Increased Lung Cancer Risk Among Patients with
Pneumococcal
Pneumonia: A Nationwide Population-Based Cohort
Study
Te-Yu Lin • Wen-Yen Huang • Jung-Chung Lin •
Cheng-Li Lin • Fung-Chang Sung • Chia-Hung Kao •
Jun-Jun Yeh
Introduction
Pneumonia is a common global disease and remains one of the major causes of death among infectious diseases [1]. Among community-acquired pneumonia conditions, approximately 50 % of causative organisms can be found. Among them, Streptococcus pneumonia is a common
pathogen [2]. Lung cancer continues to be the leading cause of cancer incidence and mortality worldwide, with more than a million estimated deaths annually [3]. Traditional risk factors of lung cancer are smoking, occupational
exposure to arsenates, nitrosamines, and asbestos, or indoor exposure to radon or fumes from fires. Recent evidence suggests that inflammatory processes may play a central role in carcinogenesis [4, 5]. The major sources of inflammation in lung tissue are chronic obstructive pulmonary
disease (including chronic bronchitis and emphysema), pneumonia, and tuberculosis. Both pneumonia and lung cancer substantially influence public health; therefore, the association between these two diseases must be clarified. Several studies have examined the association between
pneumonia and lung cancer using hospital and communitybased populations [6–18]. A meta-analysis performed by
Brenner et al. [19] revealed a positive association between pneumonia and subsequent lung cancer. However, the cited references were mostly case–control studies. Recall bias of
exposure assessment, selection bias of the control group, and the validity of the reported pneumonia diagnosis could be considered possible limitations of this type of study. In addition, the temporal relationship between pneumonia and subsequent lung cancer is difficult to assess in case–control studies. Some cohort studies have reported a positive association between pneumonia and lung cancer [20]. However, these studies might not have been able to yield generalizable results, because they have usually focused on
high-risk populations, such as heavy smokers and asbestosexposed smokers.
Pneumonia can be caused by virus, bacteria, fungi, and parasites, and the relationship between the causative organisms of pneumonia and lung cancer is less clearly
understood. Two published reports have shown that Chlamydia pneumonia infection is associated with an increased
risk of lung cancer [20, 21]. However, no publications, to the best of our knowledge, focus on the association between pneumococcal pneumonia and lung cancer. Therefore, we conducted a nationwide population-based cohort study that included numerous pneumococcal pneumonia cases to investigate the association between pneumococcal pneumonia and subsequent risk of lung cancer, as well as the temporal relationship between these two diseases.
Materials and Methods Data Source
We established a longitudinal cohort study from the Taiwan National Health Institute Research Database
(NHIRD), with whole population admission insured in Taiwan. Taiwan has implemented a compulsory singlepayer National Health Institute (NHI) program since the
beginning of 1995, covering more than 99 % of Taiwan’s 23 million residents and prospectively recorded until December 31, 2010. The Bureau of National Health Insurance (BNHI) provided the registration files and medical claims data. All data were confidential and all people were anonymous and de-identified, thus ensuring
that the NHI reimbursement data were suitable for public research. Previous studies have addressed information on the NHIRD [22–25]. The accuracy and high validity of
diagnoses identified in the NHIRD also has been demonstrated [22]. This study was exempted by the Institutional
Review Board of China Medical University (CMU-REC-101-012).
Patients with Pneumococcal Pneumonia
The International Classification of Diseases, Ninth Revision (ICD-9), was used to confirm diagnostic codes. We
selected all adult pneumococcal pneumonia (ICD-9 code 481) patients (C20 years old) from the NHIRD. In total,
22,034 patients who had a first-time diagnosis of pneumococcal pneumonia and at least one health care admission
from 1997 to 2010 were confirmed; the index date was the date of pneumococcal pneumonia registration. Patients with a history of cancer (ICD-9 codes 140–208) before the index date, or with incomplete age or sex information, were excluded. All patients without a history of pneumonia (ICD-9 codes 480–487) were selected randomly in the NHIRD. To increase the statistical power, four comparison controls (88,136) for each case were frequently matched for age (5 years each), sex, and index year. All of the study patients were monitored until the date on which they were diagnosed with lung cancer.
Identification of Lung Cancer Cases
Our newly diagnosed patients with lung cancer (ICD-9 code 162) were confirmed by the Registry for Catastrophic Illness
Patient Database (RCIPD), and the index date was the date of lung cancer registration. To apply for a catastrophic illness
certificate for reimbursement claims, patients with cancer must provide comprehensive supporting information, such as reports or evidence of cytology and pathology. At least two other oncologists carefully examine their medical and laboratory information. After the rigorous assessments are processed, patients who meet the criteria for the diagnoses are issued certificates.
Potential confounding factors were listed according to established risk factors, and analyses were performed to
establish whether these variables were substantially associated with lung cancer. Baseline comorbidities were analyzed
as follows: tuberculosis (ICD-9 code 611), asthma (ICD-9 code 493), and chronic obstructive pulmonary
disease (COPD, ICD-9 codes 490–496), which is characterized by chronic airway inflammatory process, have been
reported as being linked to lung cancers [25–27]. Metabolic syndromes also have been linked to several types of cancer [28, 29]. The relationship between diabetes mellitus (ICD-9 code 250) and lung cancers remains controversial [30,
31]. Hyperlipidemia (ICD-9 code 272) and elevated blood pressure (hypertension, ICD-9 codes 401–405) comprise the major components of metabolic syndromes; therefore, they were included in the comorbidity analysis.
Statistical Analysis
All statistical analyses were performed using the SAS statistical package (Version 9.2 for Windows; SAS Institute, Inc., Cary, NC). Statistical significance was accepted at a = 0.05. The distributions of categorical characteristics between the pneumococcal pneumonia cohort and the controls were described and compared by conducting v2
tests. The person-years were calculated to assess incidence densities (per 1,000 person-years) until lung cancer diagnosis was identified, loss of follow-up, or censored at the
end of 2010. To estimate the age effect, we divided the study patients into categories based on age (20–45, 46–55, 56–65, 66–74, and C75 years) at the index date of pneumococcal pneumonia. To estimate the relative risk (the
incidence density rate) and absolute risk (adjusted hazard ratio, aHR) of lung cancer, we proceeded as follows. First, we used the Poisson regression model to assess incidence rate ratios (IRRs), and 95 % confidence intervals (CIs)
between the pneumococcal pneumonia cohort and the controls. Second, we used the Cox proportional hazard
regression model to estimate the hazard ratios (HRs) of developing lung cancer. Furthermore, we assessed whether
the association of lung cancer varied according to the interval from the index date to the date of lung cancer diagnosis. We divided the time lag into the following two periods: B1 and[1 year. The cumulative incidence of lung cancer between the pneumococcal pneumonia cohort and the controls was assessed using the Kaplan–Meier method, and the differences were assessed with a log-rank test (Fig. 1).
Results
Demographic characteristics are shown in Table 1. No statistical difference was found in the distributions of sex and age between the pneumococcal pneumonia cohort and the controls (mean age 63.6 ± 17.7 years). Nearly twothirds of the male patients and more than half of all of the
patients in this study were older than age 66 years. The prevalence of comorbidities was greater in the pneumococcal pneumonia cohort.
The IRRs (relative risks) and aHRs (absolute risks) between the pneumococcal pneumonia cohort and the controls are shown in Table 2. The overall IRR of lung
cancer was 3.25-fold greater in the pneumococcal pneumonia cohort than in the controls (28.2 vs. 8.7 per 1,000
person-years) with an aHR of 4.24 (95 % CI 3.96–4.55) in the following 14 years. Analyses stratified by gender showed both genders have significant relative risks of developing lung cancer. For females, the incidence densities were 16.0 and 5.0 per 1,000 person-years between the two cohorts, with a 3.21-fold relative risk of developing lung cancer (95 % CI 2.88–3.57). The aHR of males of lung cancer was 85 % significantly higher than females (95 % CI 1.71–1.99). When stratified by age, the incidence density rates of lung cancer increased as age did and were the highest in the oldest patients of both cohorts (45.7 vs. 14.4 per 1,000 person-years). In the age groups, age
20–45 years had a 27.7-fold higher relative risk of developing lung cancer (95 % CI 15.8–48.8). After adjusting for
cofactors, the risk of developing lung cancer increased with age (patients 20–45 years of age were the reference group)
with an aHR of 18.7 in the C75 year age group (95 % CI 15.1–23.1). The aHR was calculated according to the interval of the follow-up period from the index date to the date of lung cancer diagnosis. The incidence density rates of lung cancer decreased with the follow-up periods among the pneumococcal pneumonia cohort. We found an 11.1-fold significantly higher relative risk of developing lung cancer within 1 year follow-up period (95 % CI
10.1–12.1).
Table 3 lists the specific analysis of comorbidities in the IRRs and aHRs between the pneumococcal pneumonia cohort and the controls. Despite excluding all cofactors, the overall IRR of lung cancer was 2.98-fold higher in the
pneumococcal pneumonia cohort than in the controls (95 % CI 2.77–3.21). Stratified by comorbidities, the
pneumococcal pneumonia cohort with hypertension, DM, hyperlipidemia, COPD, TB, and asthma had statistically significantly higher IRR of developing lung cancer than the controls. It showed that the study patients with TB had a 52 % higher risk of lung cancer than did those without (95 % CI 1.32–1.74).
Discussion
This is the first nationwide population-based cohort study to investigate the associations between lung cancer and previous pneumococcal pneumonia. Our study showed that the incidence of lung cancer is greater in patients with previous history of pneumococcal pneumonia than in the general population. Most previous studies were hospitalbased research with limited observation cases. The population-based data with age- and sex-matched control groups enabled better external validation than have previous studies. Moreover, using the NHI data to create a cohort study allowed us to observe the temporal relationship between pneumococcal pneumonia and lung cancer. Pneumococcal pneumonia is diagnosed according to clinical symptoms, physical findings, image findings, and
laboratory results (sputum culture or positive urine pneumococcus antigen test). We identified the pneumococcal
pneumonia cases based on their ICD-9-CM codes and antibiotics prescriptions. Lung cancer diagnosis entitles patients to qualify for special health care privileges in the ‘‘major critical diseases’’ class in Taiwan’s NHI system. We used both diagnostic codes and the Catastrophic Illness Registry. Histological confirmation is needed under the NHI program to apply for a Catastrophic Illness Card; therefore, the pneumococcal pneumonia and lung cancer data are valid.
Reverse causality must be considered in cases of pneumonia, because infections may have been the result of
weakening immunity caused by occult lung cancer. For
these conditions, ascertainment bias should be considered as individuals with pneumonia may have been more likely
to be diagnosed with lung cancer because of the use of additional chest X-rays in the infection workup. In the presence of occult lung cancer or the coexistence of
pneumococcal pneumonia and lung cancer, the risk of lung cancer during the first year of pneumococcal pneumonia would be magnified. Therefore, lung cancer diagnosis before or within the first year of pneumococcal pneumonia infection was excluded to lessen the possibility of reverse causality and coexistence. After excluding lung cancer occurring during the first year of pneumococcal pneumonia, we still found an increased IRR for lung cancer in the
pneumococcal pneumonia cohort.
We observed elevated rates of lung cancer among previous pneumococcal pneumonia cases compared with
controls, with crude IRR of 1.24 (95 % CI 1.15–1.35;
p\0.001). These results were consistent with many previous studies, which have reported a positive association
between pneumonia and subsequent lung cancer [6–18]. However, the relationship between pneumonia’s causative organisms and lung cancer has been less described. Most of the previous reports revealed a positive association
between Chlamydia pneumonia infection and subsequent lung cancer. Streptococcus pneumonia is an important bacterial pathogen of pneumonia; to our knowledge, this
study is the first to evaluate the relationship between pneumococcal pneumonia and lung cancer.
We observed a higher lung cancer incidence among males in our study (HR, 1.85; 95 % CI 1.71–1.99; p\0.001). In a cross-sectional survey by Tsai et al. in 2004, they found a difference in smoking behavior among males and females (45.7 vs. 4.8 %), observing that men smoked significantly more cigarettes per day (18 vs. 11) [32]. Smoking is an important risk factor for lung cancer and could account for the gender discrepancy of lung cancer incidence in our study.
The risk of lung cancer is increased in the pneumococcal pneumonia cohort, especially in the older than age 75 years group (HR, 18.7; 95 % CI 15.1–23.1; p\0.001). Epidemiology data showed that the age-based risk group of
pneumococcal pneumonia is younger than 2 years or older than 75 years. Our data were consistent with previous
reports and demonstrated the important role of pneumococcal pneumonia in subsequent lung cancer, especially in
the group older than 75 years.
Pulmonary infections have been suggested as risk factors for lung cancer. Respiratory tract infections may
contribute to lung cancer by promoting airway remodeling and causing inflammation, which could generate reactive oxygen or nitrogen species, increase cellular proliferation, upregulate antiapoptotic pathways, and stimulate angiogenesis [5, 33]. Our data showed increased lung cancer
incidence in the pneumococcal pneumonia cohort. Further studies should incorporate research that is more experimental and epidemiological to investigate the mechanisms
of lung cancer induced by Streptococcus pneumonia infection.
There are potential limitations in this study. First, the NHIRD does not contain detailed information regarding smoking habits, alcohol consumption, socioeconomic status, and family history of systemic diseases, all of which
may be risk factors for pneumococcal pneumonia or lung cancer. Second, the evidence derived from a retrospective
cohort study is generally lower in statistical quality than that from randomized trials because of potential biases related to adjustments for confounding variables. Despite our meticulous study design and control measures for
confounding factors, bias resulting from unknown confounders may have affected the results. Third, all data in
the NHIRD are anonymous. Thus, relevant clinical variables, such as blood pressure, imaging results, pathology
findings, and serum laboratory data, were unavailable regarding our study patient cases. However, the data regarding the diagnoses of pneumococcal pneumonia or lung cancer were, nonetheless, reliable.
The major strength of our study is its nationwide population-based settings that included relatively large numbers
of pneumococcal pneumonia cases, thus allowing a clear observation of the temporal relationship between
pneumonia and lung cancer. Our findings provide epidemiological evidence that pneumococcal pneumonia is a
risk factor for subsequent lung cancer and that pneumococcal pneumonia patients should be monitored carefully
