: 539-544.
http://dx.doi.Org/10.1007/s004410050839
West. GB: Brown. TH. (2005). The origin of allometric scaling laws in biology from genomes to
ecosystems: Towards a quantitative unifying theory of biological structure and organization
[Review], J Exp Biol 208: 1575-1592. http://dx.doi.org/10.1242/jeb.01589
West. S: Hildesheim. A: Dosemeci. M. (1993). Non-viral risk factors for nasopharyngeal carcinoma in
the Philippines: Results from a case-control study. Int J Cancer 55: 722-727.
http://dx.doi.Org/10.1002/iic.2910550504
WHO (World Health Organization). (1967). Manual of the international statistical classification of
diseases, injuries, and causes of death. Geneva, Switzerland.
WHO (World Health Organization). (1977). Manual of the international statistical classification of
diseases, injuries, and causes of death. Geneva, Switzerland.
WHO (World Health Organization). (1987a). Air quality guidelines for Europe (1st ed.).
Copenhagen, Denmark: World Health Organization, Regional Office for Europe.
WHO (World Health Organization). (1987b). Formaldehyde. In Air quality guidelines for Europe.
Copenhagen, Denmark: World Health Organization, Regional Office for Europe.
WHO (World Health Organization). (1989). Environmental health criteria 89: Formaldehyde.
(RISKLINE/1990090019). http://www.inchem.org/documents/ehc/ehc/ehc89.htm
WHO (World Health Organization). (2010). Guidelines for indoor air quality. Selected pollutants.
Geneva, http://www.euro.who.int/ data/assets/pdf file/0009/128169 Ze94535.pdf
Widdicombe. TG. (1998). Afferent receptors in the airways and cough [Review], Respir Physiol 114:
5-15.
Wilcox. AT. (2010). Fertility and pregnancy: An epidemiologic perspective. In Fertility and
pregnancy: An epidemiologic perspective. New York, NY: Oxford University Press.
Wilcox. AT: Hornev. LF. (1984). Accuracy of spontaneous abortion recall. Am J Epidemiol 120: 727-
733. http://dx.doi.org/10.1093/oxfordiournals.aie.all3940
Wild. C: Brennan. P: Plummer. M: Bray. F: Straif. K: Zavadil. I. (2015). Cancer risk: Role of chance
overstated [Letter], Science 347: 728-728. http://dx.d0i.0rg/l0.1126/science.aaa6799
Wilmer. TWG. M: Woutersen. RA: Appelman. LM: Leeman. WR: Feron. VI. (1987). Subacute (4-week)
inhalation toxicity study of formaldehyde in male rats: 8-hour intermittent versus 8-hour
continuous exposures. J Appl Toxicol 7: 15-16. http://dx.doi.org/10.1002 /iat2550070104
Wilmer. TWG. M: Woutersen. RA: Appelman. LM: Leeman. WR: Feron. VI. (1989). Subchronic (13-
week) inhalation toxicity study of formaldehyde in male rats: 8-hour intermittent versus 8-
hour continuous exposures. Toxicol Lett 47: 287-293. http://dx.doi.org/10.1016/Q378-
4274f89190147-l
Wilson. A: Laurenti. E: Trumpp. A. (2009). Balancing dormant and self-renewing hematopoietic
stem cells [Review], 19: 461-468. http://dx.doi.Org/10.1016/i.gde.2009.08.005
Wisniewski. TR: Zougman. A: Mann. M. (2008). N-epsilon-Formylation of lysine is a widespread
post-translational modification of nuclear proteins occurring at residues involved in
regulation of chromatin function. Nucleic Acids Res 36: 570-577.
http://dx.doi.org/10.1093/nar/gkml057
Witek. TT. Tr: Schachter. EN: Tosun. T: Beck. GT: Leaderer. BP. (1987). An evaluation of respiratory
effects following exposure to 2.0 ppm formaldehyde in asthmatics: Lung function,
symptoms, and airway reactivity. Arch Environ Health 42: 230-237.
This document is a draft for review purposes only and does not constitute Agency policy.
R-57 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Witek. TT. Tr: Schachter. EN: Tosun. T: Leaderer. BP: Beck. GT. (1986). Controlled human studies on
the pulmonary effects of indoor air pollution: Experiences with sulfur dioxide and
formaldehyde. Environ Int 12: 129-135. http://dx.doi.org/10.1016/0160-4120r86190023-l
Wodarz. D: Zauber. AG. (2015). Cancer: Risk factors and random chances [Comment], Nature 517:
563-564. http://dx.doi.org/10.1038/517563a
Wolf. DC: Gross. EA: Lvght. 0: Bermudez. E: Recio. L: Morgan. KT. (1995a). Immunohistochemical
localization of p53, PCNA, and TGF-alpha proteins in formaldehyde-induced rat nasal
squamous cell carcinomas. Toxicol Appl Pharmacol 132:27-35.
http://dx.doi.org/10.1006/taap.1995.1083
Wolf. DC: Morgan. KT: Gross. EA: Barrow. C: Moss. OR: Tames. RA: Popp. TA. (1995b). 2-YEAR
INHALATION EXPOSURE OF FEMALE AND MALE B6C3F1 MICE AND F344 RATS TO
CHLORINE GAS INDUCES LESIONS CONFINED TO THE NOSE. Fundam Appl Toxicol 24: 111-
131.
Wolffe. TAM: Vidler. 1: Halsall. C: Hunt. N: Whalev. P. (2020). A survey of systematic evidence
mapping practice and the case for knowledge graphs in environmental health & toxicology.
Toxicol Sci. http : //dx.doi.org/10.1093 /toxsci /kfaaO 2 5
Wong. V: Cash. HL: Morse. TL: Lu. S: Zhitkovich. A. (2012). S-phase sensing of DNA-protein
crosslinks triggers TopBPl-independent ATR activation and p53-mediated cell death by
formaldehyde. Cell Cycle 11: 2526-2537. http://dx.doi.org/10.4161/cc.20905
Woolf. CI: Salter. MW. (2000). Neuronal plasticity: increasing the gain in pain [Review], Science 288:
1765-1769. http://dx.doi.org/10.1126/science.288.5472.1765
Wortlev. P: Vaughan. TL: Davis. S: Morgan. MS: Thomas. DB. (1992). A case-control study of
occupational risk factors for laryngeal cancer. Br J Ind Med 49: 837-844.
http://dx.doi.org/10.1136/oem.49.12.837
Woutersen. RA: Appelman. LM: Van Garderen-Hoetmer. A: Feron. VI. (1986). Inhalation toxicity of
acetaldehyde in rats. III. Carcinogenicity study. Toxicology 41: 213-231.
http://dx.doi.org/10.1016/0300-483Xr86190201-5
Woutersen. RA: Appelman. LM: Wilmer. TWG. M: Falke. HE: Feron. VI. (1987). Subchronic (13-week)
inhalation toxicity study of formaldehyde in rats. J Appl Toxicol 7: 43-49.
http://dx.doi.Org/10.1002/iat2550070108
Woutersen. RA: van Garderen-Hoetmer. A: Bruiinties. TP: Zwart. A: Feron. VI. (1989). Nasal tumours
in rats after severe injury to the nasal mucosa and prolonged exposure to 10 ppm
formaldehyde. J Appl Toxicol 9: 39-46. http://dx.doi.org/10.1002 /iat2550090108
Wu. H: Romieu. I: Seinra-Monge. I: del Rio-Navarro. BE: Anderson. DM: Tenchura. CA: Li. H: Ramirez-
Aguilar. M: Lara-Sanchez. I: London. ST. (2007). Genetic variation in S-nitrosoglutathione
reductase (GSNOR) and childhood asthma. J Allergy Clin Immunol 120: 322-328.
http://dx.doi.Org/10.1016/i.iaci.2007.04.022
Wu. Y: You. H: Ma. P: Li. L: Yuan. Y: Li. 1: Liu. X: Yao. H: Chen. R: Lai. K: Yang. X. (2013). Role of
transient receptor potential ion channels and evoked levels of neuropeptides in a
formaldehyde-induced model of asthma in Balb/c mice. PLoS ONE 8: e62827.
http://dx.doi.org/10.1371/journal.pone.0062827
Xing Sv. LY. (2007). Toxic effect of formaldehyde on reproduction and heredity in male mice.
Journal of Jilin University - Medicine Edition 33.
Xing. SY: Ye. L: Wang. NN. (2007a). Toxic effect of formaldehyde on reproduction and heredity in
male mice. Journal of Jilin University - Medicine Edition 33: 716-718.
Xing. SY: Ye. L: Wang. NN. (2007b). [Toxic effect of formaldehyde on reproduction and heredity in
male mice]. Journal of Jilin University - Medicine Edition 33: 716-718.
Xiong. GX: Zhan. TM: Tiang. HY: Li. IF: Rong. LW: Xu. G. (2008). Computational fluid dynamics
simulation of airflow in the normal nasal cavity and paranasal sinuses [Case Report], Am J
Rhinol 22: 477-482. http://dx.doi.org/10.2500/air.2008.22.3211
This document is a draft for review purposes only and does not constitute Agency policy.
R-58 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Yamamoto. Y: Shioda. N: Han. F: Moriguchi. S: Fukunaga. K. f2010al [Donepezil-induced
neuroprotection of acetyl cholinergic neurons in olfactory bulbectomized mice] [Review],
Yakugaku Zasshi 130: 717-721.
Yamamoto. Y: Uchima. T: Konoike. Y: Nakamine. H. (2010b). Myeloid sarcoma in the nasal cavities
that developed during the course of acute myelomonocytic leukemia [Letter], J Clin Exp
Hematop 50: 167-170. http://dx.doi.org/10.3960/islrt.50.167
Yan. Y: Ye. Z: Lu. ZS: Qiao. Y: Yang. X: Li. CM. (2005). Nitric oxide level associated with gaseous
formaldehyde exposure in lungs of mice. In X Yang; B Zhao; RZhao (Eds.), Indoor Air 2005:
Proceedings of the 10th International Conference on Indoor Air Quality and Climate, vol 5
(pp. 3851-3854). Beijing, China: Tsinghua University Press.
https://www.isiaq.org/docs/PDFs/3851.pdf
Yang. W: Su. Z: Xu. Z: Yang. W: Peng. Y. ue: Li. I. (2020). Comparative study of alpha-, beta-, gamma-
and delta-Mn02 on toluene oxidation: Oxygen vacancies and reaction intermediates. Appl
Catal B-Environ 260. http://dx.doi.Org/10.1016/i.apcatb.2019.118150
Yang. WH. (2007). [Hemogram of workers exposed to low concentration of formaldehyde] (pp. 792-
799). Yang, WH.
Yang. X: Zhang. YP: Chen. D: Chen. WG: Wang. R. (2001). Eye irritation caused by formaldehyde as
an indoor air pollution--a controlled human exposure experiment Biomed Environ Sci 14:
229-236.
Yang. XR: Diehl. S: Pfeiffer. R: Chen. CI: Hsu. WL: Dosemeci. M: Cheng. YT: Sun. B: Goldstein. AM:
Hildesheim. A: Team. CaAGEoNS. (2005). Evaluation of risk factors for nasopharyngeal
carcinoma in high-risk nasopharyngeal carcinoma families in Taiwan. Cancer Epidemiol
Biomarkers Prev 14: 900-905. http://dx.doi.org/10.1158/1055-9965.EPI-04-0680
Ye. X: Ti. Z: Wei. C: Mchale. C: Ding. S: Thomas. R: Yang. X: Zhang. L. (2013a). Inhaled formaldehyde
induces DNA-protein crosslinks and oxidative stress in the bone marrow and other distant
organs of exposed mice [Abstract], Environ Mol Mutagen 54: S41.
Ye. X: Ti. Z: Wei. C: Mchale. CM: Ding. S: Thomas. R: Yang. X: Zhang. L. (2013b). Inhaled formaldehyde
induces DNA-protein crosslinks and oxidative stress in bone marrow and other distant
organs of exposed mice. Environ Mol Mutagen 54: 705-718.
http: / /dx. doi. or g/10.10 0 2 /em. 21821
Ye. X: Yan. W: Xie. H: Zhao. M: Ying. C. (2005). Cytogenetic analysis of nasal mucosa cells and
lymphocytes from high-level long-term formaldehyde exposed workers and low-level short-
term exposed waiters. MutatRes 588: 22-27.
http://dx.doi.Org/10.1016/j.mrgentox.2005.08.005
Yeatts. KB: El-Sadig. M: Leith. D: Kalsbeek. W: Al-Maskari. F: Couper. D: Funk. WE: Zoubeidi. T: Chan.
RL: Trent. CB: Davidson. CA: Boundv. MG: Kassab. MM: Hasan. MY: Rusvn. I: Gibson. TM:
Olshan. AF. (2012). Indoor air pollutants and health in the United Arab Emirates. Environ
Health Perspect 120: 687-694. http://dx.doi.org/10.1289/ehp.1104090
Yendamuri. S: Calin. GA. (2009). The role of microRNA in human leukemia: A review [Review],
Leukemia 23: 1257-1263. http://dx.doi.org/10.1038/leu.2008.382
Ying. CI: Yan. WS: Zhao. MY: Ye. XL: Xie. H: Yin. SY: Zhu. XS. (1997). Micronuclei in nasal mucosa,
oral mucosa and lymphocytes in students exposed to formaldehyde vapor in anatomy class.
Biomed Environ Sci 10: 451-455.
Ying. CI: Ye. XL: Xie. H: Yan. WS: Zhao. MY: Xia. T: Yin. SY. (1999). Lymphocyte subsets and sister-
chromatid exchanges in the students exposed to formaldehyde vapor. Biomed Environ Sci
12: 88-94.
Yon. DK: Hwang. S: Lee. SW: Tee. HM: Sheen. YH: Kim. TH: Lim. DH: Han. MY. (2019). Indoor
Exposure and Sensitization to Formaldehyde among Inner-City Children with Increased
Risk for Asthma and Rhinitis. Am J Respir Crit Care Med 200: 388-393.
http://dx.doi.org/10.1164/rccm.201810-1980LE
This document is a draft for review purposes only and does not constitute Agency policy.
R-59 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Toxicological Review of Formaldehyde—Inhalation
Yonemitsu. T: Kuroki. C: Takahashi. N: Mori. Y: Kanmura. Y: Kashiwadani. H: Ootsuka. Y: Kuwaki. T.
(2013). TRPA1 detects environmental chemicals and induces avoidance behavior and
arousal from sleep. Sci Rep 3: 3100. http://dx.doi.org/10.1038/srepQ3100
Yoshida. I: Ibuki. Y. (2014). Formaldehyde-induced histone H3 phosphorylation via JNK and the
expression of proto-oncogenes. MutatRes 770: 9-18.
http://dx.doi.Org/10.1016/i.mrfmmm.2014.09.003
Young. IT. (1981). Histopathologic examination of the rat nasal cavity. Fundam Appl Toxicol 1: 309-
312. http://dx.doi.org/10.1016/S0272-0590r81180037-l
Young. RP: Hopkins. R: Eaton. TE. (2007). Forced expiratory volume in one second: not just a lung
function test but a marker of premature death from all causes [Review], Eur Respir J 30:
616-622. http://dx.d0i.0rg/l 0.1183 /09031936.00021707
Yu. GY: Song. XF: Liu. Y: Sun. ZW. (2014). Inhaled Formaldehyde Induces Bone Marrow Toxicity via
Oxidative Stress in Exposed Mice. Asian Pac J Cancer Prev 15: 5253-5257.
http://dx.doi.org/10.7314/APTCP.2014.15.13.5253
Yu. IT: Li. AM: Goggins. W: Leung. TO: Chan. GY: Fung. CK: Chan. CK: Lau. AP. (2017). Association of
wheeze during the first 18 months of life with indoor nitrogen dioxide, formaldehyde, and
family history of asthma: a prospective cohort study. Hong Kong Med J 23 Suppl 2: 19-23.
Yu. ITS: Chin. YL: Wong. TW: Tang. TL. (2004). Deaths from nasopharyngeal cancer among waiters
and waitresses in Chinese restaurants. Int Arch Occup Environ Health 77: 499-504.
http://dx.doi.Org/10.1007/s00420-004-0543-0
Yu. MC: Lai. SH: Henderson. BE. (1986). Cantonese-style salted fish as a cause of nasopharyngeal
carcinoma: Report of a case-control study in Hong Kong. Cancer Res 42: 956-961.
Yu. R: Lai. Y: Hartwell. HI: Moeller. BC: Dovle-Eisele. M: Kracko. D: Bodnar. WM: Starr. TB:
Swenberg. TA. (2015a). Formation, Accumulation, and Hydrolysis of Endogenous and
Exogenous Formaldehyde-Induced DNA Damage. Toxicol Sci 146: 170-182.
http: / /dx. doi. or g/10.109 3 /toxsci /kfvO 7 9
Yu. R: Lai. Y: Hartwell. HI: Moeller. BC: Dovle-Eisele. M: Kracko. D: Bodnar. WM: Starr. TB:
Swenberg. TA. (2015b). Formation, Accumulation, and Hydrolysis of Endogenous and
Exogenous Formaldehyde-Induced DNA Damage - supplemental data [Supplemental Data],
Toxicol Sci 146.
Zeller. I: Neuss. S: Mueller. TU: Kiihner. S: Holzmann. K: Hogel. 1: Klingmann. C: Bruckner. T: Triebig.
G: Speit. G. (2011). Assessment of genotoxic effects and changes in gene expression in
humans exposed to formaldehyde by inhalation under controlled conditions. Mutagenesis
26: 555-561. http://dx.doi.org/10.1093/mutage/ger016
Zendehdel. R: Touni. FT: Haiipour. B: Paniali. Z: Kheiri. H: Vahabi. M. (2017). DNA damage in workers
exposed to formaldehyde at concentrations below occupational exposure limits. Toxicol
Environ Chem 99: 1409-1417. http://dx.d0i.0rg/l 0.1080/02772248.2017.1343335
Zhai. L: Zhao. 1: Xu. B: Deng. Y: Xu. Z. (2013). Influence of indoor formaldehyde pollution on
respiratory system health in the urban area of Shenyang, China. Afr Health Sci 13: 137-143.
http: / /dx. doi. or g/10.4314 /ahs. vl 3 il. 19
Zhang. B: Shi. Y: Chen. X. in: Dai. 1: Tiang. ZF. a: Li. N: Zhang. Z. (2013a). Protective effect of curcumin
against formaldehyde-induced genotoxicity in A549 Cell Lines. J Appl Toxicol 33: 1468-
1473. http: / /dx. doi. or g/10.10 0 2 /i at 2 814
Zhang. L: Lan. 0: Guo. W: Hubbard. AE: Li. G: Rappaport. SM: Mchale. CM: Shen. M: Ti. Z: Vermeulen.
R: Yin. S: Rothman. N: Smith. MT. (2011). Chromosome-wide aneuploidy study (CWAS) in
workers exposed to an established leukemogen, benzene. Carcinogenesis 32: 605-612.
http://dx.doi.org/10.1093/carcin/bgq286
Zhang. L: Tang. X: Rothman. N: Vermeulen. R: Ti. Z: Shen. M: Oiu. C: Guo. W: Liu. S: Reiss. B: Freeman.
LB: Ge. Y: Hubbard. AE: Hua. M: Blair. A: Galvan. N: Ruan. X: Alter. BP: Xin. KX: Li. S: Moore.
LE: Kim. S: Xie. Y: Hayes. RB: Azuma. M: Hauptmann. M: Xiong. 1: Stewart. P: Li. L: Rappaport.
This document is a draft for review purposes only and does not constitute Agency policy.
R-60 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Toxicological Review of Formaldehyde—Inhalation
SM: Huang. H: Fraumeni. IF. Tr: Smith. MT: Lan. 0. (2010). Occupational exposure to
formaldehyde, hematotoxicity, and leukemia-specific chromosome changes in cultured
myeloid progenitor cells. Cancer Epidemiol Biomarkers Prev 19: 80-88.
http://dx.d0i.0rg/l 0.1158/1055-9965.EPT-09-0762
Zhang. 0: Yan. W: Bai. Y: Zhu. Y: Ma. 1. (2014). Repeated formaldehyde inhalation impaired olfactory
function and changed SNAP25 proteins in olfactory bulb. Int J Occup Environ Health 20:
308-312. http://dx.d0i.0rg/l0.1179/2049396714Y.0000000079
Zhang. Y: Liu. X: Mchale. C: Li. R: Zhang. L: Wu. Y: Ye. X: Yang. X: Ding. S. (2013b). Bone marrow
injury induced via oxidative stress in mice by inhalation exposure to formaldehyde. PLoS
ONE 8: e74974. http://dx.doi.org/10.1371/iournal.pone.0074974
Zhao. B: Nakada. N: Itai. S: Hanamoto. S: Okumura. K: Tanaka. H. (2020a). Diurnal patterns of N-
nitrosodimethylamine and formaldehyde behaviors in different seasons in surface water
influenced by effluent from sewage treatment plants. J Hazard Mater 383: 121155.
http ://dx. doi. or g/10.1016 /i .ihazmat 2019.121155
Zhao. Y. un: Magana. LC: Cui. H: Huang. 1: Mchale. CM: Yang. X. u: Loonev. MR: Li. R. ui: Zhang. L.
(2020b). Formaldehyde-induced hematopoietic stem and progenitor cell toxicity in mouse
lung and nose. Arch Toxicol 95: 693-701. http: //dx.doi.Org/10.1007/s00204-020-02932-x
Zheng. W: Blot. WT: Shu. XO: Diamond. EL: Gao. YT: Ti. BT: Fraumeni. 1. R. (1992). A population-based
case-control study of cancers of the nasal cavity and paranasal sinuses in Shanghai. Int J
Cancer 52: 557-561. http://dx.doi.org/10.1002/iic.2910520410
Zhong. W: Hee. SO. (2004). Quantitation of normal and formaldehyde-modified deoxynucleosides
by high-performance liquid chromatography/UV detection. Biomed Chromatogr 18: 462-
469. http://dx.doi.0rg/lO.lOO2/bmc.337
Zhou. D: Zhang. 1: Wang. H. (2011a). Assessment of the potential reproductive toxicity of long-term
exposure of adult male rats to low-dose formaldehyde. Toxicol Ind Health 27: 591-598.
http://dx.d0i.0rg/l 0.1177/0748233710393401
Zhou. D: Zhang. 1: Wang. H: Xue. Y. (2011b). Effect of formaldehyde exposure on structure and
function of epididymis in adult rats: A histological and biochemical study. Toxicol Environ
Chem 93: 134-144. http://dx.d0i.0rg/l0.1080/02772248.2010.501145
Zhou. DX: Oiu. SD: Zhang. 1: Tian. H: Wang. HX. (2006). The protective effect of vitamin E against
oxidative damage caused by formaldehyde in the testes of adult rats. Asian J Androl 8: 584-
588. http://dx.d0i.0rg/l 0.1111/i.l 745-7262.2006.00198.x
Zhou. ES: Kane. YY: Gao. XX: Wu. LF: Lu. ZS: Yan. Y: Qiao. Y: Yang. X. (2005). A pilot investigation on
human serum formaldehyde-specific IgE. Paper presented at 10th International Conference
on Indoor Air Quality and Climate, September 4-9, 2005, Beijing, China.
Zhu. TL: Knudsen. LE: Andersen. AM: Hiollund. NH: Olsen. 1. (2005). Time to pregnancy among
Danish laboratory technicians who were a part of the National Birth Cohort Scand J Work
Environ Health 31: 108-114.
Zhu. TL: Knudsen. LE: Andersen. AM: Hiollund. NH: Olsen. 1. (2006). Laboratory work and pregnancy
outcomes: a study within the National Birth Cohort in Denmark. Occup Environ Med 63: 53-
58. http://dx.doi.Org/10.1136/oem.2005.021204
Zwart. A: Woutersen. RA: Wilmer. TWG. M: Spit. BT: Feron. VI. (1988). Cytotoxic and adaptive effects
in rat nasal epithelium after 3-day and 13-week exposure to low concentrations of
formaldehyde vapour. Toxicology 51: 87-99. http://dx.doi.org/10.1016/030Q-
483Xf 88190083-2
This document is a draft for review purposes only and does not constitute Agency policy.
R-61 DRAFT-DO NOT CITE OR QUOTE
------- |