FINAL DRAFT
FORMALDEHYDE RISK ASSESSMENT UPDATE
June 11, 1991
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C.
-------
DISCLAIMER
This document is an external draft for review purposes only
and does not constitute policy. Mention of trade names or
commercial products does not constitute endorsement or
recommendation for use.
11
-------
PREFACE
This document was prepared by the U.S. EPA Office of Toxic
Substances with the assistance of the Office of Research and
Development, and it provides an updated assessment with data that
have become available since publication of "Assessment of Health
Risks to Garment Workers and Certain Home Residents from Exposure
to Formaldehyde" by EPA in 1987. Various approaches for
estimating cancer risk from formaldehyde exposure are discussed.
Thus, this document is not intended to replace the 1987
assessment, but its purpose is to provide an evaluation of recent
data and risk assessment approaches.
111
-------
Authors, Contributors, and Reviewers
The Office of Toxic Substances was responsible for preparing
this document -with contribution of the Office of Health and
Environmental Assessment.
>v*
Principal Authors
Gary Grindstaff, M.S.P.M.
Mary Henry, Ph.D.
Oscar Hernandez, Ph.D.
Karen Hogan, M.S.
David Lai, Ph.D., D.A.B.T.
Cheryl Siegel-Scott, M.S.P.H.
Contributing Authors
Lawrence Anderson, Ph.D.
Angela Auletta, Ph.D.
Elizabeth Margosches, Ph.D., M.P.H.
Bonnie Stern, Ph.D., M.P.H.
Lorenz Rhomberg, Ph.D. (OHEA/ORD)
Reviewers
The following individuals provided peer review of this
document and/or earlier drafts of this document.
Steven P. Bayard, Ph.D.
Human Health Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, D.C.
David L. Bayliss, M.S.
Human Health Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, D.C.
Murray Cohn, Ph.D.
Health Effects Division
Directorate for Health Sciences
Consumer Product Safety Commission
Bethesda, MD
Joseph A. Cotruvo, Ph.D.
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C.
IV
-------
Ernest V. Falke, Ph.D.
Health and Environmental Review Division
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C.
William Farlarid,
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C.
Richard Hill, M.D., Ph.D.
Science and Policy Analysis Staff
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C.
John F. Martonik
Directorate of Health Standards
Occupational Safety and Health Administration
Charles H. Ris, Ph.D.
Human Health Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, D.C.
Vanessa Vu, Ph.D.
Health and Environmental Review Division
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C.
-------
TABLE OF CONTENTS
Section Page
1.0 Executive Summary 1
2.0 Background 2
3.0 Metabolism and Pharmacokinetics 5
3.1 Review of New Studies 7
3.2 Summary and Conclusions 12
4.0 Carcinogenic Hazard 16
4.1 Animal Carcinogenicity Studies 16
4.1.1 Long-Term Bioassays 16
4.1.2 Tumor Promotion and
Co-Carcinogenicity Studies 21
4.1.3 Summary and Conclusions of
Animal Studies 22
4.2 Mutagenicity 23
4.3 Possible Mechanisms of Formaldehyde-Induced
Carcinogenicity 24
4.3.1 Role of DNA-protein Cross-linking 25
4.3.2 Role of Cytotoxicity and
Cell Proliferation 25
4.3.3 Role of the Protective Effect
of the Mucociliary Apparatus 31
4.3.4 Importance of Dose Available
for Deposition 32
4.3.5 Air Concentration vs. Total Dose 32
4.3.6 Summary and Conclusions 34
4.4 Epidemiologic Studies 35
4.4.1 Case-Control Studies 37
4.4.2 Proportional Mortality Studies 39
4.4.3 Cohort Studies 40
4.4.4 Evaluation of New Studies 45
4.4.5 Non-EPA Reviews of the
Formaldehyde Literature 46
4.4.6 Discussion and Conclusions 48
4.4.6.1 Exposure Consideration 52
4.4.6.2 Evaluation of the Body
of Human Evidence 53
4.5 Weight of Cancer Evidence 56
4.5.1 Human Evidence . . ' 56
4.5.2 Animal Evidence 57
4.5.3 Supporting Evidence , 58
4.5.4 Categorization of Overall
Cancer Evidence 58
VI
-------
5.0 Estimates of Cancer Risk 59
5.1 Review of Quantitative Assessment
by EPA (1987) 60
5.2 Issues in the Use of DPX as Delivered Dose 60
5.2_. 1 Use of DPX for High-to-Low-Dose
Extrapolation 62
5.2.2 Use of DPX Data for
Interspecies Extrapolation 65
5.2.3 Summary 68
5.3 Calculation of Risk Estimates Using DPX 69
5.3.1 Estimation of Delivered Dose 69
5.3.2 Lifetime Average Daily Exposure:
Dose-Rate Considerations 73
5.3.3 Quantitative Estimation of Risk • 74
5.4 Comparison with Other Formaldehyde
Risk Estimates 78
5.4.1 Comparison with L987 Unit Risk Estimate 78
5.4.2 Quantitative Assessment by Starr (1990) 78
5.5 Discussion of Uncertainties 80
5.5.1 Uncertainties Associated with
Use of DPX 80
5.5.2 Concordance with Epidemiologic Evidence 81
5.6 Summary 83
6.0 Noncarcinogenic Hazard Effects 83
6.1 Review of New Animal Studies 83
6.1.1 Respiratory Effects 83
6.1.2 Contact Sensitization 84
6.1.3 Pulmonary Sensitization 85
6.1.4 Immunotoxicologic Studies 85
6.1.5 Summary and Conclusions 87
6.2 Review of Noncancer Epidemiologic Studies 87
6.2.1 Eye and Upper Respiratory
Tract Effects 88
6.2.1.1 Irritant Symptoms 88
6.2.1.2 Nasal Cellular Changes 95
6.2.2 Lower Airway Effects 98
6.2.2.1 Subject-Reported Symptoms 98
6.2.2.2 Lung Function Measurements 102
6.2.3 Respiratory Effects in Asthmatics 1^7
6.2.4 Immunologic Effects .. 109
6.2.5 Central Nervous System Effects ill
6.2.6 Limitations 112
6.2.7 Summary and Conclusions 113
7.0 Estimates of Upper Respiratory and Eye *
Irritation Risks 115
7.1 Incorporation of New Data 116
7.2 Discussion of the Overall Evidence 123
vii
-------
8.0 Risk Characterization 124
8.1 Noncancer Effects 124
8.4-.X Sensory Irritation 124
8.1,. 2 Mucociliary Clearance Effects 125
8.1.3 Cellular Changes 126
8.2 Carcinogenic Effects 127
8.2.1 Studies of Humans 127
8.2.2 Studies in Animals 130
8.2.3 Additional Supportive Evidence 131
8.3 Quantitative Risk Assessment 131
8.3.1 Noncancer Risk Assessment 132
8.3.2 Cancer Risk Assessment 134
8.3.2.1 Dose-Response Assessment 134
8.3.2.2 Cell Proliferation and
Carcinogenesis 136
8.3.2.3 Discussion and Conclusions 137
9.0 References 143
Appendix A Summary of Epidemiologic Studies Cited
in 1987 and 1991 EPA Formaldehyde Risk
Assessments
Appendix B Comparison of Animal Model Risk Predictions
with Epidemiologic Evidence
Appendix C The Anatomy and Physiology of the Nasal Passages
of Humans, Monkeys, and Rats
Vlll
-------
TABLES TITLE PAGE
3-1 Concentration of DNA-protein cross-links in the
Respiratory tract and Bone Marrow (femur)
of Rhesus Monkeys Exposed to H CHO 11
4-1 Incidences of Nasal Squamous Cell Carcinomas in
Rats After Inhalation Exposure to Formaldehyde 19
4-2 Effect of Formaldehyde Exposure on Cell
Proliferation in the Central Portion of the
Nasal Passages 28
4 Sele ~ed Findings from Epidemiologic Studies
Exar ing Inhaled Formaldehyde 49
5-1 Mea^ .red Delivered-r ^e Levels in N,. .
-------
FIGURES TITLE PAGE
3-1 Phannacokinetic Model for Formaldehyde
Disposition in the Rat Nasal Mucosa to
Mejtabolites (Detoxification) or to DNA-protein
Cross-links 8
'% '\
i>*
3-2 Formation of DNA-protein Cross-links in the
Turbinates and Anterior Nose of F344 RaJ^s and
Rhesus Monkeys Exposed for 6 Hours to H CHO 13
4-1 Tumor Incidence and Cell Proliferation in Rats
Exposed to Formaldehyde 30
5-1 Steps in the Derivation of Lifetime Human
Risks Based on Rat Carcinogenicity and
Administered Dose, as described in 1987. 61
5-2 Steps Involved in the Derivation of Lifetime
Human Cancer Risks Based on Rat Carcinogenicity
Data, and the Use of DNA-Protein Cross-Linking
(DPX) as Dosimeter 70
5-3 Observed and Model-Predicted Concentration of
Formaldehyde Covalently Bound to DNA and Protein
in F344 Rats (Casanova et al., 1989) and Rhesus
Monkeys (Heck et al., 1989) 72
7-1 Health Effects by HCHO in Complaint
Mobile Homes (Ritchie and Lehnen, 1987) 118
7-2 Health Effects by HCHO in Complaint
Conventional Homes (Ritchie and Lehnen, 1987) 119
7-3 Predicted Prevalence of Burning Nose Over Range
of Formaldehyde Concentrations in ppm
(Horvath et al., 1988) 120
7-4 Observed Burning/Tearing Eye Irritation
Response Over a Range of HCHO Levels
(Liu et al., in press) 121
7-5 Observed Burning/Tearing Eye Irritation
Response Over a'Range of HCHO Levels
(Liu et al., Personal Communication) 122
-------
1.0 EXECUTIVE SUMMARY
The EPA has conducted a review of epidemiologic and
toxicological data developed since the release of EPA's
"Assessment of Health Risks to Garment Workers and Certain Home
Residents from' Expdsure to Formaldehyde" (EPA, 1987) , and has
done an evaluation of the potential human health risks due to
inhalation of formaldehyde vapor released from pressed wood
products in new mobile and conventional homes.
Sensory irritation of the mucus membranes of the eyes and
the respiratory tract, and cellular changes in the nasal cavity
are the principal noncancer effects of exposure to low airborne
concentration of formaldehyde. The latter effect may lead to the
secondary complication of respiratory disease. Although there is
inadequate information to quantitate the noncancer risks to the
general population from inhaled formaldehyde, fewer and less
severe responses are expected to be associated with less frequent
and less intense exposures. Further, little risk from noncancer
effects after exposure to formaldehyde is expected in situations
where exposures are ten-fold less than a no-observed effect
level, or one hundred-fold less than a lowest-observed effect
level in humans.
Precise thresholds for the irritant effects of formaldehyde
for the eyes and upper airway have not been firmly established.
However, a large number of observations from clinical and
nonclinical settings supports a conclusion that the range of
airborne formaldehyde concentrations over which most people
respond is 0.1 to 3.0 ppm. Recent studies in humans indicate
that symptoms of lower airway irritant effects may occur at about
1 ppm or higher, which are slightly lower than levels reported
previously (EPA, 1987). Limited data in humans and laboratory
animals indicate that low level exposure to formaldehyde may
impair nasal mucociliary flow. The inhibition of nasal mucus
flow occurred at 0.4 ppm in humans and in laboratory animals at 2
ppm of formaldehyde. Repeated and long-term exposures to
formaldehyde vapor may also induce histologic changes of the
nasal epithelium (e.g. cell degeneration, inflammation,
hyperplasia, squamous metaplasia, and rarely dysplasia). In
laboratory animals, the no observed effect level for these
cellular changes is about .1 ppm.
EPA has classified formaldehyde as a "Probable Human
Carcinogen" (Group Bl) in accordance to its Guidelines for
Carcinogenic Risk Assessment (1986). Available epidemiologic
studies support the conclusion made in 1987 that there 'is
"limited" evidence to indicate that formaldehyde may be a
carcinogen in humans. The human evidence for potential
carcinogenicity is based mainly on the association of
formaldehyde exposure with modest increases of cancers of the
nasopharynx, nasal cavity, and sinus. The evidence relied in a
-------
small degree on observations of increases in lung cancer.
However, no consistent patterns between formaldehyde exposure and
lung cancer are apparent across studies. Further, there is a
suggestion from recent analyses that combined exposures to
formaldehyde and other substances in resins and molding material
processes may be involved in the development of lung cancer.
The epidemiologic evidence is corroborated by experimental
findings showing that formaldehyde is genotoxic and carcinogenic
to rats and mice. Data developed since 1987 confirm the previous
findings that formaldehyde is a nasal carcinogen in rats at high
exposure concentrations and provide a better understanding of its
mechanism of carcinogenic action, though it is still not
completely understood. Evidence from recent inhalation studies
in rats indicates that formaldehyde concentration may be more
important than the accumulated dose (i.e., concentration x time)
in the induction of cellular effects, and suggests that sustained
tissue damage and cell proliferation associated with exposure to
high concentrations of formaldehyde are important in the
induction of nasal tumors. Major advances have also been made in
determining the delivered dose in respiratory tract tissues in
experimental animals.
In the absence of quantifiable human data, dose-response
data in rats at high concentrations were used to estimate low-
dose human cancer risk for formaldehyde. The cancer quantitative
risk assessment has been modified by using the rate of formation
of DNA-protein cross links as surrogates for delivered
concentrations in the target cells. EPA has calculated an upper
bound inhalation incremental unit risk for lifetime exposure
based on rat (2.8 x 10" per ppm) and monkey (3.3 x 10"4 per ppm)
dosimetry by using the linearized multistage procedure. Of the
risk estimate options, EPA favors that based on monkey DPX
because the monkey is considered a closer surrogate to humans
than the rat. The resultant 'risk estimates are about 50-fold
less than estimates based on airborne exposure concentrations
performed in the previous EPA assessment (EPA, 1987). The
largest contribution to the reduction in the cancer risk
estimates is the use of DNA-protein cross-links data as delivered
doses. Limitations and associated uncertainties of the cancer
risk estimates are discussed.
2.0 BACKGROUND
A previous review, "Assessment of Health Risks to Garment
Workers and Certain Home Residents from Exposure to Formaldehyde"
(EPA, 1987), examined the noncancer and cancer effects associated
with formaldehyde exposure. EPA conclusions on the noncancer
effects associated with exposure to formaldehyde (EPA, 1987)
were based mainly upon already-existing reviews by the National
Research Council (1981), Consensus Workshop on Formaldehyde
(1984), and Interagency Risk Management Council (IRMC, 1984).
-------
The major noncancer human health effects posed by inhalation
exposure to formaldehyde were due to the irritating nature of the
chemical (EPA, 1987). These effects were sensory irritation and
cellular changes. The evidence of cellular damage in humans,
although limited,'-was considered important. A number of lower
airway and pulmonary effects, including asthma, may occur with
formaldehyde exposure. Formaldehyde-induced cellular changes in
the nasal passages of rats, mice, hamsters, and monkeys were
reported from a number of studies (EPA, 1987) . These changes
ranged from rhinitis and epithelial hyperplasia and squamous
metaplasia to dysplasia, depending on the duration of exposure,
the concentration, and the species tested.
There is also considerable evidence, with inhalation
exposure, for the carcinogenicity of formaldehyde in animals.
This was based on the increased incidence of a rare malignant
tumor, nasal squamous cell carcinoma in two species (rats and
mice), and in both sexes of two rat strains (Fischer 344 and
Sprague-Dawley), in multiple inhalation experiments at high
concentrations.
EPA (1987) detailed the ability of formaldehyde to induce
gene mutations in several test systems including viruses,
bacteria, yeast, fungi, Drosophila, grasshoppers, and mammalian
cells in culture. The document further described the ability of
formaldehyde to induce cellular transformation in several
systems; its co-mutagenic ability when tested with agents such as
x-rays, ultraviolet light, and hydrogen peroxide; and its
increased activity in test systems which were deficient in
excision repair mechanisms. Formaldehyde was further
characterized as inducing both chromosome aberrations (CA) and
sister chromatid exchanges (SCE) in vitro as well as causing
single-strand breaks in DNA, inhibiting the resealing of single-
strand breaks produced by ionizing radiation and inducing cross-
links between DNA and protein, in both epithelial and
fibroblastic cells. It was proposed that formaldehyde may act
both by damaging DNA and by inhibiting DNA repair in treated
cells. A review of studies on the metabolic incorporation of
formaldehyde versus covalent binding to DNA in treated cells were
also included in the document. Human studies which were
concerned with the occurrences of CA and SCE in the blood, and
mutagenic products in the urine of exposed workers, were also
detailed.
In 1987, EPA also reviewed 28 epidemiologic studies and
concluded based upon EPA's Guidelines for Cancer Risk'Assessment
(EPA, 1986) that "limited" evidence existed for an association
between formaldehyde and human cancers. The "limited"
classification recognizes that a credible argument for a causal
association can be made, but that bias, chance, and confounding
factors cannot be ruled out. The evidence weighted most heavily
for an association with cancers of the upper respiratory tract
-------
(nasopharynx, nasal cavity and sinus, and buccal cavity). A
lesser portion of evidence suggested that excesses in lung and
brain cancers, and leukemia, may have been associated with
formaldehyde-exposure in some studies. However, the biological
explanation for canceTrs beyond the site of contact (e.g. brain
cancer and leukemia) remained unsupported.
Based on limited epidemiologic evidence for an association
of upper respiratory tract cancer with formaldehyde exposure and
sufficient animal evidence for an induction of nasal tumors in
formaldehyde-exposed animals along with supporting genotoxicity
evidence, the EPA classified formaldehyde as a probable human
carcinogen (Group Bl).
The epidemiologic studies were considered inadequate for
quantitative risk assessment. Therefore, the quantitative risk
assessment of formaldehyde reported in 1987 was based on the
Kerns et al. (1983) F344 rat bioassay, in which nasal squamous
cell carcinoma incidences were increased with increasing
formaldehyde levels in both males and females (see Table 4.1).
The accounting of the tumor incidence and of the number of
animals at risk was in accord with the recommendations of the
Interagency Risk Management Council (IRMC, 1984). A five stage
model was found to fit the data adequately. This process
resulted in an upper5limit inhalation incremental unit risk
estimate of 1.3 x 10 per ug/m , or 1.6 x 10 per ppm.
Casanova-Schmitz et al. (1984) proposed to use the formation
of DNA-protein cross-links (DPX) as a surrogate dose for risk
estimates based on animal inhalation studies. These authors
developed methodology that would allow differentiation between
metabolically incorporated and covalently bound formaldehyde
(DPX). Metabolic reactions of formaldehyde via the one-carbon
pathway in vivo involve the loss of hydrogen from formaldehyde
whereas chemical reactions do not. Casanova-Schmitz et al.
(1984) had used this fact to sort out the relative amounts of 3H-
and C-labeled formaldehyde which were metabolized or reacted
chemically when administered to rats by inhalation. They
concluded that formaldehyde produced DNA-protein cross-linking
(DPX) in the rat nasal mucosa at doses of 2 ppm and greater, and
that the extent of cross-linking increased with dose in a
nonlinear manner from 2 to 6 ppm. They believed these data
suggested a potential overestimate of risk at low levels of
exposure.
t>
In 1985, the Environmental Protection Agency (EPA), Consumer
Product Safety Commission (CPSC) and National Toxicology Program
(NTP) published their reservations to the Chemical Industry
Institute of Toxicology (CUT) interpretations of the
experimental data, and these were upheld by a panel of expert
scientists. Additionally, it was found that whereas H had
little effect on the rate of formaldehyde metabolism, H had a
major slowing effect. This isotope effect reduced the"accuracy
-------
of the[ H]/[ C] ratio as a quantitative means of distinguishing
bound formaldehyde from metabolically-incorporated formaldehyde;
this observation along with other critiques led CUT to develop
new methodology (Casanova et al., 1989) in order to support the
existence and' importance of DPX in the nasal mucosa of rats dosed
with formaldehyde. The new information was responsive to the
reservations and it has been utilized in this revised assessment.
3.0 METABOLISM AND PHARMACOKINETICS
Formaldehyde may undergo several reactions in vivo. These
reactions are characterized as either metabolic or chemical.
Metabolic reactions - those which require enzyme catalysis - may
be subdivided into the processes of degradation and incorpora-
tion. The biotransformations involved in these processes are
oxidation to formic acid and carbon dioxide; and incorporation
into biosynthetic pathways leading to proteins and nucleic acids.
The latter biotransformations utilize either formaldehyde or its
metabolite, formic acid. The oxidative step is mediated by
formaldehyde dehydrogenase, a glutathione (GSH)-dependent enzyme
(Votila and Koivusalo, 1983). The initial step in the oxidative
process is formation of S-hydroxymethyl GSH (GS-CH2OH) . Hydride
abstraction from this intermediate produces S-formyl GSH (GS-
CHO) . Enzymatic hydrolysis of the thioester, mediated by S-
formylglutathione hydrolase, produces formic acid with
concomitant regeneration of GSH.
Chemical reactions do not require enzyme participation.
They involve interactions of nucleophilic amine, sulfur or
hydroxyl groups with formaldehyde. The initial product of
reaction is the addition of nucleophile to the carbonyl group of
formaldehyde leading to formation of hydroxymethyl derivatives.
Amine, sulfur and oxygen nucleophiles form aminals (R2-N-CH2OH) ,
thioacetals (RS-CH2OH) , and hemiacetals (R-O-CH2OH) ,
respectively. These reactions are reversible, the equilibrium is
dependent on pH and the nature of the nucleophile. Hydroxymethyl
derivatives of primary amines or amides can react further with
either formaldehyde to form N-bis-hydroxymethyl derivatives, or
with a second molecule of nucleophile. The latter reaction is
particularly favorable with basic amines and produces structures
of the type RjN-CH^N in which two molecules of amine are linked
by a methylene bridge. This latter transformation is also
reversible although more drastic conditions (pH and temperature)
are required to hydrolyze the methylene-1inked products relative
to the corresponding hydroxymethyl derivatives.
Formaldehyde is known to react with amino acids, proteins,
nucleotides, RNA and DNA with predominant formation of N-
hydroxymethyl derivatives which react further to form methylene-
linked products. Cross-links are formed (intermolecularly)
between proteins, protein and DNA, and between DNA molecules.
DNA-DNA cross-links have been identified in pure DNA preparations
incubated with formaldehyde (Chaw et al., 1980), but-such direct
-------
cross-links have yet to be identified following treatment of
intact cells. Some evidence suggests that indirect DNA-DNA
cross-links may occur by the linkage of two DNA strands to the
same chromatirr ±0 forjn protein bridges between DNA strands
(Wilkins and McLeod," 1976) . The reactions of formaldehyde with
DNA are of mechanistic interest because of the potential
association of these transformations with genotoxic effects.
Cross-linking of DNA and protein by formaldehyde has been
observed in a variety of mammalian cells in culture following
treatment with formaldehyde (Bedford and Fox, 1981; Ross et al.,
1981; Craft et al., 1987). Inhalation exposure of rats to
formaldehyde resulted in decreased extractability of DNA from
proteins in the respiratory mucosa. This result was interpreted
as evidence that formaldehyde forms DNA-protein cross-links (DPX)
under in vivo conditions (Casanova-Schmitz and Heck, 1983) .
CUT recognized the potential implications of DPX for
quantitative risk assessment and sought to develop methodology
that would allow quantitation of protein-DNA cross-links.
Results of this effort were published in Casanova-Schmitz et al.
(1984). The proposed method was based on the inhalation exposure
of rats to a mixture of [ C]- and [ H]-formaldehyde as a way to
differentiate between metabolic incorporation and covalent
binding. Two observations were key to this approach: first,
cross-linking of DNA to proteins decreases its extractability - a
conventional extraction procedure separates DPX material from
non-cross-linked DNA; and second, oxidative metabolism of
formaldehyde involves hydride abstraction which would result in
loss of [ H] from formaldehyde; material covalently bound would
retain its original [ H] specific activity. Macromolecules
radiolabelled by metabolic incorporation of formaldehyde would
show a lower value for tfye ratio [ H]/[ C] than thor.e labelled by
covalent binding since [ C] is constant in either situation.
Thus, by measuring the ratio of [ H]/[ C] in radiolabeled
macromolecules it would be possible to establish the relative
amount of covalently bound formaldehyde and hence DPX. ^{IT used
this fact to sort out the relative amounts of [ H]- and [ C]-
labeled formaldehyde which were metabolized or reacted chemically
when administered to rats by inhalation. They concluded that
formaldehyde produced DPX in the rat respiratory mucosa at doses
of 2 ppm and greater, and-that the extent of cross-linking
increased with dose between 2 and 6 ppm.
CPSC, EPA and NTP published their reservations to the
interpretation of the experimental data by CUT (Cohn et al.,
1985). A panel of experts, convened with the express purpose of
evaluating the CUT study, concurred with the Agency's conclusion
that several shortcomings precluded the use of DPX as a measure
of intracellular dose in the derivation of risk estimates. The
comments and recommendations given by the expert group were
summarized in the Agency's risk assessment document (EPA, 1987)
as follows: First, the methodologies must be validated to assure
-------
that the experimental assumptions were scientifically sound and
that the formaldehyde-DNA-protein complexes were identified
properly; second, the single intracellular target used in the
study may be inadequate; and third, the use of an acute exposure
model in the- C-IIT study may not be appropriate because chronic,
not acute exp61sure',1*is most relevant to risk assessment.
Specific issues raised were: the lack of validation of the
extraction procedure (chloroform/iso-amyl alcohol/phenol) in
terms of efficiency and identity of the materials separated by
this procedure; the incursion of kinetic isotope effects in the
disposition of [ H]-formaldehyde which confound the use of the
proposed approach; the use of other covalent adducts such as
protein cross-links, and hydroxmethyl adducts with proteins, DNA,
or RNA, as dosimetric markers; and insufficient evidence to state
unequivocally that all the DPX occurs in the interfacial DNA
fraction.
3 .1 Review of New Studies
The relevance attached to the use of DPX as a measure of
intracellular concentration of formaldehyde is emphasized in
recent publications by CUT. An important finding, that verified
a concern expressed by the expert panel, was the demonstration of
a [ H] isotope effect in the formaldehyde dehydrogenase oxidation
of formaldehyde to formic acid (Heck and Casanova, 1987).
Oxidation of [ H]- and [ C]-formaldehyde in rat nasal mucosal
homogenates occurred with an isotope effect equal to 1.8 + O.ll
(ratio of V^/K,,, values for corresponding dehydrogenase reactions
with [H CHO] and [HCHO]). A small isotope effect in the
formation of cross-links was determined using rat hepatic nuclei;
binding of [ HCHQJ to DNA was slightly favored (3.4%) over
reaction with [H CHO]. The ne± result of an isotope effect
during the oxidation step is [ H] enrichment of intracellular
formaldehyde relative to extracellular or administered
formaldehyde. This event reduces the accuracy of the [ H/[UC]
ratio as a quantitative means of distinguishing bound from
metabolically incorporated formaldehyde. Casanova and Heck (1987)
proposed a pharmacokinetic model for formaldehyde disposition in
the nasal mucosa. As shown in Figure 3-1, the model consists of
three components: 1) influx of HCHO and H CHO into the cells of
the nasal mucosa with rate constants characteristic for each
isotope; 2) oxidative metabolism of intracellular formaldehyde
with rate constants for each isotope; and 3) elimination of
formaldehyde by routes other than oxidation including covalent
binding to macromolecules, diffusion out of the cells and any
metabolic transformations not involving cleavage of the H-carbon
bond. Based on this model, the discrepancy between iritracellular
and extracellular isotope ratios can be attributed to the isotope
effect during oxidation. The authors further assumed that the
isotope effect in vivo is equivalent to that determined in vitro
and that the isotope effect on covalent binding, because of its
small magnitude, can be neglected. The result of these
assumptions was to establish an upper limit, equal to the
measured isotope effect, for the increase in the intracellular
-------
Km
detoxification
t
NAD
D< K
\
D
DMA
DNA-protein
cross-links
Figure 3-1. PharrnacoWnetic model for disposition of inhaled HCHO in the rat
nasai mucosa to metaboites (detoxification) or to DNA-protein cross-links.
Abbreviations are as folows: A is extracellular (administered) HCHO and 0 is
intracelutar (delivered) HCHO consisting of free (Op and QSH-bound (Cy
forms. Rate and ecMbrium constants are defined in the text, (from Casanova
and Heck. 1987)
-------
isotope ratio relative to the isotope ratio of administered
formaldehyde. The concentration response curve for DPX following
6 hr exposure to HCHO an<| H CHO was normalized using this factor
(1.8+1.1) to account for H enrichment. The resulting curve
showed that^icfnoring the isotopic effect on oxidation would lead
to overestimates df the extent of DPX formation, particularly at
low levels of formaldehyde. These findings prompted CUT to
develop new methodology in order to support the existence and
potential importance of DPX in the nasal mucosa of rats dosed
with formaldehyde.
Earlier research at CUT was interpreted as indicating that
formaldehyde cross-linked DNA with protein in the rat nasal
mucosa and that the product (DPX) was present only in the
interfacial (IF) DNA fraction. Additional studies were performed
in an effort to establish these points.
In one study (Casanova et al., 1989), triplicate groups of
four male Fischer 344 rats^ were exposed in nose-only exposure
chambers to a mixture of C- and H-labeled formaldehyde (6 ppm;
6 hr) . Animals were not pre-exposed to formaldehyde as in
previous experiments. Immediately after exposure, the rats were
decapitated and their respiratory mucosal tissue was removed.
The tissue specimens were homogenized, treated with sodium
borohydride to reduce reversibly bound Schiff base adducts to
methylamines, dissolved in buffer, and sheared by passage through
a 20-gauge needle. The sheared solution was separated into
aqueous (AQ) DNA and interfacial (IF) DNA fractions with a
mixture of chloroform, isoamyl alcohol and phenol. The two DNA
fractions were purified and assayed separately. The reaction
products of formaldehyde and DNA in rat nasal mucosa were studied
by HPLC (high pressure liquid chromatography) analyses of
hydrolysates of AQ and IF fractions. Chromatography of
hydrolyzed AQ DNA samples showed that in the AQ DNA, essentially
all of the radioactivity was eluted with the normal
deoxyribonucleosides, dG, dT, and dA, indicating that the
radioactivity in the AQ DNA fraction was almost entirely due to
metabolic incorporation. In addition, [ H] (but not [ C]) was
eluted early in the chromatography profile; this early-eluting
[ H] was considered by the authors to be caused by isotopic
exchange of [ H]-H20 with the DNA.
In the IF DNA, radioactivity was also eluted with the normal
deoxyribonucleosides, dG, dT, and dA, implying that nucleosides
in the IF DNA were also labeled by metabolic incorporation.
However, unlike the AQ DNA, a substantial quantity of both [ H]
and [ C] was eluted before the deoxyribonucleosides. This early
eluting radioactivity was identified as formaldehyde by chemical
derivatization with dimedone. The observations that [ H]- and
[C]-formaldehyde were present only in the IF DNA and that the
[ H] activity of the early-eluting peak was higher than the [ C]
activity suggested^to the authors (Casanova et al., 1989) that
formaldehyde was bound as DPX.
-------
The difference between the hydrolysates of the two DNA
fractions was that only IF DNA hydrolysate contained
formaldehyde, meaning that only IF DNA was cross-linked. No
other radioactive organic compounds were detected in the DNA
hydrolysates-. - Specifically sought and not found were w -hydroxy-
methyl-deoxyadenosine* (hm6dA), N -hydroxy-methyl-deoxyouanosine
(hm2dG), N -hydroxy-methyl-deoxycytidine (hm4dC) and w-hydroxy-
raethyl-thymidine (hm3dt) which had been synthesized and
characterized by their UV spectra and migration during reversed-
phase HPLC.
A second experiment by Casanova et al. (1989) was designed
to determine the extent of DNA-formaldehyde covaient binding in
the rat respiratory mucosa at each of five exposure levels.
Triplicate groups of four rats each were exposed for six hours in
nose-only chambers to approximately 0.3, 0.7, 2, 6 and 10 ppm of
C-formaldehyde. The mucosal tissue was collected and treated
as in the previously described double-label study to prepare
solutions of sheared DNA. Then followed incubation with
proteinase K, solvent extraction, etc. leading to the release and
assay of C-formaldehyde by HPLC and dimedone derivatization.
It should be noted that this experimental procedure did not
separate IF DNA and AQ DNA, but collected total DNA.
No evidence of formation of N-hydroxymethyl derivatives was
obtained at any exposure level. The study did show that
formaIdehyde-DNA binding resulted from all exposure levels,
including 0.3 ppm (1.4 + 0.6 pmol/mg of DNA). The latter
airborne concentration is in the range of measured ambient
concentrations of formaldehyde in urban or indoor environments.
The results obtained by this procedure were about two-fold lower
than those obtained by the [ H]/[ C] ratio appr ,\ch (Casanova-
Schmitz et al., 1984). The difference was expl ined by the
authors as a result of exposure conditions (nai e vs pre-exposed
animals) and the isotope effect in the oxidation of HCHO which
leads to overestimation of the extent of covaient binding to DNA.
The shape of the concentration-response for DPX appeared to
remain nonlinear with respect to formaldehyde exposure
concentration.
Similar findings were reported for rhesus monkeys exposed in
a head-only inhalation chamber to H CHO (0.7, 2, or 6 ppm; 6
hr). Analysis of DPX was .carried out by HPLC as described for
the rat study (Heck et al., 1989). The mean concentration of
cross-links in the turbinates and anterior nose was calculated,
with weighting according to the amount of DNA in each tissue, to
allow comparison with the corresponding rat data. Other tissues
were examined for formation of DPX, and results are summarized in
Table 3-1.
The results in^Table 3-1 show predominant formation of DPX
in the turbinates and anterior nose areas with lower levels
detected in the nasopharynx and trachea. No cross links were
detected in the sinus, proximal lung or bone marrow. "Formation
10
-------
Table 3-1
Concentration of DNA-protein Cross-links in the Respiratory Tract
and Bone Marrow (femur) of Rhesus Monkeys Exposed to H CHO.
Tissue •* Concentration of DNA-cross-links (pmol/mg DMA).
is *
0 . 7 ppin 2 . Q PPm 6 . 0 PPTO
Turbinates, 0.36 ± 0.10* 2.56 + 0.31 18.2 + 3.4
anterior nose
Sinus ND ND ND
Nasopharynx (0.09± 0.09) (0.47 ± 0.09) 5.8 + 2.4
Larynx/ ND (0.4) 9.4 ±5.4
trachea/carina
Airways** ND (0.70 ± 0.15) 5.7 ± 5.7
Proximal lung ND ND ND
Bone marrow ND ND ND
* Mean + S.E.; ND = not detectable; values in parentheses are
uncertain, as dpm for these samples were less than twice the
background.
** Major intrapulmonary airways greater than 2 mm in diameter.
11
-------
of DPX in turbinates and anterior nasal mucosa in the monkey was
established at all exposure levels with concentrations of cross-
links being lower than in rats, by as much as an order of
magnitude (Casanova et al., 1989). Both studies (Casanova et
al., 1989; H«cX et ^al., 1989) reported that DPX was stable in
unhydrolyzed DNA but'was dissociated by enzymatic hydrolysis of
the DNA which resulted in release of radiolabelled formaldehyde.
Structural characterization of specific DPX components was made
difficult by the hydrolytic lability of such adducts.
Data for each species were fitted to a pharmacokinetic model
(Casanova and Heck, 1987), and the results are shown graphically
in Figure 3-2.
The results described above indicate that (a) the formation
of DNA-protein cross-links likely is a nonlinear function of the
exposure dose in both species and (b) monkeys had lower levels of
cross-links in corresponding tissues than did rats.
Heck et al. (1989) interpreted the preceding studies as
strongly suggesting that the intracellular concentration of
formaldehyde is modulated by two saturable defense mechanisms,
namely, covalent binding to protein in nasal mucosa and
metabolism by formaldehyde dehydrogenase. They added that the
repair of DPX cross-links may also be saturable.
3.2 Summary and Conclusions
The recent CUT work has clarified most of the concerns
expressed in the expert panel report outlined earlier in this
section. The major experimental design issues concerned the
validation of the extraction procedure for separation of AQ DNA,
IF DNA and proteins; possible existence of an isotope effect in
the reactions of labelled formaldehyde; and substantiation of the
claim that all the formaldehyde bound as DPX was present in the
IF DNA. These issues were addressed by development of a method
to determine the extent of formaldehyde binding to DNA that
utilizes only H CHO and thus is not subject to isotope effects
as in the previous H/ C approach. This method is based on the
release of labelled formaldehyde from DNA by enzymatic digestion.
HPLC is used to separate deoxynucleosides from formaldehyde
released by hydrolysis. The latter provides the basis for the
quantitation of DPX, and it was also used to demonstrate that
only IF DNA contained DPX since no formaldehyde was released by
hydrolysis of AQ DNA. This method also proved to be more
sensitive than the isotope ratio procedure allowing detection of
DPX at the lowest exposure concentration used (0.3 ppm) ,
significantly lower than the previously reported detection
threshold of 2 ppm (Casanova-Schmitz et al., 1984). Levels of
DPX were determined in the turbinates and anterior nose of the
monkey at all exposure concentrations tested (0.7, 2, and 6 ppm).
12
-------
50
45 -
40 -
Q
0)30
c
o 25
n
O
5 20
r
"o
E 15
Q.
CD
10
RAT
I-'"
MONKEY
0
0
468
[HtHO] (ppm)
10
Figure 3-2. Formation of DNA-protein cross-links in the
turbinates and anterior nose of F-344 rats and rhesus monkeys
exposed for 6 hr to HtHO. Values shown are mean + S.E., D = 3
exposure groups or animals per concentration. (Data in the rat
and monkey are from Casanova et al., 1989 and Heck et al., 1989,
respectively).
13
-------
The study by Casanova et al., 1989 also addressed the
question posed by the expert panel with regard to the use of
other covalent adducts as indicators of intracellular
formaldehyde-. - No evidence was obtained for formation of
hydroxymethyl derivatives of DMA, although it may be still be
argued that formation of such derivatives at levels below the
detection limit of the current methodology cannot be ruled out.
An important finding was the demonstration that hydroxymethyl
derivatives of deoxynucleosides may be formed during hydrolysis
of DPX; hydroxymethyl adducts are stable to HPLC conditions and
detection by this analytical procedure is possible. The choice
of buffer is critical in the formation of hydroxymethyl compounds
from deoxynucleosides. The use of Tris buffer effectively
eliminates formation of these intermediates because the primary
amine group in Tris interacts with the released, "free"
formaldehyde. Buffers not "interactive" with formaldehyde allow
formation of hydroxymethyl derivatives. It is possible that a
previous report (Beland et al., 1984) on the isolation of hm6dA
in DNA hydrolyzates, isolated from Chinese hamster ovary cells
incubated with HCHO (1 mM) may have dealt with an artifact
produced during the isolation procedure.
An important issue that is not addressed by the current work
is the suitability of using an acute exposure model to predict
effects under chronic exposure. A source of uncertainty is the
replacement of normal respiratory epithelial cells by squamous
cells that takes place within a week following the first
exposure. Whether the pharmacokinetic parameters for
formaldehyde in these two different cell types are comparable
remains to be established.
An important aspect of the in vivo disposition of
formaldehyde, and not discussed in the current studies, is the
mechanism by which formaldehyde penetrates into the cell and
eventually the nucleus. It is improbable to presume that
formaldehyde could penetrate cellular membranes as free material.
It is more likely that a "latent" form of formaldehyde is carried
into the cell possibly bound to protein(s) as an N-hydroxymethyl
derivative. This "latent" formaldehyde carrier could either
deliver formaldehyde to a critical target (DNA), or react
directly with DNA. A certain selectivity governs this type of
reaction. It has been reported that only proteins that have
binding affinity for DNA form cross-links in the presence of
formaldehyde (Solomon and Varsharsky, 1985), an observation that
may suggest that perhaps N-hydroxymethyl derivatives of -proteins
may constitute an "active" form of formaldehyde. Mechanisti-
cally, reactivity considerations would favor reaction of
formaldehyde with nucleophilic groups in proteins rather than
DNA. DNA affinity as a requisite for protein cross-linking
reactions is not unreasonable, or unexpected, if one considers
that the length of the cross-link established is rather short,
thus requiring close proximity of reactants. Other observations
14
-------
from chemical reactions of formaldehyde with weak nucleophiles
such as amides indicate that reaction of N-hydroxymethyl
derivatives takes place more readily than the corresponding
reaction with formaldehyde (Walker, 1975). The latter may be
simply a matter of reversibility.
'n*
These observations could eventually be incorporated in the
design of a pharmacokinetic model of the disposition of
formaldehyde, i.e., the model could incorporate the potential
role of protein N-hydroxymethyl adducts in the formation of DPX.
The current model relates extracellular formaldehyde to the
intracellular concentration of formaldehyde not bound to GSH. A
possible scheme for the reaction of formaldehyde leading to
formation of DPX that incorporates the above observations would
involve initial formation of a protein N-hydroxymethyl derivative
-probably more than one hydroxymethyl group per protein - which
transports formaldehyde into the cell. Since N-hydroxymethyl
reactions are reversible, the same protein may not be the one
that transports formaldehyde into the nucleus, binds and
eventually reacts with DNA. The slow step in this process could
be formation of the first cross-link; reaction of other N-
hydroxymethyl groups present in the same protein should take
place at a faster rate, i.e., a contributing factor to the
nonlinear kinetics observed may also be the presence of multiple
intermediates (protein N-hydroxymethyls) reacting with DNA at
different rates.
The conclusions based on currently available evidence are
that: formaldehyde reacts in vivo with proteins and DNA to form
cross-linked adducts (DPX); formation of DPX is nonlinear with
respect to exposure concentration, the source of nonlinearity is
attributed to saturation of detoxication mechanisms that include
binding to extracellular (mucosal) proteins and oxidation to
formic acid - an alternative is to consider the role of protein
N-hydroxymethyl derivatives as an "active" form of formaldehyde
which would also be expected to exhibit nonlinear kinetics;
identification of specific adducts found in DPX remains an issue;
and, the role of N-hydroxymethyl derivatives of proteins as
described above remains unexplored.
The current pharmacokinetic analysis of DPX formation have
adequately addressed most of the reservations expressed in the
expert panel's report. The use of DPX as the surrogate dose for
risk estimates appears appropriate although some uncertainties
remain which may limit the application of these data. These
uncertainties, which have a strong pharmacodynamic component, are
associated with the use of an acute model to predict chronic
effects and the biological significance of DPX as it relates to
formaldehyde-induced carcinogenesis. These issues are discussed
in more detail in Section 5 of this document.
15
-------
4.0 CARCINOGENIC HAZARD
4 . l Animal .Carcinoaenicitv Studies
-— u*
4.1.1 Long-Term Bioassays
The principal evidence indicating that formaldehyde is
carcinogenic in laboratory animals derives from the inhalation
studies in rats and mice conducted by CUT. In the first study
(Kerns et al., 1983), groups of 120 male and 120 female Fischer
344 strain rats and C57BL/6 X C3HF1 strain mice were exposed by
inhalation to 0, 2.0, 5.6 and 14.3 ppm of formaldehyde gas for 6
hr/day, 5 days/week, for 24 months. Squamous cell carcinomas of
the nasal cavity were observed in 103 of 232 rats (44%; 51/117
males and 52/115 females) and 2 of 215 mice (0.9%) exposed to
14.3 ppm of formaldehyde and in 2 of 235 rats (0.9%; 1/119 male
and 1/116 female) exposed to 5.6 ppm formaldehyde gas. Research
recently completed at CUT (Monticello, 1990) supports the
results of the 1983 study in rats in that nasal tumors were found
with formaldehyde exposures to 10 (4/17, 23% tumor incidence) and
15 ppm (27/45, 60% tumor incidence) . At 6 ppm, no tumors were
observed in Monticello (1990). The reasons for the difference in
these results at 6 ppm compared to those in Kerns et al. (1983)
may be due to fewer animals allotted to dose groups in Monticello
(1990) .
The spontaneous incidences of nasal neoplasia in these
strains of rats and mice are extremely low. Although not
statistically significantly elevated in incidence in comparison
with matched control mice; the nasal carcinomas observed in 2
(out of a total of 215) mice exposed to a concentration of 14.3
ppm were, therefore, regarded as related to formaldehyde
exposure. The difference in susceptibility of rats and mice has
been postulated to be due, at least in part, to a greater
reduction in respiratory rate and minute volume, and thus the
"dose" available for deposition in the nasal cavity, in mice than
in rats during exposure to this irritating vapor (see Section
4.3). A lifetime inhalation study was also conducted in hamsters
exposed to 10 ppm formaldehyde for 5 hours/day, 5 days/week
Dalbey, 1982). There was no tumorigenic response in the
formaldehyde-exposed hamsters, but survival time was reduced
compared to unexposed controls.
Kerns et al. (1983) , also reported a small number, of
polypoid adenomas and other neoplasms (adenocarcinoma, poorly
differentiated carcinoma/sarcoma) of the nasal mucosa in the
rats. However, according to a Pathology Workgroup (Consensus
Workshop, 1984), these polypoid adenomas are not likely to
represent the benign counterpart of squamous cell carcinomas.
Therefore, the pol^ypoid adenomas were not combined with squamous
cell carcinomas for risk estimation.
16
-------
Knowledge of the anatomic sites of origin of formaldehyde-
induced neoplasms is important for a better understanding of the
nasal carcinogenicity of formaldehyde. In this chronic
inhalation study (Kerns et al., 1983), the tumors were reported
to originate'11 in the. anterior portion of the nasal cavity but
their precise location in the nose was not determined. Morgan et
al. (1986) have defined more precisely the localization of nasal
tumors associated with chronic exposure of F334 rats to
formaldehyde vapor by reexamining histologic sections from the
nasal passages of 98 of the 103 tumor-bearing rats (slides of 5
rat noses were missing) in the inhalation study (Kerns et al.,
1983). In their analysis (Morgan et al., 1986), the location of
121 squamous cell carcinomas in 98 rat noses (some had multiple
neoplasms) were recorded on diagrams of cross sections of the
nose. It was found that the majority of squamous cell carcinomas
occurred on the lateral side of the anterior portion of the
nasoturbinate and adjacent lateral wall (57%) or the midventral
nasal septum (26%) ; about 10% were on the dorsal septum and the
roof of the dorsal meatus. Only one squamous cell carcinoma (out
of 121 reexamined) was found on the medial aspect of the
maxilloturbinate, a region which presumably also has a high
"delivered dose" due to airflow patterns influencing regional
exposure to formaldehyde. It had also been shown that
inflammation and sustained cell proliferation but not tumors
occurred in this region following chronic exposure of
formaldehyde (see Section 4.3). These findings suggested to the
authors (Morgan et al., 1986) that not only regional exposure but
local tissues susceptibility may be important for the
distribution of formaldehyde-induced neoplasms in the nose of
rats.
Additional studies by other investigators have confirmed the
carcinogenicity of formaldehyde in Fischer 344 and other strains
of rats by inhalation exposure. In a study by Tobe et al.
(1985), groups of 32 male Fischer 344 rats were exposed to 0,
0.3, 2.0, and 15 ppm formaldehyde vapor 6 hours/day, 5 days/week,
for 28 months. Squamous cell carcinomas of the nasal cavity were
seen in 14/27 high-dose (15 ppm) rats surviving past 12 months;
no such neoplasms were observed in the control or the low-dose
groups (0.3 and 2.0 ppm). In addition, nasal papillomas,
considered to represent the benign counterpart of the squamous
cell carcinomas, were observed in 5 high-dose rats in this study.
In two reports (Albert et al., 1982; Sellakumar et al.,
1985), groups of 99-100 male Sprague-Dawley rats were exposed 6
hours/day, 5 days/week for life by inhalation to: hydrochloric
acid (HC1) alone, mixtures of HCl and formaldehyde, or
formaldehyde alone. The results show that significant incidences
(27-38%) of squamous cell carcinomas of the nasal cavity were
induced in all rats exposed to formaldehyde (14-15 ppm)
regardless of the concurrent exposure to HCl. No carcinogenic
response was observed in the rats exposed to air (control) or to
HCl alone.
17
-------
Feron et al. (1988) exposed male Wistar rats to 0, 10, or 20
ppm formaldehyde vapor 6 hr/day, 5 days/week, for 13 weeks and
then were held up to 117 weeks. A low incidence (6/132, 4.5%) of
nasal tumors- f-3 -^squamous cell carcinomas, 1 carcinoma in situ and
2 polypoid adenomas) 'was observed in the rats exposed to 20 ppm
formaldehyde vapor. Squamous cell carcinoma was also noted in
1/44 rats at the dose level of 10 ppm. No such neoplasms were
found in 45 control rats. These data indicate a carcinogenic
potential of formaldehyde after short-term exposure to very high
concentrations.
Formaldehyde-induced nasal tumors were found mainly at
exposure concentrations which also induced severe degenerative,
hyperplastic and metaplastic changes in nasal epithelium.
Therefore, it has been suggested that increased cytotoxicity and
cell replication may play an essential role in the induction of
nasal cancer by formaldehyde. To study the significance of
cytotoxicity to the nasal mucosa for the induction of nasal
tumors by formaldehyde in rats, a long-term inhalation study was
conducted in which groups of 60 male Wistar rats with either
severely damaged or undamaged nose were exposed to 0.1, 1.0, or
10 ppm formaldehyde, 6 hr/day, 5 days/week for 28 months
(Woutersen et al., 1989). The damage to the nasal mucosa was
induced by bilateral intranasal electrocoagulation. The results
of these experiments showed that compound-related hyperplasia,
rhinitis (inflammation of the nasal cells with squamous cells) in
the respiratory mucosa developed in rats with either a damaged or
an undamaged nose at 10 ppm, although a higher incidence of these
lesions occurred in the rats with a damaged nose. Exposure to 10
ppm formaldehyde for 28 months produced significant incidence
(15/58) of nasal squamous cell carcinomas in rats with damaged
nose but not in rats with intact nose. No compound-related nasal
neoplasms or cytotoxic effects were observed in rats (with either
a damaged or undamaged nose) exposed to 0.1 or 1.0 ppm
formaldehyde for 28 months. The findings of these studies,
therefore, suggest that the nasal mucosa damaged by
electrocoagulation was more susceptible to the induction of nasal
tumors by formaldehyde. The roles of cytotoxicity and cell
proliferation in formaldehyde-induced carcinogenesis are
discussed in Section 4.3.2.
As shown in Table 4-1, the combined results of four long-
term inhalation studies indicate that formaldehyde induces high
incidences of nasal tumors (38-50%) in two strains of rats at
approximately 15 ppm (Albert et al., 1982; Kerns et al.,'1983;
Tobe et al., 1985; Sellakumar et al., 1985; Monticello, 1990).
There is also weak evidence of carcinogenicity at 10 ppm or lower
concentrations. The observed tumor incidence in these studies
produces a very steep dose-response curve for the carcinogenic
effect of formaldehyde. An exception to the above is the
observation that rats with noses damaged by electrocoagulation
had a high incidence of nasal tumors following inhalation
ol
18
-------
Table 4-1
Incidencea of Nasal Squamous Cell Carcinomas in Rats
— .-. After Inhalation Exposure to Formaldehyde
Rat Strain
Fischer 344
Fischer 344
Fischer 344
Sprague-Dawley*
Wistar**
Wistar*«*
Dose Level
(ppm)
0
2.0
5.6
14.3
0
0.7
2.0
6.0
10.0
15.0
0
0.3
2.0
15.0
0
14.0-15.0
0
10.0
20.0
0
0.1
1.0
10.0
Tumor Incidence
Male Female
. 0/118 0/114
0/118 0/118
1/119 1/116
51/117 52/115
0/17
0/17
0/16
0/18
4/17
27/45
0/32
0/32
0/32
14/27
0/99
27/100-38/100
0/45
1/44
3/132
1/54
1/58
0/56
15/58
Reference
Kerns
et al.,
1983
Monticello,
1990
Tobe et
al., 1985
Albert
et al. ,
1982;
Sellakumar
et al . ,
1985
Feron
et al . ,
1988
Woutersen
et al. ,
1989
* With or without concurrent exposure to hydrochloric acid; hydrochloric acid
alone did not induce any tumors.
** Exposed for 13 weeks, then observed up to 117 weeks.
**« Rat noses were damaged by bilateral intranasal electrocoagulation. No tumors
were found in exposed rats with undamaged noses.
19
-------
exposure to formaldehyde at 10 ppm. These data strongly suggest
that increased cytotoxicity and cell proliferation play a role in
the induction of nasal cancer by formaldehyde.
Several investigators (Takahashi et al., 1986; Til et al.,
1989; Tobe et._aL., 1-9,89; Soffritti et al., 1989) have also
studied the carcinogenic activity of formaldehyde by
administration to the rat in the drinking water.
In one study (Takahashi et al., 1986), squamous cell
papillomas of the forestomach were observed in 8 of 10 Wistar
rats given 0.5% formalin (about 0.2% formaldehyde) in the
drinking water for 32 weeks.
In another study by Til et al. (1989), groups of 70 male and
70 female Wistar rats were administered 0.19% formaldehyde in
their drinking water for up to 104 weeks. Significant incidences
of hyperplastic and inflammatory changes in the fore- and
glandular stomach were reported, but there was no evidence of
tumor formation. Til et al. (1989) proposed that the differences
in findings between these two studies might be ascribed to the
use of different criteria for the classification of a lesion as a
papilloma or as papillary epithelial hyperplasia.
Similarly, Tobe et al. (1989) observed high incidences of
forestomach and glandular hyperplasia in the rats given
formaldehyde solution in drinking water for 12 months at a
concentration of 0.5%. In this study, group of 20 male and 20
female Wistar rats were given formaldehyde at concentrations of
0, 0.02, 0.10 and 0.5% in their drinking water ad libitum. Six
randomly chosen rats from each group were sacrificed after 12 and
18 months. All 12 rats (6 males and 6 females) in the 0.5% group
sacrificed at 12 months developed squamous cell hyperplasia in
the forestomach. Hyperplasia of the glandular stomach was
observed in 10 of the 12 high-dose rats (6 males and 4 females)
sacrificed at 12 months. The experiment was terminated at 24
months and all surviving rats were killed for necropsy. Various
types of tumors were observed in organs such as pituitary,
thyroid, testis, adrenals, mammary gland and skin; however, there
were no significant difference in the incidences of any tumors
among the treated and control groups. This drinking water study
may not be adequate to evaluate the carcinogenic potential of
formaldehyde because of the high early mortality of the high-dose
group and the use of a small number of animals. At 12 months,
the mortality rates in the high-dose (0.5%) group were 45% and
55% in males and females, respectively. At 18 months, the
respective mortality rates reached 67% and 70%, and by 2,4 months,
all animals in the 0.5% group had died.
In a study conducted by Soffritti et al. (1989), groups of
50 male and 50 female Sprague-Dawley rats of different ages at
the start of the study (12 day embryos — transplacental
20
-------
carcinogenesis, 7- and 25-week old rats) were administered
formaldehyde in drinking water at different concentrations (0,
0.001, 0.005, 0.01, 0.1, 0.15, and 0.25%) for 104 weeks. A low
incidence (2-21%) of gastrointestinal neoplasms was observed in
some treated animals. These gastrointestinal neoplasms, which
were not see_n in the matched controls and were reported to be
extremely rare 'in the, historical controls (1.39% incidence) of
this rat strain, included benign tumors (papillomas, acanthomas,
adenomas and leiomyomas) and malignant tumors (sguamous cell
carcinomas, adenocarcinomas and leiomyosarcomas) of the
forestomach and the small intestine (duodenum, jejunum, and
ileum). The lack of clear reporting of the test results limits
the usefulness of this study.
Collectively, the findings of these drinking water studies
with formaldehyde (Takahashi et al., 1986; Til et al., 1989; Tobe
et al., 1989; Soffritti et al., 1989) provide only suggestive
evidence for the carcinogenicity formaldehyde by the oral route.
in the rat by the oral route.
In the study of Soffritti et al. (1989), a slightly
increased incidence of leukemia was also noted in some
formaldehyde-treated rats (11-18%) as compared with the controls
treated with drinking water alone (3.5-5.5%). However, no
statistical analysis and historical control data were reported;
the biological significance of these low incidences of neoplasms
is questionable.
The carcinogenicity of formaldehyde has also been studied by
a variety of other routes of administration and/or in other
animal species including application to buccal mucosa in rabbits
(Meuller et al., 1978), subcutaneous injection in rats (Watanabe
et al., 1954; Watanabe and Sugimoto, 1955), skin painting in mice
(Iversen, 1986), and inhalation in hamsters and monkeys (Rusch et
al., 1983). Although some of the data suggest that formaldehyde
was carcinogenic, none of these studies is conclusive regarding
formaldehyde carcinogenicity because of various limitations in
the experimental protocols used (e.g., small number of animals,
short duration).
4.1.2 Tumor Promotion and Co-Carcinogenicity Studies
The tumorigenosis-promoting potential of formaldehyde has
been investigated in classical skin painting experiments .in mice
(Iversen, 1986). Two groups of 32 mice (hr/hr Oslo strain) were
painted once with 51.2 ug dimethylbenz(a)anthracene (DMBA) in 100
ul acetone; one group was followed 9 days later with applications
of 200 ul 10% formaldehyde in water twice a week, and the other
was given no further treatment. The animals were treated and/or
observed for 60 weeks. The incidence of skin tumors after DMBA
initiation followed by formaldehyde (11/32, 37.5%) was not
significantly different from that after DMBA alone (85/225,
21
-------
37.7%). However, the time of appearance of the first tumor and
the mean latency time were significantly reduced.
As benrofa ] pyr,$ne (B[a]P) is a well-known respiratory
carcinogen, the interaction of B[a]P and formaldehyde in
respiratory tract carcinogenesis was studied in rat tracheal
explants (Cosma and Marchok, 1987). The induction of
carcinogenesis was quantitated as the number of growth-altered
cell populations (designated as "tumor-initiation sites", TIS)
isolated in cell cultures generated from explants cut from the
preexposed tracheas. While exposure twice-weekly for 4.5 months
to 0.2% formaldehyde solution gave only a weak response (0.25
TIS/trachea), 2.37 TIS per trachea were detected after exposure
to 20 ug B[a]P in the same regimen. The combination of B(a]P
followed by formaldehyde had a greater response than either
compound alone (7.83 TIS/trachea), while the reverse exposure
gave 1.49 TIS per trachea, which was less than B[a]P alone.
Thus, the induction of TIS by combined exposure to B[a]P and
formaldehyde was dependent on the order of exposure. The
suppression of the cellular response to B(a]P when it was
preceded by exposure of formaldehyde was believed to be due to
the inhibition of enzymes involved in the metabolism of B[a]P.
The effects of formaldehyde on gastric carcinogenesis in
rats after initiation with N-methyl-N-nitro-N-nitrosoguanidine
(MNNG) were studied by Takahashi et al. (1986). Male outbred
wistar rats were given MNNG in the drinking water (100 mg/liter)
and a diet supplemented with 10% sodium chloride for 8 weeks.
Thereafter, they were maintained on drinking water containing
0.5% formalin (about 0.2% formaldehyde) for 32 weeks and then
sacrificed for necropsy and histological examination. The
results showed that formaldehyde significantly increased the
MNNG-induced incidences of neoplasms in the forestomach and
glandular stomach of the rats. Papillomas in the forestomach were
observed in 88% (15/17) of animals treated with MNNG and
formaldehyde; no such neoplasm was seen in 30 rats treated only
with MNNG. The incidences of glandular stomach adenocarcinomas
in rats treated with MNNG alone and in rats treated with both
MNNG and formaldehyde were 13% and 29%, respectively.
No evidence was found for any co-carcinogenic effects of
formaldehyde when C3H mice were exposed simultaneously to coal
tar aerosol and formaldehyde (Horton et al., 1963). There was
suggestive evidence of a co-carcinogenic effect of formaldehyde
on diethylnitrosamine-induced carcinogenesis in hamsters (Dalbey,
1982) , but high mortality and other factors discussed in EPA
(1987) limit the usefulness of this study.
4;1.3 Summary and Conclusions of Animal Studies
An earlier review (EPA, 1987) has indicated that there is
sufficient evidence for the carcinogenicity of formaldehyde in
22
-------
experimental animals exposed by inhalation at high
concentrations. Recent animal studies confirm previous findings
of an increased incidence of squamous cell carcinomas of the
nasal cavity in rats exposed by chronic inhalation or less than
lifetime exposure to high concentration of formaldehyde. In
addition, tfie "localization of nasal tumors in rats has been
better defined; the findings suggest that not only regional
exposure but also'local tissue susceptibility may be important
for the distribution of formaldehyde-induced neoplasms. The nasal
mucosa damaged by electrocoagulation has been found to be more
susceptible to the induction of nasal tumors by formaldehyde.
Results obtained from four studies in which formaldehyde was
administered to rats in the drinking water provide only
suggestive evidence by the oral route for the carcinogenicity of
formaldehyde.
In recent tumor promotion studies, formaldehyde enhanced the
tumor response in mouse skin, rat trachea, and rat stomach,
indicating that formaldehyde has tumor promotion potential at
least in some tissues but not in hamster respiratory tract. No
co-carcinogenic effects were seen in any available studies.
4.2 Mutagenicity
The EPA' (1987) document detailed the ability of formaldehyde
to induce mutation in several test systems, including viruses,
bacteria, yeast, fungi, Drosophila. grasshoppers, and mammalian
cells in culture. The document further described the ability of
formaldehyde to induce cellular transformation in several
systems, to act as a co-mutagen when tested with agents such as
x-rays, ultraviolet light, and hydrogen peroxide, and to have
enhanced activity in test systems deficient in excision-repair
mechanisms. Formaldehyde was further characterized as inducing
both chromosome aberrations and sister chromatid exchanges in
vitro as well as causing single-strand breaks in DNA, inhibiting
the resealing of single-strand breaks produced by ionizing
radiation and inducing cross-links between DNA and protein in
both epithelial and fibroblastic cells. Formaldehyde was
proposed to act by damaging DNA and by inhibiting DNA repair in
treated cells. Studies on the metabolic incorporation of
formaldehyde versus covalent binding to DNA in treated cells were
also included in the document.
Formaldehyde was reported neither to induce micronuclei in
vivo nor to cause sperm head abnormalities in mice (EPA, 1987).
There was question about the ability of formaldehyde to induce
mutation in the dominant lethal assay in mice and in ari in vivo
assay for sister chromatid exchanges. Limitations in these
studies were stated in the document.
23
-------
Human studies concerned with the occurrences of chromosomal
aberrations and sister chromatid exchanges in the blood and
mutagenic products ^in urine of exposed workers were also
detailed. — — * '-M,
New literature in the area of mutagenicity/genotoxicity was
reviewed and included studies on covalent binding and the
induction of DNA strand breaks; studies on induction of
chromosomal aberrations in vitro; co-mutagenesis studies of x--
rays and formaldehyde in Drosophila; cytotoxicity and mutation in
human lymphocytes and Salmonella and a study on sister chromatid
exchange formation in anatomy students exposed to formaldehyde
embalming solution (Bogdanffy et al., 1987; Casanova et al.,
1989; Craft et al., 1987; Crosby, 1988; Dresp and Bauchinger,
1988; Dowd et al., 1986; Ecken and Sobels, 1986; Liber et al.,
1989; Heck et al., 1989, in press; Schmid et al., 1986; Snyder
and Van Houten, 1986; Yager et al., 1986).
The new studies corroborate previous findings on the
genotoxicity of formaldehyde (EPA, 1987).
4.3 Possible Mechanisms of Formaldehyde-Induced
Carcinogenicity
Formaldehyde has been shown to be mutagenic in a variety of
assay systems, and there is convincing evidence that formaldehyde
is genotoxic in mammalian cells, including human lymphoblasts
(see Section on Mutagenicity; also EPA, 1987; Ma and Harris,
1988) . The mechanisms of genotoxicity of formaldehyde, however,
are not yet completely understood. Formaldehyde is chemically
reactive and has been shown to interact with various biological
macromolecules including amino acids, proteins, nucleic acid
bases, nucleosides, RNA and DNA (EPA, 1987). The reactions of
formaldehyde with DNA and its components are of particular
interest since such reactions are believed to be involved in the
genotoxic effects of chemicals (see Section 3.0). DNA-DNA cross-
links have been identified in pure DNA preparations incubated
with formaldehyde (Chaw et al., 1980), but such direct cross-
links have yet to be identified following treatment of intact
cells. Some evidence suggests that indirect DNA-DNA cross-
linking may occur by the linkage of two DNA strands to the same
chromatin to form protein bridges between DNA strands (Wilkins
and MacLeod, 1976). Snyder and Van Houten (1986) have shown that
the reaction of formaldehyde with synthetic DNA polymers in
protein-free mixtures resulted in an increased frequency of
misincorporation by polymerase, suggesting the production of
potentially mutagenic adducts.
The understanding of the mechanisms of the carcinogenic
effects is important for the risk assessment of formaldehyde.
The findings that formaldehyde is a nasal carcinogen in rats and
that the dose-response for the nasal squamous cell carcinomas is
24
-------
nonlinear (0% incidence at 0 and 2 ppm, about 1% incidence at 5.6
ppm and 44% incidence at 14.3 ppm; see Section 4.1.1) have
prompted considerable research into its mechanism(s) of
carcinogenic-.actipn. As a result, a number of factors and
hypotheses have been proposed to understand this type of
carcinogenic response. Recent mechanistic studies on
formaldehyde have focused on the roles of DNA-protein cross-
linking (see Section 3.0) and cell proliferation. Examinations
of the species differences in carcinogenic response and the role
of the nasal mucociliary apparatus has led to the conclusion that
the "dose" available for deposition in the nasal cavity and
epithelium may be critical for the carcinogenic effects of
formaldehyde.
4.3.1 Role of DNA-Protein Cross-linking
DNA-protein cross-links (DPX) are removable by excision
repair, which may increase the freguency of single-strand breaks
in DNA and represent a possible mutagenic response of cells
exposed to formaldehyde. Cross-linking of DNA and protein via a
methylene bridge N-CH2-N, has been observed in a variety of
mammalian cells in culture following treatment with formaldehyde
(Bedford and Fox, 1981; Ross et al., 1981; Craft et al., 1987).
Evidence that formaldehyde may form DPX in vivo was provided by
studies from CUT (as discussed in Section 3.0). Crosby et al.
(1988) have shown that formaldehyde mutagenesis in human
lymphoblasts is characterized by large deletions of DNA (which
may result from DPX formation) and that in E. coli, different
types of mutations (large deletions, base-pair transversions and
transitions) are induced at different concentrations of
formaldehyde. It seems possible that a shift in the type of
mutations induced by formaldehyde as a function of concentration
may be related to the highly nonlinear nature of formaldehyde
carcinogenesis in rat nasal cavity.
However, the relevance of DPX to formaldehyde carcinogenesis
needs to be further explored. For instance, the types of
mutations and the relationships between DNA-protein cross-links
and formaldehyde mutagenesis have not yet been examined in the
rat nasal mucosa; the repair of formaldehyde-induced DNA-protein
cross-links has not been investigated to define the kinetics or
mechanism of this process; and this lack of knowledge constitutes
a major gap in our understanding of this important mechanism
postulated for formaldehyde carcinogenesis.
/
4.3.2. Role of Cytotoxicity and Cell Proliferation
It is generally accepted that the multi-stage process of
carcinogenesis involves DNA damage (mutation) and clonal
expansion of the initiated or altered cells. Theoretically,
there are several roles that cell replication may play in the
process of carcinogenesis. The chance for an initiating event is
25
-------
enhanced during cell replication when the DNA is single stranded
and experiences greater exposure and susceptibility to
chemically-induced DNA damage. Furthermore, enhanced cell
replication may increase the probability of converting DNA
adducts into mutations before they can be repaired. However,
cell proliferation is an ongoing process in normal living tissue.
Enhanced cetl-proliferation is generally believed to be an
important but not a sufficient step to induce tumors. Not all
chemicals that induced cell proliferation or hyperplasia are
equivalent in their ability to promote/induce tumor formation.
Chemically-induced cell proliferation may just represent
regenerative cell growth ("adaptation"), which does not
necessarily lead to tumorigenesis. It appears that clarification
between "adaptive" versus "preneoplastic" responses is needed.
Measurement of the extent and duration of site-specific cell
proliferation, in conjunction with elucidation of the exact
biochemical and molecular alterations underlying the effects is
important in understanding the relationship of chemically-induced
cell proliferation to carcinogenesis.
Formaldehyde-induced nasal tumors were found mainly at
exposure concentrations which also induced severe degenerative,
hyperplastic and metaplastic changes in nasal epithelium.
Therefore, it has been proposed that increased cytotoxicity and
cell replication may play an essential role in the induction of
nasal cancer by formaldehyde.
Studies by Woutersen et al. (1989) in Wistar rats have
demonstrated a higher susceptibility of the nasal mucosa damaged
by electrocoagulation than the undamaged nasal mucosa to the
induction of nasal tumors by formaldehyde). Bilateral intranasal
electrocoagulation was reported to cause damage in all structures
in the anterior of the nasal cavity, with perforation of the
nasal septum and loss of turbinates being a common occurrence.
The higher susceptibility of the damaged than the undamaged
nasal mucosa to the carcinogenic action of formaldehyde may be
due to irreparable damage to nasal defense mechanisms (e.g., the
mucociliary apparatus) caused by the electrocoagulation, thus
rendering the metaplastic and hyperplastic nasal respiratory
mucosa more susceptible to formaldehyde. It is plausible that
the cytotoxic effects caused by electrocoagulation may enhance
the carcinogenic response of formaldehyde by providing an
increased opportunity for formaldehyde to interact with single-
strand DNA during cell proliferation or by promoting the
formaldehyde-initiated cells.
Studies investigating formaldehyde-induced toxicity have
shown that increased cell proliferation occurs following acute
exposure and is an early and sensitive indicator for cytotoxic
effects in the nasal passages of rats (Swenberg et al. 1983;
Zwart et al., 1988). Using a labeling technique with H-
thymidine, Swenberg et al. (1983) found a 10- to 20-fold increase
26
-------
in the labeling index (defined as the proportion of cells in a
given population that are in the process of DNA replication) in
the nasal epithelium of rats 2 or 18 hours after exposed to 6 or
15 ppm 6 hr/day, for 3-5 days; only slight increases were
detected in rats ^exposed to 2 or 0.5 ppm formaldehyde as compared
with the controls."' In mice, increased cell proliferation was
evident only at 15 ppm and was much lower than that in rats
(Table 4-2).
It should be noted, however, that increases in cell
proliferation following acute exposure to rats occurred in the
respiratory epithelium of both the central portion (the primary
site of sguamous cell carcinomas) and the most anterior portion
of the nasal passages. Furthermore, the observed cell
proliferation was substantially decreased after 10 days of
exposure (Swenberg et al., 1983).
To study in detail possible effects of low concentrations of
formaldehyde on the nasal epithelium, Zwart et al. (1988) exposed
Wistar rats to 0, 0.3, 1 and 3 ppm formaldehyde vapor for 6
hr/day, 5 days/week during the 3-day or 13-week cell
proliferation studies. A statistically significant increase in
cell proliferation in the respiratory epithelium of the central
portion (the primary site of sguamous cell carcinomas; Level III
in a 6 standard cross levels of the nose) of the nose was
observed in rats exposed to 1 and 3 ppm formaldehyde for 3 days;
the dose-response showed an approximate log-linear positive
relationship with about a lO-fold increase in the 3 ppm group.
After 13 weeks of exposure, the cell turnover rate at this nose
region, however, tended to be slightly lower than in the
controls. At the anterior part of the nose (Cross level II), on
the other hand, an increase in cell proliferation was observed in
the rats after both 3 days and 13 weeks of exposure to 3 ppm
formaldehyde only; cell turnover in rats exposed to 0.3 and 1 ppm
formaldehyde was not different from controls.
The findings of these studies (Swenberg et al., 1983; Zwart
et al., 1988), therefore, suggest that restorative cell
proliferation occurs in nasal passages of rats in response to
formaldehyde exposure and that: (1) a significant increase in
cell turnover occurred above l ppm, (2) the response lacks site
specificity with respect to tumor induction and, (3) the response
seems to be transient.
Recent acute and subacute studies have also demonstrated
that there is a correlation between cytotoxicity and cell
proliferation induced by formaldehyde in the rat nasal epithelium
and that the cell proliferation rate is concentration-dependent.
In these studies (Monticello, 1990), groups of Fischer 344 rats
were exposed to 0, 0.7, 2, 6, 10, or 15 ppm formaldehyde gas, 6
hours/day, for 1, 4, or 9 days, or 6 weeks. There were no
treatment-induced lesions or responses in cell proliferation in
27
-------
Table 4-2
Effect of Formaldehyde Exposure on Cell Proliferation
irr th,e Central Portion Of the Nasal Passages*
% of Labelled Respiratory Epithelial Cells
Exposure** Rat Mouse
Control
0.5 ppm
2 ppm
6 ppm
15 ppm
0.22±0.03***
0.38+0.05
0.33±0.06
5.40±0.82
2.83±0.81
0.12+0.02
0.09+0.04
0.08+0.04
0.15+0.06
0.97+0.04
*Swenberg et al. (1983).
**A11 animals exposed for 6 hrs/day for 3 days.
***Mean + standard error,
28
-------
the two lowest concentration groups, 0.7 and 2.0 ppm. However,
formaldehyde-induced lesions and increased cell proliferation
were detected in the nasal passages of the 6, 10, and 15 ppm
groups. Elevations, in cell proliferation were detected following
1 day at concentrations of 6 ppm or higher and remained elevated
after 6 weeks of exposure for some locations (the lateral wall
and the maxilloturbinate) of the anterior nasal cavity.
Further insights into the relationship between both
sustained cell proliferation and inflammation and formaldehyde-
induced nasal carcinogenesis have recently been provided by the
results of a chronic study in Fischer 344 rats. In these studies
(Monticello, 1990; Monticello and Morgan, 1990), regional nasal
respiratory epithelial lesions and cell proliferation (labeled
cells/mm basement membrane) were examined following 6, 12, or 18
months exposure (6 hr/day, 5 day/week), to 0, 0.7, 2, 6, 10, or
15 ppm formaldehyde. Tumor incidence was assessed for all
animals (scheduled and unscheduled sacrifices) through 18 months
exposure of this 2-year study. The data showed that
statistically significant elevations in cell proliferation were
confined to the high-dose (10 and 15 ppm) groups following
chronic exposures (6, 12 and 18 month values expressed as fold-
increase over control; 10 ppm group: 4.0-, 4.3-, and 4.0-fold,
respectively; 15 ppm group: 7.0-, 9.8-, and 6.6-fold,
respectively). Nasal epithelial cell necrosis, inflammation,
hyperplasia, metaplasia and squamous cell carcinomas also were
present only in the 10 (4/17, 23% tumor incidence) and 15 ppm
(27/45, 60% tumor incidence) groups. These results lend some
support to the hypothesis that cell proliferation plays an
important role in formaldehyde-induced nasal carcinogenesis since
there appears to be a correlation between sustained cell
proliferation and nasal tumor induction in rats exposed to
formaldehyde at high doses (see Figure 4-1). These results also
point to cell proliferation as a contributor to the observed
nonlinear dose-response for formaldehyde-induced nasal tumors in
rats; the cell proliferation data appear important for the
development of biologically-based models for quantitative cancer
risk assessment of formaldehyde.
It should be noted, however, as in the acute studies
(Swenberg et al., 1983; Zwart et al., 1988), that the cell
proliferation responses observed in these chronic studies
(Monticello and Morgan, 1990) lack site-specificity with respect
to tumor induction; similar increases in cell proliferation and
inflammation scores occurred also at the medial aspect pf the
maxilloturbinate, a region where no nasal tumors were seen.
These results, therefore, suggest that sustained increases in
cell proliferation may not be solely responsible for the increase
in the predisposition of the nasal cavity to formaldehyde
carcinogenesis.
29
-------
Figure 4-1.
Tumor incidence and cell proliferation
in rats exposed to formaldehyde
- 13
TUMOR INCIDENCE 18-MONTH STUDY
(MONT1CEULO, 1900)
TUMOR INCIDENCE 24-MONTH STUDY
(KERNS. 1983: EPA. 1987)
••-o—
CELL PROLJFERAT1ON STUDY
6-MONTH (MONT1CELLO. 1990)
-A
CELL PROLIFERATION
12-MONTH (MONT1CELLO. 1990)
CELL PROLIFERATION
1 8-^ONTH (MONT1CELLO, 1990)
4 6 8 10 12
HCHO CONCENTRATION (PPM)
14
16
30
-------
As in the rat, studies in monkeys showed that formaldehyde-
induced lesions were correlated with increases in cell
proliferation (Monticello and Morgan, 1989). Groups of three
rhesus monkeys were exposed either to room air or 6 ppm
formaldehyde for 5 days per week for 1 or 6 weeks. Formaldehyde-
induced lesions in the respiratory tract were associated with
increases in cell proliferation rates up to 18-fold over
controls. Whereas the greatest response in cell proliferation
occurred in the anterior nasal cavity, increases in cell
proliferation in the monkey were also detected in the larynx,
trachea, and carina; this is in contrast to the rat where lesions
were confined to the anterior nasal passages. Furthermore, after
6 weeks of exposure to 6 ppm formaldehyde, the labeling index in
the monkey nasal mucosa remained elevated. Based on the extent
of lesions and cell proliferation data, it was concluded that the
monkey may be more sensitive than the rat to the acute and
subacute effects of formaldehyde at 6 ppm. This is further
supported by the finding that monkeys had considerably less DPX
than rats exposed to the same formaldehyde airborne concentration
(6 ppm) (Casanova et al., 1989). Under the conditions of this
study, the maxillary sinuses of the monkey did not exhibit any
evidence of histologic responses or changes in cell proliferation
rate.
Formaldehyde-induced lesions and cell proliferation in human
tracheobronchial epithelium that has been transplanted into rat
tracheas have been studied (Klein-Szanto et al., 1989; Ura et
al., 1989). In these experiments, de-epithelialized rat tracheas
repopulated with normal human tracheobronchial epithelium were
exposed in vivo to slow-releasing devices containing 0, 0.5, 1,
and 2 mg paraformaldehyde. These studies showed that all doses
elicited a proliferative response in the human respiratory
epithelium similar to that observed in the rat epithelium when
exposed to formaldehyde; the labeling index of the human
respiratory epithelium showed dose dependence between 2 and 4
weeks after initiation of exposure. Based on these findings, the
authors (Klein-Szanto et al., 1989) concluded that the similarity
of response of the rats and human respiratory epithelia exposed
to formaldehyde further supports the cancer risk concern of
humans exposed to formaldehyde.
4.3.3 Role of the Protective Effect of the Mucociliary
Apparatus
The mucociliary apparatus presents a continuous layer of
mucus which flows over the surface of the nasal epithelium. In
the rats, the mucociliary apparatus has been suggested to play an
important role in protecting the nasal epithelium against the
cytotoxicity of formaldehyde (Zwart et al., 1988). Both in vitro
and in vivo studies (Morgan et al., 1983, 1986) have shown that
there is a clear dose-dependent effect of formaldehyde on the
mucociliary apparatus of the rats. At 15 ppm, there was
31
-------
significant inhibition of mucociliary activity whereas at 2 or 6
ppm, only slight effects were noted. At 0.5 ppm, no effects were
observed. This ran.ge appears to correspond to the range where
the dose-resporises in- nasal mucosa cytotoxicity (Wilmer et al.,
1989) and in carcinogenicity (Kerns et al., 1983) of formaldehyde
were seen. It has, therefore, been postulated that at low
formaldehyde concentrations, the mucus la'yer of the rat nasal
cavity may trap and remove much inhaled formaldehyde, thus
preventing it from reaching underlying cells. If the mucus layer
is saturated (at high concentrations), the mucociliary clearance
system could be seriously compromised and may allow a greater
amount of formaldehyde to reach the underlying respiratory
epithelium.
Although there is a dose-dependent effect of formaldehyde on
the mucociliary apparatus (and on nasal cytotoxicity) of the
rats, it is not known to what degree the mucus layer protects
against the cytotoxic effects of formaldehyde.
In humans, nasal mucociliary function was inhibited by
exposure to 0.3 ppm formaldehyde for 1-5 hours (Anderson and
Molhave, 1983). It is also known that formaldehyde at levels
below 1 ppm can be detected in the olfactory region of the human
nose, indicating that formaldehyde is not completely removed by
the mucus layer, even at low concentrations.
4.3.4 Importance of Dose Available for Deposition
The sensory irritant effects of formaldehyde are well
documented. It has been shown that mice are more sensitive and
reduce their respiratory rate and minute volume to a greater
extent than rats in response to the irritancy of formaldehyde.
When both species were exposed to 14.3 ppm formaldehyde, the mice
inhaled only half as much formaldehyde per unit time as did rats
(Chang et al., 1983). Therefore, it has been postulated that it
may be this difference in the "dose" available for deposition to
the nasal cavity and epithelium that accounts for the different
carcinogenic response in rats and mice observed in the studies of
Kerns et al. (1983). However, this factor alone may not be
totally responsible for the difference in response between rats
and mice. It is not known whether similar carcinogenic response
would be observed if mice could have been exposed to levels of
formaldehyde comparable to the amount effectively inhaled by
rats.
4.3.5 Air Concentration vs. Total Daily Dose
An inhalation toxicity study was conducted by Wilmer et al.
(1989) in order to investigate the importance of formaldehyde
concentration vs. total dose. One hundred and fifty male Wistar
rats (albino SPF) were allocated to 5 groups of 25 animals for
histopathology, with 5 additional groups of 5 animals assigned to
32
-------
a 3-day cell-proliferation study. Five rats from each group of
25 were further examined at the end of the 13-week period for
histopathology and cell-proliferation. Two groups were exposed
continuously-to 1 or 2 ppm for a total duration of 8 hours, with
concentration-time products (Ct) of 8 or 16 ppm-hr/day. Two
groups were exposed intermittently at 30 minute intervals, total
duration of 4 hours, to 2 or 4 ppm with Ct equal to 8 or 16 ppm-
hr/day. A fifth group was exposed to air.
The cell-proliferation study showed no statistically
significant difference in the number of [ H]thymidine-labeled
nasal epithelial cells between exposed and control groups after 3
days and 13 weeks formaldehyde exposure. After 13 weeks, the
rats from each of the histopathology groups were sacrificed, and
the posterior surface of the nasal cavity was sectioned and
examined. The animals in the 4 ppm group intermittently exposed
group exhibited significantly higher incidences than controls of
treatment-related histopathological changes (diffuse
disarrangement and squamous metaplasia). An increased incidence
of these lesions were not found in the group receiving the same
total dose of formaldehyde (2 ppm continuously exposed for 8
hours) or in any of the other groups. Thus the rate (the authors
call this "concentration") at which a dose is applied is more
important than the total dose in producing lesions commonly
associated in rats with formaldehyde exposure. As a general
observation, this study points to the possibly greater importance
of concentration, rather than total dose in determining the
response of rats to formaldehyde.
In another study of dose rate by Feron et al. (1988), 465
male wistar rats (albino SPF) were assigned to 9 different
groups. Exposure regimens and observation periods are as
follows.
Formaldehyde
Concentration Exposure Duration (6 hours/day, 5
days/week)
0 ppm 4 weeks 8 weeks 13 weeks
10 ppm 4 weeks 8 weeks 13 weeks
20 ppm 4 weeks 8 weeks 13 weeks
Follow-up 126 weeks 122 weeks 117 weeks
Slightly lower body weights were observed in rats exposed to
10 ppm for 8 and 13 weeks; however, those animals exposed for 4
weeks at that concentration were not significantly affected. In
all instances, rats exposed at 20 ppm had lower body weights
compared to controls. Regardless of exposure duration, rats
exposed to 10 ppm showed incidence of hyperplasia and stratified
squamous metaplasia higher than in the controls. The incidence
was higher in those animals exposed for 13 weeks. Rats exposed
33
-------
to 20 ppm showed an increased incidence of rhinitis and
replacement of the olfactory epithelium by respiratory or
regenerating olfactory epithelium (in addition to hyperplasia and
stratified squamous'\fn£taplasia) when compared with the controls.
Two squamous cell carcinomas were found in the control group
exposed for 8 weeks, 1 each in the 10 ppm group exposed' for 8
weeks and 13 weeks, and 5 in the 20 ppm group (1 each in the 4-
and 8-week exposed groups and 3 in the group exposed for 13
weeks). The 20 ppm group also exhibited 2 polypoid adenomas (1
each in the 4- and 8-week exposed groups), 1 carcinoma in situ
(13-week exposed group), 1 cystic squamous cell carcinoma (13
week-exposed group), and 1 ameloblastoma (13 week-exposed group).
(It was not possible to determine from the study report if any of
the four different types of lesions at the 20 ppm group occurred
in the same animals).
The highest incidence occurred in the groups exposed for the
longest period of time (13 weeks) and at the highest
concentration (20 ppm). There is also a slight suggestion from
the data that the total dose of formaldehyde may be a predictor
for incidence of squamous cell carcinomas; however, the actual
tumor counts are small and the periods for which the animals were
exposed are too short for definitive determinations of the
predictive potential of total dose versus concentration to be
properly assessed.
4.3.6 Summary and Conclusions
Recent mechanistic studies have focused on the roles of DNA-
protein cross-linking, cell proliferation, mucociliary apparatus,
and concentration versus dose in formaldehyde-induced
carcinogenesis.
A study by CUT demonstrated that formaldehyde produced DNA-
protein cross-linking in the rat nasal mucosa in a nonlinear
manner. In view of the chemical reactivity of the compound, a
plausible hypothesis is that formaldehyde may act by DNA-protein
cross-linking and/or other direct genome-damaging reactions.
Formaldehyde may also act in ways other than DNA-protein cross-
linking or directly damaging the genome; however, additional
issues that remain unresolved include determination of the types
of mutations and the relationship between DNA-protein cross-links
and mutagenesis in the rat nasal mucosa; and the possible role of
other effects of formaldehyde on the genome in the carcinogenesis
process.
Acute exposure studies show that cell proliferation in the
rat was enhanced significantly at formaldehyde concentrations of
1 ppm or greater. The dose response for cell proliferation shows
a steep increase from 3 to 10 ppm. An important observation is
that the cell proliferation in the rat is sustained only at high
34
-------
levels of exposure (10-15 ppm). The results of chronic studies
show an apparent correlation between sustained, increased levels
of cell proliferation and nasal tumor induction. These results
lend support to the hypothesis that cell proliferation plays an
important role in formaldehyde-induced nasal carcinogenesis.
These results also .point to cell proliferation as a contributor
to the observed nonlinear dose-response for formaldehyde-induced
nasal tumors in' rats,,' However, as in the acute studies, cell
proliferation responses observed in the chronic studies lack
site-specificity with respect to tumor induction; similar
increases in cell proliferation and cytotoxicity occurred also at
the medial aspect of the maxilloturbinate, a region where no
nasal tumors were seen. These results, therefore, indicate that
sustained increases in cell proliferation may not be solely
responsible for the increase in the predisposition of the nasal
cavity to formaldehyde carcinogenesis.
Recent studies of monkeys and of human tracheal tissues
indicate that the proliterative response in these species is
qualitatively similar to that in rats.
Although a dose-dependent effect of formaldehyde on the
mucociliary apparatus of the rats has been established, it is not
known to what degree the mucus layer protects against the
cytotoxic effects of formaldehyde, especially in humans.
Recent studies have also examined the relative importance of
air concentration and total dose in formaldehyde cytotoxicity and
cell proliferation. There is evidence to suggest that
concentration is of greater importance than total dose in
eliciting toxic responses.
4.4 Epidemioloqic Studies
The EPA (1987) review of epidemiologic evidence based on 28
studies concluded that "limited" evidence existed for an
association between formaldehyde and human cancer using criteria
as defined in EPA's Cancer Risk Assessment Guidelines (EPA,
1987). The evidence is based mainly on well-conducted cohort and
case-control studies of populations occupationally exposed to
formaldehyde during chemical production, in textiles, in plastics
production, and in various wood industries, one study examined
residency in mobile homes.
Three studies were designed specifically to detect moderate
elevations in underlying cancer risk among populations with
formaldehyde exposure. These studies (Blair et al., 1986, 1987;
Stayner et al., 1988; and Vaughan et al., 1986a, 1986b) observed
statistically significant elevations in risk of site-specific
upper respiratory cancers (nasopharynx or buccal cavity) with
measures of formaldehyde exposure. In all three studies, the
35
-------
site-specific cancer risks appeared to increase with increasing
exposure level or with increasing duration of exposure.
Other investigations (Olsen et al., 1984; Hayes et al, 1986)
showed possible increases in sinonasal cancer, while others
(Partanen et-*i\, 19,85; Bertazzi et al., 1986; Blair et al.,
1986, 1987; and Acheson et al., 1984) indicated that lung cancer
also may be associated with occupational exposure to
formaldehyde. The risk associated with sinonasal cancer appeared
to be specific for the histologic type, squamous cell carcinoma.
The relative risks observed for upper respiratory tract cancers
in all the reviewed studies ranged from just above 1.0 (a risk of
1.0 implies no association between exposure and disease) to 3.0,
depending on the site.
Several of the epidemiologic studies (Blair et al., 1987;
Hayes et al., 1987; Olsen et al., 1984), also, suggested that
nasopharyngeal and sinonasal cavity cancer risks may be enhanced
with simultaneous exposure to particulates or wood dust. There
is limited evidence for an association between wood dust exposure
itself and sinonasal cavity cancer (IARC, 1987). Both Hayes et
al. (1987) and Olsen et al. (1984) observed that the risk
(statistically significantly elevated) associated with combined
formaldehyde and wood dust exposure appeared higher than the
individual risks associated with wood dust exposure, adjusted for
formaldehyde exposure, or 'for formaldehyde exposure, adjusted for
wood dust exposure. Risks in analyses which attempted to control
for wood dust exposure, however, were generally not statistically
significant.
As discussed in EPA (1987), elevations in risks from cancers
of the nasopharynx, nasal cavity and sinuses, and lung were not
detected in every reviewed study. One reason the epidemiologic
evidence on formaldehyde can appear inconsistent is that many of
the previously reviewed studies had limited ability to detect
small to moderate elevations in site-specific cancer risks. This
is particularly true for those sites with lower incidence rates
such as cancer of the nasal cavity and sinuses. Contributing
factors that reduce a study's detection ability include small
study populations, shortened follow-up periods which did not
consider sufficient disease latency, and a low proportion of
highly exposed individuals. In addition, errors in the exposure
measurements introduce a bias which limits the detection ability.
The EPA (1987) review also discussed studies of populations
occupationally exposed to formalin, a solution containing
formaldehyde. These investigations reported excesses of
leukemia, and cancer of the brain and prostate were reported in
embalmers, morticians, anatomists, and pathologists (Harrington
and Shannon, 1975; Harrington and Oakes, 1984; Levine et al.,
1984; Matanoski, 1982; Stroup, 1984; Walrath and Fraumeni 1983;
Walrath and Fraumeni- 1984). The biological support for an
36
-------
association with cancer at these sites was weak. In addition,
the contribution of other agents such as viruses and certain
solvent exposures was considered more likely with populations
exposed to formalin than with the industrial populations.
'\ ' \
* • *
Since publication of EPA (1987) , eleven new reports have
become available. Four of these studies contain new data (Roush
et al., 1987; Stern et al., 1987; Malker et al., 1990; Hayes et
al., 1990). The remaining are either reanalyses of the Blair et
al. (1986, 1987) study (Collins et al., 1988; Sterling and
Weinkam, 1988; Robins et al., 1988; Blair et al., 1990a), or
further investigations of existing data sets (Gerin et al., 1989;
Partanen et al., 1990). Data presented by Stayner et al. (1988)
were already reviewed in EPA (1987) and will stand as discussed
there. In addition to the above reports, three reviews have been
published (IARC, 1987; UAREP, 1988; Blair et al., 1990b).
This section provides an evaluation of the new reports and
reviews, along with a discussion of exposure information and the
total epidemiologic evidence on formaldehyde. More detailed
information regarding the new reports and selected existing
studies (as reviewed in EPA (1987)) are presented in Appendix A.
The tables in Appendix A contain a description of the study
design and population, the risks associated with site-specific
cancers, and an estimate of exposure levels as the study authors
described them.
4.4.1 Case-Control Studies
Gerin et al. (1989) conducted a case-control analysis to
further investigate possible associations of occupational
formaldehyde exposure and cancer at specific sites. Cases and
controls came from a study which was originally designed to
generate hypotheses on many occupational exposures and 14 site-
specific cancers (Siemiatycki et al., 1987a). Cancers of the
nasopharynx and sinonasal cavity were not included in either the
case or control series. For the present analysis, Gerin et al.
(1989) assessed the job histories of all cases and controls for
potential formaldehyde exposure.
No odds ratios (OR) were significantly elevated in this
study. These investigators, however, observed an apparent
increased odds ratio between adenocarcinoma of the lung and long
duration-high occupational exposure to formaldehyde (OR=2.3, 95%
confidence interval: (0.9, 6.0)). The long duration-high
exposure classification was believed most likely a time-weighted-
average exposure below 1.0 ppm. This apparent increase did not
meet statistical significance at the 0.05 level in an analysis
which took into account age, ethnic group, socio-economic status,
cigarette smoking, and white-collar/blue-collar status. Although
no trend analysis was presented by Gerin et al. (1989), for those
with long durations of exposure alone, the risks for
37
-------
adenocarcinoma of the lung appeared to increase with increasing
exposure level (long-low, OR = 0.8, 95% C.I.: (0.3, 1.3); long-
medium, OR = 0.8, 95% C.I.: (0.4, 1.6); long-high, O.R. = 2.3,
95% C.I.: (0^9-v 6,a))T. The authors concluded that the apparent
increase in adenocarc'inoma of the lung may be indicative of
increased risk with formaldehyde exposure. They noted, however,
that this finding may be due to chance.
Roush et al. (1987) conducted a case-control study
specifically designed to examine associations between sinonasal
and nasopharyngeal cancer cases and possible occupational
formaldehyde exposure. They selected cases covering the last 41
years from the Connecticut Tumor Registry. Controls were drawn
from men dying in Connecticut in the same period. Occupational
exposure information was indirectly ascertained for each of the
cases and controls from death certificate and city directories.
These investigators did not observe statistically significant
increases in risk between sinonasal cavity cancer and probable
occupational formaldehyde exposure, although an apparently
elevated risk remained after taking into account a 20+ year
latency and probable high formaldehyde exposure.
An apparently elevated odds ratio was observed for
nasopharyngeal cancer when there had been probable formaldehyde
exposure at high levels 20 or more years prior to death (OR =
2.3, 95% C.I.: (0.9, 6.0)). This odds ratio appeared to increase
with age (68+ years vs. <68 years) and was statistically
significantly elevated for the older age group (OR =4.0, 95%
C.I.: (1.3, 12.0)).
These authors concluded, for the older age and higher
exposure category, that the elevated odds ratio suggested an
association between cancer at this site and formaldehyde. They
attributed the,increased risk partially to lower background
nasopharyngeal cancer risks for males in this age group. The
detection ability of this study may have been lowered by several
factors, including the availability of only indirect information
on occupational exposure. In addition, formaldehyde exposures
from other-than-occupational sources, for instance, from
residential dwellings, could not be ascertained in this study,
nor could possible confounding exposures such as smoking.
Partanen et al. (1990) examined possible associations
between formaldehyde and respiratory cancer in a nested case-
control study of 136 respiratory cancers among 7307 male Finnish
woodworkers. These men were employed in jobs in particleboard,
plywood, construction carpentry, furniture manufacturing, and
glue manufacturing plants, and in sawmills. In a previous study
(Partanen et al., 1985), these investigators found an apparently
increased odds ratio (OR=1.4, p>0.10) between respiratory cancer
and formaldehyde in a nested case-control study based upon a
cohort of 3805 of these workers. For the current study, Partanen
41
38
-------
et al. augment their original cohort to include other wood
working facilities and they update mortality an additional year,
until 1982. As in their earlier study, three controls were
matched to each case according to year of birth.
'* ' >
After accounting for a minimum latency period of 10 years,
smoking, and vital status at the time of data collection,
Partanen et al. observed an apparently elevated odds ratio for
respiratory cancer (OR=1.4, 90% C.I.: 0.4, 4.1) with exposure to
cumulative formaldehyde (either dustborne or as a gas) (£3 ppm-
months). It appeared from other analyses examining upper
respiratory and lung cancer separately, that the excess risk
associated with cumulative formaldehyde exposure concentrated
into the category of upper respiratory cancer (OR=2.4, 90% C.I.:
0.4, 13.2; adjusted for vital status and a 10 year latent period)
rather than to lung cancer (OR=0.9, 90% C.I.: 0.3, 3.0; adjusted
from vital status, a 10 year latent period, and smoking).
Analyses examining exposure to only dustborne formaldehyde were
similar to analyses examining total formaldehyde. In addition,
no exposure-response relationships were observed between
respiratory cancer and either level, duration, or cumulative
formaldehyde exposure. In all these analyses, Partanen et al.
did not control for wood dust exposure.
Partanen et al. (1990) believed these results are compatible
either with chance or with a weak elevated risk mainly due to
cancers of the upper respiratory organs. Identified limitations
in this study include low power, inconsistencies between the data
from the original cohort and the augmented cohort, possible
raisclassification of early exposure, pooling of different
"respiratory" cancers, respondent bias regarding information on
smoking, and possible confounding by phenol. Phenol exposure was
highly correlated with formaldehyde since glue formulations in
the manufacturing of plywood contained both formaldehyde and
phenol.
4.4.2. Proportional Mortality Studies
Hayes et al. (1990) conducted a proportional mortality ratio
(PMR) study of embalmers and funeral directors in the U.S. In
all, 3649 white and 397 nonwhite male deaths between 1975 and
1985 were identified from -licensing boards and state funeral
directors' associations, from occupational information on death
certificates, and from a listing of deaths among members of the
National Funeral Directors' Association. Proportions of deaths
for 1975-1985 in the U.S. population were used as referents. The
investigators provided exposure information from recently
conducted exposure monitoring of 24 embalmings in a mortuary
college. Formaldehyde levels averaged over the entire embalming
procsdure ranged from 0.98 ppm (associated with high ventilation)
to 3.99 ppm (for low,ventilation); total dust levels averaged
0.42 mg/m (range 0.0*7. to 0.78 mg/m ). Wipe samples in the
39
-------
embalming room contained blood contamination, suggesting that
aerosolization of blood and other biologic tissue may occur. The
authors use these exposure measures as qualitative indicators of
exposure for deaths in this study.
Statistically significantly elevated proportions of deaths
were found from the following causes: all cancers (whites,
PMR=107, 900 deaths, 95% C.I.: 101-115), ischemic heart disease
(whites, PMR=113, 1418 deaths, 95% C.I.: 107-119; nonwhites,
PMR=145, 135 deaths, 95% CM.: 122-172), and suicide (whites,
PMR=130, 74 deaths, 94% C.I.: 102-164). Significant decreases in
the proportion of deaths for both races were noted for diseases
of the respiratory system (whites, PMR=85, 233 deaths, 95% C.I.:
74-97; nonwhites, PMR=53, 13 deaths, 95% C.I.: 28-91). Among
neoplastic deaths, statistically significant increases were
observed for cancers of the colon (nonwhites, PMR=231, 16 deaths,
95% C.I. 132-376) and lymphatic and hematopoietic system (whites,
PMR=131, 100 deaths, 95% C,I.: 106-159; nonwhites, PMR=241, 15
deaths, 95% C.I.: 135-397). These investigators also noted that
the risk associated with nasopharyngeal cancer appeared elevated
for whites (PMR=189, 3 deaths, 95% C.I.: 39-548), but the
elevation was not statistically significant. Only one death due
to nasopharyngeal cancer was observed among nonwhites, whereas,
only 0.25 would have been expected.
The authors believed the apparent elevated proportions of
death due to nasopharyngeal cancer and leukemia were consistent
with previous observations in formaldehyde-exposed industrial
cohorts and other studies of professionals (embalmers,
morticians, pathologists, and anatomists). This study is limited
by its study design. Since the comparison is based on a
proportion, if the number of deaths other than the cause of
interest is low, then the PMR for the cause of interest may be
artificially inflated. In addition, selection bias related to
leukemia diagnosis could not be discounted and lack of specific
exposure information limits these conclusions.
4.4.3 Cohort Studies
Stern et al. (1987) conducted a cohort study of 9,365
leather tannery workers. The exposures of concern were chrome
and other numerous chemicals, including formaldehyde.
Formaldehyde was detected in measurable levels in the tannery's
finishing department. The investigators observed one death (0.4
expected) due to squamous cell carcinoma of the nasal cavity.
This person had worked more than 18 years in the finishing
department. Although formaldehyde could not be specifically
implicated, these investigators concluded that the death was most
likely attributable to an occupational origin since it occurred
sufficient years after entry to the department to be consistent
with an induction period for occupational nasal carcinoma.
40
-------
Malker et al. (1990) examined the relationship between
nasopharyngeal cancer incidence and several occupational
groupings with potential formaldehyde exposure. Incident male
cases of nasopharyngeal cancer were identified from the Swedish
Cancer-Environment Registry for the period 1961-1979. Expected
numbers of cases were derived using age- and sex-specific cancer
incidence rates applied to the distributions of the various
occupational and industrial categories. Employment was defined
for 1960 from information given to the cancer registry. The
authors observed statistically significantly elevated (p<0.05)
standardized incidence ratios (SIR) for men employed in
fiberboard plants (SIR=3.9, 95% C.I.: 1.3, 9.3) and shoe repair
(SIR=4.0, 95% C.I.: 1.0, 9.9). No statistically significantly
elevated risks were seen among textile workers, furniture makers,
or chemical workers.
These investigators consider the association between
nasopharyngeal cancer and fiberboard employment as noteworthy,
given previous reports of elevated risks between cancer at this
site and formaldehyde exposure. The authors further believe that
the lack of an association between other job categories with
exposure to formaldehyde and nasopharyngeal cancer may reflect
limitations in the job coding scheme used by the cancer registry.
Four groups of investigators have reexamined the Blair et
al. (1986, 1987) data, using either different exposure measures
or statistical methodology (Collins et al., 1988; Sterling and
Weinkam, 1989, 1990, personal communication; Robins et al., 1988;
Blair et al, 1990a). Blair et al. (1986, 1987) studied 26,561
workers from 10 U.S. plants with occupational formaldehyde
exposure through manufacture of formaldehyde, resins, laminate,
molding compounds, photographic films, or particleboard. Blair
et al. (1986) examined the relationship between mortality and
formaldehyde exposure as categorized three ways: 1) time-
weighted average (TWA), 2) peak exposure, and 3) cumulative
exposure in ppm-years.
Collins et al. (1988) reanalyzed the cohort data of Blair et
al. (1986, 1987) using two approaches. The first partitioned the
nasopharyngeal mortality experience into that of the Cy^namid
cohort and that of the remaining nine plants. The second
examined nasopharyngeal cancer mortality and simultaneous
formaldehyde and particulate exposure utilizing a revised
exposure classification scheme. Formaldehyde exposure was
considered important only when it simultaneously occurred in the
presence of particulates. Blair et al. (1986, 1987), on the
other hand, classified an individual as particulate exposed if he
was ever exposed to particulates at any time in his working
history. For a particulate-exposed study subject in Blair et al.
(1986, 1987), all formaldehyde exposure, whether incurred in the
presence or absence of particulates, was summed in the cumulative
exposure measure.
41
-------
Collins et al. concluded from the first approach that the
excess in nasopharyngeal cancer mortality observed by Blair et
al. (1986, 1987) was limited in the Cyanamid plant. By using
their exposure classification (the second approach), Collins et
al. (1988) concluded"that the dose-response relationship between
increasing cumulative formaldehyde exposure and increasing
nasopharyngeal cancer risk apparent in Blair et al. (1986, 1987),
disappeared. Collins et al. (1988) drew support for their
conclusion from the.lack of a statistically significant trend in
risks with increasing formaldehyde-particulate exposure.
The Collins et al. (1988) reanalysis appears well-conducted,
however, strong conclusions cannot be drawn. Their reanalysis is
based upon few deaths and as these deaths are distributed across
four exposure categories, the risks for each category are subject
to wide variation, and become very unstable. For this reason,
any strong interpretation of the Collins et al. (1988) reanalysis
is difficult.
Sterling and Weinkam (1988; 1989; personal communication
with Dr. James Weinkam, April 12, 1990) performed several
analyses examining deaths due to all cancers, all respiratory
cancers, and all lung cancer observed in Blair et al. (1986,
1987) by reclassifying exposure using a time-integrated exposure
score, and by comparing mortality among the higher-exposed to
that of lower-exposed (an internal comparison group). Blair and
Stewart (1989) and Blair et al. (1990a) questioned the Sterling
and Weinkam (1988; 1989) conclusions with respect to the correct
number of person-years and the correct number of deaths. For
example, Blair et al. (1986) observed 280 deaths, yet the
Sterling and Weinkam (1989) reanalysis was based upon 299 lung
cancer deaths. The 299 deaths included all respiratory system
cancer deaths, which includes cancers of the sinonasal cavity,
larynx, and lung (Weinkam, personal communication).
A reanalysis has been performed based only on the 280 lung
cancer deaths (personal communication with Dr. James Weinkam,
April 12, 1990). Cumulative exposure to formaldehyde was ranked
into four hierarchical categories and odds ratios for cumulative
exposure to formaldehyde and lung cancer were as follows: <0.1
ppm-yr, OR = 1.0; 0.1-0.5 ppm-yr, OR = 1.12, 95% C.I.: (0.76,
1.84), 0.5-2.0 ppm-yr, OR = 1.12, 95% C.I.: (0.76, 1.84), and 2+
ppm-yr, O.R. = 1.46, 95% C.I.: (0.87, 2.49). Exposure-response
relationships between lung cancer and cumulative formaldehyde
were also examined using logistic regression analysis. /'This
analysis suggested an increase in lung cancer risk with
increasing cumulative exposure, however, the estimated odds ratio
for the highest cumulative exposure was not statistically
significant. Weinkam believes that this analysis along with the
observations in thev!988 and 1989 articles supported the
conclusion that lung cancer mortality is associated with
formaldehyde exposure. This conclusion is limited since the 1988
.*
42
-------
and 1989 analyses appeared to be based upon a incorrect number of
site-specific cases, and the most recent results did not achieve
statistical significance.
Robins et al. (£988) reanalyzed the lung cancer and
nonmalignant respiratory disease mortality data of Blair et al.
(1986) using analytical methods newly developed by the authors
that accounted for the survivor effect. These analyses were
carried out to address comments that the lack of a dose-response
trend for lung cancer observed in Blair et al. (1986) may be due
to bias from -the survivor effect. The survivor effect is the
tendency of workers at increased risk of death to preferentially
terminate employment earlier than those not at increased risk.
These workers, thus, tend to have less cumulative exposure. If a
survivor effect is operating, an artifactual inverse relationship
will be seen between cumulative exposure and the site-specific
death.
By examining mortality rates among workers who terminated
employment and those that did not, Robins et al. (1988) observed
that nonmalignant respiratory disease mortality was higher among
workers who terminated employment than among those workers who
continued to worker over a 15 year period. This finding supports
the hypothesis that survivor effect bias was present in the Blair
et al. (1986) analyses examining nonmalignant respiratory
disease. In their examinations of lung cancer, Robins et al. did
not find the effects of survivor bias when formaldehyde exposure
was assessed using cumulative exposure, peak exposure, and
combined exposure to formaldehyde and particulates. They
believed these findings confirmed the Blair et al. (1986)
original conclusions that formaldehyde was not associated with
lung cancer mortality.
One limitation of the Robins et al.'s reanalysis of lung
cancer is that it is based upon the entire cohort. This includes
both salaried and wage workers. Blair et al. (1986) did not
observe an increased lung cancer risk in the entire cohort, only
among workers classified as wage workers and among wage workers
with 20 or more years since first exposure. Any small lung
cancer effect limited to a selected subgroup such as wage workers
may be diluted in analyses which examine the lung cancer
experience of the entire cohort.
Blair et al. (1990a) examined in greater detail than in
their 1986 article exposure-response relationships for lung
cancer mortality and formaldehyde, and for lung cancer and
exposure to other substances found in conjunction with
formaldehyde. These analyses were limited to wage workers. The
authors did not observe any consistent pattern of increasing
trends between lung cancer and formaldehyde exposure, as assessed
by peak, average, intensity, duration, and cumulative exposure.
They reported positive exposure-response associations between
43
-------
lung cancer and cumulative formaldehyde level for persons hired
before the study's start date, and for those workers with
exposure to formaldehyde and particulates from resin and molding
compound operations. Neither of these associations was
statistically significant.
In analyses that examine lung cancer mortality by specific
manufactured products, the SMR for plywood was 180; molding
compounds or resins, SMR=120; and plastic products, SMR=170
(confidence intervals could not be calculated due to missing
elements of observed/expected ratios in the paper). Among
specific plants (work sites), statistically significant
elevations were noted for three plants (Nos. 1, 6, and 9) for
those workers with any formaldehyde exposure. Particulate
exposure would have been expected for plants 1 and 6. Exposure-
response associations were suggested for three individual plants
(Nos. 1, 3, and 4) with cumulative exposure. The trend was
statistically significant for one plant (No. 3); however, the
trends for the other two plants were not.
In analyses which examine lung cancer mortality and exposure
to formaldehyde alone, and to substances found in conjunction
with formaldehyde, statistically significantly elevated SMRs were
noted among wage workers with a 20-year latency for exposure to
antioxidants, asbestos, hexamethylenetetramine, melamine,
plasticizers, urea, and wood dust. These SMRs were greater than
or equal to 150. In addition, lung cancer mortality was
statistically significantly elevated among formaldehyde-exposed
who also had exposure to other substances (SMR=140, 95% C.I.:
116, 167). Persons exposed only to formaldehyde had an SMR of
103 (95% C.I.: 83, 127). Statistically significant positive
exposure-response trends were noted for melamine and urea.
Positive trends were also suggested for phenol and wood dust, but
they were not statistically significant.
Blair et al. (1990a) concluded that the lack of clear
exposure-response trends for lung cancer in this study was
consistent with results of their previous study (Blair et al.,
1986). They feel that the small, but not statistically
significant positive association for persons with exposure to
formaldehyde and particulates may reflect the risks associated
with substances in resin and molding compound operations. This
conclusion was further supported by positive exposure-respoise
trends for several compounds (e.g., melamine and urea) found in
the resin and molding compound operations. They also believed
that interpretation of the apparent exposure-response gradient
for exposure to formaldehyde and particulates from resin and
molding compound operations is difficult because of small
increases in risk, lack of statistical significance, and
inconsistency among individual plants.
44
-------
4.4.4 Evaluation of New Studies
The reports received since the EPA (1987) assessment focused
on cancer o£-.-the ^respiratory tract following inhalation exposure.
Dermal exposure, i-n* addition, was probable in one study (Hayes et
al., 1990). Four new studies were added (Stern et al., 1987;
Malker et al., 1990; Roush et al., 1987; and Hayes et al., 1990).
The study of Roush et al. (1987) was specifically designed to
assess the role of formaldehyde exposure, thus, the conclusions
of this study carry greater weight than the other new studies.
The remainder were reanalyses of previously published studies
(Collins et al., 1988; Sterling and Weinkam, 1988, 1989; Robins
et al., 1988; and Blair et al., 1990), or further investigations
of existing data sets (Gerin et al., 1989; Partanen et al.,
1990).
The new studies provide small additional support to the
conclusions made in EPA (1987) that suggested an association
between formaldehyde and respiratory tract cancer. This new
evidence, however, does not provide causal evidence of human
cancer. These new studies are limited because of lack of power
to detect small to moderate excess cancer risks associated with
exposure and possible exposure misclassification.
Two of the new studies investigated sinonasal cancer (Roush
et al., 1987; Stern et al., 1987). Neither showed any indication
of a formaldehyde-associated effect. Nasopharyngeal cancer was
evaluated in several of the analyses. Roush et al. (1987) found
the highest risk among those who were older, who had the longest
latency, and with the highest occupational exposure. This
finding suggests that an occupational agent, most likely
formaldehyde, is responsible for the observed increased cancer
risk. Malker et al. (1990) observed a statistically significant
elevation in nasopharyngeal cancer risk with employment in
fiberboard plants where formaldehyde and particulate (wood dust
and resin) exposures are expected. No such observations were
noted for other industries with potential formaldehyde exposure.
Exposure is not as well characterized in this study as in those
where information exists on individual jobs held by each worker,
and limits the Malker et al. (1990) observations. Hayes et al.
(1990) noted an elevation in the proportion of nasopharyngeal
deaths among embalmers. This elevation was not statistically
significant (p>0.05).
The reanalysis of the Blair et al. (1986, 1987) data using
different exposure measures or statistical methodology, including
those done by Collins et al. (1988), Sterling and Weinkam (1988;
1989), Robins et al. (1988) and Blair et al. (1990a), enables an
examination of the stability of prior conclusions and allows the
generation of alternative interpretations and limitations around
these alternatives'^ Different conclusions than those of Blair et
al. (1986, 1987) have been drawn by two investigators (Collins et
45
-------
al., 1988; Sterling and Weinkam, 1988, 1989, personnel
communication). Conclusions from two other reanalyses (Robins et
al., 1988; Blair et al., 1990) were similar to the original
investigationr 'With\respect to lung cancer, Sterling and Weinkam
(1988, 1989, personal communication) believed their analyses
suggested an exposure-response relationship between increasing
lung cancer risk and increasing cumulative formaldehyde. On the
other hand, both Blair et al. (1990) and Robins et al. (1988)
believed their analyses did not demonstrate that an exposure-
response relationship existed with formaldehyde, with respect to
nasopharyngeal cancer, Collins et al. (1988) did not find a
statistically significant relationship between increasing
nasopharyngeal cancer risks and increasing cumulative
formaldehyde using an exposure classification scheme which
considered important only formaldehyde exposure when it
simultaneously occurred in the presence of particulates.
The EPA has carried out other analyses on data presented by
Collins et al. (1988) (see Appendix A) using a Poisson trend
statistic (Armitage, 1955; Tarone, 1982; as cited in Chapter 3 of
Breslow and Day, 1980). Analyses which are based upon exposure
as defined by interval midpoints (0 ppm/yr, 0.25 ppm/yr, 3.0
ppm/yr, and 12.5 ppm/yr) and by rank score (1, 2, 3, 4) show
statistically significant trends between increasing
nasopharyngeal cancer risk and increasing formaldehyde exposure.
These results support the Blair et al. (1987) previous
observations of an association between formaldehyde and
nasopharyngeal cancer.
New investigations on lung cancer do not provide much
support that formaldehyde, by itself, is responsible for excesses
in lung cancer. No statistically significant increases in risk
were noted in Gerin et al. (1989) and Partanen et al. (1990) nor
in reanalyses (Blair et al., 1990a; Robins et al., 1988; Sterling
and Weinkam, personal communication) of Blair et al. (1986). One
study (Blair et al., 1990), however, suggests that combined
exposures, such as those in operations involving resins and
molding materials, may be associated with excesses in lung
cancer. This analysis observed positive trends with exposure to
two resins (urea and melamine), where exposure to both gaseous
formaldehyde and particulates is expected.
4.4.5 Non-EPA Reviews of the Formaldehyde Literature
The International Agency for Research on Cancer (IARC)
(1987) reviewed essentially the same literature as EPA (1987) and
concluded that there was "limited" evidence of carcinogenicity to
humans according to the lARC's Classification of the Evidence for
Carcinogenicity. IARC defines "limited evidence of
carcinogenicity" as '>"a positive association has been observed
between exposure to the agent and cancer for which a causal
interpretation is considered to be credible, but chance, bias or
46
-------
confounding could not be ruled out with reasonable confidence"
(IARC, 1987).
IARC (1987) believed the evidence was strongest for cancers
of the noseband' nasopharynx since exposure-response relationships
had been observed in several studies. This group also noted an
excess in lung cancer in several studies, but that exposure-
response gradients were not consistently demonstrated across all
studies. The excesses in mortality from leukemia and brain
cancer were primarily limited to occupations with formalin
exposure, but IARC believed that factors other than formaldehyde
might have increased the risk for these cancers.
A panel under the auspices of the University Associated for
Research and Education (UAREP) (1988) reviewed the same body of
literature as IARC (1987) and EPA (1987) using a metanalysis
approach. The UAREP panel commented only on the determination of
causality. Unlike the IARC and EPA, the UAREP panel did not
attempt to categorize the epidemiologic evidence other than
whether causality could be established. The panel concluded that
a causal relationship has not been established for cancer at any
site. In addition, the panel noted that if such a causal
relationship exists, the excess risk must be small. The panel
noted elevated risks in nasopharyngeal cancer with formaldehyde
exposure in several studies, and concluded that the evidence for
causality was weak. With respect to observed excesses in nasal
cavity and sinus cancers and any formaldehyde exposure, several
studies suggest an approximate doubling of the risk, while other
studies could not exclude an elevation of this size. Overall, the
panel concluded that the presence or absence of an association
could not be firmly established. With respect to lung cancer,
the panel thought the evidence was not consistent and did not
indicate a causal association with formaldehyde exposure.
For sites which are not directly in contact with
formaldehyde (denoted non-topical sites in the UAREP report), the
panel stated that the rapid metabolism of formaldehyde makes it
unlikely that formaldehyde is the agent responsible for increased
brain tumors observed in the group that uses formalin. For the
excesses in leukemia observed in several studies of anatomists,
embalmers, and pathologists, the panel concluded that
socioeconomic factors influencing diagnosis may explain the
elevations observed in these groups.
Blair et al. (199Qb) performed a metanalysis on essentially
the same body of literature as reviewed by IARC (1987), UAREP
(1988), and EPA (1987) with the addition of more recent findings,
either published or in press. From the metanalysis, Blair et al.
found excesses in deaths due to cancers of the nasal cavities,
nasopharynx, lung, and brain, and due to leukemia. The
investigators believed that a causal role for formaldehyde was
47
-------
most probable for cancers of the nasopharynx and, to a lesser
extent, the nasal cavities.
Blair et al. (I990b) derived their support for the
conclusion regarding nasal cavity cancer from statistically
significant^eirBvatiqns in nasal cavity cancer risk, from the
apparent specificity of the association with squamous cell
carcinoma, and from histological changes in the nasal mucosa of
woodworkers and chemical workers which corresponded to lesions
observed in rodents. Potential confounding from other exposures
(i.e., wood dust) in these studies still remained a problem,
however. With regard to nasopharyngeal cancer, their conclusions
were based on statistically significant excesses observed in
three studies and the observation of a dose-response trend
between increasing risk and increasing formaldehyde exposure
observed in two of these studies. In addition, they believed
that the occurrence of cancer at a site with possible direct
contact with formaldehyde made sense in terms of the biological
knowledge. Small numbers, inconsistency among studies, and a
possible independent role for particulates precluded definitively
labeling formaldehyde as a nasopharyngeal, carcinogen.
Blair et al. (1990b) further concluded that the excesses in
lung cancer were difficult to interpret due to inconsistencies
among studies and lack of trends with either level or duration of
exposure. In addition, the excesses of leukemia and brain and
colon cancer observed among professionals (i.e., those with
formalin exposure) were most likely not related to formaldehyde
since similar excesses were not observed among the industrial
workers.
4.4.6 Discussion and Conclusions
The accumulated epidemiologic evidence is consistent with
the conclusion reached in EPA (1987) that there is "limited"
evidence of human carcinogenicity as defined in EPA's Cancer Risk
Assessment Guidelines (EPA, 1986). .Because collectively the data
do not conclusively demonstrate a causal relationship, the
evidence is not considered "sufficient".
Increases in respiratory tract cancer are seen in several
well-designed and well-conducted studies. Observed relative
risks have not been large, risks ranged from just above 1.0 to
3.0, depending on the site. There also appears to be a gradient
of support, with greater support for cancers near the portal of
entry (cancers of the nasal cavity and sinus, and the
nasopharynx) than further down the respiratory tract (dung
cancer). Table 4-3 presents selected results from reviewed
studies cited in Appendix A.
48
-------
Authors
(year)
Table 4-3
Selected findings from epidemiologic studies examining inhaled formaldehyde
Cancer Site
Risk
Measure
Type of Exposure
Risk Ratio
(95% Confidence
Interval)3
Hayes et al.
(1986)
Olsen et al.
(1984)
Olsen and
Asnaes (1986)
Nose and Nasal Sinuses OR1
(Squamous cell carcinoma)
Nose and Nasal Sinuses OR
Nose and Nasal Sinuses OR
(Squamous cell carcinoma)
Formaldehyde; No to low wood
dust
Classification Ac
Classification Bc
Formaldehyde; Low wood dust
Classification A
Classification B
Formaldehyde and wood dust
Formaldehyde; adjusted for
wood dust
Formaldehyde; adjusted for wood
dust
2.5* (1.2°. 3.0e)
1.6 (0.9d, 2.8e)
3.0* (1.3d 6.4e)
1.9 (1.0d. 3-:6e)
2.8* (1.3d, 6.4e)
1.6
2.3 (0.9d, 5.8e)
Vaughan et al. Nasopharynx
(1986b)
Roush et al.
(1987)
Nasopharynx
Malker et al. Nasopharynx
(-1990)
OR Mobile home residency
1-9 years
10+ years
OR Formaldehyde
High level exposure
(> 1 ppm at 20+ yrs
to death)
prior
High level exposure
(> 1 ppm at 20+ yrs. prior
to death); 68t yrs. old at
(lent h
SIR* Fiberboard manufacturing
2.1 (0.7, 6.6)
5.5* (1.6, 19.4)
2.3 (0.9, 6.0)
4.0* (1.3, 12.0)
3.9* (1.1, 10.2)
49
-------
Table
(continued)
Authors
(year)
Cancer Site
Risk
Measure
Type of Exposure
Risk Ratio
(95% Confidence
Interval)3
Blair et al.
(1986, 1987)
Nasopharynx
SMR&
Collins et al.
(1988)
(reanalysis of
Blair et al.,
1986)
Hayes et al.
(1990)h
Blair et al.
(1986)
Blair et al.
(1990a)
(reanalysis of
Blair et al. ,
1986)
Nasopharynx
SMR
Nasopharynx
Lung
Lung
PMR1
SMR
SMR
Ever exposed to formaldehyde
(ppm-year)
Ever exposed to particulates
(ppm-year)
< 0.5
0.5 -5.5
> 5.5
Simultaneous formaldehyde and
particulates (ppm-year)
0
< 0.5
0.5 -5.5
> 5.5
Embalmers and funeral directors
Ever exposed to formaldehyde
(ppm-year)
300* (HO,1 653)
192 (. 5, 1114)
403 (48v 1445)
746 (81, 2408)
215 .(25, 802)
343 (37, 1204)
216 ( 3, 1113)
826* (112, 3611)
189 (39, 548)
110 (97, 125)
Wage workers, >20 years of latency 122* (106, 139)
Formaldehyde alone (ppm-year)
< 0.5
0.5 -5.5
> 5.5
Formaldehyde and other agents
(ppm-year)
< 0.5
0.5 - 5.5
> 5.5
110 ( 80, 149)
110 ( 78, 151)
74 ( 41, 124)
154* (110,, 212)
118 ( 87, 158)
154* (113,1 207)
50
-------
Table 6-3 (continued)
Authors
(year) Cancer Site
Sterling and Lung
Weinkam
(personal
comniunication-
reanalysis of
Blair et al. ,
1986)
Risk
Measure Type of Exposure
OR Formaldehyde (ppm-year)
< 0.1
0.1 - 0.5
0.5 - 2.0
> 2.0
Risk Ratio
(95% Confidence
Interval)3
1.0
1.1 (0.8, 1.8)
1.1 (0.7, 1.8)
1.5 (0.9, 2.5)
*
Acheson et al.
(19.84)
Bertazzi et al
(1986)
Gerin et al.
(1989)
Partanen et al.
(1990)
Lung
Lung
Lung
Lung
SMR All 6 plants
BIP plant
SMR Resin workers (total cohort)
Formaldehyde (a subset)
OR Formaldehyde - Long duration
Low level
Medium level
High level
OR Formaldehyde - >3 ppm-months
105 ( 91, 120)
118* (100, 137)
236* (140, 373)
136 ( 44, 315)
0.5 (0.2, 1.3)
1.0 (0.4, 2.5)
2.2 (0.7, 7.6)
0.9 (0.3d,3.0e)
Statistically significant.
a Confidence interval as presented by the nuthors or calculated by EPA, with the exception of Olsen
et al., 1984.
b Odds ratio.
*• Work history was assessed by using two independent classifications.
° Lower 90% confidence limit.
® Upper 90% confidence limit.
*• Standard incidence ratio.
6 Standard mortality ratio.
. Inhalation and dermal exposure.
1 Proportional mortality ratio.
51
-------
4.4.6.1. Exposure Consideration
Exposure to formaldehyde in the epidemiologic studies has
been primarily occupational to formalin or in the manufacture or
production of formaldehyde or formaldehyde products. Only one
study examined,.population exposure to formaldehyde, and this was
indirectly assessed \by residency in a mobile or manufactured
home.
Exposure to formaldehyde for the industrial populations
occurs in the manufacture of formaldehyde or formaldehyde-
containing products; Exposure can result while manufacturing and
handling formaldehyde, paraformaldehyde or other formaldehyde
derivatives; from gaseous formaldehyde released from resins which
are being subjected to heat or pressure processes (e.g., making
plywood and particleboard, or plastic molding compounds); or from
off-gassing from formaldehyde-containing materials (e.g.,
manufacturing apparel from fabrics treated with formaldehyde
resins which are used to impart wrinkle and crease resistance).
For these studies, formaldehyde exposure was examined directly by
quantitating formaldehyde levels or indirectly through particular
occupations where formaldehyde exposure has been know to occur.
In the settings where formaldehyde-based resins are used to make
another product such as a molded plastic compound, plywood, or
particleboard, exposure to particulates is a potential.
Particles can either be composed of the resin itself, or in the
case of plywood or particleboard manufacturing, the particles are
wood dusts with the potential for formaldehyde to become attached
to them.
Formalin is a solution containing 5% formaldehyde.
Occupational, titles with potential formalin exposure can include
morticians, embalmers, anatomists, and pathologists. Exposure to
formaldehyde for these professional is by inhalation and by
dermal absorption. In addition to formaldehyde, this group has
diverse chemical exposures, for example, to solvents and viruses.
The levels to which the studied populations were exposed to
formaldehyde are absent in all these studies. In some cases, the
investigator inferred formaldehyde exposure to the studied
population based upon data from industrial hygiene surveys
conducted in the 1970's and 1980's of similar occupational
settings, from selected funeral homes or anatomy laboratories.
It is not known whether the individuals under study did or did
not have formaldehyde exposure at the levels identified in the
industrial hygiene surveys. In other cases, the investigators do
not even attempt to infer level of exposure. In this situation,
no information on exposure level can be obtained.
4*
Only one study (Vaughan et al., 1986) examined general
population exposure to formaldehyde. Actual exposure levels
experienced by the individual under study were not known, but the
potential for formaldehyde exposure was assessed indirectly by
•i
52
-------
examining length of residence in a manufactured or mobile home.
Longer exposure durations inferred higher cumulative formaldehyde
exposure than shorter durations of exposure. Potential exposure
to formaldehyde in manufactured or mobile homes comes from off-
gassing from formaldehyde-containing materials such as
part icleboard^-and'plywood. These materials are used as
underlayment. for flooring and in wall paneling, shelves, and
cabinets.
It is difficult to infer how exposure patterns experienced
by the individuals under study influence risk. One reason is
that information on past exposure is absent from many studies.
All that can be assumed is that past exposure was most likely
higher than indicated by present monitoring data. When past
exposures are absent, it is not known whether process changes may
have altered exposure patterns.
Another reason is the difficult in knowing the most
appropriate exposure measure. Exposure in these studies has been
defined several ways, primarily as time-weighted averages or
cumulative exposure. Several investigators present some
information on peak exposure. In the Blair et al. (1986) cohort
study of workers manufacturing formaldehyde or formaldehyde-based
products, for example, peaks were examined as the highest time-
weighted average (TWA) or peak excursion level of exposure
experienced by a study participant. Peak patterns of exposure
would also be expected for occupations using formalin where
short-term high-level formaldehyde exposure would be expected
during the embalming or autopsy processes.
4.4.6.2. Evaluation of the Body of Human Evidence
The evidence for potential human carcinogenicity associated
with formaldehyde exposure rests heavily on associations with
cancers of the nasal cavity and sinus and of the nasopharynx.
The EPA (1987) concluded that cancer of the nasal cavity and
sinus may be related to formaldehyde exposure. This conclusion
was largely supported by the finding of statistically significant
elevated risks between sinonasal cavity cancer and
formaldehyde/wood dust exposure in two well-conducted case-
control studies (Olsen et al., 1984; Hayes et al., 1986). Both
investigators attempted to adjust for wood dust exposure. The
risk observed by Olsen et al. (1984) appeared elevated but was
not statistically significant. In analyses which examined only
those with no to low wood dust exposure, Hayes et al. (1986)
observed that risk associated with formaldehyde was statistically
significantly elevated under one exposure classification but not
in another exposure classification.
Additional support came from two studies (Olsen and Asnaes,
1986; Hayes et al., 1986) which noted squamous cell carcinoma as
the histological type most specifically associated with
53
-------
formaldehyde exposure. Both studies adjusted for wood dust
exposure. In these analyses, Hayes et al. (1986) showed the
highest risks were statistically significant associated with
squamous cell carcinoma, whereas, no such association was
observed between formaldehyde exposure and adenocarcinoma. The
risks associated with squamous cell carcinoma in Olsen and Asnaes
(1986) appeared elevated, but were not statistically significant.
/
Findings from four studies (Vaughan et al., 1986b; Blair et
al., 1986, 1987; Roush et al., 1987; and Malker et al., 1990)
carried out under differing study designs and exposure settings
support the finding of a possible association between
nasopharyngeal cancer and formaldehyde exposure. Nasopharyngeal
cancer risk appeared to increase with either increasing exposure
level (Blair et al., 1986, 1987; Roush et al., 1987) or with
increasing duration of exposure (Vaughan et al., 1986b).
The observation in the human studies of elevated risks due
to cancers of the nasopharynx, and the nasal cavity and sinuses
are supported by results from toxicological studies. Rats
exposed for two years to formaldehyde developed squamous cell
carcinoma of the nasal turbinates (Kerns et al., 1983).
Likewise, DNA-protein crosslinks and nonneoplastic
histopathological changes have been found in the nasopharynx and
nasal cavity of rats and monkeys (Heck et al., 1989; Casanova et
al, 1989). These observations indicate that formaldehyde reacts
with biological material at the point of contact. Human exposure
to formaldehyde is predominately by inhalation, thus, sites along
the respiratory tract are most likely to be "at risk". DPX-
protein crosslinks and histopathological changes have not been
noted in the maxillary sinuses of monkeys (Monticello and Morgan,
1989) . This finding may indicate that pooling cancers of the
nasal cavity with those of the nasal sinuses in the epidemiologic
studies may not best reflect disease in humans.
Further biological support for a possible association
between formaldehyde and cancer of the nasal cavity and sinuses
comes from three studies (Edling et al., 1988; Holmstrom et al.,
1989; Boysen et al., 1990) examining cellular changes among
formaldehyde-exposed workers. These studies observed dysplasia
and squamous cell metaplasia in nasal biopsies of woodworkers and
chemical workers exposed to formaldehyde.
Firm conclusions regarding possible associations between
nasal cavity and sinuses cancers and nasopharyngeal cancer and
formaldehyde can not be drawn. Support for nasal cavity and
sinus cancers is based on only two studies and on four studies
for nasopharyngeal cancer, with respect to nasal cavity'and
sinus cancer, elevated risks were not detected in any of the
cohort studies. For each study, the expected number of deaths
was less than three, thus, the detection levels in these studies
were greater than the risks for sinonasal cavity cancer observed
54
-------
in Olsen and Asnaes (1984), Olsen et al. (1986), and Hayes et al,
(1986). Other cohort studies did not observe any elevations in
nasopharyngeal cancer risk. The number of participants in these
studies tended be smaller than in those of Vaughan et al.
(1986b), Blair.et al. (1986, 1987), and Roush et al. (1987) and
possible limitations in these remaining studies tended to
conservatively bias*the results.
In addition, the role of wood dust and particulate exposure
remains unclear. Even though the studies of Olsen et al. (1984),
Olsen and Asnaes (1986), and Hayes et al. (1986) tried to adjust
for wood dust exposure, its influence may not have been entirely
removed. This is important since Vaughan and Davis (1990) have
recently reported an association between wood dust exposure and
increased risk of squamous cell cancer of the sinonasal cavity.
Simultaneous particulate exposure also may have a role in the
development of nasopharyngeal cancer. Two studies (Blair et al.,
1987; Malker et al., 1990) report statistically significant
elevations in nasopharyngeal cancer risk with formaldehyde and
particulate exposure. It is more difficult to assess the
contribution of particulate exposure for the case-control studies
of Roush et al. (1987) and Vaughan et al. (1986a,b) since
information on its presence is lacking.
The classification of the epidemiologic evidence relies in
small degree on observations that lung cancer may be associated
with formaldehyde exposure. Statistically significant elevations
in lung cancer relative risks have been noted in several studies:
Blair et al. (1986) (for a subgroup of white men with greater
than 20 years of latency), Acheson et al. (1984) (for a subgroup
of workers from one plant), and Bertazzi et al. (1986). Exposure
in these studies were to formaldehyde and to particulates.
Because the elevation in lung cancer risk has not been
consistently observed across all studies, these observations
carry lesser weight. In addition, further analyses of Blair et
al. (1990a) suggest that exposure to formaldehyde and other
substances might be important in the development of lung cancer.
The role of these exposures, however, needs to be further
elucidated.
Excesses in leukemia and brain cancer are observed in
studies of populations occupationally exposed to formalin; these
are most likely related to exposures other than formaldehyde.
These populations have exposures to viruses and solvents in
addition to the formalin exposure; none of the studies adjusted
for these confounding factors. Second, the absence of excesses
of leukemia and brain cancer among the industrial cohorts argues
against a role of formaldehyde. Third, the biological support
for an association with cancers at these sites was weak.
Formaldehyde is rapidly metabolized upon contact with biological
materials. No measurable increase in formaldehyde, as indicated
by DNA-protein cross-links, have been found in the blood, and
55
-------
there is an absence of induced pathology at sites distant to the
portal of entry (Heck et al., 1989; Casanova et al, 1989;
Monticello and Morgan, 1989).
4.5 Weight of Cancer Evidence
~~~ ~~ '\
4.5.1. .Human )vEVidence
The associations observed in the newer studies between lung,
nasal cavity and sinus cancer, and nasopharyngeal cancer and
formaldehyde have not been large. Point estimates for relative
risks are of the order of 3.0 or less for high exposure
scenarios. Studies reviewed in 1987 also suggested that relative
risks associated with exposure are of this magnitude.
Increased nasopharyngeal cancer risks have been observed in
four studies (Vaughan, et al.,1986; Blair et al., 1986, 1987; and
Roush et al., 1987; Malker et al., 1990) carried out under
differing circumstances. The remaining studies did not observe
any elevations in nasopharyngeal cancer risk. Three studies
(Vaughan et al. , 1986; Blair et al., 1986, 1987; and Roush et
al., 1984) were designed to detect moderate increases in
formaldehyde risks. Limitations in these remaining studies tend
to conservatively bias the results. EPA (1987) also concluded
that cancer of the nasal cavity and sinus may be related to
formaldehyde exposure based on two well-conducted studies (Olsen
et al., 1984; Olsen and Asneas, 1986; Hayes et al., 1986).
Elevated risks for nasopharyngeal, and nasal cavity and
sinus cancers were observed among various subgroups: those who
were older, those with the greatest latency since first exposure,
those with the highest duration or level of exposure, and those
with occupational particulate exposure. These observations add
support for a possible causal association between cancer at these
sites and formaldehyde exposure. In addition, the suggested
increase in metaplasia and dysplasia in workers in formaldehyde-
related industries provides biological support for a possible
association between formaldehyde and nasal cavity and sinus
cancers.
Elevations in lung cancer relative risks have been noted in
several studies (Blair et al., 1986; Acheson et al., 1984,
Bertazzi et al., 1986). However, no consistent patterns with
increasing level or duration of exposure to formaldehyde have
been observed. In addition, one study (Blair et al., 1990a)
suggests that formaldehyde, by itself, is not associated with
lung cancer, but, in conjunction with exposure to ^
compounds found in resin and molding operations (i.e., melamine
and urea), may be responsible for elevated lung cancer risks.
The studies of Acheson et al. (1984) and of Bertazzi et al.
(1986) were of populations with combined exposures to
formaldehyde and resins.
56
-------
The excesses in leukemia and brain cancer observed in those
studies of populations occupationally exposed to formalin are
most likely related to exposures other than formaldehyde.
When examined in the context of previously reviewed studies,
the studies released since 1987 support the conclusions drawn in
EPA (1987) . — That 'i,s,, limited evidence exists for an association
between formaldehyde and human cancer. Collectively, however,
the evidence does not demonstrate conclusively a causal
relationship.
4.5.2. Animal Evidence
The principal evidence indicating that formaldehyde is able
to elicit a carcinogenic response in laboratory animals comes
from three chronic inhalation studies (Kerns et al., 1983; Albert
et al., 1982 and Sellakumar et al., 1985; Tobe et al., 1985).
All three studies showed statistically significant increases in
squamous cell carcinomas of the nasal cavity of Fischer 344 rats
or Sprague-Dawley rats at comparable incidences (38-50%),
following exposure to high concentrations of formaldehyde (about
15 ppm). Tumor induction was also seen at lower formaldehyde
concentrations but the carcinogenic dose-response curve was very
steep (Kerns et al., 1983; Monticello, 1990). A tumor incidence
of 1% was seen at 6 ppm (Kerns et al., 1983) and no tumorigenic
response was observed at 2 ppm or less (Kerns et al., 1983;
Albert et al., 1982 and Sellakumar et al., 1985). Results of a
recent inhalation study by Feron et al. (1988) confirm the
carcinogenic potential of formaldehyde at high concentrations (20
ppm) following shorter term exposure (13 weeks), with a lifetime
follow-up.
As previously discussed (EPA, 1987), there are species
differences in the response to inhalation of formaldehyde. In
mice, there was a marginal tumorigenic response. Only a small
number of nasal squamous cell carcinomas were seen in male mice
exposed to high concentrations of formaldehyde (Kerns et al.,
1983) . No tumorigenic response was found in hamsters (Dalbey,
1982) . No cancer studies have been conducted in nonhuman
primates. However, two studies demonstrated that at least for
nonneoplastic lesions (hyperplasia and squamous metaplasia) , rats
and monkeys respond similarly to inhalation of formaldehyde
(Rusch et al., 1983; Monticello et al, 1989).
Recent studies also provide suggestive evidence for the
carcinogenicity of formaldehyde based on oral exposure data.
Marginal increases in incidences of forestomach and intestinal
tumors were reported in an extensive drinking water study in rats
(Soffritti et al., 1989), and similar increases in forestomach
neoplasms were reported in another drinking water study
(Takahashi et al., 1986). In two other drinking water studies in
rats (Til et al., 1989; Tobe et al., 1989), there was no
57
-------
carcinogenic evidence but hyperplastic and inflammatory changes
in the fore- and glandular stomach were reported. These studies
further support the suggestion that formaldehyde is likely to
cause toxicity at the initial site of contact but not at distant
sites.
The combfned 'data support the conclusion in the EPA (1987)
review that there is sufficient evidence of carcinogenicity of
formaldehyde in animals based on inhalation data. This is based
on the induction by formaldehyde of an increased incidence of a
rare type of malignant tumor (i.e., nasal squamous cell
carcinomas) in both sexes of rats, in multiple inhalation
experiments, and in more than one species (i.e., rats and mice).
4.5.3. Supporting Evidence
Formaldehyde induces a variety of toxic effects other than
neoplasia. It is a potent upper respiratory tract irritant in
humans and animals. Short-term exposure to formaldehyde causes
cytotoxicity, increases rates of cell turnover and cell
proliferation in the nasal passages, and induces hyperplasia and
squamous metaplasia in the noses of rats and mice, and the upper
respiratory tract of monkeys. Formaldehyde also induces a number
of genotoxic effects in a variety of cell culture systems (e.g.
bacteria, viruses, fungi, mammalian cells) and in vivo studies in
insects. The observed effects include gene mutations, single
strand breaks, chromosomal aberrations, sister chromatid
exchanges, and cell transformation. In addition, formaldehyde
has been found to form adducts with DNA in both in vivo and in
vitro tests. Its ability to interfere with DNA repair mechanisms
has also been demonstrated (EPA, 1987)•
Recent studies indicate that formaldehyde has tumor
promotion potential, at least in some tissues. Formaldehyde
enhances the tumorigenic response in mouse skin induced by DMBA
(Iversen, 1986), rat respiratory trachea induced by
benzo[a]pyrene (Cosma and Marchok, 1987), and rat stomach induced
by MNNG (Takahashi et al., 1986).
Formaldehyde is not the only aldehyde which is carcinogenic
in animals. Acetaldehyde, the closest aldehyde to formaldehyde
in structure, is carcinogenic in rats causing nasal cancers via
inhalation exposure. It also induces cancers of the nose and
trachea in hamsters. Other aldehydes such as glycidaldehyde and
malondialdehyde have also been shown to be carcinogenic.
4.5.4. Categorization of Overall Cancer Evidence
The lines of evidence presented above, i.e. human, animal,
and supporting evidence, can be integrated into a weight of
evidence assessment of human carcinogenicity for formaldehyde.
These consist of (1) a sufficient animal evidence, mainly based
58
-------
on multiple inhalation studies demonstrating nasal cancers in
rats, and some nasal tumor response in mice; (2) limited human
evidence for an association of formaldehyde exposure with cancers
of the nose and nasopharynx, and possibly lung cancers seen in
epidemiologic studies; and (3) considerable other key evidence.
This latter-information includes (1) structure-activity
considerations that-Associate aldehyde exposure with carcinogenic
activity; (2) the known genotoxic activity of formaldehyde
including the formation of DNA adducts; and (3) the ability of
formaldehyde to injure cells and affect cell division. EPA
classifies all this information on formaldehyde as indicating
that it is a probable human carcinogen (Group Bl) according to
EPA's Guidelines for Carcinogen Risk Assessment (1986).
5.0 ESTIMATES OF CANCER RISK
Data from studies of humans are always preferred for making
numerical risk estimates. As was the case in 1987, however, the
available epidemiologic data on formaldehyde are still not
suitable for low dose quantitative cancer risk estimation, mainly
because of a lack of adequate exposure information in the
studies. Accordingly, results from animal studies are relied
upon to estimate low-dose human cancer risk.
Since the 1987 EPA review, updated data on nasal DNA-protein
cross-link (DPX) data have been reported for rats, as well as
DNA-binding data for monkeys. Risk estimates based on these data
also have been published by Starr (1990). This section reviews
the 1987 approach and provides risk estimates based on the new
information, along with an evaluation of the assumptions and
considerations in the estimation of risks.
The accuracy of risk assessment depends on the mechanistic
basis of the model used to calculate risks. Biologically based
models, of the type developed by Moolgavkar and Knudson (1981),
have potential for a more comprehensive approach to risk
estimation. Although certain aspects of the mechanism of
formaldehyde induced carcinogenesis have been established, it is
still not completely understood. A biologically based approach
has more prospects to be constructively applied when more
definitive data become available concerning cell proliferation,
dose rate effects, actual carcinogenic moieties and thei:-
mechanisms of action, and the interactions among these processes
(See discussion in Section 8.3.2.2). In the absence of these
data, the linearized multistage procedure is used as in 1987 to
yield a plausible upper limit to the cancer risk based on the
cancer incidence observed in rats at high concentrations.
Discussions of the choice of animal study for modelling, and
tumor response are found in the 1987 document.
59
-------
5.1 Review of Quantitative Assessment bv EPA (19871
The quantitative risk assessment of formaldehyde reported in
1987 (outlined in Figure 5-1) was based on the Kerns et al.
(1983) F344 rat bioassay. This study provides a dose-response
relationship-between increased nasal squamous cell carcinoma
incidences in both'-males and females with increasing formaldehyde
concentrations. The accounting of the tumor incidence and of the
number of animals at risk was in accord with the recommendations
of the Interagency Risk Management Council (IRMC, 1984); serial
sacrifice and mortality were taken into account, and squamous
cell carcinoma incidences in males and females were combined.
The administered exposures were assumed to pose equivalent
carcinogenic risk across species. A continuous lifetime average
exposure correction factor of 5.6 was used (by multiplying each
exposure rate in a 6 hours/day, 5 days/week inhalation bioassay
by a factor of (5 days/7 days) x (6 hr/24 hr) = 1/5.6). A "five-
stage" model (i.e., one incorporating a fifth degree polynomial)
was found to fit the data adequately, using GLOBAL83 (ICF Clement
Associates, Ruston, LA). This process resulted in an upper limit
inhalation Incremental unit risk estimate of 1.3 x 10 per
or 1.6 x 10" per ppm, for exposure levels below 0.6 ppm.
Since 1987, EPA has changed its policy for interpreting the
linearized multistage procedure (LMP). EPA currently uses
GLOBAL86 rather than GLOBAL83 for fitting the LMP. Rather than
fitting a polynomial of fixed degree, this new implementation
fits models of all degrees 1 through 6, and, among those that fit
the data, chooses the one with the minimum q,*. This results in
the choice of a model of different order for the 1987 data set
with administered exposures, a "three-stage" model giving the
lowest unit risk (6.1 x 10 per ppm).
5.2 Issues in the Use of DPX as Delivered Dose
Insight into the exposure of target tissues to reactive
formaldehyde can be helpful. Formaldehyde has been shown to
react with macromolecules in the target cells to form detectible
DPX. The DPX levels following certain levels of inhalation
exposure, therefore, may serve as a useful dosimeter. This is
true whether or not the cross-links are involved (directly or
indirectly) in the actual mechanisms of formaldehyde-induced
carcinogenesis. That is, under certain assumptions, the DPX can
serve as an index of the presence of the reactive compound in the
target cells. The considerations for using DPX levels in this
way may be outlined as follows.
First, one must assume that, from moment to moment; the
concentration of formaldehyde at the targets for DPX formation
(i.e., DMA and certain proteins) is equal or at least
proportional to the concentration at the target sites, whatever
they may be, for formaldehyde's carcinogenic effects. This seems
60
-------
Bioassay Exposure,
Rats (pp»)
(Kerns et al, 1983}
Rat Tumor Response
(Kama et al., 1983;
USBPA, 1987)
„ Adjustment for
daily exposure,
x (5 days/7 days)
GLOBAL83
Dos«-Response
Relationship/
q,* (p«r pp«) for
low dos«a
Continuous Daily
Human Exposure at
0.1, o.S, 1.0 ppm
Lifetime Hunan Risks
Figure 5-1: Steps in the derivation of lifetime human cancer risks
based on rat carcinogenicity and administered dose, as described in
1987.
61
-------
reasonable in the present case; if the targets of carcinogenic
action and DPX formation are not one and the same, they are at
least side-by-side within the same cells. Second, the rate of
DPX formation at any one moment should be proportional to the
formaldehyde concentration at that moment; i.e., the rate of
cross-linking .should be first order and not saturable over the
range of conditions-^encountered. This also seems reasonable; the
cross-linking reactions are spontaneous and not enzyme-mediated.
There would seem to be a large excess of linkable nucleophilic
sites on DNA and protein, so their concentrations remain
essentially constant.
Under these conditions, at any given moment the rate of new
DPX-formation in target cells is proportional to intracellular
formaldehyde concentration. As a consequence, the total number
of cross-links accumulated during the course of the tissue's
exposure - that is, the integral of the instantaneous rate of
formation from the onset of inhalation until the disposal of all
inhaled and absorbed formaldehyde - is proportional to the
integral of tissue formaldehyde concentration over that time
span. That is, the total DPX is an index of the target tissue
"area under the exposure-time curve11 (AUC) . The AUC is often
used as a measure of target tissue exposure, since it encompasses
both the amount of compound present and the duration of its
presence.
In practice, what is measured is not the total amount of
cross-linking during a bout of exposure, but the net amount,
reflecting some degree of repair during the exposure. When DPX
is used as a dosimeter of the presence of formaldehyde in the
target cells, the total rather than the net cross-link formation
is desired. Repair tends to obscure the desired datum. One must
assume that such repair is negligible, or at least that it
constitutes a constant proportion of the total cross-linking.
(In other contexts, however, the net cross-linking may be the
appropriate thing to examine - e.g., when exploring the
mechanisms whereby cross-links may affect the target cells or the
carcinogenic process.)
5.2.1 Use of DPX for High-to-Low-Dose Extrapolation
Having discussed the use of DPX levels as an indicator of
target-tissue AUC, it remains to explore how this "delivered-
dose" measure is proportional (or not proportional) to inhalation
exposure on the one hand,'and to carcinogenic effect on the
other. Turning first to the question of high-to-low-dose
extrapolation, the DPX determinations following inhalation of
different air concentrations (Casanova et al., 1989; Heck et al.,
1989) indicate that higher exposures have more-than-
proportionally higher levels of DPX. Since the dose-response
curve for formaldehyde-induced nasal carcinomas in rats is
markedly convex, the question is raised as to whether this
4 62
-------
nonlinearity can be explained as a consequence of the
pharmacokinetics of nasal tissue exposure.
A number of potential causes for pharmacokinetic
nonlinearity were discussed earlier, in Sections 3 and 4. Most
prominent is_ the presumed saturability of oxidative
detoxification mediated by formaldehyde dehydrogenase. In
addition, the role' "of the mucociliary apparatus in trapping and
removing formaldehyde may be compromised at higher exposures.
Heck et al. (1989) also mention possible saturation of protein
binding and repair of DNA-protein cross-links; if these two
factors do indeed figure prominently, the rationale for linking
DPX to intracellular formaldehyde exposures is weakened, since
basic assumptions, discussed above, would be violated.
The empirical data on DPX levels in rats (Casanova et al.,
1989) show that, whatever saturation of protective effects are
responsible, there is indeed a higher-than-proportional exposure
of nasal epithelial cells to formaldehyde at high air
concentrations. Use of DPX levels as a surrogate measure of
target-cell dose will, therefore, tend to account for its
nonproportionality to air concentration. That is, it will
empirically adjust for observed nonlinearities in formaldehyde
pharmacokinetics, removing this source of nonlinearity from the
expression of the dose-response relationship.
Examination of the rat DPX data (Casanova et al., 1989) and
the nasal carcinoma data of Kerns et al. (1983) suggests that the
convexity of tumor response is much sharper and extends to much
lower doses than does the cross-linking nonlinearity. The
convexity in DPX levels is noticeable but rather shallow, and the
data indicate substantial cross-linking at lower exposure levels
- levels at which tumor response was no longer seen. Moreover,
DPX formation tends to become linear with air concentration at
lower levels. Thus, although pharmacokinetics may explain an
important part of the convexity in the tumor dose-response
relationship, there are other sources of nonlinearity as well.
This raises the question of the proportionality of "delivered
dose" to carcinogenic effect.
Once it is taken that DPX levels provide an index of area
under the target-tissue formaldehyde concentration curve (AUC),
it should be clear that a number of factors can cause tumorigenic
response to be convex with respect to this "delivered dose."
Being an integrated, summary measure of the tissue's exposure,
the AUC contains no information about the shape of the
concentration curve over time, including its peak levels. Peak
levels and average concentration will be higher immediately after
breathing higher air concentrations.
For a mechanism of carcinogenic action in which the impact
on critical cellular targets accumulates from moment to moment at
63
-------
a rate proportional to concentration, the AUC will give a good
prediction of the cumulative effect of a given tissue exposure.
The argument is the same as that tying cross-links to the AUC for
dosimetric purposes, since cross-links accumulate in this way.
Indeed, it is literally the same if the cross-links are the sole
agent of carcijiogenicity, say, by causing transforming mutations.
If, however, the meqhanism of carcinogenic action is such that
doubling the target-tissue concentration more than doubles the
rate at which damage is done, integrating the concentration (the
AUC) will no longer be proportional to summing the damage. For
example, a receptor-mediated mechanism may have receptor
occupancy at a given time in proportion to tissue concentration
of ligand, but the carcinogenic effect of an exposure that
propels occupancy above the critical level for triggering signal
transduction will be out of proportion to the consequences of a
slightly lower concentration that just falls short of such
triggering levels. Carcinogenesis mediated by cytotoxicity would
also be expected to show such effects. Identifying and
accounting for such effects requires rather detailed
specification of how the carcinogenic process proceeds, including
how the rates of the key processes vary moment by moment as a
function of carcinogen concentration, and how the engendered
effects accumulate over time.
Judging the impact of such factors for formaldehyde is
difficult, owing to lack of comprehensive knowledge about the
mechanism of carcinogenic action. As detailed in Section 4,
there are considerable data showing cytotoxicity and at least
transitory cell proliferation at high, but not low, inhaled
formaldehyde concentrations. At the same time, the documented
genotoxic effects of formaldehyde would be expected to operate in
proportion to the AUC. Perhaps the two mechanisms operate
synergistically at higher exposure levels (resulting in the sharp
convexity), while the genotoxic effect may (or may not) operate
alone at lower levels. Just as the dosimetric nonlinearities may
be analyzed and compensated for by pharmacokinetic analysis,
these "pharmacodynamic" nonlinearities may be approached using a
biologically based approach to dose-response modeling (e.g.,
Moolgavkar and Knudson, 1981). The understanding of
formaldehyde's carcinogenic mechanisms, however, is
insufficiently developed to apply this approach at present.
It should be evident from the foregoing that the use of
measured DNA-protein cross-links as a surrogate for target-tissue
exposures to reactive formaldehyde allows one to take into
account certain pharmacokinetic nonlinearities, but does not
address what may be called pharmacodynamic nonlinearities in the
dose-response relationship. These must be approached empirically
by fitting a dose-response curve of rat nasal carcinoma
incidences (as seen in the Kerns et al. [1983] study) versus
target tissue exposure (i.e., versus DPX level). The resulting
curve will be less convex than one of tumors versus air
64
-------
concentration, by virtue of its ability to eliminate the effect
of lack of proportionality of inhaled air concentration to the
resulting exposure of cells of the nasal epithelium to reactive
formaldehyde. It will, however, retain curvature ascribable to
the variation in carcinogenic effects of high and low delivered
doses. "*" ~ ''„*
Some potential pitfalls in this analysis should be noted.
First, as discussed above, the validity of DPX levels as a
dosimeter is dependent on some assumptions, which, although
reasonable, are unproven. Second, the DPX data of Casanova et
al. (1989) are for a single 6-hour exposure to naive rats, while
rats in the Kerns et al. (1983)'bioassay received such exposures
daily for a lifetime. One must assume that the dosimetric
adjustment identified for the first day's exposure obtains
throughout life. Skepticism about this is warranted, since the
long-term bioassay rats lost their normal nasal epithelial tissue
in favor of a proliferation of squamous cells, which may have
very different metabolic abilities, formaldehyde uptake, and
detoxifying mechanisms than the epithelial cells examined in the
DPX experiments.
5.2.2 Use of DPX Data for Interspecies Extrapolation
The result of the approach outlined above is a dose-response
curve of rat nasal carcinoma risk as a function of nasal
epithelial exposure, as indexed by its surrogate, DPX. The
question then arises how such a curve should be applied to the
calculation of potential human risks from formaldehyde
inhalation. This question can be decomposed into two main
components: how to determine the relationship of inhalation
exposure to target tissue AUC in humans (i.e., the question of
human pharmacokinetics), and how to determine the expected
lifetime cancer risk for a human with a certain daily target
tissue exposure, using the rat-based dose-response curve (i.e.,
the question of human pharmacodynamics vis-a-vis the bioassay
rats). These will be considered in turn.
Since the carcinogenicity of formaldehyde is being expressed
in terms of DPX, for its application to humans one first needs a
determination of the relation between inhaled air concentration
and such target tissue exposure. There are no data on
formaldehyde-induced DNA-protein cross-links in humans, nor is
the simple pharmacokinetic model of Casanova et al. (1987)
readily scaled up to provide human DPX estimates. Two empirical
approaches may be entertained to estimate this relationship by
analogy with another species: using the observations on DPX
levels following inhalation of various formaldehyde
concentrations in rats (Casanova et al., 1989) or in rhesus
monkeys (Heck et al., 1989).
65
-------
If the inhalation-DPX relationship in rats is used to
characterize dose delivery in both rats and humans, it is clear
that differences between these two species cannot be addressed.
Such an analysis can, however, illuminate high-to-low dose
differences.^ The convexity of the dose-response relationship is
a key issue) and, 33 noted above, accounting for the lack of
proportionality between air concentration and resultant DPX can
account for much, but not all, of this convexity. The
shortcomings of this approach are several. The obvious one is
that any potential differences between humans and rats in the DPX
levels resulting from given inhalation exposures cannot be
identified or accounted for. There may be such differences not
only in the general levels of DPX, but also in the degree of
nonlinearity between high and low air concentrations.
There is also the question of pharmacokinetic dose-rate
effects. The observations of rat DPX formation are for constant
6-hour exposures followed by exposure to pure air, matching the
daily exposures in the Kerns et al. (1983) cancer bioassay. In
contrast, human exposures may arise in a variety of patterns,
from continuous exposure to episodic peaks. Since the degree of
saturation of capacity-limited processes and other sources of
pharmacokinetic nonlinearity depends on the moment-by-moment
tissue concentration profile, the DPX resulting from exposures of
different pattern yet equal ppm-hours may not be equivalent.
Fortunately, we are usually interested in rather low-level human
exposures, and the importance of dose-rate effects to
pharmacokinetics tends to be minimal under such circumstances
(Hattis, 1990).
An alternative to using the rat inhalation-DPX relationship
for characterizing the one in humans is to rely on the
observations in rhesus monkeys (Heck et al., 1989). The
respiratory tract of monkeys is more similar in anatomy and
geometry to humans than is that of the rat (see Appendix C).
Moreover, monkeys may share facultative oral-breathing with
humans, while rodents are obligate nose-breathers. These factors
would tend to suggest that observations of the degree of DPX
formation in monkeys may be more likely to be representative of
the case in humans. There are remaining differences in the
respiratory tract of humans and monkeys, however. In addition,
there are differences in size scale and respiratory rhythms that
have unknown effects on the ability of formaldehyde to be
absorbed and react. Finally, the enzymatic rates of
detoxification, binding, and cross-linking reactions must be
assumed similar in monkeys and humans, although such rates are
often considered to scale allometrically across species of
different body size. These questions bear on whether the
proportionality of DPX levels to the AUC of formaldehyde is
similar across species; i.e., whether the use of DPX as an index
of tissue exposure needs to be recalibrated. In sum, any claim
that use of the monkey DPX data introduces a correction for
66
-------
species differences in pharmacokinetics (as well as the high-to-
low dose correction) needs to be tempered by the uncertainty
about how well the monkeys actually represent the situation in
humans.
If it is granted that a useful estimate of DPX levels in
humans following a^defined exposure can be derived from either
the rat or monkey DfcX data, and that such an estimate provides an
adequate index of area under the curve of formaldehyde
concentration in the target nasal epithelial cells, it remains to
determine how the dose-response relationship of nasal carcinoma
incidence as a function of DPX—determined in rats—is to be
applied to the calculation of human risk.
EPA commonly applies so-called surface area scaling to
applied doses—i.e., administered amounts are equal in terms of
mg/kg ' /day—to determine doses of expected equal carcinogenicity
across species. Such scaling is intended to adjust for the fact
that physiological processes operate more slowly in larger
species, even when the basic anatomical and biochemical machinery
is the same. Rates of such processes as uptake, cardiac output,
metabolic activation and detoxification, and excretion of a
material from the body tend, in general, to maintain
proportionality to body weight to the 2/3 power. A consequence
of such allometric scaling of the rate of processing an
administered dose is that, again as a default expectation, the
area under the curve of the daily target tissue concentration of
the proximate carcinogen (be it the administered compound or a
metabolite) is expected to be equalized across species when the
doses are scaled in proportion to this processing rate, that is,
proportional to the 2/3 power of body weight. Since the volume
of air breathed per unit time stays roughly in proportion to body
weight to the 2/3 power, this method is essentially equivalent to
equating exposure on a ppm-basis, as was done in EPA (1987) .
As discussed above, the AUC is an index of the carcinogen's
opportunity to interact with the tissues, and measures both the
amount of compound present at the target and the duration of its
presence. When dosing that is adjusted to equalize daily AUCs
across species is continued for a lifetime, it results in equal
lifetime average concentrations of the toxicant in the target
tissue. Such lifetime average concentrations are taken to be
equivalent in lifetime cancer risk, despite the fact that a human
is exposed for a longer total amount of time. Just as
pharmacokinetic processes tend to proceed at a slower pace in
humans, so does the pace of carcinogenesis, as shown by the fact
that age-specific incidence curves of cancer and cancer latency
tend to be congruent when time is compared on a fraction-of-
lifetime basis.
In the present case, the DPX measurements have been taken as
an index of the area under the curve of formaldehyde in the
67
-------
target cells. Thus, an equal amount of newly created cross-links
per day per unit of tissue would be expected to indicate equal
daily AUCs, and hence, by assumption, equal lifetime cancer risk.
In other words, no further "cross-species scaling factor" need be
applied, since AUCs are being compared directly. This is in
contrast to-a -situation in which the pharmacokinetic analysis
produces estimates 6f,amount of a compound metabolically
activated to its carcinogenic form. The expectation from
allometric variation in pharmacokinetics is that doses scaled by
surface area will result in an equal fraction of the lower human
administered dose being metabolized. That is, amounts
metabolically activated are l£ss per unit of tissue in humans,
but equal in proportion to kg . Owing to the expectation of
slower processing of the activated material, this lesser amount
in humans may nonetheless be expected to lead to the same AUC as
in the experimental animals. In sum, empirical pharmacokinetic
data on "delivered doses," should be compared against an
expectation based on pharmacokinetic allometry - i.e., against
the pattern that one assumes for delivered doses when such case-
specific data are not available. The same target tissue
exposure, when characterized by different units of measurement,
can have different cross-species ratios. Definitions of
toxicological equivalency of tissue doses across species must
account for which measure is being used.
The question of cross-species extrapolation of carcinogenic
effects is a difficult one, subject to ongoing debate. The issue
of how to incorporate pharmacokinetic information, and of the
need for additional cross-species scaling corrections on
"internal doses," is particularly problematic. The scheme
outlined above for formaldehyde is intended to be compatible with
EPA's default methodology, modified to take special account of
the nature of the DPX data. A document presenting a fuller
discussion of EPA's cross-species scaling methods, their
scientific rationale, and their modification when using
pharmacokinetic data, is in preparation.
5.2.3 Summary
To summarize the rationale for using DPX in the quantitative
cancer risk assessment of formaldehyde, it is argued that DPX
should provide an index of the area under the curve of reactive
formaldehyde in the target nasal epithelial cells (AUC), both in
rats and in other species, This is so whether or not the DPX are
mechanistically involved in the carcinogenic process or simply
passive markers of the presence of formaldehyde in the target
cells.
General allometric arguments lead to the a priori
expectation that relative AUCs across species from a given
inhalation exposure .might tend to be equal, but the observed
facts contradict this expectation. In fact, there is a markedly
68
-------
lower DPX level in monkeys than in rats, indicating a lower
target tissue exposure to formaldehyde, which may be attributed
to differences in respiratory anatomy and airflow patterns (see
Appendix C). Moreover, there is a lack of proportionality of DPX
levels to air concentration, indicating that saturable
pharmacokinetic processes are encountered at higher air
concentrations., leading to higher-than-proportional tissue doses.
By figuring the dose-response curve in terms of DPX, rather than
externally encountered air concentration, this high-to-low dose
nonlinearity can be identified and taken into account. In
addition, if the difference between rats and humans is similar to
that between rats and monkeys, then humans are expected to have a
lower nasal epithelium exposure than rats. The curve of
carcinoma risk versus DPX is taken to apply across species, on
the assumption that equal tissue AUCs (indexed by DPX)
experienced for a lifetime may be expected to yield equal
lifetime cancer risks. A human low-dose unit risk can then be
developed by estimating the DPX level from a low unit of
exposure, and determining the projected risk for such a DPX
level. The nonlinearity of the curve of carcinoma risk versus
DPX is an empirical estimate of the nonproportionality of risk to
the various levels of target tissue exposure. Further knowledge
of the underlying biological processes of carcinogenesis as they
operate at different tissue exposure levels could illuminate the
estimation of this relationship, but such biologically based
dose-response modeling has not been attempted in the present
analysis.
5.3 Calculation of Risk Estimates Using DPX
As discussed in Section 5.2, DPX may be considered as a
measure of formaldehyde delivered to target cells. The following
sections detail the development of risk estimates using DPX data
of rats and monkeys. The risk assessment procedures are
schematically described in Figure 5-2.
5.3.1 Estimation of Delivered Dose
The first step in modelling cancer risk is to convert the
bioassay exposure' rates to delivered dose equivalents, using the
data reported by Casanova et al. (1989) and Heck et al. (1989) .
Table 5-1 reports the total amount of formaldehyde equivalents
covalently bound to DNA and protein in nasal mucosa, resulting
from a 6-hour exposure to formaldehyde at approximately 0.3, 0.7,
2, 6, or 10 ppm for F344 rats, and 0.7, 2, or 6 ppm for Rhesus
monkeys.
A variety of curve-fitting techniques are available for
relating the administered exposures to the observed DPX levels.
One approach is to use the pharmacokinetic model developed by
Casanova and coworkers (Casanova and Heck, 1987; Casanova et al.,
1989; see Figure 5-3). The model, however, has several
69
-------
Bioaaaay Exposure,
Rata (PP»)
(Kama at al, 1983)
Interpolation
among-rat DPX,
corresponding to
experimental
exposures
Dalivarad Dose, Rata
[(pmol/mg DNA)/day]
Adjustment for
daily exposure,
x (5 days/7 days)
Average Daily
Dalivarad Dose, Rats
[(pmol/mg DNA)/day]
GLOBAL86
Rat Tumor Raaponsa
(Kerns at al., 1983?
U8EPA, 1987)
Dose-Response
Relationship/
q,* (/pmol/mg DNA/day),
for low doses
DPX Experiment
Expoaurea (ppm)/
Rat or Monkey*
Interpolation
among expt'1
data points*
Dalivarad Dose,
[(pmol/mg DNA)/day],
Rat or Monkey*
Adjustment for
continuous
exposure, x 4
(24 hr/6 hr)
Expected Human Dalivared
Dose (pmol/mg DMA/day),
at O.I/ 0.5, 1.0 ppm
Continuous Daily
Exposure (based on
analogy to either
rat or monkey)
Lifetime Human Risks
* Rat dosimetry data from Casanova et al. (1989-) ; monkey
dosimetry data from Heck et al. (1989).
Figure 5-2: Steps involved in the derivation of lifetime cancer
human risks based on rat carcinogenicity data, and the use of DNA-
protein cross-linking (DPX) as dosimeter.
70
-------
Table 5-1
Measured Delivered-Dose Levels in Nasal Mucosa at Experimental
Administered Dose Levels in Rats and Monkeys, and Interpolation
Equations — - .
Administered Formaldehyde Linear
Formaldehyde Equivalents Interpolation
Concentration (opm) (pmol/ma DNA/6 hours) Relationshioc
Rats': o.o
0.32
0.71
1.93
5.92
9.87
Monkeys": 0>Q
0.7
2.0
6.0
0.0
1.4
3.9
19.9
106
266
0.0
0.36
2.56
18.2
(assumed)
~1 Y = 4.4X
+ 0.6 '
i n ji . ...-
. •+ U . 4
-i 1 "7 .—•
I C
t D
X _} U
Y = -0.65 + 6.41X
— — O.41 T 13 . IX
— —21.8 + 21. 6X
Y = -134 + 41. 5X
(assumed)
~~] Y = 0.51X
-t- 0.10 '
i n T i «...
-t U . J JL —
-L 1 A
2. J . 4
— — U .82 T 1 . byX
V RO^A.'^QIY
X O • ^1 O ' J • J7 X /W
Casanova et al., 1989: Analytical chamber concentrations were
b reported as mean + s.e. (n = 3) .
Heck et al., 1989: Analytical concentrations were not reported
for the monkey DNA-binding experiment.
Equations listed apply only to the associated range of bracketed
doses: Y = DPX in pmol/mg DNA; X = exposure in ppm.
71
-------
n
4
Z
Q
\
Q
\j
13
X
£
0
L
0
a
c
0
Administered ForrrBldehyde CPP"0
Figur« 5-3: Observed and model-predicted concentrations of DPX in
nasal mucosa of F-344 rats (Casanova et al., 1989) and Rhesus
monkeys (Heck et al., 1989).
72
-------
shortcomings, as acknowledged by the authors
"it ignores other elimination routes (mucociliary clearance,
diffusion out of the cells, metabolism by enzymes other than
FDH, and covalent binding to macromolecules other than DNA),
and it-4oes pot incorporate the possibility of DNA repair or
possible' changes in numbers or types of respiratory mucosal
cells."
While each of these processes would generally be expected to
decrease DPX levels, their relative rates and interactions are
not well characterized. These omissions in the pharmacokinetic
model, therefore, may account for the reported model's
underestimation of the mean observed DPX concentration in rats at
a nominal concentration of 2.0 ppm (analytical concentration 1.93
ppm) by approximately 25%. The model predicts about 14.8 pmol/mg
DNA, while the mean observed level was 19.9 pmol/mg DNA after 6
hours exposure. Accurate estimation of the magnitude of the
lowest dose can be critical, since the estimate of low-dose risk
can be strongly influenced by the lowest dose even when, as in
this case, there is no tumor response in the low-dose animals.
Use of delivered dose estimates based on the pharmacokinetic
model would lead to a higher unit risk than using the observed
DPX concentration, about 25% higher in this case, all other
factors remaining equal.
Alternatively, linear and quadratic regression methods both
fit the data adequately from an overall perspective (high
correlation coefficients) but neither method accurately estimates
the observations in the low-dose region, where there is the
greatest need for accuracy.
In order to better reflect the experimental data, therefore,
a linear-interpolation approach was applied to the experimental
data presented in Table 5-1. Cross-linking corresponding to a
concentration falling between experimental concentrations has
been estimated by linearly interpolating between the two
experimental concentrations closest to'that concentration. This
approach was chosen because it will not under- or overestimate
the observed points. A concentration of 0 ppm was assumed to
correspond to 0 pmol of formaldehyde bound to DNA and protein.
For concentrations greater than 10 ppm, the linear extrapolation
was based on the DPX observed at 6 and 10 ppm. The equations of
the lines connecting each consecutive pair of points are also
listed in Table 5-1.
5.3.2 Lifetime Average Daily Exposure:
Dose-Rate Considerations
In the 1987 risk assessment, an adjustment of the bioassay
doses to continuous daily exposures was made before modelling
risk estimates. This may be inappropriate for the DPX-based
73
-------
procedure; there is some qualitative evidence that dose rate,
rather than average exposure concentration, can be more critical
in determining the severity of endpoints such as cell
proliferation and nonmalignant histological changes (see Section
4.3.5).
' * * \
It is evident frtfm examination of the Casanova et al. (1987)
and Heck et al." (1989) data on DPX (Figure 5-2) that inhalation
of different levels of formaldehyde lead to DPX levels that are
not proportional to air concentration. Since the curves are
convex, the predicted DPX corresponding to a 24-hour averaged
exposure will tend to be less than for a 6-hour exposure to a 4-
times higher air concentration. Since the aim is to express
response in rats in terms of their DPX levels, the DPX
corresponding to the unaveraged, 6-hour exposures should be used.
Since such DPX levels were experienced by bioassay rats for only
5 days per week, they should be corrected by a factor of 5/7 to
produce the average daily DPX experienced for a lifetime.
The above procedure tends to account for dose-rate effects
in the formation of DPX (and hence AUC of formaldehyde in the
target tissue) by different inhalation regimens. At low dose
levels, as evidenced by the lower dose linearity of the curves in
Figure 5-2, such effects become minimal, and the delivered dose
for a 24-hour exposure is equivalent to that of a 6-hour exposure
to a 4-times higher formaldehyde air concentration that produces
the same cumulative dose. As noted previously, however, non-
proportionality of DPX to carcinogenic effect and dose-rate
effects in this relationship, are not addressed by such an
analysis.
5.3.3 Quantitative Estimation of Risk
In order to characterize the empirical relationship of rat
nasal tumors to DPX, the linearized multistage procedure, as
programmed in GLOBAL86 (ICF Clement Associates, Ruston, LA), was
applied to the interpolated rat DPX concentrations adjusted to
daily exposure levels, and the IRMC (1984) tumor incidences used
in the 1987 assessment. The modelling procedure yielded a "two-
stage" model (goodness-of-fit p = 0.07). The input data and the
resulting model parameters are summarized in Table 5-2. The low-
dose slope of the linearized upper bound curve (q^) represents
an upper bound for the incremental risk per unit of tissue dose -
that is, 1.6 x 10 per pmol of bound formaldehyde per mg of DNA
per day. As discussed in section 5.2.2, this incremental risk
per unit of nasal DPX is presumed to apply to humans as well.
s
In general, risks corresponding to lifetime daily 6-hour
exposures at rates below 0.7 ppm can be calculated using an upper
limit inhalation incremental unit risk estimate of 1.6 x 10 per
(pmol/mg DNA)/day, after adjusting the exposure rate to reflect
the relevant DPX formation rate. The full model as given in
*S
74
-------
Table 5-2
Calculation of Cancer Risk Based on Incidence of Squamous Cell
Carcinomas and on Estimates of Delivered Dose in F344 Rats
Bioassay
Exposure
(ppm)
0
2.0
5.6
14.3
"* Delivered Dose
fomol/ma DNA/dav)*
Unadjusted _
0
21.4
99.1
445.2
Adjusted"
0
15.3
70.8
318
Tumor-bearing
animals/
Number at risk0
0/156
0/159
2/153
94/140
Interpolated from the rat DPX data (see Table 5-1).
DPX concentrations were multiplied by 5/7 (see Section 5.2.2).
Effective numbers, based on recommendation of the Interagency
Risk Management Council (IRMC, 1984) .
Results of Linearized Multistage Procedure (GLOBAL86),
Goodness of fit test, p = 0.07
Maximum Likelihood Estimate of Lifetime Cancer Risk:
P(d) = 1 - exp[-(q0 + qid + (fed2)],
with qg = 0.0,
q, = 0.0,
q^ = 1.0 x 10~ (pmol/mg DNA/day) "2
Upper Bound on Lifetime Cancer Risk Estimate (based on 95%
animal upper bound):
P*(d) = 1 - exp[-(q0* + q,*d +
with qo* = 0.0
q,* = 1.6 x 10"4 (pmol/mg DNA/day)'1
qj* = 1.0 x 10"5 (pmol/mg DNA/day)'2
75
-------
Table 5-2 must be used for DPX levels corresponding to exposures
greater than 0.7 ppm, owing to the nonlinear relationship between
DPX formation and tumor response.
To calculate expected human risks from this relationship for
formaldehyde, it is necessary to relate the concentration in air
to the expected DPX level in human nasal tissue. There are two
alternatives. First", one can use the relationship of DPX to air
concentration in rats to estimate the human DPX from a given
inhalation exposure. This alternative addresses only the high-
to-low dose nonlinearity of the air concentration
DPX/relationship. Second, one could use the DPX curve in monkeys
to derive the human estimate. Since monkeys are more like humans
in their anatomy, airflow patterns, and cell distribution
characteristics in the nose, this alternative aims at allowing
for interspecies difference as well as air concentration
differences in the level of DPX formation.
Estimates of expected human DPX levels corresponding to
ambient exposures of 0.1, 0.5 and 1.0 ppm were calculated from
either the monkey or the rat DPX data by using the equations for
monkeys and rats in Table 5-1 (see Table 5-3). Next, to obtain a
DPX estimate for a human with continuous exposure to that air
concentration, the concentration of DPX formed in 6 hours was
multiplied by 4. Then the full model was used to calculate the
risks associated with lifetime exposure to continuous
concentrations of 0.1, 0.5 and 1.0 ppm (see Table 5-4).
For applying the q,* to human exposures below 0.7 ppm (and
for comparison with the 1987 unit risk), the dose units in
(pmol/mg DNA)/day are converted to ppm units of continuous
exposure. For rat dosimetry, the unit risk becomes:
1.6 x 10"" 4.4 (pmol/mg DNA)/day 24 hr 0 0 ,n-3
(pmol/mg DNA)/day x ? ppa* U~JL- x -THT ' 2'8 x 10 per Ppm;
the 4.4 pmol/mg DNA comes from the first linear relationship for
rats in Table 5-1. The unit risk based on .monkey dosimetry is
calculated similarly:
1.6 x IP"4 0.51 (pmol/mg DNA)/day 24 hr , , . n-»
; 1-7 _... i. ; .— x ^ '—a " J— x —7—r— - 3.3 x 10 per ppm;
(pmoi/mg DNA)/day ppm bnr * *v
the 0.51 pmol/mg DNA comes from the first linear relationship for
monkeys in Table 5-1. It should be noted that these unit risks
are applied only to continuous exposure or to low exposures that
have been prorated over a lifetime.
76
-------
Table 5-3
Interpolated Delivered-Dose Concentrations Corresponding to
Selected Human Environmental Exposures
Delivered-Doses (pmol/mg DNA/day)* at
Selected Human Environmental Exposures
Q. 1 ppm 0. 5 PPP 1.. 0 ppm
Rats:
6 hours daily 0.43
Continuous daily6 1.72
Monkeys :
6 hours daily 0.051
Continuous daily 0.204
Derived using the relationships in
6 hour daily exposures multiplied
2.6
10.2
0.26
1.04
Table 5-1.
by (24 hours/ 6
7.7
30.8
0.88
3.52
hours) = 4.
Table 5-4
Comparison of Estimates of Upper Bounds (and Maximum Likelihood
Estimates) of Human Lifetime Carcinogenic Risk Associated with
Lifetime Continuous Daily Exposure to Formaldehyde
Exposure 1987 Risk 1991 Risk Estimates6
Rate (ppm) Estimates3 Monkey-based Rat-based
0.1 2 E-3C (5 E-7) 3 E-5 (4 E-7) 3 E-4 (3 E-5)
.0.5 8 E-3 (5 E-4) 2 E-4 (1 E-5) 3 E-3 (1 E-3)
1.0 2 E-2 (1 E-2') 7 E-4 (1 E-4) 1 E~2 (1 E-2)
* Estimated using 1987 inhalation unit risk 1.6 x 10"2 per ppm.
Incorporated monkey or rat dosimetry data (see Section 5.3.3).
° 2 E-3 = 2 x 10 .
77
-------
5.4 Comparison with other Formaldehyde Risk Estimates
5.4.1 Comparison with 1987 Unit Risk Estimate
There is an approximately 6-fold difference between tLhe 1987
and the 1991- rat dpsimetry-based risk estimates (2.8 x 10" per
ppm vs 1.6 x 10 per*ppm). An approximate 2.5-fold difference
is due to use of DPX as an internal dosimeter in rats. This
reflects accommodation of the high-to-low dose nonlinearity in
the relation of air concentration to tissue exposure. As
discussed in Section 5.1, a 2.5-fold difference from EPA's 1987
assessment is due to a change in EPA's policy for interpreting
the linearized multistage procedure, resulting in the choice of a
model of different order.
Use of the monkey nasal DPX data as a surrogate for human
delivered dose further lowers the estimated risk to humans at low
exposures about another 9-fold, yielding an overall 50-fold
reduction of unit risk estimates compared to the 1987 unit risk
(3.3 x 10' per ppm vs 1.6 x 10" per ppm).
5.4.2 Quantitative Assessment by Starr (1990)
Starr (1990) calculated cancer risks based on the DPX
experiments of Casanova et al. (1989) in rats, and of Heck et al.
(1989) in monkeys. As in the EPA assessment, he made the species
equivalence assumption that a "given level of formaldehyde
covalently bound to DNA would pose the same lifetime cancer risk
to both rats and monkeys." All cancer risks were based on the
response data in the Kerns et al. (1983) bioassay, using a
slightly different accounting of tumor incidence than EPA used,
namely 0/160, 0/160, 2/160, and 87/160, in the 0, 2.0, 5.6, and
14.3 ppm groups, respectively. Using GLOBAL82 (ICF Clement
Associates, Ruston, Louisiana), Starr fit a "three-stage" model
using rat DPX levels interpolated from the DPX experiment to
correspond to the bioassay exposures. Predicted risks
corresponding to 0.1, 0.5, and 1.0 ppm formaldehyde in air, based
on the DNA-binding data for both rats and monkeys are reproduced
in Table 5-5. Starr also did not address the non-nasal DPX
observed in monkeys in making his calculations.
Starr concluded that point estimates of human risk (also
called maximum likelihood estimates, or MLEs) based on DPX in
monkeys were lower than those based on airborne concentrations to
rats (the basis of EPA's 1987 unit risk), by JLS much as 150,000-
fold. Starr calculated a MLE risk of 3 x 10 based on the air
concentration of 0.1 ppm administered to rats, while the
corresponding MLE human risk based on monkey DPX was 2 x 10" .
The differences between upper bounds on risk were less dramatic,
the largest difference being 25^-fold between an upper bound rat
dosimetry-based risk of 2 x 10 and an upper bound monkey
78
-------
Table 5-5
Comparison of Risk Estimates from Starr (1990), Upper Bounds and
Point (Maximum Likelihood) Estimates"
•\ ',
i-,*
Air Upper Bound and (MLE) Estimates of Risk From
Cone. Several Formaldehyde Exposure Measures
(ppm) Rat/1983"Rat/1989cMonkev/19890
0.1 2 E-4* (3 E-7) 7 E-5 (2 E-9) 8 E-6 (2 E-12)
0.5 8 E-4 (3 E-5) 4 E-4 (3 E-7) 4 E-5 (3 E-10)
1.0 2 E-3 (3 E-4) 1 E~3 (6 E-6) 1 E-4 (1 E-8)
" Continuous lifetime average exposure adjustment not used.
Kerns et al. (1983) exposure concentrations (ppm).
DNA-protein cross-links (pmol/mg DNA) from Casanova et al.
d (1989).
Using the 1989 rat DNA-binding data for the • dose-response
relationship, and the Heck et al. (1989) DNA-protein cross-
links for delivered dose at 0.1, 0.5, and 1.0 ppm.
* 2 E-4 = 2 X 10
79
-------
dosimetry-based risk of 8 x 10" at an air concentration of
0.1 ppm.
The 1991 upper bound risk estimates (unadjusted, as listed
in Table 5-4), based on either rat or monkey DPX data, are
greater than_ those estimated by Starr (1990) by a factor of 4,
reflecting the^ continuous exposure adjustment in the 1991 EPA
estimates. The MLE estimates, however, differ dramatically, with
the 1991 point estimates higher than Starr's by 3 to 4 orders of
magnitude. The significance of these differences is only slight,
owing to the instability of the maximum likelihood estimates for
such nonlinear dose-response patterns.
The EPA generally does not compare point estimates of risk
based on animal data, particularly in instances where such
estimates are based on very small numbers of animals responding,
which are expected to provide high variability in the MLE
coefficients. A reliable procedure for calculating point
estimates of risk does not yet exist, as is stated in the EPA
Guidelines for Carcinogen Risk Assessment (USEPA, 1986).
5.5 Discussion of Uncertainties
By the arguments discussed in section 5.2, use of DPX as a
dosimeter should avoid some uncertainties associated with using
administered dose, the dosimeter used in the 1987 assessment,
because the DPX should be more closely proportional to the target
dose than the ambient exposure. Uncertainties specific to the
1991 risk assessment are discussed below.
5.5.1 Uncertainties Associated with Use of DPX
There are a number of uncertainties associated with risk
estimates based on rat carcinogenicity data using dosimetry data,
i.e. DPX, from the rat and monkey. Human risk estimates based on
the rat dosimeter may be too high at a given exposure
concentration, owing to differences in breathing patterns
resulting in different exposure of the target tissue. The
observed rat DPX concentrations best predict the rate of nasal
tumors, albeit in rats. Since the DPX were lower in the monkey
nasal mucosa, use of the rat dosimetry-based estimates may
overestimate the probability of nasal tumors in primates. On the
other hand, the monkey dosimetry-based estimates may be too low
because the DPX levels measured in the nasopharynx of monkeys,
corresponding to a site having a possible association of tumor
incidence in humans, are not considered in the derivation of risk
estimates. The dynamics of carcinogenesis in these different
respiratory regions in different species have not been adequately
examined.
Although there is evidence of a dose-rate effect in short-
term experiments examining cell proliferation, nonmalignent and
•S
80
-------
nonmalignent changes, it is not possible to incorporate this
information more explicitly into the risk assessment at this
time, principally because the current studies were not carried
out for a long enough period of observation to assess
quantitatively the comparable consequences on tumor incidence.
If dose-rate-could be taken into account in the risk
calculations, 'then'^there is a suggestion that, for some human
exposure conditions, such as those of the workplace, the cancer
risk could be higher than these which rely on time-weighted
averages, although the amount of any increase cannot be predicted
at this time.
It should also be noted that the experimental conditions for
measuring DPX did not duplicate the conditions of the chronic
assay, in which the cell type that developed cancer (squamous
cells) was not seen following acute exposures, and the dynamics
of cell proliferation and cytotoxicity were not assessed. Also,
if continued exposure to formaldehyde stimulates cell
proliferation, a time when more DNA is produced, then chronically
exposed animals could have a lower concentration of DPX than
those that were exposed for one session, if respiratory surface
area and dynamics determine the amount of formaldehyde
absorption. On the other hand, the amount of available DNA could
be the determining factor in the amount of formaldehyde absorbed
by the respiratory mucosa. These factors could have a greater
impact at high concentrations (those above 6 ppm) based on
available dose-rate effect information.
5.5.2 Concordance with Epidemiologic Evidence
As mentioned above, it was acknowledged in 1987 that human-
based risk estimates could not be calculated, owing principally
to imprecise exposure estimates. Nevertheless, some rough
comparisons (Figure 7-1; USEPA, 1987) suggested that animal-based
risk estimates using airborne concentrations were concordant with
observed human risks.
The Blair et al. (1986) study was selected in this
comparative assessment since it provides the most thoroughly
characterized exposure data. The authors reported less than two
excess nasopharyngeal cancer deaths for each of the three
exposure level categories. The predicted excess cancer deaths
based on rat carcinogenicity data, using rat and monkey DPX
dosimetry, were on the order of 10' to 10" , and 10" to 10" ,
respectively (See Appendix B for details of the analysis).
A number of factors may contribute to the differences
between the predicted animal-based risk estimates and "the
observed risks in the Blair et al. (1986) study. The oronasal
pattern of respiration in humans and monkeys would be expected to
reduce the dose of formaldehyde received by the nose and the
nasopharynx, and to increase the dose delivered to the oral
81
-------
cavity and upper respiratory tract, relative to the rat's
strictly nasal pattern. This is supported by the observation of
a lower rate of DPX formation but more widespread distribution in
the respiratory tract of monkeys, particularly with higher level
exposure. Since DPX are being used as a surrogate for the
agent(s) inducing tumors, the possible contribution of non-nasal
DPX in monkeys stioul'd .be noted. The rat-tumor-based model,
however, depends only'-on nasal DPX levels, first in rats, for the
model derivation, then in monkeys, for translation across
species. Consequently, its use in confirming the experience
reported in the Blair et al. study might somewhat underestimate
total respiratory cancer at all anatomical sites.
Without carcinogenic data in monkeys, it is not possible to
assess how susceptible monkeys are to formaldehyde-induced
carcinogenesis. For noncancer effects, the monkeys appear to be
more susceptible to formaldehyde toxicity than rats, based on an
observed induction of more widespread histologic lesions in the
respiratory tract of monkeys at equivalent exposure
concentrations (6 ppm). It is difficult to evaluate the full
impact of these observations because, as stated before, no
consistent correlation has yet evolved between histopathological
changes and tumor induction. DNA binding also occurs in the
lower respiratory tract of monkeys, so that risk estimates based
only on cross-links data that correspond to those in rats may
underestimate human risk.
Several exposure measurement issues further complicate a
direct comparison of human- and animal-based risks. First, the
human exposures reported in Blair et al. (1986) included
particles onto which formaldehyde adsorbed, or which may have
contained releasable formaldehyde. The particles may have caused
some irritation which could potentiate the effect of the
formaldehyde actually delivered. Therefore, formaldehyde
exposure for these situations could have been underestimated,
tending to underestimate the number of expected deaths.
•Second, human exposures in Blair et al. (1986) were reported
as cumulative exposures. As seen from animal data (Wilmer et
al., 1989; Feron et al., 1988), dose/exposure rate may be more
predictive of cancer incidence than cumulative exposure. This is
based on the observation that noncancer effects appeared to be
correlated with exposure rate more than with cumulative exposure.
When used for comparison with the animal model predictions, which
are based on dose rate, cumulative exposures averaged over length
of employment probably lead to underestimates of effective
exposure, and consequently, underestimates of cancer risk.
Individual records of length of employment and occurrence- of peak
exposures would need to be evaluated and incorporated for a more
accurate estimation of dose rate to use with an animal model.
82
-------
It is not clear what the relative contributions of these two
aspects may be to an accurate risk estimation. As a rough
estimate, each probably affects the exposure measures by less
than an order of magnitude on average, but their relative impacts
on individual exposures and consequent health effects are much
harder to asjsass. Individual exposures would need to be
reassessed fo'r a more parallel comparison of animal and human
risks.
5.6 Summary
Recently published research concerning formaldehyde toxicity
has potential for refining the formaldehyde level estimates
associated with carcinogenesis. These data favor the use of
formaldehyde DNA-protein cross-links as an appropriate measure of
formaldehyde exposure associated with the incidence of squamous
cell carcinomas. DNA-protein cross-link data for rats were used
in the linearized multistage procedure, and estimates of risk to
humans have been calculated using DNA-protein cross-link data
from rats or monkeys.
6.0 NONCARCINOGENIC HAZARD EFFECTS
6.1 Review of New Animal Studies
6.1.1. Respiratory Effects
Animal studies reviewed in EPA (1987) demonstrated that
formaldehyde causes cellular effects in the nasal cavity of rats
and monkeys, with the extent and severity dependent on both
concentration and duration of exposure. A NOEL of 1.0 ppm for
squamous metaplasia was determined in Cynomolgus monkeys (Rusch
et al., 1983) and rats (Kerns et al., 1983) following 26 weeks
and 24 months of exposure, respectively. The cellular effects of
formaldehyde are confirmed in later studies in mice (Maronpot et
al., 1986), rats (Zwart et al., 1988; Woutersen et al.1, 1987;
Monticello and Morgan, 1989; Monticello, 1990) and monkeys
(Monticello et al, 1989). However, the NOEL for cellular changes
in rats were not consistent among these studies.
The pathologic effects of formaldehyde were determined in
male and female B6C3F1 mice exposed to 0, 2, 4, 10, 20, or 40 ppm
for 6 hr/day, 5 days/week for 13 weeks. Squamous metaplasia and
rhinitis were induced in the nasal tissues and larynx at 20 ppm,
and in the nasal tissues at 10 ppm. Minimal squamous metaplasia
was detected in the nasal mucosa of 1/20 mice exposed to 4 ppm.
No lesions were detected in mice exposed to 0 or 2 ppm. The
cellular effects on the respiratory system of mice were more
prevalent in males than in females (Maronpot et al., 1986).
To study the cellular effects of low concentrations of
formaldehyde, Zwart et al. (1989) exposed male and female albino
83
-------
Wistar rats, to 0, 0.3, 1, or 3 ppm formaldehyde vapor for 6
hr/day, 5 days/week for 13 weeks. Compound-related
histopathologic nasal changes varying from epithelial
disarrangement to epithelial hyperplasia and squamous metaplasia
were found in the 3 ppm group. These changes were not observed
in the other_ lx>w dose groups, indicating a NOEL of 1.0 ppm, which
is consistent with a'MN,OEL of 1.0 ppm determined in a chronic
study (Kerns et al., 1983) as reported previously in EPA (1987).
In contrast, in another 13 week inhalation study in male and
female albino Wistar rats (exposed to 0, 1, 10, or 20 ppm
formaldehyde) conducted by the same laboratory, minimal focal
hyperplasia or sguamous metaplasia was found in the respiratory
epithelium lining in the nasal septum and maxillary turbinates of
3 of 20 rats exposed to 1 ppm formaldehyde. Squamous metaplasia
were induced in the nasal tissues and larynx at 20 ppm, and in
the nasal tissues at 10 ppm. No lesions were found in control
animals (Woutersen et al., 1987).
A concentration-dependent response in formaldehyde-induced
cellular changes was also observed in a recent chronic study in
rats (Monticello, 1990). Male F344 rats were exposed to 0, 0.7,
2, 6, 10 or 15 ppm formaldehyde for up to 18 months. However,
only the highest two concentrations of formaldehyde (10 and 15
ppm) induced nasal epithelial cell necrosis, inflammation,
hyperplasia and metaplasia. The lack of response at 6 ppm in
this study is inconsistent with the positive effects found in a
previous chronic study in F344 rats by Kerns et al. (1983). The
reasons for the discrepant results from these two studies are not
known.
Monticello and colleagues compared the results in monkeys
with those observed in F344 rats exposed under similar conditions
(Monticello et al., 1989; Monticello and Morgan, 1989). Groups
of three rhesus monkeys were exposed to 0 or 6 ppm formaldehyde
vapor 5 days/week for 1 or 6 weeks. Groups of 6 male F344 rats
were exposed under a similar test regimen. In the rat,
formaldehyde-induced lesions were detected only in the anterior
portion of the nose. These lesions were characterized as
epithelial degeneration, hyperplasia and squamous metaplasia. In
contrast, lesions in monkeys were more widespread, being
detectable as deep as the trachea and bronchi. Slight
progression of histologic lesions occurred between 1 and 6 weeks
of exposure, and the percent of nasal surface affected was
greater in monkeys for 6 weeks than in monkeys exposed for 1
week.
6.1.2 Contact Sensitization
In a study by Lee et al. (1984), formaldehyde was tested in
guinea pigs with inhalation, dermal and injection exposure in
order to assess the route of exposure most likely to cause
84
-------
sensitization in animals and to evaluate the potency of
formaldehyde as a sensitizing chemical. In this study,
formaldehyde induced contact sensitivity by all three routes of
exposure. Dermal and injection exposure routes were more potent
than the inhalation route; as the dose of formaldehyde was
increased, both ±he degree of sensitization and the percentage of
animals sensitized increased by either route. Contact sensitivity
was detected only"at 24 hours post-challenge in 2/4 animals
exposed via inhalation to 10 ppm for 8 hours/day for 5 days.
Animals exposed to either 6 or 10 ppm for 6 hours/day for 5 days
did not exhibit evidence of either contact sensitivity or
antibody production.
Animals injected with an emulsion of formaldehyde and Freund
adjuvant responded to topical challenge with formaldehyde by
displaying extensive dermal reactions that were maximal 24 hours
after challenge. Antibodies were detected in low titer in 2/4
animals in this group. All animals in this group displayed
extensive dermal sensitivity when challenged. The degree of
contact sensitivity was determined to be dose-dependent and
increased doses resulted in a greater percentage of animals
becoming sensitized.
6.1.3 Pulmonary Sensitization
Formaldehyde did not produce pulmonary sensitivity (Lee et
al., 1984). However, the authors urge caution in concluding that
formaldehyde does not cause pulmonary sensitization, because the
protocol used in this study was developed for toluene
diisocyanate and the latent period for eliciting a sensitization
response to formaldehyde may be different. Guinea pigs induced
by inhalation did not demonstrate either immediate or delayed
onset respiratory reactions when challenged by bronchial
provocation of either 2 ppm formaldehyde for 1 hour or 4 ppm for
4 hours.
Animals sensitized by dermal application of 74 mg
formaldehyde on each of 2 days did not give any evidence of a
respiratory response when challenged with inhalation exposure to
either 4 ppm formaldehyde for 4 hours or 2 ppm for 1 hour. No
cytophilic antibodies were detected in the blood of these
animals.
6.1.4 Immunotoxicologic Studies
No papers evaluating the role of the immune system in
formaldehyde cytotoxicity were discussed in the EPA (1987). One
recent, well-designed study focuses on the role of macrophage
development and function and suggests a new mechanism through
which formaldehyde-associated cytotoxicity might be induced.
85
-------
The systemic effects of formaldehyde inhalation exposure on
macrophage development and function were studied (Adams et al.,
1987) to determine the potential role of formaldehyde in altering
the mononuclear phagocyte system (MPS). Intraperitoneal
macrophages were selected to measure effects on the MPS that were
not simply due,to Ipcal irritation and inflammation. Female
pathogen-free B6C3Fi>>mice were exposed to either 0 or
approximately 15 ppm of formaldehyde for 6 hr/day, 5 days/week
for 3 weeks. Peritoneal leukocytes were collected by peritoneal
lavage from control and exposed mice 2-3 days after final
exposure and tested to determine whether formaldehyde exposure
altered macrophage activation, produced by the prototype
activator, pyran copolymer (MVE-2). Macrophage phagocytosis was
determined using labelled sheep red blood cells. Binding and
lysis of target cells to macrophage monolayers were also
quantified. Hydrogen peroxide (H2O2) was measured to test the
capacity of macrophages to secrete reactive oxygen intermediates
(ROI), a response usually elicited by cells exposed to an
irritant, an inflammatory agent or mycobacteria. The enzyme
activity of leucine aminopeptidase (LAP) was quantified to
distinguish inflammatory from activated macrophages.
Exposure of mice to 15 ppm formaldehyde for 5 days/week for
3 weeks did not alter the number of resident macrophages
appreciably in the peritoneal cavity or the number elicited in
response to MVE-2. Overall, comparison of the functional stage
of macrophage activation between exposed and control animals
indicated that the peritoneal population of macrophages was in
the basal or resident stage of development, and was not in the
activated stage. The macrophages from formaldehyde-exposed mice
did have a lower content of the enzyme leucine aminopeptidase.
However, there was no other evidence to suggest that resident
macrophages were activated. Thus, formaldehyde exposure did not
induce significant systemic maturation of resident tissue
macrophages. Similarly, formaldehyde neither inhibited the
development of activated macrophages (induced by exposure to
pyran (MVE-2) ) , nor did it alter the two acquired functions of
these macrophages: binding of tumor cells and cytolysis.
Primed macrophages acquired .competence for release of
hydrogen peroxide, which was significantly increased in
macrophages from formaldehyde-exposed mice. Thus, formaldehyde
exposure can, on a systemic basis, significantly enhance
macrophage competence to release ROI. The biological impact of a
change in the potential of mononuclear phagocytes to release ROI
may be either beneficial or detrimental. An earlier study has
documented that exposure to formaldehyde enhances host resistance
to challenge with the microbe Listeria monocytoqenes. A''
mechanism proposed for this effect was macrophage competence for
release of ROI because it is strongly correlated with macrophage
activation for destruction of many facultative or obligate
intracellular microbes. It has also been hypothesized that
86
-------
macrophages might induce genotoxic damage of significant
mutagenic potential in bystander eukaryotic cells. Released ROI
from these macrophages, at concentration levels comparable to
those induced in this study, might contribute to the genotoxicity
and lesion induction of formaldehyde.
6.1.5 Summar'y ^and Conclusions
Results from rec'ent animal studies and those reviewed in EPA
(1987) indicate that formaldehyde is capable of inducing cellular
changes in the respiratory tract of mice, rats and monkeys. The
affected site(s) and severity of formaldehyde-induced lesions are
dependent on both concentration and duration of exposure. At low
concentrations (<3 ppm) lesions are confined to the anterior
region of the nose of mice, rats, and monkeys. However, at high
concentrations (>6 ppm), lesions may extend beyond the nasal
turbinates though the lesions at distal sites are generally less
extensive. There are clear differences in the distribution of
formaldehyde induced lesion between rats and monkeys at high
concentration, i.e., 6 ppm. Lesions are more widespread
extending to the bronchi whereas in rats, cellular changes are
confined to the anterior portion of the nasal passages. Based on
available data, it can be determined that a NOEL for formaldehyde
induced cellular changes is 2.0 ppm in mice, 1.0 ppm in monkeys,
and ranges from 0.3 - 1.0 ppm in rats.
Dermal sensitization induced by direct skin contact with
formaldehyde is well documented. Limited available information
indicate that inhalation exposure to formaldehyde may induce
contact sensitivity in guinea pigs. However, there is no
evidence that formaldehyde elicits pulmonary sensitization in
laboratory animals.
The study on macrophage function by Adams et al. (1987)
suggests that altered macrophage competence may play a role in
lesion induction after formaldehyde exposure. However, this
hypothesis is highly speculative. Neither the role of the immune
system in the induction of cellular and neoplastic changes during
inhalation exposure to formaldehyde, nor the effects of
formaldehyde inhalation exposure on pulmonary antibacterial
defenses, are well understood. These areas reguire further
elucidation.
6.2 Review of Noncancer Human Studies
This section reviews the major studies published since the
release of EPA (1987) for noncancer effects in humans associated
with inhaled formaldehyde. The effects investigated in' these
studies included irritation of the eyes and upper respiratory
tract, lower airway symptoms, changes in lung function, cellular
changes in the nasal tissues, and immunologic and neurologic
alterations.
87
-------
The reports under review included studies of occupational
populations (Alexandersson and Hedenstierna, 1988; Alexandersson
and Hedenstierna, 1990; Holness and Nethercott, 1989; Horvath et
al., 1988; Imbus and Tochilin, 1988; Kilburn et al. , 1988; Malaka
and Kodama, 1990; N.unn et al., 1990; Uba et al., 1989), residents
in mobile homes or urea-formaldehyde-fcam-insulated (UFFI) houses
(Broder et al./l98ffa^ 1988b, 1988c; Liu et al. , in press,
personal communication; Ritchie and Lehnen, 1987) and controlled
human exposures (Green et al., 1987; Sauder et al., 1986, 1987;
Schachter et al., 1986, 1987; Witek et al. , ~1986, 1987). Three
studies examined cellular changes in the nasal turbinates by
biopsy of workers (Boysen et al., 1990; Edling et al., 1988;
Holmstrom et al., 1989). Several studies examined either
immunologic (Pross et al., 1987; Thrasher et al., 1987, 1988,
1989; 1990; wilhelmsson and Holmstrom, 1987) or neurologic
(Kilburn et al., 1987) responses. Most of these studies looked
at a variety of endpoints; the context and details of the study's
design are given at its' first discussion.
6.2.1 Eye and Upper Respiratory Tract Effects
6.2.1.1 Irritant Symptoms
Results from studies of residential, occupational, and
controlled exposures supported the 1987 assessment that sensory
irritation of the eyes, nose, and throat occurred at mean
formaldehyde concentrations of 0.1 to 3.0 ppm. In general, the
newer studies showed that eye and upper airway irritant effects
have been transient and of mild to moderate intensity.
Three prevalence studies of residential exposures (Broder et
al. 1988a, 1988b, 1988c; Ritchie and Lehnen, 1987; Liu et al., in
press) examined eye and upper airway irritant effects. Broder et
al. (1988a, 1988b, 1988c) conducted a health survey of 1726
residents in 571 UFFI houses and compared health outcomes to a
control group of 720 residents in 231 homes. UFFI-homes were
identified from a Canadian Government-sponsored registry
(Consumer and Corporate Affairs). The UFFI-house group was
subdivided into three categories depending on their intention to
remove the insulation, take other remedial action, or take no
action. Controls were recruited by letters to homes selected at
random from streets adjacent to consenting UFFI-households.
Cases and controls were similar in their distributions of race,
sex, and occupation. The control population, however, was
younger, had fewer exsmokers, and spent fewer hours at home. The
refusal rate was 5% among UFFI-horae residents and 80% among
controls.
All homes and occupants were examined on two occasions
separated by a 12-month interval, during which two-thirds of the
UFFI-homes performed remedial work. At the first visit, mean
formaldehyde levels in the three groups of UFFI homes fell
*>.
88
-------
between 0.039 ppm (other remedial ^action group) and 0.045 ppm (no
action group) (range: 0.01-0.23 ppm), and for control homes,
0.035. ppm (range: o.01-0.11 ppm). Pooled subjects from all UFFI
homes showed statistically significant positive exposure-response
relationships between exposure level and the prevalence at the
first visitr^of; thr,oat discomfort, eye irritation, and nasal
problems (either ninny nose, stuffy nose, sneezing, nasal
discomfort, or nose bleeds), but the relationships were largely
dependent on a small group whose mean indoor formaldehyde levels
were greater than 0.12 ppm. These logistic regression analyses
adjusted for time of year, sex, age, race, smoking, and number of
hours spend in the house per week. Objective indicators of
health status (nasal airway resistance and detection of pyridine
odor) did not demonstrate differences between UFFI-home and
control groups.
Repeated participation rates at the second survey were 84%
for controls and 91% for the pooled UFFI-home group. Of the
pooled UFFI-home group, 90% of the intended removal, 74% of the
no action, and 66% of the other remedial action houses had
performed remedial work. Although for all three exposure
subgroups statistically significant declines were seen between
the two surveys in the prevalence of throat discomfort and nasal
symptoms, the pooled-UFFI group continued to show at the second
visit a small but statistically significantly greater amount of
eye irritation and nasal problems compared to controls. The
prevalence of sense of smell problems and results of nasal airway
resistance remained constant between visits for all groups and,
with respect to the second visit, between the pooled-UFFI and
control groups.
Mean formaldehyde levels remained constant between the first
and second visits. Exposure-response analyses which controlled
for age, sex, smoking experience, total hours spent in the house
per week, race, and outside temperature, however, showed
formaldehyde level at the second survey as a statistically
significant predictor of nasal problems. No relationships were
noted, as in the first visit, with other upper respiratory
symptoms.
In this study, the presence of respondent bias is a
possibility due to the nature of the self-selected study
population. Such bias would tend to obscure any exposure-
response relationship. Hovever, the presence of exposure-
response relationships diminishes our belief in the likelihood of
respondent bias.
Ritchie and Lehnen (1987) studied 2000 residents with
concerns about formaldehyde exposure who lived in 397 mobile and
494 conventional homes. The Minnesota Department of Health
provided names of study participants given by referring
physicians. In an analysis examining formaldehyde concentration
89
-------
(<0.1 ppm, 0.1-0.3 ppm, and >0.3 ppm) and eye and nose/throat
irritation, the prevalences of response were statistically
significantly associated with exposure level. For each residence
type, 90% of occupants exposed to 0.3 ppm or greater and 1-2%
exposed to <0.1 ppm reported eye irritation. For the 0.1 to 0.3
ppm group, between ^12% (conventional home residents) and 21%
(mobile home resideri.ts) reported eye irritation. For nose/throat
irritation, depending on age and smoking status, 74% and 99% of
respondents reported the symptoms at the high concentration range
(>0.3 ppm); 12% and 36% at the middle level (0.1-0.3 ppm); 0 and
11% at levels <0.1 ppm. The authors pointed out that the results
could have been biased by self-selection. The inclusion of some
subjects might have been related to the existence of symptoms and
approximate knowledge about the level of formaldehyde exposure,
but the extent of such bias was difficult to assess.
Liu et al. (in press) conducted a prevalence study of
randomly-selected non-complaint mobile homes in the State of
California using a house-age-stratified sampling scheme. Among
2203 letters mailed to recruit participants, a total of 663 homes
with 1394 residents completed the summer phase and 523 homes with
1096 residents finished the winter phase of the survey; 472 of
the latter homes were also part of the summer study.
Approximately 60% of the homes were manufactured between 1981 and
1983. The large sample size allowed for control of potential
confounders such as age, sex, smoking status, time spent at home,
and the effect of formaldehyde level on people with chronic
respiratory illnesses and allergy problems.
Formaldehyde concentrations varied from below the limit of
detection (0.01 ppm) to 0.46 ppm; the mean formaldehyde
concentration was 0.089 ppm in the summer and 0.088 ppm in the
winter. The authors found that cumulative exposure (ppm-hours)
better described the health effect relationships rather than
concentration (ppm). All analyses, thus, were based on
cumulative exposure as categorized into three groups: <7.0
ppm*hours, 7.0-12.9 ppm*hours, and >12.0 ppm*hours.
Logistic regression analyses showed that complaints of
burning/tearing eyes (summer and winter), stinging/burning skin
(summer), dizziness (winter), and sleeping problems (winter) were
significantly (p<0.05) associated with cumulative formaldehyde
exposure after adjusting for identified possible confounders.
Response for burning/tearing eyes during the summer and winter
phases for <7.0 ppm*hours was 10.8% to 13.3%; for 7.0 to 12.0
ppm*hours, 14.7% to 17.1%; and for >12.0 ppm*hours, 20.6% to
21.4%. The limitations associated with this study are similar to
those of Broder et al. (1988a, 1988b, 1988c) and Ritchie'and
Lehnen (1987). The low participation rate (about 52%) and the
self-selection of participants enhance the potential for
respondent bias. The presence of exposure-response relationships
diminishes this possibility, however.
90
-------
Five studies of occupationally-exposed populations (Uba et
al., 1989; Horvath et al., 1988; Alexandersson and Hedenstierna,
1988; and Holness and Nethercott, 1989) noted increased reporting
of upper irritant effects. All studies except Uba et al. (1989)
were of the cross-sectional (prevalence) design. Uba et al.
(1989) assessed acute and persistent respiratory effects in a
well-conducted" prospective study of 107 medical students exposed
to formaldehyde during a 7-month gross anatomy laboratory. The
occurrence of acute* symptoms (upper respiratory tract and lower
airway effects) was measured by the American Thoracic Society
(ATS) Epidemiology Standardization Project respiratory
questionnaire in 81 students after a gross anatomy laboratory in
which there was formaldehyde exposure, and after a microanatomy
laboratory in which no formaldehyde was used. Responses at the
beginning of the school year were compared to subjects' responses
at the end of the study to give estimates of the incidence of
persistent (chronic) effects (See Section 6.2.2.1 for lower
airway effects).
For acute effects, these investigators found statistically
significant elevations in the occurrences of eye, nose, and
throat irritation during gross anatomy laboratory than during
microanatomy laboratory. The time-weighted averages (TWA) of
formaldehyde in the gross anatomy laboratory ranged from 0.16-
0.93 ppm; peak exposures ranged from 0.1 to 5.0 ppm, with a mean
of 1.9 ppm.
Horvath et al. (1988) evaluated upper airway effects in 109
workers exposed to formaldehyde (mean=10 years) at a plant
producing particleboard or molded products, and in 254 control
subjects from food-processing plants. Controls were similar to
exposed on most demographic descriptors except the control group
contained more females than males. Subjects completed a modified
ATS questionnaire before and after the monitored work shift, in
addition to estimating the intensity of symptoms by a visual
analog scale. Personal monitoring showed formaldehyde levels
(TWA) in the molded products/particleboard processes ranged from
0.17- 2.93 ppm, with a mean of 0.69 ppm. The molded products and
particleboard processes also contained nuisance particulates;
approximately one-third were of respirable sizes. The mean level
at the control food processing plants was 0.05 ppm.
In this well-conducted study, these investigators showe-1
statistically significantly higher prevalences of eye irritation
(46% exposed vs 24% controls) and irritation of the nose and
throat (35% vs 13%) among exposed subjects compared to controls.
Exposed positive responders reported moderate intensity of
burning eyes; however, considerable variability existed between
individuals. It is difficult to draw comparative inferences
regarding intensity since no such assessment was made in the
nonexposed control population. Logistic regression analysis
showed formaldehyde concentration to be a statistically
91
-------
significant predictor of the following symptoms: burning eyes,
itchy eyes, burning nose, stuffy nose, itchy nose, and sore
throat. While this study is limited by possible respondent bias,
The authors believed the exposure-response relationships
demonstrated a qualitative hazard.
-— - ', ,:^
Alexandersson and Hedenstierna (1988) investigated
respiratory effects associated with formaldehyde exposure among
38 workers involved in painting work using acid-hardening
lacquers and in 18 nonexposed controls. These investigators
found statistically significantly increased prevalences for
irritation of the eyes (66% versus 17%) and of the nose/throat
(39% versus 0%). Mean formaldehyde concentrations for exposed
averaged 0.32 ppm (range of 0.12-2.08 ppm) , with peaks up to 0.56
ppm. Formaldehyde-exposed subjects also had exposures to low
concentrations of solvents, i.e., 10 to 100 lower than the TWA.
The exposed population could be considered a survivor group
since participants were chosen if they were working at the time
of the study and had been exposed during the preceding 12 months.
The mean number of years worked, 7.8, is not surprising given the
eligibility criteria. These investigators do not describe
selection criteria for controls. Cases and controls appeared
similar with respect to age; however, the control population
contained fewer smokers. No information was provided on sex.
Holness and Nethercott (1989) studied 84 funeral service
workers exposed to formaldehyde and 38 controls. Funeral service
workers were selected from a funeral directors' association list
of homes expressing interest in the study. Among funeral service
workers, 66% were active licensed embalmers, 20% were currently
inactive licensed embalmers, and 14% were apprentices. Inactive
embalmers were older and had worked a longer number of years than
active embalmers. Apprentices were younger than active workers,
and had just started embalming regularly during the past year.
Controls were recruited from a large service organization and
from a group of paid student volunteers. Exposed and controls
were similar in age and sex; however, the exposed population had
accumulated more pack-years smoked and had higher body weight.
All analyses controlled for the effects of smoking. These
investigators obtained information on respiratory complaints
using the ATS questionnaire. The mean formaldehyde level
measured during the period of study was 0.36 ppm with a range
from 0.08-0.81 ppm. Average formaldehyde level for controls was
0.02 ppm.
These investigators observed statistically significant
increases in the prevalence of nose and of eye irritation over
controls (44% vs 16%; 42% vs 21%, respectively) in logistic
regression analyses which accounted for smoking. Exposed
subjects also reported throat irritation more often than controls
(17% vs 5%), but this difference was not statistically
92
-------
significant (p=0.iO). No differences in irritative symptoms
existed between apprentices and active licensed embalmers;
however, active members reported eye irritation more frequently
(73% vs 71%, p
-------
Using a similar study design as Schachter et al. (1986,
1987), wite'k et al. (1986) found more frequent reporting of
irritation of the eyes, nose, and throat and odor perception with
formaldehyde exposure than with room air among 15 healthy and 15
asthmatic subjects.. Asthmatics generally had a higher responses
to these symptoms than healthy subjects, although both groups had
similar perceptions of* severity. In a subsequent study of only
asthmatics (n=15), Witek et al. (1987) noted that unusual taste,
odor, nasal discomfort, and eye irritation were commonly reported
complaints, and were of mild to moderate severity. As in the
reports of Schachter et al. (1986, 1987), the investigators did
not perform statistical testing to determine whether the
differences in response between formaldehyde exposure and room
air were statistically significant.
In a series of studies, Sauder et al. (1986, 1987) and Green
et al. (1987) exposed healthy nonsmoking or asthmatic
participants to 3 ppm formaldehyde for three hours. Each subject
served as his or her own control and received clean air for three
hours. Participants in both Sauder (1986) and Green et al.
(1987) performed exercise during exposure. Symptom
questionnaires in all studies assessed the presence and severity
of odor, nose or throat irritation, and eye irritation.
These studies found statistically significant increases in
upper airway complaints. Sauder et al. (1986) observed
statistically significant increases in response for odor, nose
and throat irritation, and eye irritation in their study of 9
healthy nonsmoking participants during formaldehyde exposure than
during exposure to room air. The individual severity scores
ranged from none to moderate. Likewise, Sauder et al. (1987)
noted in 9 asthmatic volunteers statistically significant
increases in nose and throat irritation and in eye irritation.
with formaldehyde exposure than during exposure to clean air.
Study participants reported irritation of the nose and throat
after 30 minutes, whereas, increased response to eye irritation
did not occur until one hour after start of exposure. In a study
of 22 healthy nonsmoking and 16 asthmatics subjects, Green et al.
(1987) reported statistically significant increases in perceived
odor, nose and throat irritation, and eye irritation. The
severity of response appeared highest after each exercise stint.
In summary, the newer studies demonstrated that the eye and
upper airway irritation occurs with formaldehyde exposure.
Results from studies of mobile home populations indicated that
eye and upper respiratory tract irritation complaints increase at
formaldehyde concentrations above 0.1 ppm, as reported
previously. Some studies reported higher prevalences or''
incidences of eye irritation whereas other studies report nose
and throat irritation as the more frequent sensory irritant
effects of formaldehyde. Different categorizations of symptoms
by individual investigators and the characteristics of the
94
-------
studied population (i.e., healthy volunteers or self-selected
individuals) may account for these disparate observations. The
finding of dose-response relationships for eye as well as for
nose and throat irritation supports a tenet that complaints of
irritant symptoms are exposure related, and not an artifact of
reporting bias.
•> * %
6.2.1.2. Nasal^'cellular Changes
Three prevalence studies of formaldehyde-exposed
occupational populations (Edling et al., 1988; Holmstrom et al.,
1989; Boysen et al., 1990) and one prevalence study of UFFI-house
occupants examined cellular alterations in tissue from the nasal
turbinates. Edling et al. (1988) assessed histopathologic
changes in nasal biopsies of 75 male workers at two particleboard
processing plants and one laminate plant. Seventy-two percent of
the total eligible workforce participated in this study. Of
participating workers, mean exposure duration was 11 years.
Histological findings from the formaldehyde-exposed workers were
compared with a referent group of 25 men, matched for age and
smoking habits, who had no industrial exposure to formaldehyde.
Concurrent formaldehyde levels in the three plants ranged from
0.08 ppm to 0.88 ppm (TWA) with peaks up to 4 ppm. Higher
exposure had occurred in the past. Exposure to wood dust was
also likely in the particleboard plants, but not in the laminant
plant, where only formaldehyde exposure occurred. Concurrent
wood dust levels ranged from 0.6 to 1.1 mg/m .
Gross examination revealed normal nasal mucosa in 75% of the
exposed, whereas 25% had swollen and/or dry changes of the nasal
mucosa. Histological abnormalities were noted in 96% of all
exposed. The histologic grading, using a scale from 0 (normal)
to 8 (carcinoma), showed a higher score among exposed when
compared to unexposed (2.9 vs 1.8, p<0.05). The histologic
examination of exposed showed 4% with normal epithelium, 11% with
loss of cilia and goblet cell hyperplasia, 56% with squamous
metaplasia, and 8% with mild or moderate dysplasia. These
investigators also assessed the effect of wood dust and
formaldehyde in comparison to only formaldehyde. No difference
in average histopathological score was found between
particleboard workers and laminate workers. No relationship was
observed with duration of exposure. Duration of exposure may be
a poor surrogate for exposure; the investigators pointed out that
dose rate may be more important and may account for the lack of
an observed exposure-response relationship.
Holmstrom et al. (1989) investigated histologic effects in
nasal turbinate biopsies associated with formaldehyde ^exposure
alone and combined with wood dust. Two groups of formaldehyde-
exposed workers were examined; one group of 70 chemical
manufacturing workers exposed only to formaldehyde or
formaldehyde resins (FA) and another group of 100 workers from
95
-------
furniture factories with exposure to formaldehyde resins and wood
dust (WD-FA). Histological scores of these workers were compared
to 36 controls, primarily clerks, with no occupational
formaldehyde exposure. All study participants had long
employment durations (means ranged from 9 to 15 years). Average
formaldehyde_ levels among exposed were 0.04 to 0.40 ppm (median
0.24 ppm), with frequent peaks over 0.80 ppm, for the FA group,
and around 0.16 to 0.20 ppm, with levels never exceeding 0.40
ppm, for the WD-FA group. Mean wood dust levels for the WD-FA
group ranged between 1 to 2 mg/m , with the majority of the
particles being of respirable size. Controls had exposure to
formaldehyde at background levels, with a measured mean of 0.07
ppm.
These investigators observed higher mean histologic scores
for exposed, FA, 2.2 (range 0-4) and WD-FA, 2.1 (range 0-6),
compared to controls, 1.6 (range 0-4), on a scale of 0 (normal)
to 8 (carcinoma). The difference between exposed and controls
was statistically significant for the exposed groups combined and
for the FA group. In addition, these investigators noted that
loss of cilia, goblet cell hyperplasia, and cuboidal and squamous
cell metaplasia replacing columnar epithelium were more frequent
in the group exposed to FA than in controls. The mean histologic
score of the WD-FA group alone was not statistically
significantly different from the control group. However, higher
mean scores appeared for WD-FA workers grinding wood for more
than four hours a day when compared to whose grinding for less
than one hour daily, although the difference was not
statistically significant (p>0.05). Two subjects (2%) of the WD-
FA group had mild or moderate dysplasia (a score of 6); both
workers ground wood for more than 4 hours daily.
These investigators did not find a relationship between mean
histologic score and levels of exposure to formaldehyde (ppm) or
wood dust, of duration of exposure, or of cumulative formaldehyde
.exposure (ppm-year). In addition, no relationship was observed
with smoking.
Boysen et al. (1990) evaluated histopathologic changes in
biopsy samples of the nasal turbinates from 37 workers producing
formaldehyde and formaldehyde resins and 37 age-matched
referents. In addition to the histologic examination,
rhinoscopic examinations and nasal complaints were also assessed.
Referents were selected from the office staff of two chemical
industries and from outpatients at an ear, nose, and throat
department of a local hospital. Exposed workers represented one-
half of the current workforce. Exposure levels were estimated
from recent measurements, from knowledge of the production
process, and from previous and present subject-reported symptoms.
Boysen et al. grouped exposed workers into two exposure
categories, 0.5 to 2 ppm and >2 ppm; however, all analyses were
performed using the aggregated exposure group.
96
-------
The histologic classification was based on a 5 point system,
with dysplasia ranked the highest; 92% of exposed had scores
higher than 0 (normal epithelium) compared to 86% of controls.
More pronounced metaplastic alterations were found among exposed
workers than among controls (scores of 1.9, range 0-5, and 1.4,
range 0-3, respectively), but the difference was not
statistically significant (p>0.05). Three exposed workers (8%)
who were categorized into the 0.5 to 2 ppm category had sguamous
cell epithelial dysplasia; none of the referents had dysplasia.
The rhinoscopic findings appeared to show a higher
prevalence of hyperplastic nasal mucosa in the exposed than in
the control group, but the difference was not statistically
significant. These investigators showed statistically
significant increases in the prevalence of subjective nasal
complaints (the identify of which was not described by the
investigators) among exposed workers. The authors pointed out
that although metaplastic changes are generally considered as
non-specific, by being induced by a number of factors including
temperature, humidity, dust, and age, the finding of dysplasia
among exposed individuals was noteworthy.
Broder et al. (1988b, 1988c) assessed possible cellular
changes of the nasal cavity of UFFI-house occupants and controls.
Swab samples of the nasal epithelial surface were acquired from
the nostril and not from any internal structure (e.g., the
turbinate) as obtained in the occupational studies. The intended
UFFI-removal group, but not other UFFI-house residents, showed a
statistically significant small magnitude increase in the mean
number of squamous metaplastic cells per subject relative to the
controls. In the follow-up survey one year later (Broder et al.,
1988c), all groups (exposed: UFFI-removal, other action, and no
action; and controls) showed a statistically significant decrease
in total cells recovered in the swab and in squamous metaplastic
cells. No differences, however, existed between the exposed
groups and controls.
In summary, results of the newer studies indicate that
formaldehyde exposure along with other exposures encountered in
the studied occupational settings may enhance the severity of
cellular damage. The major cellular changes in the nasal
epithelium of formaldehyde-exposed workers were loss of cilia,
squamous metaplasia, and mild or moderate dysplasia. The
magnitude of the squamous-metaplastic changes was more pronounced
among exposed than among the referent populations. A low
prevalence of dysplasia also existed among exposed workers which
were absent among referents.
These findings are based upon few individuals, and this
limits the conclusions. In addition, the prevalent design of
these studies may over- or under-estimate the effects. The lack
of exposure-response relationships in these studies may also be a
97
-------
reflection of the prevalent study design; the studied populations
could be considered survivor populations. In addition, exposure
raisclassification may also be present, thus, limiting the ability
to detect exposure-response relationships.
6.2.2 JLower Airway Effects
'» '<
6.2.2.1 Subj eci'-Reported Symptoms
Lower airway irritation is characterized by cough, phlegm,
wheezing, and chest discomfort, and has been reported often in
people exposed to 5-30 ppm formaldehyde (EPA, 1987) . Several
recent studies suggested increased prevalences of lower airway
symptoms at lower exposure levels than reported in EPA (1987).
Two studies of residential exposures (Broder et al., 1988b,
1988c; Liu et al., in press) evaluated lower airway irritant
effects. Broder et al. (1988b) showed statistically significant
positive relationships between formaldehyde level and the
prevalence in the first survey of cough and sputum in UFFI-house
occupants (mean formaldehyde levels of 0.039 to 0.045 ppm) using
logistic regression analyses. These relationships remained even
when the subgroup exposed to levels greater than 0.06 ppm were
deleted. The UFFI-house group was subdivided into three
categories depending on their intention to remove the insulation,
take other remedial action, or take no action. In the logistic
regression analyses, the respondent's intent to take remedial
action and formaldehyde level together produced a statistical
interaction in which the estimated prevalence of cough was larger
for both variable combined than for only formaldehyde level by
itself. Results from a repeat survey one year later (Broder et
al., 1988c) showed a decrease in symptoms among only the subgroup
of UFFI subjects who had undertaken remedial work, however, the
positive relationship (p<0.05) remained between formaldehyde
level and cough. No statistically interaction such as that
observed for cough in the first survey was noted in the follow-up
survey. Formaldehyde levels between these surveys did not change
(0.040 to 0.044 ppm).
Liu et al. (in press), using logistic regression analyses,
did not find any relationship between cumulative formaldehyde
exposure (ppm-hour) and either cough or wheezing. Rather, these
symptoms were related to smoking, sex, and presence of chronic
disease. Mean formaldehyde exposures in this study were higher,
0.89 ppm, than in Broder et al. (1988b, 1988c).
Six studies of occupational populations (Horvath et al.,
1988; Malaka and Kodama, 1990; Uba et al., 1989; s
Alexandersson and Hedenstierna, 1988; Holness and Nethercott,
1989; Nunn et al., 1990) evaluated lower airway irritant effects.
Horvath et al. (1988) observed statistically significantly
increased prevalences of cough (35% vs 19%), phlegm (27% vs 10%),
98
-------
and chest discomfort (9% vs 2%) in particleboard and molded
products workers exposed to formaldehyde when compared to
prevalences in workers from food-processing plants (Horvath et
al., 1988). In addition, these investigators observed
statistically significant dose-response relationships for cough,
chest complaints, ajrid phlegm. The severity of chest symptoms
among positive respdnders was mild and was accompanied by
considerable individual variation. Formaldehyde levels (TWA) in
the molded products/particleboard processes ranged from 0.17 to
2.93 ppm, with a mean of 0.69 ppm, and in food processing, 0.05
ppm (mean). The molded products and particleboard processes also
contained nuisance particulates, approximately one-third were of
respirable sizes.
Another study (Malaka and Kodama, 1990) also reported
statistically significant increases in the prevalences of cough
and phlegm among 93 formaldehyde-exposed Indonesian male plywood
workers compared to 93 non-exposed working referents. These
results were obtained using logistic regression analyses which
controlled for age, smoking status, and dust exposure. Referents
were age, sex, and smoking-habit-matched workers not
occupationally exposed to formaldehyde. Mean formaldehyde levels
across the plywood plant ranged, for exposed, from 0.22 to 3.48,
and for referents, from level of detection to 0.07 (background
levels). Both formaldehyde-exposed subjects and referents had
respirable dust exposures above the threshold limit value. The
increased prevalence of lower airway symptoms is difficult to
interpret. This population also had a high prevalence of chronic
upper respiratory infectious diseases which may have contributed
to the lower airway symptoms, but due to the cross-sectional
design of this study, it is not known whether formaldehyde may
have aggravated any praexisting condition.
In a prospective study of medical students currently
enrolled in a gross anatomy laboratory course (Uba et al., 1989),
cough, wheezing, and dyspnea among 81 students completing an ATS
respiratory questionnaire were not statistically significantly
related to formaldehyde. However, the prevalence of chest
tightness was of borderline significant (5% vs 0%, p=0.05).
Questionnaires were administered once after a gross anatomy
laboratory in which there was formaldehyde exposure (TWA ranged
from 0.16 to 0.93 ppm) and once after a control microanatomy
laboratory in which no formaldehyde was used.
These investigators also assessed persistent symptoms among
103 students associated with prolonged exposure to formaldehyde
over a 7-month period. Medical students reported cough more
often (p<0.05) at the end of the 7-month exposure than at the
beginning (8% vs 1%). Conversely, wheezing both with and without
dyspnea were reported more frequently (p<0.05) at the beginning
than at the end of the anatomy course. Formaldehyde
concentration (TWA) was highest at the beginning of the study
99
-------
period (up to 0.8 ppm) and declined steadily over the 7-month
period to an average of 0.1 ppm.
In 38 workers exposed to formaldehyde in acid-curing paints
and lacquers when compared to 18 nonexposed controls,
Alexandersson _and Hedenstierna (1988) noted a higher prevalence
of dyspnea dur'ing work (11% vs 0%), "chest oppression" (11% vs
0%) , and cough (5% vs 0%) , but the difference was not
statistically significant (p>0.01). Mean formaldehyde
concentrations for exposed averaged 0.32 ppm (range of 0.12 to
2.08 ppm), with peaks averaging 0.56 ppm. Formaldehyde-exposed
subjects also had exposures to low concentrations of solvents,
(10 to 100 lower than the TWA).
In a study by Holness and Nethercott (1989), no
statistically significant differences existed in the prevalence
of cough, sputum presence, and chest tightness among 84 funeral
service workers (active and retired embalmers, and apprentices)
when compared to 38 controls. Formaldehyde-exposed funeral home
workers, however, more frequently (p<0.05) reported chronic
bronchitis and shortness of breath. Mean formaldehyde level for
the funeral home workers was 0.36 ppm with a range from 0.08 to
0.81 ppm. The average formaldehyde level for control workers was
0.02 ppm.
Nunn et al. (1990) did not find any differences in the
prevalence of lower airway irritant symptoms (chronic bronchitis,
shortness of breath or wheezing) among 125 chemical workers with
free formaldehyde exposure over a six-year period when compared
to 95 other workers at this plant who did not have free
formaldehyde exposure. The prevalence in free formaldehyde-
exposed and nonexposed workers with self-reported symptoms was
similar; 12% and 16% reported breathlessness on hurrying and 26%
and 20% for wheezing, respectively. None of the formaldehyde
exposed, but 4% of the control group, had chronic bronchitis.
Mean (geometric) formaldehyde concentration as measured
semiannually from routine industrial hygiene monitoring over the
6 year study period were less than 1 ppm, except one measurement
which was approximately 1.4 ppm. Peak levels were as high as 5.4
ppm.
Workers included in this study were participants in an
annual health assessment for workers exposed to free formaldehyde
and workers exposed to other materials such as phenolic resins,
epoxy resins, and carbon fibers (the controls). Workers were
identified in 1980 and included 164 workers exposed to
formaldehyde and 129 nonexposed personnel. During the six year
study period, workers had either died (5 exposed, 1 nonexposed)
or left employment (39 exposed, 26 nonexposed) . Among those
workers who had left employment, only 9 exposed and four
nonexposed attended the final examination. Results from 10
100
-------
workers (4 exposed, 6 nonexposed) with unacceptable spirometric
traces were excluded from the analyses.
Although this study identifies workers with exposure six
year earlier, follow-up was limited mainly to workers currently
employed. 9nl_Y 9 put of 39 exposed workers who had left
employment during this study participated in the final
examination. Workers with adverse respiratory effects associated
with formaldehyde may have left employment early so that the
remaining workers would be considered a survivor group. Nunn et
al. presented data that suggested this bias was present. Of 12
men in the exposed group not seen six years later, the FEV, at
the start of the study was less than predicted in 75% compared to
36% of the 117 who were followed over the six years. In
addition, use of chemical workers as a non-formaldehyde referent
group may introduce another source of bias since respiratory
health in this group may be compromised by exposure to
potentially harmful agents. This referent population had annual
respiratory screening because of their exposure to phenolic
resins, epoxy resins, and carbon fibers. These two limitations
would make it difficult to see differences between exposed and
controls.
None of the controlled-exposure studies of formaldehyde
exposure to 2 ppm for 40 minutes (Witek et al., 1986, 1987;
Schachter et al., 1986) or 3 ppm for 3 hours (Sauder et al.,
1986; Green et al., 1987), at rest and with exercise, reported
symptoms of lower airway effects.
In summary, data from controlled-exposure studies showed
that acute exposures to formaldehyde up to 3 ppm did not elicit
symptoms of lower airway effects. Four studies (Broder et al.,
1988b, 1988c; Horvath et al., 1988; Uba et al. 1989; Malaka and
Kodama, 1990) documented increased reporting of lower airway
irritation with longer-term formaldehyde exposure (occupational
and UFFI-house) in the presence of other exposures (for example,
particulates of wood dust or resins in the occupational studies) .
Other studies (Alexandersson and Hedenstierna, 1988; Holness and
Nethercott, 1989; Nunn et al., 1990) examining formaldehyde
exposures at levels less than 5 ppm did not show any
associations. The small number of exposed subjects examined
limits the ability to detect small differences between exposed
and controls. In addition, these studies were unable to examine
all workers who had left employment (i.e., Nunn et al., 1990),
increasing the possibility of underestimating any underlying
exposure-related-effects since workers who left employment may be
considered a more sensitive subpopulation.
Despite these limitations of the overall data base', it is
concluded that some exposed* subjects will experience lower airway
effects but the precise level at which the effects are initiated
is uncertain. The levels in which lower airway symptoms have
101
-------
been observed are lower than the lower limit previously reported
with formaldehyde-associated lower airway effects, 5.0 ppm (EPA,
1987). In the occupational studies reporting lower airway
symptoms, the mean level over a workshift was reported as <1.0
ppm. The range of exposure concentrations in the newer studies,
however, were ^betwee/i 0.17 and 3.48 ppm, and the frequency of
peak exposures above 1.0 ppm was not provided. The number of
excursions above 1.0 ppm during a working day may influence the
prevalence of lower airway irritation complaints.
6.2.2.2 Lung.Function Measurements
Lung function measurements as objective health status
indicators of formaldehyde effects on the lower airways were
assessed in a number of studies. Overall, chronic decrements do
not appear to be associated with formaldehyde exposure, although
small transient decreases were observed at formaldehyde levels
around 1.0 ppm and above.
Broder et al. (1988b) found no differences in pulmonary
function between UFFI-home occupants and their referents after
adjustment for age and other covariates in their two surveys
separated by a year later. Formaldehyde levels in this study
were very low, from 0.039 to 0.045 ppm.
Imbus and Tochilin (1988) observed no differences between
pulmonary function tests conducted in 99 particleboard workers
before the work shift and five hours into the shift. These
workers had exposure to formaldehyde gas (<0.02-0.06 ppm),
phenol-formaldehyde resins, and wood dust. It is not known
whether the baseline spirometry measurements in these workers'
values were within or deviated from normal limits since no
comparison with a standard population was made. In addition,
these investigators did not state whether they followed ATS-or
other board-recommended procedures. This limits interpretation
of the observed results.
Holness and Nethercott (1989) did not observe any
differences in preshift (baseline) or cross-shift changes in lung
function of 22 embalmers compared to controls. Lung function was
assessed by FVC, FEV,, FEV1 o/FVC, and FEF50X, FEF^, nitrogen
washout, and closing volume. Spirometry tests were administered
to embalmers prior to and following an embalming, and for
controls two times during.a three-hour period. The mean
formaldehyde level in this study measured 0.36 ppm with a range
from 0.08 to 0.81 ppm.
A longitudinal study of 125 (116 currently-employed) workers
exposed daily to free formaldehyde during the production of urea-
formaldehyde resin found no evidence of a formaldehyde-related
decline in FEV, over a six year period (Nunn et al., 1990).
Smoking explained the expected rate of decline in FEV1 among
102
-------
nonexposed. The mean rate of decline in the exposed group was
similar, however, between smokers and nonsmokers. This finding
suggested that some other factor besides smoking was responsible
for the observed decrease in FEV1. This study included few
nonsmokers, and results from this subgroup are subject to large
statistical"vari>atian% These investigators, in addition, did not
find any association 'a'mong exposed between the rate of decline in
FEV, and level or duration of formaldehyde exposure.
Lung function measurements (FEV1 and FVC) were part of the
annual health assessment for workers exposed to free formaldehyde
and workers exposed to other materials such as phenolic resins,
epoxy resins, and carbon fibers (the controls). These
investigators did not present results on FVC in the paper. Mean
(geometric) formaldehyde concentrations as measured from biyearly
routine industrial hygiene monitoring (i.e., accidents and spill
situations were not included) were less than 1 ppm except for one
with a TWA of approximately 1.4 ppm. Peak exposures were as high
as 5.4 ppm. The possibility that the observed results might
reflect bias which was discussed in Section 6.2.2.1 (survivor
population and inappropriate control population) limits the
ability to draw strong inferences from this study.
Occupational exposures at higher levels (around 1.0 ppm and
above) appeared to produce transient decreases in volume and/or
flow parameters. Preshift (baseline) pulmonary function studies
were considered to reflect chronic effects of long-term exposure
to formaldehyde and post- or cross-shift changes were considered
acute effects. Three studies (Alexandersson and Hedenstierna,
1988; Kilburn et al., 1989; Malaka and Kodama, 1990) observed
differences in preshift tests, but not in post-shift tests, while
three others (Horvath et al., 1988; Uba et al., 1989;
Alexandersson and Hedenstierna, 1989) observed the opposite,
post-shift changes but not preshift changes.
Alexandersson and Hedenstierna (1988) observed statistically
significant decreases in baseline lung function (FVC and FEV,)
among 38 formaldehyde-exposed acid-hardening lacquer workers
compared with sex-, age-, and height-adjusted reference values as
assessed before work on Monday after 2 days of no exposure.
Ncnsmokers had larger decreases than smokers. No changes were
noted in volume or flow parameters over a work shift.
The deviations in preshift FVC and FEV1 were not correlated
to either peak formaldehyde exposure or mean formaldehyde
exposure/length of employment period. Mean formaldehyde
concentrations for exposed averaged 0.32 ppm (range of O.<12 to
2.08 ppm), with peaks averaging 0.56 ppm. Formaldehyde-exposed
subjects also had exposures to low concentrations of solvents (10
to 100 lower than the TWA). These investigators were uncertain
as to the exact nature of the functional impairments, but noted
that the degree of functional disturbance was small and should
103
-------
not cause any measurable reduction of physical capacity. They
further noted that since impairment was observed after a 2-day
nonexposure interval, the possibility existed for a short-term or
chronic functional impairment. This latter possibility is
important since the -cross-sectional design of the study may have
underestimated" esffec,ts of formaldehyde exposure because the most
sensitive workers may»tiave left employment and would not have
been included.
Kilburn et al. (1989) observed steeper age-related
decrements in vital capacity and flows (FVC, FEV,, and FEF2^.-ft%)
in a cross-sectional study of 280 non-smoking female histology
technicians compared to the expected values derived from a random
sample of the female population in Michigan. These analyses
adjusted for the effects of age and height. In addition, non-
smoking histology technicians had lower flows (FEF^.^ and FEF^.
85X) than the corresponding stratified random sample of Michigan
women. Information on individual study participants regarding
the specific levels of formaldehyde exposure was not provided.
Formaldehyde levels as assessed through industrial hygiene
measurements obtained from a nonrepresentative sample of
laboratories ranged from 0.2 to 1.9 ppm with peaks up to 5 ppm.
Solvents such as xylene, toluene, and chloroform were also
present.
Malaka and Kodama (1990) also observed statistically
significant differences in preshift spirometric tests (FEV,,
FEV,/FVC, and FEF^,,.^) among 93 formaldehyde-exposed Indonesian
plywood workers than among 93 non-exposed Indonesian plywood
workers. No post-shift changes were observed. Spirometric
procedures recommended by the ATS were followed. Formaldehyde
exposure among exposed ranged from 0.22 to 3.48 ppm, and for the
non-exposed, from 0.003 to 0.07 ppm. Formaldehyde-exposed also
had respirable dust exposures (i.e., to wood dust and resins).
These investigators also observed a positive exposure-
response relationship between cumulative formaldehyde (ppm-year)
and FEV,, FEV^FVC, and FEF^.^. The predicted decrements for
these parameters per, unit of cumulative exposure were small. As
mentioned previously, the high prevalence of abnormal chest X-
rays among study subjects indicated chronic upper respiratory
tract infection, the decrements in lung function may reflect the
overall poorer health status of these workers.
Horvath et al. (1998) observed post-shift decrements in both
volume (FVC) and flow parameters (FEF^.^, FEF5W, FEF^) among
particleboard manufacturing workers with no changes observed
among unexposed workers employed in food-processing facilities.
In addition, these investigators observed statistically
significant positive exposure-response relationships between
formaldehyde level and post-shift changes in FEV^FVC,
FEF50X, and FEF^j The correlation coefficients for these
104
-------
relationships, however, were small- No differences were observed
in preshift lung function parameters between formaldehyde-exposed
and controls and duration of formaldehyde exposure was not a
statistically significant predictor of baseline lung function
among exposed. Analyses adjusted for the effects of age, sex,
height, smoking, mpbile home residency, and workshift.
'n* '
The authors also observed higher prevalences of workers
exhibiting the greatest post-shift declines in lung function
(more than one standard deviation greater than the mean changes
as calculated from controls) among formaldehyde-exposed than
among the controls. Horvath et al. termed this group "reactors".
Further analyses showed that formaldehyde concentration was a
statistically significant predictor of reactor status for
FEV^FVC, FEF2SV75X, FEF50X, and FEF^,
In this well-conducted study, investigators followed ATS
testing procedures and methodology throughout this study.
Formaldehyde levels (TWA) in the particleboard processes ranged
from 0.17 to 2.93 ppm, with a mean of 0.69 ppm, and in food
processing, 0.05 ppm (mean). The molded products and
particleboard processes also contained nuisance particulates,
approximately one-third were of respirable sizes. Based upon the
study's results, Horvath et al. (1988) concluded that
formaldehyde, perhaps combined with other airborne substances,
might exert a small effect on the airways as measured by post-
shift tests, but no evidence existed for permanent respiratory
impairment.
Uba et al. (1989) also found among 96 medical students
enrolled in a gross anatomy laboratory class a statistically
significant larger post-shift decrease in FVC with formaldehyde
exposure than the post-shift change observed in spirometry
testing conducted at the beginning of the school term (a period
of no formaldehyde exposure). No chronic decrements were noted,
however, over a 7-month laboratory period. Spirometry was
conducted on three occasions: before start of the laboratory
session, 2 weeks into the laboratory session, and at the end of
7-month laboratory period. On each test day, spirometry was
performed before the start of laboratory and at the end, 4 hours
later. Testing procedures conformed with ATS recommendations.
Formaldehyde concentration (TWA) during the 7-month laboratory
period ranged from limit of detection (0.05 ppm) to 0.93 ppm,
with peaks up to 5.0 ppm (mean=1.9 ppm) during dissection. The
authors did not consider the crossshift decreases in FVC to be
physiologically significant because of their small size, and that
the observation may represent an acute irritant response which
was not accentuated by repeated exposure.
Alexandersson and Hedenstierna (1989) observed statistically
significant small changes in post-shift lung function (FEV^FVC,
CV%) among 21 formaldehyde-exposed woodworkers compared to sex-,
•V
105
-------
age-, height,- and weight-adjusted reference values of normal
individuals. These effects were mainly confined to a subgroup of
10 nonsmokers. In addition, nonsmokers also showed a
statistically significant small decrease in FEF^.^j when
compared to reference values. The lung function decrements did
not appear to persist during a period of no formaldehyde
exposure. After a 4-week period of work interruption, lung
function among ndnsmoke1.0 ppm) had sharply declined in
the second examination (mean=0.55 ppm) due to process changes and
automation.
Post-shift declines in lung function has also been assessed
in controlled-exposure studies. Sauder et al. (1986) noted
statistically significant decreases in two lung function
parameters (FEV1>0 and FEF^.^) at 30 minutes into a 3 hour
exposure scenario in a study of nine healthy nonsmokers exposed
to 3 ppm formaldehyde. These decreases were not present at 60 or
180 minutes, and pulmonary function was normal at 24 hours.
Similar findings were reported in a study by Green et al. (1987).
Healthy subjects (n=22) performing heavy exercise during exposure
for 1 hour to 3 ppm formaldehyde showed small but statistically
significant decrements in FEV1 0, FVC, and FEV3 0. A similar
pattern of decrease in FEF25X.^X was observed, but the change was
not statistically significant. Three other controlled-exposure
studies (Witek et al., 1986; Schachter et al., 1986; Schachter et
al., 1987) did not find any measurable deficits in lung function.
Exposures in these studies were lower (2 ppm) and for a shorter
time (40 minutes).
106
-------
In summary, formaldehyde, possibly along with other agents,
appears to produce transient lung function changes, but does not
appear to produce any chronic decrements. Five of the newer
studies (Horvath et al., 1988; Uba et al., 1989; Alexandersson
and Hedenstierna, 1989; Sauder et al., 1987; Green et al., 1987 )
demonstrated transient lung function changes after short-term
controlled exposures or over a working shift. Although deficits
in volume parameters (FEV or FVC) were noted in all four studies,
no consistent deficit* in a single parameter (e.g., FEV) was noted
across the studies. In those studies reporting statistically
significant deficits in lung function, exposure levels were
higher in the controlled-exposure studies (3 ppm) than in the
occupational studies where average values were approximately 1
ppm, with higher peak exposures. The populations studied by
Horvath et al. (1988) and Alexandersson and Hedenstierna (1989)
also had exposures to particles from wood dust. Exposure to
particles in combination with formaldehyde may be important in
initiating the post-shift changes in lung function observed at
the lower formaldehyde levels, however, the role of combined
exposures cannot be determined from these studies.
Three studies of occupationally-exposed populations (Malaka
and Kodama, 1990; Kilburn et al., 1989; Alexandersson and
Hedenstierna, 1988) reported declines in baseline tests. It is
difficult, however, to draw firm conclusions from these
observations since the better designed and conducted studies did
not demonstrate any chronic deficits. Other factors may explain
the observations. It is difficult to separate the effects of
formaldehyde from any possible contribution from other exposures,
such as solvents for the studies of Kilburn et al. (1989) and
Alexandersson and Hedenstierna (1988). In addition, the study
population in Malaka and Kodama (1990) had a high prevalence of
respiratory infectious diseases which may have accounted for some
of the lung function decrement. Due to the cross-sectional
design of the study, it cannot be determined whether formaldehyde
may have enhanced any lung function effects related to
preexisting chronic disease.
6.2.3 Respiratory Effects in Asthmatics
Information on possible formaldehyde-related respiratory
effects in asthmatics was derived from controlled-exposure
studies and one occupational study. Asthmatics were studied to
determine whether they may be a more sensitive subpopulatioi .
Asthmatics are known to be more sensitive to other environmental
agents such as SO2 (Witek et al., 1986).
In the"prospective study of medical students (Uba et al.,
1989) , 12 of the 103 subjects identified, themselves as having a
previous history of asthma, but no clinical verification was
performed. Asthmatics reported a statistically significantly
greater prevalence of burning and watering eyes with acute
formaldehyde exposure when compared self responses in the absence
of formaldehyde exposure. No differences existed between
107
-------
asthmatics and nonasthmatics regarding the prevalence of eye
irritation associated with acute exposure, nor were differences
noted in the development of respiratory symptoms from chronic
exposure over a 7-month period to formaldehyde, with respect to
lung function testing, no clear evidence of a bronchoconstrictive
response existed for asthmatics. As these investigators noted,
the absence of clinically significant bronchoconstriction among
the 12 asthmatic subjects does not provide strong evidence
regarding the brondhbconstrictive effects of formaldehyde since
the number of asthmatics in the study was small and the diagnoses
were not verified.
Asthmatic subjects in controlled-exposure studies exposed to
2 ppm formaldehyde for 40 minutes at rest and with exercise
(Witek et al., 1986, 1987), 3 ppm formaldehyde for 1 hour at rest
or with exercise (Green et al., 1987), or 3 ppm formaldehyde for
3 hours at rest (Sauder et al., 1987) reported upper respiratory
symptoms and eye irritation, however, the incidence of response
was not greater for asthmatics than for healthy individuals.
Witek et al. (1986, 1987), Sauder et al. (1987), and Green
et al. (1987) observed that asthmatic individuals exposed to
formaldehyde at 3 ppm for 1 or 3 hours or 2 ppm for 40 minutes
had common irritative symptoms as that reported previously
(Section 6.2.1) for healthy individuals, i.e., unusual odor,
sore throat, nasal discharge or stuffiness, and eye irritation.
The incidence of nose/throat irritation and eye irritation as
reported by Sauder et al. (1987) and by Green et al. (1987) was
statistically significantly elevated. Likewise, lower
respiratory symptoms were either mild or absent (Witek et al.,
1987) . In comparisons with healthy subjects, no difference
existed in the incidence of response for eye and upper
respiratory tract irritation between asthmatics and nonasthmatics
(Witek et al. , 1987; Green et al., 1987).
Acute formaldehyde exposure did not appear to induce acute
airway obstruction or bronchoconstriction as measured by lung
function testing during or following exposure (Witek et al.,
1986, 1987; Sauder et al., 1987; Green et al., 1987). Witek et
al. (1987) in a study of 15 asthmatics did not observe any
deficits in either volume or flow parameters. Rather,
formaldehyde exposure appeared to lower the dose of methacholine
necessary to cause a 20% drop in FEV,, as assessed in the
methacholine inhalation challenge (MIC) test, in eight of the 12
subjects. Witek et al. (1987) concluded that formaldehyde
exposure may alter nonspecific airway hyperresponsiveness in
asthmatics, but does not cause any functional impairment.
Sauder et al. (1987) observed similar results from,-
spirometry testing with exposure to 3 ppm formaldehyde for 3
hours at rest, i.e., no deficits in lung function were detected.
Unlike Witek et al. (1987), however, Sauder et al. (1987) did not
observed any statistically significant changes in lung function
parameters for airway reactivity (MIC). In addition, Green et
108
-------
al. (1987), in a study comparing the responses of 16 asthmatics
and 22 nonasthmatics to 3 ppm formaldehyde for one hour during
mild and heavy exercise, reported no differences in severity of
effects for lung function parameters and upper respiratory
symptoms. These investigators also did not find any changes in
nonspecific airway reactivity.
In summary, results from these studies indicate that
asthmatics, like healthy individuals, experienced eye and upper
respiratory tract irritant symptoms. Formaldehyde exposure does
not appear to exaggerate the incidence of response. Further,
these studies do not provide evidence that asthmatics are more
sensitive than nonasthmatics to any formaldehyde-related effects
on the lower respiratory tract.
6.2.4 Immunol'ogic Effects
The presence of IgG antibodies against formaldehyde-human
serum albumin conjugates and .human serum albumin (HSA) was
described previously in EPA (1987). Several reports published
since 1987 measured antibodies to formaldehyde-HSA in mobile home
dwellers and in workers with potential exposure to formaldehyde.
Cell-mediated immunity alterations were also observed in
residents of mobile, UFFI, or recently redecorated homes.
Wilhelmsson and Holmstrom (1987) examined IgE levels in 30
workers in a factory producing formaldehyde. Roughly 40% of
these workers had nasal symptoms such as rhinitis, nasal
obstruction, and nasal discharge. Immunoglobulins other than IgE
were not assayed in this study. These investigators observed
that two of the 30 workers had elevated levels of IgE to
formaldehyde-HSA. Both these workers had complaints of nasal
symptoms clearly associated with the workplace. Wilhelmsson and
Holmstrom (1987) believed these two cases indicated that long-
term inhalation exposure to formaldehyde may cause sensitization
and trigger a classical IgE-mediated allergic reaction.
In a series of reports, Thrasher et al. (1987, 1988, 1989,
1990) examined immunologic parameters among patients with
symptoms which the investigators believed were related to
multiple chemical sensitivity. In the latest report, Thrasher et
al. (1990) assessed immune activation and autoantibodies in four
groups of subjects which included previously studied subjects
(Thrasher et al., 1987, 1988, 1989), in addition to other
symptomatic individuals. A control group, composed of
asymptomatic subjects, had acute exposure to formaldehyde some 12
months previously. The investigators considered the symptomatic
subjects in the other four groups to represent chronic
formaldehyde exposure. The five groups were: 28 asymptomatic
chiropractic students (controls) exposed to formaldehyde for 13
hours/week for 28 weeks one year prior to the study; 19 current
mobile symptomatic home residents, some of which were studied by
Thrasher et al. (1987); 21 office workers with multiple health
complaints, some were studied by Thrasher et al. (1989) ; 21
•V
109
-------
patients with multiple symptoms who had been removed from mobile
homes for at least one year; and eight occupationally-exposed
patients, 6 of whom were studied by Thrasher et al. (1988).
Formaldehyde levels were directly measured for two groups, in
current mobile home residents (0.05 to 0.5 ppm) and in the
removed mobile home residents (0.14 to 0.81 ppm), and were
inferred frojn other monitoring data for office workers (0.01 to
0.77 ppm) and for consols (0.43 ppm).
Thrasher et al. (1990) observed a statistically
significantly greater proportion of positive anti-formaldehyde-
HSA titers (either IgM, IgE, or IgG) in each of the symptomatic
groups (75 to 100%) when compared to asymptomatic controls (39%) .
This finding was also reported in the earlier studies (Thrasher
et al., 1987, 1988, 1989). The number of Tal cells was
statistically significantly increased in all four symptomatic
groups compared to the controls, whereas only two of the four
symptomatic groups showed statistically significant increases in
IL2+ receptor (acute antigenic stimulation) (mobile home
residents and removed mobile home patients) and number of B cells
(office workers and removed mobile home patients). Previous
reports showed increased numbers and percent of Tal cells among
occupationally-exposed subjects (Thrasher et al., 1988) and no
difference in B cell number or blastogenesis in current mobile
home dwellers (Thrasher et al., 1987). Of note, even though
differences existed between symptomatic groups and controls, all
values for IL2 + and B cells fell within the expected reference
ranges. With respect to autoantibodies, antiparietal cell was
the most frequently found autoantibody in all groups, with the
lowest percentage in controls.
Pross et al. (1987) in a controlled-exposure study examined
the effects of UFFI off-gassing products (free-formaldehyde) on
hematologic and immunologic parameters in 23 asthmatics living in
UFFI-homes and four asthmatics from conventional homes.
Asthmatics were considered to be a possibly sensitive
subpopulation. The subjects, who served as their self control,
received exposure in an environmental chamber to room air,
formaldehyde gas (1.0 ppm for 3 hours), and formaldehyde-free
UFFI off products (0.5 particles/ml, 4 um, 3 hours) in sequence.
Blood samples for testing were obtained before the exposure
series, one day after completion, and 7 days later. The
hematologic and immunologic parameters examined by these
investigators included: complete blood count and differential,
erythrocyte sedimentation rate, lymphocyte subpopulations (E-
rosetting, T3, T4, T8, FcR,' and Fc receptor positive lymphocytes
and large granular lymphocytes), lymphocyte response to
phytohemagglutinin and formalin-treated red blood cells (Form-
RBC), serum antibody against the Thomsen-Friedenreich RBC''antigen
and against Form-RBC, natural killer (NK), interferon-boosted NK,
and antibody-dependent cell-mediated cytotoxicity.
These investigators found that long-term chronic exposure to
UFFI off-gassing products had no effect on the immunologic
110
-------
parameters studied, i.e., no differences existed between groups
in the preexposure blood samples. Rather, short-term acute
exposure resulted in minor immunologic changes among the UFFI-
insulated home subjects. Analyses comparing before and after
UFFI chamber exposure demonstrated small but statistically
significant increases in the percentage of eosinophils,
basophiles,.^nd T8 positive cells in the UFFI-insulated home
group only. T,he authors commented that the mechanism of the
eosinophilia in chrtfnic allergic states has not been established,
and the observed levels were not remarkable compared to levels
reported in patients with chronic hypersensitivity reactions.
They also noted that the small change (10% increase) in T8
positive cells may not be biologically relevant.
These studies are inadequate for determining whether the
immunologic changes are formaldehyde related. First, certain
antibodies to formaldehyde appear in the sera of exposed
individuals, but not always the same antibodies. Some exposed
subjects had elevations in IgE levels whereas other exposed
subjects had elevations in IgG. Second, it cannot be determined
from these studies whether the changes in immunologic parameters
resulted from the health effects, whether they represent only
markers of exposure, or whether they are part of the mechanism in
the production of chemical hypersensitivity. Third, none of the
reports provide information concerning exposure levels at which
immunologic responses might have been initiated. Fourth, three
(Wilhelmsson and Holmstrom, 1987; Thrasher et al., 1988, 1989)
did not contain comparison groups. The lack of comparison groups
makes it difficult to determine the magnitude of any observed
effect.
6.2.5 Central Nervous System Effects
Reports in the literature link formaldehyde with a number of
behavioral and physiological effects such as thirst, dizziness
and apathy, inability to concentrate, and sleep disturbances
(EPA, 1987). Only one report published since 1987 assessed
neurological effects and formaldehyde exposure. Kilburn et al.
(1987) studied 305 histology technicians attending two annual
conferences. Attendance at either of the two conferences made
the technician eligible for inclusion in this study. Increased
daily hours of exposure to formaldehyde, as assessed over the
duration of the technician's employment in any histology
laboratory, were correlated with poor performance on several
indicators of neurological functioning which reflected dexterity,
balance, coordination, and choice performance. While age was
also related to poor performance on these indicators, performance
was not related to years of cigarette smoking, educational level,
or solvent exposure.
These investigators did not believe that solvent exposure
was responsible for the observed deficits in psychological
parameters attributable to formaldehyde, although solvent
exposure and formaldehyde exposure were individually
111
-------
statistically significantly associated with memory loss. The
effect of solvent exposure on memory loss was not as great as the
formaldehyde exposure contribution.
The Kilburn et al. (1987) study is inadequate for hazard
identification. First, neurological functioning was not assessed
in an unexposed group. Second, although the researchers do not
believe that solvent\epcposure is as important as formaldehyde, it
is still difficult to" separate the effects related to solvent
exposure from those of 'formaldehyde exposure. Third, this
prevalence study may be biased by self-selection. Reasons for
volunteering for this study are not known. Fourth, the
formaldehyde levels at which decrements in neurological
functioning are observed cannot be determined from this study.
Formaldehyde was measured in only four histology laboratories.
Levels found in these laboratories may not be characteristic of
all formaldehyde exposures encountered by these histology
technicians.
6.2.6 Limitations
The data on the irritant effects of formaldehyde came from
controlled-exposure studies, prospective (cohort) studies,
prevalence (cross-sectional) studies or case reports. Case
reports such as Thrasher et al. (1987, 1988, 1989, 1990) are
often used to generate hypotheses, but they have serious
limitations for making causal inferences. Case-reports lack
information with which to decide the magnitude of the reported
symptom or any association with exposure. In this situation, the
size and characteristics of the population at risk, from which
the case was drawn, are not known, neither how it differs from a
population without risk.
The majority of the studies were designed as cross-sectional
studies, also known as surveys (Alexandersson and Hedenstierna,
1988; Holness and Nethercott, 1989; Horvath et al., 1988; Imbus
and Tochilin, 1988; Kilburn et al., 1988; Malaka and Kodama,
1990; Broder et al., 1988a, 1988b, 1988c; Liu et al., in press,
personal communication; Ritchie and Lehnen, 1987; Boysen et al.,
1990; Edling et al., 1988; Holmstrom et al., 1989; Wilhelmsson
and Holmstrom, 1987; Kilburn et al., 1987). In these studies,
study individuals are not followed over a period of time and
health effects are assessed at the same time as exposure. It is
difficult to determine whether exposure produced the observed
effect. In addition, the study of Wilhelmsson and Holmstrom
(1987) did not incorporate an unexposed comparison or control
group. Without a control group, the attributable magnitude of a
reported symptom cannot be determined.
Prospective studies (Uba et al., 1989; Nunn et al., 1990;
Alexandersson and Hedenstierna, 1990) and controlled-exposure
studies (Green et al., 1987; Sauder et al., 1986, 1987; Schachter
et al., 1986, 1987; Witek et al., 1986, 1987; Pross et al., 1987)
follow a group of individuals, thus, disease development can be
112
-------
inferred subsequent to exposure. These later studies carry the
greatest weight in an assessment of causality.
There is uncertainty regarding whether the statistically
significant elevations in an observed effect is real or simply
reflecting statistical variability related to multiple testing
(i.e., the observation is a false positive). The majority of the
reviewed studies assessed several different endpoints such as eye
and upper tract respiratory effects, lower airway symptoms, and
lung function. Due to the many statistical comparisons, a larger
probability exists that several of these relationships will be
falsely positive, but their identify can not be determined.
Confounding and bias introduce other uncertainties in these
studies. Wood dust and particulates were found simultaneously
with formaldehyde in several studies ((Alexandersson and
Hedenstierna, 1988; Alexandersson and Hedenstierna, 1990; Horvath
et al., 1988; Imbus and Tochilin, 1988; Malaka and Kodama, 1990;
Boysen et al., 1990; Edling et al., 1988; Holmstrom et al., 1989;
Wilhelmsson and Holmstrom, 1987). Unless noted, results were not
adjusted for these exposures and an observed effect may not be
due to formaldehyde alone.
With respect to non-occupational factors such as sex, age,
and smoking, only the studies of Liu et al. (in press) and Broder
et al. (1988a, 1988b, 1988c) adjusted for potentially confounding
effects in their statistical analyses. In addition, these
factors are not considered problems in studies where subjects
served as served as their own control (Uba et al., 1989; Nunn et
al., 1990; Alexandersson and Hedenstierna, 1990; Green et al.,
1987; Sauder et al., 1986, 1987; Schachter et al., 1986, 1987;
Witek et al., 1986, 1987; Pross et al., 1987).
Bias may have be introduced in the controlled-exposure
studies since the steady subjects studied were presumably self-
selected and may not be representative of the general population.
Likewise, the cross-sectional studies of occupational populations
(i.e, Nunn et al., 1990), by nature of their design, did not
include individuals who left employees. Survivor bias can not be
completely ruled out in these studies.
6.2.7 Summary and Conclusions
The irritative effects of formaldehyde on the eyes and upper
respiratory tract are well-documented (EPA, 1987); a large number
of observations of people-from controlled-exposure and
observational settings support a conclusion that the generally
observed range over which most people experience irritation of
the eyes and upper airways is 0.1-3.0 ppm of formaldehyde. The
studies published since 1987 added further evidence of acute
sensory irritation of the eye and upper respiratory airways at
levels identified previously. In addition, exposure-response
relationships for eye, throat, or nose irritation were present in
three studies (Ritchie and Lehnen, 1987; Horvath et al., 1988;
113
-------
Liu et al., in press), and provided support that formaldehyde
induces these effects. These findings corroborated the exposure-
response gradients observed in several studies reviewed earlier
by EPA (1987) . See Section 7 for further analysis of the dose-
response relationships.
Inhalation exposure to formaldehyde may also cause increased
severity of 'Tifstoiogical changes (metaplasia and dysplasia) in
the nasal epithelial mucosa of occupationally-exposed workers
compared to non-formaldehyde exposed occupational controls
(Edling et al., 1988; Holmstrom et al., 1989; Boysen et al. ,
1990) . Formaldehyde levels in these exposed populations ranged
from 0.04-2.0 ppm. In addition to formaldehyde, these
occupational populations had exposure to particles from
formaldehyde resins or from wood dust. No exposure-response
relationship was noted between histologic score and duration of
employment. Duration of exposure is often an imprecise surrogate
with which to examine exposure trends. Caution must be employed
when making inferences from these observations due to biases
introduced by the small number of exposed persons, the cross-
sectional design of the studies (essentially survivor populations
were studied and the most susceptible individuals were not
available), and the inability to identify a possible contribution
from the presence of particles.
Four of the reviewed studies (Broder et al., 1988b, 1988c;
Horvath et al., 1988; Uba et al. 1989; Malaka and Kodama, 1990)
reported lower airway irritation symptoms at mean exposure levels
less than 1 ppm. These effects were reported among particleboard
workers, medical students in anatomy laboratory, and UFFI-home
populations. The occupationally-exposed groups also experienced
peak exposures to concentrations greater than 1 ppm, but less
than 5 ppm. Other cross-sectional studies of chronic exposure
and controlled-exposure studies of acute exposures to 2 ppm
formaldehyde for 40 minutes or 3 ppm formaldehyde for 3 hours did
not produce symptoms of lower airway effects. Results from these
studies may be limited by the small number of subjects studied
and by selection bias. In addition, exposures in some of the
cross-sectional studies were close to background, 0.05 ppm.
Overall, the results of studies published to date suggest that
onset of symptoms of lower airway irritation due to formaldehyde
exposure may occur at concentrations ranging somewhere in the
range of 1 ppm and 5 ppm. These exposure levels are lower than
had been reported previously (EPA, 1987).
Small transient (acute) effects on volume and flow lung
function parameters were suggested in four studies (Horvath et
al., 1988; Uba et al., 1989; Sauder et al, 1987; Green et al.,
1987). These more recent observations appeared consistent with
those observed in previously reviewed studies (EPA, 1987). Two
studies of chronic exposure, a prospective study of medical
students (Uba et al., 1989) and a prevalence study of
particleboard workers (Horvath et al., 1988), observed
statistically significant decreases in FVC (both studies) and FEF
114
-------
(Horvath et al. , 1988) between pre-exposure and post-exposure
tests. Two acute exposure studies (the controlled-exposure
studies Sauder et alt (1987) and Green et al. (1987)) observed
statistically significant deficits in FEV (both studies), FEF
(Sauder et al., 1986), and FVC (Green et al., 1987) between pre-
exposure and post-exposure tests. Mean formaldehyde levels in
the Uba et al. (1989) and Horvath et al. (1988) were around 1
ppm, with peaks up to,5 ppm (Uba et al., 1989). Subjects in the
controlled-exposure studies received 3 ppm formaldehyde for 3
hours.
EPA weighed the above four studies more heavily due to their
better designs (except for Horvath et al. (1988), all were
controlled-exposure or prospective studies) and their careful
conduct (i.e., use of ATS-recommended procedures and control for
possibly confounding variables). The transient effects on lung
function have been slight, possibly explaining why other cross-
sectional studies reporting upper respiratory effects did not
observe these effects. In addition, these other studies
contained limitations related to fewer numbers of subjects,
exposures close to background, or selection bias, which may
partly explain the differences in observations. There was no
evidence for chronic decrements in pulmonary function as
reflected by no observed changes in pre-shift lung function
measurements in occupationally-exposed populations. The reviewed
studies, however, are limited in their ability to adequately
assess any possible relationship.
Data examining possible associations between immunological
changes and poorer neurological functioning and inhalational
formaldehyde exposure were considered inadequate for hazard
identification.
7.0 ESTIMATES OF UPPER RESPIRATORY AND EYE IRRITATION RISKS
In 1987, the EPA reviewed several studies for determining
the dose-response characteristics for upper respiratory and eye
irritation in populations exposed to formaldehyde. The upper
respiratory system is a likely target of formaldehyde's effects
since inhalation is a primary route of exposure. At that time,
response over a range of formaldehyde levels was estimated for
four studies (Hanrahan et al., 1985; Andersen and Molhave; 1984;
Bender et al, 1983; Kulle, 1985) using logistic regression
analyses. None of these studies proved adequate data to.quantify
population risks for the irritant effects of formaldehyde. These
studies were limited due to possible selection bias, their design
(some measured only prevalence) and to small numbers of
participants. These studies, however, provided a qualitative
estimate of population response over a wide exposure range and
quantitative estimates of response for very select populations.
115
-------
7.1 Incorporation of New Data
Three studies (Ritchie and Lehnen, 1987; Horvath et al.,
1988; Liu et al, in press) released since 1987 demonstrate dose-
response relationships between formaldehyde exposure and the
prevalence of irritation of the eyes and/or nose and throat.
Prevalence is the number of cases existing with the outcome at a
single point -in time.
\v*
Ritchie and Lehnen (1987) examined health complaints in a
study of mobile and convention home residents. Study
participants were identified from complaints the Minnesota
Department of Health received from physicians. These
investigators observed positive dose-response relationships
between formaldehyde concentration and eye irritation,
nose/throat irritation, headaches, and skin rash. In these
analyses, formaldehyde concentration was categorized as £0.1 ppm,
0.1-0.3 ppm, and >0.3 ppm. Age, sex, and smoking (either active
smoking or exposure through passive smoke) were identified as
statistically significant confounders, and were controlled in the
analyses.
The Horvath et al. (1988) study of particleboard workers
observed a concentration (ppm)-dependent relationship between
formaldehyde and the prevalence of burning of the nose. These
workers also had exposures to nuisance particles, which included
respirable particles; nuisance particles were not adjusted for in
the analyses. Eight-hour time weighted averages (TWA) for
formaldehyde ranged from 0.2-2.93 ppm (mean 0.69 ppm). The me^an
respirable dust level was 0.11 mg/m (range of 0.025-1.06 mg/m ) .
Liu et al. (in press) observed a positive exposure-response
relationship between tearing and/or burning eyes and cumulative
formaldehyde exposure (ppm-hour) among occupants from mobile
homes in California. Mobile homes in this study were randomly
selected from a list of all mobile homes registered by the
California Department of Housing and Community Development.
Approximately 60% of the home were manufactured after 1980.
Average formaldehyde concentration in this study was around 0.09
ppm (range: level of detection, 0.01 ppm,-0.46 ppm). Additional
data provided to EPA by Liu (personal communication with C.
Siegel Scott, 14 February 1991) suggested a concentration-
dependent effect (Figure 7-3). Statistical tests for linear
trend were not statistically significant (p>0.05), however.
These authors accounted for two possibly confounding
exposures in their logistic regression analysis examining
cumulative exposure; sex and previous chronic illness were
statistically significantly associated with burning eyes. Since
the prevalence of response was higher in summer than in winter,
season, also, appears to be a modifying factor. The analyses
performed by EPA examining the relationship with concentration
did not adjust for these confounders and are limited by this
fact.
116
-------
Figures 7-1 through 7-7 present the dose-response
relationships observed in Ritchie and Lehnen (1987) (Figures 7-1
and 7-2), Horvath et al. (1988) (Figure 7-3), and Liu et al. (in
press; personal communication) (Figures 7-4 and 7-7). The dose-
response relationships estimated for the data of Horvath et al.
(1988) was obtained using logistic regression analyses, and
presented by the authors. Estimated prevalences of response
simultaneously" taking into account possibly confounding factors
such as sex, gender, and smoking were not presented by Ritchie
and Lehnen or by Liu et al. (in press; personnel communication),
even though these investigators discussed results of regression
analyses controlling for these possibly confounding factors. The
figures associated with these two studies (Figures 7-1, 7-2, 7-4,
and 7-5), therefore, show the unadjusted data for the study
participants.
One can relate responses in the newly reviewed studies to
those predicted from the data identified in Section 8 of EPA
(1987). It must be acknowledged that this comparison contains
limitations since each study characterized endpoints differently,
although all endpoints can be broadly defined as irritation of
mucous linings. As identified in EPA (1987), Hanrahan et al.
(1984) presented data for eye discomfort only, both Andersen and
Molhave (1984) and Kulle (1985) assessed the combined symptoms of
eye, nose, and throat irritation, and Bender et al. (1983)
measured eye irritation.
Response for irritation of the eye and of the nose/throat as
reported by Ritchie and Lehnen (1987) (Figures 7-1 and 7-2) show
a marked increase as exposure increases from <0.1 ppm to >0.3
ppm. The prevalence of eye irritation and nose/throat irritation
associated with exposure between 0.1-0.3 ppm appears to be
similar to the predicted prevalence at around 0.2 ppm reported by
Hanrahan et al. (1984) for burning eyes and by Andersen and
Molhave (1984) for the combined effects of eye, nose, and throat
irritation. In contrast, the slope of the curve of Horvath et
al. (Figure 7-3) for burning nose is not as steep, i.e., response
associated with a particular exposure level is lower, as curves
identified in Section 8 of EPA (1987) for Hanrahan et al. (1984),
Andersen and Molhave (1984), or Bender et al. (1983). Rather,
response in Horvath et al. is more similar to that of Kulle
(1985) for the combination of eye, nose, and throat irritation.
Liu et al. (in press) found that cumulative exposure (ppm-
hour) (Figure 7-4) best described the relationship between
burning and/or tearing eyes. Due to differences in exposure
units, however, a strict comparison can not be made with the
other studies. It is best to base a comparison on common
exposure units, concentration (ppm), as shown in Figure 7-5. The
data of Liu et al. (personal communication) can only be compared
to the data of Ritchie and Lehnen (1987) since exposure levels in
these two studies roughly overlap. All the other studies
examined irritant effects associated with higher levels of
formaldehyde exposure. For exposure <0.10 ppm, Liu et al.
117
-------
100
90 -
80 -
70 -
.2 60 -
03
O) cn i
c 50 -
O
a
CD
40 H
30 -
20 -
10 -
Figure. 7-1. Health Effects by HCHO
in Complaint Mobile Homes
Reference: Ritchie and Lehnen, 1987
Eye Nose/Throat Headache
< 0.1 ppm
0.1 to 0.3 ppm
Skin Rash
> 0.3 ppm
118
-------
100
90 -
80 -
70 -
•2 60 H
03
O
D.
CD
40 H
30 -
20 -
10 -
Figure 7-2". Health Effects by HCHO
in Complaint Conventional Homes
Reference: Ritchie and Lehnen, 1987
Eye Nose/Throat Headache Skin Rash
< 0.1 ppm
0.1 to 0.3 ppm Iff > 0.3 ppm
119
-------
NJ
o
Figure 7-3. Predicted prevalence of burning nose
over range of formaldehyde concentrations in ppm
80 -q
60 -5
Q)
o
c
-------
Figure 7-4. Observed burning/tearing
eye Irritation response over a range
of HCHO levels
25
20
15
Q)
O)
03
C
Q)
O
10
0
Data from Liu, in press
A
0-7.0 7.0-12.0 >12.0
HCHO Exposure (ppm x hrs)
121
-------
25
Figure 7-5. Observed burning/tearing
eye irritation response over a range
,, of HCHO levels
(for those whose average time
spent at home is > 14 hours)
Data from Liu (personal communication)
20
15 h
o>
O)
2
c
Q)
U
U.
0)
0_
10 |-
5 h
0
•A
*
0
<0.05 0.05-0.10 >0.10
HCHO Exposure (ppm)
122
-------
(Figure 7-5) reported a higher prevalence of burning/tearing eye
irritation than that reported by Ritchie and Lehnen (1987) for
eye irritation.
7-2 Discussion of the Overall Evidence
A range- of ^predicted responses for a given formaldehyde
concentration is obtained when the seven studies (Bender et al.,
1983; Hanrahan "et al., .1984; Horvath et al., 1988; Kulle, 1985;
Liu et al., in press; Molhave and Andersen, 1984; and Ritchie and
Lehnen, 1987) are examined comparatively. This may be due to
several limitations which prevent the use of these studies to
infer the magnitude of general population risks. Three of these
studies (Kulle, 1984; Bender et al., 1983; Andersen and Molhave,
1984) were of controlled-exposures. These studies did not
utilize a randomization scheme. Study participants were self-
selected and may not be representative of the general population.
Confounding may influence the magnitude of response. Only
the studies of Ritchie and Lehnen (1987) and Liu et al. (in
press) were able to account for possibly two confounding
variables, smoking and gender. Unfortunately, predictions of
response in analyses controlling for these confounding variable
could not be obtained from either study. In addition, Liu et al.
further suggest that season and average time spent at home may be
important covariates. None of the other studies accounted for
these factors.
The population studied by Horvath et al. (1988) had nuisance
dust exposure, and it is not known how this may confound
response. The response predicted by the Horvath et al. data,
however, is not as steep as that predicted from the data of
Hanrahan et al. (1984) and Andersen and Molhave (1984). Nuisance
dust particles, therefore, does not appear to greatly influence
the prevalence of irritation response.
Most important, a wide variation in response across all
studies is not uncommon since these studies did not incorporate
similar categorizations of symptoms. Each study identifies eye
or upper respiratory tract involvement, however, each study
defines and groups those effects differently. For example,
response to tearing and/or burning of the eyes may be guite
different than for only burning eyes. Also, these studies report
subjective and self-reported symptoms. These symptoms have not
been medically verified, and thus, results may be biased by over-
or under-reporting.
In conclusion, caution must be taken in inferring the
results in EPA (1987) and in Figures 7-1 through 7-5 to the
general population. Limitations in these studies at the present
prevent the inference of eye and upper respiratory risks. None
of the studies reviewed in this document and in EPA (1987)
provide adequate data to precisely quantify general population
risks for eye and upper respiratory effects associated with a
123
-------
specific formaldehyde concentration. Nevertheless, these studies
document eye and upper respiratory tract effects at levels
previously identified, 0.1 ppm to 3 ppm. Even though the
prevalence of exposure can not be precisely estimated for a given
formaldehyde concentration, these studies support the conclusion
that the number of individuals responding in a population will
increase with -increasing formaldehyde concentration.
8.0 RISK CHARACTERIZATION
This section presents the major conclusions of EPA's risk
assessment of formaldehyde. Some subsections have been taken
from the 1987 assessment document and modified as indicated by
subsequent developments. The current document reviews the
underlying scientific foundation for the findings,, describes the
strengths and weaknesses of the supporting data, and discusses
the uncertainties attending EPA's interpretation of the data and
projection of risk. The risk characterization discusses the
qualitative aspects of the risk assessment and the quantitative
risk estimations at exposure levels relevant for the population
of concern.
8.1. Noncancer Effects
The major noncancer effects posed by exposure to
formaldehyde are due to the irritating nature of the chemical.
These effects include sensory irritation which is readily
perceived by the exposed individual; inhibitory effects on the
mucociliary system and cellular changes of the nasal cavity,
which in turn may result in increased susceptibility to
respiratory infections. Newer studies have failed to provide
evidence linking formaldehyde exposure to either immunological,
neurologic, or pulmonary sensitization effects. This document,
however, has not included an exhaustive examination of the total
spectrum of health points of possible concern.
8.1.1. Sensory Irritation
The well-documented health effects from acute and repeated
inhalation exposures to formaldehyde are concentration dependent.
These effects include irritation of the eyes, nose, throat and
lower airways, the intensity of which is dependent upon the
extent and duration of exposure, and may result in extreme
discomfort and inability to function normally at work or in
routine daily activities.
Dose-response relationships for irritation effects obtained
from studies of selected populations cannot be generalized to
give the probability of response in larger populations. However,
qualitative evidence supports the expectation that the number of
persons who respond in a population will increase wich increasing
concentrations of formaldehyde. A large number of observations
of people in various clinical and nonclinical settings support a
conclusion that the generally observed range over which most
124
-------
people experience irritation of the eyes and upper airway is 0.1-
3.0 ppm of formaldehyde.
Symptoms of eye, nose and throat irritation above background
incidence frequently occur at about 0.1 ppm and become widespread
at concentrations near 3.0 ppm. Exposures greater than 3.0 ppm
are generally intolerable for more than short periods. Lower
airway irritation1 has'-.been noted at formaldehyde concentrations
that range from around" 1 ppm to below 5 ppm. The latter
irritation effects were transient and there were no indications
of chronic lower airway impairment. Tolerance to low levels of
formaldehyde can occur in individuals after 1-2 hours of
exposure, but symptoms can return if exposure is interrupted and
then resumed. These acute irritation effects are usually
reversible upon removal of the source of exposure.
8.1.2. Mucociliary Clearance Effects
A major function of the nose is to prepare the inhaled air
for the lungs. This includes warming, moistening, and filtering
the inspired air. Dust and many bacteria found in the inspired
air are precipitated in the mucus that bathes the mucous membrane
and are moved outward by the action of the cilia of the nasal
passage. Research indicates that formaldehyde has a number of
effects on the workings of this mucociliary apparatus.
Very limited data in humans have shown that nasal
mucociliary function is inhibited by formaldehyde exposure at
levels as low as 0.3 ppm (Anderson and Molhave, 1983). The
stopping of mucous flow (mucostasis) followed by cessation of
ciliary activity (ciliastasis) was clearly shown in male rats
following repeated exposure to 15 ppm formaldehyde; only slight
effects were noted in animals exposed at lower concentrations;
and at 0.5 ppm no effects were observed (Morgan et al., 1983,
1986).
Overall, the animal data show that at low formaldehyde
concentrations (somewhere less than 2 ppm), the mucus layer of
the nasal cavity can trap and remove a large proportion of
inhaled formaldehyde. At higher formaldehyde concentrations
(especially above 6 ppm), there is a pronounced inhibition of
mucociliary activity. The results obtained indicate an apparent
nonlinear dose-dependent on the rat mucociliary apparatus which
suggests possible saturation. Once the mucus layer is saturated,
the mucociliary clearance system may be seriously compromised and
allow a greater amount of formaldehyde to reach the nasal
epithelium.
A reduction in the efficient operation of the mucociliary
system can increase the risk of persons exposed to formaldehyde
to develop respiratory disease since it is an important defense
mechanism.in the removal of foreign particles and bacteria.
125
-------
8.1.3. Cellular Changes
The primary point of contact of formaldehyde upon exposure
by inhalation is the internal lining of the nose. It appears
that inhalation of formaldehyde above some threshold level, which
varies from person to person, causes a number of cellular effects
which can impair the normal functioning of the nose and are
dependent on the concentration and duration of exposure.
Cellular changes considered in this analysis include
cytotoxicity, cell proliferation, and histological changes.
Evidence of cellular damage in humans is limited to results
of biopsies of the nasal turbinates. Several studies (Edling et
al., 1988; Edling et al., 1985; Holmstrom et al., 1989; Boysen,
et al., 1990) in which humans were occupationally exposed from
nine to twenty years to formaldehyde in the range of 0.04-2 ppm,
time-weighted average (TWA) concentration with peak exposures up
to 4 ppm, showed loss of cilia, development of squamous
metaplasia, and a few workers exposed for the longest durations
showed cellular dysplasia. These findings are consistent with
the known irritating properties of formaldehyde. Overall, the
severity of the squamous metaplasia induced in these workers was
only sligthly greater than that observed in control groups. The
incidence of cellular dysplasia— a lesion that may be considered
precursor to tumor development— was low and it showed no
consistent correlation with either concentration or duration of
exposure. Finally, caution must be used when generalizing
observations from these studies because of the small number of
exposed persons examined and the possibility of confounding
exposure to wood dust and resin particles.
The cumulative evidence from animal studies indicates that
the effect of formaldehyde on cell proliferation in nasal
passages is time and concentration dependent. Short-term
formaldehyde exposure (6 ppra) of rhesus monkeys causes
significant increases in the rates of cell proliferation in the
damaged nasal regions (squamous metaplasia); cell proliferation
rates remain elevated at the end of 6 weeks of exposure which is
a departure from the results obtained in the rat. There was no
apparent effect of formaldehyde in the maxillary sinuses of the
monkey. A lower degree of cell proliferation and tissue damage
also occur in the trachea and bronchi in monkeys exposed to
formaldehyde at this same concentration (6 ppm).
\
In chronic studies, cellular effects such as rhinitis
(inflammation of the nasal' mucosa), epithelial dysplasia
(displacement of one cell type with another one), and squamous
metaplasia (replacement of normal mucosal cells with squamous
cells), developed in the nasal cavities of rats and monkeys after
exposures for 12 months and 26 weeks, respectively, to 2-3 ppm of
formaldehyde. After 24 months of exposure, the incidence of
squamous metaplasia in rats increased to nearly 100 percent. In
contrast to the rat, where formaldehyde-induced lesions were
confined to the anterior nasal passages, these lesions were more
126
-------
widespread in the nasal passages of the monkey. In both rats and
monkeys, a NOEL (no-observed-effect level) of 1.0 ppm for
squamous metaplasia was determined, with a LOEL (lowest observed
effect level) of 2.0 ppm in rats and 3.0 ppm in monkeys.
8.2. Carcinogenic Effects
EPA has classified formaldehyde as a "Probable Human
Carcinogen" (Group Bl) under its Guidelines for Carcinogen Risk
Assessment. This classification is based on:
o limited evidence of carcinogenicity in humans;
o sufficient evidence of carcinogenicity in animals; and
o additional supportive evidence (i.e., mutagenicity and
structure activity and mechanistic considerations)
8.2.1. Studies of Humans
In 1987 the EPA examined 28 epidemiologic studies relevant
to formaldehyde. Based upon a review of these studies, EPA
concluded that "limited" evidence existed that formaldehyde may
be a carcinogen in humans. Eleven additional studies were
reviewed in the present assessment. These newer studies support
the conclusions drawn in 1987 regarding "limited" evidence. The
evidence for potential human carcinogenicity associated with
formaldehyde exposure rests heavily on associations with cancers
of the nasal cavity and sinus and of the nasopharynx, and to a
small degree on observations of elevated risks between lung
cancer and combined formaldehyde and particulate exposures.
Epidemiologists use several criteria for judging, in a
collection of studies, whether exposure represents a carcinogenic
hazard to humans. These criteria include: 1) antecedent
exposure (disease occurs after exposure), 2) strength of the
association 3) coherence of the observations across different
studies and exposure assessments, 4) specificity of the
association (i.e., whether elevated relative risks are confined
to specific disease subcategories or exposure subgroups), and 5)
consistency with existing knowledge of the biological behavior of
the agent. Antecedent exposure is satisfied because all the
epidemiologic studies reviewed in the 1987 and in this document
examined health consequences of previous exposure.
With regard to the strength of the association, the
associations observed in the epidemiologic studies between
formaldehyde exposure and either cancers of the sinonasal cavity,
nasopharynx, or lung have not been large. The question of
whether bias may influence the magnitude of risk becomes /
important. Bias is not thought to artificially inflate the
observed relative risks. Several studies (Vaughan et al., 1986,
Roush et al., 1987) took into consideration such factors as age,
smoking, sex, or socioeconomic status. Statistically significant
elevations in risks between nasopharyngeal cancer and
127
-------
formaldehyde exposure were noted after taking into account some
of these possibly confounding factors. Three other studies that
adjusted for simultaneous wood dust exposure (Hayes et al., 1986;
Olsen and Asnaes, 1986; Olsen et al., 1984) observed apparently
elevated, but not statistically significant, risks between
sinonasal cavity cancer and formaldehyde exposure. Smoking is
not considered a factor in one other study (Blair et al., 1986,
1987) since examination of sites most likely associated with
smoking did not show*a consistent pattern of elevations.
In contrast, the excesses in leukemia and brain cancer
observed among populations that are occupationally exposed to
formalin are most likely related to exposure other than
formaldehyde. Elevations in these site-specific risks have not
been observed in studies of industrially-exposed populations.
Viruses and solvents were also potential exposures for the
formalin-exposed, and may be important with respect to the
development of leukemia and brain cancer.
The observation of similar results across different studies
or across different exposure assessments can support a causal
association between exposure and disease. Increased
nasopharyngeal cancer risks have been observed in four studies
(Vaughan et al., 1986a, 1986b; Blair et al., 1986, 1987; Roush et
al., 1987; and Malker et al., 1990) that were carried out under
differing study designs and exposure scenarios. Exposure-
response gradients were observed in two of these studies (Vaughan
et al., 1986a, 1986b); Blair et al., 1987), and suggested in
another (Roush et al., 1987). In addition, one reanalysis
(Collins et al., 1988), which adopted a different exposure
assessment appears to support formaldehyde and particulates
together as risk factors for nasopharyngeal cancer.
Two studies (Hayes et al., 1986; Olsen et al., 1984)
reviewed in the 1987 document reported an association between
sinonasal cavity cancer and formaldehyde and wood dust exposure.
Similar trends in risk were noted in analyses that controlled for
wood dust exposure, however, these risks were not statistically
significant. In addition, risk appeared to be specific for the
histologic type, sguamous cell carcinoma (Hayes et al., 1986;
Olsen and Asnaes, 1986). Two studies (Roush et al., 1987; Stern
et al., 1987) did not observed any associations between cancer of
the nasal cavity and sinuses and formaldehyde exposure.
Not all of the reviewed studies reported elevated cancer
risks due to sinonasal cavity or nasopharyngeal cancer.
Limitations in these remaining studies tended to prevent the
results from revealing differences from the referent population
(a conservative bias).
Elevations in lung cancer relative risks have been observed
in several studies (Blair et al., 1986; Sterling and Weinkam,"
1989; Acheson et al., 1984; Stayner et al., 1988; Bertazzi et al,
1986; Partanen et al, 1984; and Gerin et al, 1989). Only the
'*
128
-------
elevations in Blair et al., Acheson et al, and Bertazzi et al.
were statistically significant; exposure in these studies were to
formaldehyde and particulates. Because the elevations in lung
cancer risk have not been consistently observed across the
studies, these observations do not carry great weight.
A finding: of 'the highest risk among those with the longest
latency and the hicfHest occupation exposure suggests that an
occupational agent, here most likely formaldehyde, is reasonable
for the observed increased cancer risk. First, elevated risks
for nasopharyngeal, and nasal cavity and sinus cancers are
observed among various subgroups: those who are older, those
with the greatest latency since first exposure, those with the
highest duration or level of exposure, and those with
occupational particulate exposure. Second, relative risks for
cancer of the nasal cavity and sinuses associated with
formaldehyde exposure are specific for one histologic type. Two
studies (Hayes et al., 1986; Olsen et al., 1984) observed the
largest risks for squaraous cell carcinoma with formaldehyde
exposure. Wood dust does not enter into this observation since
wood dust exposure was associated only with adenocarcinoma and
not with squamous cell carcinoma.
The associations between formaldehyde and cancers of the
nasopharynx and sinonasal cavity are persuasive in light of a
biological interpretation of how formaldehyde may be acting.
Experimental data of Kerns et al. (1983) indicate that
formaldehyde reacts with biological material at the point of
contact. Human exposure to formaldehyde is predominately by
inhalation, thus, sites along the respiratory tract are expected
to be "at risk". The observation of dysplasia in nasal biopsies
among formaldehyde-exposed workers (Edling et al., 1988;
Holmstrom et al, 1989; Boysen., 1990) adds further biological
support for a possible association between formaldehyde and
cancer of the nasal cavity and sinuses.
The relevance of combining cancers of the nasal cavity with
those of the sinuses in studies of humans for hazard
identification has been questioned. Formaldehyde DNA-protein
adducts have been detected in the monkey in the upper respiratory
tract (nose, nasopharynx, larynx, trachea, and carina) and, to a
lesser degree, in major intrapulmonary airways greater than 2 mm
diameter (Heck et al., 1989). Experimental conditions in
monkeys, however, do not show evidence of formaldehyde DNA-
protein adducts (Heck et al., 1989) or of cellular proliferation
(Monticello and Morgan, 1989) in the maxillary sinuses.
Insufficient information is presented in the reviewed
studies to examine whether or not formaldehyde exposure is
associated with cancers of the nasal sinuses. The epidemiologic
studies can neither support nor reject this hypothesis. Any bias
which would be introduced by grouping cancer sites unrelated to
exposure with those that are would be similar to the bias
introduced by misclassification cf exposure; it would more likely
129
-------
influence the risk towards seeing no association between exposure
and disease. Several well-conducted case-control studies
examining nasal cavity and sinus cancer have detected elevated
risks in cancer at this aggregated site with formaldehyde
exposure, although it is recognized these elevations could always
reflect random events. Therefore, these studies are considered
pertinent to. hazard identification.
••, ',
In summary, the'*studies released since 1987 support the
conclusions drawn by EPA (1987) and do not alter the evaluation
that "limited" evidence exists for an association between
formaldehyde and human cancer according to the EPA's cancer
assessment guidelines (EPA, 1986). Although the common exposure
in the reviewed studies was formaldehyde, the epidemiologic
evidence does not conclusively demonstrate a causal relationship;
possible exposure to other agents may have confounded the
findings of excess site-specific cancers. In addition, excesses
in nasopharyngeal and nasal and sinus cavity cancers are based on
a small number of deaths. It is for this reason that the
evidence is called "limited" rather than "sufficient".
8.2.2. Studies in Animals
Based upon a review of studies conducted by the Chemical
Industry Institute of Toxicology (CUT) (Kerns et al., 1983) and
by Albert et al. (1982) and Tobe et al. (1985), EPA (1987)
concluded that there is "sufficient" evidence of
carcinogenicity of formaldehyde in animals. This finding is
based on the induction by formaldehyde of an increased incidence
of a rare type of malignant tumor (i.e., nasal squamous-cell
carcinoma) in both sexes of rats, in multiple inhalation
experiments, and in multiple species (i.e., rats and mice). In
these long-term laboratory studies, tumors were not observed
beyond the initial site of nasal contact nor have other mammalian
in vivo tests shown effects at distant sites. A subsequent
report by CUT (Monticello, 1990) lends further support for the
carcinogenicity of formaldehyde in rats.
EPA's Guidelines for Carcinogen Risk Assessment define
limited evidence of carcinogenicity in humans as indicating that
"...a causal interpretation is credible, but that alternative
explanations, such as chance, bias, or confounding, could not
adequately be excluded."
EPA's Guidelines for Carcinogen Risk Assessment define
sufficient evidence of carcinogenicity from studies in experimental
animals as indicating that "...there is an increased incidence of
malignant and benign tumors: (a) In multiple species or strains;
or (b) in multiple experiments, preferably with different routes of
administration or using different dose levels); or (c) to an
unusual degree with regard to incidence, site or type of tumor,
dose-response effects, as well as information from short-term tests
or on chemical structure."
130
-------
In contrast to the inhalation data, results obtained from
several carcinogenicity studies with formaldehyde given to rats
in drinking water provide only suggestive evidence for
carcinogenic potential via the oral route. The target tissue in
these studies was the forestomach, which is consistent with
observations from the inhalation studies in that tumors develop
at the site .ot 4nitial contact. In recent tumor promotion
studies, formaldehyde'enhanced the tumor response in mouse skin,
rat trachea, and rat stomach (indicating that formaldehyde has
tumor promotion potential at least in some tissues) but not in
hamster respiratory tract— consistent with the observed lack of
carcinogenic response to formaldehyde in hamsters reported by
Dalbey (1982) and discussed in EPA (1987).
, 8.2.3. Additional Supportive Evidence
Tests for point mutations, numerical and structural
chromosome aberrations, DMA damage/repair, and in vitro cell
transformation provide evidence for the potential mechanisms of
carcinogenicity. A battery of tests which measure different
endpoints helps to characterize the chemical's response spectrum.
In general, the wider the range and the greater the intensity of
response of a substance in applicable short-term tests, the more
likely it is that the substance may cause cancer.
Formaldehyde is mutagenic in numerous bacterial test systems
and test systems using fungi and insects (Drosophila). It also
transforms cells in culture and causes DNA cross-linking, sister
chromatid exchanges (SCE) and chromosome aberrations. In
addition, formaldehyde has been shown to bind with DNA and with
proteins in both in vivo and in vitro test systems. Its ability
to interfere with DNA repair in human cells has also been shown.
Mutagenicity data obtained since the 1987 assessment confirm the
original conclusions.
8.3. Quantitative Risk Assessment
The risk assessment has identified two biological effects
for which the data are sufficient for quantitative examination.
These are sensory irritation and cellular effects of the upper
respiratory tract and cancer. Results obtained from studies in
animals and humans were used to assess the sensory irritation and
cellular effects. Cancer risk estimates were derived by modeling
data obtained from studies in animals.
This assessment is focused on the population constituted by
persons who reside in mobile and conventional homes constructed
using "significant amounts" of urea-formaldehyde (UF) pressed-
wood (i.e, homes in which UF pressed wood is used for floor
underlayment and, in some cases, for wall paneling). The
exposure scenarios on which the quantitation is based are those
described in the 1987 risk assessment document.
131
-------
8.3.1. Noncancer Risk Assessment
Table 8-1 illustrates the relationship between the dose
levels associated with sensory irritation and cellular effects in
the nasal cavity and the exposure levels for a number of
population groups. In keeping with Agency practice, instead of
using high-to-low-dose extrapolation models, the degree of
concern from" these effects is approximated by comparing existing
exposures to rto-obse>rved-effeet-level (NOEL) or lowest-observed-
effect-level (LOEL)'1 values. Generally, little risk is expected
in cases where exposures are ten-fold less than a NOEL, or one
hundred-fold less than a LOEL when based on human observations.
An additional uncertainty factor of ten is applied to account for
interspecies differences in response if animal-derived NOEL and
LOEL values are used.
The results shown in Table 8-1 illustrate the spectrum of
responses that humans and animals exhibit on exposure to
formaldehyde. Human mucociliary function may be inhibited at
formaldehyde concentrations of 0.3 ppm and above; comparable
effects are reported in animals, with a NOEL of 0.5 ppm in rats
(Table 8-1). The NOEL for nasal squamous metaplasia in rats and
monkeys is 1.0 ppm for both species, with LOELs of 2.0 ppm (rats)
and 3.0 ppra (monkeys). Similarly, studies of humans showed nasal
cavity effects in some persons exposed in the range of 0.08-0.9
ppm (Edling, et al., 1985, 1988; Holmstrom et al., 1989) with
reported peak exposures of >4 ppm. The lack of adequate exposure
data in the epidemiologic studies as well as confounding exposure
to wood dust and resin particles do not allow a conclusive
determination of the exposure level(s) eliciting human response
relative to animal observations.
From the above, it appears that humans and animals may
respond similarly to the cellular effects of formaldehyde in the
nose. Formaldehyde exposures in mobile and conventional homes
(0.1-0.2 ppm with peak exposures up to 0.4 ppm) fall below the
NOELs and LOELs for cellular effects determined from studies in
animals (Table 8-1). Since the anticipated exposures in the
identified populations are close— within a factor of ten— to
those associated with effects in humans occupationally exposed
and animals, home residents are likely to be be at some risk of
experiencing these cellular effects.
From the data in Table 8-1, it is apparent that there is no
margin between existing exposures and levels of formaldehyde that
are associated with irritation of the eye and upper airway in
some humans. Since the available data do not allow the
development of a well-defined dose-response relationship for
these irritation effects, only qualitative assertions may be made
with regard to population response. The people at greatest risk
of experiencing discomfort due to formaldehyde-induced irritation
would be homeowners during the first year of occupancy,
particularly under conditions of high temperature and humidity
which favor elevated levels of formaldehyde in these homes.
A.
132
-------
TABLE 8-1
HEALTH EFFECTS AND ASSOCIATED FORMALDEHYDE
EXPOSURE CONCENTRATIONS
Formaldehyde"
Concentration
(ppm)
<0.05
0. 1
0.3
>1.0
2.0
3 .0
5.0
15.0
Health Effects
0*
Eye irritation observed in
some people
Upper airway irritation; most
people experience eye
irritation; human mucociliary
inhibition
Lower airway irritation
reported; nasal squamous
metaplasia NOEL (rat and
monkey)
Rat squamous metaplasia and
mucociliary system LOELC
Human (most) experience nose
and throat irritation
Rat observed 1% cancer
incidence
Rat observed 50% cancer
incidence;
Mouse observed 1% cancer
incidence
Exposure Comments
Indoor ambient
background
New mobile homes
10-year average
concentration
8
HUD standard =
0 . 4 ppm
Current OSHA PEL=
1.0 ppm (8 hr TWA)
Highest recorded
level in homes
House and Urban Development.
NOEL= no observed effect level.
^LOEL= lowest observed effect level.
urea-formaldehyde (UFFI) foam insulated home.
133
-------
Although quantitative estimates of risk are not possible,
the frequency and severity of response are concentration related.
Fewer responses are expected to be associated with less frequent
and less intense exposure. In addition, both sensory irritation
and cellular effects are expected to be reversible once
formaldehyde exposure is reduced.
8.3.2. Cancer,,;Risk Assessment
In principle, data from studies of humans are preferred for
derivation of numerical risk estimates. However, the available
epidemiologic data on formaldehyde were not suitable for low-dose
quantitative cancer risk estimation, mainly because of a lack of
adequate exposure information in the studies. Accordingly,
results from studies in animals have been used to estimate low-
dose human cancer risk.
8.3.2.1. Dose-Response Assessment
As detailed in EPA (1987), of the carcinogenicity studies
with formaldehyde in animals, EPA selected the CUT study in rats
(Kern et al., 1983) as the best study for cancer risk
extrapolation. This study was well designed, well conducted,
included multiple doses, and used a large number of animals per
dose.
Data relevant to selecting a model for extrapolation of
cancer risk associated with exposure to formaldehyde were
reviewed; some of the biological information support a direct
relationship between exposure and carcinogenicity while other
data are consistent with a nonlinear response. The Agency,
however, concludes that insufficient information is available at
present to propose an extrapolation model for formaldehyde
different from the one recommended by EPA's Guidelines for
Carcinogen Risk Assessment (i.e., linearized multistage
procedure). The Agency presented various other models in its
1987 assessment for comparative purposes.
Biologic evidence on mechanism of action, which can aid in
model selection, largely is inferred from a variety of types of
studies. These are limited and suggestive of several mechanisms
for formaldehyde. Thus, the ability of formaldehyde to cause
point mutations, chromosome aberrations and DNA damage is
consistent with the chemical's ability to initiate the
carcinogenic reaction. Mutagenicity studies suggest a direct
relationship (i.e., a linear one) between exposure to
formaldehyde and carcinogenicity. The steep curvilinearity of
the rat nasal carcinoma dose-response data in the CUT study
(Kerns et al., 1983) suggests, however, that cancer development
is limited at low dose levels and greatly accentuated above
certain concentrations.
134
-------
The CUT also conducted molecular dosimetry experiments
attempting to relate ambient exposures to formaldehyde with
tissue-specific levels of formaldehyde-DNA adducts as DNA-protein
cross-links (DPX). In its previous analysis, EPA (1987) used the
administered dose to calculate carcinogenic risk from
formaldehyde" exposure because of perceived deficiencies in
experimental design In the DPX dosimetry approach. However,
recent evidence with an improved technique indicates that the use
of DNA-protein cross-links as a measure of intracellular dose may
provide a better indicator of target tissue exposure than would
airborne formaldehyde levels. The use of an intracellular
dosimeter (DPX) in the derivation of risk estimates would reflect
the impact of both mucociliary clearance and metabolism, whose
combined influence effectively reduces the amount of formaldehyde
available to the nasal epithelial cells. Particularly at low
dose levels, much of the inhaled formaldehyde does not reach
nasal tissues, as it is rapidly cleared by the mucociliary system
and oxidative metabolism. At sufficiently high exposure
concentrations the detoxification mechanisms become overwhelmed,
making a greater amount of formaldehyde available for
interactions with DNA and other cellular macromolecules. DPX
formation increases nonlinearly with increasing airborne exposure
concentrations in a manner similar to the observed carcinogenic
response in the rat cancer bioassay. These findings suggest
possible saturation of detoxification processes at high
formaldehyde concentration. The conclusion derived from the
above discussion is that the use of airborne concentration in the
derivation of risk estimates would, by ignoring important
contributions from detoxifying processess, lead to an
overestimate of exposure and concomitant risk.
While formation of DPX may lead to a number of genotoxic
effects, its role—if any— in the induction of nasal cancer is
not completely understood. The Agency recognizes that the
available DNA-binding data are derived from acute or subacute
exposure. The latter may either underestimate or overestimate
binding levels under chronic exposure conditions where cell
proliferation may be significant, particularly at high exposure
concentrations. The argument for the use of DPX as a surrogate
dose is that this measure provides an index of the area under the
curve of a reactive formaldehyde species in the target cells,
both in rats and other species. This argument applies whether
DPX are mechanistically involved in the carcinogenic process or
are simply an indicator of intracellular exposure.
Although toxicological studies on formaldehyde have shown
that adverse effects are a function of exposure, there is
increasing evidence that the exposure parameter that more closely
associates with observed toxicities is airborne formaldehyde
concentration rather than total—cumulative—• daily dose (the
latter is used in the 1987 quantitative estimates of cancer
risk). Studies in rats by various authors have measured cell
135
-------
turnover using constant cumulative daily dose (ppm-hr) but
varying the concentration of formaldehyde and duration of
exposure. Cell turnover rates are found to increase
significantly with increasing concentrations at nasal regions
where neoplasms develop. In contrast, in certain nasal sites,
cell proliferation appears to be a function of total daily dose.
The combined results of several toxicologic studies further
support the importance of airborne formaldehyde concentration in
the induction of toxic effects. Inhalation exposure to a low
level of formaldehyde for a long duration of daily exposure (1
ppm for 22 hours) over the course of 6 months does not cause
lesions in the rat nose. In contrast, subchronic or chronic
inhalation exposure to higher formaldehyde concentrations (2-4
ppm) but with a shorter duration of daily exposure (6 hours)
produces varying degree of nasal damage in rats. Similar
observations are found in a recent subchronic study showing that
cell proliferation and nasal damage are only seen with the animal
group treated intermittently to high formaldehyde concentration
(intermittent exposure for a total of 4 hours at 4 ppm). No
responses are found in animals receiving the same total daily
dose at a lower concentration (2 ppm for 8 hours). All these
findings emphasize that airborne formaldehyde concentration is an
important exposure parameter and imply that the utilization of
lifetime average daily concentrations for risk quantification
purposes may overestimate risk potential by arbitrarily lowering
the exposure concentration at which adverse effects are expected.
In sum, a number of modifications, relative to the 1987
assessment, have been incorporated in the current quantitative
risk assessment. These include the following: (1) the use of
GLOBAL86 (as opposed to GLOBAL83) which contains EPA's current
interpretation of the linearized multistage procedure; (2) the
omission of lifetime average daily exposure adjustments to arrive
at daily levels prior to dose-response modelling; this lifetime
adjustment was used in the 1987 risk assessment; and (3) the use
of an intracellular dosimeter- DPX binding data, instead of
airborne concentrations of formaldehyde.
8.3.2.2 Cell Proliferation and Carcinogenesis
As discussed in section 4.5.2, the results of chronic
studies point to cell proliferation is a contributor to the
observed nonlinearity in nasal cancer induction; thus, the
postulated cancer risk for low exposures to formaldehyde may have
been overestimated by not incorporating cell proliferation in the
derivation of such estimates. However, a number of data
important for factoring cell proliferation into quantitative
cancer risk assessment are still lacking; the potential magnitude
of the cell proliferation effect on formaldehyde carcinogenesis
cannot be determined at this time. For example, a clear
relationship between formaldehyde-induced regional cell
proliferation and tumor formation has not been established. The
136
-------
causes for the discrepancy in cytotoxicity/cell proliferation and
carcinogenic responses at the medial aspect of the
maxilloturbinate need to be explored. As the nasal respiratory
epithelium comprises numerous cell types, the cell type of origin
of the squamous cell carcinomas and cell proliferation remains to
be determinadv Such information is important for the
quantitative assessment of dose to critical target cell
populations. Furthermore, the relative roles of
cytotoxicity/cell proliferation and mutation in the multistep
process of formaldehyde carcinogenesis is still unknown. Whether
formaldehyde affects mutation rate and increases initiated cells
or promotes spontaneous mutation without involving direct DNA
reactivity (or both) has not been studied in the rat nasal
cavity. Quantitative data on mutation rate per unit of DPX of
the nasal cavity at different stages of the carcinogenic process
and at different concentrations of formaldehyde are needed. In
addition, information on the effects of formaldehyde exposure on
cell cycle length (kinetics of cell proliferation) in control,
non-neoplastic, pre-neoplastic and neoplastic tissues of the
nasal cavity is also needed for meaningful modelling of cancer
risk. There are ongoing studies at CUT to generate the
necessary data for the development of biologically-based risk
assessment models for formaldehyde. This biologically-based risk
assessment approach should improve the human risk assessment of
formaldehyde.
8.3.2.3 Discussion and Conclusions
Formaldehyde is a component of normal metabolism in the
biosynthesis of cellular molecules including amino acids, lipids,
and nucleotides. Because of its high chemical reactivity and
rapid rate of metabolism, inhaled formaldehyde is essentially not
transported into the body from the site of contact, at least at
environmental concentrations. Enzyme participation in the
biological disposition of formaldehyde is only evident in a
detoxication step, the GSH-dependent formaldehyde dehydrogenase.
Toxic effects occurring upon exposure to formaldehyde appear
to be attributable primarily to the interactions of the chemical
molecule with contact tissue. The absence of enzyme
participation in transformations that may potentially lead to
toxic events (DNA binding, tissue irritation, etc.) introduces an
element of predictability in the biological behavior that this
chemical may exhibit in different species. There is site
concordance among mice, rats and monkeys in the lesions inflicted
by formaldehyde under inhalation exposure conditions; there is
similarity between rats and monkeys in the shape of the dose
response curve for both cellular effects and rate of DPX
formation. Comparable human data are more limited but do provide
indications that the behavior of formaldehyde is consistent with
the chemical characteristics elucidated in animal experiments.
Interactions of formaldehyde with cells result in cellular
137
-------
irritation and destruction, DNA damage and possible mutations.
These mutations along with cellular proliferation could lead to
the development of cancer, an endpoint that is observed in test
animals and for which there is limited evidence in humans.
The mechanisms of formaldehyde induced-carcinogenesis are
not completely understood; there is, however, knowledge about
some factors.which"correlate with tumor response and which have
plausible roles in cancer development. There is ample evidence
that formaldehyde is a mutagen, which is one possible component
to the carcinogenicity of formaldehyde. The focus of recent
studies on formaldehyde has been on the effects of concentration,
total daily dose, and the length of exposure period on various
biological endpoints— mucociliary clearance, cell proliferation,
cytotoxicity, DNA-protein binding, and pathology— that may help
to explain the pronounced nonlinear carcinogenic response
observed in the cancer studies in rats.
As stated earlier, the epidemiologic data base on
formaldehyde does not provide information sufficient to calculate
cancer dose-response. An attempt was made to project limits
based on human data that could put risk estimates derived from
animal data into perspective, however. The study by Blair et al.
(1986) was used to gauge the behavior of the current risk
estimates while recognizing that the exposure data are inadequate
as a basis for unit risk estimation. The excess deaths
attributed to nasopharyngeal cancer are approximately one for
each exposure category (range 0.05- to 89-ppm-year) defined by
Blair et al. The corresponding upper bound expected excess
lifetime cancer deaths obtained by using the 1991 animal-based
unit risk, derived from either rat or monkey dosimetry, are in
the range of 10 to 10" . A number of issues, including the use
of cumulative exposure categories, confounding exposure to other
carcinogenic agents, and anatomical and physiological differences
among species, modulate the magnitude of this difference. The
concentration(s) at which these toxic effects become evident
cannot be firmly established at this time, although it can be
offered that based on the chemical characteristics of
formaldehyde and dose response data in animals the dose response
curve for humans is expected to be similar in shape, i.e.,
relatively shallow at low concentrations followed by a steep
increase after an undetermined concentration. The information
currently available do not allow a reasonable estimate of the
latter. As mentioned earlier, the EPA recognizes the need to
develop quantitative risk assessment procedures which further
attempt to incorporate biologic data, particularly with regard to
pharmacokinetics and mechanism, that can yield estimates of risk
that appear to conform better to all facets of the information in
hand.
In this document, human risk estimates, as stated earlier,
are based on animal data. Dosimetry data obtained in rats and
monkeys were used as alternative means to adjusting exposure
138
-------
concentrations for human subjects to better reflect exposure of
the target cells in the nasal epithelium. The Agency supports
the use of monkeys as a more suitable animal model than rats for
human extrapolation since monkeys have similar anatomic and
morphologic structures and breathing pattern to humans (see
Appendix C) ._ Anatomical and physiological similarities that
support the us"e of">the monkey as a model for dosimetry estimates
for humans include the oronasal mode of breathing; comparable
relative nasal surface area; mucociliary clearance routes; and
inspiratory airflow routes (see Appendix C for a complete
description). The rat is an obligate nose breather, exhibits a
greater relative nasal surface area than humans (or monkeys),
shows differences in mucociliary clearance routes, and a
different proportion distribution in nasal surface area covered
by different epithelia. Anatomical and physiological differences
are expected to influence both the rate and pattern of
distribution of inhaled formaldehyde. This observation is of
particular relevance when applied to formaldehyde which, as
described earlier, is expected to react chemically at the site of
contact. Thus, the breathing mode and large relative nasal
surface area in the rat would be expected to lead to a high
accumulation of formaldehyde in this area, which is consistent
with the experimental findings. The oronasal pattern of
respiration in humans and monkeys would be expected to reduce the
dose of formaldehyde received by the nose and the nasopharynx,
and to increase the dose delivered to the oral cavity and upper
respiratory tract. This is supported by the observation of a
lower DPX formation but more widespread distribution in the
respiratory tract of monkeys.
The risk estimates for the linearized multistage procedure
upper bound (UB) at various exposure levels are presented in
Table 8-2. These risk estimates are derived utilizing the
monkey-dosimetry unit risk. Risks at any exposure level may
range from the upper bound to zero. An established procedure
does not yet exist for making "most likely" or "best" estimates
of risk within the range of uncertainty defined by the upper
bound and zero. The upper bound estimate for excess lifetime
risk of developing cancer is 5 x 10" [Group Bl] for residents of
mobile homes who are exposed for 10 years to an average level of
0.20 ppm; and 2 x 10 [Group Bl] for residents of some
conventional homes who are exposed for 10 years to an average
level of 0.07 ppm. For residents of older, conventional homes
exposed for 10 years to an average level of 0.05 ppm (upper limit
indoor level) the projected cancer risk is 1 x 10 . The
formaldehyde levels in the first two settings are expected to
decline and eventually approach background levels (< 0.05 ppm).
139
-------
TABLE 8-2
SUMMARY OF CANCER RISK ESTIMATES ASSOCIATED WITH FORMALDEHYDE
EXPOSURE UNDER SELECTED INDOOR ENVIRONMENTS
Population Segment,
(Exposure Level)
Projected Upper Bound Lifetime
Excess Individual Risk*
Mobile Home Residents
(0.20 ppm 10-yr average)
5x10
Conventional Home Residents
(0.07 ppm lO-year average)
2 x 10
Older Homes/Indoor Ambient
Background Upper Limit (0.05 ppm
10 year)
1 x 10
a!991 Unit Risk based on monkey dosimetry: 3.3 x 10" /ppm.
Exposures were assumed to cover 14 hr/day and were adjusted to
70-year lifetime.
bFor homes containing substantial amounts of urea-formaldehyde
pressed wood (e.g., floor underlayment and/or paneling)
140
-------
The human carcinogenic risks estimates based on the monkey
DPX data are approximately ten-fold lower than corresponding
risks based on rat DPX data at a given formaldehyde exposure
concentration. The analysis based on rat DPX data can be
interpreted as forgoing the use of the DPX dosimeter for cross-
species extrapolation on the grounds that rat-human differences
may be poorly, illuminated by reference to rat-monkey differences.
That is, the unit risk based on rat DPX reflects a correction
only for high-to-low dose differences in tissue exposure that
results from the apparent saturation of detoxification processes
at the high exposure concentrations used in the bioassay.
The use of the monkey DPX data in quantitative risk
estimation aims at taking into account not only high-to-low
nonlinearity in tissue dose levels, but also dosimetry
differences between rodents and primates. Incorporating these ,
dosimetry differences, however, leaves open the question of
relative sensitivity of rodents and primates to the toxic effects
of formaldehyde. Tn the absence of any carcinogenic data in
monkeys, it is not known how susceptible monkeys are to
formaldehyde-induced carcinogenesis. For noncancer effects, the
monkeys appear to be more susceptible to formaldehyde toxicity
than rats, as shown by an induction of more widespread histologic
lesions in the respiratory tract at equivalent exposure
concentrations (6 ppm), however, the relevance of this
observation to cancer development is uncertain since not all such
lesions develop into tumors. In addition, it is also true that a
gradient of effects is observed with decreasing exposure
concentrations where at 3 ppm cellular changes are confined to
the nasal cavity, and are not observed at concentrations of 1 and
0.2 ppm. This gradient of effects has been reported for both
rats and monkeys, and a similar profile is anticipated in humans.
Cross-sectional studies of workers in formaldehyde-related
industries show higher histological scores (more metaplasia and
dysplasia) ^with longer exposure; unfortunately, only nasal
turbinates could be examined. It is difficult to evaluate the
full'impact of these observations because as stated before, no
consistent correlation has yet evolved between nonneoplastic
changes and tumor induction. DNA binding also occurs in the
lower respiratory tract of monkeys at 6 ppm but it is not
measurable at lower exposure concentrations with exception of the
nasal area.
The forgoing argument is based on estimates of risk to the
nasal epithelium only. Monkeys, as stated above, show DPX
formation deeper in the respiratory tract in regions for which
rats show neither DPX nor tumor response. Whether these further
DPX engender additional risk, and whether humans are subject to
such risk from formaldehyde .inhalation is not considered in the
present analysis. It is possible that basing risk estimates only
on cross-links data corresponding to those in rats for the nose
may lead to an underestimate of risk in humans. Epidemiologic
information provides limited evidence for an association between
141
-------
formaldehyde exposure and cancer of the nasal cavity and
nasopharynx but a weaker case is available for the lung. Under
the anticipated human exposure scenarios in the current
assessment (airborne concentrations of about 0.3 ppm and lower)
the cumulative evidence indicates that areas of the respiratory'
tract that come in most immediate contact with inhaled
formaldehyde are the most likely targets for formaldehyde-induced
toxicity. In*conclusion, EPA believes that the predicted cancer
risk estimates depicted in this document provide a plausible
description of the actual risks for low level exposures to
formaldehyde, as encountered in new mobile and conventional
homes.
142
-------
9.0 REFERENCES
Acheson ED, Gardner MJ, Pannett B, Barnes HR, Osmond C, Taylor
CP. 1984. Formaldehyde in the British Chemical Industry (an
occupational cohort study). The Lancet 1: 611-616.
Adams DO, Hamilton -TA, Lauer LD, Dean JH. 1987. The effect of
formaldehyde exposure* upon the mononuclear phagocyte system of
mice. Toxicol. Appl. Pharm. 88:165-174.
Albert RE, Sellakmur 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.
Alexandersson R, Hedenstierna G. 1989. Pulmonary function in wood
workers exposed to formaldehyde: A prospective study. Arch.
Environ. Health 44:5-11.
Alexandersson R, Kolmodin-Hedman B, Hedenstierna G. 1982.
Exposure to formaldehyde: Effects on pulmonary function. Arch.
Environ. Health 37:279-284.
Anderson I, Molhave L. 1984. Controlled Human Studies with
Formaldehyde. In: Gibson, J.E. (ed.), "Formaldehyde Toxicity",
Hemisphere Publishing Corp., New York. pp. 154-165.
Bedford P, Fox BW. 1981. The role of formaldehyde in methylene
dimethanesulphonate-induced DNA cross-links and its relevance to
cytotoxicity. Chem. Biol. Interact. 38:119-126.
Beland FA, Fullerton NF, Heflich RA. 1984. Rapid isolation,
hydrolysis, and chromatography of formaldehyde-modified DNA. J.
Chromatogr. 308:121-131.
Bender JR, Mullin LS, Graepel GJ, Wilson WE. 1983. Eye irritation
response of humans to formaldehyde. Am. Ind. Hyg. Assoc. J.
44:463-465.
Bertazzi PA, Zochetti C, Pesatori A, Radice L, Vai T. 1986.
Exposure to formaldehyde and cancer mortality in a cohort of
workers producing resins. Scand.. J. Work Environ. Health 12:461-
468.
Blair A, Stewart PA, Hoover, RN 1990a. Mortality from lung
cancer among workers employed in formaldehyde industries. 'Am. J.
Ind. Med. 17:683-699.
143
-------
Blair A, Saracci R, Stewart PA, Hayes RB, Shy C. I990b.
Epidemiologic evidence on the relationship between formaldehyde
exposure and cancer. Scand. J. Work Environ. Health 16:381-393.
Blair A, Stewart PA 1989. Comments on the reanalysis of the
National Cancer Institute study of workers exposed to
formaldehyde. J. Occup. Med. 31:881.
»*
Blair A, Stewart PA, Hoover RN, Fraumeni JF Jr., Walrath J,
O'Berg M, Gaffey W. 1987. Cancers of the nasopharynx and
oropharynx and formaldehyde exposure. J.Natl. Cancer Inst.
78:191-193.
Blair A, Stewart P, O'Berg M, Gaffey W, Walrath J, Ward J, Bales
R, Kaplan S, Cubit D. 1986. Mortality among industrial workers
exposed to formaldehyde. J. Natl. Cancer Inst. 76:1071-1084.
Bogdanffy MS, Morgan PH, Starr TB, Morgan ET. 1987. Binding of
formaldehyde to human and rat nasal mucus and bovine serum
albumin. Toxicol. Lett. 38:145-154.
Boysen M, Zadig E, Digernes V, Abeler V, Reith A. 1990. Nasal
mucosa in workers exposed to formaldehyde: a pilot study. Br. J.
Ind. Med. 47:116-121.
Breslow NE, Day NE. 1980. Statistical Methods in Cancer Research.
Vol. 1. The analysis in case-control studies. IARC Scientific
Publication No. 32.
Breslow NE, Day NE. 1987. Statistical Methods in Cancer Research.
Vol. 2. The design and analysis of cohort studies. IARC
Scientific Publication No. 82, p.96-98.
Broder I, Corey P, Cole P, Lipa M, Mintz S, Nethercott JR. 1988a.
Comparison of health of occupants among control homes and homes
insulated with urea formaldehyde foam. I. Methodology. Environ.
Res. 45:141-155.
Broder I, Corey P, Cole P, Lipa M, Mintz S, Nethercott JR. 1988b.
Comparison of health of occupants and characteristics 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.
Broder I, Corey P, Brasher P, Lipa M, Cole P. 1988c. 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.
Brooks SM, Weiss MA, Bernstein IL. 1985. Reactive airways
dysfunction syndrome. J. Occup. Med. 27:473-476.
•s
144
-------
Casanova M, Deyo DF, Heck Hd'A. 1989. Covalent binding of inhaled
formaldehyde to DNA in the nasal mucosa of Fisher 344 rats:
Analysis of formaldehyde and DNA by high-performance liquid
chromatography and provisional pharmacokinetic interpretation.
Fundam. Appl. Toxicol. 12:397-417,
Casanova M, Heck H6MA. 1987. Further studies pf the ^etabolic
incorporation and coH^alent binding of inhaled H and C
formaldehyde in Fischer-344 rats: Effects of glutathione
depletion. Toxicol. Appl. Pharmacol. 89:105-121.
Casanova-Schmitz M, Heck Hd'A. 1983. Effects of formaldehyde
exposure on the extractability of DNA from proteins in the rat
nasal mucosa. Toxicol. A.ppl. Pharmacol. 70:121-132.
Casanova-Schmitz M, Starr TB, Heck Hd'A. 1984. Differentiation
between metabolic incorporation and covalent binding in the
labeling of macromplecules in the rat nasal mucosa and bone
marrow by inhaled C and H formaldehyde. Toxicol. Appl.
Pharmacol. 76:26-44.
Chang JCF, Gross EA, Swenberg JA, Barrow CS. 1983. Nasal cavity
deposition, histopathology, and cell proliferation after single
or repeated formaldehyde exposure in B6C3F1 mice and F-344 rats.
Toxicol. Appl. Pharmacol. 68: 161-176.
Chaw YFM, Crane LE, Lange P, Shapiro R. 1980. Isolation and
identification of cross-links from formaldehyde treated nucleic
acids. Biochemistry 19:5525.
Cohn MS, DiCarlo FJ, Turturro A, Ulsamer AG. 1985. Letter to the
editor. Toxicol. Appl. Pharmacol. 77:363-364.
Collins JJ, Caporossi JC, Utidjian HMD. 1988. Formaldehyde
exposure and nasopharyngeal cancer: re-examination of the
National Cancer Institute study and an update of one plant. J.
Natl. Cancer Inst. 80:376-377.
Consensus Workshop on Formaldehyde. 1984. Report on the Consensus
Workshop on Formaldehyde, Environ. Health Persp. 58: 323-381.
Cosma GN, Marchok AC. 1987. The induction of growth altered cell
population (tumor-initiation sites) in rat tracheal implants
exposed to benzo[a]pyrene and formaldehyde. Carcinogenesis 8:
1951-1953.
Craft TR, Bermudez E, Skopek TR. 1987. Formaldehyde mutagenesis
and formation of DNA-protein crosslinks in human lymphoblasts in
vitro. Mutat. Res. 176:147-155.
145
-------
Crosby RM, Richardson KK, Craft TR, Benforado KB, Liber HL,
Skopek TR. 1988. Molecular analysis of formaldehyde-induced
mutations in human lymphoblasts and E. coli. Environ. Mol.
Mutagen. 12:155-166.
Dalbey WE. t&8-2. 'Formaldehyde and Tumors in Hamster Respiratory
Tract. Toxicology '2*4: 9-14.
Dowd MA, Gaulden ME, Proctor BL, Seiber GB. 1986. Formaldehyde-
induced acentric chromosome fragments and chromosome stickiness
in Chortophaga neuroblasts. Environ. Mutagen. 8:401-411.
Dresp J, Bauchinger M. 1988. Direct analysis of the clastogenic
effect of formaldehyde in unstimulated human lymphocytes by means
of the premature chromosome condensation technique. Mutat. Res.
204:349-352.
Ecken JCJ, Sobels FH. 1986. The effect of X-irradiation and
formaldehyde treatment of spermatogonia on the reversion of an
unstable, P-element insertion mutation in Drosophila
melanogaster. Mutat. Res. 1765:61-65.
Edling C, Hellguist H, Odkvist L. 1988. Occupational exposure to
formaldehyde and histopathological changes in the nasal mucosa.
Br. J. Ind. Med. 45:761-765.
EPA. 1986. US Environmental Protection Agency Guidelines for
Carcinogen Risk Assessment. 51 FR 33992.
EPA. 1987. Assessment of Health Risks to Garment Workers and
Certain Home Residents from Exposure to Formaldehyde. Office of
Pesticides and Toxic Substances, U.S. Environmental Protection
Agency.
Feron VJ, Bruyntjes JP, Woutersen PA, Immell HR, Appelman, LM.
1988. Nasal tumors in rats after short-term exposure to a
cytotoxic concentration of formaldehyde. Cancer Lett. 39: 101-
111.
Gerin M, Slemiatycki J, 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.
Green DJ, Sauder LR, Kulle TJ, Bascom R. 1987. Acute response to
3.0 ppm formaldehyde in exercising healthy nonsmokers and
asthmatics. Am. Rev. Respir. Dis. 135:1261-1266.
Hanrahan LP, Dally KA, Anderson HA, Kanarek MS, Rankin J. 1984.
Formaldehyde vapor in mobile homes: A cross sectional survey of
concentrations and irritant effects. Am. J. Public Health
74 : 1026-1027.
146
-------
Hardell L, Johansson B, Axelson O. 1982. Epidemiological study of
nasal and nasopharyngeal cancer and their relation to phenoxy
acid or chloropheno exposure. Am. J. Ind. Med. 3:247-257.
Harrington J, Oakes D. 1984. Mortality study of British
pathologists. ~Br. J. Ind. Med. 41:188-191.
Harrington J/'Shahnpn H. 1975. Mortality study of pathologists
and medical laboratory technicians. Br. Med. J. 4:329-332.
Hattis D. 1990. Phannacokinetic principles for dose-rate
extrapolation of carcinogenic risk from genetically active
agents. Risk Anal. 10:303-316.
Hayes RB, Raatgever JW, Gerin M. 1986. Cancer of the nasal cavity
and paranasal sinuses, and formaldehyde exposure. Int. J. Cancer
37: 487-492.
Hayes RB, Blair A, Stewart PA, Herrick RF, Mahar H. 1990.
Mortality of US embalmers and funeral directors. Am. J. Ind. Med.
.18:641-652.
Heck Hd'A, Casanova M. 1987. Isotope effects and their
implications for the covalent binding of inhaled H and UC
formaldehyde in the rat nasal mucosa. Toxicol. Appl. Pharmacol.
89:122-134.
Heck Hd'A, Casanova M, Starr TB. 1990. Formaldehyde toxicity-new
understanding. CRC Crit. Rev. Toxicol. 20:397-426.
Heck Hd'A, Cassanova M, Steinhagen WH, Everitt JI, Morgan ET,
Popp JA. 1989. Formaldehyde toxicity. DNA-protein cross-linking
studies in rats and nonhuman primates. In: Feron VJ and Bosland
MC (eds.), "Nasal Carcinogenesis in Rodents: Relevance to Human
Health Risk", Pudoc Wageningen, The Netherlands, p.159-164.
Holmstrom M, Wilhelmsson B, Hellquist H, Rosen G. 1989.
Histological changes of the nasal mucosa in persons
occupationally exposed to formaldehyde alone and in combination
with wood dust. Acta Otolaryngol. 107:120-129.
Holness DL, Nethercott JR. 1989. Health status of funeral service
workers exposed to formaldehyde. Arch. Environ. Health 44:222-
228.
Horton AW, Tye 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: 31-43.
Horvath EP, Anderson H, Pierce WE, Hanrahan L, Wendlick, JD.
1988. Effects of formaldehyde on the mucous membranes and lungs.
J. Am. Med. Assoc. 259:701-707.
147
-------
IARC. 1987. Monographs on the Evaluation of Carcinogenic Risks to
Humans. Suppl. 7:211-216.
Imbus HR, Tochilin SJ. 1988. Acute effect upon pulmonary function
of low level- *xpos,ure to phenol-formaldehyde resin coated wood.
Am. Ind. Hyg. Assoo*,* J. 49:434-437.
IRMC (Interagency Risk Management Council), 1984. Draft Report
of the Risk Assessment Subgroup of the IRMC-Formaldehyde
Workgroup.
Iversen OH. 1986. Formaldehyde and skin carcinogenesis.
Environ. Int. 12:541-544.
Kerns WD, Pavkov KL, Donofrio DJ, Gralla EJ, Swenberg JA. 1983.
Carcinogenicity of formaldehyde in rats and mice after long-term
inhalation exposure. Cancer Res. 43:4382-4392.
Kilburn KH, Warshaw R, Thornton JC. 1987. Formaldehyde impairs
memory, equilibrium, and dexterity in histology technicians:
Effects which persist for days after exposure. Arch. Environ.
Health 42:117-120.
Kilburn KH, Warshaw R, Thornton JC. 1989. Pulmonary function in
histology technicians compared with women from Michigan: Effects
of chronic low dose formaldehyde on a national sample of women.
Br. J. Ind. Med. 46:468-472.
Klein-Szanto AJP, Ura H, Resau J. 1989. Formaldehyde induced
lesions of xenotransplanted human nasal respiratory epithelium.
Toxicol. Pathol. 17:33-37.
Kulle TJ, Letter to C.S. Scott, US EPA, August 28, 1985.
Lee HK, Alarie Y, Karol MH. 1984. Indication of formaldehyde
sensitivity in guinea pigs. Toxicol. Appl. Pharmacol. 75:147-155.
Levine RJ, DalCorso RD, Blunden PB, Battigelli MC. 1984. The
effects of occupational exposure on the respiratory health of
West Virginia morticians. J. Occup. Med. 26:91-98.
Liber HL, Benforado K, Crosby RM, Simpson D, Skopek TR. 1989.
Formaldehyde-induced and spontaneous alterations in human hprt
DMA sequence and mRNA expression. Mutat. Res. 226:31-37.
Liu KS, Huang FY, Hayward SB, Wesolowski J, Sexton K. Irritant
effects of formaldehyde exposure in mobile homes. Environ. Hlth.
Perspec. (in press).
Ma T-H, Harris MN. 1988. Review of the genotoxicity of
formaldehyde. Mutat. Res. 196:37-59.
148
-------
Malaka T, Kodama AM. 1990. Respiratory health of plywood workers
occupationally exposed to formaldehyde. Arch. Environ. Health
45:288-294.
Malker HSR, McLaughlin JK, Weiner JA, Silverman DT, Blot WJ,
Ericsson JL£j _Fraumeni JF Jr. 1990. Occupational risk factors for
nasopharyngeal '^cancer; in Sweden. Br. J. Ind. Med. 47:213-214.
Maronpot RR, Miller RA, Clarke WJ, Westerberg RB, Decker JR, Moss
OR. 1986. Toxicity of formaldehyde vapor in B6C3F1 mice exposed
for 13 weeks. Toxicology 41:253-266.
Meuller R, Raabe G, Schumann D. 1978. Leukoplakia induced by
repeated deposition of formalin in rabbit oral mucosa: Long-term
experiments with a new "Oral Tank". Exp. Pathol. 16: 36-42.
Monticello TM. 1990. Formaldehyde-induced pathology and cell
proliferation. Ph.D. Dissertation submitted to the Department of
Pathology, Duke University, North Carolina.
Monticello TN, Morgan ET, Everitt JI, Popp JA. 1989. Effects of
formaldehyde gas on the respiratory tract of Rhesus monkeys. Am.
J. Pathol. 134:515-527.
Monticello TN, Morgan KT. 1990. Correlation of cell proliferation
and inflammation with nasal tumors in F-344 rats following
chronic formaldehyde exposure. Am. Assoc. Cancer Res. 31:139.
Moolgavkar SH, Knudson AG Jr. 1981. Mutation and cancer: a model
for human carcinogenesis. J. Natl. Cancer Inst. 66:1037-1052.
Morgan ET, Jiang X-Z, Starr TB, Kerns WD. 1986a. More precise
localization of nasal tumors associated with chronic exposure of
F-344 rats to formaldehyde gas. Toxicol. Appl. Pharmacol. 82:
264-271.
Morgan ET, Patterson DL, Gross EA. 1983. Formaldehyde and the
Nasal Mucociliary Apparatus. In: Formaldehyde: Toxicology-
Epidemiology Mechanisms. Clary JJ, Gibson JE, and Waritz RS,
(eds.)i Marcel Dekker, New York, p.193-209.
Morgan KT, Patterson DL, Gross EA. 1986b. Response of the Nasal
Mucociliary Apparatus of F-344 Rats to Formaldehyde Gas. Toxicol.
Appl. Pharmacol. 82:1-13.
National Research Council, Committee on Aldehydes, Board on
Toxicology and Environmental Health Hazards. 1981. Formaldehyde
and other aldehydes. Washington, DC, National Academy Press.
Nunn AJ, Craigen AA, Darbyshire JH, Venables KM, Newman Taylor
AJ. 1990. Six year follow up of lung function in men
occupationally exposed to formaldehyde. Br. J. Ind. Med. 47:747-
752. >;
149
-------
Olsen JH, Asnaes S. 1986. . ormaldehyde and the risk of squamous
cell carcinoma of the sinonasal cavities. Br. J. Ind. Med.
43:769-774.
Olsen JH, Jensen SP, Hink M, Faurbo K, Breum NO, Jensen ON. 1984.
Occupational f-ormaldehyde exposure and increased nasal cancer
risk in man. Int. J. Cancer 34:639-644.
Partanen T, Kauppinen T, Hernberg S, Nickels J, Luukkoner R,
Hakulinen T,. Pukkala* MA. 1990. Formaldehyde exposure and
respiratory cancer among woodworkers - an update. Scan. J. Work
Environ. Health 16:394-400.
Partanen T, Kauppinen T, Nurminen M, Nickels J, Hernberg S,
Hakulinen T, Pukkala E, Savonen E. 1985. Formaldehyde exposure
and respiratory and related cancers: A case-referent study among
Finnish woodworkers. Scand. J. Work Environ. Health 11:409-411.
Pross HF, Day JH, Clark RH, Lees REM. 1987. Immunologic studies
of subjects with asthma exposed to formaldehyde and urea-
formaldehyde foam insulation (UFFI) off products. J. Allergy
Clin. Immunol. 79:797-810.
Ritchie IN, Lehnen RG. 1987. Formaldehyde-related health
complaints of residents living in mobile and conventional homes.
Am. J. Public Health 77:323-328.
Robins J, Pambrum M, Chute C, Blevins D. 1988. Estimating the
effect of formaldehyde exposure on lung cancer and non-malignant
respiratory disease (NNRD) mortality using a new method to
control for the healthy worker survivor effect. In: Progress in
Occupational Epidemiology. Hogstedt, C and Reuterwall, C (Eds).
Exerpta Medica, Amsterdam, p. 75-78.
Ross WE, McMilan DR, Ross CF. 1981. Comparison of DNA damage by
methylmelamines and formaldehyde. J. Natl. Cancer Inst. 67:2-17.
Rothman KJ, Boice JD. 1982. Epidemiologic analysis with a
programmable calculator, 2nd Edition. Boston, Epidemiology
Resources, Inc.
Roush G, Walrath J, Stayner L, Kaplan S, Flannery JT, Blair A.
1987. Nasopharyngeal cancer, sinonasal cancer, and occupations
related to formaldehyde: A case-control study. J. Natl. Cancer
Inst. 79:1221-1224.
Rusch GM, Clary JJ, Rinehart WE, Bolte HE. 1983. A 26-week
inhalation toxicity with formaldehyde in the monkey, rat, and
hamster. Toxicol. Appl. Pharmacol. 68:329-343.
Sauder LR, Chatham MD, Green DJ, Kulle TJ. 1986. Acute pulmonary
response to formaldehyde exposure in healthy nonsmokers. J.
Occup. Med. 28:420-424.
150
-------
Sauder LR, Green DJ, Chatham ND, Kulle TJ. 1987. Acute pulmonary
response of asthmatics to 3.0 ppm formaldehyde. Toxicol. Indus.
Health. 3:569-578.
Schachter EN, Witek TJ, Brody DJ, Tosun T, Beck GJ, Leaderer BP.
1987. A study .of respiratory effects from exposure to 2.0 ppm
formaldehyde in occupationally exposed workers. Environ. Res.
44:188-205. -> ''n, .
Schachter EN, Witek TJ, Tosun T, Beck GJ. 1986. A study of
respiratory effects from exposure to 2 ppm formaldehyde in
healthy subjects. Arch. Environ. Health 41:229-239.
Schmid E, Goggelmann W, Bauchinger N. 1986. Formaldehyde-induced
cytotoxic, genotoxic and mutagenic response in human lymphocytes
and Salmonella typhimurium, Mutagenesis 1:427-431.
Sellakumar AR, Snyder CA, Solomon JJ, Albert RE. 1985.
Carcinogenicity of formaldehyde and hydrogen chloride in rats.
Toxicol. Appl. Pharmacol. 81:401-406.
Siemiatycki J, Richardson L, Gerin M, Goldberg N, Dewar R, Desy
M, Campbell S, Wacholder S. 1986. Associations between several
sites of cancer and nine organic dusts: Results from an
hypothesis-generating case-control study in Montreal, 1979-1983.
Am. J. Epidemiol. 123:235-249.
Siemiatycki J, Wacholder S, Richardson L, Dewar R, Gerin N.
1987a. Discovering carcinogens in the occupational environment:
Methods of data collection and analysis of a large case-referent
monitoring system. Scand. J. Work Environ. Health 13:4860492.
Siemiatycki J, Dewar R, Nadon L, Gerin M, Richardson L, Wacholder
S. 1987b. Associations between several sites of cancer and twelve
petroleum-derived liquids: Results from a case-referent study in
Montreal. Scand. J. Work. Environ. Health 13:486-492.
Snyder RD, Van Houten B. 1986. Genotoxicity of formaldehyde and
an evaluation of its effect on the DNA repair process in human
diploid fibroblasts. Mutat. Res. 165:21-30.
Soffritti M, Maltoni C, Maffei F, Biagi R. 1989. Formaldehyde:
An experimental multipotent carcinogen. Toxicol. Indust. Health
5:699-730.
Solomon NJ, Varshavsky A. 1985. Formaldehyde-mediated DNA-protein
crosslinking: A probe for in vivo chromatin structures. Proc.
Natl. Acad. Sci. 82:6470-6474.
Starr TB. 1990. Quantitative cancer risk estimation for
formaldehyde. Risk Anal. 10:85-91.
151
-------
Stayner LT, Elliott L, Blade L, Keenlyside 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.
Sterling TD,~ Weinkam JJ. 1988. Reanalysis of lung cancer
mortality in a'National Cancer Institute study on mortality among
industrial workers exposed to formaldehyde. J. Occup. Med.
30:895-901.
Sterling TD, Weinkam JJ. 1989. Reanalysis of lung cancer
mortalityin a National Cancer Institute study of "Mortality among
industrial workers exposed to formaldehyde": additional comments.
J. Occup. Med. 31:881-884.
Stern FB; Beaumont II, Halperin WE, Nurthy LI, Hills BW, Fajen
JM. 1987. Mortality of chrome leather tannery workers and
chemical exposures and tanneries. Scand. J. Work Environ. Health
13:108-117.
Stroup N, Blair A, Erickson G. 1984. Brain cancer and other
causes of death in anatomists. Presented at the Society for
Epidemiologic Research 17th Annual Meeting, Houston, Texas, June
13-15. Am. J. Epidemiol. 120:500.
Swenberg JA, Barrow CS, Boreikol CJ, Heck Hd'A, Levine RJ, Morgan
ET, Starr TB. 1983. Non-linear biological responses to
formaldehyde and their implications for carcinogenic risk
assessment. Carcinogenesis 4:945-952.
Takahashi M, Hasegawa R, Furukawa F, Toyoda K, Sato H, Hayashi Y.
1986. .Effects of ethanol, potassium metabisufite, formaldehyde
and hydrogen peroxide on gastric carcinogenesis in rats after
initiation with N-methyl-N-nitro-N-nitrosoguanidine. Jpn. J.
Cancer Res. (Gann) 77:118-124.
Thrasher JD, Broughton A, Micevich P. 1988. Antibodies and immune
profiles of individuals occupationally exposed to formaldehyde:
Six case reports. Am. J. Ind. Med. 14:479-488.
Thrasher JD, Broughton A, Madison R. 1990. Immune activation and
autoantibodies in humans with long-term inhalation exposure to
formaldehyde. Arch. Environ. Health 45:217-223.
Thrasher JD, Madison R, Broughton A, Card Z. 1989. Building-
related illness and antibodies to albumin conjugates of
formaldehyde, toluene diisocyanate, and trimellitic anhydride.
Am. J. Ind. Med. 15:187-195.
Thrasher JD, 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.
-------
Til HP, Woutersen PA, Feron VJ, Hollanders VHM, Falke HE. 1989.
Two-year drinking-water study of formaldehyde in rats. Fd. Chem.
Toxicol. 27:77-87.
Tobe M, Kaneko T, Uchida Y, Kamata E, Ogawa Y, Ikeda Y, Saito M.
1985. Studies-of the inhalation toxicity of formaldehyde.
National Sanitary and Medical laboratory Service (Japan), pp.
1294.
\ ,
Tobe M, Naito K, Kurokawa Y. 1989. Chronic toxicity study on
formaldehyde administered orally to rats. Toxicology 56:79-86.
Tollefson L, Lorentzen RJ, Brown RN, Springer JA. 1990.
Comparison of the cancer risk of ethylene chloride predicted from
animal bioassay data with the epidemiologic evidence. Risk Anal.
10:429-435.
Uba G, Pachorek D, Bernstein J, Garabrant DH, Balmes JR, 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.
Universities Associated for Research and Education in Pathology,
Inc. 1988. Epidemiology of chronic occupational exposure to
formaldehyde: report of the ad hoc panel on health aspects of
formaldehyde. Toxicol. Indus. Health 4:77-90.
Ura H, Nowak P, Litwin S, Watts P, Bonfil RD, Klein-Szanto AJP.
1989. Effects of formaldehyde on normal xenotransplanted human
tracheobronchial epithelium. Am. J. Pathol. 134:99-106.
Vaughan TL, Davis S. 1990. Wood dust exposure and squamous cell
cancers of the upper respiratory tract. Am. J. Epidemiol.
133:560-564.
Vaughan TL, Strader C, Davis S, Daling JR. 1986a. Formaldehyde
and cancers of the pharynx, sinus and nasal cavity: I.
Occupational exposures. Int. J. Cancer 38:677-683.
Vaughan TL, Strader C, Davis S, Daling JR. 1986b. Formaldehyde
and cancers of the pharynx, sinus and nasal cavity: II.
Residential exposures. Int. J. Cancer 38:685-688.
Walker JF. 1975. Formaldehyde. Robert E. Kriegeiz Publishing
Company, Huntington, New York, pp. 373-404.
Walrath J, Fraumeni J. 1983. Mortality patterns among embalmers.
Int. J. Cancer 31:407-411.
Walrath J, Fraumeni JF. 1984. Cancer and other causes of death
among embalmers. Cancer Res. 44:4638-4641.
153
-------
Watanabe F, Natsunaga T, Soejima T, Iwata, Y. 1954. Study on
aldehyde carcinogenicity. Part I. Experimentally induced rat
sarcomas by repeated injection of formalin. Gann 45:451-452.
Watanabe F, Sugimoto S. 1955. Studies on aldehyde
carcinogenicity. Part II. Seven cases of transplantable sarcomas
of rats developed in-.^he area of repeated subcutaneous injections
of Urotropin (Hexame£hylenetetramine). Gann 46:365-367.
Wilhelmsson B, Holmstrom M. 1987. Positive formaldehyde-RAST
after prolonged formaldehyde exposure by inhalation. The Lancet
2:164 .
Wilkins RJ, MacLeod HD. 1976. Formaldehyde induced DNA-protein
crosslinks in Escherichia coli. Mutat. Res. 36:11.
Wilmer JWGM, Woutersen PA, Appleman IN, Leeman VTR, Feron VJ.
1989. Subchronic (13-week) inhalation toxicity study of
formaldehyde in male rats: 8-hour intermittent versus 8-hour
continuous exposure. Toxicol. Lett. 47:287-293.
Witek TJ, Schachter EN, Tosun T, Beck GJ, 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.
Witek TJ, Schachter EN, Tosun T, Leaderer BP, Beck GJ. 1986.
Controlled human studies on the pulmonary effects of indoor air
pollution: experiences with sulfur dioxide and formaldehyde.
Environ. Int. 12:129-135.
Woutersen PA, van Garderen-Hoetroer A, Bruijntjes JP, Zwart A,
Feron VJ. 1989. Nasal tumors in rats after severe injury to the
nasal mucosa and prolonged exposure to 10 ppm formaldehyde. J.
Appl. Toxicol. 9:39-46.
Yager JW, Cohn KL, Spear RC, Fisher JN, Morse L. 1986. Sister-
chromatid exchanges in lymphocytes of anatomy students exposed to
formaldehyde-embalming solution. Mutat. Res. 17:135-139.
Zwart A, Woutersen PA, Appelman LM, Wilmer JWGM, Spit BJ. 1988.
Cytotoxic and adaptive effects in rats nasal epithelium after 3-
day and 13-week exposure to low concentrations of formaldehyde
vapour. Toxicology 51:87-99.
154
-------
APPENDIX A:
Summary of Epidemiologic Studies Cited
1987 and 1990 EPA Formaldehyde Risk Assessments
-------
fable A 1
Risks Observed in SMR Studies
Study Size
FORKAL IN- EXPOSED COHORTS
Mataooski (1982) 1336
pathologists
. no exposure data
pathologists 1439
. no exposure data
Harrington and Shannon (1975) 2079
pathologists
„ no exposure data
Harrington and Oakes (1984) 2307
nale pathologists
. no exposure data
Levine et ai , (1984) 1477
morticians
„ TUA: 0.02 pp» (1980's)
. Mean TUA while eabalning;
0.3 - 0.9 ppn (1980's)
„ Peak: 0.4 - 2.1 ppm (1980's)
Cancer Site
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
buccal cavity and pharynx
lung
colon
brain
tymphopoiet ic
leukemia
nasal
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
Observed
NG8
NG
NG
5
7
NG
0
NG
6
NG
1
5
NG
0
NG
11
NG
NG
8
1
0
NG
9
NG
4
2
1
0
1
19
NG
3
8
4
0
Expected SMR
NG
NG
MGK
1.7*
9.5b
NG
NG
NG
8.1b
NG
'•^
6.2b
NG
NG
NG
27.9
NG
NG
4.0
1.6
NG
NG
22.0
NG
1.2
3.0
1.1
0.1e
2.1
20.2
NG
2.6
6.5
2.5
0.2
296
74
...
S '_
"'' 74
*
82
81
*39
*200
63
•
**1
*331
67
91
48
94
115
124
160
95X Confidence
L imi ts
Lower Upper
96
30
...
27
2
26
22C
100C
3C
21C
114C
12C
5C
1
57
24
53
44
686
152
161
464
188
65d
361d
. 296d
71d
..."
763d
210d
431d
2996d
265
147
337
243
410
1844
-------
Iable^TT-1
Risks Observed in SMR Studies
Study Size
Stroup (1984) 2239
anatomists
. 1 ppm • 3 ppm (1974-1980)
. Higher peak exposure
INDUSTRIAL COHORTS
Stayner et •(. (1988) 11030
garment workers
Concurrent:
. Average TUA, 0.15 ppn
(1981-1984)
. Higher past exposures
Acheson et aLt (1984) 7716
formaldehyde and resin
production workers
. particulates present
. < 0.1 ppN - > 2.0 ppn
BIP plant (subset of cohort)
Cancer Site
buccal cavity and pharynx
lung
colon
brain
lymphopoietic
leukemia
nasal
brain
leukemia
buccal cavity and pharynx
buccal cavity
pharynx
lung
colon
brain
tymphopoiet ic
leukemi a
connective tissue
nasal
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
lung
nose
Observed
1
12
20
10
18
10
0
11
8
6
4
2
39
NG
5
18
9
4
0
5
205
NG
5
20
9
0
166
0
Expected
6.8
43.0
18.5.
3.7
14.4
6.7
0 4
1>
3.8b
3.9
1.2
1.8
34.1
NG
7.0
19.8
7.9
1.1
0.6
4.3
196.0
NG
12.5
26.3
11.4
1.1
141.0
0.7
SXR
•15
•28
108 •
•271
125
148
...
•579 -•
212
-
•i
155
•343
113
114
...
71
91
114
•364
116
105
40*
76
79
...
118'
95X Confidence
L imi ts
Lower upper
0
14
66
130
74
72
...
' 289
91
68C
118C
20C
86C
28C
60C
123C
38
91
13
46
36
...
100
82
49
167
497
196
274
922
1036
415
107**
786d
359d
149d
...
149*^
135d
200d
825d
499**
271
120
94
117
150
335
137
3527
. > 2.0 ppn
. particulars
*-?
-------
. 0.13 pp» - 2.5 ppm (1974-1979)
Range:
. q.5*ppw • 9.8 ppM (1974-1979)
. participates present
Blair et al.fl (1986. 1987)
formaldehyde and resin
production workers
(white (tales)
. > 0.1 pp», TWA
(average exposure)
Table A-1 (continued)
Bisks Observed in SMR Studies
Study
per tan i et al. (1985. 1986)
for»aldehyde-
exposed
retfn yorkert
Mean:
Size Cancer Site
4462 buccal cavity and pharynx
lung . . . .
colon
brain
lymphopoietic
leukemia
nasal
Observed
NG
5
NG
NG
3
NG
0
Expected
NG
3.7
NC
NG
1.1
NG
NG
95X Confidence
L lltii ts
SMR Lower Upper
136 44 315
273 ; 56 797
26561
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
18
20J
42
17
56
19
2
19
192
48
21
62
24
2.2
96
111
87
81
91
80
91
57
96
63
47
62
48
11
150
127
117
130
184
124
328
< 0.5 ppn-yr - > 5.5 ppm-yr
(cunul ,,i ive exposure)
with > 20 years latency (subset of
exposure group)
simultaneous paniculate
exposure
buccal cavity and pharynx
nasopharynx
lung
colon
brain
lymphopoietic
leukemia
lung
nasopharynx
19
6
236
55
NG
64
23
146
22
2
214
57
NG
73
28
109
86
300*
110
96
88
82
133*
52
110
97
68
52
113
135
653
125
112
123
157
0.5 ppm-yr
0.5 - < 5.5 pp»-yr
5.5 ppm-yr
0.5
0.5
0.3
192
403
746
5
48
81
1114
1445
2408
-------
A-1 (continued)
Risks Observed in SMR Studies
Study
Collins et al, (1988)
(bated on Blair et al.
1986. 1987)
Sterling and We Ink am
personal connunication
(based on Blair et al.
1986)
Blair et al^ (1990a)
(based on Blair et at.
0
< 0.
0.
> 5.
< 0.
0.
0.
2+
5
5
5
1
1
5
Size
7 case
subset
Cancer Site
nasopharynx
pp»-yr » port icutstes
- < 5.5 ppm-yr
* particulates
ppm-yr + particulates
280 case
subset
ppm-yr
- 0.5 ppm-yr
- 2 ppm-yr
lung
ppn-yr
212 case
subset
lung
Observed Expected SMR
2 .9 215
2 .6 343
1 .5 216
2 .2 826'
*.
-
•A
1.00"
1.12"
1.K"
1.46"
95X Confidence
Limits
Lower Upper
25
37
3
112
0.76"
0.73"
0.87"
802
1204
1113
3611
_ .
1.84"
1.77"
2.49"
1986)
exposure to fornatdebyde
only
exposure to formaldehyde
and other agents
> 0 • <0.5 pprn-yr
0.5 - 5.5 ppw-yr
> 5.5 ppn-yr
0 • <0.5 ppm-yr
0.5 - 5.5 ppm-yr
5.5 ppm yr
40
35
13
39
43
42
36.2
31.8
17.5
25.2
36.3
27.2
110
110
74
154«
118
154«
80
78
41
110
87
113
149
151
124
212
158
207
A-4
-------
Table A-1 (continued)
Risks Observed in SMR Studies
Study Size Cancer Site
Stern et al, (1987) . 9,365 buccal cavity and pharynx
- chrome leather lung
tannery worker* . colon
brain
. 0.5 pp» - 7 pp» (1981), lymphopoietic
finishing department leukemia
. Hean TUA, 2.5 pp» (1981) nasal
Malker et al. (1990)
*i no exposure data 471 incident cases nasopharnyx
cases from Swedish fiber board
Environment -Cancer manufacture
Registry textile workers
furniture workers
chemical workers
shoe repa i r
Observed
1
24
NG
NG
14
.7
19
4
1
3
1
5
Expected
• NG
34.3
NG
NG
13.2
5.6
.49
1.0
2.5
3.8
1.7
1.3
SMR
70
106
125
250 ,
s '
r
3 9 ^
O.V
0.8'
0.6*
•4.01
95X Confidence
L irai ts
Lower Upper
45
•-
58
50
6
1.08'
0.01'
0.16'
0.01'
1.24'
105
178
258
1393
10.24'
2.23'
2.31'
3.271
8.971
Footnotes:
* p<0.05.
* NG, observed or expected nuifcer of deaths not given in paper.
b Age-specific mortality rates of psychiatrists used as the comparison group.
c Lower 90X confidence interval.
d Upper 90X confidence interval.
* As described in Levine et al. (1984).
* Observed and expected nuit>ers are for workers in the finishing department.
8 Given in text.
h Odds ratio.
' standardized incidence ratio.
Cohorts studied by Wong (1983) end Tabershaw Associates (1982) are included in the Blair et al. (1986) study and thus, are not included in this table.
Confidence interval* were obtained from the published papers. In those cases where this information was lacking, EPA caluculated the confidence
interval.
A-5
-------
Study
Size
Cancer Site
Observed
Expected
PUR
95 X Confidence
L inits
Lower Upper
FORMAL IM - EXPOSED:
Ualrath and Fraumeni (1983)
NT enbalmers
aod funeral
directors
Presumed exposure levels
. Peaks: up to 5.3 ppm (197*0
. Average: 0.25 - 1.4 pp» (1975)
* Average concentration:
0.25 - 1.* pp» (1975)
Wat rath and fratmeni (1984)
California
enbalners
Presumed exposure levels
. Peaks: up to 5.3 ppm (1975)
. Average: 0.25 - 1.4 ppn (1975)
* Average concentration:
0.25 - 1.4 ppM (1975)
Hayes et al. (1990)
embalmers and funeral
directors (uhite)
Presumed exposure levels
. Average: 0.98 ppm. 3.99 ppm
. Total dust: 0.07 - 0.78 3
1132
1050
3649
buccal cavity and pharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
buccal cavity and pharynx
lung
colon
brain
Iymphopoict ic
leukemia
nasal
buccal cavity and pharynx
nasopharynx
lung
colon
brain
lymphopoiet ic
leukemia
nasal
8
72
29
9
25
12
0
8
41
30
9
1!
26
3
285
95
24
100
NG
NG
7.1
66.8
20.3
5.8
20.6
8.5
0.5C
6.1
42.9
16.0
4.7
0.6
21.8
1.6
294.0
80.5
19.4
76.4
NG
NG
113
108
•143
156
121
''140
131
96
•187
•193
119
189
97
118
123
131'
56"
88
102
81
84
81
65a
72a
1358
100s
1^8a
100a
78
39
86
95
80
106
88
203
131
194
271
170
229
599b
237*"
124b
254b
334^
282b
499
174
548
109
144
184
259
605
161
INDUSTRIAL EXPOSED:
Del lei I end Grufferman
textile
workers
. exposure data lacking
4462
buccal cavity and pharynx
lung
colon
brain
lywphopoiet ic
leukemia
nasal
18
106
115
17
121
45
NG
18.0
117.8
115
18.9
64.2
37.5
NG
100
90
100
90
188
120
59
73
83
52
156
144
88
158
108
120
144
225
605
161
A-6
-------
Table A-2 (continued)
Footnotes:
* p<0.05
* Lower 90X confidence interval.
b Upper 90X confidence interval.
c As publUhed In Levin* et •!. (19M). '
i
The individuals studied by Harsh (1983) and liebltng et at. (1984) are included in the Blair et al. (1986) study, and thus are not listed separately
here.
A-7
-------
Table A-J
Odds Ratios Observed in Case-Control Studies
Study
Hardetl et at. (1982)
nasal and nasopharyngeal
case* in Sweden
. exposure level* lacking
Size Cancer Site Exposure Ratio = P
44 cases/ nasal cavity part icteboard
541 controls production:
males (0.8X)
Odds
Ratio
5.8*
95X Confidence
Limits
Lower Upper
1.2 25.9
01 sen et al. (1984)
nasal <«xJ nasopharyngeal cancer
cases in Denmark
„ exposure levels lacking
Hayes et al . (196%, 1986)
nasal and nasal
sinus cases in
the Netherlands
. exposure levels lacking
839 cases/
2465 controls
144 cases/
353 controls
(1:2 match)
nasal cavity
and sinuses
nasopharynx
nasal cavity
and sinuses
formaldehyde:
females (0.1X)
males (4.2X)
formaldehyde:
females (0.1X)
males (4.2X)
formaldehyde:
males with no or low
level wood dust
exposure:
classi ficat ion A
(6.2X)
classi f icat ion B
(27.3X)
males with high
level wood dust
exposure:
classi fication A
(47.1X)
classi ficat ion B
(91.2X)
formaldehyde:
males, controlled
for smoking:
classification A
class if ication B
2.8
2.8*
2.6
0.7
2.5*
1.6
1.9
0.5"
1.3"
0.3a
0.3"
1.28
0.9a
0.78
14.3°
4.3b
21.9b
5.0C
2.8t
5.5C
2.2*
1.6
1.1*
0.98
2.8
A-8
-------
Table A-3 (continued)
Study
Fayerweathtr et al . ( 1 982 )
cancer deaths
in cheaical
workers
Partenen et al. (1985)
nested respiratory
cancer case-control
study
Partcnen et al . (1990)
nested respiratory
cancer case-control
study
Vaughan et al . (1986 o.b)
sinonasal and
pharyngeal cases in
Washington State
.exposure levels lacking
^,
'""
.'
Sife Cancer Site Exposure Ratio » PQ
481 cases/ lung, bronchus formaldehyde:
481 controls and trachea male workers
(1:1 match) intermittent
level #1 (10X)
(<2.0 ppm)
intermi ttant
level *2 (13X)
(>2.0 ppm)
cont inuous
level #1 (2X)
(<0.1 ppm)
cont inuous
level #2/(T3 (7X)
(0.1 ppm - 2.0 ppm)
55 cases/ respiratory formaldehyde:
169 controls system ever exposed (26. 6X)
level of exposure
0.1 - 1.0 ppm (16X)
> 1.0 ppm (7.7X)
136 cases/ respiratory formaldehyde:
408 controls system >3 ppm-months'
118 cases/ lung formaldehyde:
354 controls >3 ppm-months'
53 cases/ nas.al cavity occupational formaldehyde:
552 controls and sinuses emulative exposure
5 - 9 years (6.3X)
10 + years (10.9X)
occupational :
resins, glues and
adhesives
low exposure (6.5X)
high exposure (2.3X)
domestic:
mobile home residence
1 + years (12. OX)
27 cases/ nasopharyngeat occupational formaldehyde:
552 controls no. of years exposed
1 - 9 (25. OX)
10 + (10.3)
domestic:
mobile home residence
1 - 9 years (12. OX)
10 + years (3.7X)
Odds
Ratio
0.9
0.9
1.7
1.1
1.4
1.5
1.4
1.4
0.9
0.7
0.3
2.0
3.8*
0.6
1.2
1.6
2.1
5.5"
95X Confidence
Interval
Lower Upper
0.4
0.5
0.3
0.4
0.7"
0.7"
0.4a
0.78
0.3°
0.3
0.1
0.7
1.1
0.2
0.5
0.4
0.7
1.6
2.0
6.4
10.9
2.6
3.0b
3 4b
5>
4.1b
3.0b
1.4
1.9
•. .«.
5.6
12.9
1.7
3.1
5.8
6.6
19.4
A-9
-------
InbIc A i (continued)
Study
Gerin et el .(1989)
cancer cases among
males aged 35-70
in Montreal
Roush et al . (1987)
nasal sinus and
Size Cancer Site Exposure Ratio = P
174 cases/ oro-hypo- occupational formaldehyde:
552 controls pharyngeal no. of years exposed
1 - 9 (25. OX)
10 + (10. 3X)
occupational :
resins, glues.
adhesives
low exposure0
high exposure
domest ic :
mobile home residence
1 - 9 years (12. OX)
10 « years (3.7X)
875 cases/ adenocorc inoma formaldehyde:
1525 controls of the lung long duration -
low level
medium level
high level
(TWA <1 .0 pprn)
371 cases/ nasal sinus formaldehyde:
605 controls high levels, >1 ppm,
Odds
Ratio
0.6
1.2
1.3
3.9*
0.9
0.8
0.5
1.0
2.2
95X Confidence
Interval
Lower Upper
0.3
0.7
0.6
1.5
0.5
0.2
0.2
0.4
0.7
1.0
2.5
3.0
10.1
1.8
2.7
1.3
2.5
7.6
nasopharyngeal cancer
cases in Connecticut
nasopharyngeal
at 20+ years prior
to death (3.3X)
high level, ^ 1 ppm,
at 20» years prior
to death (3.3X)
1.5
2.3
0.6
0.9
3.9
6.0
* p<0.05, except as noted.
, s
8 Lower 90X confidence interval.
" Upper 90X confidence interval.
c Prevalence of the exposure among the controls not cited by Partenen et al. (1990) nor Vaughan et al. (As reported in SA1C, 1986).
Confidence intervals were obtained from the published papers. In those cases where this information was lacking, EPA caluculated the confidence
interval.
-------
APPENDIX B
Comparison of Animal Model Risk Predictions
with Epidemiologic Evidence
An analysis was carried out to compare the epidemiologic data
base and the 1990 animal-based risk estimates, to see if light
can be shed on the procedure used to derive the 1990 risk
estimates. An analogous comparison was carried out in 1987. In
that comparison, site-specific excess risks were calculated by
multiplying the excess risk observed in the epidemiologic
studies, that is the excess above a risk of one, by the site-
specific mortality ratios in the concurrent general population.
These lifetime excess risks were then compared to upper bound
lifetime excess risks calculated by applying the 1987 unit risk
to representative exposure levels from the epidemiologic studies.
Risks were found to be comparable by the two approaches, with
rat-based excess risks tending to be within an order of magnitude
lower than the corresponding epidemiologically based excess
risks.
Such comparisons are approximate, and do not require the
epidemiologic data to be of sufficient quality to derive risk
estimates. Nevertheless, the data affording the most information
on exposure groupings were selected. Blair et al. (1986, 1987)
give information on relative cumulative exposure, and on observed
and expected numbers of nasopharyngeal cancer cases.
The current comparison was carried out following the approach
of Tollefson et al. (1990). This approach is equivalent to the
1987 approach, but focuses on numbers of excess nasopharyngeal
cancer deaths, rather than on risk of nasopharyngeal cancer. The
steps are outlined in Table B-l, and are described below.
»• Steps a - c. Excess nasopharyngeal cancer deaths observed in
the Blair et al. (1986) study were calculated as the
difference between the observed deaths (step a in Table B-l)
and the deaths expected based on 1980 US rates for the period
covered by the study (step b).
>
•• Under the hypothesis that the animal-based 1991 unit risk
estimates apply, the excess nasopharyngeal cancer risks to be
expected in the Blair cohort were predicted by applying the
monkey dosimetry unit risk, 3^3 x 10" /ppm, and the rat
dosimetry unit risk, 2.8 x 10 /ppm, to the exposures reported
in Blair et al. (1986):
+ Step d. Cumulative exposures were reported in intervals, <
0.5, 0.5-5.5, and > 5.5 ppm-yr. Representative values for
the first 2 intervals were taken to be the upper limits 0.5
and 5.5 ppm-yr, ,-while the median of the range for the > 5
B-l
-------
ppm-yr interval (range 5-89 ppm-yr), 12.6 ppm-yr, was
used. These are conservative representations of exposure,
tending to elevate the risk predictions from animal data.
»• Steps e, f. These representative cumulative exposures were
converted- to average exposure rates by dividing each by the
average employment duration of the employees in each
exposure category. This approach ignores periods of zero
exposure before11/ and perhaps after, occupational exposure.
Accounting for these periods would tend to lower the animal-
based risk predictions.
*• Steps g, j. The average exposure rates were multiplied by
the 1991 unit risks, adjusted for 8-hour occupational
exposures. This yielded excess lifetime nasopharyngeal
cancer risks for each exposure category.
*• Steps h, k. Total excess nasopharyngeal cancer deaths were
predicted by applying these animal-based risk estimates to
the total number of employees in each exposure category.
This total represents the situation in which nasopharyngeal
cancer would lead to early enough death that all such cases
would have been observed during the period of this study.
*• Steps i, 1. Excess nasopharyngeal cancer deaths for the
period this study covered were predicted by adjusting the
total deaths expected by the proportion of the study group
which was expected to die during that time, based on
concurrent age-specific US mortality rates. Overall,
approximately 11% of the entire study group was expected to
have died by the time the study was concluded. Implicit in
this approach is the assumption that the rate of
nasopharyngeal cancer incidence is constant over lifespan.
The results are listed in Table B-l. The excess deaths
observed in the study due to nasopharyngeal cancer were
approximately 1 in each exposure category. The predicted upper
bounds on expected cancer deaths were on the order of 10" to 10
deaths.
B-2
-------
Table B-1: Comparison of Excess Nasopheryngeal Cancer (NPC) Deaths Observed in Blair et al. (1986),
with those Expected Using 1991 An ime I Models
Exposure Category (ppm-yr)
Excess human deaths-observed ,-^
in Blair et al. (1986): \^f
a Observed MPC deaths
b No. NPC deaths expected (US rates)
c Observed excess deaths (a - b)
Unexposed
1
0.2
0.8
< 0.5
2
0.7
1.3
0.5-5.5 5.5-89
2 2
0.8 0.5
1.2 1.5
Exposure rate calculations for use with 1991 unit risks:
d Cumulative Exposure Level
used for Comparison (ppx-yr)
e Average employment duration (years)
f Average Exposure Rate (ppm) (oVe)
U.O
0.0
0.5
11.8
0.042
5.5
15.7
0.350
12.6*
17.9
0.704
Excess predicted deaths using 1991 monkey dosimetry
unit risk:
g Individual lifetime excess risk
(f x 8/24 x 3.3x10'")
h Total number predicted NPC deaths
(g x N)
i Heater predicted excess
MFC deaths (h x D/N)
0
0
0.0
5x10- 4x10-' 8x10''
0.08
0.004
0.4
0.04
0.3
0.05*
Excess predicted deaths using 1991 rat otosimetry
unit risk:
j Individual lifetime excess risk
(f x 8/24 x 2.8x10 ')
k Total number predicted NPC deaths
(j x N)
I Muster predicted exeats
*K deaths (k x D/N)
0.0
4x10'' 3x10''
0.7
0.05
3.4
0.5
7x10"
2.5
0.5*
Characteristics of Hair tt al. (1986) cohort:
N - Number of employees 2638 17769 10484 3767
0 - Expected deaths, all causes (US rates) 353 12S4 1414 783
* 12.6 was the median exposure level (fro* transcripts of 1987 OSHA formaldehyde hearing).
' Using the upper limit of the highest exposure range, the number of predicted excess NPC deaths would be
(average continuous exposure rate « 89/17.9 x 8/24 « 1.66)
Monkey dosimetry - Total excess NPC deaths: 1.66 x 3.3x10'4 x 3767 « 2.1
Among expected deaths: 1.66 x 3.3x10'' x 783 > 0.4
•at dosimetry - Total excess NPC deaths: 1.66 x 2.8x10' x 3767 * 17.5
Among expected deaths: 1.66 x 2.8x10'' x 783 * 3.6
B-3
-------
Appendix C
The Anatomy and Physiology of
the Nksal'passages of Humans, Monkeys, and Rats
-------
ANATOMY AND PHYSIOLOGY
\ '\
Overview of the Upper Respiratory Region
The respiratory system consists of a specialized interface for exchange of gases between the
blood and air, and a series of passageways leading from the interface to the ambient air. Clinically,
the respiratory system is often divided into upper and lower respiratory regions based upon the gross
anatomical location of the structures. While this division has been criticized as being vague when
attempting to study deposition of paniculates in various portions of the airways (Bates et al.. 1966), it
is easily visualized. Figure 1 illustrates the various components of the respiratory system. As
depicted in figure 1, those structures contained within the thorax compose the lower respiratory
region; extrathoracic structures constitute the upper respiratory region.
The upper respiratory airways serve as a conduit from the ambient air to the lungs. The
particular anatomical territories included in the upper airways are the nasal passages, pharynx, larynx.
and trachea. In higher primates, including humans (figure 2) and monkeys (figure 3), the arrangement
i
of the upper airways is "L" shaped (Schreider, 1986). This arrangement is due to the orientation of
the face, which is perpendicular relative to the body, causing the upper respiratory airways to exhibi
a 90 degree turn in the pharynx (Patra, 1986). In this configuration, the larynx is inferior to the oral
cavity, and the epiglottis (a cartilaginous flap that is attached to the anterior wall of the larynx and is
used to seal off the trachea during swallowing) does not appose the soft palate. Consequently, since
both the oral cavity and nasal passages open into the pharynx, higher primates are able to breathe
through both their mouths and their noses (oronasal breathers [Negus, 1958; Reznik.1990]).
In the nt (figure 4), the layout of the upper respiratory structures is tandem such that the air
could travel in a nearly linear fashion from the nose to the bifurcation of the trachea (Schreider,
1986). The linear arrangement of the upper airways allows the larynx of rats to lie close to the
posterior edge of the oral and nasal cavities. In this condition, the epiglottis lies against the soft
c-i
-------
/Olfactory
fi V\ ,Phtrrnc««l orifice of
rO\/ / auditory tubt
UPPER
RESPIRATORY
RZGIONS
Pbtrynx
LOWER
RESPIRATORY
REGIONS
Broncb
Figure 1. Diagram of the human respiratory system, illustrating the major gross anatomical structures
and the division of these structures into upper and lower respiratory regions. (Modified
from Andenon, 1978).
C-2
-------
Figure 2. Sagittal section through the head of a human. Within the cranial cavity, note the small
olfactory bulb (OB), which receives and processes nervous stimuli from the olfactory nerves
in the roof of the nasal passage. The cerebral cortex (Q is extremely Urge. The external
nose is outlined in white. Note that most of the nasal passages is located within the skull (as
opposed to the"external nose). The nasal septum has been removed to reveal structures on
the lateral wall of the nasal cavity. The three turbinate bones are simple In structure and
appear as vertical shelves. The more caudal respiratory structures make a right angle turn
inrerioriy in the pharynx. The epiglottis is associated with the base of the tongue and does
not approach the soft palate. Consequently, the oral and nasal cavities are in open
communication with the pharynx, thereby permitting oronasal breathing. Dashed line
represents the plane of section through nasal cavity in Figure 5. C « cerebral cortex;
Cb * cerebellum; FS = frontal sinus; M = medulla oblongata; OB * olfactory bulb;
SC = spinal cord; SpS * sphenoid sinus.
C-3
-------
Soft Palate
Esophagus
\ \
C-4
-------
Turbinates
Soft Palace
Epigl
Esophagus
MONKEY
Figure 3. Sagittal section through the head of a monkey with the nasal sepcum removed to reveal
structures of the lateral nasal wall. The shape of the nasal cavity U similar to that of the
human. The three turbinate bones appear rather simple in structure; the middle turbinate is
plow shaped, protruding towards the nostrils; the superior turbinate is small. The caudal
upper respiratory structures make a virtual right angle turn at the region of the pharynx; note
the inferior position of the trachea relative to the pharynx and the separation between the
epiglottis and soft palate, thereby allowing oronasal breathing as in humans. Dashed line
represents the plane of section through nasal cavity in Figure 5.
C-5
-------
Turbinates
-rio-
turbinate
Soft Palate
Esophagus
RAT
Figure 4. Sagittal section through the head of rat with the nasal septa removed to reveal structures on
the lateral walls of the nasal cavity. The cranial cavity is proportionately smaller than that
of the human. Despite that, the olfactory bulb (OB) is large especially when it U compared
to the cerebral cortex (C). The nasal caviry is elongated. The major turbinate bones appear
complex in structure; note also the presence of the anteriorly placed atrioturbinate bone.
Note the nearly tandem arrangement of the respiratory structures and the consequent
apposition of the epiglottis to the soft palate, closing off the oral cavity from the pharynx
and making the rat an obligatory nose breather. Dashed line represents the plane of section
through nasal caviry in Figure 5. C * cerebral cortex; Cb » cerebellum; M » medulla
oblongata; NPC * Nasopalatine Canal; OB * olfactory bulb; SC * spinal cord
C-6
-------
palate. The apposition of the epiglottis to the soft palate in the resting condition isolates the oral
cavity from the respiratory airways and makes the rat an obligatory nose breather (Negus, 1958;
Proctor and Chang. 1983).
V.*
Nasal Passages
While the overall arrangement of the upper respiratory airways differs among these species in
that the rat is an obligate nose breather whose design is nearly linear whereas higher primates are
oronasal breathers whose upper airways exhibit a right angle turn, even greater differences occur in
the anatomy of the nasal passages. Despite the differences in nasal anatomy, the nasal passages of all
three species serve as the site for olfactory stimulation and for conditioning air (i. e., humidifying,
warming, removal of inhaled paniculate matter [Negus, 1958; Proctor and Chang, 1983; Schreider,
1986]) prior to its access to the gaseous exchange interface in the lungs. The following paragraphs
will outline the common features of the nasal passages shared by all three species and will point out
the major anatomical differences. The anatomical features are depicted in figures 2 (human),
3 (monkey), and 4 (rat).
In all three species, me nose contains a pair of openings (nares) leading into the vestibule of the
nasal cavity. The vestibule is relatively large in humans, and entry into the nasal cavity proper in
humans occurs through the nasal valve, a narrow opening flanked by cartilage and muscle (Proctor
and Chang. 1983; Reznik, 1990). A homologous structure does not exist in the rat While the rat
does not possess a muscular nasal valve, it does exhibit small cartilaginous projections into the
vestibule just inside the nares. These projections are the atrioturbinatss (Proctor and Chang, 1983;
Hebel and Stromberg, 1986). Similar projections do not exist in the vestibules of the primates. The
rat also possesses a small diameter, patent nasopalatine canal that connects the nasal cavity and the
oral cavity (Negus, 1958). While the higher primates do possess nasopalatine canals as embryos, they
give rise to the incisive canals in adults (Warwick and Williams, 1973). The incisive canals of higher
primates are covered by mucosa and are not patent conduits between the oral and nasal cavities
(Negus, 1958).
C-7
-------
The nasal cavity proper is divided into paired passageways by a vertical cartilaginous and bony
septum. The nasal passages extend into the head, superior to the roof of the mouth, posterioriy as far
as the pharynx. Only a portion of the nasal passages are contained in the external nose; the remainder
is located within the skull and is not readily visible. When viewed from the front (see figure 5), a
coronal section through the head reveals each of the nasal passages to be nearly triangular in shape
with the vertex oriented superiorly. The floor of the nasal passage (made up of the hard and soft
palates, which separate the oral and nasal cavities) and the medial wall (the nasal septum) are
relatively flat, with no remarkable surface features. In the rat and other rodents, the inferior aspect of
the nasal septum contains structures termed swell bodies that are composed of several vascular spaces
subjacent to the epithelium (Proctor and Chang, 1983; Reznik, 1990). These swell bodies are capable
of rapid engorgement with blood (Sorokin, 1988). When the swell bodies are in a collapsed
condition, the nasal septum exhibits a flat, vertical profile and the nasal passage is maximally open.
This allows inspired air to take a rather direct course through the nasal passage, which is especially
important when the animal is physically active. When the swell bodies are engorged, the inferior
aspect of the nasal septum encroaches on, and nearly obliterates, the inferior portion of the nasal
passage. In this condition, air is forced into the upper portion of the nasal airway, which houses the
olfactory apparatus. Swell bodies do not exist in the higher primates (Negus, 1958; Reznik, 1990).
In contrast to the relatively featureless contours of the septum and floor of the nasal cavity, the
lateral walls of the nasal passages possess a series of bony processes that project into the passageway.
The processes are termed turbinate bones or conchae; there are three major turbinates. While the
specific names differ among species, they will be referred to according to the human nomenclature for
ease of comparison. The three turbinates are designated the inferior, middle, and superior according
to their positions relative to the floor of the nasal passages and each other.
\
The most striking anatomical differences can be seen in the structure of the major turbinate
bones (Negus, 1958; Schreider. 1986). When examining the sagittal sections of the human and
monkey, three turbinates are readily discerned. The rat, however, appears to exhibit additional
C-8
-------
HUMAN
MONKEY
RAT
Figure 5. Coronal (frontal) sections through the left nasal passage of human, monkey, and rat Planes
of section are denoted by dashed lines on Rgures 2. 3. and 4, respectively. Note the simple,
comma-shaped profile of the turbinate bones in human and monkey. The rat turbinate bones
are complex, exhibiting a "t" shaped profile with scroll-like foldings at the ends. Note the
comparable size of the maxillary sinus (MS) in human and monkey; the MS of rat Is much
smaller. Only humans possess ethmoid sinuses (EC). Note also the position of the
(collapsed) swell body on the ventral aspect of the nasal septum or rat. Primates do not
possess septal swell bodies
-------
curb mates. What appears to be additional turbinates in the rat is caused by the existence of dorsal and
^ f %
ventral lamellae for each turbina'te'(Hebel and Stromberg. 1986). The complexity of the nasal
turbinate bone in the rat can best be appreciated by studying diagrams of coronal (frontal) sections
taken midway through the nasal cavities of humans, monkeys, and rats (figure 5). In each diagram,
only the left nasal passage is depicted. The plane of section is denoted by the dashed vertical lines on
the sagittal section of the appropriate species (figures 2-4). In all three species, the nasal septa are
thin with no prominent surface features and the lateral walls possess turbinate bones that protrude into
the nasal passage. In the monkey and human, the turbinate bones exhibit simple "comma" shaped
profiles that extend inferiorly into the nasal passage (Negus, 1958; Warwick and Williams. 1973;
Anderson, 1978). Both primate species possess large maxillary sinuses (MS) that open into the
middle meatus (inferior to the middle turbinate bone). Humans possess small sinuses (ethmoid air
cells [EC]) in the ethmoid bone forming part of the roof of the nasal passage; corresponding structures
do not exist in either monkeys or rats. Due to the plow shaped structure of the middle turbinate of the
monkey (see figure 4; Harkema et al., 1987; Harkema, 1990), the attachment site of the middle
turbinate to the lateral wall is not depicted in this diagram. In contrast to the primates, the nasal
passage of the rat exhibits elaborate turbinate bones that exhibit V shaped profiles when viewed in
cross section (Hebel and Stromberg, 1986; Schreider, 1986). The distal edges of the V often appear
scrolled. The effect of this arrangement of rurbinate bones in the rat is to greatly increase the surface
area of the nasal passage compared to monkey and human. The maxillary sinus (MS) of rats is very
small (Hebel and Stromberg, 1986).
Morphometry
The differences in anterior-posterior extent and complexity of the turbinates in the three
species result in differences in the amount of surface area that lines the nasal passages. Table 1
presents morphometric data for each species concerning the nasal cavity volumes, surfaces of the
nasal passages and other upper airways, as well as body weights, body surface areas, and parameters
of pulmonary physiology.
C-10
-------
Table 1. Morphometric and Physiological Data Tor the Nasal Passages, Upper Respiratory Tract, and
Pulmonary Systems of Humans, Monkeys, and Rats
Nasal Caviiv
Upper Respiratory
Pulmonary Phvsiolotiv
Species
Human
Rhesus
Monkey
Rai
Body
Weight
(Kg)
w»
„
0.25<4>
Body
Surface Area
(m2)
1.85<4>
0.35
0.045<4>
Relative Nasal
Volume Surface Area Surface Area (l)
(cm3) (cm2)
25<5> I60(3) 6.4
S(2> 62^) 7.75
0.26<9> I3.44<9> 51.7
Surface Areas (cm2)<2)
Pharynx Larynx Trachea
46.6 29.5 82.5
..
1.2 0.17 3
Tidal
Volume
(cm3)
W6>
70
2<6)
Breaths Mini
per minute Volu
(IVlM
J5<6) p(
34(6) 2 Calculated from data in Patra, 1986
<3> Snydcrctal., 1975
<4> Calabrcse. 1983
<3> Harkema. 1990
<6> Alarie. 1982
C7) Schrieder, 1986
<8> Heck elal.. 1989
W Gross el al.. 1982
-------
The human nasal passage proper has a volume of 25 cm3 (Harkema, 1990) with a surface area
of ,160 cm2, excluding the vestibule (Snyder, et al. 1975). The human vestibule has a surface area of
40 cm2 (Snyder et al., 1975); an estimate of the vestibular volume has not been published. Because
')-.'
the human vestibule is so large compared to those of rat and monkey, and because it shares so much
in common with the skin of the face, the vestibule was not considered as part of the nasal passages
proper for the interspecies comparisons in this paper. The values for surface areas and volumes for
the rat and monkey include the vestibule. The vestibules of these animals are considerably smaller
than that of the human, and specific surface area and volume data have not been published.
The absolute volume and surface area of the nasal passages in humans are, not unexpectedly,
larger th-~ those of both the rat and the monkey. In an attempt to compare these data among all three
species, relative nasal surface area ratios were calculated by dividing the nasal surface area by the
nasal cavity volume for each species. The relative nasal surface areas of humans and monkeys were
nearly equal (6.4 and 7.75, respectively). The relative nasal surface area of the rat (51.7) was
eight-fold larger than that of humans. This means that the rat has eight times more surface area per
unit volume in the nose than either of the primate species. Inspection of figure 5 reveals that this is
not an unexpected finding.
It is interesting to note that the surface area of the nasal passage is larger than that of the
trachea. In humans, the nasal passages exhibit twice the surface area of the trachea. In rats, due to
their elaborate turbinate structure, the surface area of the nasal passages is 4_5 times that of the
trachea.
Patra et al., (1986) performed a morphometric analysis of the nasopahryngeal airways of rat,
rhesus monkey, baboon, beagle, guinea pig, and a human child. Using nasal casts and literature
values, the authors calculated the cross sectional areas and perimeters of the airways at various
distances from the nares. The data was normalized to correct for differences in body size and
compared graphically. The graphs of the monkey and the baboon most closely resembled thoseofthe
child. The graphs of the rat and the guinea pig were similar to each other, but were significantly
C-12
-------
differeni from the primates. The graphs of the beagle differed from those of both the primates and the
rodents. From this analysis of.normalized data it can be concluded that, while the geometry of the
nasal passages and upper airways within taxonomic orders are similar, there are significant
differences in nasal cavity geometry, (beyond mere size) when one compares rats and humans.
Mucosa
In all three species, the nasal passages are lined by a well vascularized mucous membrane, or
mucosa, that serves as a barrier to the entry of inhaled materials into the nasal walls. The mucosa is
composed of an epithelial sheet, lying on a basement membrane and underlain with a thin layer of
loose connective tissue that contains blood and lymphatic vessels. The mucosa is named after the
type of overlying epithelium.
Epithelia. There are three major types of e pi the Li a in the nasal passages proper squamous,
respiratory, and olfactory (Bloom and Fawcett, 1968; Gross et al., 1982; Sorokin. 1988). The entry to
the nose is the vestibule, which is covered by modified skin (keratinized, stratified squamous
epithelium) containing sebaceous glands and hairs. Within the nasal cavity proper, the epithelial
covering becomes progressively more delicate, in terms of its ability to respond to stress, and more
specialized, in terms of its functions of conditioning air and olfaction.
Beyond the vestibule, the anterior nasal passages are covered by a moist stratified squamous
epithelium, consisting of multiple layers of cells that become thin and plate-like as they approach the
surface. While the surface of this epithelium is covered by mucus secretions emanating from the
adjacent respiratory epithelium, the squamous epithelium itself is non-secretory. Moist stratified
squamous epithelium is not as resilient as the modified skin of the vestibule. The number of layers of
cells decreases, and the squamous epithelium thins, as one travels deeper into the nasal passages. At
approximately the anterior end of the turbinate bones, the epithelium changes to respiratory
epithelium, which is more delicate than the moist squamous epithelium. The respiratory epithelium is
a mucus-secreting, ciliated, pseudostratified columnar epithelium with goblet cells. The underlying
C-13
-------
connective tissue contains numerous muco-serous glands. The secretions of these glands together
with that of the respiratory epithelium itself serve to keep the nasal passages moist In the roof of the
nasal cavity, the mucous membrane is covered by the highly specialized olfactory epithelium.
•> •",
Olfactory epithelium appears-yellow in the fresh condition. It is a tall pseudostratified columnar
epithelium that is higher than the respiratory epithelium and contains neither cilia nor goblet cells.
The olfactory cells are bipolar neurons that exhibit very long, nonmotile processes (stereocilia) that lie
on the epithelial surface and house the receptors for olfaction. The olfactory cells are surrounded by
sustentacular cells which exhibit numerous apical microvilli and continuously secrete a serous fluid.
The glands in the connective tissue are also serous glands.
The approximate distribution of these epithelia within the nasal passages of the human,
monkey, and rat is depicted in figure 6. While relatively defined lines of demarcation between areas
of the nasal passages covered by the various epithelia have been published (Hebel and Stromberg,
\
1986; Harkema, 1990), the zone of transition between adjacent areas of squamous and respiratory
epithelia is usually not abrupt and the transitional epithelia in these zones share the characteristics (in
graded fashion) of the adjacent epithelial types (Menco, 1983; Harkema et al.. 1987; Morgan and
Monticello, 1990).
Under conditions of physiological stress, the more delicate respiratory epithelia can be replaced
by a more hardy epithelium. This change usually involves the replacement of the respiratory
epithelium with moist, stratified squamous epithelium and is termed squamous metaplasia (Bloom
and Fawcett, 1968).
Measurements of the surface area of the nasal passages covered by the three major epithelia in
rats have been made by Gross et al. (1982). Such measurements for humans have not been made
except for the olfactory epithelium (Negus. 1958; Bloom and FtwcetL 1968). No surface &ea
measurements could be found for the monkey. The available surface area data for nasal epithelia are
presented in table 2. Based upon the percentage of total nasal epithelium that is olfactory, animals
such as the rat (50 percent olfactory epithelium) have been designated macrosmatic, whereas primates
c-u
-------
DISTRIBUTION OF EPITHELIA
HUMAN
MOlfKEY
RAT
Figure 6. Sagittal sections through the heads of human, monkey, and rat to fflusoite the extent of the
major types of epithelia in the nasal passages. Note the relatively large olfactory area in rat
compared to human and monkey. Stippled area « olfactory; clear • respiratory; griped -
015
-------
Table 2. Epithelial Surface Areas of the Nasal Passages
in Rats and Humans
Total Surface Olfactory Respiratory Squamous
Species Area (cm2) Area (cm2) % Area (cm2) % Area (cm2) %
Rat 13.44 6.75<» 50 6.23<» 47 0.440) 3
Human 160™ 12.5™ 8 NF - NF -
(J) Gross etal., 1982
(2) SnyderetaJ., 1975
(3) Negus, 1958
NF = not found
C-16
-------
including humans (8 percent olfactory epithelium) and monkeys are classified as microsmaoc (Negus,
1958; Reznik.1990). The relative importance of olfaction to rats versus humans can perhaps be
appreciated by comparing the. size of the olfactory bulb (OB) to the cerebral cortex (Q in each of the
species as depicted in figure 2 (human) and figure 4 (rat).
Vascular Supply. The mucosa of the nasal passages contains a rich vascular supply. An
unexpected feature of the arterial supply is that the olfactory and respiratory mucosae receive blood
from different sources (Warwick and Williams, 1973; Sorokin, 1988). The olfactory mucosa is
supplied by the posterior and anterior ethmoidal arteries, which are very small calibre branches of the
ophthalmic artery. The respiratory mucosa receives arterial supply from the sphenopalatine artery,
which is the terminal branch of the maxillary artery (the great artery of the face), as well as the greater
palatine artery, another branch of the maxillary. Both of these arteries are of greater calibre than the
ethmoidal arteries. Based on the arterial diameters, the blood supply to the respiratory epithelium
appears to be richer than that of the olfactory epithelium. This may be related to the secretory rates of
these epithelia, as discussed later. The epithelium of the vestibule and anterior nasal passages receive
blood supply from the nasal branches of the superior labial artery (a branch of the facial artery).
The swell bodies located on the ventral aspect of the nasal septum in rats are formed by a
superficial plexus of thin walled veins that resemble erectile tissue, except that they lack muscular
septa (Negus, 1958). While the higher primates do not possess swell bodies, thin walled venous
plexuses are present on the anterior aspect of the inferior and (sometimes) middle turbinates (Proctor
and Chang, 1983; Sorokin, 1988). Distension and collapse of these vascular structures can occur
extremely quickly (Negus, 1958) and, due to their location, these structures can alter both the course
of airflow through the nasal passages (Proctor and Chang, 1983) and the volume of the airway in the
nasal passages (Guilmette et al, 1989).
Nasal Secretions. Since the functions of the nasal passages include wanning and
humidification of air, removal of paniculate matter, and olfaction, it is not surprising that the nasal
mucosae are covered with secretions which are continuously produced, transported through the nasal
C-17
-------
passages, and removed. In humans, it has been estimated that the thickness of nasal secretions on the
mucosa is 0.5 mm (Snyder et al., 1975); that secretions are replaced every 10 minutes over the
posterior two-thirds of the nasal passages and once an hour over the anterior one-third (Negus, 1958);
and that the daily production volume of nasal secretions is 0.5-1.0 liters, most of which is transported
to the nasopharynx and swallowed (Snyder et al.. 1975).
The great majority of the nasal passage secretions is produced by the respiratory epithelium.
While this secretion is frequently referred to as "mucus", the secretion really consists of two distinct
components (Proctor and Chang, 1983). Immediately adjacent to the epithelial surface and extending
above it 5-30 UJTI is a layer of periciliary fluid. This periciliary fluid is a transudate of plasma
(Sorokin, 1988) that provides a low viscosity medium in which the cilia of the respiratory can beat in
a regular and coordinated manner, thereby creating a current for the transport of secretions. The
mucous secretions of the goblet cells and mixed sero-mucous glands of the lamina propria are
deposited on the surface of the periciliary fluid, where they form a continuous layer in the nasal
passages (Proctor and Chang, 1983).
The course of transport by the mucociliary activity of the respiratory epithelium in humans
CNegus, 1958; Proctor and Chang, 1983), monkeys (Lucas. 1932). and rats (Negus. 1958; Morgan et
al., 1986a) are illustrated in figure 7. In both monkeys and humans, the predominant direction of
transport is posterioriy towards the nasopharynx. A small amount of secretion is transported toward
the nostril, but the rate of transport is very slow (1-2 mm per hour in the anterior nasal cavity versus
8-10 mm per minute in the posterior region (Snyder et al.. 1975]) allowing dehydration of the
secretion and eventual removal from the nose through nose blowing or by digital cleaning. In the case
of the rat, a much greater amount of secretion is directed towards the nostril Morgan et al. (1986a)
measured the rate of transport of particles in various parts of the rat nasal cavity. While their
measurements are not directly comparable to those given for humans, the rat did not show such a
great disparity in rates of transport when comparing anterior and posterior directions. It has been
suggested that the anterior mucociliary clearance in the rat allows more rapid transport of paniculate
C-18
-------
ROUTES OF MUCUS FLOW
HUMAN
MONKEY
RAT
Figure 7. Sagittal sections through the heads of human, monkey, and rat to fflustnte the flow of
mucus in the nasal passages. The predominant direction of mucus flow in human and
monkey is toward the posterior of the nasal passage! In rat, a greater amount of mucus flow
is directed anteriorly along the lateral nasal wall and septum toward the ooaril.
C-19
-------
laden secretions to the vestibule where the secretions may be removed by licking or sneezing (Proctor
and Chang. 1983).
'» ' >
It should be,noted that the secretions of the respiratory mucosa form a physiological barrier
which inhaled foreign materials (dust particles, aerosol droplets, vapors, etc.) must traverse before
they can interact with the cells of the nasal mucosa. The nasal secretions are aqueous fluids that
contain a continuous external layer of mucus, which is a glycoprotein. Aqueous soluble, reactive
toxicants are likely to be dissolved in, and to react with, the secretions. Such toxicants are unlikely to
gain access to the nasal mucosae until the potential targets in the nasal secretions are overwhelmed.
Consequently, the large daily volume of nasal secretions are an important pan of the. nasal cavity's
defense mechanism, providing a physical threshold which must be surpassed before exposure can
occur.
The secretions of the olfactory region are serous (watery) and primarily serve as a solvent for
odiferous materials that enables interaction between the material and the chemoreceptors of the
olfactory cells. In contrast to the respiratory region, there are no ciliated cells to rapidly transport the
secretions away from their site of origin, although they follow the same transport pathways described
above. Not surprisingly, the rate of secretion in the olfactory region is not particularly high.
It is interesting to note that the arterial vascular supplies to the olfactory and respiratory
mucosae differ in a way that is consistent with the differences in secretory volumes of the respective
mucosae. The respiratory mucosa, which produces the majority of the nasal secretions, receives the
larger blood supply (via the large calibre sphenopalatine artery). Not only does the respiratory
mucosa receive t greater volume of blood than the olfactory mucosa, but also the superficial
capillaries of the respiratory region are fenestrated, thereby further facilitating transudation of plasma,
which is the source of the peritiliary fluid (Sorokin, 1988).
C-20
-------
Airflow
Due to the complex bony framework surrounding and protruding into the nasal cavity, it is
difficult to describe the airways of the nasal passages and associated structures in words. Despite the
complex folding of the surface of the lateral walls of the nasal passages, several regions have been
described by relating the position of the airway to nearby structures such as the turbinate bones and
nasal septum (Proctor and Chang, 1983). Figure 8 depicts a coronal section through a generalized
human nasal passage to reveal four regions: main (respiratory) airway, olfactory airway, meatuscs,
and paranasal sinuses. The clear regions between the nasal septum and both the lower portion of the
middle turbinate (MT) and the inferior turbinate (IT) are the main airways. The stippled region in the
superior regions are the olfactory airways; the horizontally striped territories are the meatuses. Note
the overlap of olfactory region with meatus in the vicinity of the superior turbinaw (ST). The cross-
hatched regions lateral and superior to the nasal cavity are the paranasal sinuses.
During nasal inspiration in humans, it has been determined that most air traverses a corridor
between the medial edges of the middle and inferior turbinates and the nasal septum (Negus, 1958;
Proctor and Chang, 1983). This corridor corresponds to the main (respiratory) airway in figure 8.
The rate of airflow through both the olfactory airway and the meatuses is considerably less than that
of the main airway. The rate of air exchange through the openings into the paranasal sinuses is so
slight that it can considered to be nonexistent (Negus, 1958).
The regions in the nasal cavities of rats and monkeys are named by similar relationships,
however, differences exist in the ancillary structures. Neither monkeys nor rats possess the sphenoid
sinuses (SpS) and the ethmoid air cells (EC), and the rat possesses a very small maxillary sinus (MS)
(Negus. 1958; Hebel and Stromberg, 1986; Reznik, 1990). During nasal inspiration in the monkey,
air traverses corridors thai are similar to the human (Negus, 1958; Proctor and Chang, 1983). The
situation is more complex in the rat due to the presence of the atrioturbinates and the swell bodies
which deflect inspired air upward and laterally into the meatuses and the olfactory airway.
C-21
-------
Figure 8. Diagram illustrating the regions of the generalized arrangement of airways in the nasal
cavity and ancillary territories. The dear regions are the main airway*. The stippled region
in the superior regions are the olfactory airways; the horizontally striped territories are the
meatuse*. The cross-hatched regions lateral and superior to the nasal cavity are the
paranasal sinuses. EC * ethmoid air cells; IT • inferior turttnate; MS • maxillary sinus;
MT * middle turbtnate; SpS » sphenoid sinuses; ST « superior turbinaie.
C-22
-------
Figure 9 depicts the anterior-posterior routes of airflow through the nasal passages of humans,
monkeys, and rats during normal nasal inspiration. In humans, air enters the nasal vestibule travelling
cephalicaUy and turns approximately 60 degrees as it enters the nasal cavity proper to traverse the
major nasal passage. The air remains close to the floor of the nasal passage between the inferior
turbinates and the nasal septum (respiratory region) until it reaches the pharynx where it strikes the
posterior wall of the nasopharynx, makes a right angle mm infcrioriy, and passes into the larynx and
trachea. In monkeys, air enters the nasal passage in a nearly horizontal orientation. The first major
structure encountered is the anterior end of the middle turbinate, which projects towards the anterior
end of the nose. As in humans, the air traverses the nasal passage close to the floor and between the
middle and inferior turbinates. The air strikes the posterior wall of the nasopharynx and turns sharply
downward to enter the larynx and trachea. In rats, air enters the nose in a nearly horizontal
orientation, but strikes the atrioturbinates just inside the nostril. The atrioturbinates deflect the air
laterally and superiorly so that the air traverses the olfactory region of the nasal cavity including the
lateral meatus and the upper lamellae of the superior and middle turbinates. At the posterior end of
the nasal passage, the air strikes the oblique wall of the respiratory pharynx and makes a mm of about
45 degrees to enter the larynx and trachea.
From the diagrams, it can be predicted that toxicants inspired by humans will first experience a
major interaction with tissues at the posterior of the nasal passages in the nasopharynx. Some
interaction is also likely to occur with the edge of the inferior turbinate as the toxicants traverse the
main airway. In monkeys, the anterior protrusion of the middle turbinate provides the first site of
interaction for toxicants As in the hitman, the nasopharynx will also be a major site for toxicant-
tissue interaction. In the case of the rat, the deflection of the inspired air by the atrioturbinates results
in turbulence close to the entry into the nasal cavity proper. The toxicant-tissue interaction is likely to
occur near the transition from squamous to respiratory epithelium at the region of the lateral meatus.
Recent studies by Morgan et aL (I986b), Heck et at (1989), and Monticello (1990) provide
limited agreement with the predictions of the preceding paragraphs. In these studies, rets and
monkeys were exposed for six hours by inhalation to varying concentrations of 14C-labeUed
C-23
-------
HUMAN
MONKEY
RAT
Figure 9. Diagrams illustrating the major inspirator? flow routes of air during normal (nasal)
breathing in humans, monkeys, and rats. The airflow in monkeys and humans is
predominantly across the lower portions of the nasal passages. In contrast, airflow in the rat
is deflected by the atrkxurbinates to the superior (olfactory) airway.
C-24
-------
formaldehyde, a water soluble toxicant Following exposure regions of the upper respiratory airways
were examined for DNA-protein cross-links as indications for the sites of interaction between tissue
and formaldehyde. Briefly, the results showed that, in the monkey, the major sites of formaldehyde-
tissue interaction were the antefip.&portion of the middle and inferior turbinate, and, to a lesser degree,
the nasopharynx. In the rat, the major site of formaldehyde-tissue interaction was in anterior portion
/
of the inferior turbinate near the lateral meatus. In both species, the amount of formaldehyde-tissue
interaction was a nonlinear function of the exposure concentration. This could be interpreted to mean
that the physical barrier of nasal secretions were sufficient to protect both species from formaldehyde
at low exposure concentrations, but the barrier was overwhelmed at higher concentrations allowing
the interactions to occur.
, Data reported by Torjussenet al. (1979) on histopathological changes in the nasal cavities from
workers in the nickel refining industry also lend credence to the airflow predictions. They reported a
variety of histopathological changes on the anterior curvature of the middle turbinate in active and
retired nickel workers. These changes included nasal carcinoma (3/343) and epithelial dysplasia
(45,343), which are relatively rare among the general population (control incidences were 0/57 and
1/57, respectively).
C-25
-------
REFERENCES CITED
Acheson, E. D., E. H: Hadfield, and R. G. Macbeth, 1967, "Carcinoma of the Nasal Cavity and
Accessory Sinuses in Woodworkers," Lancet, Vol. i, pp. 311-312.
Acheson. E. D.. R. H. Codwell, and B. Jolles, 1970, "Nasal Cancer in the Northampshire Boot and
Shoe Industry." British Medical Journal, Vol. i, pp. 385-393.
Alarie, Y., 1982, "Inhalation Toxicology," a lecture given at the Mid-American Toxicology Course in
Kansas City. MO, April. 1982.
Anderson. J. E.. 1978, Grant's Atlas of Anatomy, 7th edition, Baltimore: The Williams & Wilkins
Company.
Balakrishnan, V., and P. Gangadharan, 1983, "Cancer of the Nasopharynx in Man: Younger Age
Peak and Related Aspects," Nasal Tumors in Animals and Man, G. Reznik and S. F. Stinson, eds.,
Boca Raton: CRC Press, Inc., Vol. l.pp. 187-231.
Bates, D. V., et al., 1966, "Deposition and Retention Models for Internal Dosimetry of the Human
Respiratory Tract," Health Physics, Vol. 12, pp. 173-207.
Blair, A., et al., 1986, "Mortality among Industrial Workers Exposed to Formaldehyde," Journal of
the National Cancer Institute, Vol. 76, pp. 1071-1084.
Bloom, W., and D. W. Fawcetu 1968, A Textbook of Histology, 9th edition, Philadelphia: W. B.
Saunders Company.
Brinton, L. A., et al., 1984. "A Case-control Study of Cancers of the Nasal Cavity and Paranasal
Sinuses", American Journal of Epidemiology, Vol. 119. pp. 896-906.
Calabrese, E. J.. 1983, Principles of Animal Extrapolation, New York: John Wiley &. Sons.
Doll, R.. 1958, "Cancer of the Lung and Nose in Nickel Workers," British Journal of Industrial
Medicine, VoL 15, pp. 217-223.
Edling. C, R HeUquist, and L. Odkvist, 1988, "Occupational Exposure to Formaldehyde and
Histopathotogical Changes in the Nasal Mucosa," British Journal of Industrial Medicine, Vol. 45,
pp. 761-765.
Gross, E. A., et aL. 1982, "Comparative Morphometry of the Nasal Cavity in Rats and Mice," Journal
of Anatomy, Vol. 135, pp. 83-88.
C-26
-------
Guilmene. R. A., J. D. Wicks, and R. K. Wolff, 1989, "Morphometry of Human Nasal Airways In
Vivo Using Magnetic Resonance Imaging," Journal of Aerosol Medicine, Vol. 2, No. 4.
Harkcma, J. R., ct al., 1987, "Nonolfactory Surface Epithelium of the Nasal Cavity of the Bonnet
Monkey: A Morphologic and'Mqrphometric Study of the Transitional and Respiratory Epithelium,"
The American Journal of Anatomy, Vol. 180, pp. 266-279.
Harkema, J. R., T. M. Monticello, and J. A. Hotchkiss, 1989, "Inhaled Toxicant-Induced Proliferative
Responses in Nasal Epithelium of Laboratory Animals," Nasal Carcinogenesis in Rodents:
Relevance to Human Health Risk, V. J. Feron and M. C. Bosland, eds., Amsterdam: Pudoc
Wageningen, pp. 42-47.
Harkema, J. R., 1990. "Comparative Pathology of the Nasal Mucosa in Laboratory Animals Exposed
to Inhaled Irritants," Environmental Health Perspective, Vol. 85, pp. 231-238.
Hebel, R., and M. W. Stromberg. 1986, Anatomy and Embryology of the Laboratory Rat, Worthsee,
Germany: BioMed Veriag.
Heck, H. d'A, et al.. 1989, "Formaldehyde Toxiciry - DNA-Protein Cross-Linking Studies in Rats and
Nonhuman Primates," Nasal Carcinogenesis in Rodents: Relevance to Human Health Risk,
Netherlands: Pudoc Wageningen, pp. 159-164.
Holmstrom, M., et al., 1989, "Histological Changes in the Nasal Mucosa in Persons Occupational!y
Exposed to Formaldehyde Alone and in Combination with Wood Dust," Acta Otolaryngolica
(Stockholm), Vol. 107, pp. 120-129.
Kerns. W. D., et al., 1983, "Carcinogenicity of Formaldehyde in Rats and Mice after Long-term
Inhalation Exposure," Cancer Research, Vol. 43, pp. 4382-4392.
Lucas, A. M., 1932. The Nasal Cavity and Direction of Fluid by Ciliary Movement in Macacus
Rhesus," The American Journal of Anatomy, Vol. 50, pp. 131-177.
Macbeth, R., 1965, "Malignant Disease of the Paranasal Sinuses," Journal of Laryngology, Vol. 79,
pp. 592-612.
Menco, B. Ph. M., 1983, The Ulirastrucrure of Olfactory and Nasal Respiratory Epithelium
Surfaces," Nasal Tumors in Animals and Man, G. Resnik and S. F. Stinson, eds., Boca Raton: CRC
Press, Inc., VoL 1, pp. 45-102.
Monocello, T. M., 1990, "Formaldehyde-Induced Pathology and Cell Proliferation," Dissertation,
Department of Pathology, Duke University, Durham, NC.
C-27
-------
Morgan, K. T., D. L. Patterson, and E. A. Gross, 1986a. "Responses of the Nasal Mucociliary
Apparatus of F-344 Rats to Formaldehyde Gas," Toxicology and Applied Pharmacology, Vol. 82,
pp. 1-13.
'i '\
IV*
Morgan, K. T., et al., 19865, "More Precise Localization of Nasal Tumors Associated with Chronic
Exposure of F-344 Rats to Formaldehyde Gas," Toxicology and Applied Pharmacology, Vol. 82,
pp. 264-271.
Morgan, K. T., and T. M. Monticello, 1990, "Formaldehyde Toxicity: Respiratory Epithelial Injury
and Repair," Biology, Toxicology, and Carcinogens is of Respiratory Epithelium, D. G. Thomassen
and P. Nettesheun, eds., New York: Hemisphere Publishing Corporation, pp. 155-171.
Negus, V., 1958, The Comparative Anatomy and Physiology of the Nose and Paranasal Sinuses,
London: E. & S. Livingstone Ltd.
Patra, A. L., 1986, "Comparative Anatomy of Mammalian Respiratory Tracts: The Nasopharyngeal
Region and the Tracheobronchial Region," Journal of Toxicology of Environmental Health, Vol. 17,
pp. 163-174.
Patra, A. L., A. Gooya, and M. G. Menache, 1986, "A Morphometric Comparison of the
Nasopharyngeal Airway of Laboratory Animals and Humans," The Anatomical Record, Vol. 215,
pp. 42-50.
Proctor, D. F., and J. C. F. Chang, 1983, "Comparative Anatomy and Physiology of the Nasal
Cavity," Nasal Tumors in Animals and Man, G. Resnik and S. F. Stinson, eds., Boca Raton: CRC
Press, Inc., Vol. l.pp. 1-34.
Proctor, D. F., 1989. "Comparative Aspects of Nasal Morphology and Physiology," Nasal
Carcinogenesis in Rodents: Relevance to Human Health Risk, V. J. Feron and M. C. Bosland, eds.,
Amsterdam: Pudoc Wageningen, pp. 5-10.
Reith, A., M. Boysen, and R. Voss, 1988. "Environmental Pathology of the Upper Respiratory Tract,"
Pathology of Head and Neck, D. R. Gnepp, ed.. New York: Churchill Livingstone, pp. 313-333.
Reznik, G. K., 1990, "Comparative Anatomy, Physiology, and Function of the Upper Respiratory
Tract," Environmental Health Perspective, VoL 85. pp. 171-176.
Schreider, J. P., 1986, "Comparative Anatomy and Function of the Nasal Passages," Toxicology of the
Nasal Passages, C. S. Barrow, ed.. New York: Hemisphere Publishing Corporation.
Snyder, W. S., et al., 1975, Report of the Task Group on Reference Man, New Yorfc Pergamon Press.
C-2b
-------
Sorokin, S. P.. 1988, The Respiratory System," Cell and Tissue Biology, 6th edition, L. Weiss, ed.,
Baltimore: Urban & Schwarzenberg, chapter 25, pp. 753-813.
Swenbcrg, J. A., et al.^1980,-Induction of Squamous Cell Cancinomas of the Rat Nasal Cavity by
Inhalation Exposure to Formaldehyde Vapor," Cancer Research, VoL 1, pp. 3398-3402.
Torjussen, W., L. A. Solberg, and A. C. Hogetveit, 1979, "Histopathological Changes of the Nasal
Mucosa in Active and Retired Nickel Workers," British Journal of Cancer, Vol. 40, pp. 568-580.
Universities Associated for Research and Education in Pathology, Inc. (U.AJRJEJ*.), 1982,
"Epidemiology of Chronic Occupational Exposure to Formaldehyde: Report of the (f=7>ad hoc
Panel on Health Aspects of Formaldehyde," Toxicology and Industrial Health, Vol. 4, pp. 77-90.
Warwick. R., and P. L. Williams. .1973, Cray's Anatomy, 35th British edition, Philadelphia: W. B.
Saunders Company.
C-29
------- |