Effect of Emphysema on Lung Cancer Risk in Smokers: A Computed Tomography–Based Assessment

Pulmonary emphysema is a pathologic lesion, defined as abnormal permanent enlargement of the airspaces distal to the terminal bronchioles accompanied by destruction of their walls without obvious fibrosis (1). Along with chronic bronchitis, emphysema has been recognized as one of the two main forms of chronic obstructive pulmonary disease (COPD; ref. 2). More often in clinical practice, emphysema has been lumped into the COPD category and the unique effect of parenchymal destruction caused by emphysema was not adequately emphasized. Many studies have shown that emphysema could be an independent risk factor for lung cancer (3–10); however, in most of these studies, the assessment of emphysema was based on self-reported clinical diagnosis (3, 8–10). The recall bias of such self-reported diagnosis has been a major concern and well recognized (8); however, the accuracy of clinical diagnosis for the anatomic emphysema remains unknown. Misclassification between clinically diagnosed emphysema and chronic bronchitis exists, because the distinction between the two diseases in clinical practice is difficult (11, 12). Emphysema can be present without detectable airflow obstruction, particularly in the early stages; therefore, people with emphysema may not be diagnosed until the disease is advanced; specific emphysema diagnosis might be missed or left off unless it was considered “clinically significant.” A more accurate assessment of emphysematous change in the lung and its impact on lung cancer risk is needed, especially among smokers who are more likely suffering from both diseases (13).

The use of an adequate resolution computed tomography (CT) scan of the lung has been advocated as the Gold standard for noninvasive detection of emphysema (14). CT can provide excellent anatomic detail to quantitatively detect and characterize the presence and the severity of emphysema. By using CT, lung cancer risk due to lung tissue damage caused by emphysema can be more accurately determined (15). Two recent lung cancer screening studies using low-dose spiral CT have shown that emphysema on CT scan was an independent risk factor for lung cancer (16, 17). However, in our previous matched case–control study, no significant association between overall percentage of emphysema as determined by CT and lung cancer risk was found (18). To resolve the discrepancy and further assess the association between CT-diagnosed emphysema and lung cancer risk, we conducted this clinic-based case–control study. Complete CT scan data and clinical information were available for all enrolled subjects, allowing us to evaluate how well the clinical diagnosis for emphysema conforms to the CT diagnosis and to assess the accurate contribution of emphysema on lung cancer risk.

Consecutive primary lung cancer cases were recruited from the Epidemiology and Genetics of Lung Cancer Study conducted at Mayo Clinic between 1997 and 2004 (19, 20). Controls were chosen from a screening study for lung cancer with a spiral CT conducted at Mayo Clinic within the same time period between 1999 and 2004 (21). The following eligibility criteria were maintained for this study. (a) Standard-dose CT scan film at the time of lung cancer diagnosis (case) or during the study (control) was available. Individuals with only low-dose CT scan films were excluded to eliminate the evaluation bias caused by different CT settings. (b) Study subjects were cigarette smokers who had smoked at least 20 pack-years. Controls were frequency matched with cases on age (±5 years), gender, race, smoking status, and residential area. A total of 565 cases and 450 controls were identified in this study. The study protocol was approved by the Mayo Clinic Institutional Review Board, and written informed consent was obtained from each subject.

Data on demographic characteristics, personal smoking history, family history of lung cancer (in first degree relatives), detailed lung disease history (including emphysema, chronic bronchitis, and unspecified COPD), and other medical information were carefully collected for all subjects via a combination of a structured subject interview, self-administered questionnaire, and medical records (22). Former smokers were defined as having quit smoking for 6 months or more before lung cancer diagnosis for cases or the date of answering a study questionnaire (controls). All the clinical diagnoses of emphysema, chronic bronchitis, and unspecified COPD for each subject were determined on the basis of the explicit diagnoses recorded in the medical documentation from the pulmonologists. In addition, an answer to the clinical diagnosis of chronic bronchitis from individuals who reported a physician-diagnosed chronic bronchitis was explicitly required, with the specific question, “Have you had an episode of cough and sputum production lasting 3 or more months for 2 consecutive years?”

Radiological diagnosis and quantification of emphysema was made through direct interpretation and evaluation of the CT examinations (all via volumetric technology) by an experienced thoracic radiologist (S.J.S.; refs. 18, 21, 23, 24), who was blinded to case–control identity and clinical diagnosis of emphysema. A detailed visual analysis of the entire lung examination with an estimate of the percentage of lung tissue destroyed by emphysema was done. The analyses were completed on digitally archived chest CT examinations using a standard protocol (25, 26). Grading of emphysema was done by comparing research subject CT examinations to computer-aided quantitative standard images generated by using −950 HU as the threshold for emphysema (18, 27, 28). Emphysema was analyzed as a categorical variable by using the following categories: 0%, >0% and <5%, ≥5% and <10%, and ≥10%. Supplementary Figure S1 illustrates the differential diagnosis of between emphysema and non-emphysema and between 5% and 10% emphysema on CT films. Quality control was assured by repeating calls of a random, blind sample. Quality control procedures were implemented to ensure accuracy and completeness of all data collected.

Pearson's χ² test was used to test the differences between cases and controls for categorical variables. Student's t test was used to test the difference for continuous variables. Wilcoxon–Mann–Whitney test was used to test the difference in median pack-years of cigarette smoking. The effects of the duration (years smoked) and intensity (cigarettes per day) of smoking history and years since quitting among former smokers were also assessed. The performance of clinical diagnosis for emphysema was evaluated using CT diagnosis as the Gold standard. Three different criteria were used to determine the existence of emphysema by CT: >0%, ≥5%, and ≥10%; if the percentage of lung destroyed by emphysema met the criteria, emphysema diagnosis of the subject was determined as “Yes,” otherwise “No.” Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy of clinical diagnosis for emphysema were calculated.

To evaluate the independent role of emphysema on lung cancer risk, an unconditional multivariable logistic regression analysis was done to control for confounding variables through backward selection. All statistical analyses were done using SAS, version 9.2 (SAS Institute).

There was no significant difference between the cases and the controls in terms of age, gender, race, smoking status, and unspecified COPD (Table 1). Caucasians represented 94.5% of the cases and 96.9% of the controls. Pack-years, years smoked, and cigarettes per day were significantly higher in the cases than in the controls. A significantly higher percentage of the cases had a family history of lung cancer compared with the controls. A significant difference was observed between the cases and the controls on chronic bronchitis. CT-diagnosed emphysema was significantly higher in the cases than in the controls regardless of the criterion (P < 0.0001).

Of the 565 lung cancer cases, 71 (12.57%) were small-cell lung cancer (SCLC) and 494 (87.43%) were non–small-cell lung cancer (NSCLC). Stage information is presented in Table 2.

Performance of clinical diagnosis for emphysema compared with CT diagnosis was assessed in both the cases and the controls, by using 3 different CT criteria as defined in the Methods section (Fig. 1A–F). In 565 lung cancer cases, if using the >0% criterion, 77% of the cases had emphysema by CT diagnosis; however, clinical diagnosis only detected 36% (Fig. 1A). If using the ≥5% criterion, 43% of the cases had real emphysema, whereas clinical diagnosis detected 27% (Fig. 1B). If using the ≥10% criterion, 28% of the cases had emphysema, and clinical diagnosis detected 20% (Fig. 1C). In 450 controls, there were much more striking differences in emphysema diagnosis between CT and clinical assessment. Using the criterion >0%, ≥5%, or ≥10%, only 7% (4.4% versus 60%), 20% (4% versus 20%), or 18% (2% versus 11%) of the CT-diagnosed emphysema could be clinically diagnosed, respectively (Fig. 1D–F). As Figure 2 shows, the sensitivity of the clinical diagnosis was much lower in the controls than in the cases: nearly 80% of healthy smokers with emphysematous changes in the lung were underdiagnosed. Overall, the diagnostic performance of clinical assessment compared with CT diagnosis was low (Supplementary Table S1).

Due to the very small sample size of non-Caucasians in this study, multivariable analyses were applied to the Caucasian population only to estimate the independent role of emphysema on lung cancer risk.

There was a significant association between emphysema and lung cancer risk (Table 3). Among subjects with any evidence of emphysema on CT, lung cancer risk increased 2.79-fold (95% CI: 2.05–3.81; P < 0.0001). Lung cancer risk increased up to 3.80-fold if emphysema percentage on CT was 5% or more (95% CI: 2.78–5.19; P<0.0001), but did not increase further when emphysema percentage on CT became 10% or more (OR = 3.33; 95% CI: 2.30–4.82; P < 0.0001).

Stratified analyses by age, pack-years, cell type, and stage were also conducted by using best comprised cutoff at 5% or more as the CT diagnosis criterion for emphysema. As Figure 3 and Supplementary Table S2 show, among heavy smokers (cigarette smoking ≥40 pack-years), lung cancer risk contributed by emphysema was much higher than in moderate smokers (20–40 pack-years; OR: 4.46 versus 2.84). Among the older age group (≥65 years old), the emphysema-associated lung cancer risk was estimated at 3.47-fold; whereas among the younger age group (<65 years old), the emphysema-associated lung cancer risk was 4.62-fold. Regarding cell type, having emphysema increased the risk of all 3 major cell types of lung cancer (adenocarcinoma, squamous cell carcinoma, and SCLC) significantly, with the strongest effect on SCLC (OR = 5.62; 95% CI: 3.13–10.09; P < 0.0001).

Finally, smoking cessation did not change the significant effect of emphysema on lung cancer risk, via stratified analysis of current smokers versus former smokers, with the adjustment of years smoked, cigarettes per day, and quit years for former smokers (data not shown).

The primary purpose of this study was to accurately assess the effect of emphysema on lung cancer risk. Emphysema predicted a 3- to 4-fold increased lung cancer risk, with higher contributions in the younger age group, heavy smokers, and SCLC. Our results also showed that clinical assessment has a fairly low performance on detecting anatomic emphysema compared with that characterized by CT.

The relationship between emphysema and lung cancer has been reported in numerous case–control studies. However, most studies that attempted to estimate the burden of emphysema did not have an accurate diagnosis, but only estimated the “physician-diagnosed” emphysema (3–10). Recall bias and physiologic inaccuracy could cause significant underdiagnosis. Emphysema is an indolent process that can remain subclinical. Comparison between CT analysis and pulmonary function test results has revealed that up to one third of the lung can be destroyed by emphysema before lung function becomes impaired (29, 30). In our study, we included all the moderate and heavy smokers whose chest CT films at the time of lung cancer diagnosis (cases) or during the study (controls) were available. We evaluated the emphysema presence for each subject by direct CT analysis. The results showed that, compared with CT diagnosis, nearly 80% of emphysema cases by CT in healthy smokers were not clinically diagnosed. CT should be used to aid a more accurate diagnosis for emphysema.

The strong relationship between emphysema by any CT criterion and lung cancer risk suggests that the presence of emphysema on CT scan is an independent predictor of lung cancer, regardless of disease severity, which is consistent with the findings of recent lung cancer screening studies by CT (16, 17). The controversy caused by a previous matched case–control study conducted in our institution could be explained by 3 pitfalls of that study (18): (a) the small sample size (24 cases versus 96 controls), which may not be able to detect the statistical significance; (b) an automated scoring system using 3-dimensional volume rendering was used in the previous study, which could misclassify other air-space lesions, such as usual interstitial pneumonia or pneumonitis (UIP) as emphysema (Supplementary Fig. S1C); (c) greater than 0% was used as the cutoff to define “emphysema,” which could include unmeaningful changes. In fact, if 10% or higher was used as the cutoff, there would have been 58.3% in cases versus 43.8% in controls to be defined as emphysema, a rather different result (18).

Of particular importance from this study, emphysema on CT was a risk factor for lung cancer even after adjusting for smoking status and pack-years of cigarette smoking. In addition, smoking cessation did not change the significant effect of emphysema on lung cancer risk. It suggests that emphysema as a lung cancer risk factor may not entirely depend on smoking.

The association between emphysema and lung cancer was stronger among heavy smokers; on the basis of CT-diagnosis, heavy smokers with emphysema would possess even higher lung cancer risks compared with individuals without emphysema. The same interpretation holds for our finding that emphysema was more strongly associated with lung cancer among the younger age group. Emphysema was associated with a greater risk of SCLC and squamous cell carcinoma, supporting emphysema imposing a higher risk of developing a specific histologic subtype of lung cancer that is associated more strongly with cigarette smoking. Although pulmonary function damage has been shown as a risk factor for lung cancer development (18, 31), the accurate diagnosis of emphysema provides the opportunity to clarify the relationship between lung structural damage and lung cancer development. This could be helpful to even earlier cancer detection because anatomic changes of emphysema can show up without functional impairment. As the debate over the utility of CT for lung cancer screening continues, our results along with previous studies may suggest that incorporating risk factors, such as emphysema could aid in a better risk assessment of higher-risk individuals in terms of lung cancer screening (17). The emphysema diagnosis should not be missed or left off even without pulmonary function impairment.

Several mechanisms have been suggested to explain the strong association between emphysema and lung cancer. First, both lung cancer and emphysema are associated with cigarette smoking, which, by generating reactive oxidant species, induces a chronic inflammatory state in the lung. A chronic inflammation environment could lead to emphysema and lung cancer (32, 33). Second, mucociliary clearance is impaired with emphysema. During the clearing process, carcinogens tend to pool in areas with impaired mucociliary clearance, leading to lung cancer development (34). Third, lung cancer and COPD (including emphysema) may share some common genetic factors, independent of smoking (35, 36). Particularly, our previous study showed that lung cancer patients were significantly more likely to carry the mutated α₁-antitrypsin allele than the general population (20). The α₁-antitrypsin deficiency results in unneutralized neutrophil elastase and an associated breakdown of elastin in lung tissue, leading to early-onset and severe emphysema (37). An imbalance between neutrophil elastase and α₁-antitrypsin has been hypothesized to contribute to the development of emphysema and lung cancer (38).

Our study has several strengths. First, the sample size is reasonably large. Second, detailed epidemiologic and clinical information were carefully and completely collected, with thorough quality control procedures. Clinical diagnosis of emphysema was verified by medical records from pulmonologists, which minimized the potential recall bias of self-reported diagnosis. Third, cases and controls were better balanced in terms of age, gender, race, smoking status, area of residency, and recruited time period. Lastly and most importantly, the diagnoses of emphysema for all subjects were corrected by careful analyses of the standard-dose CT films at the time of lung cancer diagnosis (cases) or during the study (controls), allowing accurate estimation of the real contribution of emphysema to lung cancer risk.

Our study has several limitations. First, we used visual scoring for assessment of emphysema rather than objective quantification. This is due to the case–control study design, and we only used the digitally archived chest CT films. However, the visual emphysema score has been described as highly correlated with objective volume-based computerized assessment for the whole lung (39) and could be more accurate to distinguish the nature of air-space lesions, for example, emphysema from UIP. Second, although the radiologist was blinded to the case–control identity, visible presence of lung cancer on the CT may have influenced his estimate of emphysema. Third, pulmonary function results were not incorporated in the multivariable analyses due to the unavailability of the data. However, exploring the effect of pulmonary morphology instead of function on lung cancer risk is the main purpose of this study. Also as previous studies showed (16, 17), CT-diagnosed emphysema remained a significant lung cancer risk factor with and without additional adjustment for airflow obstruction. Therefore, we believe our results are reliable even without incorporating lung function data. Last, we have had only 13 cases and 18 controls with pack-years less than 10; therefore, we are unable to evaluate the effect of mild or light exposure to cigarette smoking on the emphysema–lung cancer association. On the contrary, we have tried our best to minimize the confounding effect of smoking on the results.

In conclusion, our results underscore the importance of the real and precise relationship between two of the most common and deadly diseases among cigarette smokers. Emphysema detected in smokers could be used to identify individuals who may need additional follow-ups, regardless of pulmonary function status. Although not in any position to advocate using CT to screen for emphysema at large, our study did show the low reliability of clinical assessment for emphysema status in smokers. Future studies are needed to identify and validate biomarkers and molecular signatures for individuals at the highest risk for lung cancer (37).

No potential conflicts of interest were disclosed.

We thank Susan Ernst for her technical assistance with the manuscript.

This work was supported by grants from the U.S. National Institutes of Health (R01 CA80127, R01 CA84354, and R01 CA115857 to P. Yang), Mayo Clinic institutional funds to P. Yang, and the Chinese Government Scholarship for Graduates to Y. Li.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.