IJC

International Journal of Cancer

Tuberculosis, smoking and risk for lung cancer incidence

and mortality
Seri Hong1,3, Yejin Mok2,3, Christina Jeon1,3, Sun Ha Jee3 and Jonathan M. Samet4
1
Department of Public Health, Yonsei University Graduate School, Seoul, Republic of Korea
2
Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD
3
Department of Epidemiology and Health Promotion, Institute for Health Promotion, Graduate School of Public Health, Yonsei University, Seoul, Republic of
Korea
4
Department of Preventive Medicine, Keck School of Medicine, Institute for Global Health, University of Southern California, Los Angeles, CA

Among the exposures associated with risk for lung cancer, a history of tuberculosis (TB) is one potentially important factor,
given the high prevalence of TB worldwide. A prospective cohort study was conducted to evaluate the associations of preexist-

Cancer Epidemiology
ing pulmonary TB with lung cancer incidence and mortality. The cohort consisted of 1,607,710 Korean adults covered by the
National Health Insurance System who had a biennial national medical examination during 1997–2000. During up to 16 years
of follow-up, there were 12,819 incident cases of lung cancer and 9,562 lung cancer deaths. Using Cox proportional hazards
models and controlling for age, cigarette smoking and other covariates, the presence of underlying TB was significantly asso-
ciated with increased risk for lung cancer incidence (HR 1.37 in men with 95% CI 1.29–1.45; HR 1.49 in women with 95% CI
1.28–1.74) and mortality (HR 1.43 in men with 95% CI 1.34–1.52; HR 1.53 in women with 95% CI 1.28–1.83). We also
observed a dose-response relationship between number of cigarettes smoked daily and lung cancer risk. There was no evi-
dence for synergism between a history of TB and smoking. The elevation in risk is relatively modest, particularly in compari-
son to that from smoking, and a prior history of TB is not likely to be useful risk indicator for clinical purposes. In
populations with high prevalence of TB, it can be considered for incorporation into models for lung cancer risk prediction.

Introduction obstructive pulmonary disease (COPD), interstitial lung dis-

Tobacco smoking has long been the predominant cause of
lung cancer in most locales, regardless of histological type;
and lung cancer is now the leading worldwide cause of can-
cer death.1,2 There are numerous additional causes of lung
cancer including other environmental agents, such as work-
place exposures and air pollution; endogenous risk factors
have also been identified including underlying lung diseases
with scarring. Epidemiological research has identified chronic
Key words: tuberculosis, lung cancer, smoking, cohort study, effect
modifier
Additional Supporting Information may be found in the online
version of this article.
Grant sponsor: Ministry of Food and Drug Safety; Grant number:
14182MFDS977; Grant sponsor: Korean Health Technology R&D
Project, Ministry of Health & Welfare, Republic of Korea; Grant
number: HI14C2686
DOI: 10.1002/ijc.30384
History: Received 27 Oct 2015; Accepted 4 Aug 2016; Online 13
Aug 2016
Correspondence to: Sun Ha Jee, Ph.D., M.P.H., Department of
Epidemiology and Health Promotion, Graduate School of Public
Health, Yonsei University, 50-1 Yonsei-ro, Seodaemun-gu, Seoul
03722, Republic of Korea, Tel.: 82-2-2228-1523, Fax: 82-2-392-8133,
E-mail: jsunha@yuhs.ac
ease and tuberculosis (TB) as associated with increased lung
cancer risk.3–9
With regard to TB, numerous studies have addressed the
association of a history of TB with increased risk for lung
cancer incidence6,9–12 or mortality.12–14 A 2009 systematic
review on the relationship between TB and lung cancer risk
found significantly elevated risk associated with TB, regard-
less of smoking status or elapsed time from the diagnosis of
TB.15 In a systematic review and meta-analysis published in
2011, a previous diagnosis of TB was associated with
increased lung cancer risk (relative risk [RR] 5 1.76 [95%
CI 5 1.49 to 2.08]) with little variation by smoking status.7
Similarly, a pooled analysis from the International Lung Can-
cer Consortium found a lung cancer RR of 1.48 (95%
CI 5 1.17 to 1.87) associated with a history of TB, controlling
for smoking status.8 Recent results from large cohort studies
have also provided evidence supporting an association
between prior TB and lung cancer risk.16–21
Despite a well-developed healthcare system and a high
standard of living for most citizens, Korea still has a substan-
tial TB burden. Data from the national passive surveillance
system show that Korea’s notification rates for TB of all types
were 78.5 and 71.4 per 100,000 population in 2012 and 2013,
respectively.22 However, annual incidence rates calculated by
the World Health Organisation (WHO) were much higher,

Int. J. Cancer: 139, 2447–2455 (2016) V

C 2016 UICC
 2448
What’s new?
Tuberculosis is a leading cause of death from infectious disease worldwide. It also is associated with an increased risk of
lung cancer, the leading source of cancer deaths globally. In Korea, both smoking and tuberculosis occur at high rates, creat-
ing an opportunity to evaluate relationships between these factors and lung cancer. This prospective cohort study of Korean
adults shows that while smoking and tuberculosis are independently associated with increased lung cancer risk, they do not
act synergistically to raise risk. Moreover, compared to smoking, the risk of lung cancer from tuberculosis was modest.
at 100 and 97 per 100,000 for the same years, respectively.
The estimated prevalence rates of active (bacteriologically
confirmed) pulmonary TB were 146 per 100,000 in 2012 and
143 per 100,000 in 2013, comparable with rates in ‘high-TB
burden countries’ in the WHO report.23,24 Overall prevalence
is much higher with consideration of those with prior or
Cancer Epidemiology
latent TB infections who have potential risk for reactivation.
The prevalence of latent TB infection may be as high as 33%
in Korea, similar to worldwide estimates.25,26 Korea also has
a high prevalence rate of smoking, reported as 24.2% in
adults in 2014, largely reflecting smoking among men (43.1%
in males; 5.7% in females).27 Thus, the Korean population
affords an opportunity to further characterize the association
between TB and lung cancer and to explore the joint effects
of smoking and TB.
In this prospective cohort study, we assessed the effects of
preexisting pulmonary TB and smoking on lung cancer inci-
dence and mortality in the Korean Cancer Prevention Study
(KCPS) data for 1997 to 2013. We focused on the indepen-
dent and combined effects of these two risk factors for lung
cancer, estimating the independent effects of smoking and
history of TB and exploring whether there is synergism such
that people with a history of TB are at increased lung cancer
risk from smoking.
Materials and Methods
Study population
The original KCPS study design and baseline characteristics
have been previously described.28 The National Health Insur-
ance Service (NHIS), Korea’s national health care coverage
system, provides biennial health examinations that include
medical examinations and a medical and lifestyle behavioural
questionnaire covering alcohol, tobacco use or history of vari-
ous illnesses. This NHIS database covered about 10% of the
South Korean population at the time of the KCPS baseline.
For this study, we used a similar approach to that of an
earlier analysis28 to assemble a new analytic cohort that com-
prised Korean men and women aged 20 years and above
who were either part of the original cohort or new enrollees.
Both the prior and the new groups had participated in a
NHIS medical evaluation between 1997 and 2000. A total of
1,839,861 possible participants had an examination between
1997 and 2000, with 369,388 enrolled in 1997, 1,290,094 in
1998, 112,821 in 1999 and 67,558 in 2000. All NHIS enrollees
Tuberculosis and lung cancer
underwent standardized examinations (described below) at
local hospitals.
Data collection
The standardized biennial examinations at the local hospitals
collected the following:
Standardized questionnaire. During the biennial visits at
baseline (1997–2000), participants reported on their health
status including information on history of past or current
medical morbidity, alcohol consumption and previous or cur-
rent smoking status, and amount and duration of current
smoking.
Medical examination. In conducting the biennial NHIS
medical examinations, each hospital adhered to internal and
external quality control procedures directed by the Korean
Association of Laboratory Quality Control.28 Participants had
a chest radiograph (X-ray) at each visit. The findings were
classified by a radiologist according to a standard reading as
normal, non-active TB, light pulmonary TB, moderate pul-
monary TB, severe pulmonary TB, suspected pulmonary TB,
or non-TB disease.29 Additional data were also available on
follow-up through the records of the NHIS.
Socio-economic status. Participants’ socioeconomic status
was indirectly assessed using the medical insurance premium
paid per year, in South Korean ‘Won’ (KRW) (1,280
KRW 5 1.00 US Dollar, on Jan 2001) as an index. The pre-
mium is the average amount of money paid by the employee
to their medical insurer, which is assessed based on the
employee’s income and assets, such as ownership of property
and automobiles.30
Mortality data. The vital status of all participants was
tracked from the date of cohort entry through December 31,
2013, and the underlying cause of death for deceased partici-
pants was recorded as reported to the National Statistical
Office.
Follow-up, exposure and outcome classification
The primary exposure variable was having current or prior
TB at baseline, as documented by chest X-ray or past hospi-
talisation for TB. TB categories for the chest radiography
reading included non-active TB and light, moderate or severe
TB, and suspected pulmonary TB. For the main analysis, we
defined preexisting TB using both the chest radiography
Int. J. Cancer: 139, 2447–2455 (2016) V
C 2016 UICC
Hong et al.
finding and the hospitalisation record. All inpatient records
from 1997 through 2013 were captured by the NHIS. Hospi-
talisations for TB were identified by the International Classi-
fication of Diseases, Tenth Edition (ICD-10) codes A15 to
A19. To assess the sensitivity of the findings to exposure clas-
sification, we used alternative classifications of the chest
radiograph, excluding a finding of ‘suspected TB’, and also
using chest radiograph findings only without consideration of
hospitalisation history.
Participants were classified as current cigarette smokers,
never-smokers and ex-smokers if they had ever smoked but
were not considered current smokers. Current smokers were
further classified by the average number of cigarettes smoked
per Day (1–9, 10–19 and 20 or more), and all participants
were asked to report the total duration of smoking in their
lifetime (0, 1–4, 5–9, 10–19, 20–29 and 30 or more years).
Alcohol consumption was assessed by self-reported
amount of mean daily alcohol intake (g/day ethanol). Partici-
pants with insurance premium data were classified into four
groups by quartiles of the premium amount. Underlying dia-
betes as a comorbidity was defined by a combination of hos-
pitalisation due to diabetes (at least one hospitalisation for
diabetes before or during the baseline period), outpatient
treatment for diabetes (at least three visits for diabetes care
within 1 year, before or during the baseline period), self-
reported past history of diabetes and detected hyperglycemia
(fasting glucose  126 mg/dl) in health examination result.
ICD-10 codes E10 to E14 were used to define hospitalisations
as well as outpatient utilisations for diabetes. History of hos-
pitalisation for respiratory diseases was defined as at least
one hospitalisation due to asthma, COPD or pneumonia,
which were identified by ICD-10 codes J42 to J44 (COPD),
J45 to J46 (asthma) and J12 to J18 (pneumonia), before
enrollment.
Primary outcomes, coded as ICD-10 code C34, were inci-
dent lung cancer and mortality from lung cancer between the
date of entry and December 31, 2013. Analyses for these two
outcomes were limited to participants who had provided
information on their smoking history during 1997–2000
(N 5 1,650,908). Incident lung cancer was defined by at least
two or more hospitalisations for lung cancer during the
follow-up period. Mortality from lung cancer was based on
the underlying cause of death on death certificates as
reported to the National Statistical Office.
The follow-up period was up to 16 years, through Decem-
ber 31, 2013. Only the year of enrollment was recorded so
that accrual of follow-up time began on January 1 of the cal-
endar year following the year of survey completion. We
excluded those under age 20 (n 5 1,291), and those with
missing or uncertain responses on baseline smoking status
(n 5 187,662), and ambiguous or absent information on chest
X-ray result (n 5 36,047). We further excluded enrollees with
a past history of any cancer (n 5 5,237) or history of incident
lung cancer or death from lung cancer if the cancer was
detected before enrollment (n 5 1,914). The final sample
Int. J. Cancer: 139, 2447–2455 (2016) V
C 2016 UICC
2449
included 1,607,710 participants. Informed consent was not
specifically obtained because the study involved routinely col-
lected data. The Institutional Review Board of Yonsei Univer-
sity approved the study.
Statistical analysis
Analyses were conducted separately by sex and adjusted for
age at enrollment (using age and the square of age in the
models) and cigarette smoking status, including amount of
smoking. Cox proportional hazards models were used to
evaluate the association of TB at enrollment with risk for
lung cancer. The combined effects of smoking status and
amount with history of TB was assessed by the v2 difference
test for Akaike information criterion (AIC) with one degree
of freedom, comparing the model without interaction terms
Cancer Epidemiology
between the smoking and TB variables and the model with
all interaction terms. To avoid bias from any preexisting dis-
ease that could affect subsequent mortality risk, we carried
out a sensitivity analysis that excluded 514 participants who
died during the first two years of follow-up. All Cox
models were tested for and met the proportional hazards
assumption.
Additionally, we calculated the population attributable
fractions for the lung cancer incidence and mortality attribut-
able to preexisting TB in our study population by smoking
status.31 The exposure prevalence of pulmonary TB and the
associated risk ratio for lung cancer in each smoking status
group were obtained from the study data.
All analyses were conducted using SAS version 9.2 (SAS
Institute Inc, Cary, NC).
Results
The study population of 1,607,710 participants (1,040,794
men and 566,916 women) was largely early middle-aged
(median age 5 42 years) at enrollment (Table 1). At baseline,
60,081 men (5.8% of total) and 19,217 women (3.4% of total)
had preexisting TB as defined, and smoking was substantially
more common in men than in women. Table 2 shows a
follow-up of 23,379,734 person-years and mean follow-up
time of 14.5 years, 9,562 deaths coded as primary carcinoma
of lung and 12,819 incident cases.
Table 2 also gives age- and other covariates-adjusted haz-
ard ratios of lung cancer incidence and mortality, in relation
to preexisting TB and status and amount of cigarette smok-
ing. The findings show that preexisting TB and having ever
smoked were associated with increased risk for lung cancer
incidence and mortality. Current smokers were at greater risk
than former smokers, and among the current smokers, the
hazard ratio increased with amount smoked. The presence of
underlying TB was associated with a significantly increased
risk for both lung cancer incidence and mortality. In current
smokers, the hazard ratios for lung cancer incidence and
mortality were two or three times higher than those for
never-smokers in both sexes. Among current smokers, there
was a dose-response relationship with number of cigarettes
 Cancer Epidemiology

2450

Table 1. Baseline characteristics of participants in the Korean Cancer Prevention Study, 1997–2000
Men (N 5 1,040,794) Women (N 5 566,916)
Preexisting TB Preexisting TB
Yes No p values Yes No p values
1
Age, mean (SD) 50.2 (12.7) 42.5 (11.9) <0.001 50.0 (14.3) 43.7 (13.8) <0.0011
Smoking status (%)
Never-smoker 33.0 31.6 95.2 97.0
Ex-smoker 16.9 13.3 1.6 1.1
Current smoker 50.1 55.1 <0.001 3.2 2.0 <0.001
(Amount/day) 1–9 cigarettes 9.5 9.1 1.8 1.1
10–19 cigarettes 27.9 31.0 1.1 0.7
20 cigarettes 12.6 15.0 <0.001 0.3 0.2 <0.001
Smoking duration (years) (%), among current and former smokers
1–4 7.3 9.6 15.8 20.7
5–9 11.4 18.8 10.8 14.9
10–19 27.3 36.1 16.5 17.7
20–29 23.9 21.0 14.9 14.9
30 26.9 12.5 <0.001 32.7 23.3 <0.001
Missing 3.3 2.1 9.3 8.4
Alcohol consumption (g/day), mean (SD)
12.7 (22.0) 13.7 (21.8) <0.0011 0.7 (3.8) 0.8 (3.6) 0.0011
Socio-economic status (Insurance premium, KRW), mean (SD)
103090.8 (56581.7) 103474.2 (52434.1) 0.1251 104722.6 (52831.8) 100390.3 (50245.2) <0.0011
Morbidity conditions (yes) (%)
Diabetes mellitus 6.4 4.6 <0.001 4.4 3.4 <0.001
Respiratory diseases (Hospitalisation) 1.2 0.7 <0.001 0.9 0.5 <0.001
Total, N (%) 60,081 (5.8) 980,713 (94.2) 19,217 (3.4) 547,699 (96.6)
1
p-Values assessed by student’s t test for the mean difference between two groups.
The others were assessed by v2 test for the difference in distribution of participants between two groups.
Abbreviations: TB: tuberculosis; HR: Hazard Ratio

Int. J. Cancer: 139, 2447–2455 (2016) V

Tuberculosis and lung cancer

C 2016 UICC
Table 2. Respective effects of preexisting TB and smoking on lung cancer incidence and mortality, Korean Cancer Prevention Study 1998–2013
Men Women
Incidence Mortality Incidence Mortality
Person- Person- Person- Person-
years No. HR1 95% CI years No. HR1 95% CI years No. HR1 95% CI years No. HR1 95% CI
Preexisting TB
No 14172364.3 8,976 1.00 Ref 14192653.5 6,697 1.00 Ref 8104368.4 2,270 1.00 Ref 8110625.9 1,550 1.00 Ref
Yes 825402.1 1,398 1.37 1.29–1.45 828260.7 1,182 1.43 277599.1 175 1.49 1.28–1.74 278226.1 133 1.53 1.28–1.83
1.34–1.52
Smoking status
Never-smoker 4767269.8 1,903 1.00 Ref 4772049.1 1,385 1.00 Ref 8147945.9 2,106 1.00 Ref 8154272.4 1,361 1.00 Ref
Ex-smoker 2011821.3 1,362 1.41 1.32–1.51 2015320.1 1,006 1.43 84099.8 71 1.92 1.51–2.45 84240.9 50 1.81 1.35–2.41
1.32–1.55
Current smoker (8218675.3) (7,109) (3.04) (8233545.0) (5,488) (3.40) (149921.8) (268) (2.75) (150338.7) (272) (3.63)
(Amount /day) 1351743.0 978 1.90 1.76–2.05 1353565.9 815 2.15 86320.3 111 1.98 1.63–2.42 86531.7 118 2.71 2.23–3.30
1–9 cig. 1.98–2.35
10–19 cig. 4637348.8 3,667 2.95 2.79–3.12 4644950.1 2,876 3.38 50993.2 127 3.77 3.13–4.55 51137.9 128 4.99 4.13–6.03
3.16–3.60
20 cig. 2229583.6 2,464 4.64 4.37–4.94 2235029.0 1,797 5.14 12608.4 30 3.99 2.78–5.74 12669.1 26 4.45 3.01–6.59
4.78–5.53
1
Results were adjusted for age, age squared, alcohol drinking, socio-economic status and other morbidities as well.
Abbreviations: TB: tuberculosis; HR: Hazard Ratio

Cancer Epidemiology
 2452
Cancer Epidemiology
Figure 1. Combined effects of preexisting TB with smoking status and amount on lung cancer incidence and mortality, Korean Cancer Pre-
vention Study 1998–2013. (a) Incidence, Men. (b) Incidence, Women. (c) Mortality, Men. (d) Mortality, Women. *Incident case was defined
by at least two or more hospitalisations for lung cancer during the follow-up period.
smoked per day, except among female heavy smokers. The
hazard ratios for lung cancer incidence and mortality, with
never-smokers as the reference category, increased as the
amount smoked increased in both sexes. The full results for
all variables of this multivariate model are provided in Sup-
porting Information Table S1, and the results of the basic
model without any covariates other than age, are presented
in Supporting InformationTable S2.
We further explored the combined effects of preexisting
TB and smoking status or amount of smoking on risk of
lung cancer incidence and mortality by gender. Table 3 and
Supporting Information Table S3 provide the multivariate-
and age-adjusted results, respectively, in strata defined by
presence of preexisting TB and smoking status. Independent
associations of both risk factors with lung cancer risk are evi-
dent in men and women. However, there was no indication
of statistically significant effect modification, confirming the
pattern apparent on inspection of the table, that is, no indica-
tion of synergism. Figure 1 (also Supporting Information
Table S4) adds additional and similar findings regarding
amount smoked per day.
All results from the sensitivity analysis, which was con-
ducted by excluding participants whose deaths occurred
Tuberculosis and lung cancer
during the first two years after enrollment (Supporting Infor-
mation Tables S5-S7) and using other categories to define
preexisting TB (Supporting Information Tables S8-S10 and
S11-S13), were generally similar to those of original analyses
and are provided in the Supporting Information Appendix.
The proportions of lung cancer incidence and mortality
attributable to prior TB are provided in Table 4. The esti-
mates range from several percent to approximately 6% with
similar findings for incidence and mortality.
Discussion
In this large prospective cohort study of 1.6 million Korean
men and women, we found that preexisting TB and smoking
status or amount were independent risk factors for lung can-
cer incidence and mortality in both men and women. With
regard to the combined effect, we found no evidence of syn-
ergism on the multiplicative scale. Importantly, we show that
TB is associated with lung cancer risk in a high-income
country. The magnitude of the increment in risk, about 40%
in males and 50% in females, is quite consistent with the
findings of the recent meta-analyses.7,8
We interpret the association of TB with lung cancer as
causal. It has been documented consistently and is robust to
Int. J. Cancer: 139, 2447–2455 (2016) V
C 2016 UICC
Table 3. Combined effects of preexisting TB and smoking status on lung cancer incidence and mortality, Korean Cancer Prevention Study 1998–2013
Men Women
Incidence Mortality Incidence Mortality
Person- Person- Person- Person-
years No. HR1 95% CI years No. HR1 95% CI years No. HR1 95% CI years No. HR1 95% CI
Preexisting Smoking
TB status
No Never-smoker 4491971.6 1,639 1.00 Ref 4496105.2 1,171 1.00 Ref 7881851.1 1,969 1.00 Ref 7887604.1 1,263 1.00 Ref
No Ex-smoker 1873547.0 1,128 1.39 1.29–1.50 1876617.2 816 1.41 1.28–1.54 80064.8 65 1.95 1.51–2.51 80190.0 45 1.82 1.34–2.46
No Current smoker 7806845.7 6,209 3.08 2.91–3.25 7819931.1 4,710 3.45 3.23–3.68 142452.6 236 2.66 2.31–3.07 142831.8 242 3.56 3.08–4.12
Yes Never-smoker 275298.2 264 1.44 1.26–1.64 275943.9 214 1.51 1.30–1.75 266094.8 137 1.43 1.20–1.70 266668.3 98 1.47 1.20–1.81
Yes Ex-smoker 138274.3 234 2.19 1.91–2.51 138702.9 190 2.32 1.99–2.70 4035.0 6 2.40 1.07–5.34 4050.9 5 2.55 1.06–6.14
Yes Current smoker 411829.7 900 4.01 3.70–4.36 413613.9 778 4.72 4.31–5.17 7469.3 32 5.34 3.71–7.68 7506.9 30 6.36 4.83–9.23
AIC for model without 262801.9 194560.9 60862.5 40690.0
interaction terms
AIC for model with all 262799.7 194560.1 60863.8 40693.2
interaction terms
p for interaction2 0.3329 0.6703 N/A3 N/A3
1
Results were adjusted for age and age squared, alcohol drinking, socio-economic status, and other morbidities as well.
2
Test of difference for Akaike information criterion (AIC) with one degree of freedom between the models without and with interaction terms, by preexisting TB and smoking status variable.
3
N/A means goodness-of-fit was better (i.e., lower AIC) in basic model without interaction, than in joint model considering interaction.
Abbreviations: TB: tuberculosis; HR: Hazard Ratio

Cancer Epidemiology
 2454
Table 4. Lung cancer incidence and mortality attributable to TB
Men Women
Incidence Mortality Incidence Mortality
(%) (%) (%) (%)
Never smokers 2.9 3.2 1.4 1.6
Ex-smokers 3.9 4.4 1.1 1.9
Current smokers 1.5 1.8 6.0 4.9
control for potential confounding, as found in this study (Table
2). As a cause of TB, smoking is a potential confounding factor
of specific concern.32 We controlled for potential confounding
by smoking with a multivariate model that included adjustment
for cigarette smoking status and amount smoked for current
Cancer Epidemiology
smokers. Additionally, the cohort was sufficiently large to show
increased risk associated with TB in never smokers (Table 3).
With a number of studies have shown significant associations
between previous TB history and lung cancer risk in never
smokers, the 2011 meta-analysis documented that the overall
RR of TB among never smokers was slightly higher than that
of the entire population.7 Another pooled analysis from large
international consortium, once described above, found higher
risk ratios of TB as a risk factor for lung cancer in never smok-
ers, compared to former and current smokers. The authors also
specified that even after adjusting for secondhand smoke in
never smokers, as a potential confounder, the results remained
with slightly changed risk estimates.8
There are several potential mechanisms by which prior
TB and chronic lung disease generally might increase risk
for lung cancer. One postulated general mechanism by
which previous lung disease may increase lung cancer risk
is local chronic inflammation and oxidative stress.1,33–37
The presence of fibrosis may be key in maintaining inflam-
mation. The presence of fibrosis has been proposed as a
mechanism underlying the association of COPD with lung
cancer, also well documented.1 Additionally, the increased
lung cancer risk associated with asbestos may be in part
mediated through the causation of pulmonary fibrosis;
whether the presence of asbestosis, the specific pneumoco-
niosis caused by asbestos, is requisite for lung cancer risk
to be increased in asbestos-exposed workers has long been
controversial.1 Finally, immune mechanisms may also be
relevant. Prior TB might indicate lesser capability for
immunosurveillance and hence increased risk for lung can-
cer as a result.38
Not surprisingly, we again documented the increased risk
associated with smoking in this cohort.39 The RRs that we
observed in this lengthier follow-up were somewhat lower
than those in follow-up of the original KCPS cohort from
1992 to 2001. The RRs associated with cigarette smoking in
Korea are still well below those observed in the most recent
findings from cohort studies in the United States.32 We have
previously explained the comparatively low risk associated
with smoking as reflecting historical patterns of smoking,
Tuberculosis and lung cancer
particularly the relatively low number of cigarettes smoked
daily in past decades.39 We still observed far lower risk in
comparison with contemporary US cohorts; perhaps because
more recent cohorts of Korean males have smoked less heavi-
ly than previous cohorts.39–41
We explored potential synergism between TB and ciga-
rette smoking in the multivariate models. We found no evi-
dence for statistical interaction on the multiplicative scale.
There are several mechanisms by which TB might increase
the risk of lung cancer in smokers: impairing local clearance
of tobacco smoke components, particularly particulate matter
and creating susceptible areas in the epithelium through
inflammation. Given the high prevalence of TB worldwide, if
there were synergism with smoking, TB would be an even
more important contributor to lung cancer burden. We did
not find evidence for synergism, similar to other studies,
which have also not found strong indication of effect modifi-
cation between TB history and smoking status or amount
smoked.6,19,42,43
In spite of the large sample size, there are potential limita-
tions of this study, largely related to potential misclassifica-
tion in the baseline data. We relied on self-report and a
single chest X-ray to establish the presence of prior TB. To
assess misclassification arising from this definition of expo-
sure, we tested alternative classifications for TB status and
found that alternative definitions gave similar results. Infor-
mation on smoking status and amount, also based on self-
report, is also subject to potential misclassification. Previous-
ly, this information in the NHIS data base was shown to
have satisfactory validity based on comparison with cotinine
measurements.29,44 With regard to the outcome, incident
lung cancer or mortality from lung cancer, we anticipate little
misclassification. Incident lung cancer was defined by at least
two hospitalisations for this malignancy, which should
exclude hospitalisations for diagnostic purposes only. With
regard to the accuracy of death certificate diagnoses of lung
cancer in Korea, a previous report found relatively high accu-
racy for death certificate diagnosis a sensitivity of 91.7% and
a confirmation rate of 79.7% compared to cancer registration
data, including primary site of cancer.45 A more recent study
suggested improving accuracy of cause of death informa-
tion.46 Because of a lack of information, we did not explore
associations within strata of histological type of lung cancer.
In conclusion, we have shown that prior TB increases
lung cancer risk in both smokers and never smokers. The ele-
vation in risk is relatively modest, particularly in comparison
to that from smoking, and a prior history of TB is not likely
to be useful risk indicator for clinical purposes. In popula-
tions with high prevalence of TB, it should be considered for
incorporation into models for risk prediction.
Author Contributions
Dr. Jee had full access to all of the data in the study, and he
takes responsibility for the integrity of the data and the accu-
racy of the data analysis. Study concept and design: Jee,
Int. J. Cancer: 139, 2447–2455 (2016) V
C 2016 UICC
Hong et al.
Hong, Samet, Data acquisition: Jee, Data analysis and inter-
pretation of data: Jee, Hong, Samet, Drafting the manuscript:
Hong, Jee, Samet, Critical evaluation of the manuscript for
References
1. Alberg AJ, Brock MV, Ford JG, et al. Epidemiolo-
gy of lung cancer: diagnosis and management of
lung cancer, 3rd ed: American College of Chest
Physicians evidence-based clinical practice guide-
lines. Chest 2013; 143:e1S–29S.
2. Khuder SA. Effect of cigarette smoking on major
histological types of lung cancer: a meta-analysis.
Lung Cancer 2001; 31:139–48.
3. Samet JM, Humble CG, Pathak DR. Personal and
family history of respiratory disease and lung
cancer risk. Am Rev Respir Dis 1986; 134:466–70.
4. Mayne ST, Buenconsejo J, Janerich DT. Previous
lung disease and risk of lung cancer among men
and women nonsmokers. Am J Epidemiol 1999;
149:13–20.
5. Brownson RC, Alavanja MC. Previous lung dis-
ease and lung cancer risk among women (United
States). Cancer Causes Control 2000; 11:853–8.
6. Brenner AV, Wang Z, Kleinerman RA, et al. Pre-
vious pulmonary diseases and risk of lung cancer
in Gansu Province, China. Int J Epidemiol 2001;
30:118–24.
7. Brenner DR, McLaughlin JR, Hung RJ. Previous
lung diseases and lung cancer risk: a systematic
review and meta-analysis. PLoS One 2011; 6:
e17479
8. Brenner DR, Boffetta P, Duell EJ, et al. Previous
lung diseases and lung cancer risk: a pooled anal-
ysis from the International Lung Cancer Consor-
tium. Am J Epidemiol 2012; 176:573–85.
9. Alavanja MC, Brownson RC, Boice JD Jr, et al.
Preexisting lung disease and lung cancer among
nonsmoking women. Am J Epidemiol 1992; 136:
623–32.
10. Zheng W, Blot WJ, Liao ML, et al. Lung cancer
and prior tuberculosis infection in Shanghai. Br J
Cancer 1987; 56:501–4.
11. Gao YT, Blot WJ, Zheng W, et al. Lung cancer
among Chinese women. Int J Cancer 1987; 40:
604–09.
12. Steinitz R. Pulmonary tuberculosis and carcinoma
of the lung. A survey from two population-based
disease registers. Am Rev Respir Dis 1965; 92:758–
66.
13. Campbell AH, Guilfoyle P. Pulmonary tuberculo-
sis, isoniazid and cancer. Br J Dis Chest 1970; 64:
141–9.
14. Shieh SH, Probst JC, Sung FC, et al. Decreased
survival among lung cancer patients with co-
morbid tuberculosis and diabetes. BMC Cancer
2012; 12:174
15. Liang HY, Li XL, Yu XS, et al. Facts and fiction
of the relationship between preexisting tuberculo-
sis and lung cancer risk: a systematic review. Int J
Cancer 2009; 125:2936–44.
16. Yu YH, Liao CC, Hsu WH, et al. Increased lung
cancer risk among patients with pulmonary
tuberculosis: a population cohort study. J Thorac
Oncol 2011; 6:32–7.
17. Wu CY, Hu HY, Pu CY, et al. Pulmonary tuber-
culosis increases the risk of lung cancer. Cancer
2011; 117:618–24.
Int. J. Cancer: 139, 2447–2455 (2016) V
C 2016 UICC
important intellectual content: Samet, Statistical analysis: Jee,
Hong, Mok, Jung, Obtaining funding: Jee, Study supervision:
Samet.
18. Engels EA, Shen M, Chapman RS, et al. Tubercu-
losis and subsequent risk of lung cancer in Xuan-
wei, China. Int J Cancer 2009; 124:1183–87.
19. Shiels MS, Albanes D, Virtamo J, et al. Increased
risk of lung cancer in men with tuberculosis in
the alpha-tocopherol, beta-carotene cancer pre-
vention study. Cancer Epidemiol Biomarkers Prev
2011; 20:672–8.
20. Leung CC, Hui L, Lee RS, et al. Tuberculosis is
associated with increased lung cancer mortality.
Int J Tuberc Lung Dis 2013; 17:687–92.
21. Heuvers ME, Aerts JG, Hegmans JP, et al. Histo-
ry of tuberculosis as an independent prognostic
factor for lung cancer survival. Lung Cancer 2012;
76:452–6.
22. Korea Centers for Disease Control and Prevention.
Annual report on the notified tuberculosis cases
patients in Korea 2013. Sejong: Korea Centers for
Disease Control and Prevention, 2014.
23. World Health Organization. Global Tuberculosis
Report. Geneva: World Health Organization, 2014.
24. World Health Organization. WHO TB burden
estimates. In: Global Tuberculosis Report,
Geneva: WHO, 2014. Available at: http://www.
who.int/tb/country/data/download/en/ (last access
on 30 Jun 2016)
25. Kang YA, Lee HW, Yoon HI, et al. Discrepancy
between the tuberculin skin test and the whole-
blood interferon gamma assay for the diagnosis
of latent tuberculosis infection in an intermediate
tuberculosis-burden country. JAMA 2005; 293:
2756–61.
26. U.S. Department of Health and Human Services.
Latent tuberculosis infection: a guide for primary
health care providers. Atlanta, GA: Centers for
Disease Control Prevention, 2013
27. Korea Centers for Disease Control and Prevention.
Summary Results of the Korea Health and Nutrition
Examination Survey (KHANES) 2014. Sejong:
Division of Health and Nutrition Survey, Center for
Disease Prevention, Korea Centers for Disease
Control and Prevention, 2015.
28. Jee SH, Ohrr H, Sull JW, et al. Fasting serum glu-
cose level and cancer risk in Korean men and
women. JAMA 2005; 293:194–202.
29. Jee SH, Golub JE, Jo J, et al. Smoking and risk of
tuberculosis incidence, mortality, and recurrence
in South Korean men and women. Am J Epide-
miol 2009; 170:1478–85.
30. Jee SH, Kivimaki M, Kang HC, et al. Cardiovas-
cular disease risk factors in relation to suicide
mortality in Asia: prospective cohort study of
over one million Korean men and women. Eur
Heart J 2011; 32:2773–80.
31. Gordis L. Epidemiology, 5th edn., Philadelphia,
PA: Saunders, 2013.
32. U.S. Department of Health and Human Services.
The Health Consequences of Smoking—50 Years
of Progress. A Report of the Surgeon General.
Atlanta, GA: Centers for Disease Control and
Prevention, National Center for Chronic Disease
Prevention and Health, Office on Smoking and
Health, 2014.
2455
33. O’Callaghan DS, O’Donnell D, O’Connell F, et al.
The role of inflammation in the pathogenesis of
non-small cell lung cancer. J Thorac Oncol 2010;
5:2024–36.
34. Yang SR, Chida AS, Bauter MR, et al. Cigarette
smoke induces proinflammatory cytokine release
by activation of NF-kappaB and posttranslational
modifications of histone deacetylase in macro-
phages. Am J Physiol Lung Cell Mol Physiol 2006;
291:L46–57.
35. Chang SH, Mirabolfathinejad SG, Katta H, et al.
T helper 17 cells play a critical pathogenic role in
Cancer Epidemiology
lung cancer. Proc Natl Acad Sci USA 2014; 111:
5664–9.
36. Sato T, Arai E, Kohno T, et al. Epigenetic cluster-
ing of lung adenocarcinomas based on DNA
methylation profiles in adjacent lung tissue: its
correlation with smoking history and chronic
obstructive pulmonary disease. Int J Cancer 2014;
135:319–34.
37. Chaturvedi AK, Caporaso NE, Katki HA, et al.
C-reactive protein and risk of lung cancer. J Clin
Oncol 2010; 28:2719–26.
38. Ichim CV. Revisiting immunosurveillance and
immunostimulation: implications for cancer
immunotherapy. J Transl Med 2005; 3:8.
39. Jee SH, Samet JM, Ohrr H, et al. Smoking and
cancer risk in Korean men and women. Cancer
Causes Control 2004; 15:341–8.
40. Yun YD, Back JH, Ghang H, et al. Hazard ratio of
smoking on lung cancer in Korea according to his-
tological type and gender. Lung 2016; 194:281–9.
41. Korea Centers for Disease Control and
Prevention. Trend of the number of cigaretts
daily smoked in current smokers, 1998-2014. In:
Korea Health and Nutrition Examination Survey
(KHANES), Sejong: Korea Centers for Disease
Control and Prevention, 2014. Available at:
http://kosis.kr/statHtml/statHtml.do?or-
gId5117&tblId5DT_11702_N003&conn_
path5I3 (last access on 30 Jun 2016)
42. Wang SY, Hu YL, Wu YL, et al. A comparative
study of the risk factors for lung cancer in
Guangdong, China. Lung Cancer 1996; 14:S99–
105.
43. Cocco P, Rice CH, Chen JQ, et al. Non-malignant
respiratory diseases and lung cancer among Chi-
nese workers exposed to silica. J Occup Environ
Med 2000; 42:639–44.
44. Park SS, Lee JY, Cho SI. [Validity of expired car-
bon monoxide and urine cotinine using dipstick
method to assess smoking status]. J Prev Med
Public Health 2007; 40:297–304. (in Korean)
45. Lee D-H, Shin H-R, Ahn D-H, et al. [Accuracy
of Cancer Death Certificates in Korea: a compari-
son between diagnoses in the central cancer regis-
try and certified underlying causes of death].
Cancer Research and Treatment 1999; 31:210–19.
(in Korean)
46. Jo MW, Khang YH, Yun S, et al. [Proportion of
death certificates issued by physicians and associ-
ated factors in Korea, 1990-2002]. J Prev Med
Public Health 2004; 37:345–52. (in Korean)