American Journal of Epidemiology
ª The Author 2011. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: [email protected].
Vol. 173, No. 6
DOI: 10.1093/aje/kwq426
Advance Access publication:
February 18, 2011
Original Contribution
Effect of Mustard Gas Exposure on Incidence of Lung Cancer: A Longitudinal
Study
Mihoko Doi, Noboru Hattori*, Akihito Yokoyama, Yojiro Onari, Masashi Kanehara, Kenji Masuda,
Tetsuji Tonda, Megu Ohtaki, and Nobuoki Kohno
* Correspondence to Dr. Noboru Hattori, Department of Molecular and Internal Medicine, Graduate School of Biomedical Sciences,
Hiroshima University, 1-2-3 Kasumi, Minami-ku, Hiroshima, 734-8551, Japan (e-mail: [email protected]).
Initially submitted April 11, 2010; accepted for publication November 9, 2010.
Sulfur mustard, an agent used in chemical warfare, is an alkylating substance with carcinogenic potential.
However, the precise long-term carcinogenic effects of mustard gas are unclear. Since 1952, the authors have
conducted health surveys of former workers who were employed from 1929 to 1945 in a poisonous gas factory in
Okuno-jima, Hiroshima, Japan. This prospective study was undertaken from 1952 to 2005 to examine the incidence of lung cancer among the workers who were exposed to mustard gas (n¼ 480), lewisite (n¼ 55), and/or
diphenylcyanarsine (n ¼ 178), as well as the incidence among unexposed workers (n ¼ 969). The stochastic
relation between exposure and lung cancer was explored on the basis of multistage carcinogenesis by using an
accelerated hazard model with a transformed age scale. Mustard gas exposure was found to transform the age
scale for developing lung cancer. One year of exposure in subjects 18 or >18 years old at first exposure shifted
the age scale down by 4.9 years and 3.3 years, respectively. On the basis of the long-term follow-up of former
workers in the poisonous gas factory, the authors concluded that sulfur mustard decreased the age at which people
were at risk of developing lung cancer and that the effect declined with aging.
carcinogens; cohort studies; inhalation exposure; lung neoplasms; mustard gas
Abbreviation: CI, confidence interval.
whereas <1 death (actual expected value ¼ 0.9) was predicted on the basis of Japanese national mortality rates in
males with the same age distribution as the workers.
Vesicant agents, such as mustard gas and the organic
arsenical lewisite, are easily dispersed in the air and cause
severe chemical burns upon direct contact with tissue. Moist
tissues such as the eyes, respiratory tract, and axilla are
particularly affected, and in Okuno-jima, erythema and
edema of the skin, as well as conjunctivitis, were frequently
observed in the former workers when their protective clothing was breached (11, 12).
At the factory in Okuno-jima, sulfur mustard (bis(2chloroethyl)sulfide) was produced at a peak rate of 450
tons/month (12). Sulfur mustard gas is a DNA-alkylating
agent that is extremely cytotoxic at low concentrations
and causes gene mutations in a dose-dependent fashion.
Formation of O6-ethylthioethyl guanine induced by mustard
Previous epidemiologic studies of workers at poisonous
gas factories (1–7) and soldiers who served during World
War I (8, 9) have demonstrated that exposure to mustard gas
can cause lung cancer. Mustard gas is classified as a group 1
carcinogen by the International Agency for Research on
Cancer. On Okuno-jima, a small island in Hiroshima Prefecture in Japan, a poisonous gas factory established by the
Japanese army was in operation from 1929 until 1945. Mustard gas was the main product, but lewisite (chlorvinylarsine), diphenylcyanarsine, hydrocyanic acid, phosgene, and
chloracetophenone were also produced. We previously reported that a number of former workers suffered from acute
injuries, including erosion and bulla formation in the skin,
as well as chronic bronchitis and lung cancer after a period
of latency (1–6, 10, 11). In 1968, Wada et al. (1) showed that
33 of the workers who were engaged in manufacturing mustard gas died from cancers of the respiratory system,
659
Am J Epidemiol 2011;173:659–666
660 Doi et al.
gas results in a poor substrate for the DNA repair system,
and DNA alkylation seems to be primarily responsible for
the mutagenic consequences of cellular exposure (6, 13).
Tumorigenesis of sulfur mustard has been confirmed by an
increasing occurrence of pulmonary tumors via inhalation
and injection in laboratory mice (6, 14).
In Okuno-jima, the arsenical gases diphenylcyanarsine
and lewisite were also manufactured. Arsenic is categorized
as a respiratory carcinogen by the International Agency for
Research on Cancer, and inorganic arsenic acid and its derivatives have been implicated as carcinogens of the skin
and the respiratory system. In contrast to the evidence about
mustard gas, however, the evidence that organic arsenicals,
including lewisite, are carcinogenic, mutagenic, or teratogenic is very weak (14, 15).
For over 50 years, we have investigated and provided
health care to the former workers employed in the poisonous
gas factory in Okuno-jima. In the present study, to clarify
the transitional effects of poison gases (mustard gas, lewisite, and diphenylcyanarsine) on the incidence of lung cancer, we conducted a prospective survey in which the risk of
lung cancer was analyzed in the former factory workers who
were exposed or unexposed to these gases.
MATERIALS AND METHODS
Study population and data collection
Because factory employment records were missing, subjects who worked in the poisonous gas factory were identified
through the use of several alternative methods. Since 1966,
questionnaires about work history at the factory have been
distributed in a house-to-house canvass of the area where most
of the former workers were thought to have lived, and spot
announcements were repeated on television throughout Hiroshima Prefecture to urge former factory employees to contact the investigators. Workers who came forward were asked
whether they had information about other persons who had
worked at the factory. In addition, persons who were admitted
to hospitals in the area surrounding Okuno-jima were asked
whether they had worked at the poisonous gas factory. Since
1952, persons identified as having worked at the factory have
been invited to receive a medical examination at least once
a year at registered hospitals, and financial support for their
medical care and/or living costs has been provided by the
government based on the severity of their disorder due to
poisonous gas exposure. The identified workers were registered, and complete records of their medical examinations
were maintained in our department. Vital status and cause
of death were determined by death certificates and notifications from the hospitals or public health authorities. As of
March 31, 2005, there were a total of 6,851 subjects registered
as former workers at the poisonous gas factory in Okuno-jima.
Study cohorts
Because birth-cohort trends in lung cancer varied considerably depending on gender and geographic area (16), we
restricted the study sample to males who had not lived outside
of Hiroshima Prefecture since retirement from the poisonous
gas factory. Person-years were calculated using the period
from the date of entry as a worker at the factory to the earliest
of either March 31, 2005, the date of death, or the last day of
follow-up. Only individuals who could be followed for >2
years were included in the cohort. Subjects were excluded if
a diagnosis of lung cancer preceded the date of entry into the
study. Information on smoking status was obtained from
a clinical interview or chart review and was assessed repeatedly during the follow-up period. To analyze the effects of
poisonous gas exposure and smoking on lung cancer incidence, subjects for whom information about smoking history
was missing were excluded. Lung cancer morbidity rates
were derived from clinical records, postmortem examinations, or notification from the hospital or public health authority. Moreover, the definition of lung cancer for purposes
of our analysis was restricted to cases determined by pathologic confirmation by histologic or cytologic examinations;
diagnoses based only on medical certificates or imaging were
excluded. Tumor histology was determined using lung biopsy specimens obtained by using bronchoscopy, computed
tomography-guided aspiration, or surgical resection.
Exposed and control groups
We had information on the job titles of the subjects who
were identified as former workers at the poisonous gas factory in Okuno-jima. On the basis of this information, we first
selected the subjects who directly engaged in manufacturing
poisonous gases, called ‘‘manufacturers.’’ The manufacturers
were categorized into 5 groups on the basis of the gases with
which they dealt: 1) mustard gas, 2) lewisite, 3) both mustard
gas and lewisite, 4) diphenylcyanarsine, and 5) other gases,
such as hydrocyanic acid, phosgene, and chloracetophenone.
Because only mustard gas, lewisite, and diphenylcyanarsine
are established carcinogens, the exposed group was selected
from the subjects categorized in groups 1, 2, and 4 and then
classified into subcohorts consisting of subjects who were
engaged in producing mustard gas, lewisite, or diphenylcyanarsine. The subjects categorized in group 3 were excluded
from the exposed group. As the control group, we selected
subjects from the manufacturers categorized in group 5 and
former workers whose job titles were classified into activities
other than manufacturing poison gases, as follows: carriers,
construction workers, clerks, cleaning men, and medical
staff. The ambient concentration of mustard gas in the factory
is estimated to have ranged from 50 mg/m3 to 70 mg/m3 (12,
14). Furthermore, acid corrosion, equipment breakdown, and
exhaust-fan troubles occurred frequently. A considerable
number of workers seem to have continued working under
these conditions, with constant or frequent exposure to small
amounts of poisonous gases (12). In the present study, we
assumed that the manufacturers had constant exposure to the
poisonous gases during their employment, so we utilized
duration of employment as a surrogate for level of exposure
to poisonous gases.
Statistical analysis
Armitage and Doll (17) established the multistage model
of carcinogenesis, which states that the accumulation of
Am J Epidemiol 2011;173:659–666
Influence of Mustard Gas on Lung Cancer Incidence
661
Enrollment up to March 31, 2005
(n = 6,851)
Females
(n = 3,006)
Males
(n = 3,845)
Inhabitants outside Hiroshima Prefecture
(n = 614)
Inhabitants of Hiroshima Prefecture
(n = 3,231)
Inappropriate follow-up criteria
(n = 390)
Appropriate follow-up criteria
(n = 2,841)
Unclassifiable subjects
(n = 1,159)
Control group
(n = 969)
Mustard gas
(n = 480)
Lewisite
(n = 55)
Diphenylcyanarsine
(n = 178)
Exposed Group
Figure 1. Identification of the study cohort in a prospective study of the effect of poisonous gas exposure on incidence of lung cancer in former
workers at a poisonous gas factory, Hiroshima, Japan, 1952–2005. Subjects were excluded if the diagnosis of lung cancer preceded the date of
entry into the study and/or the follow-up period was <2 years. Unclassifiable subjects were subjects with unidentified occupation (n ¼ 1),
employees who manufactured both mustard gas and lewisite (n ¼ 86), and employees other than manufacturers who may have been exposed
to gas (n ¼ 1,072).
mutations in target cells explains the age-time patterns of
excess cancer. They considered cancer to be the end result
of the accumulation in a normal cell of a critical number (k)
of independent mutations through a series of intermediate
states (17). We modified this basic multistage model to formulate an estimate of whether exposure to a specific mutagen shifts the age scale, that is, whether the exposure
accelerates the age at risk of carcinogenesis (18, 19). The
risk ratio between exposed and unexposed subjects increases in proportion to the intensity of exposure and decreases with time after exposure. This model conforms
remarkably well to observations from cohort studies of the
atomic-bomb survivors, miners with prolonged exposure to
radon, and cigarette smokers who stopped smoking at various ages (20). In the present study, we applied the generalized Armitage-Doll multistage model (18, 19) in which
age, duration of exposure, and age at first exposure are
specified as effect modifiers and smoking status is treated
as a main effect in the baseline rate model. When a subject
born at time b receives a continuous exposure of dose D
beginning at age a, the hazard of lung cancer at time t is
expressed as the function h(tjb, a, D, C) a (t b w þ
baD þ cC)k1, where ba is the coefficient of the effect of gas
exposure, c is the coefficient of smoking status C (1, current
or ex-smoker; 0, nonsmoker), and w is the tumor growth
time (the time from the final step in carcinogenesis to a clinically detected tumor). As suggested by Collins et al. (21),
the number of cancer cells increases exponentially during
division. However, as described in a previous report from
Am J Epidemiol 2011;173:659–666
our group (22), a certain proportion of cancer cells die in the
process of tumor growth; therefore, in many deep-growthtype cancers, there is a quite large difference between the
generation time (about 7 days) and the doubling time (1–3
months) of cancer cells. Judging from the data in the previous report, we set the doubling time of lung cancer cells
as 2 months. We also defined a clinically detectable tumor
in the lung as one having a diameter of 10 mm, which is
considered to contain 230 cancer cells. On the basis of these
assumptions, we calculated the amount of time it takes a single cancer cell to proliferate into 230 cells and concluded
that w was 5 years. The unknown parameters (k, ba, c) were
estimated by maximizing a partial likelihood (23), and k was
optimized with the constraint that it be an integer value. The
exposure accelerates the aging process, and the relative risk,
the rate ratio between the exposed and control groups, is
expressed as relative risk ¼ {1 þ (baD þ cC)/(t b w)}k1. The relative risk was considered to be significantly
different from 1.0 when its 95% confidence interval did not
include 1.0.
This study was approved by the Hiroshima University
Ethics Committee.
RESULTS
Study population and characteristics of the cohorts
Figure 1 illustrates the subject selection process used in
this study. To exclude the effects of gender and
662 Doi et al.
Table 1. Baseline Characteristics of the Exposure Group and Controls in Former Workers at a Poisonous Gas
Factory, Hiroshima, Japan, 1952–2005
Exposure Group (n 5 713)
Controls
(n 5 969)
Mean (SD)
Mustard Gas
(n 5 480)
%
Mean (SD)
Lewisite
(n 5 55)
%
Mean (SD)
Diphenylcyanarsine
(n 5 178)
%
Mean (SD)
Age at first employment,
years
19.3 (8.6)
25.4 (8.6)
21.7 (7.4)
22.8 (7.1)
Duration of employment,
years
1.2 (2.2)
4.0 (3.1)
3.0 (2.2)
1.9 (1.5)
18.9 (11.3)
24.1 (10.8)
26.8 (11.6)
22.5 (10.1)
Period of observation,
years
%
Smoking history
Smoker
40.9
33.3
41.8
39.9
Ex-smoker
38.0
25.8
29.1
30.9
Nonsmoker
Unknown
8.8
6.3
9.1
5.6
12.4
34.6
20.0
23.6
Abbreviation: SD, standard deviation.
environmental carcinogens such as air pollutants, we restricted the study population to 3,231 men who had never
lived outside of Hiroshima Prefecture after retirement from
the poisonous gas factory. Among them, 2,841 men met the
criteria of having a follow-up time >2 years. Moreover,
some subjects were excluded from the analysis because they
were considered unclassifiable: 1 whose occupation at the
factory was not verifiable, 86 who manufactured both mustard gas and lewisite, and 1,072 who may have had contact
with poisonous gases (factory inspection, incineration, and
destruction of the factory or disposal of poisonous gases
after the war) but did not work in the manufacture of poisonous gases. Consequently, 1,682 subjects were assigned to
the exposed and control groups. The exposed group included subjects who were engaged in manufacturing carcinogenic gases and was divided into 3 subcohorts: mustard
gas (480 subjects), lewisite (55 subjects), and diphenylcyanarsine (178 subjects). The control group consisted of 969
subjects who were not engaged in the production of carcinogenic gases. The characteristics of each group are shown
in Table 1. Age at employment was younger and duration of
employment and length of follow-up were shorter in the
control group than in each of the exposed subcohorts (Table
1). Smoking history was obtained from about 80% of the
subjects. The proportion of subjects with unknown smoking
history was higher in the cohort of workers exposed to mustard gas than in the other groups.
Lung cancer incidence in 10-year intervals
Table 2 shows incidence rates and histologic classifications of lung cancers observed in each group in 10-year
intervals from 1955 to 2005. In the control group, the incidence of lung cancer showed a marked increase in recent
years; on the other hand, among the workers exposed to
mustard gas, the incidence of lung cancer was apparently
higher than in the controls in the early years and showed
a slow increase in recent years (Table 2). Table 3 shows vital
statistics and incidence rates of lung cancer assessed on
March 31, 2005, in the subjects exposed to mustard gas
and the control group, by age at first employment. As of
March 31, 2005, only 17% of subjects in the mustard gas
group had survived, but more than half of the subjects in the
control group were alive. During follow-up, a total of 77
incidences of lung cancer occurred in the mustard gas (n ¼
39) and control (n ¼ 38) groups. When the incidence was
stratified by age at first employment (10-year strata), the rate
of lung cancer was higher in subjects exposed to mustard
gas than in the control group (Table 3).
Effects of gas exposure and smoking on the age scale
Under the assumption that malignancy occurs upon the
kth mutation, we estimated the relative contributions of
carcinogenic-gas exposure and smoking to the carcinogenic
process of lung cancer. We analyzed the following 2 groups:
the entire gas exposure group, which consisted of subjects
exposed to mustard gas, lewisite, or diphenylcyanarsine, and
the group of subjects exposed only to mustard gas. Using the
multistage model of carcinogenesis modified with an accelerating effect, we assumed that an accumulation of 6 mutations was required in both the entire gas exposure group and
the mustard gas only exposure group (Table 4). In the entire
gas exposure group, the effect of gas exposure for 1 year was
estimated to shift the age scale down by 5.2 years (95%
confidence interval (CI): 2.1, 8.2) and 2.9 years (95% CI:
0.7, 5.0) for subjects whose ages at first exposure were 18
or >18 years, respectively, after adjustment for attained age
and smoking status. In the mustard-gas-only exposure group,
the effect of gas exposure for 1 year was estimated to shift
the age scale by 4.9 years (95% CI: 1.7, 7.9) for those 18
years of age at first exposure and 3.3 years (95% CI: 1.2,
5.5) for those >18 years of age at first exposure. On the
other hand, the effect of smoking was estimated to shift
the age scale by 26.2 years in the entire gas exposure group
and 23.5 years in the mustard-gas-only exposure group after
Am J Epidemiol 2011;173:659–666
Influence of Mustard Gas on Lung Cancer Incidence
663
Table 2. Incidence Rates and Histologic Findings for Lung Cancer in 10-Year Intervals in Former Workers
Employed From 1929 to 1945 in the Poisonous Gas Factory in Okuno-Jima, Hiroshima, Japan, 1955–2005
Time Period
1955–1964
1965–1974
1975–1984
1985–1994
1995–2005
Controls
Person-years
No. of cases/incidence
rate per 1,000 person-years
9,695
9,578
9,106
7,816
4,829
0/0
0/0
5/0.55
8/1.02
25/5.18
1/1/3/0
5/3/0/0
13/8/2/2
Sq/Ad/Sm/others
No. of cases
Exposed workers
Mustard gas
Person-years
No. of cases/incidence
rate per 1,000 person-years
4,745
4,431
3,573
2,423
1,069
3/0.63
7/1.58
13/3.64
10/4.13
6/5.61
1/1/1/0
2/0/3/2
8/3/1/1
6/0/3/1
2/1/2/1
550
536
481
420
209
0/0
1/1.86
0/0
1/2.38
0/0
Sq/Ad/Sm/others
No. of cases
Lewisite
Person-years
No. of cases/incidence
rate per 1,000 person-years
Sq/Ad/Sm/others
No. of cases
1/0/0/0
1/0/0/0
Diphenylcyanarsine
Person-years
No. of cases/incidence rate
per 1,000 person-years
1,781
1,751
1,570
1,202
504
0/0
0/0
1/0.64
5/4.16
1/1.98
1/0/0/0
3/0/2/0
1/0/0/0
Sq/Ad/Sm/others
No. of cases
Abbreviations: Ad, adenocarcinoma; Sm, small-cell carcinoma; Sq, squamous-cell carcinoma.
adjustment for attained age and the effect of gas exposure.
An analysis limited to subjects exposed to lewisite or diphenylcyanarsine could not be conducted because the numbers
of subjects in those groups were too small. To illustrate the
age-dependent effects of exposure to only mustard gas for 1
year on the incidence of lung cancer, we estimated the
relation between attained age and relative risk for subjects whose ages at first exposure were 18 or >18 years
(Figure 2). The estimated relative risk decreased with increasing attained age and was higher in the subjects whose
age at first exposure was 18 years.
DISCUSSION
In the present prospective study, we demonstrated that exposure to mustard gas accelerated the age at risk of developing lung cancer in former poisonous gas factory workers. The
relative risk of lung cancer in those who were engaged in the
manufacture of mustard gas decreased substantially with increasing age at start of exposure and with attained age, similar
to what has been seen in the pattern of excess risk of solid
cancer among the atomic-bomb survivors (20, 24).
In the multistage model of carcinogenesis for most cancers, accumulation of 5–7 mutations in a normal stem cell
is thought to result in the development of malignant cells
Am J Epidemiol 2011;173:659–666
(25). In the present study, it was estimated that cancer results
after the accumulation of 6 mutations (Table 4, k) in the
former poisonous gas factory workers who suffered from
lung cancer. This finding is supported by the presence of
epithelial hyperplasia and dysplasia, as well as by the fact
that lung cancer was frequently observed in the bronchial
trees of the workers (5). These pathologic findings suggest
the occurrence of multistep carcinogenesis beginning with
hyperplasia, moving through metaplasia, dysplasia, and in
situ carcinoma, and ultimately becoming invasive cancer
(6). In addition, genomic instability in preneoplastic lesions
and the presence of p53 mutations in lung cancers have been
reported in the former workers (26). On the other hand,
Tokuoka et al. (5) reported significant correlations between
the incidence of atypical lesions and mustard gas exposure
in a multivariate analysis, though the incidence of atypical
lesions was also influenced significantly by age, smoking,
and chronic bronchitis. In the present study, we explored the
relative effects of exposure to the carcinogenic gases and
smoking on the age-specific incidence of lung cancer. Exposure to mustard gas was shown to act as a mutagen with
a long-term effect on the development of lung cancer that
was related to cumulative exposure.
Regarding the association between mustard gas exposure
in the former workers of the factory at Okuno-jima and the
664 Doi et al.
Table 3. Vital Status and Incidence Rates of Lung Cancer as of
March 31, 2005, by Age at Employment (Age at First Exposure), in
Former Workers Employed From 1929 to 1945 in the Poisonous Gas
Factory in Okuno-Jima, Hiroshima, Japan
Age at First
Employment,
years
Alive
Subjects,
No. of
No. of Person
Cases
Cases Years
Table 4. Estimated Number of Mutations Required for Lung Cancer
and the Accelerating Effects of Gas Exposure and Smoking on Aging
Scale in a Prospective Study of the Effect of Poison Gas Exposure on
Incidence of Lung Cancer, Hiroshima, Japan, 1952–2005
Lung Cancer
Incidence
No. of
Rate Per 1,000
Cases
Person Years
0–9
Mustard gas
0
0
0
0
0
Control
9
8
454
0
0
Mustard gas
167
71
7,448
12
1.61
Control
664
459
32,287
30
0.93
Mustard gas
164
12
6,155
13
2.11
Control
149
28
6,155
4
0.65
10–19
Entire Gas
Exposure
ka
6
6
bab at age 18 years
Year
5.2
4.9
95% CI
2.1, 8.2
1.7, 7.9
Year
2.9
3.3
95% CI
0.7, 5.0
1.2, 5.5
bab at age >18 years
cc
20–29
Year
30–39
Mustard Gas
Exposure
95% CI
26.2
8.1, 57.8
23.5
6.3, 53.1
Abbreviation: CI, confidence interval.
k is the estimated number of mutations required for lung cancer.
b
ba is the aging effect of gas exposure for 1 year in the subjects
whose ages at the first gas exposure were 18 years and >18 years,
and is adjusted for attained age and smoking status.
c
c is the aging effect of the presence of the smoking history, and is
adjusted for attained age and gas exposure.
a
Mustard gas
113
0
3,341
13
3.89
Control
107
5
3,811
4
1.05
40–49
Mustard gas
30
0
644
0
0
Control
36
0/
1,145
0/
0
Mustard gas
3
0
62
0
0
Control
1
0
45
0
0
Mustard gas
3
0
36
1
Control
3
0
44
0
50
Unknown
27.81
0
Total
Mustard gas
480
83
17,686
39
2.21
Control
969
500
43,943
38
0.86
histologic types of lung cancer, there have been several
previous reports (1–5) demonstrating that squamous-cell
and undifferentiated carcinomas distributed from the upper
airway to the hilar region predominated, whereas adenocarcinoma and other types were rare. As shown in Table 2,
however, the relative proportion of squamous-cell carcinoma did not differ between the workers with mustard gas
exposure (19 squamous-cell carcinomas out of 39 lung cancers) and the control group (19 squamous-cell carcinomas
out of 38 lung cancers). We believe that this discrepancy is
mainly caused by the inclusion criteria. The survey of former poisonous gas factory workers in our department started
in 1952 after a 30-year-old man who had worked at the
factory for 16 months beginning in 1942 was diagnosed with
bronchogenic cancer (1, 2). Because employment records
were missing, identification of former workers at the factory
was difficult, and enrollment of former workers into the
survey was insufficient until 1966, when the distribution
of questionnaires regarding work history began in
a house-to-house canvass of the area where most of the
former factory workers were thought to have lived. As
a consequence, persons with major cases of respiratory carcinoma reported in the previous studies from our department
were recognized as former workers at the factory after the
diagnosis of lung cancer or were not followed for sufficient
periods of time and, thus, were excluded from the present
study. In addition, to show the agent-specific effect on the
incidence of lung cancer, we excluded workers who were
engaged in producing both mustard gas and lewisite. This
exclusion may have resulted in an underestimation of the
occurrence of lung cancer.
Figure 2. Relative risk of lung cancer among former workers with
mustard gas exposure in Hiroshima, Japan, by age at employment,1952–2005.
Am J Epidemiol 2011;173:659–666
Influence of Mustard Gas on Lung Cancer Incidence
Although ethnic and racial differences have been reported,
smoking is the most well-established risk factor for lung
cancer (27, 28). The effects of smoking vary depending on
the amount and duration of smoking, age at starting or quitting smoking (29), and the product smoked (cigarettes, cigars,
or pipes). We sought detailed smoking information from all
subjects included in the cohort, but it was fragmentary and of
limited accuracy except for presence or absence of smoking.
Therefore, we limited our investigation to the effect of this
simple variable. Smoking data were more frequently missing
for subjects in the group exposed to mustard gas because
subjects in the earlier period of the survey comprised many
of those with mustard gas exposure, and smoking information
from the medical chart was frequently insufficient in the
earlier period. Because most subjects smoked, we reanalyzed
the data by labeling as a ‘‘smoker’’ all subjects with missing
smoking data; the result was essentially the same. In general,
long-term smoking elevates the risk of lung cancer 10–30fold over one’s lifetime compared with not smoking (30), and
the occurrence of lung cancer in smokers rises in the middleto-late 40s (31). In the present study, smoking was found to
independently shift the age scale of lung cancer occurrence
down by 24–26 years in comparison with never smokers.
When this result was referred to the accelerated hazard model
on the transformed age scale used in the previous study that
investigated the effects of smoking cessation on lung cancer
mortality in a cohort of 1,000,000 subjects (20), it was estimated that smoking duration was about 30 years in that
cohort.
There are several limitations to the present study. First,
we failed to include workers who developed lung cancer in
the early period after the start of the survey. This is mainly
because of inability of the system to enroll former workers
and is likely to have resulted in underascertainment of lung
cancers. Second, our control group might also have been
slightly exposed to poisonous gases because of the lack of
industrial hygiene in the factories, which led to contamination by poisonous gases of the air, water, and soil around the
factory (12). Thus, the effect of poisonous gases on the incidence of lung cancer in the exposed group might be underestimated. Moreover, we need to consider the unobserved
interindividual variability arising from either exposure to carcinogens, such as mustard gas or smoking, and background
characteristics, because the stochastic model we applied was
based on an assumption of population homogeneity.
To the best of our knowledge, this is the first study to report
a long-term effect of poisonous gas exposure on the incidence of
lung cancer. In the Iran-Iraq conflict between 1980 and 1988,
approximately 45,000 military and civilian casualties were associated with sulfur mustard gas, and not only acute consequences after inhalation but also a series of chronic destructive
pulmonary sequelae have been reported (32). We hope that our
present investigation will be beneficial for the future health care
of all persons exposed to these chemical agents.
ACKNOWLEDGMENTS
Author affiliations: Department of Molecular and Internal
Medicine, Graduate School of Biomedical Sciences, Hiroshima
Am J Epidemiol 2011;173:659–666
665
University, Hiroshima, Japan (Mihoko Doi, Noboru Hattori,
Yojiro Onari, Masashi Kanehara, Kenji Masuda, Nobuoki
Kohno); Department of Hematology and Respiratory Medicine,
Kochi University, Kochi, Japan (Akihito Yokoyama); and
Department of Environmetrics and Biometrics, Research Institute for Radiation Biology and Medicine, Hiroshima University,
Hiroshima, Japan (Tetsuji Tonda, Megu Ohtaki).
This work was partly supported by a grant for the group
conducting ‘‘Research on the Aftereffects and Prognosis of
Poison Gas Injuries’’ from the Ministry of Health, Labour,
and Welfare, Japan.
The authors thank all contributing medical staff who have
provided health care to the former workers in the poisonous
gas factory, especially Dr. Masato Yukutake.
Conflict of interest: none declared.
REFERENCES
1. Wada S, Miyanishi M, Nishimoto Y, et al. Mustard gas as
a cause of respiratory neoplasia in man. Lancet. 1968;1(7553):
1161–1163.
2. Yamada A. On the late injuries following occupational inhalation of mustard gas, with special references to carcinoma of
the respiratory tract. Acta Pathol Jpn. 1963;13(3):131–155.
3. Nishimoto Y, Yamakido M, Ishioka S, et al. Epidemiological
studies of lung cancer in Japanese mustard gas workers.
Princess Takamatsu Symp. 1987;18:95–101.
4. Yamakido M, Ishioka S, Hiyama K, et al. Former poison gas
workers and cancer: incidence and inhibition of tumor formation by treatment with biological response modifier NCWS. Environ Health Perspect. 1996;104(suppl 3):
S485–S488.
5. Tokuoka S, Hayashi Y, Inai K, et al. Early cancer and related
lesions in the bronchial epithelium in former workers of
mustard gas factory. Acta Pathol Jpn. 1986;36(4):533–542.
6. Inai K, Takeshima Y. Pathology of respiratory tract in former
workers in mustard gas factory. A review. J Toxicol Pathol.
1995;8(4):343–353.
7. Easton DF, Peto J, Doll R. Cancers of the respiratory tract in
mustard gas workers. Br J Ind Med. 1988;45(10):652–659.
8. Case RA, Lea AJ. Mustard gas poisoning, chronic bronchitis,
and lung cancer; an investigation into the possibility that
poisoning by mustard gas in the 1914–18 war might be a factor
in the production of neoplasia. Br J Prev Soc Med. 1955;
9(2):62–72.
9. Beebe GW. Lung cancer in World War I veterans: possible
relation to mustard-gas injury and 1918 influenza epidemic.
J Natl Cancer Inst. 1960;25:1231–1252.
10. Nishimoto Y, Burrows B, Miyanishi M, et al. Chronic obstructive lung disease in Japanese poison gas workers. Am Rev
Respir Dis. 1970;102(2):173–179.
11. Inada S, Hiragun K, Seo K, et al. Multiple Bowen’s disease
observed in former workers of a poison gas factory in Japan,
with special reference to mustard gas exposure. J Dermatol.
1978;5(2):49–60.
12. Wada S, Nishimoto Y, Miyanishi M, et al. Review of OkunoJima poison gas factory regarding occupational environment.
Hiroshima J Med Sci. 1962;11(3):75–80.
13. Ludlum DB, Kent S, Mehta JR. Formation of O6-ethylthioethylguanine in DNA by reaction with the sulfur mustard,
chloroethyl sulfide, and its apparent lack of repair by
666 Doi et al.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
O6-alkylguanine-DNA alkyltransferase. Carcinogenesis.
1986;7(7):1203–1206.
Watson AP, Griffin GD. Toxicity of vesicant agents scheduled
for destruction by the chemical stockpile disposal program.
Environ Health Perspect. 1992;98:259–280.
Goldman M, Dacre JC. Lewisite: its chemistry, toxicology, and
biological effects. Rev Environ Contam Toxicol. 1989;110:
75–115.
Sobue T, Tsukuma H, Oshima A, et al. Lung cancer incidence
rates by histologic type in high- and low-risk areas;
a population-based study in Osaka, Okinawa, and Saku
Nagano. Japan. J Epidemiol. 1999;9(3):134–142.
Armitage P, Doll R. The age distribution of cancer and a multistage theory of carcinogenesis. Br J Cancer. 1954;8(1):1–12.
Ohtaki M, Fujita S, Hayakawa N, et al. The age distribution of
human adult cancer and an initiation-manifestation model for
carcinogenesis. Jpn J Clin Oncol. 1985;15(suppl 1):S325–S343.
Ohtaki M, Niwa O. A mathematical model of radiation carcinogenesis with induction of genomic instability and cell
death. Radiat Res. 2001;156(5):672–677.
Pierce DA, Vaeth M. Age-time patterns of cancer to be anticipated from exposure to general mutagens. Biostatistics.
2003;4(2):231–248.
Collins VP, Loeffler RK, Tivey H. Observations on growth
rates of human tumors. Am J Roentgenol Radium Ther Nucl
Med. 1956;76(5):988–1000.
Ohtaki M. A stochastic model for proliferation of cancer cells
and growth of tumor—a Monte Carlo simulation study. Jpn
J Biom. 1993;14:47–54.
Cox DR. Regression models and life tables (with discussion).
J R Stat Soc B. 1972;34:187–220.
24. Preston DL, Shimizu Y, Pierce DA, et al. Studies of mortality
of atomic bomb survivors. Report 13: solid cancer and noncancer disease mortality: 1950–1997. Radiat Res. 2003;
160(4):381–407.
25. Doll R. The age distribution of cancer: implications for models
of carcinogenesis. J R Stat Soc A. 1971;134:133–155.
26. Takeshima Y, Inai K, Bennett WP, et al. p53 mutations in lung
cancers from Japanese mustard gas workers. Carcinogenesis.
1994;15(10):2075–2079.
27. Danaei G, Vander Hoorn S, Lopez AD, et al. Causes of cancer
in the world: comparative risk assessment of nine behavioural
and environmental risk factors. Comparative Risk Assessment
Collaborating Group (Cancers). Lancet. 2005;366(9499):
1784–1793.
28. Haiman CA, Stram DO, Wilkens LR, et al. Ethnic and racial
differences in the smoking-related risk of lung cancer. N Engl
J Med. 2006;354(4):333–342.
29. Lubin JH, Caporaso N, Wichmann HE, et al. Cigarette
smoking and lung cancer: modeling effect modification of
total exposure and intensity. Epidemiology. 2007;18(5):
639–648.
30. Samet JM, Wiggins CL, Humble CG, et al. Cigarette smoking
and lung cancer in New Mexico. Am Rev Respir Dis. 1988;
137(5):1110–1113.
31. Burns DM. Cigarette smoking. In: Aisner J, Perry MC,
Green MR, et al, eds. Comprehensive Textbook of Thoracic
Oncology. Baltimore, MD: Williams & Wilkins; 1996:
51–65.
32. Ghanei M, Harandi AA. Long term consequences from
exposure to sulfur mustard: a review. Inhal Toxicol. 2007;
19(5):451–456.
Am J Epidemiol 2011;173:659–666