United State* • Science Advisory EPA-SAB-EHC-82-021
Environmental Board (A-101) September 1992
Protection Agency
»EPA AN SAB REPORT:
FORMALDEHYDE RISK
ASSESSMENT UPDATE
REVIEW OF THE OFFICE OF
TOXIC SUBSTANCES'S
DRAFT FORMALDEHYDE
RISK ASSESSMENT UPDATE
BY THE ENVIRONMENTAL
HEALTH COMMITTEE
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
EPA-SAB-EHC-92-021
September 15, 1992
Honorable William K. Reilly
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Subject: Science Advisory Board review of the Office of Toxic Substances draft
Formaldehyde Risk Assessment Update document.
Dear Mr. Reilly:
In 1987, the Office of Toxic Substances examined the non-cancer and cancer
effects associated with formaldehyde exposure (Assessment of Health Risks to
Garment Workers and Certain Home Residents from Exposure to Formaldehyde).
EPA conclusions on the non-cancer effects associated with exposure to
formaldehyde were based mainly upon existing reviews by the National Research
Council, the Consensus Workshop on Formaldehyde, and the Interagency Risk
Management Council. The major non-cancer human health effects posed by inhala-
tion exposure to formaldehyde were attributed to the irritating nature of the chemical.
There was also considerable evidence advanced for the carcinogenicity of inhaled
formaldehyde in animals. EPA also reviewed 28 epidemiologic studies and concluded
that "limited" evidence existed for an association between formaldehyde and human
cancers. The evidence weighted most heavily for an association with cancers of the
upper respiratory tract (nasopharynx, nasal cavity and sinus, and buccal cavity).
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 B1). The epidemiologic studies were considered inadequate for
quantitative risk assessment, so that the quantitative risk assessment of formaldehyde
Pitted on (»p*r in* eontln
• 1MB 7S% ncytMd Gb«
-------
reported in 1987 was based on rat bioassay data in which nasal squamous cell
carcinoma incidence increased with increasing formaldehyde levels in both males and
females.
An updated Formaldehyde Risk Assessment was presented to the Environ-
mental Health Committee for review by the OTS on July 17, 1991. the assessment
incorporated information that had become available since the release of the 1987
report on formaldehyde. The updated document reviews the evidence bearing on both
the cancer and non-cancer effects of inhaled formaldehyde vapor.
The major issues identified for discussion were:
a. The weiqht-of-evidence support for a classification of formaldehyde in
• . - '- v~-v*si. ...--..:/:•''•;•• •
Group 81.
b. Adequacy of the evidence for the use of the nasal DPX (DNA-protein
cross-links) data as a measure of intracellular dose for quantitative risk
assessment.
c. Justification for not using interspecies scaling factors.
d. Support for use of the monkey, rather than the rat DPX
data as dose parameters for human risk estimation.
e. Support for not using the cancer epidemiologic studies for quantitative
risk assessment.
f. Adequacy of the treatment of the non-cancer effects from animal and
epidemiologic studies.
The Committee found the draft Update to be a generally well-written document;
our following comments are intended to aid in making the Update a more balanced,
accessible,and comprehensive document.
The Update reaffirms the 1987 classification of formaldehyde as a Group B1
carcinogen. Animal experimental data are unequivocal, demonstrating in rats that
formaldehyde is a nasal carcinogen. The epidemiological evidence is currently judged
to be less certain than the experimental evidence, largely due to questions of expo-
sure, and the Committee recognizes the basis for denoting the human evidence as
"limited" and applying the Bl classification. Some Committee Members noted,
-------
however, the high relative risk estimates for nasal cancer seen in certain epidemic-
logic studies, and suggestions of a dose-response relationship.
As in 1987, a quantitative risk assessment was derived from the animal data.
The most notable difference between the current document and the 1987 assessment
lies in the increased reliance on a biomarker of formaldehyde exposure rather, than on
ambient chamber concentration, as the source of dose-response information. A
number of questions about this biomarker, DNA-protein cross-links (DPX), were raised
by a panel of experts when its use was first proposed in 1984. Some of the questions
have been resolved by results from a research program undertaken by the Chemical
Industry Institute of Toxicology (CUT). These newer data indicate that the regional
concentrations of DPX serve as surrogates for formaldehyde dose to target cells.
DPX data from monkeys were obtained to provide a species showing greater corre-
spondence with human breathing patterns than do rats. The resulting upper bound
inhalation unit risk (based on the linearized multistage procedure) was calculated as
2.0 x 1CT3 per ppm for the rat data and 3.3 x 1CT4 per ppm for the monkey data. These
values are considerably lower, for both species, than the values based on airborne
exposure concentrations as calculated for the 1987 document and generate several
concerns, as noted below.
The Committee agrees that a risk assessment based on DPX data is a useful
exercise, in that it offers a comparison between risk estimates based on environmental
levels and those based on a biomarker of exposure. It also agrees that the monkey
provides a model more readily extrapolated to humans than does the rat. Questions
still persist, however, about the application of DPX measures to risk assessment,
except as a measure of exposure. For example, the Committee shares the Agency's
concerns about the absence of DPX data based on chronic exposures, and about the
inability to procure information about the correlation between topographical DPX
variations in the monkey and possible tumor sites.
The Committee recognizes the advances in exposure assessment stemming
from DPX measures, and commends the Office of Toxic Substances (OTS) for explor-
ing this approach. At this time, however, the use of DPX measures in quantitative risk
assessment remains equivocal, except as a measure of exposure. The Committee
recommends that the risk estimates based on animal DPX data be compared to those
derived from the most appropriate human studies; several Committee Members
suggested that the subjects followed in the American Cyanamid Corporation studies
(Blair et al, 1986; Marsh, unpublished communication submitted to EPA's Office of
Toxic Substances, t991) might provide such a source. With estimates based on rats,
-------
monkeys, and humans, a revised Update document could more cohesively compare
the advantages and disadvantages of each source and model.
In addition, the Committee believes that the joint effects of particulates and
formaldehyde exposure, which appear to confound the epidemiologic studies, warrant
more extensive discussion. The current draft Update acknowledges this possibility,
but, given the suggestions from the epidemiological data, devotes insufficient attention
to this problem. The importance is of this issue is underlined by experimental studies
of air pollutants which indicate that particulates may serve as efficient carriers for toxic
materials and modify both exposure and pharmacokinetic parameters, and the fact
that exposure to airborne formaldehyde typically occurs in the presence of particu-
lates.
Non-cancer risk assessment was addressed in detail, but the Committee-
recommends that some issues be further expanded. These include sub;c!inicai^ffects,
potentially sensitive subpopulations, full presentations of data, toleranceVdevelopf^nt,
the contributions of particulates and exercise, and methods for the precise psycho-
physical measurement of irritant responses.
In summary, the Committee believes that a revised Update document, address-
ing the issues noted above, would significantly improve the Agency's ability to
quantitate the risks of exposure to formaldehyde, and recommends that such a
revision be undertaken.
We look forward to receiving your response to our comments.
Dr. Raymond Loejir^-Cfiairrnan
Science Advisory Board
Dr. Bernard Weiss, Acting Chairman
Environmental Health Committee
ENCLOSURES
-------
ABSTRACT
An updated draft formaldehyde risk assessment was presented to the Environ-
mental Health Committee on July 17, 1991, incorporating information that had become
available since the 1987 EPA report (Assessment of Health Risks to.Garment Workers
and Certain Home Residents from Exposure to Formaldehyde). The updated docu-
ment reviews the evidence bearing on both the cancer and non-cancer effects of
inhaled formaldehyde vapor. The Committee found the draft update to be a generally
well-written document, but raised issues and provided suggestions on several aspects
of the Update.
The current report reaffirms the 1987 classification of formaldehyde as a Group
B1 (Probable Human) carcinogen. Animal data are unequivocal, demonstrating in rats
.that formaldehyde is a nasal carcinogen. The epidemiological evidence is currently
judged to be less certain than the experimental evidence, primarily because of ;
problems in identifying exposure. Some Committee Members noted, however,'th'e3?'-
high relative risk estimates for nasal cancer in certain epidemiologic studies and
suggestions of a dose-response relationship.
As in 1987, a quantitative risk assessment was derived from animal data, but
there was increased reliance on a biomarker of formaldehyde exposure (DNA-protein
cross-links, or DPX) rather, than on ambient chamber concentration as the source of
dose-response information. DPX data from monkeys were obtained to provide a
species showing greater correspondence with human breathing patterns than do rats.
The resulting upper bound inhalation unit risk (based on the linearized multistage
procedure) was calculated as 2.0 x IO*3 per ppm for the rat data and 3.3 x 10"* per
ppm for the monkey data. These values are considerably lower (for both species)
than the values based on airborne exposure concentrations as calculated for the 1987
document and generate a variety of questions about the application of DPX measures
to risk assessment, except as a measure of exposure. The Committee is concerned
(as is EPA) about the absence of DPX data based on chronic exposures, and about
the inability to procure information about the correlation between topographical DPX
variations in the monkey and possible tumor sites.
The Committee recognizes the advances in exposure assessment stemming
from the use of DPX measures, but views their application to quantitative risk assess-
ment, except as a measure of exposure, as equivocal. The Committee recommends
that the risk estimates based on animal DPX data be compared to those derived from
the most appropriate human studies, particularly on those subjects followed in the
American Cyanamid Corporation studies (Blair et a/, 1986; Marsh, unpublished
communication submitted to EPA's Office of Toxic Substances,. 1991).
In addition, the Committee believes that the joint effects of particulates and
formaldehyde warrant more extensive discussion, since particulates may serve as
-------
efficient carriers for toxic materials and modify both exposure and pharmacokinetic
parameters. Further, the rationale for selecting the monkey model-congruency with
human breathing pattern and respiratory system structure-also invokes the possible
contribution of exercise because it engenders a shift toward oral breathing. Both rats
(which are obligate nose breathers) and monkeys can be induced to exercise, and
consequent shifts in respiration patterns could yield useful new information about the
applicability of DPX.
Non-cancer risk assessment was addressed in detail, but the Committee
recommends that some issues be further expanded. These include subclinical effects,
potentially sensitive subpopulations, full presentations of data, tolerance development,
the contributions of particulates and exercise, and methods for the precise psycho-
physical measurement of irritant responses.
KEYWORDS: Formaldehyde; risk assessment; cancer; DPX.
-------
NOTICE
This report has been written as a part of the activities of the Science Advisory
Board, a public advisory group providing extramural scientific information and advice
to the Administrator and other officials of the Environmental Protection Agency. The
Board is structured to provide balanced, expert assessment of scientific matters
related to problems facing the Agency. This report has not been reviewed for
approval by the Agency and, hence, the contents of this report do not necessarily
represent the views and policies of the Environmental Protection Agency, nor of other
agencies in the Executive Branch of the Federal government, nor does mention of
trade names or commercial products constitute a recommendation for use.
in
-------
ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
ENVIRONMENTAL HEALTH COMMITTEE
Formaldehyde Review Panel, July 17-18, 1991
ACTING CHAIRMAN
Dr. Bernard Weiss, University of Rochester School of Medicine and Dentistry
MEMBERS & CONSULTANTS
Dr. Gary Carlson, School of Pharmacy, Purdue University
Dr. David Gaylor, National Center for Toxicological Research
Dr. Philip Landrigan, Dept-. of Community Medi^.a, i^iount Sinai School of Medtjeine.
-•••••- .,-...„ ' &%*jjgjfr- -
Dr. Nancy K. Kim, New York Department of Health
Dr. Fred Miller, Chemical Industry Institute of Toxicology
Dr. Richard Monson, Harvard University School of Public Health
Dr. Miriam Poirier, National Cancer Institute
Dr. Martha J. Radike, University of Cincinnati Medical Center
Dr. David Wegman, University of Massachusetts at Lowell
Dr. Ronald Wyzga, Electric Power Research Institute
DESIGNATED FEDERAL OFFICIAL
Mr. Samuel Rondberg, Environmental Health Committee, Science Advisory Board,
U.S. Environmental Protection Agency
STAFF SECRETARY
Ms. Mary L. Winston, Science Advisory Board, U.S. Environmental Protection Agency
ASSISTANT DIRECTOR
Mr. Robert Flaak, Science Advisory Board, U.S. Environmental Protection Agency
DIRECTOR
Dr. DonaJd G. Barnes, Science Advisory Board, U.S. Environmental Protection Agency
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 4
2.1 Background 4
2.2. Charge To The Committee 6
3. FINDINGS 7
3.1 Classification of Formaldehyde 7
3.2 Epidemiologic Data And Classification 7
3.3 Use of DPX Qata . 8
3.4 Interspecies Scaling Factors . . . . .:....• : \ . 9
• •.. . 3.5 Use. of Monkey .vs. Rat DPX r ......,...;.... iSy* .?:.1.P'
3.6 "Use of Epidemiologic/Animal Data For Cancer Risk Assessment :^W? "12
3.7 Use of Epidemiologic/Animal Data For Non-Cancer Risk Assessment . 13
4. FINDINGS AND RECOMMENDATIONS 17
5. REFERENCES 20
-------
1. EXECUTIVE SUMMARY
An updated Formaldehyde Risk Assessment was presented to the Environ-
mental Health Committee on July 17, 1991. The assessment incorporated information
that had become available since the release of the 1987 EPA report on formaldehyde
(Assessment of Health Risks to Garment Workers and Certain Home Residents from
Exposure to Formaldehyde). The updated document reviews the evidence bearing on
both the cancer and non-cancer effects of inhaled formaldehyde vapor. The Commit-
tee found the draft Update to be a generally well-written document; our comments, as
expressed in this report, are intended to aid in making the Update a more balanced,
accessible,and comprehensive document.
The current report reaffirms the classificutun of forrnaL^yd^, p^crita^^l^e-
1987 report, as a Group B1 (Probable Human) carcinogen. Animal experimental data
are unequivocal, demonstrating in rats that formaldehyde is a nasal carcinogen. The
epidemiological evidence is currently judged to be less certain than the experimental
evidence. Some Committee Members noted the high relative risk estimates for nasal
cancer in certain studies and suggestions of a dose-response relationship. In addition,
the Committee requested that further attention be given to the analysis and interpreta-
tion of studies involving concurrent formaldehyde and particulate exposures. The
Committee recognizes the basis for denoting the human evidence as "limited" and for
applying the Bl classification.
As was done in 1987, a quantitative risk assessment was derived from animal
data. The most notable difference between the current document and the 1987
assessment lies in the increased reliance on a biomarker of formaldehyde exposure
rather, than on ambient chamber concentration, as the source of dose-response infor-
mation. A number of questions about this biomarker, DNA-protein cross-links (DPX),
were raised by a panel of experts when its use was first proposed in 1984. Some of
the questions have been resolved by results from a research program undertaken by
the Chemical Industry Institute of Toxicology (CUT). These newer data indicate that
the regional concentrations of DPX serve as surrogates for acute formaldehyde dose
to target cells. DPX data from monkeys were obtained to provide a species showing
greater correspondence with human breathing patterns than do rats. The resulting
upper bound inhalation unit risk (based on the linearized multistage procedure) was
calculated as 2.0 x IO"3 per ppm for the rat data and 3.3 x 10"4 per ppm for the monkey
data. These values are considerably lower, for both species, than the values based
1
-------
on airborne exposure concentrations as calculated for the 1987 document and
generate a variety of issues, noted as follows.
The Committee agrees that a risk assessment based on DPX data is a useful
exercise, and that the monkey provides a model more readily extrapolated to humans
than does the rat. Questions still persist, however, about the application of DPX
measures to risk assessment, except as a measure of exposure. For example, the
Committee is concerned (as is EPA) about the absence of DPX data based on chronic
exposures, and about the inability to procure information about the correlation between
topographical DPX variations in the monkey and possible tumor sites.
. The Committee recognizes the advances in exposure assessment stemming
from DPX measures, and commends the Office of Toxic Substances (OTS) forexplor-1
•.-..• *~~«gj- '••
ing this approach. At this time, however, we can not unequivocally advise for or~"
against the use of DPX measures in quantitative risk assessment, except as a
measure of exposure. The Committee recommends that the risk estimates based on
animal DPX data be compared to those derived from the most appropriate human
studies; several Committee Members indicated that the subjects followed in the
American Cyanamid Corporation studies (Blair et at, 1986; Marsh, unpublished
communication submitted to EPA's Office of Toxic Substances, 1991) might provide
such a source. With estimates based on rats, monkeys, and humans, a revised
document could more cohesively compare the advantages and disadvantages of each
source and model.
In addition, the Committee believes that the joint effects of particulates and
formaldehyde warrant more extensive discussion, since exposure to airborne formalde-
hyde typically occurs in the presence of particulates. The current draft Update ac-
knowledges this possibility, but, given the suggestions from the epidemiological data,
devotes insufficient attention to this problem. Its importance is underlined by experi-
mental studies of air pollutants which indicate that particulates may serve as efficient
carriers for toxic materials and modify both exposure and pharmacokinetic parameters,
which for formaldehyde, would occur in the upper, rather than the lower, respiratory
tract. Further, the rationale for selecting the monkey model-congruency with human
breathing pattern and respiratory system structure-also invokes the possible contri-
bution of exercise because it engenders a shift toward oral breathing. Both rats
(which are obligate nose breathers) and monkeys (which are not) can be induced to
-------
exercise, and consequent shifts in respiration patterns could yield useful new informa-
tion about the applicability of DPX.
Non-cancer risk assessment was addressed in detail, but the Committee
recommends that some issues be further expanded. These include subclinical effects,
potentially sensitive subpopulations, full presentations of data, tolerance development,
the contributions of particulates and exercise, and methods for the precise psycho-
physical measurement of irritant responses.
-------
2. INTRODUCTION
2.1 Background
In an earlier document, /Assessment of Health Risks to Garment Workers and
Certain Home Residents from Exposure to Formaldehyde (EPA, 1987), the Office of
Toxic Substances examined the non-cancer and cancer effects associated with
formaldehyde exposure. EPA conclusions on the non-cancer effects associated with
exposure to formaldehyde 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 non-cancer human health effects posed by inhalation
formaldehyde were attributed to the irritating nature of the chemical. 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 asthmatic-like responses, 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. These
changes ranged from rhinitis, epithelial hyperplasia, and squamous metaplasia, to
dysplasia, depending on the duration of exposure, the concentration, and the species
tested.
There was also considerable evidence advanced for the carcinogenicity of
inhaled formaldehyde in animals. This evidence was based on the increased inci-
dence 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.
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 respirato-
ry 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 biologi-
cal explanation for cancers beyond the site of contact (e.g. brain cancer and leukemia)
remained unclear.
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 carcino-
gen (Group B1).
The epidemiologic studies were considered inadequate .for quantitative risjk
assessment. Therefore, the.quantitative riska^bs^-ut of formaldtt^cie
1987 was based on rat bioassay data in which nasal squamous cell carci
incidence was increased with increasing formaldehyde levels in both males and
females.
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 They
believed these data suggested a potential overestimate of risk at low levels of
exposure.
In 1985, the Environmental Protection Agency (EPA), Consumer Product Safety
Commission (CPSC) and National Toxicology Program (NTP) published their reserva-
tions to interpretations of these DPX data made by the Chemical Industry Institute of
Toxicology (CUT); these reservations were upheld by a panel of expert scientists.
These (and 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 reserva-
tions raised by the Agency and it was utilized in the updated assessment brought to
the Environmental Health Committee for review in 1991.
-------
2.2. Charge To The Committee
The issues identified for discussion were:
a. Does the weight-of-evidence support a classification of formaldehyde in
Group 61 (probable human carcinogen)? Is the overall epidemiologic
record adequately summarized in the narrative?
b. Does the presentation on the nasal DNA-protein cross-links (DPX) as a
measure of intracellular dose provide sufficient evidence to favor use of
the DPX data for quantitative risk assessment?
c. Is there sufficient justification for not using interspecies scaling fad^cs?
d. Has the preference for use of the monkey rather than the rat DPX data
as dose parameters for human hsk estimation been adequately ex-
plained?
e. Is there sufficient explanation of the limitations of the cancer epidemio-
logic studies that preclude their use in quantitative risk assessment?
Has the issue of lack of concordance between risk estimates based on
epidemiologic and animal data been adequately addressed?
f. Have the non-cancer effects from animal and epidemiologic studies been
adequately described? Is there sufficient explanation of the limitations of
the epidemiologic studies in quantitative risk assessment?
-------
3. FINDINGS
In general, the Committee believes the direction taken by the risk assessment
update will prove useful, although it will recommend numerous modifications of its
content. Past SAB reviews of various agents have emphasized the desirability of
incorporating as much relevant biological information as possible in risk assessments.
The updated formaldehyde document takes this course by choosing to rely on a selec-
tive biomarker of exposure, DNA-protein cross-links (DPX), rather than on ambient
airborne exposure levels.
3.1 Classification of Formaldehyde ,
The weight of the evidence is consistent with the classification of.formeMis^yeie-
as a B1 (probable human) carcinogen. The animal data are unequivocal, and,
moreover, the tumors are found in an unusual site and correlated with exposure
markers. Formaldehyde cannot be classified as a confirmed (Category A) human
carcinogen because of ambiguity in the epidemiological evidence; for example, the
small number of nasal sinus cancers, which makes estimates of relative risks highly
uncertain, and exposure issues, such as the possible contribution of concurrent
paniculate exposure. Denoting such evidence as "limited" is currently the most appro-
priate description. Therefore, the lower bound of uncertainty in a quantitative risk
model must include zero.
3.2 Epidemiologlc Data And Classification
The lack of sufficient formaldehyde exposure information, coupled with the
possible confounding with simultaneous particulate exposures presents problems in
using the available epidemiologic data for quantitative risk assessment. Although the
Committee is uncertain about how well the epidemiologic data would support a
quantitative risk assessment, we recommend that tables similar to those displayed in
the 1987 document (Figure 7-1 and Table 7-5) be included. These provide estimates
of the lifetime excess risk conditional on the assumption that the epidemiologic data
reflect a causal association. As noted later, these calculations provide a consistency
check for the estimates in the current draft based on DPX. The. Tables should note
that, because of uncertainty in interpretation, the lower bound for each of these
estimates is zero, and the upper bound is based on the upper confidence interval for
attributable risk.
-------
. Perhaps the greatest source of complications in interpretation arise from situa-
tions in which exposure occurs concurrently to particulates (wood dust, pigments) and
formaldehyde. The document should describe more dearly how EPA tried to ascer-
tain possible synergistic effects in these studies. For example, have the available
studies on the association between formaldehyde in the presence of particulates and
cancer been adequately considered? This is an important question because it is now
clear from experimental studies of air pollutants that particulates may serve as efficient
carriers for toxic materials, modifying both exposure and pharmacokinetics parame-
ters.
Another generic question bearing on and discussed in the current report is how
best to characterize exposure: Is peak exposure (what the Update terms dose rate) a
toxicblogicai variable independent of some measure of total (integrated) exposure?
The DPX data indicate that this biomarker, derived from acute exposures, is not
linearly related to exposure concentration. Do studies with other agents yield some
clues? Pharmacokinetics explorations with certain organic solvents, for example,
indicate that non-linearity is common. But such data, too, like the DPX data, are
based largely on acute exposures.
3.3 Use of DPX Data
The use of DNA-protein cross links (DPX) as the basis for formaldehyde risk
assessment is one approach designed to improve estimates of internal tissue expo-
sure that are closer to the biologically effective dose than are measures of external
concentration. Exposure levels described by chamber concentrations are universally
recognized as only gross approximations to the values prevailing at biologically
relevant tissues; this is one of the reasons EPA has emphasized the need for biomar-
kers of exposure.
Despite the progress represented by DPX measures, however, several key
questions remain:
a. Although DPX are offered as a biomarker of exposure, not of carcino-
genic activity, a convincing correlation between DPX concentration based
on chronic exposure in an animal model, and resulting tumor incidence
would strengthen its role as a relevant dose index in risk assessment. It
would be important to understand the special biological or physical prop-
8
-------
erties (fluid dynamics, for example) at the site of the higher tumor rates,
because such understanding may help predict events occurring during
human exposure.
b. Chronicity represents another conundrum. Lifetime exposures yield a
nonlinear dose-response function for nasal tumors in rats. Adduct
analyses for animals exposed chronically were lacking at the time of the
update document. Data on the kinetics of DPX processing (formation,
removal, etc.) would help clarify the validity of DPX in predicting the
consequences of chronic exposures. Because DPX reach steady-state
values in a relatively short time, they do-not:; provide an inde.^.o.f cumula-
tive exposure. Cell proliferation has been adduced as th.e,intervening
•. Step. The Update notes, however, that the proliferative responseis^nly
a partial reflection of DPX distribution. *-*f-:'.T.
c. Further questions have been stimulated by data showing DPX induced,
at least in vitro, by metal species such as chromium VI, and by agents
such as methyl chloride.
d. It may prove useful to conduct an analysis based on total number of
DPX, rather than their concentration in a specified area.
3.4 Interspecies Scaling Factors
In developing risk assessments, the EPA usually applies an interspecies scaling
factor based upon body surface area (equating applied doses in different species as a
function of (mg/kg) w /day). This scaling factor is designed to reflect differences in
exposure at the target organ, and differences in the rates of metabolic activation,
detoxification, and other biological processes associated with toxic mechanisms.
The reasoning behind the EPA's decision not to apply its usual scaling factor is
presented in the Update assessment document (pages 67-68). The argument for
using the DPX measure in place of the surface area correction to equate delivered
dose across species is valid and cogently presented. In addition, such an approach is
consistent with recent initiatives on the part of the Agency to characterize inhalation
exposures, more precisely; they stress the site of contact rather than indirect indices
such as body surface area.
-------
The EPA document, however, does not clearly state how the scaling factor may
be related to other processes involved in carcinogenesis once the carcinogen reaches
the target organ. For example, the document observes that the pace of carcinogene-
sis is quite species dependent. For formaldehyde, tumors in the rat were observed
after 18 months; in humans the latency period could extend for decades. Yet the
temporal differences in DPX measures (and presumably duration of target organ expo-
sure) between monkeys and rats were relatively modest. As for carcinogens in
general, more direct measures of process are needed to account for differences in the
rates of carcinogenesis between species. Because joint application of the usual
interspecies scaling factor and DPX could be interpreted as a double adjustment for
exposure differences across species, one alternative to a choice between the two
methods might be the DPX measure and an inter-species scaling factor less tfran that
usually applied; (e.g., the square root of the usi:al factor). .This is suggested, because
the usual scaling factor is designed to compensate for differences in bioavailability and
for differences in species sensitivity; the DPX seating factor is assumed to compensate
for differences in target organ dose.
3.5 Use of Monkey vs. Rat DPX
The OTS report provides an extensive discussion of the pros and cons of
selecting monkey, rather than rat, DPX data as dose parameters for human risk
estimation. As observed earlier, DPX concentrations may provide a more relevant
exposure marker than nominal chamber concentration. The subsequent step, extrap-
olation to humans, evokes numerous questions about the appropriate tactics vis-a-vis
selecting between rat and monkey DPX data, were they to be used for dose estima-
tion.
The findings which support the use of the monkey-derived data are:
a. Monkeys, as primates, are anatomically and physiologically more closely
related to humans; in particular, the oronasal breathing pattern of mon-
keys resembles that of humans. In contrast, the rat is an obligate nasal
breather.
b. Available data, on formaldehyde-induced tissue changes in the rat indi-
cate that they and DPX are found only in the nasal area; in monkeys,
however, DPX were observed in the nasal tissues, the trachea, and the
10
-------
bronchi. Lesions were observed at the same sites, again providing a
closer correspondence to locations evoking concern about human
cancers. That is, rats exposed to formaldehyde (by inhalation) over an
estimated lifetime showed an elevated cancer incidence only at nasal
sites.
c. The anatomical sites of origin of tumors classified as lung cancer in
epidemiologic studies are the bronchi, which corresponds to the binding
sites in the monkey.
Objections to the arguments above have been raised, however. ^
* ** . 'SirtS*"" '"•'
l^v~-
a. Unlike experiments in the rat, monkey d.fctr,cannot.r-c I'ssckto. £'^$:evv
the correlation between topographical DPX variations anU tumor sites: •
Such experiments, for obvious reasons, are not carried out.
b The quantitative risk assessment based on DPX dosimetry in the monkey
indicates a potential discrepancy with the epidemiologic data. However,
epidemiology can serve as a consistency check for the risk assessments
based on animal data. Most of the time, this Committee is faced with a
situation in which the unit risk calculations based on animal data embody
a rather conservative value compared to available human data. With
formaldehyde, estimates based on animal experiments predict fewer
cancers than those observed in human studies. Furthermore, nasopha-
ryngeal and sino-nasal cancers are relatively rare in the general popu-
lation compared to lung cancer, so that risk figures can be swayed signif-
icantly by even a single case. Given this unease, even if it arises from
disputed interpretations of the human data, the update document should
recognize its source. The document should then explicitly respond by,
for example, indicating how different formulations of a DPX model might
be modified to approximate the epidemiologic findings were the latter
assumed as valid. Would it be feasible to try to model, given what is
known about particle deposition in the respiratory tract, how particulates
such as wood dust, carrying formaldehyde, (cf., Brain and Valberg, 1979)
might modify DPX distribution and formaldehyde carcinogenicity? Such
an exercise would provide a useful focus to the current debate.
11
-------
c. Acute exposure models, as noted earlier, may not be relevant to chronic
exposures. The total pharmacokinetics evaluation, for which DPX is the
first entry, might be modified significantly with extended exposure, as
noted in the Update. Because chronic exposures are the source of
concern about carcinogenic potential, the update document should
discuss more extensively, under a specific heading, the full implications
of this problem. The document's treatment of this problem now appears
in a number of scattered comments.
On balance, despite the reservations above, the Committee tends to agree
with the arguments favoring the monkey DPX data rather than those of the rat.
The update, however, should make certain that readers appreciate the uncertainties.
involved in the unit risk calculations, and that they recognize that DPX serve asxa
measure of exposure rather than as an index of carcinogenesis. The aim of using the
monkey to model human exposure is to characterize, in essence, the effect on DPX of
specified ambient formaldehyde levels so that the risk estimates, derived from
standard models such as the linearized multistage, can be based on a more direct
measure of dose.
3.6 Use of Epldemiologic/Animal Data For Cancer Risk Assessment
The draft OTS Update document does not provide sufficient explanation and
discussion of why the limitations of the available cancer epidemiologic studies should
preclude their use in quantitative risk assessment, nor does it adequately address the
concordance, or lack of concordance, between the epidemiologic and animal data, as
noted earlier. The Committee recognizes that much of this discussion is contained in
the 1987 document. Many readers, however, will come to the Update document
without access to the previous review. Committee members themselves, and even
EPA staff, have experienced difficulties in obtaining earlier documents on various
topics because of restrictions on archiving Agency reports. The update needs to be a
more complete document and to include the more extensive analyses contained in the
1987 report.
As outlined there, and in the Update, the epidemiologic information available to
date is based on relatively small numbers of sino-nasal or naso-pharyngeal cancers.
As a result there is considerable uncertainty as to whether the excesses seen reflect a
causal association with exposure to formaldehyde. In addition, although some studies
12
-------
show an excess of lung cancer (a more common cancer), any excess reported reflects
a weak relative risk. Under these circumstances, considerable uncertainty in the
interpretation of epidemiologic studies is certain to persist. A further complication,
alluded to earlier, is the possible confounding of joint formaldehyde and particulate
exposure, a situation especially prevalent in certain occupational settings, but also
characteristic of home environments, some of which may even exceed occupational
levels. Such confounding may be multiplied by the possible contribution of exercise,
perhaps more likely in occupational settings, and its influence on breathing patterns
and dose parameters. Enhanced oral breathing, a consequence of exercise, elevates
the dose to the lung and probably shifts relative DPX distribution. The monkey DPX
model could be modified by studying exercising animals to yield such information.
In summary, the epidemiologic data should be utilized, at a minimum,
consistency check in any assessment of the risks of exposure to formaldehyde based
on animal data, perhaps by comparison with the 1987 calculations. And, even if EPA
now concludes that a quantitative risk assessment is simply not feasible with the
current human data, an overall point estimate can be made (with the accompanying
wide bounds of uncertainty). The significance of the large difference between
estimates based on animal and epidemiologic data is softened considerably by the
uncertainty attached to the latter, the possibility of modification by simultaneous
exposure to particulates, and by the uncertainties attached to the carcinogenic
significance of the DPX measures themselves.
3.7 Use of Epidemiologic/Animal Data For Non-Cancer Risk Assessment
The Committee reviewed the OTS document and identified several concerns
with the section on non-cancer effects.
Evaluation of the update led to the recommendation that the section on non-
cancer risk assessment be revised, taking the following issues into consideration:
a. The current discussion treats non-cancer effects as if they were health
endpoints, but subsequently minimizes the importance of some of them.
Acute respiratory symptoms (e.g., irritation) and chronic respiratory
symptoms (e.g., cough, wheeze) affect quality of life, and, in addition,
may be indicators of developing disease. Moreover, abnormal values
(for example in FEV,) may be indicators of disease risk without being
13
-------
defined as conventional evidence of disease. The document's discussion
of these issues, however, implies that they are of little importance
because they are not sufficiently aberrant to be considered clinically
significant. The Committee recommends that the document be
revised so that the potential importance of these effects is dis-
cussed more broadly, and cogently. EPA's position on hazardous air
pollutants stands in contrast to the current evaluation.
b. A related concern is that the current discussion emphasizes mean
values.. As a practical result of study design and analysis in available
published reports, the assessment is based largely on studies of working
- populations, medical students, arid controlled exposures.with healthy
1 adults. This makes examination of variability in these largely youog iand
healthy study groups all the more important in estimating exposure
effects for members of more sensitive sub-populations. Extreme re-
sponses in such populations, noted in other contexts, may yield clues to
the response of particularly susceptible populations. The Committee
recommends that additional information on population variability in
responsiveness be added.
c. Most of the presentation of non-cancer effects does not include display
of the actual results from the studies, but rather only whether the results
were (statistically) significantly different. This practice has two problems.
First, it places undue importance on statistical testing without addressing
(1) the preponderance of the evidence, (2) consistency across studies,
and (3) indications of trends. Second, it does not provide the reader with
adequate detail, even at the level of a summary of findings. The Com-
mittee recommends that additional information addressing this
concern be added. In addition, the health implications of tolerance
to the irritant effects of formaldehyde should be addressed. Toler-
ance is not an index of risk reduction.
d. The summary of the data in each of the sub-sections in section 6.2 is too
discursive. There is no attempt to present the strengths and weakness-
es of the study designs, population sizes, outcome measures, and data
analyses. We recommend that additional data be added to address
this concern and that the studies before 1987 should be Included as
14
-------
well, because, as noted previously, the earlier document may be
unavailable to many readers.
e. At least one of the Alexandersson studies (1989) is prospective and not
identified as such. The benefits of this approach in this study should be
incorporated in the evaluation of pulmonary function.,
f. One problem is the focus on effects uniquely attributable to formalde-
hyde. Concurrent exposure to other agents, such as particulates,
frequently occurs in almost any. occupational study. Ignoring the effects
of formaldehyde in the presence of particulates addresses ansunrealistic-
circumstance. Only in an exposure chamber will formaldehyde oxpp-. .-.--.
sures be free of particulates. In tio general- environment (indoor o|,. ...
outdoor) particulates will accompany any exposure. The Committee
recommends further attempts to interpret study results in terms of
the independent and combined effects of formaldehyde and particu-
lates, rather than to treat particulate exposure simply as a "con-
founder." Moreover, as noted in Section 3.6, the additional contri-
bution of exercise, and how it modifies both dose and breathing
pattern, needs to be considered.
g. There is a lack of consistency between the discussion of detailed expo-
sure estimates as they relate to non-cancer effects, and the parallel
discussion related to cancer effects. Much more detailed exposure data
are available for non-cancer effects. The discussion of the overall
evidence suggests that no epidemiologic study could ever provide
enough exposure-effect information to "precisely quantify general popula-
tion risks for eye and upper respiratory effects...". This is too conserva-
tive a reading of the existing data.
h. The scientific community would benefit if this risk assessment were to
identify issues needing investigation. One particular concern is the very
minimal attention placed, as yet, on the use of psychophysical measures
of irritant responses. The only report attempting to apply modem mea-
sures of subjective response is that of Horvath, et al., 1988 (.cited in
document). This lack of attention to an important research tool
15
-------
should be noted and the Committee recommends that a section be
added listing research needs.
i. The exposure descriptions for both the animal and human epidemiologic
studies should be discussed in sufficient detail to enable the reader to
evaluate better the data and their applicability for risk assessment.
j. The explanation of both the advantages and the limitations of the epide-
miologic studies for quantitative risk assessment needs to be expanded,
both in the text and in the executive summary. The reasons why the
studies may not show an effect when one may exist and why the studies
may be showing an effect that may not be related to formaldehyde,need
to be expanded. Furthermore, it is important to estimate the relajjye -
likelihood that findings from the epidemiologic studies are misleading.
The executive summary also needs a brief statement qualifying the
findings. The previous paragraphs suggest specific items that need to
be added.
16
-------
4. FINDINGS AND RECOMMENDATIONS
It is difficult to read the current document which updates the 1987 risk assess-
ment without a summary of the documentation available at that time. An evaluation of
the literature for both cancer and non-cancer effects, and how they relate to the
respective risk assessment, cannot be made from the June, 1991 Update document
alone. A careful summary of the earlier findings would be a useful and important
addition to the Final Draft risk assessment. The Committee recommends that such
a summary be added.
The current document proposes the classification of formaldehyde, as did the
1987 report, as a Group Bl (Probable Human) carcinogen. The primaryevidence for
human carcinogenicity, the reported association between cancers of the nasopharynx
and buccal cavity and formaldehyde exposure, infrequently complicated by simulta-
neous exposure to particulate matter. The Committee agrees that Hie human evi-
dence is "limited" and supports the Bl classification.
The current update places considerable emphasis on the use of a biomarker of
formaldehyde exposure, DPX, rather than on ambient chamber concentration as the
source of dose-response information. A number of questions about this biomarker,
DNA-protein cross-links (DPX), were raised by a panel of experts when it was first
proposed in 1984. A program to resolve these reservations was subsequently
undertaken by the Chemical Industry Institute of Toxicology (CUT), and has met most
of them to the satisfaction of EPA. These newer data indicate that regional concentra-
tions of DPX may be able to serve as surrogates for formaldehyde dose to target
cells. In addition, because rats are obligate nose-breathers, DPX data from monkeys
were obtained to provide comparisons with species showing greater correspondence
with human breathing patterns. The upper bound inhalation unit risk based on the
linearized multistage procedure was calculated as 2.8 x IO*3 per ppm for the rat and
3.3 x 10"4 per ppm for the monkey. These values.in the case of both species, are
much lower than the values based on airborne exposure concentrations as calculated
for the 1987 document.
The Committee views a risk assessment based on DPX data as a useful
comparison with risk estimates based on environmental concentration and with
the epidemioiogical data; for this purpose, the monkey provides a more suitable
model than the rat. The Committee is concerned.as is EPA, about the absence
17
-------
of DPX data for chronic exposures, and about the inability to obtain information
about the correlation between topographical DPX variations in the monkey and
possible tumor sites.
The Committee recognizes the advances in exposure assessment stem-
ming from DPX measures, and commends the OTS for pursuing this course. At
this time, however, we can not provide an unequivocal answer to item (b) of the
Charge, namely the use of DPX data for quantitative risk assessment. There are
simply too many issues as yet unresolved, to offer a definitive recommendation. This
position is based on the following observations:
«•
a. . The presence of DPX may be taken as an index of acute formaldehyde
exposure. In the rat, the relationship between formaldehyde concentra-
tion and DPX shows a clear non-linearity. Whether the form of this
relationship will persist during chronic exposure conditions remains to be
settled,
b. Definitive information demonstrating that DPX are relevant to carcinogen-
esis is not available.
c. Under these circumstances, DPX continue to serve as a useful interme-
diate step pending further explorations of chronic exposures and their
connection to carcinogenesis.
We recommend that the OTS continue to pursue this issue, and suggest
comparing risk estimates based on DPX data to those derived from the most
appropriate human studies; several Committee Members indicated that the subjects
followed in the American Cyanamid Corporation studies might contribute important
new epidemiological information. With estimates based on rats, monkeys, and
humans, a revised assessment document could more cohesively compare the
advantages and disadvantages of each source and model. The Committee is aware
of the assumptions about exposure and other variables required of such a treatment
of the human data, but believes that the resulting document will attain more cogency
by such an effort.
In addition, the Committee believes that the joint effects of particulates and
formaldehyde warrant more extensive discussion, since exposure to airborne formalde-
18
-------
hyde typically, occurs in the presence of participates. The current draft update ac-
knowledges the possibility of interactions, but, given the suggestions from the epidemi-
ological data, inadequate attention is directed to possible joint effects. Particulates are
so ubiquitous that a "pure" formaldehyde exposure situation may be impossible to find
in humans. The importance of joint effects is reinforced by experimental studies of air
pollutants that demonstrate particulates may serve as efficient carriers for toxic
materials, modifying both exposure and pharmacokinetic parameters. In the case of
formaldehyde, the upper, rather than the lower, respiratory tract would be the source
of concern. Further, the rationale for selecting the monkey model-congruency with
human breathing pattern and respiratory system structure-also should invoke the
possible contribution of exercise because it engenders a shift tov/ard oral .breathing,
Both rats and monkeys can be induced to exer^se, aiitj consequent shifts iivr^sRirar
tion patterns could yield useful new information about t! Vc applicability pfDPX; . .
While acknowledging that the state-of-the-art is a rapidly moving target, the
Committee nevertheless advocates that future revisions of the document incorporate
the latest information. In particular, the updated Cyanamid analyses (currently
unpublished update of the Blair, et al. 1986 study by Marsh, 1991), recent work on
airflow patterns in rat and monkey nasal passages (Morgan et al., 1991), exercise
effects on dose and dose distribution (Kleinman and Mautz, 1991), and on occupation-
al risk factors for sinonasal cancer in woodworkers (Luce et al., 1992).
Non-cancer risk assessment was addressed in detail by the Update, but
the Committee recommends that some issues be further expanded. These
include subclinical effects, potentially sensitive subpopulations, full presenta-
tions of data, tolerance development, the contributions of particulates and
exercise, and methods for the precise measurement of irritant responses.
In closing, the Committee recommends that the Agency revise the Update to
address the issues noted above. These revisions would enhance significantly EPA's
ability to quantitate the risks of exposure to formaldehyde.
19
-------
5. REFERENCES
Alexandersson R, Hedenstierna G. 1989 Pulmonary function in wood workers exposed to
formaldehyde: A prospective study. Arch. Environ. Health 44:5-11.
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.
76:10071-1084.
Brain, J. D., and Valberg, P.A. Deposition of aerosol in the respiratory tract. Amer. Rev.
Respirat. Dis. 120:1325-1373, 1979.
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-Schmitz, M, Starr, TB, Heck Hd'A. 1984 Differentiation between metabolic
incorporation and covalent binding in the labeling of macromolecules in the rat nasal mucosa
and bone marrow by induced u C arid 3 H formaldehyde. Toxicol. App. Phramacol. 76:26-44
Consensus Workshop on Formaldehyde. 1984. Report on the Consensus Workshop on
Formaldehyde. Environ. Health Persp. 58: 323-381
ERMC (Interagency Risk Management Council), 1984. Draft Report of the Risk Assessment
Subgroup of the IRMC-Formaldehyde Workgroup.
Kleinman, M.T., Mautz, W.J. The effects of exercise on dose and dose distribution on inhaled
automotive pollutants. Health Effects Institute, Research report Number 45, 1991.
Luce, D, et al. Occupational risk factors for sinonasal cancer: a case-control study in France.
Amer J. Occup. Med., 21:163-175, 1992.
Marsh, G. Unpublished report on lung cancer rates and formaldehyde exposure. Submitted to
U.S. EPA Office of Toxic Substances May 3, 1991 (8EHQ-0591-0924).
Morgan, K.T. et al. Studies of inspiratory airflow patterns in the nasal passages of the F344
rats and rhesus monkey using nasal molds: relevance to formaldehyde toxicity. Tox. Appl.
Pharmacol. 110:223-240, 1991
National research Council, Committee on Aldehydes, Board on Toxicology and Environmental
Health Hazards. 1981. Formaldehyde and other aldehydes. Washington DC, National
Academy Press.
20
-------
U.S. Environmental Protection Agency, Office of Toxic Substances. Assessment of health
risks to garment workers and certain home residents from exposure to formaldehyde.
Washington, D.C., 1987
U.S. Environmental Protection Agency, Office of Toxic Substances. Formaldehyde Risk
Assessment Update (Final Draft) Washington D.C., June 11, 1991.
21
-------
Distribution List
Administrator
Deputy Administrator
Assistant Administrators
Deputy Assistant Administrator for Research and Development
Director, Office of Pollution Prevention and Toxics
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
------- |