Content uploaded by Kenneth Mundt
Author content
All content in this area was uploaded by Kenneth Mundt on Feb 16, 2018
Content may be subject to copyright.
Contents lists available at ScienceDirect
Regulatory Toxicology and Pharmacology
journal homepage: www.elsevier.com/locate/yrtph
Commentary
Six years after the NRC review of EPA's Draft IRIS Toxicological Review of
Formaldehyde: Regulatory implications of new science in evaluating
formaldehyde leukemogenicity
Kenneth A. Mundt
a,∗
, P. Robinan Gentry
a
, Linda D. Dell
a
, Joseph V. Rodricks
a
, Paolo Boffetta
b
a
Environment and Health, Ramboll Environ, Amherst MA, United States
b
Icahn School of Medicine at Mount Sinai, New York, NY, USA
ARTICLE INFO
Keywords:
Regulatory science
Hazard evaluation
Evidence integration
Epidemiology
Toxicology
Mechanistic studies
ABSTRACT
Shortly after the International Agency for Research on Cancer (IARC) determined that formaldehyde causes
leukemia, the United States Environmental Protection Agency (EPA) released its Draft IRIS Toxicological Review
of Formaldehyde (“Draft IRIS Assessment”), also concluding that formaldehyde causes leukemia. Peer review of
the Draft IRIS Assessment by a National Academy of Science committee noted that “causal determinations are
not supported by the narrative provided in the draft”(NRC 2011). They offered recommendations for improving
the Draft IRIS assessment and identified several important research gaps. Over the six years since the NRC peer
review, significant new science has been published. We identify and summarize key recommendations made by
NRC and map them to this new science, including extended analysis of epidemiological studies, updates of
earlier occupational cohort studies, toxicological experiments using a sensitive mouse strain, mechanistic studies
examining the role of exogenous versus endogenous formaldehyde in bone marrow, and several critical reviews.
With few exceptions, new findings are consistently negative, and integration of all available evidence challenges
the earlier conclusions that formaldehyde causes leukemia. Given formaldehyde's commercial importance, en-
vironmental ubiquity and endogenous production, accurate hazard classification and risk evaluation of whether
exposure to formaldehyde from occupational, residential and consumer products causes leukemia are critical.
1. Introduction
Classification and regulation of human carcinogens is a key com-
ponent to the protection and improvement of public health. However,
proper regulation of industrial chemicals hinges on both valid hazard
identification and quantitative risk assessment. Increasingly, hazard
identification –at least where adequate scientific evidence is available –
draws on critically assessing and integrating evidence across lines of
inquiry including animal and human toxicology (e.g., pharmacokinetic,
mechanistic studies) and epidemiology. Quantitative risk assessment
requires reasonably accurate characterization of exposure, which is
complicated, especially where historical measures are sparse or do not
exist. Where adequate evidence from some or all of these is lacking, and
where important uncertainties remain, policy-driven approaches fa-
voring precaution are warranted. On the other hand, as evidence ac-
cumulates, more science-focused methods can be employed, reducing
uncertainties, leading to sounder conclusions. Nevertheless, confident
conclusions are sometimes drawn prematurely, as discussed in this
commentary. Recent evaluations of formaldehyde, coupled with
improved critical review and evidence integration expectations and
new, more focused scientific evaluations, illustrate the dynamic nature
of scientific inquiry, the need for parallel refinement of hazard char-
acterization, and subsequently, stronger risk assessment.
In this paper, we illustrate the evolution of new scientific evidence
on formaldehyde as a potential human leukemogen. The impetus for the
new science summarized below is derived from the International
Agency for Research on Cancer's (IARC) 2009 classification of for-
maldehyde as a known cause of leukemia in Monograph 100F (Baan
et al., 2009; IARC, 2012), the US Environmental Protection Agency's
(EPA's) similar classification in the Draft IRIS (Integrated Risk Informa-
tion System) Toxicological Review of Formaldehyde –Inhalation Assessment
(hereafter referred to as “Draft IRIS Assessment”)(EPA, 2010), and the
criticisms and recommendations presented in two National Academy of
Science (NAS), National Research Council (NRC) expert reviews –one
on the Draft IRIS Assessment and one on the IRIS process itself (NRC,
2011; NRC, 2014a). Various organizations and agencies have con-
tributed to or sponsored the new science, including governments and
universities, as well as industry. In revising and finalizing the Draft IRIS
https://doi.org/10.1016/j.yrtph.2017.11.006
Received 7 April 2017; Received in revised form 27 October 2017; Accepted 15 November 2017
∗
Corresponding author.
E-mail address: kmundt@ramboll.com (K.A. Mundt).
Regulatory Toxicology and Pharmacology 92 (2018) 472–490
Available online 20 November 2017
0273-2300/ © 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T
Table 1
Summary of major formaldehyde carcinogenicity classifications and noted scientific basis.
Year Agency Carcinogenicity Classification Findings
1981 NTP (1981) Anticipated to be a human
carcinogen
Epidemiological evidence. Not discussed
Toxicological evidence. One study cited (Swenberg et al., 1980).
Nasal cancers: “While a full evaluation of the carcinogenicity of formaldehyde
vapor must await completion of studies at the Chemical Industry Institute of
Toxicology, evidence presented to date demonstrates that inhalation of
formaldehyde results in a high incidence of nasal cancers in rats (Swenberg et al.,
1980).”
1981
a
IARC (1982a; b) Possibly carcinogenic to humans
(Group 2B)
Epidemiological evidence. Inadequate (6 epidemiology studies)
Toxicological evidence. Sufficient, formaldehyde is carcinogenic to rat, causes
nasal cancers.
1982 NTP (1982) Anticipated to be a human
carcinogen
Epidemiological evidence. Inadequate (cites IARC, 1982a; b)
Toxicological evidence. Sufficient, formaldehyde is carcinogenic to two strains of
rats.
Nasal cancers. One test in mice did not produce statistically significant results.
Other studies in animals (mice and hamsters by inhalation exposure) were
considered inadequate for evaluation.
1987
b
IARC (1987) Probably carcinogenic to humans
(Group 2A)
Epidemiological evidence. Limited
Nasal cancers: Reported epidemiological evidence is strongest for nasal and
nasopharyngeal cancer, noted limitations with small numbers of exposed cases and
inconsistent reports.
Leukemia: “Excess mortality from leukemia and cancer of the brain was generally
not seen among industrial workers, which suggests that the excess for these cancers
among professionals is due to conditions other than formaldehyde. The slight
excesses of cancer among professionals noted in several studies generally did not
display the patterns of increasing risk with various measures of exposure (i.e.,
latency, duration, level, or cumulative) usually seen for occupational carcinogens.
No other cancer showed a consistent excess across the various studies.”
Toxicological evidence. Sufficient
No changes in information reported from IARC (1982b)
Supporting data.
“In single studies of persons exposed to formaldehyde, increases in the frequencies
of chromosomal aberrations and sister chromatid exchanges in peripheral
lymphocytes have been reported, but negative results have also been published.
The interpretation of both the positive and negative studies is difficult due to the
small number of subjects studied and inconsistencies in the findings (IARC, Suppl
6, 1987).”
1991 EPA (1991) Probable human carcinogen
(Group B1)
Epidemiological evidence.Limited (28 studies considered)
Nasal cancers: “Human data include nine studies that show statistically significant
associations between site-specific respiratory neoplasms and exposure to
formaldehyde or formaldehyde-containing products.”(p.7)
Leukemia: “Analysis of the remaining 19 studies indicate that leukemia and
neoplasms of the brain and colon may be associated with formaldehyde exposure.
The biological support for such postulates, however, has not yet been
demonstrated.”(p. 8)
Toxicological evidence.Sufficient, nasal squamous cell carcinomas
Increased incidence of nasal squamous cell carcinomas observed in rats and mice in
long-term inhalation studies.
Supporting data. “The classification is supported by in vitro genotoxicity data and
formaldehyde's structural relationships to other carcinogenic aldehydes such as
acetaldehyde.”(p. 7)
1994
c
IARC (1995) Probably carcinogenic to humans
(Group 2A)
Epidemiological evidence. Limited
Nasal cancers: Lack of consistency between cohort and case-control studies of
cancers of the nasal cavities and paranasal sinuses.
Leukemia: “The studies of industrial cohorts also showed low or no risk for
lymphatic or hematopoietic cancers; however, the cohort studies of embalmers,
anatomists and other professionals who use formaldehyde tended to show excess
risks for cancers of the brain, although they were based on small numbers. These
findings are countered by a consistent lack of excess risk for brain cancer in the
studies of industrial cohorts, which generally included more direct and quantitative
estimates of exposure to formaldehyde than did the cohort studies of embalmers
and anatomists.”(p. 334)
Toxicological evidence. Sufficient (nasal squamous cell carcinomas)
Squamous cell carcinomas of nasal cavities, at highest exposure. No evidence of
carcinogenicity in hamsters. Mice showed no effect or were inadequate for
evaluation.
Supporting data. Genotoxic in variety of experimental systems in vivo. Induced
DNA-protein cross-links, DNA single-strand breaks, chromosomal aberrations,
sister chromatid exchange, gene mutation in human and rodent cells in vitro.
(continued on next page)
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
473
Table 1 (continued)
Year Agency Carcinogenicity Classification Findings
2004
d
IARC (2006) Carcinogenic to humans (Group 1) Epidemiological evidence. Sufficient, based on nasopharyngeal cancer
Leukemia: “There is strong but not sufficient evidence for a causal association
between leukemia and occupational exposure to formaldehyde. Increased risk for
leukemia hasconsistently been observed in studies of professionalworkers and in two
of three of the most relevant studies of industrial workers. These findings fall slightly
short of being fully persuasive because of some limitations in the findings from the
cohorts of industrial and garment workers in the USA and because they conflict with
the non-positive findings from the British cohort of industrial workers.”(p.276)
Toxicological evidence. Sufficient (nasal squamous cell carcinoma)
Supporting data. Mechanism for inducing myeloid leukema is not known. Possible
mechanisms considered included clastogenic damage to circulatory stem cells.
“The Working Group was not aware of any good rodent models that simulate the
occurrence of acute myeloid leukemia in humans. Therefore, on the basis of the data
available at this time, it was not possible to identify a mechanism for the induction of
myeloid leukemia in humans.”(p. 280)
2009
e
IARC (2012) Carcinogenic to humans (Group 1) Epidemiological evidence. Formaldehyde causes cancer of the nasopharynx and
leukemia.
“The Working Group was not in full agreement on the evaluation of formaldehyde
causing leukemia in humans, with a small majority viewing the evidence as sufficient
of carcinogenicity and the minority viewing the evidence as limited.”(p. 430)
Toxicological evidence.
“Studies of bone marrow cells in formaldehyde-exposed animals have been
inconsistent.”(p.427) “Pancytopenia has not been among the haematological
findings in experiments with laboratory animals exposed to relatively high doses of
formaldehyde, including classic long-term safety assessment studies.”(p.428)
Inconsistent genotoxic effects in blood lymphocytes from animals exposed to
formaldehyde via inhalation.
Supporting data. “Particularly relevant to the discussions regarding sufficient
evidence was a recent study accepted for publication which, for the first time,
reported aneuploidy in blood of exposed workers characteristic of myeloid
leukeaemia and myelodysplastic syndromes, with supporting information suggesting
a decreased in the major circulating blood-cell types and in circulating
haematological prescursor cells. The authors and Working Group felt that this study
needed to be replicated.”(p. 430)
“Three possible mechanisms, all focused around genotoxicity, are moderately
supported as the underlying mechanism for induction of haematological
malignancies in humans. Further research is needed to decide which of the
mechanisms is the most important.”(p. 430)
2010 Draft IRIS Assessment (EPA, 2010) Carcinogenic to humans Epidemiological evidence.Sufficient. “Human epidemiological evidence is
sufficient to conclude a causal association between formaldehyde exposure and
nasopharyngeal cancer, nasal and paranasal cancer, all leukemias, ML and
lymphohematopoietic cancers as a group”(p. 6–46).
All LHM combined: “Given the consistency and strength of the positive associations
for all LHP [lymphohematopoietic] cancer mortality in professional cohorts
(embalmers, anatomists and pathologists) taken together with the strong positive
results of the NCI cohort, human epidemiologic evidence are [sic] sufficient to
conclude that there is a causal association between formaldehyde exposure and
mortality from all LHP malignancies (as a group.)”(p. 4–180).
All leukemias as a group: “While the epidemiologic evidence for a causal
association between formaldehyde and all leukemia as a group is not at [sic] strong
as for all LHP as a group, the repeated identification of an association in multiple
meta-analyses taken together with the clear causal association between myeloid
leukemia demonstrated by Hauptmann et al. (2009) and the consistent evidence
reported by Beane Freeman et al. (2009) are sufficient to conclude that there is a
causal association between formaldehyde exposure and mortality from all
leukemia as a group.”(p. 4–182)
Myleoid leukemia: “Given the consistency of the positive associations for
formaldehyde with myeloid leukemia cancer mortality across five of the six studies
(Hauptmann et al., 2009; Pinkerton et al., 2003; Hayes et al., 1990; Stroup et al.,
1986; Walrath and Fraumeni, 1984, 1983; but not Beane Freeman et al., 2009), the
statistically significant meta-analysis by Zhang et al. (2009) and the convincing
results from Hauptmann et al. (2009), the human epidemiologic evidence is
sufficient to conclude that there is a causal association between formaldehyde
exposure and mortality from myeloid leukemia.”(p. 4–185)
Toxicological evidence. Limited evidence to support conclusion that
formaldehyde exposure causes leukemia. Four studies evaluated the leukemic
potential of formaldehyde.
“Inhalation exposure of formaldehyde increased lymphoma in female mice and
leukemia in female F344 rats, but not male rats (Battelle Columbus Laboratories,
1981). No increases in leukemia or lymphoma were seen in male Wistar rats when
exposed to formaldehyde in drinking water (Til et al., 1989) or male rats after
chronic inhalation exposures (Sellakumar et al., 1985).”(p. 6–21)
Supporting data.“Chromosomal damage in blood-borne immune cells, relevant to
agent-induced lymphohematopoietic cancers has been coumented in formaldehyde
exposed workers, including increased micronuclei and chromosomal aberrations,
increased incidence and aneuploidy in hematopoietic stem cells.”(p. 6–22)
(continued on next page)
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
474
Table 1 (continued)
Year Agency Carcinogenicity Classification Findings
2012 NTP (2011) Known to be a human carcinogen Epidemiological evidence. Causes nasopharyngeal cancer, sinonasal cancer, and
myeloid leukemia
“Epidemiological studies have demonstrated a causal relationship between
exposure to formaldehyde and cancer in humans. Causality is indicated by
consistent findings of increased risks of nasopharyngeal cancer, sinonasal cancer,
and lymphohematopoietic cancer, specifically myeloid leukemia among
individuals with higher measures of exposure to formaldehyde (exposure level or
duration), which cannot be explained by chance, bias, or confounding. The
evidence for nasopharyngeal cancer is somewhat stronger than that for myeloid
leukemia.”(p. 195)
Toxicological evidence. No specific evidence cited regarding leukemia beyond
the following: “Hemolymphoreticular tumor (combined types) in rats of both sexes
also were significantly increased after long-term exposure of adults; however, it is
unclear whether these turmos were exposure-related, because of limitations in the
reporting of these tumors (Soffritti et al., 2002).”(p. 198)
Supporting data.“Lymphohematopoietic cancers are a heterogeneous group of
cancers that arise from damage to stem cells during hematopoietic and lymphoid
development (Greaves, 2004). Blood cells arise from a common stem cell, which
forms two progenitor cells, the common myeloid stem cell and the common
lymphoid stem cell. Most agents known to cause leukemia are thought to do so by
directly damaging stem cells in the bone marrow. In order for a stem cell to become
malignant, it must acquire genetic mutations and genomic instability (Zhang et al.,
2010a). Because formaldehyde is highly reactive and rapidly metabolized, a key
question is how it can reach the bone marrow or cause toxicity or genotoxicity at
distal sites. The endogenous concentration in the blood of humans, monkeys, and
rats is about 2–3μg/g, and the concentration does not increase after inhalation of
formaldehyde from exogenous sources (Heck et al., 1985; Casanova et al., 1988;
Heck and Casanova, 2004). Moreover, N2-hydroxymethyl-dG–DNA adducts have
not been detected at distal sites in rats (such as the bone marrow, white blood cells,
lung, spleen, liver, or thymus) (Lu et al., 2010). For these reasons, the plausibility
of formaldehyde's causing cancer at distal sites, such as myeloid leukemia, has been
questioned (Golden et al., 2006; Pyatt et al., 2008).
However, systemic effects have been observed after inhalation or oral exposure,
and although the mechanisms by which formaldehyde causes myeloid leukemia in
humans are not known, a number of plausible mechanisms have been advanced.
These include (1) theoretical mechanisms for the distribution of formaldehyde to
distal sites and (2) proposed mechanisms of leukemogenesis that do not require
formaldehyde to reach the bone marrow. In addition, there is some evidence that
formaldehyde causes adverse haematological effects in humans.”(p. 199)
2012 RAC (2012) Carc. 1B - H50
f
May cause cancer
Epidemiological evidence. Limited
“In conclusion, while some studies have found increased rates of leukemia, the
epidemiology data do not show consistent findings across studies for leukemia
rates. The inconsistent findings across job types and exposure groupings, and the
lack of biological plausibility argue against formaldehyde as the cause of the
increased rates. The findings of slightly increased leukemia rates among
embalmers, pathologist and anatomists, but not among industrial workers, suggests
the possibility of confounding factors that bear investigation. Results based on
cohort and case-control studies do not suggest an association between
formaldehyde exposure and leukemia.”(p.41)
Toxicological evidence.“No indication of carcinogenic potential on organs/
tissues distant from the site of contact (respiratory tract) including
lymphohaematopoietic tumors in inhalation study of rats and mice (Kerns et al.,
1983).”(p.22)
Supporting data.“Physiologically, formaldehyde occurs in most organisms,
tissues and cells at very low concentrations. In mammals, formaldehyde is found at
values of about 0.1 mM in blood (man, monkey, rat). The physiological blood
formaldehyde levels in humans, rats and monkeys were not elevated after
parenteral exposure, indicating a very low systemic tissue and organ distribution of
formaldehyde. These findings support evidence that formaldehyde shows local
reactivity and elicits its toxic potential focally and predominantly at deposition
areas such as epithelia of the upper respiratory tract, the oro-gastric tract as well as
the skin. (BfR-Wissenschaft, 2006). Thus, it may be expected that carcinogenic
effects are not found at anatomical sites distant from the port of entry.”(p.44)
(continued on next page)
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
475
Assessment (EPA, 2010), EPA now has the opportunity to incorporate
the new evidence in addressing many of the issues raised by the NRC
reviews.
2. Formaldehyde cancer hazard evaluation
The carcinogenicity of formaldehyde has been evaluated by several
agencies since the early 1980s, including the IARC, the National
Toxicology Program (NTP) of the National Institute for Environmental
Health Sciences (NIEHS), the EPA, and most recently, the Committee
for Risk Assessment (RAC) of the European Chemicals Agency (ECHA),
and the Scientific Committee on Occupational Exposure Limits (SCOEL)
of the European Commission (Table 1). Except for the RAC review
(RAC, 2012) and the SCOEL review (Bolt et al., 2016), which re-
classified formaldehyde as a Carcinogen Category 1B (i.e., presumed to
have carcinogenic potential for humans) and a Category C carcinogen
(i.e., genotoxic carcinogen with a mode of action based threshold),
respectively, these reviews classified formaldehyde as a known human
carcinogen, primarily based on NPC but also on lymphohematopoietic
malignancies (LHM) as a group and/or all leukemias as a group, and all
myeloid leukemias (ML) as a group (EPA, 2010; IARC, 2012; NTP,
2011). Differences between NTP (2011) and EPA draft classifications
(final version of the EPA review is pending) have been highlighted by
Rhomberg (2015a) and differences between the IARC (2012) and the
RAC (RAC, 2012) evaluations have been discussed by Marsh et al.
(2014).
The reviews by authoritative bodies acknowledged that hazard
identification for formaldehyde was not straightforward, especially
with respect to possible leukemogenicity, in part due to its endogenous
production and high reactivity. This prompted closer scrutiny regarding
the methods used to critically evaluate the strength and quality of sci-
entific studies, and ultimately, how best to integrate evidence across
lines of inquiry such as animal, mechanistic and epidemiological eva-
luations.
IARC first classified formaldehyde as “carcinogenic to humans”(i.e.,
Group 1) in 2005 (Cogliano et al., 2005; IARC, 2006), revising the
previous evaluation in 1995 that formaldehyde is “probably carcino-
genic to humans”(i.e., Group 2A) (Table 1). The 2005 evaluation
(Cogliano et al., 2005; IARC, 2006) concluded that formaldehyde
causes NPC, based primarily on results from animal studies, with ad-
ditional evidence from “the largest and most informative cohort study
of industrial workers”(i.e., Hauptmann, et al., 2004). Results from
animal studies demonstrated that formaldehyde in direct contact with
nasal passage tissues induced tumors at formaldehyde concentra-
tions > 2 parts per million (ppm) as summarized by Nielsen et al.
(2013) and later by Nielsen et al. (2017). This was considered con-
sistent with formaldehyde's demonstrated genotoxicity, and with the
“sufficient epidemiological evidence that formaldehyde causes naso-
pharyngeal cancer in humans”(IARC, 2006).
IARC (2012) concluded that formaldehyde also causes leukemia,
and in particular ML, although the Working Group noted that it was a
“small majority”who found the evidence to be sufficient. Neither
Hauptmann et al. (2003) nor the subsequently updated study (Beane
Freeman et al., 2009) published results specifically for acute myeloid
leukemia (AML). The Working Group noted a study reporting aneu-
ploidy in the blood of exposed workers (Zhang et al., 2010a), recently
accepted for publication, provided supporting data, with the caveat that
the study needed to be replicated (IARC, 2012). Indeed, proper re-
plication of this study is still needed, because the study protocol was not
consistent with adequate cell counting standards, including the authors’
earlier descriptions of the OctoChrome FISH method (Zhang et al.,
2005; Zhang et al., 2011) and other standards (American Society of
Medical Genetics, 2006). One particular challenge is that occupational
exposure limits in North America, Europe and in many countries around
the world protect workers from the levels of occupational formaldehyde
exposures that were studied by Zhang et al. (2010a) in China making
replication of the study logistically difficult. Proper replication of this
study also will require use of methods to successfully distinguish be-
tween aneuploidy arising in vivo from aneuploidy that arises during the
period of in vitro culture, as discussed in section 3.3.3 below.
Following the IARC review and classification, the National
Toxicology Program (NTP) concluded in the 12th Report on
Carcinogens (12th RoC) that formaldehyde causes nasopharyngeal
cancer and myeloid leukemia (NTP, 2011)(Table 1). The 12th RoC
stated “The most informative studies for evaluation of the risk of ML are
the large cohort studies of industrial workers (the NCI, NIOSH, and
Table 1 (continued)
Year Agency Carcinogenicity Classification Findings
2016 Scientific Committee on Occupational
Exposure Limits for Formaldehyde (Bolt
et al., 2016)
Carcinogen Group C
(genotoxic carcinogen with a mode-
of-action based threshold)
Epidemiological evidence. Limited.
Leukemias: “A possible induction of myeloid leukaemias by FA in humans is not so
easy to explain, but there are indications that FA might induce this kind of
malignancy. However, this would require that FA would act systemically and reach
the bone marrow, which is the target tissue. Such an action would not be possible
within a range where the external dose does not change the physiological level of
FA.”(p.45)
Toxicological Evidence. “In essence, new experimental data, reported since 2008,
clearly indicate that systemic genotoxic action of inhaled FA is not likely, even at
exposure concentrations leading to nasal malignancies in the rat.”(p.49)
Supporting Data.“A plethora of arguments suggests that FA concentrations below
1 or 2 ppm would not increase the risk of cancer in the nose or any other tissue, or
affect FA homeostasis within epithelial cells (Swenberg et al., 2013).“(p. 49)
a
IARC Working Group met February 1981. IARC Preamble (1982): “For many of the chemicals evaluated in the first 29 vol of the/ARC Monographs for which there is sufficient
evidence of carcinogenicity in animals, data relating to carcinogenicity for humans are either insufficient or nonexistent. In the absence of adequate data on humans, it is reasonable, for
practical purposes, to regard chemicals for which there is sufficient evidence of carcinogenicity in animals as if they presented a carcinogenic risk to humans. The use of the expressions
'for practical purposes' and 'as if they presented a carcinogenic risk' indicates that at the present time a correlation between carcinogenicity in animals and possible human risk cannot be
made on a purely scientific basis, but only pragmatically. Such a pragmatical correlation may be useful to regulatory agencies in making decisions related to the primary prevention of
cancer.”
b
IARC Working Group met March 1987.
c
IARC Working Group met October 1994; monograph published 1995.
d
IARC Working Group met June 2004; monograph published 2006.
e
IARC Working Group met October 2009; monograph published 2012.
f
EU harmonized classification and labelling.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
476
British cohorts) and the NCI nested case-control study
1
of lymphohe-
matopoietic cancer in embalmers”and specifically that “Three of these
four studies found elevated risks of myeloid leukemia among in-
dividuals with high exposure to formaldehyde, as well as positive ex-
posure-response relationships”. However, the NTP also noted “In the
large cohort of British chemical workers, no increased risk of leukemia
was found for formaldehyde exposure”and that in the only case-control
study examining ML (Blair et al., 2000)“an excess risk was found for
chronic (but not acute) myeloid leukemia”(NTP, RoC, 12th edition,
“Formaldehyde”, p.3).
2.1. Environmental Protection Agency integrated risk assessment program
(IRIS)
Formaldehyde had been classified by the EPA as a “probable”
human carcinogen (Group B1) in 1991 (Table 1). An updated assess-
ment for public review and comment was first released in June 2010,
12 years after the EPA announced the re-evaluation, and the draft as-
sessment reported that formaldehyde causes NPC, nasal and paranasal
cancer, lymphohematopoietic cancers, all leukemias, and ML (Table 1).
The EPA (2010) also derived a draft inhalation unit risk (IUR) value of
8.1 × 10
−2
per ppm (6.6 × 10
−5
per μg/m
3
)
2
based on the upper
bound on the sum of the risk estimates for NPC, Hodgkin lymphoma,
and leukemia (combined risks) based on part of the results reported in
Beane Freeman et al. (2009). For rationale, the EPA said the classifi-
cation “is supported by cohort analyses of embalmers, pathologists and
anatomists (Hall et al., 1991; Hayes et al., 1990; Levine et al., 1984;
Matanoski, 1989; Stroup et al., 1986; Walrath and Fraumeni, 1983,
1984)”despite the observation that “… SMR analyses of the large in-
dustrial cohorts do not indicate a similar association (Beane Freeman
et al., 2009; Coggon et al., 2003; Pinkerton et al., 2004)”(EPA, 2010;
page 4–180). The EPA also cited three meta-analyses (Bosetti et al.,
2008; Collins and Lineker, 2004; Zhang et al., 2009) that largely in-
cluded the same studies as providing additional evidence. Repeatedly
reporting the same results, however, does not constitute independent or
additional evidence. Similarly, all meta-analyses included earlier ver-
sions of the NCI cohort workers and embalmers studies and therefore,
the meta-analyses, too, are redundant with the updated analyses of the
NCI cohort workers and embalmers studies.
The conclusions in the Draft IRIS Assessment specific to myeloid
leukemia are as follows:
“Given the consistency of the positive associations for formaldehyde
with myeloid leukemia cancer mortality across five of the six studies
(Hauptmann et al., 2009; Hayes et al., 1990; Pinkerton et al., 2004;
Stroup et al., 1986; Walrath and Fraumeni 1983, 1984; but not
Beane Freeman et al., 2009), the statistically significant meta-ana-
lysis by Zhang et al. (2009) and the convincing results from
Hauptmann et al. (2009), the human epidemiologic evidence is
sufficient to conclude that there is a causal association between
formaldehyde exposure and mortality from myeloid leukemia.”
(EPA, 2010; pages 4–184, 4–185)
Again, because of the significant overlap between Hauptmann et al.
(2009) and the three PMR studies of funeral directors and embalmers
(Hayes et al., 1990; Walrath and Fraumeni, 1983; 1984) these reports
do not constitute independent evidence or consistency across studies.
Hauptmann et al. (2009) has been judged to have severe methodolo-
gical flaws (Cole et al., 2010a; b). Separately, the Zhang et al. (2009)
meta-analysis combined different exposure metrics (peak, average in-
tensity, cumulative exposure, duration), and thus, the exposure metrics
were not comparable across studies. A more methodologically rigorous
approach would be to perform meta-analyses for similar exposure me-
trics, that is, a meta-RR for cumulative exposure, meta-RR for average
exposure, meta-RR for duration of exposure (only one study reported
results in relation to peak exposure, precluding a meta-analysis for peak
exposure). As such, the Zhang et al. (2009) meta-analysis results are
difficult to interpret and methodologically flawed. Finally, combining
data in a meta-analyses does not overcome any systematic biases in the
underlying studies (Greenland and Longnecker, 1992).
2.2. National academies peer-review process
The NRC of the NAS, at the request of the EPA, formed an expert
Committee to perform the peer-review of the Draft IRIS Assessment.
Following a series of meetings during the second half of 2010, the NRC
issued the final peer-review report on April 8, 2011 (NRC, 2011)asa
pre-publication copy. The Committee identified numerous constructive
criticisms and data gaps, and provided recommendations for improving
IRIS reviews in general (NRC, 2011). Though not directly charged to
evaluate the Draft IRIS Assessment conclusions, the peer review raised
important questions regarding the underlying methods giving rise to
several conclusions, including the basic causal conclusions:
“EPA evaluated the evidence of a causal relationship between for-
maldehyde exposure and several groupings of LHP cancers—“all
LHP cancers,”“all leukemias,”and “myeloid leukemias.”The com-
mittee does not support the grouping of “all LHP cancers”because it
combines many diverse cancers that are not closely related in
etiology and cells of origin. The committee recommends that EPA
focus on the most specific diagnoses available in the epidemiologic
data, such as acute myeloblastic leukemia, chronic lymphocytic
leukemia, and specific lymphomas.”(NRC, 2011; page 11)
The Committee concluded that EPA's claims that formaldehyde
causes leukemia, ML or related hematopoietic cancers were not sup-
ported in EPA's assessment, appeared to be subjective in nature, and
that no clear scientific framework had been applied by EPA in reaching
that conclusion (NRC, 2011). The absence of such a framework was
judged by the committee as problematic:
“As with the respiratory tract cancers, the draft IRIS assessment does
not provide a clear framework for causal determinations. As a result,
the conclusions appear to be based on a subjective view of the
overall data, and the absence of a causal framework for these can-
cers is particularly problematic given the inconsistencies in the
epidemiologic data, the weak animal data, and the lack of me-
chanistic data. Although EPA provided an exhaustive description of
the studies and speculated extensively on possible modes of action,
the causal determinations are not supported by the narrative pro-
vided in the draft IRIS assessment. Accordingly, the committee re-
commends that EPA revisit arguments that support determinations
of causality for specific LHP cancers and in so doing include detailed
descriptions of the criteria that were used to weigh evidence and
assess causality. That will add needed transparency and validity to
its conclusions.”(NRC, 2011; page 11)
The NRC peer review further pointed out that the EPA (2010)
conclusion that formaldehyde causes ML was based primarily on se-
lected epidemiological studies, and other streams of evidence (animal,
mode of action) were not considered beyond studies conducted by
Zhang et al. (2009, 2010a).
In the 7th and final chapter of its review, entitled, “A Roadmap for
Revision,”the NRC provided recommendations in two categories:
“Critical Revisions of the Current Draft IRIS Assessment of
1
This study technically is not a “nested case-control study”but rather a pooled re-
analysis of death certificate data from several published proportionate mortality ratio
(PMR) analyses, using a case-control approach. Thus, it carries the same limitations of
death certificate analyses performed outside of a well enumerated cohort, and therefore is
not “nested”in any true cohort that could be accurately enumerated.
2
This is 15 times higher than the inhalation unit risk (IUR) derived by EPA for vinyl
chloride (4.4 × 10
−6
per μg/m
3
)(EPA, 2000; page 50), a chemical for which the evi-
dence clearly supports a causal association between exposure and effects in both animals
and humans.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
477
Formaldehyde,”and “Future Assessments and the IRIS Process”(NRC,
2011). NRC (2011) specifically identified the systematic review stan-
dards adopted by the Institute of Medicine (IOM), as being appropriate
for such an analysis (IOM, 2011).
Following the release of the NRC (2011) peer review, Congress is-
sued House Report No. 112–151 (US U.S. House, 2011), and directed
EPA to incorporate recommendations of Chapter 7 of the NRC (2011)
peer-review report into the IRIS process. In 2014, NRC released an
additional report on the IRIS process (NRC, 2014a), and emphasized the
importance of evidence integration for hazard identification, in which
studies of higher quality and low risk of bias are given greater weight in
drawing conclusions regarding causality.
As part of their response to the NRC reviews, the EPA convened a
state-of-the-science workshop on formaldehyde on April 30 and May 1,
2014 in Arlington, Virginia. This workshop focused on three themes:
•Evidence pertaining to the influence of formaldehyde that is pro-
duced endogenously (by the body during normal biological pro-
cesses) on the toxicity of inhaled formaldehyde, and implications for
the health assessment;
•Mechanistic evidence relevant to formaldehyde inhalation exposure
and lymphohematopoietic cancers (leukemia and lymphomas); and
•Epidemiological research examining the potential association be-
tween formaldehyde exposure and lymphohematopoietic cancers
(leukemia and lymphomas).
(From: https://www.epa.gov/iris/formaldehyde-workshop)
A second workshop was announced at the meeting but never con-
vened. Since then, the EPA submitted a progress report to Congress in
2015 (EPA, 2015) in response to a request from Congress (U.S. House,
2014, p. 59). Most recently, House Report No. 114–632 (U.S. House,
2016; page 57–59) and Senate Report No. 114–281 (U.S Senate, 2016;
page 62) have requested the allocation of funds for NRC to peer review
the revised IRIS Toxicological Review of Formaldehyde, to ensure that
recommendations raised by the NRC (2011) were implemented.
3. New studies published since the 2011 NRC peer review of the
draft IRIS assessment
Numerous studies and updated analyses have been published since
the 2011 NRC peer review of the Draft IRIS Assessment, the findings of
which, at least in part, fill many of the “data gaps”and address several
key methodological issues highlighted in the NRC Committee re-
commendations (NRC, 2011). Below we summarize this new research,
organized around the data streams (e.g., epidemiological, toxicological,
and mode of action) for evidence integration and quantification of
potential leukemia risks, specifically responsive to the following NRC
recommendations (2011) (page reference provided):
•Epidemiological Evidence
•Discussion of the specific strengths, weaknesses and incon-
sistencies in several key studies, as the draft IRIS assessment relies
solely on epidemiologic studies to determine causality. (p.113)
•Clarification of the basis of the EPA's interpretations of the Beane
Freeman et al. (2009) results regarding the various dose metrics
(peak versus cumulative) and the various LHP cancers. (p.113)
•Evaluation of the most specific diagnoses available in the epide-
miologic data (i.e., acute myeloblastic leukemia, chronic lym-
phocytic leukemia, and other specific lymphomas). (p. 113)
•Toxicological Evidence
•Paucity of evidence of formaldehyde-induced LHP cancers in an-
imal models. EPA's unpublished re-analysis of the Battelle chronic
experiments in mice and rats (Battelle Columbus Laboratories,
1981), although intriguing, provides the only positive findings
and thus does not contribute to the weight of evidence of caus-
ality. (p.110)
•Mode of Action Evidence
•Improving the understanding of when exogenous formaldehyde
exposure appreciably alters normal endogenous formaldehyde
concentrations. (p. 58)
•Reconciliation of divergent statements regarding systemic de-
livery of formaldehyde, (p.59) as direct evidence of systemic de-
livery of formaldehyde is generally lacking. (p.5)
•Data are insufficient to conclude definitively that formaldehyde is
causing cytogenetic effects at distant sites. (p. 5)
•Dose-Response Assessment
•Independent analyses of the dose-response models to confirm the
degree to which the models fit the data appropriately. (p. 14)
•Consideration of the use of alternative extrapolation models for
the analysis of the cancer data. (p.14)
•Further justification of the selection and use of the NCI cohort
(Beane Freeman et al., 2009) for calculation of unit risk because
the cumulative exposure metric (used in the calculation of unit
risk) was not related to leukemia risk in the NCI cohort. (p.112)
•Methods for Evidence Integration
•Development of an approach to weight of evidence that includes
“a single integrative step after assessing all of the individual lines
of evidence”. Although a synthesis and summary are provided, the
process that EPA used to weigh different lines of evidence and
how that evidence was integrated into a final conclusion are not
apparent in the draft assessment and should be made clear in the
final version. (p. 113)
A summary of each of these recommendations and data gaps, along
with the new science that has been conducted to address them is pro-
vided in Table 2 and discussed in the following sections.
3.1. Epidemiological evidence
The NRC peer review called attention to the EPA's sole reliance on
epidemiological studies to determine causality, rather than integrating
epidemiology data with the toxicological and mechanistic evidence.
When inferring causation from epidemiology studies, the evidence is
critically assessed and synthesized across a body of individual studies,
with greater weight assigned to studies of higher quality (rather than
assigning equal weight to each). Better epidemiological studies are
those that implement individual level exposure data, and minimize the
potential for systematic bias and confounding. The ascertainment of
outcome and analysis using accurate (and specific) diagnosis are also
critical in the causal evaluation. The NRC peer review noted that the
grouping of “all LHPs”comprises 14 biologically distinct diagnoses in
humans and should not be used in determinations of causality. There is
some evidence that these diseases may originate from the same stem
cell line (Gluzman et al., 2015; Goldstein, 2010) and could therefore
arise from direct effects on these cells. There are no studies, however,
that demonstrate an effect on these stem cells following exposure to
formaldehyde. The largest population of these stem cells would be
found in the bone marrow, and, based on the available evidence, in-
haled formaldehyde appears incapable of reaching the bone marrow
(see Section 3.3.2). The affected cells would need to be circulating stem
cells that encounter formaldehyde at the portal of entry (i.e., the nose or
upper airways) and then return to the bone marrow.
After the NRC peer review was published, Checkoway et al. (2012)
critically reviewed the epidemiological evidence and reported incon-
sistent and sporadic associations between formaldehyde exposure and
various specific LHM, including ML. Only a few epidemiology studies
considered AML specifically. Since the critical review (Checkoway
et al., 2012), several additional epidemiological studies have been
published that provide insights on formaldehyde exposure and AML risk
and address other specific issues raised by the 2011 NRC peer review.
The key strengths and limitations of these studies are highlighted
below.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
478
Table 2
Summary of NRC (2011) comments or identified data gaps and new formaldehyde science by lines of inquiry.
NRC (2011) Comment/Identified Data Gap New Formaldehyde Science
A. Epidemiological Evidence
Evaluation of the most specific diagnoses available in the epidemiologic data
(i.e., acute myeloblastic leukemia, chronic lymphocytic leukemia, and other
specific lymphomas). (NRC, p. 113)
•New analyses of the NCI formaldehyde workers cohort specifically for AML are reported.
Results do not support the hypothesis that formaldehyde causes AML. See:Checkoway
et al., 2015
•Associations seen between formaldehyde exposure and Hodgkin lymphoma and chronic
myeloid leukemia (CML) have not been observed in other studies and are not considered
plausible. See:Checkoway et al., 2015
Because the draft IRIS assessment relies solely on epidemiologic studies to
determine causality, further discussion of the specific strengths, weaknesses,
and inconsistencies in several key studies is needed. (NRC, p. 113)
•A critical review of the epidemiological literature indicated no consistent or strong
epidemiologic evidence that formaldehyde is causally related to any lymphohematopoetic
malignancies. The absence of established toxicological mechanisms further weakens any
arguments for causation. See:Checkoway et al., 2012
Clarification of the basis of its interpretations of the results regarding the various
dose metrics (peak versus cumulative) and the various LHP cancers. (NRC, p.
112–113)
•Acute myeloid leukemia (AML) was unrelated to cumulative, average or peak exposure,
and few deaths occurred within 20 or more years of last peak exposure. Suggestive
associations with peak exposure were observed for chronic myeloid leukemia, based on
very small numbers. Hodgkin lymphoma relative risk estimates suggested trends for both
cumulative (p
trend
= 0.05) and peak (p
trend
= 0.003) exposures. However, no other
lymphohematopoietic malignancy was associated with either cumulative or peak
exposure. See:Checkoway et al., 2015
The selection and use of the NCI cohort (Beane Freeman et al., 2009) should be
further justified. (NRC, p. 112)
•Extended follow-up of a cohort of 14,008 chemical workers at 6 factories in England and
Wales, covering the period 1941–2012. Results provide no support for an increased hazard
of myeloid leukemia from formaldehyde exposure. See:Coggon et al., 2014
•Extended follow-up of 11,098 employees of three garment manufacturing facilities. Results
demonstrated limited evidence for formaldehyde exposure and any LHM including AML,
based on 14 observed cases. See:Meyers et al., 2013
B. Toxicological Evidence
Paucity of evidence of formaldehyde-induced LHP cancers in animal models.
EPA's unpublished re-analysis of the Battelle chronic experiments in mice
and rats (Battelle Columbus Laboratories, 1981), although intriguing,
provides the only positive findings and thus does not contribute to the
weight of evidence of causality. (NRC, p. 110)
•No cases of leukemia or lymphohematopoietic neoplasia were seen. FA inhalation did not
cause leukemia in genetically predisposed C3B6·129F1-Trp53tm1Brd mice. See:Morgan
et al., 2017
•FA inhalation did not cause leukemia or lymphohematopoietic neoplasia in genetically
predisposed p53-Haploinsufficient mice. See:Morgan et al., 2017
C. Mode of Action Evidence
Improve understanding of when exogenous formaldehyde exposure appreciably
alters normal endogenous formaldehyde concentrations. (NRC, p. 58)
•Endogenous formaldehyde in nasal tissues did not significantly affect flux or nasal uptake
predictions at exposure concentrations > 500 ppb; however, reduced nasal uptake was
predicted at lower exposure concentrations. See:Schroeter et al. (2014)
•With the application of highly sensitive instruments and accurate assays, inhaled
formaldehyde was found to reach nasal respiratory epithelium, but not other tissues distant
to the site of initial contact. In contrast, endogenous adducts were readily detected in all
tissues examined with remarkably higher amounts present. Moreover, the amounts of
exogenous formaldehyde-induced adducts were 3- to 8-fold and 5- to 11-fold lower than the
average amounts of endogenous formaldehyde-induced adducts in rat and monkey nasal
respiratory epithelium, respectively. See:Yu et al., 2015
Reconcile divergent statements regarding systemic delivery of formaldehyde
(p.59); direct evidence of systemic delivery of formaldehyde is generally
lacking. (NRC, p.5)
•Based on a sensitive analytical method that can measure endogenous versus exogenous
formaldehyde DNA adducts, the multiple studies demonstrated that inhaled exogenous
formaldehyde only reached rat or monkey noses, but not tissues distant to the site of initial
contact. Also, new evidence suggests that endogenous formaldehyde in bone marrow is
toxic and carcinogenic, and may cause leukemia (but not exogenous formaldehyde). See:
Lai et al., 2016; Pontel et al., 2015; Yu et al., 2015; Edrissi et al., 2013; Moeller et al.,
2011; Lu et al., 2011
Data are insufficient to conclude definitively that formaldehyde is causing
cytogenetic effects at distant sites. (NRC, p. 5)
•Critical review of the genotoxicity literature found no convincing evidence that exogenous
exposures to FA alone, and by inhalation, induce mutations at sites distant from the portal
of entry tissue as a direct DNA reactive mutagenic effect –specifically, not in the bone
marrow. Review of the existing studies of hematotoxicity, likewise, failed to demonstrate
myelotoxicity in any species–a probable prerequisite for leukemogenesis. See:Albertini
and Kaden, 2016
•Reanalysis of selected raw data from the Zhang et al. (2010a) study do not support a causal
association between formaldehyde and myeloid leukemia or lymphoid malignancies.
Because of the significant methodological limitations, unless the results can be confirmed
using appropriate methodologies designed to detect in vivo events, the reanalysis of the
results provided by Zhang et al. (2010a) raise sufficient questions that limit the use of Zhang
et al. (2010a) to support the hypothesis that formaldehyde exposure is causally related to
leukemia or lymphoid malignancies. See:Gentry et al. (2013)
•Additional analyses were performed on the study data obtained from the original study
(Zhang et al., 2010a) including individual average formaldehyde exposure concentration
measurements performed for each exposed worker. The objective was to evaluate
haematological parameters and aneuploidy in relation to quantitative exposure measures of
formaldehyde. Results showed that differences in white blood cell, granulocyte, platelet, and
red blood cell counts were not exposure-dependent. Furthermore, among formaldehyde-
exposed workers, no association was observed between individual average formaldehyde
exposure estimates and frequency of aneuploidy, suggested by the original study authors to
be indicators of myeloid leukemia risk. See:Mundt et al., 2017
D. Dose-Response Assessment
Independent analysis of the dose-response models is needed to confirm the
degree to which the models fit the data appropriately. (NRC, p. 14)
•The documentation of the methods applied in the Draft IRIS Assessment (EPA, 2010) lacks
sufficient detail for duplication of the unit risk estimates provided, even with the
availability of the raw data from the NCI cohort study (Beane Freeman et al., 2009). This
(continued on next page)
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
479
3.1.1. Key studies and their strengths and limitations
Since the update of mortality in the US formaldehyde users and
producers cohort (Beane Freeman et al., 2009), two other large in-
dustrywide cohort mortality studies have been updated: the NIOSH
garment workers (Meyers et al., 2013) and the UK industry-wide for-
maldehyde producers and users (Coggon et al., 2014). In addition, a
large population registry-based case-control study of incident AML
cases in the Nordic countries, a small occupational study in Italy and a
large multicenter European study of occupational exposures in a cohort
established to study nutritional and metabolic risk factors in cancer
risks have been published (Pira et al., 2014; Saberi Hosnijeh et al. 2013;
Talibov et al., 2014).
3.1.1.1. NIOSH cohort study of garment workers.Meyers et al. (2013)
updated mortality from 1960 through 2008 for 11,043 US garment
workers exposed to formaldehyde who worked for at least three months
between 1955 and 1983 at three US factories. A total of 36 leukemia
deaths was reported (SMR = 1.04, 95% CI 0.73–1.44, compared to US
mortality rates), of which 21 were ML (14 AML, 5 chronic myeloid
leukemia (CML), 2 other and unspecified ML). Although this study did
not link quantitative estimates of formaldehyde exposure to study
subjects, an industrial hygiene survey during the early 1980s reported
that formaldehyde concentrations were similar across all departments
and facilities, and the overall geometric mean was 0.15 ppm with a
geometric standard deviation of 1.90 (Stayner et al., 1988). The
formaldehyde resins used to treat permanent press fabrics had been
reformulated over time, and as a result, the formaldehyde
concentrations measured in the early 1980s were believed to be
lower than the approximately 4 ppm estimated by NRC for years
prior to 1970 (NRC, 2014b). Meyers et al. (2013) reported an SMR for
AML of 1.22 (95% CI 0.67–2.05), noting that NIOSH investigators
“continue to see limited evidence of an association between
formaldehyde and leukemia”and that “the extended follow-up did
not strengthen previously observed associations.”All 14 AML deaths
occurred 20 or more years after first exposure to formaldehyde. The
NIOSH study is a large cohort with adequate follow up but limited
industrial hygiene measurements of historical formaldehyde
concentrations, as most workers were first exposed prior to 1970.
Therefore, the study did not assign individual estimates of cumulative
or peak exposure, and analyses for mortality due to various LHM
including AML were performed using duration of exposure as a proxy
for cumulative exposure. Information on smoking was also lacking.
3.1.1.2. Registry-based case control study of AML in Nordic
countries.Talibov et al. (2014) analyzed 15,332 incident cases of
AML diagnosed in Finland, Norway, Sweden, and Iceland from 1961
to 2005. The investigators matched 76,660 controls to cases by year of
birth, sex, and country. Job titles and dates of assignment were linked
to a job-exposure matrix (JEM) to estimate quantitative exposure to 26
workplace agents, including formaldehyde. No association was seen
between risk of AML and increasing cumulative exposure to
formaldehyde, after adjusting for exposure to solvents (aliphatic and
alicyclic hydrocarbon solvents, benzene, toluene, trichloroethylene,
methylene chloride, perchloroethylene, other organic solvents) and
radiation (hazard ratio (HR) 0.89, 95% CI 0.81–0.97 for workers
exposed to ≤0.171 ppm-years; HR 0.92, 95% CI 0.83–1.03 for
workers exposed to 0.171–1.6 ppm-yrs, and HR 1.17, 95% CI
0.91–1.51 for > 1.6 ppm-years, compared to workers not exposed to
formaldehyde). The strengths of this study were its exposure assessment
based on a validatedJEM and the comprehensive ascertainment of
incident AML cases (i.e., not deaths), resulting in high statistical power
to detect increased risks, avoidance of survival bias, and the ability to
consider and control for other possible leukemogens. One major
limitation is the lack of data on smoking, which also is known to
cause leukemia. This study failed to find an association between
benzene and AML; however, increased risk of AML may be limited to
those with exposure to very high concentrations that historically
occurred only in a few occupational settings, e.g., the rubber
hydrochloride industry (Infante et al., 1977; Schnatter et al., 2012).
3.1.1.3. European prospective investigation into cancer and nutrition
(EPIC) cohort study.Saberi Hosnijeh et al. (2013) followed 241,465
subjects from 1992 through 2010 for a prospective study of lymphoid
and myeloid leukemia risk in relation to occupation, nutrition and
Table 2 (continued)
NRC (2011) Comment/Identified Data Gap New Formaldehyde Science
lack of transparency and detail may result in different estimates of unit risks, especially as
initial analyses resulted in a lack of a significant dose-response relationship for selected
endpoints. See: Van Landingham et al., 2016
BBDR models developed by Conolly and co-workers should be used. (p.58) These
models are biologically motivated and mechanistic; requiring that all
relevant data be reconciled with the model. (NRC, p.57)
•Expansion of the model to incorporate recent data on endogenous levels of formaldehyde
is in development. This will incorporate the most recent science to better understand when
exogenous formaldehyde exposure appreciably alters normal endogenous formaldehyde
concentrations. Work in progress: Clewell et al., unpublished
Consideration of the use of alternative extrapolation models for the analysis of
the cancer data. (NASNRC, p.14)
•Results of the “bottom-up “approach indicate that recent top-down risk extrapolations
from occupational cohort mortality data for workers exposed to formaldehyde are overly
conservative by substantial margins. See:Starr and Swenberg, 2013
•Updated “bottom-up”risk estimates heighten the marked contrasts that are present between
the previous estimates and the corresponding USEPA estimates, with the larger difference
for leukemia being due primarily to the significantly improved detection limit for the
analytical method used in quantitating DNA adduct numbers. See:Starr and Swenberg, 2016
E. Methods for Evidence Integration
EPA's approach to weight of evidence should include “a single integrative step
after assessing all of the individual lines of evidence.”Although a synthesis
and summary are provided, the process that EPA used to weigh different
lines of evidence and how that evidence was integrated into a final
conclusion are not apparent in the draft assessment and should be made
clear in the final version. (NRC, p. 113)
•A hypothesis-based weight-of-evidence (HBWoE) approach was conducted to evaluate the
large body of evidence regarding formaldehyde and leukemogenesis, attending to how
human, animal, and mode-of-action results inform one another. Upon comparison of
alternative proposals regarding what causal processes may have led to the array of
observations, it was concluded that the case for a causal association is weak and strains
biological plausibility. Instead, apparent association between formaldehyde inhalation
and leukemia in some human studies is better interpreted as due to chance or
confounding. See: Rhomberg et al., 2011
•Additional frameworks have been developed to integrate evidence. See:Adami et al., 2011;
Lavelle et al., 2012; Linkov et al., 2015; Rhomberg 2015b; Rooney et al., 2014; Woodruff
and Sutton, 2014.
•Other agencies or advisory bodies have conducted assessments of the carcinogenicity of
formaldehyde in a transparent manner. See:RAC, 2012; Bolt et al., 2016; Nielsen et al., 2017
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
480
metabolic risk factors. The European Prospective Investigation into
Cancer (EPIC) investigators studied occupational risk factors among
477 incident leukemia cases (201 ML, including 113 AML, 237
lymphoid leukemia, and 39 other or unspecified leukemias) in
France, Oxford (UK), the Netherlands, Sweden, Norway, and Italy
(Saberi Hosnijeh et al., 2013). Occupational exposures were estimated
using a general population JEM that classified occupational codes of
study subjects by categories of high, low, and no exposure for 11
specific agents (e.g., benzene, trichloroethylene) or groups of agents
(e.g., pesticides, chlorinated solvents). However, the authors reported
that work histories were missing on a large number of cohort members,
and these individuals had to be excluded. Study investigators lacked
detailed job histories (job tasks and duration) for others, and the
resulting exposure misclassification would be expected to be non-
differential, attenuating risk estimates. On the other hand, this is one
of the few studies examining specific subtypes of leukemia with risk
estimates adjusted for smoking and other risk factors. AML risk was not
increased among the formaldehyde low-exposure group (HR 1.01, 95%
CI 0.65–1.57) after adjusting for sex, smoking status, alcohol intake, age
at recruitment and country, and no AML cases occurred among
individuals in the high-exposure category. An HR for chronic
lymphocytic leukemia of 1.45 (95% CI 0.46–4.56) was reported
among those with high exposure to formaldehyde, but this was based
on 3 or fewer cases. ML risks were increased among those employed in
chemical laboratories and shoe and leather workers, and weakly
increased among those exposed to benzene but not those exposed to
ionizing radiation (Saberi Hosnijeh et al., 2013).
3.1.1.4. UK formaldehyde users and producers cohort study.Coggon et al.
(2014) updated mortality through 2012 for the UK cohort of 14,008
formaldehyde users and producers; however, the analysis grouped all
ML and did not analyze AML mortality separately. Similar to other large
industrial cohorts (Beane Freeman et al., 2009; Meyers et al., 2013),
industrial hygiene measurements were not available in the early years
and investigators estimated averages for job titles based on irritant
symptoms and later measurements. Exposures were estimated to range
from background (< 0.1 ppm), low exposure (0.1–0.5 ppm), moderate
exposure (0.6–2.0 ppm) and high exposure (> 2 ppm). These exposure
categories were similar to those estimated by Stewart et al. (1986) and
applied in Beane Freeman et al. (2009). Moreover, a larger proportion
(and greater number) of the UK cohort was exposed to high
concentrations of formaldehyde (approximately 18% of the cohort)
than the US cohort (approximately 4% of the cohort). Coggon et al.,
2014 reported no increased mortality from ML (SMR 1.16, 95% CI
0.60–2.20 for background exposure; SMR 1.46, 95% CI 0.84–2.36 for
low/moderate exposure; and SMR 0.93, 95% CI 0.450–1.82 for high
exposure). In a nested case-control analysis of 45 ML (diagnosis from
underlying or contributing cause of death or as a cancer registration)
and 450 controls matched on factory and age, no significantly increased
risk of leukemia was seen. Although ML risk was non-statistically
significantly increased among workers exposed to high concentrations
for < 1 year (OR 1.77, 95% CI 0.45–7.03), workers exposed to high
concentrations ≥1 year showed no increased risk (OR 0.96, 95% CI
0.24–3.82) (Coggon et al., 2014).
3.1.1.5. Extended analysis of the NCI cohort study to evaluate specific types
of myeloid leukemia.Checkoway et al. (2015) obtained the data from
the NCI formaldehyde industrial workers cohort to further investigate
specific types of leukemias, including AML (which had never been
reported for this cohort), as well as performing an alternative analysis
of peak exposure. The investigators reported that AML mortality was
unrelated to cumulative exposure or peak exposure. Twelve of 34 AML
deaths and 6 of 13 CML deaths occurred among study subjects with less
than one year of employment. For workers employed at least one year,
the risk of AML was highest (but not statistically significant) among
workers with peak exposures of ≥2.0 to < 4 ppm (HR 1.78, 95% CI
0.61–5.25) and no trend was seen with increasing category of peak
exposure (p for trend 0.37). In contrast, CML risks were greater,
although the estimates were imprecise (HR 4.83, 95% CI 0.64–36.42
for peak exposure ≥2.0 to < 4 ppm based on 2 CML deaths and HR
5.32, 95% CI 0.81–34.90 for peak exposure ≥4 ppm based on 2 CML
deaths).
3.1.2. Synthesis of epidemiology studies: exposure assessment issues
identified by NRC
One of the major issues highlighted by the NRC peer review is that
one exposure metric (peak exposure) was used to determine causality in
the draft IRIS assessment, while a different exposure metric (cumulative
exposure) was used for the dose-response evaluation to calculate an
inhalation unit risk.
The NRC (2011) review of the Draft IRIS Assessment stated “the
reliance on the peak exposure metric to determine causality rather than
the more conventional dose metric of cumulative exposure should be
further justified particularly in the absence of established modes of
action”[p.112]. NRC further elaborated:
“In the absence of evidence regarding exposure-disease mechanisms,
as in the case of formaldehyde and LHP cancers, cumulative ex-
posure is typically the default dose metric applied in epidemiologic
analyses and risk assessment. But the most significant results were
found for peak exposures, which have the greatest associated un-
certainty. In view of the importance of this study, EPA should clarify
the basis of its interpretations of the results regarding the various
dose metrics and the various LHP cancers. Despite those concerns,
the committee agrees that the NCI study is the most appropriate
available to carry forward for calculation of the unit risk.”(NRC,
2011, pp. 112–113)
The NRC recommended that the quality of exposure assessment
relied upon in epidemiological evaluations should be explicitly eval-
uated when weighting and synthesizing epidemiological evidence.
Where known causal relationships have been observed, exposure-re-
sponse relationships often are seen with various exposure metrics, with
stronger associations seen when more relevant metrics and exposure
time windows are examined. Results such as those reported by Beane
Freeman et al. (2009) are a good example of conflicting findings: the
conventional exposure metric, cumulative exposure, demonstrated no
association with risk of ML, whereas a surrogate of ‘peak’exposure
suggested one (Beane Freeman et al., 2009). When evaluating differ-
ences between cumulative exposure and peak exposure, and comparing
risks associated with these, several differences should be highlighted.
NCI investigators (Beane Freeman et al., 2009; Blair et al., 1986;
Hauptmann et al., 2003)defined peak exposure as the maximum peak,
and the NCI investigators substituted the time-weighted average (TWA)
for jobs without assigned peak exposures (Stewart et al., 1986). The
authors reported a significant test for trend between peak formaldehyde
exposure and leukemia, but only when unexposed subjects were in-
cluded. Increased risk was not seen for higher peak exposure categories
(2.0 to < 4.0 ppm, or ≥4.0 ppm) when compared to the lower peak
category (> 0 to < 2.0 ppm). No association was reported with fre-
quency of peak exposure, average intensity of exposure or with cu-
mulative exposure to formaldehyde (“There was little evidence among
formaldehyde workers of association for any lymphohematopoietic
malignancy (LHM) with average intensity or cumulative exposure at the
end of follow-up in 2004.”(Beane Freeman et al., 2009, p. 751). In fact,
a 10% deficit of ML deaths (acute and chronic types combined) was
reported when compared to US population mortality rates. In an in-
ternal analysis, Beane Freeman et al. (2009) reported that ML deaths
were not associated with the number or frequency of peaks. If there
were a true association between peak exposure and leukemia, one
would expect to see an association with number of peaks and not only
ever having a (perhaps single) peak exposure. Hauptmann et al. (2003)
acknowledged that “no measurements of peak exposure were available
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
481
in this study. Peak exposures were therefore estimated by an industrial
hygienist from knowledge of the job tasks and a comparison with the 8-
hour time-weighted average”(Hauptmann et al., 2003, p. 1616;
Stewart et al., 1986). Stewart et al. (1986) reported that the exposure
reconstruction included rating confidence (i.e., confident, less con-
fident, not confident) in the exposure estimate; however, the “con-
fidence”category appeared to apply to the “rank”exposure and not the
“peak exposure.”For example, if an IH specified “not confident”for an
average exposure estimate, it is not clear how or if this information
applied to the estimate of peak exposure (categorized during data col-
lection as 1 = none, 2 = 0.1–0.5, 3 = 0.51–2.0, 4 = 2.1–4.0,
5 = > 4.0, 9 = unknown) (Stewart et al., 1986).
In extended analyses of the NCI cohort study, Checkoway et al.
(2015) refined the classification of peak exposure. Workers who did not
work in jobs identified as likely having peak exposures were classified
as not exposed to peaks, and became the referent group. A total of 3478
cohort members were classified as having worked in jobs with esti-
mated peak exposure of 2- < 4 ppm, and 2907 worked in jobs with
estimated peak exposure of ≥4 ppm. Analysis by ML subtype (i.e., AML
and CML deaths, separately) found no association between peak ex-
posure and AML mortality (HR 1.71, 95% CI 0.72–4.07 and HR 1.43,
95% CI 0.56–3.63, respectively) (Checkoway et al., 2015). However, 13
of the 34 AML deaths were classified as having worked in jobs likely
having peak exposure > 2.0 ppm, only 4 of which worked in these jobs
within the 20 years preceding their AML death (i.e., latest exposure),
and only one occurred (similar to the number expected) within the
typical AML latency window of 2–15 years. Upon fuller analyses of
these data, Checkoway et al. (2015) subsequently found that only a
third of all the AML deaths were among cohort members assigned to
categories with any peak exposure (i.e., > 2.0 ppm), nearly all of whom
had their last peak exposure more than 20 years earlier, well outside of
the maximum latency window.
Coggon et al. (2014) also reported that limited IH data were
available for the UK formaldehyde users and producers cohort, pre-
venting the derivation of quantitative metrics. Nevertheless, the in-
vestigators expressed high confidence that the high exposure category
corresponded to average concentrations of at least 2 ppm. Industrial
hygiene data also were limited in the US NCI industrial workers study,
although the investigators used them as part of a detailed exposure
reconstruction using best practices for such a reconstruction at the time.
Stewart et al. (1986) reported that historical exposure levels were es-
timated because most companies did not begin sampling until the mid-
1970's: they also monitored “present day”(i.e., early 1980's) operations
to help extrapolate historical exposures. The NCI investigators relied
upon exposure rank (six levels of TWA): trace, < 0.1 ppm, 0.1–0.5 ppm,
0.51–2.0 ppm and > 2 ppm.
One criticism leveled at the UK worker cohort study (Acheson et al.,
1984; Coggon et al., 2003, 2014; Gardner et al., 1993) was that the
“authors reported a concern about the quality of data when they made
exposure assignments”(NRC, 2014b). This criticism seems to stem from
the appropriate identification and discussion of study limitations by
earlier UK investigators: Gardner et al. (1993) reported “when jobs
were being placed into qualitative categories of exposure in the British
study, some disagreement occurred as to which of two adjacent grades
was most appropriate-for example, high or moderate? To achieve
consistency across all the factories, the higher of the two was always
used. It is not clear how differences were resolved in the United States
study.”Thus, there are no essential differences in the approach used by
the UK investigators and the US investigators: both studies reported
that limited data were available on quantitative exposure measures
using existing industrial hygiene data (from the 1980s); both exposure
assessments allowed for the consideration of changes in processes and
exposure controls during the period of the study; and both used ranked
categories of exposure, developed before the estimation process, based
somewhat on subjective sensory experiences encountered in the job
(e.g., odor occasionally present), and both used eye irritation and odor
throughout the day to identify the highest intensity of exposure jobs
(Acheson et al., 1984; Stewart et al., 1986).
Ultimately, the Beane Freeman et al. (2009) study alone does not
(and cannot) provide reliable support for a conclusion that peak for-
maldehyde exposure causes ML or AML, especially considering the
absence of peak measurement data in the US study, the results of the re-
analysis by Checkoway et al. (2015), and the updated results from the
UK study (Coggon et al., 2014), which used a more conservative ap-
proach to exposure estimation.
3.1.3. Synthesis of epidemiology studies: evaluation of the most specific
diagnosis
The NRC (2011) raised the issue that diverse types of leukemias and
lymphomas should not be grouped “because it combines many diverse
cancers that are not closely related in etiology and cells of origin. Al-
though the draft IRIS assessment explores specific diagnoses—such as
AML and CML, as well as Hodgkin lymphoma and multiple myeloma
(see, for example, EPA 2010, Tables 4–92)—the determinations of
causality are made for the heterogeneous groupings of “all LHP can-
cers,”“all leukemias,”and “ML”. When results for heterogeneous
groupings are presented, there is no evidence of increased risk of all
LHP cancers (Meyers et al., 2013; Beane Freeman et al., 2009) or all
leukemias combined (Coggon et al., 2014; Meyers et al., 2013; Beane
Freeman et al., 2009) in industrial cohorts when compared to general
mortality rates. In addition, there is no evidence of exposure-response
associations between all LHPs combined (or all leukemias combined)
and cumulative exposure or average exposure (Beane Freeman et al.,
2009) or duration of exposure (Meyers et al., 2013; Coggon et al.,
2014).
Interestingly, the Draft IRIS Assessment noted that “Acute leukemias
(ALL and AML), believed to arise from transformation of stem cells in
the bone marrow, are less plausible. In contrast chronic lymphatic
leukemia, lymphomas, multiple myelomas (from plasma B cells), and
unspecified cancers may involve an etiology in peripheral tissues to
include cells, cell aggregates, germinal centers, and lymph nodes. An
association of these cancers to an exogenous agent acting at the POE
[portal of entry] is biologically plausible”(EPA, 2010; page 4–190).
While the etiologies of most LHM are poorly understood, the pos-
sible role of environmental agents is plausible for AML, which has been
linked with benzene, tobacco smoking, ionizing radiation and various
cancer treatment agents, such as cisplastin, all of which have been
classified by IARC as known human carcinogens that cause AML. It
should be stressed that evidence exists that these agents, or their car-
cinogenic components, are capable of reaching the bone marrow.
However, only six epidemiological studies of workers substantially
exposed to formaldehyde published to date have published AML-spe-
cific results (Blair et al., 2001; Checkoway et al., 2015; Hauptmann
et al., 2009; Meyers et al., 2013; Saberi Hosnijeh et al. 2013; Talibov
et al., 2014), four of which were not available at the time of the IARC
review or the release of the Draft IRIS Assessment. Saberi Hosnijeh et al.
(2013) reported no association between “low”formaldehyde exposure
and incidence of myeloid leukemia (HR 1.02, 95% CI 0.72–1.42 based
on 49 cases exposed to formaldehyde and 130 unexposed cases). No
differences were seen between subtypes: AML (HR 1.01, 95% CI
0.65–1.57) or CML (HR 0.92, 95% CI 0.46–1.84). No myeloid cases
(and therefore no AML cases or CML cases) occurred among those
classified as having “high”formaldehyde exposure (Saberi Hosnijeh
et al., 2013). Talibov et al. (2014) found no association between for-
maldehyde and incident AML, after adjusting for exposure to specific
solvents and ionizing radiation (HR 1.17, 95% CI 0.91–1.51 for 136
workers and 628 controls exposed to > 1.6 ppm-yrs). Meyers et al.
(2013) reported a SMR for AML of 1.22 (95% CI 0.67–2.05) based on 14
observed AML deaths. Checkoway et al. (2015) performed AML-specific
analysis using the NCI cohort, which had provided results only for all
ML combined (Beane Freeman et al., 2009). When compared to US
referent rates, AML mortality risk was decreased among workers
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
482
exposed to formaldehyde (SMR 0.80, 95 %CI 0.46–1.14) and internal
analysis of exposure reported no trend with increasing cumulative ex-
posure or peak exposure categories (Checkoway et al., 2015). Thus,
new analyses of the NCI formaldehyde workers cohort specifically for
AML detract from the hypothesis that formaldehyde causes AML.
The associations reported by Beane Freeman et al. (2009) between
formaldehyde exposure and Hodgkin lymphoma and CML have not
been observed in other studies (Meyers et al., 2013; Saberi Hosnijeh
et al., 2013) and are less plausible, given the lack of known associations
with Hodgkin lymphoma or CML and other chemicals or agents, such as
benzene (Checkoway et al., 2015). Saberi Hosnijeh et al. (2013) re-
ported a RR of 0.92 (95% 0.46 to 1.84) based on 46 CML cases. Meyers
et al. (2013) reported a SMR of 1.35 (95% CI 0.44–3.15), based on 5
CML cases through 2008. The absence of established toxicological
mechanisms for formaldehyde exposure and any of the LHM further
weakens arguments for causation (Checkoway et al., 2012, 2015),
especially given that inhaled formaldehyde appears incapable of
reaching the bone marrow (discussed in Section 3.3).
3.2. Toxicological evidence
3.2.1. Animal evidence of formaldehyde-induced LHM
With regard to animal evidence of formaldehyde-induced LHM, the
Draft IRIS Assessment (EPA, 2010) stated that the available animal
evidence is limited, discussing mainly the results from the Battelle
Columbus Laboratories (1981) study. The Draft IRIS assessment in-
dicated that this study provides the only evidence of formaldehyde-
induced LHM in animal models. However, the NRC (2011) peer review
noted that although intriguing, EPA's unpublished re-analysis of the
Battelle chronic experiments in mice and rats (Battelle Columbus
Laboratories, 1981) contributed little to the weight of evidence eva-
luation.
In rats, Battelle Columbus Laboratories (1981) reported the in-
cidence of leukemia (most of which were diagnosed as undifferentiated
leukemia found sporadically in various organs) in male and female
Fischer 344 rats following exposure to concentrations of 0, 2, 6, or
15 ppm for 24 months, followed by 6 months with no exposure. No
concentration-related increases in the incidences of leukemia in either
sex of rats were reported by Battelle Columbus Laboratories (1981),
when a standard Fisher-Irwin exact test was applied (males p = 0.0972;
females p = 0.2316).
Because of a significant number of early deaths in the high con-
centration group of both males and females, Battelle Columbus
Laboratories (1981) also applied Tarone's extension to the Cox log-rank
test (Tarone, 1975) to evaluate the leukemia incidence data. This test
accounts for the number of animals at risk at each time point when the
response of interest is observed. This adjustment assessed the prob-
ability of developing the endpoint of interest in those animals that did
not survive until the termination of the study. The results of Tarone's
extension indicated that the incidence among female rats in the high
concentration group was statistically significant (p = 0.0056, not
0.0003 as reported
3
); however, no association was seen in the male rats
exposed at high concentrations (p = 0.6891). No concentration-related
increase in leukemia was observed in the female rats exposed at either
2 ppm or 6 ppm, and no survival problems were noted. Even after
application of Tarone's extension, leukemia in male or female rats was
not identified in the Battelle Columbus Laboratories (1981) study as an
endpoint related to formaldehyde exposure, nor was it so designated in
two publications citing this study (Kerns et al., 1983; Swenberg et al.,
2013).
More contemporary statistical methods, such as the Cochran-
Armitage and the Poly3 (Bailer and Portier, 1988; Peddada and
Kissling, 2006) trend tests, have replaced those used in the early 1980's.
The Poly3 trend test is a survival-adjusted quantal-response procedure
that modifies the Cochran-Armitage linear trend test to take inter-group
survival differences into account. Importantly, the Poly3 test is the test
currently used by the National Toxicology Program (NTP) to evaluate
incidence data both for trend and pair-wise comparisons, to assess the
probability of the response in the presence of inter-current mortality.
The results of the application of these tests indicated p values of 0.43
and 0.82 for the Poly3 and Cochran-Armitage, respectively, demon-
strating no association.
In mice, the Draft IRIS Assessment (EPA, 2010) suggested that the
“adjusted”incidence of lymphoma in female mice, when the 6-month
sacrifice animals were removed from consideration (because tissues
outside of the respiratory tract were not examined), was statistically
significant (p < 0.05) in animals exposed to 15 ppm formaldehyde,
compared to untreated controls. However, as indicated in the methods
for the Battelle Columbus Laboratories (1981) study, statistical sig-
nificance, when applying the Tarone extension of the Cox test, is
achieved with a p value of 0.05 divided by the number of dose groups.
In the case of the Battelle Columbus Laboratories (1981) study for the
mouse data, statistical significance would be p < 0.0167, as noted in
the summary tables (Table 8 of the Battelle Columbus Laboratories
(1981) report); therefore, based on this criterion, this endpoint was not
considered statistically significant. As with the leukemia incidence in
rats, the Battelle study authors did not report lymphoma in mice as an
endpoint related to formaldehyde exposure.
Since 2010, two short-term carcinogenicity studies have been con-
ducted and published (as a Technical Report) by the NTP of NIEHS in
strains of genetically predisposed mice (male C3B6·129F1-
Trp53tm1Brdp53 haplo-insufficient mice and male B6.129-
Trp53tm1Brd) (Morgan et al., 2017). These short-term carcinogenicity
studies were conducted to test the hypothesis that formaldehyde in-
halation would result in an increased incidence and/or shortened la-
tency to nasal and lymphohematopoietic tumors and to investigate
hypotheses that formaldehyde may induce leukemia by a mechanism
not involving DNA adduct formation. This proposed mechanism as-
sumes that inhaled FA could cause significant genetic damage to stem
cells in the nasal epithelium or circulating in local blood vessels. These
damaged stem cells could reach the general circulation, home to tissues
that support the hematopoietic niche, undergo lodgement and become
leukemic stem cells. The animals were exposed to 7.5 or 15 ppm for-
maldehyde 6 hours/day, 5 days/week, for 8 weeks. The investigators
reported that because the doubling time for hematopoietic stem and
progenitor cells (HSPCs) is between 2 and 4 weeks, and the entire HSPC
pool turns over every 8 weeks, an 8 week exposure duration was con-
sidered sufficient to investigate the hypothesized mechanism for indu-
cing leukemia. Following the 8-week inhalation exposure, mice were
monitored for approximately 32 weeks (until approximately 50 weeks
of age). At the highest concentrations, significant cell proliferation and
squamous metaplasia of the nasal epithelium were observed; however,
no nasal tumors were observed. No cases of leukemia were seen in ei-
ther strain and a low incidence of lymphoma in exposed mice was not
considered related to exposure. In addition, no significant changes in
haematological parameters were noted. Under the conditions of these
studies, the authors concluded that formaldehyde inhalation did not
cause leukemia in these strains of genetically predisposed mice (Morgan
et al., 2017).
Overall, the weight of evidence from animal studies reported in the
Draft IRIS Assessment (EPA, 2010) did not support an association be-
tween formaldehyde exposure and LHM. Since that time, additional
studies (Morgan et al., 2017) have provided evidence that suggests a
lack of association between formaldehyde exposure and LHM. In ad-
dition, no evidence of changes in blood parameters that might be
3
This appears to be a misreading of the Battelle report. In the Battelle Report Volume A
Table 10 –Analysis of Effects of Formaldehyde in Female Rats - reports a p-value of
0.0056 from the Adjusted Cox/Tarone pair-wise comparison of the control to 15 ppm for
Leukemia, all. The next row in that table with an endpoint of Uterus, Endometrial Stromal
Polyp is the one that reports a p-value of 0.0003 for the pair-wise analysis of control to
15 ppm.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
483
associated with leukemias has been reported in any animal studies
exposed to formaldehyde at high concentrations following both acute
and chronic durations (Appelman et al., 1988; Dean et al., 1984;
Johannsen et al., 1986; Kamata et al., 1997; Kerns et al., 1983; Til et al.,
1988, 1989; Tobe et al., 1989; Vargova et al. 1993; Woutersen et al.,
1987). Among these studies, Vargová et al. (1993) reported increased
red blood cell counts and increased proportions of lymphocytes and
monocytes in rats, rather than decreases, following exposure to for-
maldehyde by gavage at 80 mg/kg/day for 28 days.
3.3. Mode of Action Evidence
3.3.1. Improve understanding of when exogenous formaldehyde exposure
appreciably alters normal endogenous formaldehyde concentrations
NRC (2011) recommended that one key improvement to the science
would be an understanding of when exogenous formaldehyde exposure
altered normal endogenous formaldehyde concentrations. Because
formaldehyde is endogenously present, it is important to differentiate
levels that are due to normal metabolic processes from levels that might
be present as a result of exogenous exposure. A number of studies have
applied sensitive methods to differentiate exogenous and endogenous
levels of formaldehyde in tissues (Casanova-Schmitz et al., 1984; Lu
et al., 2010, 2011; Moeller et al., 2011; Swenberg et al., 2011).
The results of these studies with highly sensitive instruments and
accurate assays indicate that inhaled formaldehyde was present in the
nasal respiratory epithelium, but not other tissues beyond the site of
initial contact. In contrast, endogenous adducts were readily detected in
all tissues examined. Moreover, the amounts of exogenous for-
maldehyde-induced adducts were 3- to 8-fold and 5- to 11-fold lower
than the average amounts of endogenous formaldehyde-induced ad-
ducts in rat and monkey nasal respiratory epithelium, respectively (Yu
et al., 2015).
An additional study conducted in rats exposed to
13
C-formaldehyde
(Kleinnijenhuis et al., 2013) provided results consistent with those from
studies focused on measuring endogenous versus exogenous DNA ad-
ducts. In this study, Sprague-Dawley rats were exposed nose-only to
10 ppm
13
C-formaldehyde for 6 hours and blood concentrations eval-
uated during exposure and for 30 minutes following exposure. This
study was conducted specifically to investigate the mechanism pro-
posed by Zhang et al. (2010a) that formaldehyde is absorbed during
respiration and could reach any target tissue, such as the bone marrow,
via the blood in the form of methanediol to exert its genotoxic activity.
Exogenous
13
C-formaldehyde was not detectable in the blood of rats
either during or up to 30 min after the exposure. The authors concluded
that “it is highly unlikely that the mechanism proposed by Zhang et al.
(2009), that exposure to FA by inhalation may lead to an increased FA
concentration in blood and as such may cause leukemia, is true”
(Kleinnijenhuis et al., 2013).
New studies have been conducted to investigate the potential toxi-
city/carcinogenicity of endogenous formaldehyde. The most recent
studies demonstrate that endogenous formaldehyde in bone marrow is
toxic, and probably carcinogenic, and therefore may increase leukemia
risk (Pontel et al., 2015; Lai et al., 2016).
3.3.2. Reconcile divergent statements regarding systemic delivery
Multiple studies in rats (Lu et al., 2011; Yu et al., 2015; Edrissi et al.,
2013) and monkeys (Moeller et al., 2011; Yu et al., 2015) conducted
with sensitive analytical methods that can measure endogenous versus
exogenous formaldehyde DNA or protein adducts have demonstrated
that inhaled exogenous formaldehyde is not systemically absorbed or
reaches sites distant from the point of initial contact. In addition to
these studies, the available data on the toxicokinetics of formaldehyde
suggest that no significant amount of “free”formaldehyde would be
transported beyond the portal of entry.
In addition to studies supporting the lack of systemic delivery of
formaldehyde, anatomically accurate computational fluid dynamics
(CFD) models of the rat, monkey, and human have been applied to
evaluate the effects of endogenously present formaldehyde on uptake
from the respiratory tract. The consideration of endogenous for-
maldehyde concentrations in nasal tissues did not affect flux or nasal
uptake predictions at exposure concentrations > 500 parts per billion
(ppb); however, reduced nasal uptake was predicted at lower exposure
concentrations (Schroeter et al., 2014).
3.3.3. Data are insufficient to conclude formaldehyde is causing cytogenetic
effects at distant sites
The modes of action that have been proposed in the Draft IRIS
Assessment (EPA, 2010) to cause leukemogenesis rely strongly on the
hypothesis that exposure to inhaled formaldehyde can result in cyto-
genetic effects at sites distant from the portal of entry. While the NRC
(2011) noted that numerous studies have shown genotoxic effects in
cells exposed in vitro, and a few studies have shown positive cytogenetic
effects in circulating blood lymphocytes in heavily-exposed workers,
they also noted that it is unlikely that these effects are relevant to a
possible leukemogenic effect of formaldehyde, particularly at low ex-
posure levels. The potential leukemogenic effect and exposure-response
relationships at lower exposure levels have been comprehensively
evaluated by Nielsen et al. (2013, 2017).
One key study cited in multiple agency evaluations as providing
evidence of cytogenetic events in the development of leukemias is by
Zhang et al. (2010a, 2010b) compared the prevalence of markers of
hematopoietic function and chromosomal aneuploidy among workers
occupationally exposed to formaldehyde with those of a group of un-
exposed workers in China. Ninety-four workers were included, with 43
workers occupationally exposed to formaldehyde and 51 workers un-
exposed to formaldehyde as controls. The authors reported a higher
prevalence of monosomy 7 (loss of a chromosome) and trisomy 8 (gain
of a chromosome) in metaphase spreads prepared from cultures of CFU-
GM colony cells. The authors suggested that this demonstrated that
formaldehyde exposure was associated with an increase in leukemia-
specific chromosomal aneuploidy in vivo in the hematopoietic pro-
genitor cells of the exposed workers. However, no direct in vivo meta-
phases had been examined in workers blood. Furthermore, this was a
cross-sectional comparison of blood and cytogenetic measures between
two groups, and observed differences could not be established as re-
sulting from formaldehyde exposure or due to other overall differences
between the two groups.
Two re-analyses of the underlying data from the Zhang et al.
(2010a) study have been published (Gentry et al., 2013; Mundt et al.,
2017). The first (Gentry et al., 2013) relied upon selected underlying
data provided through a Freedom of Information Act request that in-
cluded: 1) individual data on blood cell counts in both formaldehyde-
exposed and unexposed individuals including any data on health status
of these individuals; 2) individual data on the FISH results for
monosomy 7 and trisomy 8 for cultures of samples obtained from 10
formaldehyde-exposed workers and 12 unexposed controls; 3) data on
additional chromosomal abnormalities examined and/or observed; and
4) details of the methods sufficient for a qualified scientist to replicate
the results reported in the Zhang et al. (2010) study. The results of this
reanalysis suggested that factors other than formaldehyde exposure
likely contributed to the reported findings. In addition, although the
authors stated in their paper that “all scorable metaphase spreads on
each slide were analyzed, and a minimum of 150 cells per subject was
scored,”this protocol was not followed specifically for chromosome 7
or chromosome 8 (recent correspondence indicates a minimum of 150
total metaphases were scored for 24 chromosomes per subject). Far too
few cells were counted to draw any meaningful conclusions, and far
fewer than the approximately 400 per chromosome cited in previous
analyses in which the protocol was described (Zhang et al., 2005,
2011). In addition, the assays used (CFU-GM) do not actually measure
the proposed events in primitive cells involved in the development of
AML. Evaluation of these data indicates that the aneuploidy measured
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
484
could not have arisen in vivo, but rather arose during in vitro culture.
In 2014, Mundt et al. requested the individual exposure measure-
ment data for each of the participants in the Zhang et al. (2010a) study
from NCI. In 2016, the request was in part granted and the mean for-
maldehyde estimate for each exposed worker (but not the individual
exposure measurement values) was provided via a Technology Transfer
Agreement (TTA) with NCI. Using these data, the Gentry et al. (2013)
reanalysis was extended to include exposure-response analyses. Results
of this second reanalysis showed that differences seen at the group
comparison level, i.e., comparing the prevalence of white blood cell,
granulocyte, platelet, and red blood cell counts at the group level in fact
were independent of measured formaldehyde exposure level. Among
exposed workers, no association was observed between individual
average formaldehyde exposure estimates and frequency of aneuploidy,
suggested by the original study authors to be indicators of ML risk.
Differences between the two groups of workers, other than for-
maldehyde exposure, were therefore likely to explain the results re-
ported by Zhang et al. (2010a).
Subsequent studies of the same population of formaldehyde-exposed
and non-exposed workers in China (Lan et al., 2015; Seow et al., 2015;
Bassig et al., 2016) have been suggested by the authors to confirm the
results of Zhang et al. (2010a); however, many of these studies report
results from the same biological samples as Zhang et al. (2010a) and
therefore, do not provide replication of the results. The repeated use of
the original Zhang et al. (2010a) data, and its implications, have been
reiterated (Pira et al., 2017; Gentry et al., 2013; Speit et al., 2010) and
the original authors have responded to some of the criticisms (Rothman
et al., 2017; Lan et al., 2015; Zhang et al., 2010b). Replication of the
Zhang et al. (2010a) results will require replication in an independent
population of formaldehyde-exposed workers, and where methodolo-
gical issues are adequately addressed. An attempt to replicate the re-
sults could be conducted in the same population of workers as Zhang
et al. (2010a) and Lan et al. (2015) in which the median exposures to 43
workers were 1.28 ppm (10th and 90th percentile: 0.63, 2.51 ppm).
However, as noted previously (Section 3.1.1), no evidence of an asso-
ciation between formaldehyde exposure and leukemias has been re-
ported in multiple recent epidemiological studies with large numbers of
subjects that have been exposed to concentrations > 2.0 ppm. The in-
creasing evidence that inhaled formaldehyde does not move beyond the
portal of entry (Section 3.3.2) also calls into question many of the
conclusions from Zhang et al. (2010a).
Albertini and Kaden (2016) reviewed the body of data that re-
portedly indicates genetic changes in circulating blood cells and in
blood-borne hematopoietic precursor cells (HPCs). These changes have
been considered to be indicators that systemic genotoxicity occurs after
human inhalation exposure to formaldehyde, although the mechanisms
by which this could occur remain unknown. For each study, the authors
examined the sources of exposure, possible co-exposures, biomarkers
for internal exposures and genetic signatures of formaldehyde effects.
In reviewing the available studies, many genetic changes in blood
cells were noted by Albertini and Kaden (2016), with a contrast in re-
sults between animal and human studies: the majority of animal studies
were negative and the majority of human studies were positive. This
pattern was attributed to the difference in target cell being studied,
with bone marrow cells studied in animals and peripheral blood lym-
phocytes studied in humans. Exposure of human cells to formaldehyde
at sites of contact in vivo could provide opportunities for exposure of T-
lymphocytes to formaldehyde or products of oxidative stress, which
could result in the genetic changes observed in peripheral blood cells.
However, these results are inconsistent with results from controlled
animal studies, discussed previously, that demonstrate - by labeling
administered formaldehyde - inhaled (exogenous) formaldehyde does
not travel beyond the portal of entry (Casanova-Schmitz et al., 1984; Lu
et al., 2010, 2011; Moeller et al., 2011; Swenberg et al., 2011).
Therefore, these types of genetic changes reported in human studies do
not provide evidence that formaldehyde moves beyond the portal of
entry to the bone marrow, which would be necessary to result in direct
induction of chromosome-level mutations in the bone marrow. Despite
the apparent inability of exogenous formaldehyde to reach the bone
marrow, the mutagenic effects of formaldehyde in bone marrow have
not been tested in humans.
Albertini and Kaden (2016) concluded that overall, the available
literature on genetic changes following formaldehyde exposure did not
provide convincing evidence that exogenous exposure, and specifically
exposure by inhalation, induces mutations as a direct DNA-reactive
effect at sites distant from the portal-of-entry tissue. This would include
proposed mode of actions that involve a stem cell effect at the portal of
entry with circulation back to the bone marrow. Such exposures have
not been shown to induce mutations in the bone marrow or in any other
tissues beyond the point of contact. Thus, the weight of scientific evi-
dence does not provide biological plausibility of lymphohematopoietic
cancers, as proposed by EPA (2010) and NTP (2011).
3.4. Dose-response assessment
Several NRC (2011) peer-review comments were raised regarding
the dose-response assessment conducted by EPA in the Draft IRIS As-
sessment (2010). One comment highlighted the need to conduct in-
dependent analyses of the dose-response models, using the data from
the Beane Freeman et al. (2009) study to confirm which models fit the
data appropriately (NRC, 2011). Using the original data from the key
study (Beane Freeman et al., 2009) and documentation provided in the
Draft IRIS Assessment, Van Landingham et al. (2016) attempted to
duplicate the reported inhalation unit risk (IUR) values for Hodgkin
lymphoma and all leukemias and address the NRC Committee's ques-
tions regarding application of the appropriate dose-response model.
Overall, there was difficulty duplicating the IURs reported by EPA
(2010), largely due to a lack of critical information provided in the IRIS
documentation. Perhaps most problematic, the first step of the analysis
did not determine significant exposure-response relationships between
formaldehyde and lymphohematopoietic endpoints for the metric (cu-
mulative exposure) needed in the estimation of an IUR. The authors
concluded that the resulting analysis, while it could be mechanically
performed, provided no valid or useful insights on the risks of for-
maldehyde exposure. The lack of apparent exposure-response re-
lationships for selected endpoints raises the question whether quanti-
tative analyses are appropriate for these endpoints, and if so, how
results are to be interpreted.
The NRC (2011) also noted the need to consider alternative extra-
polation models for analyzing the cancer data. In 2013, Starr and
Swenberg proposed a novel “bottom-up”approach for bounding low-
dose human cancer risks using formaldehyde as an example (Starr and
Swenberg, 2013). This approach requires information on background
risk, background or endogenous exposure and the additional exogenous
exposure of interest. The results of this approach provided estimates of
risk (< 3.9 × 10
−6
) that were more than 14,000-fold lower than the
corresponding Draft IRIS Assessment (EPA, 2010) estimate for all leu-
kemias (5.7 × 10
−2
) and considers the impact of background en-
dogenous formaldehyde concentrations, which is not considered in the
Draft IRIS Assessment (EPA, 2010). In 2016, Starr and Swenberg pro-
vided an update to this approach, incorporating new formaldehyde-
DNA adduct data, and allowing for uncertainty in two of the parameters
(background cancer risk and background endogenous concentrations of
formaldehyde) (Starr and Swenberg, 2016). Consideration of the sta-
tistical uncertainty in these two parameters resulted in estimates of risk
for leukemias that were even smaller than those initially estimated in
Starr and Swenberg (2013). The authors concluded that these estimates
provide a reality check for the IUR presented in the Draft IRIS Assess-
ment (EPA, 2010). In addition, the large discrepancy between results
using an approach that relies on molecular dosimetry data (i.e., the
bottom up approach) versus one that relies upon uncertain retro-
spective occupational exposure reconstructions (i.e., the approach
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
485
relied upon in EPA (2010) call into question the credibility of attri-
buting increases in human mortality from leukemias to occupational
exposure to formaldehyde.
3.5. Methods for evidence integration
The NRC (2011) noted that the Draft IRIS Assessment's (EPA, 2010)
approach to weight of evidence should include “a single integrative step
after assessing all of the individual lines of evidence”. Although a
synthesis and summary are provided, the process that EPA used to
weigh different lines of evidence and how that evidence was integrated
into a final conclusion are not apparent in the draft assessment and
should be made clear in the final version.
Since the Draft IRIS Assessment (EPA, 2010) and the NRC (2011)
peer review, several frameworks have been developed to integrate
evidence across different lines of scientific inquiry including epide-
miology, toxicology and mode of action studies (Adami et al., 2011;
Lavelle et al., 2012; Linkov et al., 2015; Rhomberg, 2015b; Rooney
et al., 2014; Woodruffand Sutton, 2014). The EPA has also proposed
preliminary approaches for integrating evidence in response to the NRC
(2011) peer review of formaldehyde (EPA, 2013a).
Rhomberg et al. (2011) applied a hypothesis-based weight of evi-
dence approach to evaluate formaldehyde and leukemogenesis, con-
sidering how human, animal and mode of action results inform one
another. In comparing the potential alternative proposals for causality,
the authors concluded that the evidence for a causal association be-
tween formaldehyde exposure and leukemia is not only weak but
strains biological plausibility (Rhomberg et al., 2011).
Nielsen et al. (2017) also considered the body of formaldehyde re-
search while re-evaluating the WHO (2010) formaldehyde indoor air
quality guideline for cancer risk assessment. Nielsen et al. (2017) iter-
ated that although formaldehyde is genotoxic and causes DNA adduct
formation, it is also clastogenic. Exposure-response relationships from
both animal and human data were nonlinear, and relevant genetic
polymorphisms had not been identified. Although one epidemiological
study had reported an association with nasopharyngeal cancer and
others reported inconsistent associations with leukemias, relative risks
were not increased below 1 ppm (mean exposures). Because inhaled
formaldehyde does not pass beyond the respiratory epithelium, any
direct effects are limited to portal-of-entry effects (Nielsen et al., 2017).
Other reviews and syntheses of evidence focused on epidemiological
studies, and this body of literature has been most variably interpreted.
In 2014, an independent National Research Council committee was
charged with performing a peer review of the NTP evaluation of for-
maldehyde for the 12th edition of the RoC (NRC, 2014b). This NRC
committee produced a new definition for “sufficient evidence”of car-
cinogenicity as demonstrated by two or more strong or moderately
strong epidemiological studies with different study designs and popu-
lations showing associations between formaldehyde exposure and a
specific cancer type. In this approach, “strong”epidemiology studies do
not refer to the magnitude of the association, but relect a judgment of
study quality and utility made by reviewers who considered chance,
bias, and confounding as alternative explanations for the observed as-
sociation and found these were not reasonable explanations. Further,
“strong”epidemiology studies comprised large populations with long
durations of exposure and an adequate follow up period to allow for
latency, and had exposure assessments that were able to discriminate
between “high”and “low”formaldehyde exposure categories. This
“strength of evidence”approach contrasts with a “weight of evidence
approach.”Although each epidemiology study was classified as one of
three categories (strong, moderately strong, or weak), this approach
suggests that 2 or more strong or moderately strong studies with po-
sitive results are enough to conclude sufficient evidence of carcino-
genicity exists, and discounts epidemiology and animal studies that are
negative or contradictory.
Meta-analyses are often used to synthesize findings across many
epidemiology studies, identifying sources of potential heterogeneity
which then can be explored in interpreting the overall evidence. In the
Draft IRIS Assessment (EPA, 2010), meta-analyses conducted by several
investigators were considered (Zhang et al., 2009; Collins and Lineker,
2004; Bosetti et al., 2008). Since then, two additional meta-analyses
were conducted (Bachand et al., 2010; Schwilk et al., 2010). Bachand
et al. (2010) excluded lower-quality studies and reported a meta-RR of
1.05 (95% CI 0.93–1.20) based on 16 cohort studies and a meta-OR of
0.99 (95% CI 0.71–1.37) based on 2 case-control studies for all leu-
kemia, reported separately due to heterogeneity. Schwilk et al. (2010)
published a meta-analysis of the epidemiological findings on myeloid
leukemia, but limited to the highest-exposed sub-group reported in four
studies (three cohort and one case-control): RR = 2.47; 95% CI, 1.42 to
4.27. Checkoway et al. (2012) conducted a critical review and synthesis
of the epidemiological evidence and concluded that results from epi-
demiological studies were not consistent and did not show strong re-
sults or exposure-response associations. None of these reviews, how-
ever, included the results from the extended follow up of the NIOSH
garment workers study (Meyers et al., 2013), the extended follow up of
the UK producers and users (Coggon et al., 2014) or the extended
analyses of the NCI cohort (Checkoway et al., 2015). In addition, meta-
analyses and/or critical reviews of epidemiological literature require
further integration with other lines of evidence.
4. Conclusions
It has been seven years since the release of the Draft IRIS
Toxicological Review of Formaldehyde (EPA, 2010). In peer-reviewing
this draft report, an NRC Committee raised many substantive questions
related specifically to the conclusions drawn in the document and the
quantitative estimates of potential toxicity (NRC, 2011). This Com-
mittee was tasked with reviewing and commenting on information
provided in the draft assessment, and did not independently conduct a
review of the primary literature, but did determine that many of EPA's
conclusions were not supported by the information and studies cited in
the draft assessment. The committee also identified general methodo-
logic problems with the Draft IRIS Assessment, and provided specific
comments related to the evaluation of specific studies and conclusions
based on the available evidence. The comments related to a causal as-
sociation between formaldehyde exposure and LHM largely involved
the interpretation of the available evidence at that time and the fra-
mework in which it was evaluated by EPA (2010). The committee found
that EPA's preliminary conclusion that formaldehyde causes leukemia,
ML or related hematopoietic cancers appeared to be “subjective’in
nature, and that no clear scientific framework had been applied by EPA
in reaching that conclusion. The absence of such a framework was
judged by the committee as troublesome, given that the scientific evi-
dence on the question was weak (NRC, 2011).
Since the NRC (2011) peer review, significant additional scientific
evidence has become available that addresses many of the questions
raised by the NRC Committee regarding a causal association between
formaldehyde exposure and LHM. Some of these new studies and ana-
lyses were conducted in response to the NRC (2011) comments and
recommendations, while others reflect ongoing work and updates of
studies on this topic. All add to the scientific evidence surrounding the
potential causal relationship between formaldehyde inhalation ex-
posure and LHM, and should be addressed in the critical evaluations
and integration of evidence presented in an updated IRIS Assessment.
Also since the NRC (2011) peer review, the EPA has proposed en-
hancements to the IRIS process (EPA, 2013b) that incorporate many of
the general recommendations made by the NRC (2011) related to
methodological issues. This process involves the evaluation and
synthesis of evidence within separate streams of evidence (human,
animal and mechanistic). However, in a critical review of the process
conducted by a separate NRC Committee, while there was improvement
in guidelines for evaluation and synthesis of evidence within an
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
486
evidence stream, the NRC Committee still noted limitations in synthe-
sizing or integrating evidence across streams or categories (NRC,
2014a).
Nearly all of the recently available evidence from multiple lines of
evidence, especially those studies that have been focused on addressing
comments from the NRC Committee reviewing the Draft IRIS
Assessment (NRC, 2011), have increased the weight of evidence fa-
voring a conclusion of a lack of a causal association between for-
maldehyde exposure and LHM. The Checkoway et al. (2015) re-analysis
using the data from the Beane Freeman et al. (2009) study was able to
address directly several questions and comments from the NRC (2011)
Committee, as the Draft IRIS Assessment (2010) was highly dependent
on this study for drawing both qualitative and quantitative conclusions
related to formaldehyde leukemogenicity and risk of LHM following
inhalation exposure to formaldehyde. The Checkoway et al. (2015)
reanalysis provides several results and insights relevant for assessing
the risk of specific LHM. Not the least of these, the AML-specific results
provide no support for the conclusion that formaldehyde causes AML.
Associations seen between formaldehyde exposure and Hodgkin lym-
phoma and CML are inconsistent with other studies and also lack a
plausible biological mechanism (Checkoway et al., 2015). NTP (2011)
also noted that because the evidence for Hodgkin lymphoma is mainly
limited to the NCI cohort study, a causal association cannot be estab-
lished. No other LHM was associated with either cumulative or peak
formaldehyde exposure. These results of the fuller analysis of the data
from Beane Freeman et al. (2009) are consistent with recent epide-
miological studies (Meyers et al., 2013; Saberi Hosnijeh et al. 2013;
Talibov et al., 2014) which report no significant increase in LHM,
specifically AML, among cohorts of workers exposed to formaldehyde.
The available animal evidence did not support a causal association
between formaldehyde exposure and LHM at the time the Draft IRIS
Assessment (EPA, 2010) was released. Since that time, additional stu-
dies have been conducted by the NTP using two sensitive assays in mice
genetically predisposed to develop cancer following short-term ex-
posure to a chemical (Morgan et al., 2017). These studies provided no
evidence of changes in endpoints related to LHM or the presence of any
LHM following exposure to high concentrations (15 ppm) of for-
maldehyde.
Studies conducted to evaluate potential mechanisms associated with
formaldehyde exposure and LHM have demonstrated a lack of evidence
for exogenous formaldehyde to move beyond the portal of entry.
Multiple studies conducted in multiple species using highly sensitive
techniques (Edrissi et al., 2013; Lu et al., 2011; Moeller et al., 2011; Yu
et al., 2015) have demonstrated that while endogenous formaldehyde is
present in all tissues, exogenous formaldehyde following inhalation
exposure is not transported systemically. While some mechanisms for
the development of LHM following inhalation exposure to for-
maldehyde have been hypothesized (EPA, 2010; Zhang et al., 2009,
2010a), there is no evidence to support these proposed mechanisms and
the NRC Committee noted that:
“Although EPA postulated that formaldehyde could reach the bone
marrow either as methanediol or as a byproduct of nonenzymatic
reactions with glutathione, numerous studies described above have
demonstrated that systemic delivery of formaldehyde is highly un-
likely at concentrations below those which overwhelm metabolism
according to sensitive and selective analytic methods that can dif-
ferentiate endogenous from exogenous exposures.”(NRC, 2011;
page 45)
The more recent research all but confirms this. Several modes of
action have been proposed, relying primarily on data reported by Zhang
et al. (2010a) as well as subsequent evaluations of the same population
of Chinese workers (Bassig et al., 2016; Lan et al., 2015; Seow et al.,
2015). These include a mode of action in which risk of ML is increased
due to immune suppression resulting from formaldehyde exposure
(Bassig et al., 2016; Seow et al., 2015). The speculated modes of action,
however, assume systemic delivery of formaldehyde except one, which
is a hypothesized mode of action in which hematopoietic cells in the
nasal epithelium that are impacted by exposure to formaldehyde return
to the bone marrow. The NRC Committee considered this proposed
mode of action and concluded that:
“As a result, EPA could only speculate that circulating hemato-
poietic stem cells that percolate through nasal capillary beds or
nasal-associated lymphoid tissues may be the target cells for muta-
tions and clastogenic effects that eventually result in lymphohe-
motopoietic cancers. Experimental evidence of [this] mechanism is
lacking.”(NRC, 2011; page 45)
This currently leaves no acceptable proposed mode of action for the
development of LHM following inhalation exposure to formaldehyde
that can be scientifically substantiated.
The available toxicokinetic data also do not support the transport of
inhaled formaldehyde from the portal of entry. The studies by
Swenberg and colleagues unequivocally demonstrate that exogenous
formaldehyde exposure does not increase formaldehyde concentrations
measured in any internal tissues over those in unexposed animals, i.e.,
endogenously produced formaldehyde is the predominant if not only
source of internal formaldehyde (Edrissi et al., 2013; Lu et al., 2010,
2011; Moeller et al., 2011; Swenberg et al., 2011; Yu et al., 2015).
The biological plausibility of a mode of action for the development
of LHM following inhalation exposure to formaldehyde has relied
heavily upon the incompletely reported results from the Zhang et al.
(2010a) study in which the authors report differences between groups
of formaldehyde exposed and unexposed groups in the frequency of
monosomy 7 (loss of chromosome) and trisomy 8 (gain of chromo-
some), based on metaphase spreads prepared from culture of CFU-GM
colony cells. However, reanalysis of the underlying raw data in two
studies (Gentry et al., 2013; Mundt et al., 2017) have identified
methodological problems with this study that challenge these conclu-
sions, as well as demonstrate a lack of association between level of
formaldehyde exposure and the observed aneuploidy (or any of the
haematological measures).
Overall, the quality and amount of evidence relevant to the un-
derstanding of a potential causal relationship between formaldehyde
inhalation exposure and risk of LHM has increased substantially since
the completion of the Draft IRIS Assessment (EPA, 2010) and release of
the NRC peer review (NRC, 2011). New evidence has been published in
each of the major streams of evidence (i.e., human, animal and me-
chanistic) that consistently indicates a lack of a causal association be-
tween formaldehyde exposure and LHM, and specifically AML. These
new studies have addressed many of the NRC (2011) scientific criti-
cisms surrounding the evaluation of a combination of cancer types, as
well as increased our understanding of the potential impact of exo-
genous exposure on endogenous levels, which is critical in attempting
to understand the potential hazards or risks from formaldehyde ex-
posure. Regardless of which of the several similar approaches to in-
tegrating the available evidence between formaldehyde inhalation ex-
posure and the potential for leukemia risk, there is at most only limited
suggestive positive evidence, in contrast with the bulk of evidence
suggesting no such association. Therefore, a conclusion of causation is
not justified scientifically. The scientific landscape into which EPA will
release its long-anticipated revised IRIS Toxicological Review of For-
maldehyde –Inhalation Assessment is very different from that of the 2010
Draft IRIS Assessment, both in terms of improved methodological ap-
proaches and the available epidemiological, toxicological and me-
chanistic evidence. Given formaldehyde's commercial importance,
ubiquity in the environment and endogenous production, accurate de-
termination of whether occupational, residential, or consumer exposure
to formaldehyde causes leukemia or any type of human neoplasm is
critical.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
487
Funding
This work was supported in part by Hexion, Inc., a leading manu-
facturer of thermoset resins, based in Columbus, Ohio USA. No em-
ployee or representative of Hexion, Inc. participated in the preparation
of this article, or influenced its contents or interpretations, which are
exclusively those of the authors.
Acknowledgements
The authors gratefully acknowledge the valuable contributions and
insights of Dr. Michael J. Thirman, University of Chicago Medical
Center, on clinical and research aspects of leukemia etiology. We also
thank the reviewers for their detailed and very helpful comments.
Transparency document
Transparency document related to this article can be found online at
http://dx.doi.org/10.1016/j.yrtph.2017.11.006.
References
Acheson, E.D., Barnes, H.R., Gardner, M.J., Osmond, C., Pannett, B., Taylor, C.P., 1984.
Formaldehyde in the British chemical industry. An occupational cohort study. Lancet
1 (8377), 611–616.
Adami, H.O., Berry, S.C., Breckenridge, C., Smith, L., Swenberg, J., Trichopoulos, D.,
Weiss, N.S., Pastoor, T., 2011. Toxicology and epidemiology: improving the science
with a framework for combining toxicological and epidemiological evidence to es-
tablish causal inference. Toxicol. Sci. 122 (2), 223–234.
Albertini, R.J., Kaden, D.A., 2016. Do chromosome changes in blood cells implicate
formaldehyde as a leukemogen? Crit. Rev. Toxicol. 47 (2), 1–40. http://dx.doi.org/
10.1080/10408444.10402016.11211987.
American College of Medical Genetics, 2006. Standards and Guidelines for Clinical
Genetics Laboratories, 2006 Edition. Section E. Clinical Cytogenetics. Availalble at.
https://www.acmg.net/Pages/ACMG_Activities/stds-2002/stdsmenu-n.htm Accessed
20 October 2017.
Appelman, L.M., Woutersen, R.A., Zwart, A., Falke, H.E., Feron, V.J., 1988. One-year
inhalation toxicity study of formaldehyde in male rats with a damaged or undamaged
nasal mucosa. J. Appl. Toxicol. JAT 8 (2), 85–90.
Baan, R., Grosse, Y., Straif, K., Secretan, B., El, G.F., Bouvard, V., et al., 2009. A review of
human carcinogens–Part F: chemical agents and related occupations. Lancet Oncol.
10 (12), 1143–1144.
Bachand, A.M., Mundt, K.A., Mundt, D.J., Montgomery, R.R., 2010. Epidemiological
studies of formaldehyde exposure and risk of leukemia and nasopharyngeal cancer: a
meta-analysis. Crit. Rev. Toxicol. 40 (2), 85–100.
Bailer, A.J., Portier, C.J., 1988. Effects of treatment-induced mortality and tumor-induced
mortality on tests for carcinogenicity in small samples. Biometrics 44, 417–431.
Bassig, B.A., Zhang, L., Vermeulen, R., Tang, X., Li, G., Hu, W., Guo, W., Purdue, M.P.,
Yin, S., Rappaport, S.M., Shen, M., Ji, Z., Qiu, C., Ge, Y., Hosgood, H.D., Reiss, B., Wu,
B., Xie, Y., Li, L., Yue, F., Freeman, L.E., Blair, A., Hayes, R.B., Huang, H., Smith, M.T.,
Rothman, N., Lan, Q., 2016. Comparison of hematological alterations and markers of
B-cell activatin in workers exposed to benzene, formaldehyde and trichloroethylene.
Carcinogenesis 37 (7), 692–700.
Battelle Columbus Laboratories, 1981. Final Report on a Chronic Inhalation Toxicology
Study in Rats and Mice Exposed to Formaldehyde. Prepared by Battelle Columbus
Laboratories, Columbus, OH, for the Chemical Industry Institute of Toxicology (CIIT),
Research Triangle Park, NC CIIT Docket No. 10922.
Beane Freeman, L.E., Blair, A., Lubin, J.H., Stewart, P.A., Hayes, R.B., Hoover, R.N.,
Hauptmann, M., 2009. Mortality from lymphohematopoietic malignancies among
workers in formaldehyde industries: the national cancer Institute cohort. J. Natl.
Cancer Inst. 101 (10), 751–761.
Blair, A., Stewart, P., O'Berg, M., Gaffey, W.R., Walrath, J., Ward, J., Bales, R., Kaplan, S.,
Cubit, D., 1986. Mortality among industrial workers exposed to formaldehyde.
J. Natl. Cancer Inst. 76 (6), 1071–1084.
Blair, A., Zheng, T., Linos, A., Stewart, P., Zhang, Y., Cantor, K., 2000. Occupation and
leukemia: a population-based case-control study in Iowa and Minnesota. Am. J. Ind.
Med. 40, 3–14.
Blair, A., Zheng, T., Linos, A., Stewart, P.A., Zhang, Y.W., Cantor, K.P., 2001. Occupation
and leukemia: a population-based case-control study in Iowa and Minnesota. Am. J.
Ind. Med. 40 (1), 3–14.
Bolt, H.M., Johnson, G., Nielsen, G.D., Papameletiou, D., Klein, C.L., 2016. SCOEL/REC/
125 Formaldehyde Recommendation from the Scientific Committee on Occupational
Exposure Limits. Adopted 30 June 2016. Publicatins Office of the European Union,
Luxembourg.
Bosetti, C., McLaughlin, J.K., Tarone, R.E., Pira, E., La Vecchia, C., 2008. Formaldehyde
and cancer risk: a quantitative review of cohort studies through 2006. Ann. Oncol. 19
(1), 29–43.
Bundesinstitut für Risikobewertung (BfR-Wissenschaft), 2006. Assessment of the
Carcinogenicity of Formaldehyde [CAS No. 50-00-0]. Available at. http://www.bfr.
bund.de/cm/350/assessment_of_the_carcinogenicity_of_formaldehyde.pdf.
Casanova-Schmitz, M., Starr, T.B., Heck, H.D., 1984. Differentiation between metabolic
incorporation and covalent binding in the labeling of macromolecules in the rat nasal
mucosa and bone marrow by inhaled [14C]- and [3H]formaldehyde. Toxicol. Appl.
Pharmacol. 76 (1), 26–44.
Casanova, M., Heck, H.D., Everitt, J.I., Harrington Jr., W.W., Popp, J.A., 1988.
Formaldehyde concentrations in the blood of rhesus monkeys after inhalation ex-
posure. Food Chem. Toxicol. 26 (8), 715–716.
Checkoway, H., Boffetta, P., Mundt, D.J., Mundt, K.A., 2012. Critical review and synthesis
of the epidemiologic evidence on formaldehyde exposure and risk of leukemia and
other lymphohematopoietic malignancies. Cancer Causes Control 23 (11),
1747–1766.
Checkoway, H., Dell, L.D., Boffetta, P., Gallagher, A.E., Crawford, L., Lees, P.S., Mundt,
K.A., 2015. Formaldehyde exposure and mortality risks from acute myeloid leukemia
and other lymphohematopoietic malignancies in the US national cancer Institute
cohort study of workers in formaldehyde industries. J. Occup. Environ. Med. 57 (7),
785–794.
Coggon, D., Harris, E.C., Poole, J., Palmer, K.T., 2003. Extended follow-up of a cohort of
british chemical workers exposed to formaldehyde. J. Natl. Cancer Inst. 95 (21),
1608–1615.
Coggon, D., Ntani, G., Harris, E.C., Palmer, K.T., 2014. Upper airway cancer, myeloid
leukemia, and other cancers in a cohort of british chemical workers exposed to for-
maldehyde. Am. J. Epidemiol. 179 (11), 1301–1311 additional charts).
Cogliano, V.J., Grosse, Y., Baan, R.A., Straif, K., Secretan, M.B., El, G.F., 2005. Meeting
report: summary of IARC monographs on formaldehyde, 2-butoxyethanol, and 1-tert-
butoxy-2-propanol. Environ. Health Perspect. 113 (9), 1205–1208.
Collins, J.J., Lineker, G.A., 2004. A review and meta-analysis of formaldehyde exposure
and leukemia. Regul. Toxicol. Pharmacol. 40 (2), 81–91.
Cole, P., Adami, H.O., Trichopoulos, D., Mandel, J.S., 2010a. Re: mortality from lym-
phohematopoietic malignancies and brain cancer among embalmers exposed to for-
maldehyde. J. Natl. Cancer Inst. 102 (19), 1518–1519.
Cole, P., Adami, H.O., Trichopoulos, D., Mandel, J., 2010b. Formaldehyde and lympho-
hematopoietic cancers: a review of two recent studies. Regul. Toxicol. Pharmacol. 58
(2), 161–166.
Dean, J.H., Lauer, L.D., House, R.V., Murray, M.J., Stillman, W.S., Irons, R.D., Steinhagen,
W.H., Phelps, M.C., Adams, D.O., 1984. Studies of immune function and host re-
sistance in B6C3F1 mice exposed to formaldehyde. Toxicol. Appl. Pharmacol. 72 (3),
519–529.
Edrissi, B., Taghizadeh, K., Moeller, B.C., Kracko, D., Doyle-Eisele, M., Swenberg, J.A.,
Dedon, P.C., 2013. Dosimetry of N
6
-formyllysine adducts following [
13
C
2
H
2
]-for-
maldehyde exposures in rats. Chem. Res. Toxicol. 26 (10), 1421–1423.
EPA, 1991. Formaldehyde; CASRN 50-00-0. Integrated Risk Information System (IRIS)
Chemical Assessment Summary. National Center for Environmental Assessment,
Washington, D.C Available online at. http://www.epa.gov/iris.
EPA, 2000. Toxicological Review of Vinyl Chloride (CAS No. 75-01-4) in Support of
Summary Information on the Integrated Risk Information System (IRIS). EPA/635R-
00/0004. Environmental Protection Agency, Washington, DC May 2000.
EPA, 2010. Toxicological Review of Formaldehyde - Inhalation Assessment (CAS No. 50-
00-0). U.S. Environmental Protection Agency, Washington, DC June 2010. www.epa.
gov/iris.
EPA, 2013a. Materials Submitted to the National Research Council Part I: Status of
Implementation of Recommendations. U.S. Environmental Protection Agency:
Integrated Risk Information System Program Available at. https://www.epa.gov/
sites/production/files/201406/documents/iris_program_materials_to_nrc_part_1.pdf
Accessed 17 May 2017.
EPA, 2013b. Enhancements to EPA's Integrated Risk Information System Program.
Available at. https://www.epa.gov/sites/production/files/2014-06/documents/
irisprocessfactsheet2013.pdf.
EPA, 2015. EPA's Integrated Risk Information System (IRIS) Program: Progress Report
and Report to Congress. November 2015. Available at. https://www.epa.gov/sites/
production/files/2015-12/documents/iris_report_to_congress_nov2015.pdf Accessed
30 March 2017.
Gardner, M.J., Pannett, B., Winter, P.D., Cruddas, A.M., 1993. A cohort study of workers
exposed to formaldehyde in the British chemical industry: an update. Br. J. Ind. Med.
50 (9), 827–834.
Gentry, P.R., Rodricks, J.V., Turnbull, D., Bachand, A., Van Landingham, C., Shipp, A.M.,
Albertini, R.J., Irons, R., 2013. Formaldehyde exposure and leukemia: critical review
and reevaluation of the results from a study that is the focus for evidence of biological
plausibility. Crit. Rev. Toxicol. 43 (8), 661–670.
Gluzman, D.F., Sklyarenko, L.M., Zavelevich, M.P., Koval, S.V., Ivanivskaya, T.S., 2015.
Leukemic blast cells and controversies in models of hematopoiesis. Exp. Oncol.
37, 2–4.
Golden, R., Pyatt, D., Shields, P.G., 2006. Formaldehyde as a potential human leuke-
mogen: an assessment of biological plausibility. Crit. Rev. Toxicol. 36 (2), 135–153.
Goldstein, B.D., 2010. Benzene as a cause of lymphoproliferative disorders. Chem. Biol.
Interact. 184, 147–150.
Greaves, M.F., 2004. Biological Models for Leukaemia and Lymphoma IARC Sci Publ, vol.
157. pp. 351–372.
Greenland, S., Longnecker, M.P., 1992. Methods for trend estimation from summarized
dose-response data, with applications to meta-analysis. Am. J. Epidemiol. 135,
1301–1309.
Hall, A., Harrington, J.M., Aw, T.C., 1991. Mortality study of British pathologists. Am. J.
Ind. Med. 20 (1), 83–89.
Hauptmann, M., Lubin, J.H., Stewart, P.A., Hayes, R.B., Blair, A., 2003. Mortality from
lymphohematopoietic malignancies among workers in formaldehyde industries. J.
Natl. Cancer Inst. 95 (21), 1615–1623.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
488
Hauptmann, M., Lubin, J.H., Stewart, P.A., Hayes, R.B., Blair, A., 2004. Mortality from
solid cancers among workers in formaldehyde industries. Am. J. Epidemiol. 159 (12),
1117–1130.
Hauptmann, M., Stewart, P.A., Lubin, J.H., Beane Freeman, L.E., Hornung, R.W., Herrick,
R.F., Hoover, R.N., Fraumeni, J.F., Blair, A., Hayes, R.B., 2009. Mortality from
lymphohematopoietic malignancies and brain cancer among embalmers exposed to
formaldehyde. J. Natl. Cancer Inst. 101 (24), 1696–1708.
Hayes, R.B., Blair, A., Stewart, P.A., Herrick, R.F., Mahar, H., 1990. Mortality of U.S.
embalmers and funeral directors. Am. J. Ind. Med. 18 (6), 641–652.
Heck, H.D., Casanova-Schmitz, M., Dodd, P.B., Schachter, E.N., Witek, T.J., Tosun, T.,
1985. Formaldehyde (CH2O) concentrations in the blood of humans and Fischer-344
rats exposed to CH2O under controlled conditions. Am. Ind. Hyg. Assoc. J. 46
(1), 1–3.
Heck, H.D., Casanova, M., 2004. The implausibility of leukemia induction by for-
maldehyde: a critical review of the biological evidence on distant-site toxicity. Regul.
Toxicol. Pharmacol. 40 (2), 92–106.
IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans, 1982a.
Chemicals, Industrial Processes, and Industries Associated with Cancer in Humans: an
Updating of IARC Monographs Volumes 1 to 29. Supplement 4. World Health
Organization; International Agency for Research on Cancer, Lyon, France.
IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans, 1982b.
Some Industrial Chemicals and Dyestuffs. Vol. 29, 1982. World Health Organization;
International Agency for Research on Cancer, Lyon, France.
IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans, 1987.
Overall Evaluations of Carcinogenicity: an Updating of IARC Monographs Volumes 1
to 42. Supplement 7. World Health Organization; International Agency for Research
on Cancer, Lyon, France.
IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans, 1995.
Wood Dust and Formaldehyde, vol. 62 World Health Organization; International
Agency for Research on Cancer, Lyon, France 1995.
IARC. IARC Monographs on the Evaluation of the Carcinogenic Risks to Humans, 2006.
Formaldehyde, 2-Butoxyethanol and 1-tert-butoxypropan-2-ol, vol. 88 World Health
Organization (WHO); International Agency for Research on Cancer, Lyon, France.
IARC, 2012. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. A
Review of Human Carcinogens Part F: Chemical Agents and Related Occupations, vol.
100 World Health Organization (WHO); International Agency for Research on
Cancer, Lyon, France.
Infante, P.F., Rinsky, R.A., Wagoner, J.K., Young, R.J., 1977. Leukaemia in benzene
workers. Lancet 2 (8028), 76–78.
IOM. Institute of Medicine, 2011. Finding what Works in Health Care: Standards for
Systematic Reviews. DC. National Academies Press, Washington.
Johannsen, F.R., Levinskas, G.J., Tegeris, A.S., 1986. Effects of formaldehyde in the rat
and dog following oral exposure. Toxicol. Ltrs 30 (1), 1–6.
Kamata, E., Nakadate, M., Uchida, O., Ogawa, Y., Suzuki, S., Kaneko, T., Saito, M.,
Kurokawa, Y., 1997. Results of a 28-month chronic inhalation toxicity study of for-
maldehyde in male Fisher-344 rats. J. Toxicol. Sci. 22 (3), 239–254.
Kerns, W.D., Pavkov, K.L., Donofrio, D.J., Gralla, E.J., Swenberg, J.A., 1983.
Carcinogenicity of formaldehyde in rats and mice after long-term inhalation ex-
posure. Cancer Res. 43 (9), 4382–4392.
Kleinnijenhuis, A.J., Staal, Y.C., Duistermaat, E., Engel, R., Woutersen, R.A., 2013. The
determination of exogenous formaldehyde in blood of rats during and after inhalation
exposure. Food Chem. Toxicol. 52, 105–112.
Lai, Y., Yu, R., Hartwell, H.J., Moeller, B.C., Bodnar, W.M., Swenberg, J.A., 2016.
Measurement of endogenous versus exogenous formaldehyde-induced DNA-protein
crosslinks in animal tissues by stable isotope labeling and ultrasensitive mass spec-
trometry. Cancer Res. 76 (9), 2652–2661.
Lan, Q., Smith, M.T., Tang, X., Guo, W., Vermeulen, R., Ji, Z., Hu, W., Hubbard, A.E.,
Shen, M., McHale, C.M., Qiu, C., Liu, S., Reiss, B., Beane-Freeman, L., Blair, A., Ge, Y.,
Xiong, J., Li, L., Rappaport, S.M., Huang, H., Rothman, N., Zhang, L., 2015.
Chromosome-wide aneuploidy study of cultured circulating myeloid progenitor cells
from workers occupationally exposed to formaldehyde. Carcinogenesis 36 (1),
160–167.
Lavelle, K.S., Schnatter, A.R., Travis, K.Z., Swaen, G.M., Pallapies, D., Money, C., Priem,
P., Vrijhof, H., 2012. Framework for integrating human and animal data in chemical
risk assessment. Regul. Toxicol. Pharmacol. 62 (2), 302–312.
Levine, R.J., Andjelkovich, D.A., Shaw, L.K., 1984. The mortality of Ontario undertakers
and a review of formaldehyde-related mortality studies. J. Occup. Med. 26 (10),
740–746.
Linkov, I., Massey, O., Keisler, J., Rusyn, I., Hartung, T., 2015. From “weight of Evidence”
to Quantitative Data Integration Using Multicriteria Decision Analysis and Bayesian
Methods. Altex 32.1:3.
Lu, K., Collins, L.B., Ru, H., Bermudez, E., Swenberg, J.A., 2010. Distribution of DNA
adducts caused by inhaled formaldehyde is consistent with induction of nasal car-
cinoma but not leukemia. Toxicol. Sci. 116 (2), 441–451.
Lu, K., Moeller, B., Doyle-Eisele, M., McDonald, J., Swenberg, J.A., 2011. Molecular
dosimetry of N2-Hydroxymethyl-dG DNA adducts in rats exposed to formaldehyde.
Chem. Res. Toxicol. 24 (2), 159–161.
Marsh, G.M., Morfeld, P., Collins, J.J., Symons, J.M., 2014. Issues of methods and in-
terpretation in the National Cancer Institute formaldehyde cohort study. J. Occup.
Med. Toxicol. 9, 22.
Matanoski, G., 1989. Risk of Pathologists Exposed to Formaldehyde. Final Report. John
Hopkins University; Department of Epidemiology, School of Hygiene and Public
Health, Baltimore, MD.
Meyers, A.R., Pinkerton, L.E., Hein, M.J., 2013. Cohort mortality study of garment in-
dustry workers exposed to formaldehyde: update and internal comparisons. Am. J.
Ind. Med. 59 (9), 1027–1039.
Moeller, B.C., Lu, K., Doyle-Eisele, M., McDonald, J., Gigliotti, A., Swenberg, J.A., 2011.
Determination of N2-hydroxymethyl-dG adducts in the nasal epithelium and bone
marrow of nonhuman primates following 13CD2-formaldehyde inhalation exposure.
Chem. Res. Toxicol. 24 (2), 162–164.
Morgan, D.L., Dixon, D., King, D.H., Travlos, G.S., Herbert, R.A., French, J.E., Tokar, E.J.,
Waalkes, M.P., Jokinen, M.P., 2017. NTP Research Report on Absence of
Formaldehyde-induced Neoplasia in Trp53 Haploinsufficient Mice Exposed by
Inhalation. NTP RR 3. National Toxicology Program, Research Triangle Park, NC,
pp. 1–30 (3).
Mundt, K.A., Gallagher, A.E., Dell, L.D., Natelson, E.A., Boffetta, P., Gentry, P.R., 2017.
Does occupational exposure to formaldehyde cause hematotoxicity and leukemia-
specific chromosome changes in cultured myeloid progenitor cells? Crit. Rev. Toxicol.
47 (7), 592–602.
Nielsen, G.D., Larsen, S.T., Wolkoff, P., 2013. Recent trend in risk assessment of for-
maldehyde exposures from indoor air. Arch. Toxicol. 87 (1), 73–98.
Nielsen, G.D., Larsen, S.T., Wolkoff, P., 2017. Re-evaluation of the WHO (2010) for-
maldehyde indoor air quality guideline for cancer risk assessment. Arch. Toxicol. 91
(1), 35–61.
NRC (National Research Council), 2011. Review of the Environmental Protection
Agency's Draft IRIS Assessment of Formaldehyde. DC. The National Academies Press,
Washington.
NRC, 2014a. Review of EPA's Integrated Risk Information System (IRIS) Process. The
National Academies Press, Washington, D.C.
NRC, 2014b. Review of the Formaldehyde Assessment in the National Toxicology
Program 12th Report on Carcinogens Washington, DC. National Academies Press,
978-0-309-31227-1pp. 1–224.
NTP (National Toxicology Program), 1981. Second Annual Report on Carcinogens.
December 1981. U.S. Department of Health and Human Services, Public Health
Service, Research Triangle Park, NC.
NTP, 1982. Third Annual Report on Carcinogens. December 1982. U.S. Department of
Health and Human Services, Public Health Service, Research Triangle Park, NC.
NTP, 2011. Report on Carcinogens. twelfth ed. U.S. Department of Health and Human
Services, Public Health Service, Research Triangle Park, NC.
Peddada, S.D., Kissling, G.E., 2006. A survival-adjusted quantal-response test for anlysis
of tumor incidence rates in animal carcinogenicity studies. Environ. Health Perspect.
114 (4), 537–541.
Pinkerton, L.E., Hein, M.J., Stayner, L.T., 2004. Mortality among a cohort of garment
workers exposed to formaldehyde: an update. Occup. Environ. Med. 61 (3), 193–200.
Pira, E., Romano, C., Verga, F., La, V.C., 2014. Mortality from lymphohematopoietic
neoplasms and other causes in a cohort of laminated plastic workers exposed to
formaldehyde. Cancer Causes Control 25 (10), 1343–1349.
Pira, E., Romano, C., La Vecchia, C., Boffetta, P., 2017. Hematologic and Cytogenetic
Biomarkers of Formaldehyde Exposure and Leukemia Risk [Letter to the Editor].
Published Online. Carcinogenesis bgx072. https://doi.org/10.1093/carcin/bgx072.
Pontel, L.B., Rosado, I.V., Burgos-Barragan, G., Garaycoechea, J.I., Yu, R., Arends, M.J.,
Chandrasekaran, G., Broecker, V., Wei, W., Liu, L., Swenberg, J.A., Crossan, G.P.,
Patel, K.J., 2015. Endogenous formaldehyde is a hematopoietic stem cell genotoxin
and metabolic carcinogen. Mol. Cell. 60 (1), 177–188.
Pyatt, D., Natelson, E., Golden, R., 2008. Is inhalation exposure to formaldehyde a bio-
logically plausible cause of lymphohematopoietic malignancies? Regul. Toxicol.
Pharmacol. 51 (1), 119–133.
RAC (Risk Assessment Committee), 2012. Opinion Proposing Harmonised Clasification
and Labelling at EU Level of Formaldehyde. European Chemicals Agency, Helsinki 30
November 2012. https://echa.europa.eu/documents/10162/254a73cf-ff8d-4bf4-
95d1-109f13ef0f5a Accessed 30 March 2017.
Rhomberg, L.R., 2015a. Contrasting directions and directives on hazard identification for
formaldehyde carcinogenicity. Regul. Toxicol. Pharmacol. 73 (3), 829–833.
Rhomberg, L., 2015b. Hypothesis-based weight of evidence: an approach to assessing
causation and its application to regulatory toxicology. Risk Anal. 35 (6), 1114–1124.
Rhomberg, L.R., Bailey, L.A., Goodman, J.E., Hamade, A.K., Mayfield, D., 2011. Is ex-
posure to formaldehyde in air causally associated with leukemia?–A hypothesis-based
weight-of-evidence analysis. Crit. Rev. Toxicol. 41 (7), 555–621.
Rooney, A.A., Boyles, A.L., Wolfe, M.S., Bucher, J.R., Thayer, K.A., 2014. Systematic
review and evidence integration for literature-based environmental health science
assessments. Environ. Health Perspect. 122 (7), 711–718.
Rothman, N., Lan, Q., Smith, M.T., Vermeulen, R., Zhang, L., 2017. Response to Letter to
the Editor of Carcinogenesis by Pira et al., 2017 regarding Lan, et al. 2015,
Carcinogenesis. Published Online: Carcinogenesis, bgx111. https://doi.org/10.
1093/carcin/bgx111.
Saberi Hosnijeh, F., Christopher, Y., Peeters, P., Romieu, I., Xun, W., Riboli, E., Raaschou-
Nielsen, O., Tjønneland, A., Becker, N., Nieters, A., Trichopoulou, A., Bamia, C.,
Orfanos, P., Oddone, E., Luján-Barroso, L., Dorronsoro, M., Navarro, C., Barricarte,
A., Molina-Montes, E., Wareham, N., Vineis, P., Vermeulen, R., 2013. Occupation and
risk of lymphoid and myeloid leukaemia in the european prospective investigation
into cancer and nutrition (EPIC). Occup. Environ. Med. 70 (7), 464–470.
Schnatter, R.A., Glass, D.C., Tang, G., Irons, R.D., Rushton, L., 2012. Myelodysplastic
syndrome and benzene exposure among petroleum workers: an international pooled
analysis. J. Natl. Cancer Inst. 104, 1724–1737.
Schroeter, J.D., Campbell, J., Kimbell, J.S., Conolly, R.B., Clewell, H.J., Andersen, M.E.,
2014. Effects of endogenous formaldehyde in nasal tissues on inhaled formaldehyde
dosimetry predictions in the rat, monkey, and human nasal passages. Toxicol. Sci.
138 (2), 412–424.
Schwilk, E., Zhang, L., Smith, M.T., Smith, A.H., Steinmaus, C., 2010. Formaldehyde and
leukemia: an updated meta-analysis and evaluation of bias. J. Occup. Environ. Med.
52 (9), 878–886.
Sellakumar, A.R., Snyder, C.A., Solomon, J.J., Albert, R.E., 1985. Carcinogenicity of
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
489
formaldehyde and hydrogen chloride in rats. Toxicol. Appl. Pharmacol. 81 (3 Pt 1),
401–406.
Seow, W.J., Zhang, L., Vermeulen, R., Tang, X., Hu, W., Bassig, B.A., Ji, Z., Shiels, M.S.,
Kemp, T.J., Shen, M., Qiu, C., Reiss, B., Beane Freeman, L.E., Blair, A., Kim, C., Guo,
W., Wen, C., Li, L., Pinto, L.A., Huang, H., Smith, M.T., Hildesheim, A., Rothman, N.,
Lan, Q., 2015. Circulating immune/inflammation markers in Chinese workers occu-
pationally exposed to formaldehyde. Carcinogenesis 36 (8), 852–857.
Soffritti, M., Belpoggi, F., Lambertin, L., Lauriola, M., Padovani, M., Maltoni, C., 2002.
Results of long-term experimental studies on the carcinogenicity of formaldehyde and
acetaldehyde in rats. Ann. N. Y. Acad. Sci. 982, 87–105.
Speit, G., Gelbke, H.-P., Pallapies, D., Morfeld, P., 2010. Occupational exposure to for-
maldehyde, hematotoxicity and leukemia-specific chromosome changes in cultured
myeloid progenitor cells [Letter to the Editor]. Cancer Epidemiol. Biomarkers Prev.
19 (7), 1882–1884.
Starr, T.B., Swenberg, J.A., 2013. A novel bottom-up approach to bounding low-dose
human cancer risks from chemical exposures. Regul. Toxicol. Pharmacol. 65 (3),
311–315.
Starr, T.B., Swenberg, J.A., 2016. The bottom-up approach to bounding potential low-
dose cancer risks from formaldehyde: an update. Regul. Toxicol. Pharmacol. 77,
167–174.
Stayner, L.T., 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 (6), 667–681.
Stewart, P.A., Blair, A., Cubit, D., Bales, R., Kaplan, S.A., Ward, J., Gaffey, W., O'Berg,
M.T., Walrath, J., 1986. Estimating historical exposures to formaldehyde in a retro-
spective mortality study. Appl. Ind. Hyg. 1 (1), 34–41.
Stroup, N.E., Blair, A., Erikson, G.E., 1986. Brain cancer and other causes of death in
anatomists. J. Natl. Cancer Inst. 77 (6), 1217–1224.
Swenberg, J.A., Kerns, W.D., Mitchell, R.I., Gralla, E.J., Pavkov, K.L., 1980. Induction of
squamous cell carcinomas of the rat nasal cavity by inhalation exposure to for-
maldehyde vapor. Cancer Res. 40 (9), 3398–3402.
Swenberg, J.A., Lu, K., Moeller, B.C., Gao, L., Upton, P.B., Nakamura, J., Starr, T.B., 2011.
Endogenous versus exogenous DNA adducts: their role in carcinogenesis, epide-
miology, and risk assessment. Toxicol. Sci. 120 (Suppl. 1), S130–S145.
Swenberg, J.A., Moeller, B.C., Lu, K., Rager, J.E., Fry, R.C., Starr, T.B., 2013.
Formaldehyde carcinogenicity research: 30 Years and counting for mode of action,
epidemiology, and cancer risk assessment. Toxicol. Pathol. 41 (2), 181–189.
Talibov, M., Lehtinen-Jacks, S., Martinsen, J.I., Kjaerheim, K., Lynge, E., Sparén, P.,
Tryggvadottir, L., Weiderpass, E., Kauppinen, T., Kyyrönen, P., Pukkala, E., 2014.
Occupational exposure to solvents and acute myeloid leukemia: a population-based,
case-control study in four Nordic countries. Scand. J. Work Environ. Health 40 (5),
511–517.
Tarone, R., 1975. Tests for trend in life table analysis. Biometrika 62, 679–682 as cited in
Battelle Columbus Laboratories [1981].
Til, H.P., Woutersen, R.A., Feron, V.J., Clary, J.J., 1988. Evaluation of the oral toxicity of
acetaldehyde and formaldehyde in a 4-week drinking-water study in rats. Food
Chem. Toxicol. 26, 447–452.
Til, H.P., Woutersen, R.A., Feron, V.J., Hollanders, V.H., Falke, H.E., Clary, J.J., 1989.
Two-year drinking-water study of formaldehyde in rats. Food Chem. Toxicol. 27 (2),
77–87.
Tobe, M., Naito, K., Kurokawa, Y., 1989. Chronic toxicity study on formaldehyde ad-
ministered orally to rats. Toxicology 56 (1), 79–86.
U.S. House, 2011. Committee on Appropriations. Department of the Interior,
Environment, and Related Agencies Appropriation Bill, 2012, (To Accompany H.R.
2584) Together with Dissenting Views. H. Rept. 112-151. GPO, Washington Available
at. https://www.gpo.gov/fdsys/pkg/CRPT-112hrpt151/pdf/CRPT-112hrpt151.pdf.
U.S. House, 2014. Committee on Appropriations. Department of the Interior,
Environment, and Related Agencies Appropriation Bill, 2015, (To Accompany H.R.
5171) Together with Minority Views. H. Rept. 113-551. GPO, Washington Available
at. https://www.gpo.gov/fdsys/pkg/CRPT-113hrpt551/html/CRPT-113hrpt551.
htm.
U.S. House, 2016. Committee on Appropriations. Department of the Interior,
Environment, and Relatd Agencies Appropriation Bill, 2017, (To Accompany H.R.
5538) Together with Dissenting and Separate Views.H. Rept. 114–632. GPO,
Washington Available at. https://www.gpo.gov/fdsys/pkg/CRPT-114hrpt632/html/
CRPT-114hrpt632.htm.
U.S Senate, 2016. Committee on Appropriations. Department of the Interior,
Environment, and Relatd Agencies Appropriations Bill, 2017, (To Accompany S.
3068). S. Rept. 114–281. GPO, Washington Available at. https://www.gpo.gov/
fdsys/pkg/CRPT-114srpt281/html/CRPT-114srpt281.htm.
Van Landingham, C., Mundt, K.A., Allen, B., Gentry, P., 2016. The need for transparency
and reproducibility in documenting values for regulatory decision making and
evaluating causality: the example of formaldehyde. Regul. Toxicol. Pharmacol. 81,
512–521.
Vargová, M., Wagnerová, J., Lísková, A., Jakubovský, J., Gajdová, M., Stolcová, E.,
Kubová, J., Tulinská, J., Stenclová, R., 1993. Subacute immunotoxicity study of
formaldehyde in male rats. Drug Chem. Toxicol. 16 (3), 255–275.
Walrath, J., Fraumeni Jr., J.F., 1983. Mortality patterns among embalmers. Int. J. Cancer
31 (4), 407–411.
Walrath, J., Fraumeni, J.F., 1984. Cancer and other causes of death among embalmers.
Cancer Res. 44 (10), 4638–4641.
WHO, World Health Organization, 2010. Kaden, D.A., Mandin, C., Nielsen, G.D., Wolkoff,
P. Formaldehyde. In: WHO Guidelines for Indoor Air Quality: Selected Pollutants.
Europe.
Woodruff, T.J., Sutton, P., 2014. The navigation guide systematic review methodology: a
rigorous and transparent method for translating environmental health science into
better health outcomes. Environ. Health Perspect. 122 (10), 1007.
Woutersen, R.A., Appelman, L.M., Wilmer, J.W., Falke, H.E., Feron, V.J., 1987.
Subchronic (13-week) inhalation toxicity study of formaldehyde in rats. J. Appl.
Toxicol. 7 (1), 43–49.
Yu, R., Lai, Y., Hartwell, H.J., Moeller, B.C., Doyle-Eisele, M., Kracko, D., Bodnar, W.M.,
Starr, T.B., Swenberg, J.A., 2015. Formation, accumulation, and hydrolysis of en-
dogenous and exogenous formaldehyde-induced DNA damage. Toxicol. Sci. 146 (1),
170–182.
Zhang, L., Lan, Q., Guo, W., Li, G., Yang, W., Hubbard, A.E., Vermeulen, R., Rappaport,
S.M., Yin, S., Rothman, N., Smith, M.T., 2005. Use of OctoChrome fluorescence in situ
hybridization to detect specific aneuploidy among all 24 chromosomes in benzene-
exposed workers. Chem. Biol. Interact. 153–154, 117–122.
Zhang, L., Steinmaus, C., Eastmond, D.A., Xin, X.K., Smith, M.T., 2009. Formaldehyde
exposure and leukemia: a new meta-analysis and potential mechanisms. Mutat. Res.
681 (2–3), 150–168.
Zhang, L., Tang, X., Rothman, N., Vermeulen, R., Ji, Z., Shen, M., Qiu, C., Guo, W., Liu, S.,
Reiss, B., Beane Freeman, L., Ge, Y., Hubbard, A.E., Hua, M., Blair, A., Galvan, N.,
Ruan, X., Alter, B.P., Xin, K.X., Li, S., Moore, L.E., Kim, S., Xie, Y., Hayes, R.B.,
Azuma, M., Hauptmann, M., Xiong, J., Stewart, P., Li, L., Rappaport, S.M., Huang, H.,
Fraumeni Jr., J.F., Smith, M.T., Lan, Q., 2010a. Occupational exposure to for-
maldehyde, hematotoxicity, and leukemia-specific chromosome changes in cultured
myeloid progenitor cells. Cancer Epidemiol. Biomarkers Prev. 19 (1), 80–88.
Zhang, L., Ji, Z., Guo, W., Hubbard, A.E., Galvan, N., Xin, K.X., Azuma, M., Smith, M.T.,
Tang, X., Qiu, C., Ge, Y., Hua, M., Ruan, X., Li, S., Xie, Y., Li, L., Huang, H., Rothman,
N., Shen, M., Beane Freeman, L., Blair, A., Alter, B.P., Moore, L.E., Hayes, R.B.,
Hauptmann, M., Stewart, P., Fraumeni Jr., J.F., Lan, Q., Vermeulen, R., Reiss, B., Liu,
S., Xiong, J., Kim, S., Rappaport, S.M., 2010b. Occupational exposure to for-
maldehyde, hematotoxicity, and leukemia-specific chromosome changes in cultured
myeloid progenitor cells. Response to Letter to Editor by Speit, et al.. Cancer
Epidemiol. Biomarkers Prev. 19 (7), 1884–1885.
Zhang, L., Lan, Q., Guo, W., Hubbard, A.E., Li, G., Rappaport, S.M., McHale, C.M., Shen,
M., Ji, Z., Vermeulen, R., Yin, S., Rothman, N., Smith, M.T., 2011. Chromosome-wide
aneuploidy study (CWAS) in workers exposed to an established leukemogen, ben-
zene. Carcinogenesis 32, 605–612.
K.A. Mundt et al. Regulatory Toxicology and Pharmacology 92 (2018) 472–490
490
