AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 30~398-406( 1 996)

Lung Cancer and Asbestos Exposure:

Asbestosis i s Not Necessary

David Egilman, MD MPH, and Alexander Reinert

Recent commentaries on the issue of asbestos-related lung cancer have raised important
points. One major question is whether lung cancer can be attributed to asbestos exposure in
the absence of asbestosis. This review attempts to place the debate in the proper context for
establishing causation. Relevant epidemiologic and pathologic studies are analyzed, as well
as the scientific basis for each position in the debate. The assertion that asbestosis must be
present in order to attribute a lung cancer to asbestos exposure does not meet accepted
standards for establishing causation. In addition, some evidence has been incorrectly cited
in support of this position. This discussion can benefit from clearer definitions of asbestosis,
a more thorough evaluation of the available scientific information, and a proper context for
determining causation. This review of the available evidence indicates that lung cancers can
occur as a result of asbestos exposure, in the absence of clinical or histologic asbestosis.
Causation in an individual should be assessed by considering duration of exposure, intensity
of exposure, and appropriate latency. 0 1996 Wiley-Liss, Inc.

KEY WORDS: lung cancer, asbestosis, causation, cancer without asbestosis, epistemol-
ogy, epidemiology, pathology

INTRODUCTION is necessary position, standing for asbestosis is necessary

for attribution of a lung cancer to asbestos exposure.
We have read the recent commentaries on the relation-
ship between asbestos exposure, asbestosis, and lung cancer DEFINITIONAL ISSUES
with great interest [Roggli et al., 1994; Abraham, 19941.
Some important issues need further clarification. First, the
Lack of clarity regarding the definition of asbestosis
definition of the terms of the debate has led to unnecessary
has clouded the debate. Is asbestosis diagnosed by X-ray, or
confusion, and should be elucidated. Second, the debate
by histology? Braun and Truan [ 19581 first recognized this
should be put in an accepted framework for establishing
confusion and offered two definitions. Macroscopic asbes-
causation. Third, a review of the relevant literature reveals
tosis was visible on an X-ray or diagnosed through clinical
many studies that support the position that asbestos expo-
symptoms. Microscopic asbestosis was visible only on bi-
sure can cause an increase in lung cancer risk in the absence
opsy or autopsy. They offered this clarification because
of asbestosis. Finally, some of the advocates of the position
there was a complete lack of definition of terms as used in
that asbestosis must be present to attribute lung cancer to
the published literature. This confusion is compounded by
asbestos exposure take a stand which contains inherent con-
the fact that early literature on asbestos, asbestosis, and lung
tradictions, especially related to the diagnosis of asbestosis.
cancer usually defined asbestosis macroscopically. Some
For brevity, we will refer to this position as the asbestosis
patients in these studies may have had parenchymal asbes-
tosis detectable on both levels, while others may have had
only microscopically detectable asbestosis. Any attempt to
Brown University, Providence, RI. use these early studies in the current debate is fraught with
Address reprint requests to David Egilman, South Shore Health Center, 759 difficulty. In this paper, clinical or radiological is
Granite St., Braintree, MA 02184.
used to mean macroscopic, and histologic or patho-
Accepted for publication August 1, 1995. logic to mean microscopic.

0 1996 Wiley-Liss, Inc.


ESTABLISHING CAUSATION
We begin this discussion with the basic problems in the
assumptions underlying the asbestosis is necessary
model, and proceed to a more focused elaboration of the
causal framework which we propose to use in assessing
causation. To accept the asbestosis is necessary position,
one or both of the following must be true: (1) asbestosis is
a required intermediary stage in the causal pathway between
asbestos exposure and the development of cancer; (2) there
exists a threshold dose for the induction of cancer from
asbestos exposure such that the dose required to produce
asbestos-induced cancer is greater than or equal to the dose
necessary to induce asbestosis. It should be clear that the
second point does not necessarily imply the first: if both
diseases are dose-related, and the dose required for their
induction is similar, then one would expect to see the dis-
eases associated with one another, even if there were no
causal relationship. In addition, our position does not ex-
clude the possibility of the first assertion being partially
true. In other words, a study which demonstrates an in-
creased lung cancer risk for asbestotics does not disprove
our hypothesis. On the other hand, if the asbestosis is
necessary causal model is accurate studies which find an
increased lung cancer risk among people exposed to as-
bestos, but without asbestosis, should not exist. Just as
the discovery of a single black swan disproves the hypoth-
esis that all swans are white, the discovery of an increased
risk of lung cancer in any exposed cohort without as-
bestosis nullifies the asbestosis is necessary hypothesis. where on the line between 0 and 100 in our model. The
From our review of the literature, black swans are in abun-
dance.
While these are the general issues concerning the cau-
sation of asbestos-related lung cancer, it would be helpful to
place the discussion in the context of accepted frameworks
for establishing causation. First, it is important to consider
all of the evidence available: epidemiological, pathological,
and experimental, along with the best evidence from mo-
lecular understanding.
Many times, different parties have attempted to elevate
one entire class of data to the level of final arbiter of causal
proof. For instance, tobacco companies used to argue that
cigarettes had not been proven to cause lung cancer because
of the lack of animal studies establishing such a relation-
ship. Some chemical companies argued that dioxin was not
a carcinogen because of the dearth of epidemiologic studies
(this was a lack of data, not a lack of positive data). It is
reasonable for different people to assign different values to
different classes of studies. It is not reasonable to ignore
entire classes of data, or to elevate one class of data to carry
the ultimate burden of proof, especially in making public
policy decisions regarding potential health risks. This Bay-
esian approach to causation, which utilizes all available sets
of data, is represented in Figure 1, where the likelihood of
Cancer Without Asbestosis: Causal Issues 399
I Available Evidence I / x x zrelationship
zGq
cause-elfect
Molecular Shldies
FIGURE 1. Accepted model for causation analysis.
a cause-effect relationship is measured on a scale of 0 to
100.
This method of causation assessment is directly related
to accepted models: Rothmans general model and the ex-
trapolation of that model into concrete decisions [Rothman,
19861. This framework asks two questions. The first is,
Does the substance cause the effect in general? In other
words, does asbestos exposure, in general, cause lung can-
cer, or does the presence of asbestosis, in general, cause
lung cancer? This can be answered, at least theoretically, by
considering Hills criteria for causation as modified by
Rothman, and using those criteria to place the agent some-
second question is, Has the substance contributed to the
effect in a particular individual? No guidelines to ap-
proach this question are proposed here, but we recognize
that it is the subtext of the discussion of general causation.
For an overall model, Rothmans modification of Hills
criteria for causation is a useful framework. Of Hills nine
criteria, the only categories which are pertinent to this dis-
cussion are strength of association, biologic gradient, con-
sistency, plausibility, coherence, experimental evidence,
and analogy. Specificity requires that each cause have a
single effect. Because asbestos exposure is recognized to
have multiple health effects, this category is not useful.
Temporality is not an issue for either theory. That is, in all
of the studies considered, the patients had had prior expo-
sure to asbestos, whether or not they had developed asbes-
tosis before developing lung cancer. While temporality does
relate to the asbestosis is necessary position in terms of
the sequence of asbestosis and lung cancer, there are no
current data to address this point. It is possible, however, to
imagine a scenario which reveals the inadequacy of this
position with respect to temporality. A patient with a history
of asbestos exposure presents with lung cancer, without
evidence of clinical asbestosis. Upon excision of the lobe
with the tumor, no signs of histologic asbestosis are present.
400 E g i l m a n and R e i n e r t
TABLE I. Sir Bradford Austin Hills Criteria for Plausibility of Causation According to the
Two Models
Hills criteria
Consistency
Coherence
Plausibility
Strength of association
Biologic gradient
Experimental evidence
Analogy
Ten years later, the patient presents with lung cancer, and
this time treatment is not successful. The autopsy reveals
signs of histologic asbestosis. According to the asbestosis
is necessary hypothesis, the second lung cancer would be
related to asbestos exposure, but not the first.
The criteria of Hills which we will attempt to satisfy
are consistency, plausibility, coherence, strength of associ-
ation, biologic gradient, experimental evidence, and anal-
ogy (Table I). Of these, most of our efforts will be aimed
toward demonstrating consistency (repeated observation of
an association under different circumstances) through a re-
view of the epidemiologic and pathologic studies, and the
molecular evidence which has been summarized by both
Churg and Walker et al. in support of our position. The issue
of biological plausibility has been covered by Roggli et al.
[1994], Churg (19931, and Walker et al. [1992], and it
would be repetitious to provide another summary. Biologi-
cal evidence indicates that asbestosis is related to inflam-
mation and the immune response, while lung cancer is prob-
ably caused by the interaction of asbestos fibers with
nuclear material, resulting in mutation [Montizaan et al.,
19891. Since biological evidence indicates that asbestos
causes lung cancer and fibrosis through different mecha-
nisms, there is no reason to posit asbestosis as an interme-
diate step in asbestos cancer causation. The category of
coherence, which asks if the theory is consistent with what
is already known of the disease, is not satisfied in the as-
bestosis is necessary model. Scarring is not considered a
necessary prerequisite to lung cancers related to carcino-
genic exposures other than asbestos. If scarring were in-
volved in the etiology of asbestos-related lung cancers, ad-
enocarcinomas would be more common in asbestos-
exposed individuals. This is not the case [Mark and Shin,
19921. In addition, evidence indicates that scars in scar
carcinomas actually develop secondary to the tumor, sug-
gesting that such scars are not etiologically significant in the
development of cancer [Cagle et al., 19851.
Strength of association requires that any proposed
cause-effect relationship have a large measure of effect.
Asbestosis is Asbestos is a carcinogen in
necessary position the absence of asbestosis
Does not meet standard Meets standard
Does not meet standard Meets standard
Does not meet standard Meets standard
Indeterminate Meets standard
Does not meet standard Meets standard
Indeterminate Indeterminate
Does not meet standard Meets standard
While Rothman [I9861 has pointed out the inadequacy of
using this criterion, it is satisfied by the fact that many of the
studies reviewed report a rate ratio of greater than 2 for the
relationship between asbestos exposure and lung cancer in
the absence of asbestosis [Cheng and Kong, 1992; Edge,
1976, 1979; Huilan and Zhiming, 1993; Martischnig et al.,
1977; Talcott et al., 19891. Biologic gradient requires the
evidence of a dose-response curve for the proposed cause-
effect relationship. Dement and colleagues have published
recent updates of the mortality of a cohort of textile workers
first reported on in 1981 which provide evidence that excess
mortality from lung cancer attains significance at lower cu-
mulative exposure levels than excess mortality from pneu-
moconiosis in this asbestos-exposed cohort [Brown et al.,
1994; Dement et al., 19941. This suggests that asbestos can
cause lung cancer in the absence of asbestosis. This study
and many others establish a dose-response relationship be-
tween asbestos exposure and lung cancer [Barbone et al.,
1992; Huilan and Zhiming, 1993; McDonald and Liddell,
1979; McDonald et al., 1980; Morabia et al., 1992; Peto,
19801.
The experimental evidence, as Roggli et al. [ 19941 have
pointed out, has not proved supportive of either stance.
There are two issues relating to analogy which should be
considered. First, it is accepted that asbestos can cause me-
sothelioma in the absence of fibrosis. Second, the presence
of emphysema or bronchitis is not considered necessary to
attribute a lung cancer to cigarette smoking. Of Hills cri-
teria, we now turn to expand on the evidence for consis-
tency, by examining the relevant epidemiologic and patho-
logic literature.
STUDIES OF CLINICAL ASBESTOSIS
If all of the epidemiological evidence is considered, the
first conclusion is that most studies have not considered the
problem of coincident asbestosis and lung cancer. The ma-
jority of studies, by far, do not offer information which
would help us resolve this debate [Armstrong et al., 1988;

Berry et al., 1985; Barbone et al., 1992; Bovenzi et al.,
1993; Haider and Neuberger, 1980; Hammond et al., 1979;
Hughes and Weill, 1980; Hughes et al., 1987; Jarvholm et
al., 1993; McDonald and Liddell, 1979; McDonald et al.,
1980; Morabia et al., 1992; Moulin et al., 1993; Nicholson
et al., 1979; Peto, 1980; Raffn et al., 1993; Selikoff et al.,
1968, 19791.
Hughes and Weill, reporting in 1991, offer the only
evidence of an increased risk of lung cancer only among
clinically diagnosed asbestotics. Their prospective cohort
study of 839 asbestos cement workers found that workers
with an ILO classification greater than or equal to 110 had
an increased risk of lung cancer, but not workers without
fibrosis. Hughes and Weill conclude from their study that
asbestosis is necessary for the attribution of lung cancer to
asbestos exposure [Hughes and Weill, 19911. It should be
noted that among workers with small opacities, whose
X-rays were ILO classification 0/1, SMR from lung cancer
was 177.5, but this was not significantly raised. In ad-
dition, the study had sufficient power (greater than 80%) to
detect only a relative risk of 2.7 among workers with X-rays
classified as 0/1. Finally, the authors state that the exposure
history of long-term workers without X-ray abnormalities
was similar to the workers with small opacities greater
than or equal to 1/0. Our analysis indicates that the differ-
ence in cumulative exposure between these two groups was
statistically significant at the 90% level (two-tailed t test: t
= 1.87, 0.10 > p > 0.05). This test makes the conservative
assumption that all of the workers with small opacities
greater than or equal to 1/0 had worked in the factory for
more than 21.5 years. The result raises the possibility that
the observed increase in lung cancer risk associated with
fibrosis was related to differences in asbestos exposure.
Most other studies that have collected data on lung
cancer without asbestosis indicate that both asbestotics and
nonasbestotics suffer an increased risk of developing lung
cancer. In 1972, Fletcher conducted a mortality study of
shipyard workers with and without pleural plaques. In this
study, Fletcher compared 408 men with pleural plaques and
404 controls (with no X-ray evidence of plaques) to Barrow
men in the same age group. He found 16 bronchial carci-
nomas among the patients with plaques compared to 6.74
expected, and seven bronchial carcinomas among the con-
trol group compared to 5.61 expected. Note: The incidence
of carcinoma among patients with pleural plaques was 2.4
times the expectedfigure and 1.2 times the expected number
among the control patients. While smoking was not con-
trolled, it is unlikely, according to Fletcher, that different
smoking habits explain the findings since the number of
cigarettes consumed by those with and without pleural
plaques was found to be almost identical [Fletcher, 19721.
Fletcher also found that of the 16 cases of bronchial carci-
noma among men with pleural plaques, none had any ra-
diological evidence of pulmonary fibrosis. Furthermore,
Cancer Without Asbestosis: Causal Issues 401
only one case of slight fibrosis was determined on histo-
logical examination of the five cases in which carcinoma
was diagnosed at necropsy. This study supports the asser-
tion that asbestos exposure, absent even macroscopic asbes-
tosis, increases ones risk of developing lung cancer.
Martischnig et al. conducted a case control study of 201
patients with confirmed bronchial carcinoma in order to
determine the relationship between lung cancer and asbes-
tos exposure [Martischnig et al., 19771. They found that 58
of the 201 lung cancer patients who did nut have clinical
asbestosis had a history of occupational exposure to asbes-
tos. Only 29 of 201 control patients (with matching age,
smoking history, residential area, and diseases other than
lung cancer) had a history of occupational exposure to as-
bestos. That is, twice as many lung cancer patients without
asbestosis as patients with other diseases had a history of
asbestos exposure (OR = 2.4). The increased risk of carci-
noma due to asbestos exposure prevailed regardless of the
level of smoking.
A case-control study of 535 men with primary lung
cancer and 659 controls further supports the assertion that
asbestos is a carcinogen separate from its fibrogenic prop-
erties [Blot et al., 19781. The relative risk of developing
lung cancer among coastal Georgia shipyard employees was
1.6, after adjustment for smoking, other occupations, age,
race, and county of residence. Given that a review of hos-
pital and pathology records revealed only two cases of as-
bestosis, almost all of the excess cases of cancer were
among men without asbestosis.
Liddell and McDonald [ 19801 reported on a prospec-
tive cohort study of 4,559 Quebec asbestos miners and mill-
ers. The purpose of the study was to determine the extent to
which chest X-rays taken dui-ing employment could predict
mortality rates. This study is often referenced in support of
the asbestosis is necessary position [Churg, 1993; Roggli
et al., 19941. In fact, Liddell and McDonalds study pro-
vides strong evidence that many lung cancers caused by
asbestos occur in the absence of clinical asbestosis. There
were 52 excess lung cancers in the cohort (1 I8 total) attrib-
uted to asbestos exposure. In the entire cohort, there were
39 X-rays with abnormal parenchymal changes; 27 of these
changes were graded 1/0 or greater according to the ILO
classification. Without further information about the 12
cases with X-ray changes less than l/O, it is inappropriate to
categorize those cases as asbestotics. Assuming that all of
the abnormal X-rays occurred among the excess lung can-
cers, at most 52% of the excess lung cancers were associ-
ated with changes consistent with asbestosis. It is likely that
not all of the abnormal X-rays were concentrated in the
excess lung cancers, so that at least half of the excess lung
cancers occurred in the absence of radiological signs of
asbestosis. The authors conclude that most, but not nec-
essarily all, cases of lung cancer attributable to chrysotile
exposure in mining and milling probably have small paren-
402 Egilman and Reinert
chymal opacities before death. While this study as well as
the Sluis-Cremer and Bezuidenhout study (see below) have
consistently been referenced as supportive of the asbesto-
sis is necessary position, the data actually support the
opposite hypothesis [Churg, 19931.
A cohort study of men who were employed in a ciga-
rette manufacturing company offers further evidence that
clinical asbestosis is not necessary to attribute lung cancer
to asbestos exposure [Talcott et al., 19891. This study of 33
men found that the relative risk for dying from lung cancer
was 13.1 (eight deaths from lung cancer were observed).
While the authors do not present data explicitly related to
the coincidence of asbestosis and lung cancer, they do give
enough information to determine that none of the men who
died from lung cancer had clinical asbestosis. Of the 15 men
who died with asbestosis, only three were diagnosed clini-
cally, the remaining 12 being diagnosed at autopsy. The
three clinical asbestotics were among the five asbestotics
whose certified cause of death was asbestosis. Therefore, 10
men who died of other causes (possibly including some of
the lung cancer cases) had microscopic asbestosis. None of
the eight men who died of lung cancer had clinical asbes-
tosis. Thus, among asbestos-exposed men without radio-
logic asbestosis, the relative risk of dying from lung cancer
was 13.1. Smoking data were collected, but it is not appar-
ent that the authors controlled for differences in smoking
history.
Cheng and Kong, in 1992, reported on a cohort of 662
men and 5 10 women employed in asbestos textiles, friction
material, and asbestos cement manufacturing. They found a
statistically significant excess in lung cancer deaths among
asbestotics and nonasbestotics (SMR 320 and 306, respec-
tively). They also found that the presence of asbestosis in-
creases ones risk of developing lung cancer, but they do not
control for differences in exposure level, so this finding is of
questionable significance. Other weaknesses include that
the study does not explain how asbestosis was diagnosed.
The tobacco smoke intake of this cohort was relatively
low, according to the authors, although they did not con-
trol for smoking.
A recent study by Wilkinson and coworkers [I9951
further supports the assertion that asbestos exposure in-
creases the risk of lung cancer in the absence of fibrosis.
This case-referent study evaluated 27 1 primary lung cancer
patients and 678 referents for evidence of asbestos exposure
and radiological asbestosis. Small opacities found on X-ray
(ILO 110 or greater) were considered evidence of asbestosis.
Individuals with ILO 0/1 or less were considered to be free
of fibrosis. After adjustment for age, sex, smoking history,
and area of referral, the odds ratio relating the risk of con-
tracting lung cancer to past asbestos exposure for those with
fibrosis was 2.03 (95% C.I. I .OO-4.13). For those with ILO
scores of 0/1 or less, the odds ratio was 1.56 (95% C.I.
1.02-2.39). The authors conclude that these data suggest
that asbestos [exposure] is associated with lung cancer even
in the absence of radiologically apparent fibrosis.
Two studies of men with pleural plaques without ra-
diological evidence of asbestosis have found an increased
risk of lung cancer among this population [Edge, 1976,
1979; Hillerdal, 19941. Edge studied over 400 men with
pleural plaques, selected from a population made up pre-
dominantly of shipyard workers at random. He found be-
tween a twofold and 2.5-fold risk of dying of carcinoma of
the bronchus compared with the general population. Roggli
et al. have already summarized Hillerdals [1994] study as
well as another supportive study by Irvine et al. [ 19931. The
latter study found that of 19,000 lung cancer cases in West
Scotland, 5.7% were asbestos related. Because of the low
annual incidence of asbestosis, the authors concluded that
lung cancer caused by asbestos exposure could occur in the
absence of fibrosis. Hillerdal [ 19941found that patients with
pleural plaques but without radiologic evidence of asbesto-
sis experienced an increased risk of lung cancer, after con-
trolling for smoking habits.
STUDIES OF HISTOLOGIC ASBESTOSIS
Among studies that have considered the incidence of
lung cancer with or without pathologic asbestosis, one has
found an increased risk only among those with asbestosis
[Sluis-Cremer and Bezuidenhout, 19891. This was a nec-
ropsy study of 339 asbestos miners which found an odds ratio
of 4.45 for patients with slight pathological asbestosis and 5.2
for moderate to pronounced asbestosis. The authors them-
selves state that [ilt must be emphasized that these results
should not affect compensation bodies dealing with living
subjects exposed to asbestos as slight asbestosis is com-
monly, and moderate asbestosis occasionally, undetected
radiologically. In addition, a reanalysis of the data, stim-
ulated by a letter from Rudd [1990], revealed that if the
variable for years of exposure is entered into the regression
analysis before the variable for presence of asbestosis, years
of exposure accounted for most of the variation (asbestosis
was still a significant risk factor for lung cancer) [Sluis-
Cremer and Bezuidenhout, 19901. This would support the
contention that the apparent correlation of asbestosis and
lung cancer is really a function of two dose-related diseases
caused by the same agent.
The first pathologic study to consider this question was
published almost 40 years ago [Williams, 19571. In this
article, Williams proposed that if asbestosis was an inter-
mediate step between asbestos exposure and lung cancer,
then alveolar metaplasia would be more common in asbesto-
tics with lung cancer than in other occupations with elevated
risks of lung cancer. He found no significant difference
between occupational groups in the prevalence of alveolar
metaplasia, although the rate of alveolar metaplasia in as-
bestotics without lung cancer was slightly increased com-
 Cancer Without Asbestosis: Causal Issues 403

TABLE II. Summary of Relevant Studies Relating to Clinical Asbestosis

Reference Study population Data on lung cancer and asbestos exposure

Fletcher, 1972
Edge, 1976
Martischnig et al., 1977
Blot et al., 1978
Edge, 1979
Liddell and McDonald, 1980
Talcott et al., 1989
Hughes and Weill, 1991
Cheng and Kong, 1992
Hillerdal, 1994
Wilkinson et al., 1995
408 shipyard workers with pleural plaques; 404 shipyard
workers without pleural plaques
235 shipyard workers with pleural plaques
201 cancer patients and 201 controls
535 lung cancer cases, 659 controls
429 men with pleural plaques, predominantly shipyard
workers, without clinical asbestosis
4,559 Quebec miners and millers
33 cigarette filter workers
839 asbestos cement workers
662 males and 510 females employed in asbestos
cement manufacturing
1596 men with pleural plaques
271 lung cancer patients and 678 referents
Workers with pleural plaques had 2.4 times expected rate
of lung cancer; workers without pleural plaques had 1.2
times expected rate of lung cancer
RR of 2.4.
OR = 2.4 for asbestos exposed without fibrosis (our
analysis)
OR of 1.6, where a review of hospital and pathology
records found only two cases of asbestosis. Almost all
of the excess cancers occurred among men without
asbestosis
Found a twofold increased risk among men with pleural
plaques
At least 48% of excess lung cancers had ILO category 0/1
or less
RR of 13.1 for workers without clinical fibrosis: NA for
pathological fibrosis (our reanalysis)
Increased risk only among workers with clinical fibrosis
Increased risk among workers with or without asbestosis
(unclear as to diagnostic method)
RR = 1.4 for patients with pleural plaques, without
asbestosis
OR = 1.56 for asbestos-exposed individuals with ILO grade
of 0/1 or less

pared to the overall rate of alveolar metaplasia [Williams, lo5 and lo6 fibers per gram of dry lung). The authors con-
19571. Williams concluded that his evidence did not support clude from this study that, because large fiber burdens may
a role for alveolar metaplasia in the etiology of lung cancer. not always result in asbestosis, such fibrosis may be a poor
Interestingly enough, a more recent study confirms Wil- marker of fiber-related lung cancer [Wamock and Isen-
liams early observations [Merrill et al., 19911. In this study, berg, 19861.
airway metaplasia was seen with equal frequency among Huilan and Zhiming published a study in 1993 of 5,893
asbestos-exposed subjects separated by ILO X-ray category. persons employed in asbestos factories in China. In com-
In other words, more severe fibrosis did not lead to an parison to a control group, the cohort had a relative risk of
increase in observed airway metaplasia. 5.3 for dying from lung cancer, with a standardized relative
An autopsy study compared asbestos body counts from risk of 4.2. Among 148 cases of pathologic asbestosis, there
100 patients without lung cancer and 30 patients with lung were 33 cases of coincident lung cancer. Assuming that all
cancer [Wamock and Churg, 19751. The patients with lung of these cases were classified under the death certificate as
cancer had a significantly higher number of asbestos bodies, a lung cancer death, the relative risk for lung cancer without
although only one patient had a history of occupational coincident asbestosis would be adjusted downward to 2.68,
exposure to asbestos. The authors suggest from this evi- with a standardized relative risk of 2.13.
dence that even extremely low levels of asbestos exposure Karjalainen et a]., in 1993, offered implicit evidence
may have a carcinogenic effect. Fibrosis was absent or that histologically diagnosed asbestosis is not necessary in
minimal in all of the cancer patients. order for asbestos to cause lung cancer. They considered the
Wamock and Isenberg [ 19861 published a similar fiber lobe of origin in 108 lung cancer patients, and attempted to
burden study of men with lung cancer. All of the men stud- relate it to a history of asbestos exposure, and the presence
ied had a history of asbestos exposure, and they were di- or absence of asbestosis. They found that a lower lobe origin
vided into three groups according to the amount of fiber of lung cancer was found 63% of the time in asbestos-
burden. Five of 19 men in the high group (greater than exposed patients and only 25% of the time in unexposed
lo6 fibers per gram of dry lung) had microscopic asbestosis, patients. This difference was statistically significant. When
as did seven of 29 in the intermediate group (between the eight patients with histologic asbestosis were excluded
404 Egilman and Reinert

TABLE 111. Summary of Relevant Studies Relating t o Histologic Asbestosis

Reference Study population Data on lung cancer and asbestos exposure

Williams, 1957
Warnock and Churg, 1975
Warnock and Isenberg, 1986
Sluis-Cremer and Bezuidenhout,
1989
Merrill et al., 1991
Kishimoto, 1992
Huilan and Zhiming, 1993
Karjaiainen et al., 1993
Examined lungs of four occupational groups thought
to be at increased risk of lung cancer
Autopsy study of 100 controls and 30 patients with
lung cancer
75 men with lung cancer
Necropsy study of 339 asbestos miners
50 asbestos-exposed men
104 lung cancer patients
5,893 asbestos factory employees
108 lung cancer patients
Found that alveolar metaplasia was not related to the
development of lung cancer
Found higher fiber counts in lung cancer patients;
fibrosis was absent or minimal in all of the cancer
patients
Found that high fiber burdens did not always result in
asbestosis; concluded that asbestosis was a poor
marker for asbestos-related lung cancer
OR for slight pathological asbestosis = 4.45; 5.2 for
moderate pronounced pathological asbestosis.
Years of exposure accounted for most of variation
Airway metaplasia was seen with equal frequency
among asbestos-exposed subjects separated by ILO
X-ray category. More severe fibrosis did not lead to
an increase in observed airway metaplasia
21 of 39 lung cancers attributed to asbestos exposure
did not have histologic asbestosis
Increased risk (RR = 2.68) for workers without
pathologic asbestosis (our reanalysis)
Asbestos-exposed patients without histologic
asbestosis had significantly more lower lobe lung
cancers than unexposed

from the analysis, 57% of the asbestos-exposed patients had
lower lobe lung cancers. The difference between this group
and the unexposed group was still statistically significant (p
= 0.003). The authors interpret this study as evidence that
asbestos may increase the risk of lung cancer in the absence
of histologic asbestosis.
Finally, Kishimoto conducted a microscopic study in
Japan in which he similarly found no correlation between
histologic asbestosis and lung cancer incidence [Kishimoto,
19921. Kishimoto counted the fibers in the lungs of a sample
of I04 lung cancer cases and found 39 who had high levels
of asbestos fibers. In these cases, he attributed the cancers to
asbestos exposure. Of these 39, only 18 had histologic as-
bestosis, while 21 did not [Kishimoto, 19931. Kishimoto
attributed these lung cancers to asbestos exposure regard-
less of the presence of asbestosis.
DISCUSSION
The epidemiological evidence relevant to the question
of the relationship between asbestos, asbestosis, and lung
cancer has been summarized (Tables 11, 111). A causal
framework has been provided which demonstrates that the
notion that asbestosis is necessary for attributing a lung
cancer to asbestos exposure is not consistent with accepted
standards of causation. One well executed study which finds
an increased risk for lung cancer among people exposed to
asbestos, but without asbestosis, disproves the hypothesis.
At the very least we have shown that the asbestosis is
necessary position can possibly be supported by only one
study [Hughes and Weill, 19911. We now turn to the inher-
ent contradictions in the advocates of this position.
Contradictions in the Asbestosis is
Necessary Position
Some proponents of the asbestosis is necessary po-
sition have used different definitions of asbestosis in eval-
uating epidemiologic studies and in evaluating asbestosis in
individuals. For example, one author defined clinical asbes-
tosis as chest film showing lower zone interstitial disease
of category 1/1 or greater and evidence of restrictive lung
disease and decreased diffusing capacity [Churg, 19931.
The study principally relied on to support this authors view
that asbestosis is necessary in order to attribute lung cancer
to asbestos exposure defined asbestosis as X-ray changes
greater than or equal to ILO 1/0, and contained no infor-
mation on pulmonary function [Hughes and Weill, 1991;
Churg, 19931. Hughes and Weill [ 19911 did not present data
regarding the cancer experience of individuals with X-rays
classified as 1/1 or greater. We would be interested to see
the results of this analysis.

The advocates of the asbestosis is necessary model
use a very restrictive definition of asbestosis when they
evaluate individual patients. On the other hand, they rely on
studies which use encompassing definitions of to
state that an individuals lung cancer is not related to as-
bestos exposure because the individual does not meet their
narrow clinical criteria for asbestosis. This leads to an un-
derdiagnosis of asbestos-related lung cancer, This manipu-
-
lation is bad science, unfair to individuals being evaluated
for asbestos-related illnesses, and leads to poorly informed
policy decisions regarding risks of asbestos-related disease
to the public.
CONCLUSION
Pathological reports and epidemiological studies have
been reviewed to show that asbestos has the potential to act
as a carcinogen independent of pathologic or clinical asbes-
tosis. Asbestos meets accepted criteria for causation of lung
cancer in the absence of clinical or histologic parenchymal
asbestosis. The possibility of a separate causal pathway for
asbestos-induced lung cancer that involves fibrotic change
should be left open, although there is little evidence to ex-
clusively support this hypothesis. Dr. Abraham [ 19941 was
correct in stating that this debate has been motivated more
by litigation than by medical relevance. We would add,
however, that this discussion is extremely important in
terms of using appropriate standards to determine causa-
tion, of evaluating the risk of lung cancer for low levels of
exposure to asbestos, and of making reliable public policy
decisions regarding asbestos-related disease. The practice of
attributing lung cancer to asbestos exposure only in the
presence of asbestosis should be questioned. Rather than the
presence of asbestosis, causation should be determined by
considering duration of exposure, intensity of exposure, and
appropriate latency.
REFERENCES
Abraham JL (1994): Asbestos inhalation, not asbestosis, causes lung can-
cer. Am J Ind Med 26:829-842.
Armstrong BK, de Klerk NH, Musk AW, Hobbs MST (1988): Mortality in
miners and millers of crocidolite in Western Australia. Br J Ind Med
45:5 -1 3.
Barbone F, Delzell E, Austin H, Cole P (1992): A casexontrol study of
lung cancer at a dye and resin manufacturing plant. Am J Ind Med 22:
835 -849.
Berry G, Newhouse ML, Antonis P (1985): Combined effect of asbestos
and smoking on mortality from lung cancer and mesothelioma in factory
workers. Br J Ind Med 4212-18.
Blot WJ, Harrington JM, Toledo A, Hoover R, Heath CW, Fraumeni JF
(1978): Lung cancer after employment in shipyards during World War 11.
N Engl J Med 299:620-624.
Cancer Without Asbestosis: Causal Issues 405
Bovenzi M, Stanta G, Antiga G, Peruzzo P, Cavallieri F (1993): Occupa-
tional exposure and lung cancer risk in a coastal area of Northeastern Italy.
Int Arch Occup Environ Health 65:35-41.
Braun D, Truan D ( 1958): An epidemiological study of lung cancer in
asbestos miners. AMA Arch Ind Health 17:634-653.
Brown DP, Dement JM, Okun A (1994): Mortality patterns among female
and male chrysotile asbestos textile workers. J Occup Med 36:882-887.
Ca&lem, Cohle SD, Greenberg SD ( 1985): Natural history of pulmonary
scar cancers. Cancer 56903 1-2035.
Cheng W, Kong J (1992): A retrospective mortality cohort study of
chrysotile asbestos products workers in Tianjin 1972-1987. Environ Res
59:271-278.
Churg A (1993): Asbestos-related disease in the workplace and the envi-
ronment: Controversial issues. Monogr Pathol 36:54-77.
Churg A, Wiggs B (1989): The distribution of amosite asbestos fibers in
the lungs of workers with mesothelioma or carcinoma. Exp Lung Res
15:771-783.
de Klerk NH, Musk AW, Cookson WOCM, Glancy JJ, Hobbs MST
(1993): Radiographic abnormalities and mortality in subjects with expo-
sure to crocidolite. Br J Ind Med 50:902-906.
Dement JM, Brown DP, Okun A (1994): Follow-up study of chrysotile
asbestos textile workers: Cohort mortality and case-control analyses. Am
J Ind Med 26:431-447.
Edge JR ( 1976): Asbestos related disease in Barrow-in-Furness. Environ
Res 11:24.1-247.
Edge JR (1979): Incidence of bronchial carcinoma in shipyard workers
with pleural plaques. Ann NY Acad Sci 330:289-294.
Fletcher DE (1972): A mortality study of shipyard workers with pleural
plaques. Br J Ind Med 29:142-145.
Haider M, Neuberger M ( 1980): Comparison of lung cancer risks for dust
workers, asbestos-cement workers, and control groups. IARC Sci Pub 30:
973-971.
Hammond EC, Selikoff IJ, Seidman H (1979): Asbestos exposure, ciga-
rette smoking and death rates. Ann NY Acad Sci 330:473-490.
Hillerdal G (1994): Pleural plaques and risk for bronchial carcinoma and
mesothelioma. Chest 105:144-150.
Hillerdal G, Nou E (1979): Occupation and bronchial carcinoma. Scand J
Respir Dis 60:76-82.
Hughes J, Weill H (1980): Lung cancer risk associated with manufacture
of asbestos-cement products. IARC Sci Pub 30:627-635.
Hughes JM, Weill H (I991 ): Asbestosis as a precursor of asbestos related
lung cancer: Results of a prospective mortality study. Br J Ind Med 48:
229-233.
Hughes JM, Weill H, Hammad YY (1987): Mortality of workers employed
in two asbestos cement manufacturing plants. Br J Ind Med 44:161-174.
Huilan 2, Zhiming W (1993): Study of occupational lung cancer in asbes-
tos factories in China. Br J Ind Med 50:1039-1042.
de vos Irvine H, Larnont DW, Hole DJ, Gillis CR ( 1993): Asbestos and
lung cancer in Glasgow and the west of Scotland. Br Med J 306:1503-
1506.
Jtirvholm B, Larsson S, Hagberg S, Olling S, Ryd W, Toren K (1993):
Quantitative importance of asbestos as a cause of lung cancer in a Swedish
industrial city: A case-referent study. Eur Respir J 6: 1271-1275.
Karjalainen A, Anttila S, Heikkila L, Kyyronen P, Vainio H (1993): Lobe
of origin of lung cancer among asbestos-exposed patients with or without
diffuse interstitial fibrosis. Scand J Work Environ Health 19: 102-107.
406 Egilman and Reinert
Kishimoto T (1992): Cancer due to asbestos exposure. Chest 101:58-63.
Kishimoto T (1993): The attribution of lung cancer to asbestos [Letter to
the Editor]. Chest 103:654-655.
Liddell FDK, McDonald JC (1980): Radiological findings as predictors of
mortality in Quebec asbestos workers. Br J Ind Med 37:257-267.
Mark EJ, Shin D-H (1992): Asbestos and the histogenesis of lung carci-
noma. Semin Diagn Pathol 9: 110-1 16.
Martischnig KM, Newell DJ, Barnsley WC, Cowan WK, Feinman EL,
Oliver E (1977): Unsuspected exposure to asbestos and bronchogenic car-
cinoma. Br Med J 1:746-749.
McDonald JC, Liddell FDK (1979): Mortality in Canadian miners and
millers exposed to chrysotile. Ann NY Acad Sci 33O:l-9.
McDonald JC, Liddell FDK, Gibbs GW, Eyssen GE, McDonald AD
(1980): Dust exposure and mortality in chrysotile mining, 1910-1975. Br
J Ind Med 37: 11-24.
Merrill WW, Carter D, Cullen MR (1991): The relationship between large
airway inflammation and airway metaplasia. Chest 100:131-1 35.
Montizaan GK, Knaap AGAC, van der Heijden CA (1989): Asbestos:
Toxicology and risk assessment for the general population in the Nether-
lands. Food Chem Toxicol 27:53-63.
Morabia A, Markowitz S , Garibaldi K, Wynder EL (1992): Lung cancer
and occupation: Results of a multicentre casexontrol study. Br J Ind Med
49:72 1-727.
Moulin JJ, Wild P, Haguenoer JM, Faucon D, de Gaudemaris R, Mur JM,
Mereau M, Gary Y , Toamain JP, Birembaut Y, Blanc M, Debiolles MP,
Jegaden D, Laterritre, Lkonard M, Marini F, Massardier C, Moulin M,
Reure M, Riga1 L, Robert G, Viossat M (1993): A mortality study among
mild steel and stainless steel welders. Br J Ind Med 50:234-243.
Nicholson WJ, Selikoff IJ, Seidman H, Lilis R, Formby P (1979): Long-
term mortality experience of chrysotile miners and millers in Thetford
Mines, Quebec. Ann NY Acad Sci 33O:ll-21.
Pet0 J (1980): Lung cancer mortality in relation to measured dust levels in
an asbestos textile factory. IARC Sci Pub 30:829-836.
Raffn E, Lynge E, Korsgaard B (1993): Incidence of lung cancer by his-
tological type among asbestos cement workers in Denmark. Br J Ind Med
50:85-89.
Roggli VL, Hammar SP, Pratt PC, Maddox JC, Legier J, Mark EJ, Brody
AR (1994): Does asbestos or asbestosis cause carcinoma of the lung? Am
J Ind Med 26:835-838.
Rothman KJ (1986): Modern Epidemiology. Boston: Little, Brown &
Co., pp 1-21.
Rudd RM (1990): Relation between asbestos and bronchial cancer in am-
phibole asbestos miners [letter]. Br J Ind Med 47:215.
Selikoff IJ, Hammond EC, Churg J (1968): Asbestos exposure, smoking,
and neoplasia. JAMA 204106-1 10.
Selikoff IJ, Hammond EC, Seidman H (1979): Mortality experience of
insulation workers in the United States and Canada, 1943-1976. Ann NY
Acad Sci 330:91-116.
Sluis-Cremer GK, Bezuidenhout BN (1989): Relation between asbestosis
and bronchial cancer in amphibole asbestos miners. Br J Ind Med 46:537-
540.
Sluis-Cremer GK, Bezuidenhout BN (1990): Relation between asbestosis
and bronchial cancer in amphibole asbestos miners [letter]. Br J Ind Med
47:2 15-2 16.
Talcott JA, Thurber WA, Kantor AF, Gaensler EA, Danahy JF, Antman
KH, Li FP (1989): Asbestos-associated diseases in a cohort of cigarette-
filter workers. N Engl J Med 321:1220-1223.
Walker C , Everitt J, Barrett JC (1992): Possible cellular and molecular
mechanisms for asbestos carcinogenicity. Am J Ind Med 21:253-273.
Warnock ML, Churg AM (1975): Association of asbestos and bron-
chogenic carcinoma in a population with low asbestos exposure. Cancer
35: 1236-1242.
Warnock ML, Isenberg W (1986): Asbestos burden and the pathology of
lung cancer. Chest 89:20-26.
Wilkinson P, Hansel1 DM, Janssens J, Rubens M, Rudd RM, Taylor AN,
McDonald JC (1995): Is lung cancer associated with asbestos exposure
when there are no small opacities on the chest radiograph? Lancet 345:
1074 -1 078.
Williams WJ (1957): Alveolar metaplasia: Its relationship to pulmonary
fibrosis in industry and the development of lung cancer. Br J Cancer
18:322-351.