|
3.4
Cancer
3.4.1
Background
The possible
carcinogenicity of crystalline silica
dust became a subject of considerable and intense debate in the scientific
community in the 1980s, especially after
- publication
of new information presented at a 1984 symposium in North Carolina
[Goldsmith et al. 1986],
- epidemiologic
studies by Westerholm [1980] and Finkelstein et al. [1982], and
- a
literature review by Goldsmith et al. [1982] (see McDonald [1989,
1995] and Graham [1998]).
Many epidemiologic
studies of cancer mortality and morbidity in silica-exposed occupational
groups were published later, but the issue remained unresolved. In October
1996, an IARC expert working group reviewed the published experimental
and epidemiologic studies of cancer in animals and workers exposed to
respirable crystalline silica. The
working group concluded that there is sufficient evidence in humans
for the carcinogenicity of inhaled crystalline silica in the form of
quartz or cristobalite from occupational
sources [IARC 1997]. In June 1996, the directors of the ATS adopted
an official statement of their Committee of the Scientific Assembly
on Environmental and Occupational Health. This statement, prepared at
the request of the American Lung Association Occupational Health Expert
Advisory Group [ATS 1997], described the adverse health effects of exposure
to crystalline silica, including lung cancer. The ATS found the following:
- The
available data support the conclusion that silicosis produces increased
risk for bronchogenic carcinoma.
- However,
less information is available for lung cancer risk among silicotics
who never smoked and workers who were exposed to silica but did not
have silicosis.
- Whether
silica exposure is associated with lung cancer in the absence of silicosis
is less clear.
NIOSH
concurs with the conclusions of the IARC working group and the ATS.
These conclusions agree with NIOSH testimony to OSHA, in which NIOSH
recommended that crystalline silica be considered a potentional occupational
carcinogen [54 Fed. Reg. 2521 (1989)].
This section,
like the IARC review, focuses on lung cancer and discusses the epidemiologic
studies that were the least likely to have results affected by confounding
and selection biases. In mixed environments such as ceramics,
pottery, or brick manufacturing, where exposure may be to two or more
polymorphs of crystalline silica, epidemiologic studies have usually
not identified specific exposures to quartz or cristobalite. Therefore,
excess lung cancers that occurred in these environments cannot be associated
with exposure to a given polymorph but only with exposure to respirable
crystalline silica. The epidemiologic studies of cancer have mainly
investigated workers exposed to respirable crystalline silica in
- ore
mining,
- quarrying
and granite works,
- ceramics,
pottery, glass, refractory brick, and diatomaceous earth industries,
or
- foundries.
The other
major study group was workers with silicosis, usually identified from
national or local registries. Studies of workers and silicotics that
were not discussed in this document because they failed to meet the
least confounded criterion have been criticized for the following reasons
[Checkoway 1995; McDonald 1995, 1996; Morgan and Reger 1995; Weill and
McDonald 1996; Seaton 1995; Weill et al. 1994; Agius et al. 1992]:
- Inadequate,
incomplete, or invalid exposure assessment
- Potential
selection and confounding biases in the cohort studies of compensated
silicotics
- Inadequate
control of confounding from cigarette smoking and from concurrent
workplace exposures (e.g., potential exposure to radon progeny, arsenic,
or diesel exhaust in ore mines and potential exposure to polycyclic
aromatic hydrocarbons in foundries)
- Inability
to distinguish differences in fibrogenic and carcinogenic potencies
of the various silica polymorphs
- Lack
of evidence of an exposure-response relationship
3.4.2
Epidemiologic Studies of Lung Cancer
Following
a comprehensive review of the large body of published epidemiologic
studies, IARC [1997] found that the following studies provide the least
confounded investigations of an association between occupational exposure
to crystalline silica and lung cancer:
- U.S.
gold miners [Steenland and Brown 1995b]
- Danish
stone industry workers [Gunel et al. 1989]
- U.S.
granite shed and quarry workers [Costello and Graham 1988]
- U.S.
crushed stone industry workers [Costello et al. 1995]
- U.S.
diatomaceous earth industry workers [Checkoway et al. 1993, 1996]
- Chinese
refractory brick workers [Dong et al. 1995]
- Italian
refractory brick workers [Merlo et al. 1991; Puntoni et al. 1988]
- U.K.
pottery workers [McDonald et al. 1995,1997; Cherry et al. 1995, 1997;
Burgess et al. 1997]
- Chinese
pottery workers [McLaughlin et al. 1992]
- Cohorts
of registered silicotics from North Carolina [Amandus et al. 1991,
1992] and Finland [Kurppa et al. 1986; Partanen et al. 1994]
Although
a few of these studies did not find a statistically significant association
between occupational exposure to crystalline silica and lung cancer
(Table 15), most of the studies did.
Study results are often not uniform when a large number of epidemiologic
studies are reviewed and a variety of populations and work environments
are studied [IARC 1997]. In addition, IARC noted that the carcinogenicity
of quartz or cristobalite may be dependent on inherent characteristics
of the crystalline silica or on external factors affecting its biological
activity or distribution of its polymorphs [IARC 1997].
Some of
the least confounded studies reported that lung cancer risk tended to
increase with
- cumulative
exposure to respirable silica [i.e., Checkoway et al. 1993, 1996],
- duration
of exposure [i.e., Merlo et al. 1991; Partanen et al. 1994; Costello
and Graham 1988; Costello et al. 1995; Dong et al. 1995],
- peak
intensity of exposure [Burgess et al. 1997; Cherry et al. 1997; McDonald
et al. 1997],
- the
presence of radiographically defined silicosis [Amandus et al. 1992;
Dong et al.1995], and
- length
of followup time from date of silicosis diagnosis [Partanen et al.
1994] (see Table 15).
These
observed associations, including the exposure-response associations,
are unlikely to be explained by confounding or other biases. Thus overall,
the epidemiologic studies support increased lung cancer risks from occupational
exposure to inhaled crystalline silica (i.e., quartz and cristobalite)
[IARC 1997].
3.4.2.1
Updated or New Studies Since the IARC Review
Two studies
discussed in this section have recently been updated: Checkoway et al.
[1997, 1999] updated their previous mortality studies of diatomaceous
earth workers [Checkoway et al. 1993, 1996] by including deaths after
1987 and through 1994, and by analyzing lung cancer risk among workers
with radiographic silicosis. Lung cancer mortality risk was highest
in the highest category of cumulative exposure to respirable crystalline
silica (rate ratio with no exposure lag period=2.11; 95% CI=1.074.1;
rate ratio for 15-year exposure lag period=1.05; 95% CI=0.991.11).
The rate ratios were adjusted for the effects of age, calendar year,
duration of followup, and ethnicity. Among workers with radiological
silicosis (ILO category >1/0
or large opacity; n=81), the lung cancer SMR
was 1.57 (95% CI= 0.434.03) [Checkoway et al. 1999]. For workers
without silicosis (ILO category <1/0), the SMR was 1.19 (95% CI=0.871.57).
The SMRs were adjusted for age and calendar year and were based on the
expected number of deaths for white U.S. males. For the nonsilicotic
workers, a statistically significant, positive dose-response relationship
(P=0.02) was observed between SMRs for lung cancer and category
of cumulative respirable silica exposure. The SMRs ranged from 1.05
in the lowest exposure category (<0.5 mg/m3·year,
13 deaths, 95% CI=0.561.79) to 2.40 in the highest exposure category
(>5.0 mg/m3·year; 12 deaths, 95% CI=1.244.20).
For the 81 workers with radiographic silicosis, an SMR >1.0 was observed
only in the highest exposure category (i.e., >5.0 mg/m3·year)
(4 deaths observed; SMR=2.94; 95% CI=0.807.53). These results
suggest that silicosis may not be a necessary condition for silica-related
lung cancer. However, radiographic surveillance of this cohort did not
extend beyond the dates of employment termination, and autopsies were
not routinely conducted [Checkoway et al. 1999].
Cherry
et al. [1998] finalized the preliminary results of a nested case control
study of 52 lung cancer deaths in 5,115 pottery workers (see Burgess
et al. [1997], Cherry et al. [1997], and McDonald et al. [1997] in Table
15). After adjustment for smoking and inclusion of a 20-, 10-, or
0 year lag period, mean respirable silica concentration
(i.e., estimated daily 8-hr TWA airborne concentrations
in µg/m3) was associated with lung cancer (P<0.008
for each lag period):
Lag
|
OR
|
|
20
yr
|
1.60
|
1.11-2.31
|
10
yr
|
1.66
|
1.14-2.41
|
0
yr
|
1.67
|
1.13-2.47
|
However,
exposure duration and cumulative silica dust exposure were not significantly
associated with lung cancer mortality, regardless of lag time [Cherry
et al. 1998]. The presence of small, parenchymal radiographic opacities
(ILO category >1/0) was not related to lung cancer mortality
before adjustment for smoking (P=0.78) or after adjustment for
smoking and mean silica concentration
(P=0.68). The authors concluded that crystalline silica
may well be a human carcinogen [Cherry et al. 1998].
Other
studies published since the IARC review also investigated exposure-response
associations for lung cancer and exposure to crystalline silica. Rafnsson
and Gunnarsdóttir [1997] reported that the incidence
of lung cancer cases among 1,346 diatomaceous earth workers in Iceland
was not statistically significant for workers who had 9 years before
start of followup and who were employed >5 years (standardized
incidence ratio [SIR] based on 3 cases observed=2.70; 95% CI=0.567.90)
or employed <5 years (SIR based on 2 cases observed=1.19;
95% CI=0.144.30).
de Klerk
and Musk [1998] conducted a cohort study of 2,297 surface and underground
gold miners in western Australia who participated in surveys of respiratory
symptoms, smoking habits, and lung function in 1961, 1974, and 1975.
Eighty-nine percent of the cohort was traced to the end of 1993 for
trachea, bronchus, and lung cancer mortality and incidence of compensated
silicosis (i.e., compensation awarded by the Pneumoconiosis Medical
Board). A nested case-control analysis of the 138 lung cancer deaths
found that lung cancer mortality was related to log total cumulative
silica dust exposure after adjustment for smoking (cigarette, pipe,
or cigar) and for the presence of bronchitis at survey (relative rate=1.31;
95% CI=1.011.70). However, the effect of cumulative silica dust
exposure on lung cancer mortality was not significant after adjustment
for smoking, bronchitis, and compensation for silicosis (relative rate=1.20;
95% CI=0.921.56). Other silica exposure variables (i.e., duration
of underground or surface employment and intensity of underground or
surface exposure) were not significantly related to lung cancer mortality
(P>0.15) after adjustment for smoking and bronchitis. Cigarette
smoking (relative rate=32.5; 95% CI=4.4241.2 for >25
cigarettes smoked per day), incidence of a compensation award for silicosis
after lung cancer diagnosis (relative rate=1.59; 95% CI=1.102.28),
and presence of bronchitis at survey (relative rate=1.60; 95% CI=1.092.33)
were significantly related to lung cancer mortality [de Klerk and Musk
1998]. The results of this study do not support a relationship between
lung cancer and silica exposure in the absence of silicosis (i.e., a
compensation award for silicosis after lung cancer diagnosis). However,
controlling for silicosis compensation and bronchitis may have masked
a silica effect because both are markers of silica exposure.
Hnizdo
et al. [1997] conducted a nested case-control study of lung cancer deaths
in a cohort of 2,260 white South African underground gold miners. (A
lung cancer mortality cohort study had been conducted earlier [Hnizdo
and Sluis-Cremer 1991]). The mineral content of the rock in the gold
mines was mostly quartz (70%-90%),
silicates (10%-30%), pyrite (1%-4%), and heavy minerals with grains
of gold and uranium-bearing minerals (2%-4%). Seventy-eight miners who
died from lung cancer (69 of the 78 had a necropsy) during 1970-1986
were matched by year of birth with 386 control subjects from the same
cohort [Hnizdo et al. 1997]. Conditional logistic regression models
were used to analyze the relationship of lung cancer mortality with
cigarette smoking (pack-years), cumulative dust exposure
(mg/m3·year), years of underground mining, incidence
of radiographic silicosis (ILO category >1/1 diagnosed up
to 3 years before death of a matched case), and uranium production or
uranium grade of the ore in the gold mine. Radon progeny measurements
in the gold mines were not available.
Lung cancer
mortality was associated with cigarette smoking, cumulative dust exposure
(lag time was 20 years from death), duration of underground mining (lag
time was 20 years from death), and silicosis. The best-fitting model
predicted relative risks of 2.45 (95% CI=
1.25.2) for silicosis and the following relative risks for various
pack-years of smoking:
Pack-years
|
|
Relative
risk
|
<6.5
|
|
1
|
6.5-20
|
0.7-16.8
|
3.5
|
21-30
|
1.3-25.8
|
5.7
|
>30
|
3.1-56.2
|
13.2
|
The authors
stated that variables representing uranium mining were not significantly
related to lung cancer mortality (modeling results for these variables
were not presented) [Hnizdo et al. 1997]. The authors proposed three
explanations for their results:
- Miners
with high dust exposure who develop silicosis have increased lung
cancer risk.
- High
silica dust exposure concentrations
are important in the pathogenesis of lung cancer, and silicosis is
coincidental.
- High
silica dust exposure concentrations
are a surrogate measure of exposure to radon progeny [Hnizdo et al.
1997].
3.4.2.2
Lung Cancer Meta-Analyses
Meta-analysis
and other systematic literature review methods are useful tools for
summarizing exposure risk estimates from a large amount of information
[Mulrow 1994]. Meta-analyses or summary reviews of epidemiologic studies
of silicotics with lung cancer have been conducted by investigators
in the United States [Steenland and Stayner 1997; Smith et al. 1995]
and Japan [Tsuda et al. 1997]. IARC is performing a pooled analysis
of epidemiologic data from several cohorts to investigate lung cancer
risks in nonsilicotic workers.
Steenland
and Stayner [1997] and IARC [1997] found that the majority of studies
of silicotics reported statistically significant excess lung cancer
risks across different countries, industries, and time periods while
controlling for the effects of cigarette smoking [Steenland and Stayner
1997; IARC 1997]. Exposure-response gradients were also observed. The
summary relative risk was 2.3 (95% CI=2.22.6) for 19 cohort and
case-control studies of silicoticsexcluding studies of miners
and foundry workers because of potential exposure to other carcinogens,
and omitting autopsy studies and proportionate mortality studies because
of possible selection biases [Steenland and Stayner 1997]. Fifteen of
the 19 studies directly or indirectly controlled for the effects of
smoking. The summary relative risk of 16 cohort* and case-control studies
of silica exposed workers was 1.3 (95% CI= 1.21.4)a moderate
and statistically significant relative risk estimate [Steenland and
Stayner 1997]. Eight of the 16 studies controlled for the effects of
smoking, either directly or indirectly.
*Cohort
size ranged from 969 to 6,266 workers.
Another
meta-analysis of 23 lung cancer studies of silicotics (including 14
of the studies analyzed by Steenland and Stayner [1997]) reported a
pooled risk estimate of 2.2 (95% CI= 2.12.4) [Smith et al. 1995].
The statistically significant pooled risk estimates from both meta-analyses
strongly support an association between silicosis and lung cancer. The
increased risk of lung cancer for silicotics is also supported by the
following [IARC 1997]:
- The
magnitude of the risk estimates (i.e., most studies reported risks
greater than 2.0 for silicotics after adjusting for the effects of
cigarette smokingcompared with exposed nonsilicotics or the
general population)
- The
observation of exposure-response gradients with various indicators
of exposure
- Consistent
findings of excess risk in different countries, industries, and time
periods
- Two
studies that provided reasonable evidence for an unconfounded association
(i.e., Amandus et al. [1991, 1992, 1995] and Partanen et al. [1994],
an update of Kurppa et al. [1986])
Tsuda
et al. [1997] conducted a lung cancer meta-analysis of pneumoconiosis
or silicosis studies (excluding asbestosis). Lung cancer risk estimates
were pooled from 32 mortality studies published from 1980 to 1994. The
estimated rate ratios were similar to those reported by Steenland and
Stayner [1997] and Smith et al. [1995]:
Rate
ratio
|
|
All
studies (32) |
2.74
|
2.60-2.90
|
Cohort
studies only (25 of 32) |
2.77
|
2.61-2.94
|
Case-control
studies (5 of 32) |
2.84
|
2.25-3.59
|
Two
of the studies are proportionate mortality studies for which rate ratios
were not reported.
3.4.3
Other Cancers
Mortality
studies of workers have reported statistically significant excesses
of deaths from stomach or gastric cancer in iron ore miners [St. Clair
Renard 1984; Lawler et al. 1985; Mur et al. 1987], Canadian gold miners
[Muller et al. 1983; Shannon et al. 1987; Miller et al. 1987; Kusiak
et al. 1993b], lead and zinc miners [Belli et al. 1989], brick production
workers [Katsnelson and Mokronosova 1979], foundry and other metal workers
[Neuberger and Kundi 1990], jewelry workers [Hayes et al. 1993; Dubrow
and Gute 1987; Sparks and Wegman 1980], farmers (reviewed by Blair and
Zahm [1991]), and farm workers [Stubbs etal. 1984] (reviewed by Zahm
and Blair [1993]). A recent case-control study of 250 male hospital
patients in Canada found a statistically significant excess of pathologically
confirmed stomach cancer among the 25 patients who reported a history
of substantial occupational exposure to crystalline
silica compared with 2,822 controls (OR=1.7; 95% CI=1.12.7
after adjusting for the effects of age, birthplace, education, and cigarette
smoking) [Parent et al. 1998]. However, in a review of epidemiologic
studies of gastric cancer and dusty occupations, Cocco et al. [1996]
noted that because most studies did not adjust for the effects of confounding
factors or assess a dose-response relationship, evidence was insufficient
to conclude that silica is a gastric carcinogen.
For workers
who may have been exposed to crystalline silica, there have been infrequent
reports of statistically significant excesses of deaths or cases of
other cancers such as nasopharyngeal or pharyngeal cancer [Chen et al.
1992; Carta et al. 1991], salivary gland cancer [Zheng et al. 1996],
liver cancer [Chen et al. 1992; Hua et al. 1992], bone cancer [Forastiere
et al. 1989; Steenland and Beaumont 1986], pancreatic cancer [Kauppinen
etal. 1995], skin cancer [Partanen et al. 1994; Rafnsson and Gunnarsdóttir
1997], esophageal cancer [Pan et al. 1999; Xu et al. 1996; Belli et
al. 1989], cancers of the digestive system [Decoufle and Wood 1979],
intestinal or peritoneal cancer [Amandus et al. 1991; Goldsmith et al.
1995; Costello et al. 1995], lymphopoietic or hematopoietic cancers
[Redmond et al. 1981; Silverstein et al. 1986; Steenland and Brown 1995b],
brain cancer [Rafnsson and Gunnarsdóttir 1997], and bladder cancer
[Bravo et al. 1987]. Again, an association has not been established
between these cancers and exposure to crystalline silica.
|