American Journal of Epidemiology
Copyright © 1999 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol. 149, No. 1
Printed in U.S.A.
Previous Lung Disease and Risk of Lung Cancer among Men and Women
Nonsmokers
Susan T. Mayne,1 Joan Buenconsejo,1 and DwightT. Janerich2
From 1982 to 1984, the authors conducted a population-based case-control study of lung cancer in men and
women nonsmokers in New York State. In-person interviews were completed for 437 lung cancer cases (197
never smokers, 240 former smokers) and 437 matched population controls. Cases and controls were asked to
report any history of physician-diagnosed nonmalignant lung disease; cases were more likely than controls to
report such a history. Statistically significant associations were found for emphysema (odds ratio (OR) = 1.94,
95% confidence interval (Cl) 1.10-3.43), chronic bronchitis (OR = 1.73, 95% Cl 1.10-2.72), and the combined
endpoint of emphysema, chronic bronchitis, or asthma (OR = 1.82, 95% Cl 1.26-2.63). After adjustment for
active and passive tobacco smoke exposure, emphysema, chronic bronchitis, and asthma (each condition and
the combined endpoint) were significantly associated with lung cancer risk. The risk was more marked for
squamous cell carcinomas and for subjects who were diagnosed at older ages, and it remained significant when
surrogate interviews were excluded. These results are consistent with the hypothesis that certain prior lung
conditions increase the risk of lung cancer in men and women nonsmokers. Am J Epidemiol 1999;149:13-20.
asthma; bronchitis; case-control studies; emphysema; lung diseases; lung neoplasms
A number of nonmalignant lung conditions, such as
chronic bronchitis, emphysema, asthma, pneumonia,
tuberculosis, silicosis, and asbestosis, have been associated with an increased risk of lung cancer (1-9).
Some of these conditions, such as emphysema and
chronic bronchitis, are strongly associated with tobacco smoking, with an obvious potential for residual
confounding. The problem of residual confounding by
active smoking can be overcome by studying previous
lung diseases as a risk factor for lung cancer in nonsmokers. While it is difficult to assemble a case group
of nonsmokers with lung cancer, a few studies of nonsmokers, particularly women, have been done. For
example, Wu et al. (1) reported that in the United
States, women lifetime nonsmokers with lung cancer
had a statistically significant excess of any previous
lung disease, asthma, and chronic bronchitis compared
with a control group of women nonsmokers. Alavanja
et al. (2) also studied US women who were nonsmokers and found significant risks for asthma and pneu-
monia among lifetime nonsmokers and for emphysema and tuberculosis among former smokers.
We completed a population-based case-control
study of lung cancer in men and women nonsmokers in
the United States; some of the findings have been published previously (10, 11). This study provided a
unique opportunity to examine risk factors for lung
cancer in a relatively large number of men and women
nonsmokers. Moreover, as these data were originally
obtained primarily to evaluate passive smoke exposure
as a risk factor for lung cancer in nonsmokers, this
study enabled us to consider the confounding effects of
environmental tobacco smoke, as well as a number of
other potential confounders, when examining the association between previous lung disease and lung cancer
risk. This report summarizes the major findings of our
evaluation of previous lung diseases as a risk factor for
lung cancer in men and women nonsmokers.
MATERIALS AND METHODS
A population-based, individually matched casecontrol study of lung cancer in nonsmokers was conducted in New York State from 1982 to 1984. The data
collection methods have been described previously
(10, 11). The procedures used were approved by the
Institutional Review Board of the New York State
Department of Health. Informed consent was obtained
from all participants.
Received for publication December 9, 1997, and accepted for
publication May 5, 1998.
Abbreviations: Cl, confidence interval; OR, odds ratio.
1
Department of Epidemiology and Public Health, Yale University
School of Medicine, New Haven, CT.
2
Foundation for Blood Research, Portland, ME.
Reprint requests to Dr. Susan T. Mayne, Department of
Epidemiology and Public Health, Yale University School of Medicine,
60 College Street, New Haven, CT 06520-8034.
13
14
Mayne et al.
The study area comprised eight Standard
Metropolitan Statistical Areas (23 counties) of upstate
New York. A special system to enable rapid ascertainment of lung cancer cases was established in this
region so that diagnosed patients could be identified
and interviewed as soon as possible. Patients were typically identified via the medical records department,
the pathology department, or the tumor registry at the
institutions in this region.
To be included as a case in the study, a patient had
to reside in the 23-county region, be aged 20-80 years,
have never smoked more than 100 cigarettes (nonsmokers), or have smoked at some time but not more
than 100 cigarettes in the 10 years prior to diagnosis
(former smokers). Cases also must have been given a
diagnosis of primary lung cancer between July 1,
1982, and December 31, 1984. Initial diagnoses were
confirmed by clinical records and by reexamination of
pathologic specimens. Slides or blocks of tissue were
available for all but five cases. All materials were
reviewed by investigators, who were masked with
respect to the patient's initial diagnosis, smoking history, and other risk factors. Interviews were conducted
with 76 percent of the eligible patients or their closest
available relatives or friends (surrogates).
One population control was selected for each case.
Controls were selected randomly from the New York
State Department of Motor Vehicles' file of licensed
drivers. For each case, six potential controls were identified on the basis of age (±5 years), sex, and county of
residence. Potential controls were called and screened
for smoking status. The first potential control whose
smoking status matched that of the case and who
agreed to participate in the study was included. On
average, two potential controls had to be contacted to
obtain one who matched the case and was willing to
participate.
The field staff conducted face-to-face interviews in
the homes of all cases and controls. Information was
collected by using a precoded questionnaire that took
approximately 1 hour to complete. Cases and controls
were interviewed in the same manner and with the
same questionnaire. No attempt was made to mask
interviewers to case-control status because of the
often-obvious evidence of disease among lung cancer
patients.
Despite efforts to identify and interview all cases
rapidly, surrogate interviews were needed for approximately one third of all cases. When a surrogate respondent was used for the case, a surrogate respondent was
also used for the matched control (with the exception
of nine matched controls for whom surrogate control
interviews were not matched for surrogate case interviews). Thus, cases and controls were matched for sex,
age, smoking history, county of residence, and interview type (self or surrogate).
Cases and controls were asked whether a physician
had ever told them that they had ever had certain lung
conditions. If they answered yes, they were asked to
report the number of times that they had been told, the
year in which the condition was first diagnosed, the
year in which they were first hospitalized for the condition, the number of times that they were hospitalized
for the condition, and whether they received medication for the condition.
Descriptive statistics were first used to characterize
the study population. Case-control differences in variables associated with prior lung diseases (number of
events, number of years since first hospitalization,
etc.) were evaluated by using either the Student's t test
or the Wilcoxon's test as appropriate to test statistical
significance. Odds ratios for prior lung diseases were
then estimated by conditional logistic regression, as
described by Holford et al. (12). This method is based
on the linear logistic model described by Cornfield et
al. (13) and takes into account pairwise matching in
case-control studies. To determine whether the relation
between prior lung diseases and lung cancer risk differed according to interview type, smoking status, sex,
age, or histologic type, the relevant case-control pairs
from each subgroup were examined separately, and
odds ratios were calculated as described previously.
The attributable risk percentage for the population was
calculated as described by Kelsey et al. (14).
To address the issue of confounding, several potential confounders (interview type, smoking history, sex,
county of residence, and age) were matched and thus
were not included in the logistic models. A variety of
measures of prior active smoking among the former
smokers were evaluated for potential confounding,
including the number of cigarettes smoked per day,
years of smoking, pack-years of exposure, and years
since quitting. Only the first measure was found to differ by case-control status and to affect the risk estimates; therefore, it was included in multivariate models. Passive smoke exposure was quantified as
described previously (10) and was included in multivariate models. Similarly, dietary variables were quantified as described previously (11) and were included in
multivariate models. Education was dichotomized for
adjustment on the basis of tree-based regression analyses (15). All other adjustment variables were modeled
as continuous variables. The final multivariate models
included the following potential confounders: tobaccorelated variables (cigarettes per day, lifetime passive
smoke exposure), dietary variables (frequency of consumption of fresh fruit and vegetables, whole milk, and
supplemental vitamin E), and education.
Am J Epidemiol
Vol. 149, No. 1, 1999
Previous Lung Disease and Lung Cancer Risk
It is conceivable that the cases might have had early
symptoms of lung cancer that could have been diagnosed incorrectly as prior lung disease, which could
artificially increase the magnitude of any association
between prior lung diseases and lung cancer risk. To
examine this possibility, we reran the analyses by
restricting the data set to those previous lung diseases
that were first diagnosed more than 5 years before the
date of interview.
C\j h-;
CO CO
C\J CM
b
n
1—
i
V4J
« CO
o ^
RESULTS
The sociodemographic characteristics of the 437
case-control pairs included in this analysis are summarized in table 1. Of the cases, approximately one half
were female and 45 percent were never smokers.
Adenocarcinoma was the predominant (51 percent)
histologic type among the cases. This distribution of
lung cancer by histologic type in nonsmokers differed
from the overall distribution in the early 1980s, when
squamous cell carcinoma was predominant (16). The
duration of smoking and the number of years since
quitting were similar for cases and controls who were
former smokers, although among those who had
smoked previously, the intensity of smoking was significantly greater among the cases than the controls
(mean number of cigarettes per day, 28.2 vs. 23.7).
As shown in table 2, cases were more likely than
their matched controls to report a history of physiciandiagnosed nonmalignant lung diseases. Odds ratios
were elevated for all conditions about which they were
queried except for punctured lung, reported by only
two cases and four controls, and for influenza, reported by 86 cases and 86 controls. Physician-diagnosed
emphysema (odds ratio (OR) = 1.94, p = 0.02) and
chronic bronchitis (OR = 1.73, p = 0.02) were significantly associated with lung cancer risk. Physiciandiagnosed asthma also notably increased lung cancer
risk (OR = 1.85), although this effect was not statistically significant in univariate analyses (p = 0.07). A
total of 21 percent of cases (90/437) but only 12 percent of controls (54/437) reported a history of emphysema, chronic bronchitis, or asthma (OR = 1.82, p =
0.001). When all prior lung conditions were considered together, a greater proportion of cases than of
controls reported being affected (54 vs. 49 percent).
Data on the risk associated with a history of emphysema, chronic bronchitis, or asthma, by sex, smoking
status, age, interview type, and histology, are shown in
table 3. The odds ratios indicate that in men, chronic
bronchitis was associated with a larger risk (OR =
2.08) than either emphysema (OR = 1.53) or asthma
(OR = 1.43). However, in women, the converse was
true; emphysema (OR = 4.00) and asthma (OR = 2.33)
were associated with greater risk than chronic bronchiAm J Epidemiol
Vol. 149, No. 1, 1999
s:
%
<n 6
cq cq
oi r^
CM C\J
31
fl
CO
CD
<§ £
V
m <
m 2
< g.
<o T -
ii
£ cu ii
o II £. -2 II S
"g-Sto
cSw
CD CO
O
O
1 a§
l o oo | o o
15
16
Mayne et al.
TABLE 2. Previous nonmalignant lung diseases and risk of lung cancer among men and women
nonsmokers, New York Lung Cancer Study, 1982-1984
Cases
(no.)
Controls
(no.)
Odds
ratio*
95%
confidence
interval
Tuberculosis
Yes
No
12
421
10
425
1.20
0.52-2.79
0.67
Pneumonia
Yes
No
123
307
104
322
1.23
0.90-1.68
0.19
Emphysema
Yes
No
36
398
19
417
1.94
1.10-3.43
0.02
Influenza
Yes
No
86
340
86
345
1.00
0.70-1.42
1.00
Chronic bronchitis
Yes
No
56
379
33
402
1.73
1.10-2.72
0.02
Asthma
Yes
No
26
411
15
421
1.85
0.95-3.59
0.07
Punctured lung
Yes
No
2
433
4
432
0.50
0.09-2.75
0.42
Bulbar polio
Yes
No
2
410
1
422
1.98
0.18-21.85
0.57
Lung infection
Yes
No
66
368
56
373
1.19
0.80-1.76
0.38
Any prior lung condition
Yes
No
235
202
214
223
1.23
0.93-1.63
0.14
Emphysema, chronic bronchitis,
or asthma
Yes
No
90
347
54
383
1.82
1.26-2.63
0.001
Any prior lung conditionf
Yes
No
207
203
192
245
1.16
0.88-1.52
0.30
Condition
P
value
* Crude odds ratio from conditional logistic regression analyses,
t Excluding emphysema, chronic bronchitis, or asthma.
tis (OR = 1.50). Among never smokers (many of
whom were women), emphysema increased risk threefold. Asthma was not an important risk factor in never
smokers (OR = 1.10) but was associated with substantial risk in former smokers (OR = 4.33). All three conditions were more important risk factors for lung cancer diagnosed at older versus younger ages, and all
three were significantly associated with lung cancer
risk when analyses were restricted to those pairs interviewed in person. Both emphysema and chronic bronchitis more than doubled the risk of squamous cell can-
cers (p < 0.05) but contributed little toward the risk of
adenocarcinomas in this study population.
Since the intensity of smoking among cases who
were former smokers was greater than that of their
matched controls, some of these associations could be
due to residual confounding by active smoking. Also,
exposure to environmental tobacco smoke during childhood increases the frequency of physician-diagnosed
bronchitis (17), and exposures during adulthood
reduce pulmonary function (17). Therefore, multivariate analyses were done to generate risk estimates, and
Am J Epidemiol Vol. 149, No. 1, 1999
Previous Lung Disease and Lung Cancer Risk
TABLE 3. History of emphysema, chronic bronchitis, or
asthma and risk of lung cancer among men and women
nonsmokers, New York Lung Cancer Study, 1982-1984
Pairs*
(no.)
Odds
ratiot
95%
confidence
interval
P
value
217
216
1.53
4.00
0.80-2.94
1.13-14.18
0.20
0.03
Smoking status
Former smoker
Never smoker
239
194
1.81
3.01
0.98-3.34
0.61-14.86
0.06
0.18
Age at interview (years)
>70
<70
187
220
2.30
1.57
1.09-4.83
0.61^1.05
0.03
0.35
Interview type
Self
Proxy
294
128
2.18
1.57
1.07-4.45
0.61-4.05
0.03
0.35
Histology
Squamous
Adenocarcinoma
Other
108
220
105
2.43
1.14
2.50
1.01-5.86
0.41-3.15
0.78-7.97
0.04
0.80
0.12
218
215
2.08
1.50
1.05-4.15
0.83-2.72
0.04
0.18
Smoking status
Former smoker
Never smoker
238
195
1.88
1.57
1.02-3.44
0.80-3.07
0.04
0.19
Age at interview (years)
>70
<70
186
221
2.20
1.42
1.04-4.65
0.79-2.56
0.04
0.24
Interview type
Self
Proxy
293
129
1.73
2.00
1.02-2.92
0.81^.95
0.04
0.13
Histology
Squamous
Adenocarcinoma
Other
108
220
105
2.86
1.22
2.00
1.21-6.76
0.66-2.28
0.68-5.85
0.02
0.53
0.21
Emphysema
Sex
Male
Female
Chronic bronchitis
Sex
Male
Female
Asthma
Sex
Male
Female
219
217
1.43
2.33
0.54-3.75
0.90-6.07
0.47
0.08
Smoking status
Former smoker
Never smoker
239
197
4.33
1.10
1.23-15.21
0.47-2.59
0.02
0.83
Age at interview (years)
>70
<70
188
222
2.25
1.56
0.69-7.31
0.67-3.59
0.18
0.30
Interview type
Self
Proxy
295
130
2.50
0.80
1.10-5.68
0.21-2.98
0.03
0.74
Histology
Squamous
Adenocarcinoma
Other
108
220
105
2.25
1.63
2.00
0.69-7.31
0.67-3.92
0.18-22.05
0.18
0.28
0.57
* One population control was selected for each case.
t Crude odds ratio from conditional logistic regression analyses.
we took into account these and other potential confounding variables. As shown in table 4, emphysema,
chronic bronchitis, and asthma were all significantly
associated with lung cancer risk following adjustment
for active and passive tobacco smoke exposure (model
1) but were no longer significant after additional
adjustment for dietary variables and education. The
Am J Epidemiol Vol. 149, No. 1, 1999
17
TABLE 4. Multivariate analysis of history of emphysema,
chronic bronchitis, or asthma and risk of lung cancer among
men and women nonsmokers, New York Lung Cancer Study,
1982-1984
Pairs*
(no.)
Odds
ratiot
95%
confidence
interval
P
value
Emphysema
Univariate
Model 1f
Model 2%
Model 3§
Model 4H
433
433
384
384
378
1.95
1.89
1.82
1.64
1.48
1.11-3.45
1.06-3.36
0.98-3.40
0.87-3.12
0.77-2.84
0.02
0.03
0.06
0.12
0.24
Chronic bronchitis
Univariate
Model 1
Model 2
Model 3
Model 4
433
433
383
383
376
1.73
1.68
1.46
1.33
1.30
1.10-2.72
1.07-2.64
0.90-2.35
0.82-2.16
0.79-2.14
0.01
0.03
0.13
0.25
0.30
Asthma
Univariate
Model 1
Model 2
Model 3
Model 4
436
436
387
387
380
1.84
2.05
1.74
1.87
1.97
0.95-3.58
1.03-4.06
0.82-3.71
0.87-4.00
0.90-4.32
0.07
0.04
0.15
0.11
0.09
Emphysema, chronic
bronchitis, or asthma
Univariate
Model 1
Model 2
Model 3
Model 4
437
437
387
387
380
1.82
1.81
1.62
1.53
1.44
1.26-2.64
1.25-2.62
1.09-2.42
1.02-2.30
0.95-2.19
0.001
0.002
0.02
0.04
0.08
* One population control was selected for each case.
t Adjusted for smoking-related variables (cigarettes smoked/day, lifetime
passive smoke exposure).
% Adjusted for dietary variables (frequency of consumption of fresh fruits
and vegetables, whole milk, and vitamin E supplements).
§ Adjusted for smoking-related variables and dietary variables.
H Adjusted for smoking-related variables, dietary variables, and
educational level (<15 years vs. >15 years).
three conditions combined remained statistically significant in all models except the final one (model 4).
The loss of significance after adjustment may partially
reflect a loss of statistical power (a decrease in the
number of pairs with informative data); however, the
point estimates for chronic bronchitis and emphysema,
but not for asthma, were reduced after adjustment.
Cases and controls who reported having a given
lung condition were queried about the number of
events, the year in which they were first diagnosed
with the condition, the year in which they were first
hospitalized for the condition, the number of hospitalizations for the condition, and whether medication was
given for the condition. The reports from cases and
controls with a given lung condition were similar for
these variables, and only two of the case-control comparisons were statistically significant. First, cases were
significantly older at first hospitalization for pneumonia than were controls (median 55.5 years vs. 40.3
years, p = 0.04). Second, the proportion of cases and
controls with a given lung condition who received
medication for that condition was similar for all conditions except emphysema (28 percent of cases vs. 58
18
Mayne et al.
percent of controls received medication for that condition, p = 0.035).
We also performed analyses in which we eliminated
data on previous lung diseases that reportedly were
first diagnosed less than or equal to 5 years before the
date of interview. As the 5-year cutpoint is entirely
arbitrary, these results should be interpreted cautiously. Nonetheless, this analytic approach should provide
a more conservative estimate of effect. The numbers of
cases and controls with a given lung condition diagnosed more than 5 years before the date of interview
were as follows: asthma, 19 cases versus 14 controls
(OR = 1.41, 95 percent CI 0.68-2.97); emphysema, 24
cases versus 15 controls (OR = 1.60, 95 percent CI
0.84-3.05); and chronic bronchitis, 39 cases versus 26
controls (OR = 1.54, 95 percent CI 0.92-2.58).
DISCUSSION
This study adds to the evidence that certain previous
nonmalignant lung diseases are a risk factor for lung
cancer. As this study was done with nonsmokers and
included assessment of potential confounding by both
active and passive smoking, it would seem that the
associations observed are not likely to be attributable
to tobacco smoke exposure. When all lung diseases
were considered together, a history of any previous
lung disease increased the odds of lung cancer by 23
percent (nonsignificant). The point estimates indicate
that three conditions—emphysema, chronic bronchitis,
and asthma—increased lung cancer risk by 50-100
percent. This increased risk was statistically significant for all three conditions after taking into account
smoking history (former smokers) and passive smoke
exposure (table 4, model 1). Further adjustment for
dietary variables (models 2 and 3) and for education
(model 4) reduced the magnitude of the risk associated with emphysema and chronic bronchitis, but adjustment did not reduce the risk associated with asthma.
However, in this situation the appropriateness of
adjustment for diet and education, a proxy for socioeconomic status, is debatable, given that these variables may well be part of the causal path (i.e., dietary
factors are associated with airways obstruction
(18-19)). Thus, models 2-4 are included for the sake
of completeness. However, perhaps model 1 is the
most appropriate and the one most comparable with
other reports on this topic, most of which did not adjust
for dietary variables and only some of which adjusted
for education.
In the control population, the prevalence of asthma
(3.4 percent) and emphysema (4.4 percent) was relatively low and was in the range reported in other studies (summarized by Wu et al. (1)). The prevalence of
chronic bronchitis was somewhat higher (7.6 percent)
and was also in the range reported in other studies (1).
Given the relatively low prevalence of asthma and
emphysema and the similar risk conferred by these
conditions and by chronic bronchitis, the three conditions were grouped together in univariate and multivariate analyses to increase statistical power. There is
a considerable diagnostic overlap for asthma, bronchitis, and chronic obstructive pulmonary disease, making a combined endpoint logical. However, as this
grouping was not a priori, results pertaining to the
combined grouping should be interpreted cautiously.
Results indicate that having any of these three conditions increases the risk of lung cancer by 82 percent (p
= 0.001 (table 2)). Specifics of selected prior lung conditions are described below.
In the full population, in women, in those older than
age 70 years at the time of interview, and for squamous
cell cancers, previous emphysema emerged as a significant risk factor for lung cancer. Others have also
reported that the effect of emphysema is stronger in
older subjects (5). In our study, emphysema was unassociated with the risk of adenocarcinoma of the lung
(OR = 1.14); this finding is consistent with reports by
others (2, 6, 8). The observation that cases with
emphysema were significantly less likely to be given
medication for the condition than were controls with
emphysema may in part reflect the lower educational
(table 1) and income status (data not shown) of cases
versus controls in this study.
Chronic bronchitis emerged as a significant risk factor for lung cancer in the full population, in men, in
former smokers, in those older than age 70 years at the
time of interview, and for squamous cell carcinomas.
The finding that the risk was greater for older subjects
is consistent with findings from other reports (1,5), as
is the lack of risk for adenocarcinomas (2, 6, 8).
Some studies have reported that asthma is associated with a reduced risk of lung cancer (5), whereas others have found that asthma increases the risk (1, 2).
Some studies of smokers suggest effect modification
by sex, with asthma increasing lung cancer risk among
men but not among women (20, 21). Some of the
inconsistencies may derive from the fact that asthmatics may voluntarily abstain from using tobacco, ultimately showing a lower incidence of lung cancer than
the general population. Thus, they would appear to be
protected when compared with the general population.
However, when compared with a control population of
nonsmokers, the risk becomes evident. The finding
that asthma was less strongly associated with adenocarcinoma is consistent with findings from other studies of nonsmokers (2).
Asthma has become increasingly common among
American youth, raising the question of whether lifeAm J Epidemiol
Vol. 149, No. 1, 1999
Previous Lung Disease and Lung Cancer Risk
long asthma increases the risk of lung cancer in nonsmokers. In this study, cases with asthma were more
likely to be diagnosed at older ages than were controls
with asthma (age 31 vs. 18 years, p = 0.13), suggesting
that latency of disease is not an important risk factor.
Cases were also older at first hospitalization (age 46
vs. 21 years, p = 0.19). These differences, while nonsignificant, run counter to those expected, suggesting
that the risk associated with asthma should be interpreted with caution.
This study found that none of the other lung conditions significantly increased the risk of lung cancer.
For example, a history of tuberculosis increased the
risk only slightly (OR =1.2 (table 2)) and nonsignificantly (p = 0.67). However, others have found that a
history of tuberculosis is significantly associated with
lung cancer risk (22). The lack of effect observed in
our study may reflect the low (2.3 percent) prevalence
of tuberculosis in our control population.
How might asthma, emphysema, and chronic bronchitis increase the risk of lung cancer? First-degree relatives of patients with lung cancer or chronic obstructive pulmonary disease have been shown to have
significantly higher age-, sex-, race-, and smokingadjusted rates of impaired pulmonary function (23)
that might enable an increase in circulating toxins and
carcinogens. This might explain our finding that former smokers with asthma, but not never smokers with
asthma, were at a markedly increased risk of lung cancer due to impaired clearance of tobacco carcinogens.
Alternatively, pulmonary diseases cause tissue damage, which results in the release of peptide growth factors from injured cells and inflammatory cells (24).
Inflammatory cells also produce interleukins, interferons, and a host of other molecules that control cell proliferation (24). Cell proliferation (mitogenesis)
increases mutagenesis (25), and a proliferative state
might form a microenvironment favoring clonal
expansion of preneoplastic foci (24, 25).
If the association between certain prior lung conditions and lung cancer risk is causal, and if our risk estimates reflect the true magnitude of the associations,
then attributable risk calculations indicate that asthma
may account for 2.8 percent, emphysema for 4.0 percent, chronic bronchitis for 5.3 percent, all three conditions combined for 9.2 percent, and any prior lung
condition for 10.1 percent of the lung cancer cases in
nonsmokers. These attributable risks are relatively
low, not unexpected given the low prevalence in the
control population, coupled with risk estimates of <2.0
for all conditions in the full population.
The strength of this study lies in the fact that we
were able to analyze data on a relatively large number
of lung cancer cases who were nonsmokers. Also, the
Am J Epidemiol Vol. 149, No. 1, 1999
19
population-based study design is an important
strength. Detailed histories of tobacco smoke exposure
were obtained, and cases were carefully classified by
histologic type. However, the exposure of interest
(previous lung conditions) was based on self-report; no
effort was made to abstract medical records of these
cases to confirm previous diagnoses of nonmalignant
lung conditions. Thus, recall bias is of concern, but the
increased risks associated with only some of these conditions but not all argue against a strong influence of
recall bias in these data. Also, it is conceivable that
some of the excess in prior lung diseases among the
cases may in fact reflect preclinical lung cancer.
Results of analyses excluding events first diagnosed
within 5 years of the date of interview (an arbitrary
cutpoint) show a reduction in the risk estimates;
nonetheless, the excess risk in cases relative to controls was not eliminated when this more conservative
analytical approach was used. Another limitation is
that smoking status was based on self-report; no objective marker such as cotinine was used. Thus, it is possible that cases and controls who truly were smokers
were misclassified as nonsmokers.
In conclusion, this population-based case-control
study of lung cancer in nonsmokers suggests that previous lung disease, specifically emphysema, chronic
bronchitis, and possibly asthma, may increase the risk
of lung cancer. These associations remained significant following adjustment for prior active smoking and
previous exposure to passive smoking, adding further
weight to the suggestions of others that these associations are not due to residual confounding by tobacco.
ACKNOWLEDGMENTS
Supported by grant R01 CA 32088 from the National
Institutes of Health and by institutional funds from the Yale
Cancer Center in New Haven, Connecticut.
REFERENCES
1. Wu AH, Fontham ETH, Reynolds P, et al. Previous lung disease and risk of lung cancer among lifetime nonsmoking
women in the United States. Am J Epidemiol 1995,141:
1023-32.
2. Alavanja MCR, Brownson RC, Boice JD Jr, et al. Preexisting
lung disease and lung cancer among nonsmoking women. Am
J Epidemiol 1992;136:623-32.
3. Wu AH, Henderson BE, Pike MC, et al. Smoking and other
risk factors for lung cancer in women. J Natl Cancer Inst
1985;74:747-51.
4. Samet JM, Humble CG, Pathak DR. Personal and family history of respiratory disease and lung cancer risk. Am Rev
RespirDis 1986;134:466-70.
20
Mayne et al.
5. Osann KE. Lung cancer in women: the importance of smoking, family history of cancer, and medical history of respiratory disease. Cancer Res 1991;51:4893-7.
6. Gao Y-T, Blot WJ, Zheng W, et al. Lung cancer among Chinese
women. Int J Cancer 1987;40:604-9.
7. Wu AH, Yu MC, Thomas DC, et al. Personal and family history of lung disease as risk factors for adenocarcinoma of the
lung. Cancer Res 1988;48:7279-84.
8. Wu-Williams AH, Dai XD, Blot W, et al. Lung cancer among
women in north-east China. Br J Cancer 1990;62:982-7.
9. Blot WJ, Fraumeni JF Jr. Cancers of the lung and pleura. In:
Schottenfeld D, Fraumeni JF Jr, eds. Cancer epidemiology and
prevention. 2nd ed. New York, NY: Oxford University Press,
1996:637-65.
10. Janerich DT, Thompson WD, Varela LR, et al. Lung cancer
and exposure to tobacco smoke in the household. N Engl J
Med 1990;323:632-6.
11. Mayne ST, Janerich DT, Greenwald P, et al. Dietary betacarotene and lung cancer risk in nonsmokers. J Natl Cancer
Inst 1994;86:33-8.
12. Holford TR, White C, Kelsey JL. Multivariate analysis for
matched case-control studies. Am J Epidemiol 1978; 107:
245-56.
13. Cornfield J, Gordon T, Smith W. Quantal response curves for
experimentally uncontrolled variables. Bull Int Statist Inst
1961;38:97-115.
14. Kelsey JL, Thompson WD, Evans AS. Methods in observational epidemiology. New York, NY: Oxford University Press,
1986.
15. Zhang H, Bracken MB. Tree-based risk factor analysis for
preterm delivery and small-for-gestational-age birth. Am J
Epidemiol 1995;141:70-8.
16. Devesa SS, Shaw GL, Blot WJ. Changing patterns of lung cancer incidence by histological type. Cancer Epidemiol
Biomarkers Prev 1991 ;1:29-34.
17. US Department of Health and Human Services. The health
consequences of involuntary smoking: a report of the Surgeon
General. Rockville, MD: US Department of Health and
Human Services, Office on Smoking and Health, 1986.
(DHHS publication no. (CDC) 87-8398).
18. Morabia A, Sorenson A, Kumanyika SK, et al. Vitamin A,
smoking and airway obstruction. Am Rev Respir Dis
1989;140:1312-16.
19. Schwartz J, Weiss ST. Relationship between dietary vitamin C
intake and pulmonary function in the First National Health and
Nutrition Examination Survey (NHANES I). Am J Clin Nutr
1994;59:110-14.
20. Vena JE, Bona JR, Byers TE, et al. Allergy-related diseases
and cancer: an inverse association. Am J Epidemiol 1985;122:
66-74.
21. Reynolds P, Kaplan GA. Asthma and cancer. (Letter). Am J
Epidemiol 1987; 125:539^0.
22. Zheng W, Blot WJ, Liao ML, et al. Lung cancer and prior
tuberculosis infection in Shanghai. Br J Cancer 1987;56:
501-4.
23. Cohen B, Diamond EL, Graves CG, et al. A common familial
component in lung cancer and chronic obstructive pulmonary
disease. Lancet 1977;2:523-6.
24. Sporn MB, Roberts AB. Peptide growth factors and inflammation, tissue repair, and cancer. J Clin Invest 1986;78:329-32.
25. Ames BN, Gold LS. Too many rodent carcinogens: mitogenesis increases mutagenesis. Science 1990;249:970-l.
Am J Epidemiol
Vol. 149, No. 1, 1999