Computed Tomography–Diagnosed Emphysema, Not Airway Obstruction, Is Associated with the Prognostic Outcome of Early-Stage Lung Cancer

Lung cancer is the most common malignancy in the world and the leading cause of cancer-related mortality (1). It is generally accepted that cigarette smoking is the number one cause of this devastating disease (2). The oxidants in smoke are responsible for the chronic biological damage, including injury to the DNA (3–6), predisposing to lung carcinogenesis. Throughout the carcinogenic pathway, premalignant cells are exposed to repeated cytogenetic damage, which eventually results in the aggressive clinical characteristics of smoking-related lung cancer (7–9).

Smoking also causes emphysema, probably with similar pathogenesis, because the oxidants damage tissue either directly or by partially inactivating α1-plasminogen inhibitor (10). Regardless of the recognized pathogenesis, cigarette smoking plays a part in the development of emphysema in only a proportion of smokers. Because the susceptibility to smoking-induced tissue damage varies among individuals, we questioned if the clinicopathologic characteristics of lung cancer arising in smokers differ according to the individual's predisposition to the development of emphysema. Other investigators have compared the clinicopathologic characteristics of smoker's lung cancer based on spirometric measurements instead of lung structural damage and, unfortunately, found no specific findings. Spirometry relies on the detection of increased airway resistance and decreased surface area of the alveolar-capillary membrane (11, 12). This test is relatively insensitive and nonspecific for identifying the severity of emphysema. In contrast, computed tomography densitometry is an objective technique, which uses computer software to distinguish voxels with abnormally low attenuation, representing emphysema, from those representing normal lung (13, 14). Because emphysema is defined by pathologic criteria as “abnormal permanent enlargement of air spaces distal to the terminal bronchioles, accompanied by destruction of the alveolar walls and without obvious fibrosis” (15), the potential of computed tomography in diagnosing emphysema may be superior to that of spirometry.

In the present study, we attempted to clarify the clinicopathologic features and prognostic importance of smoking-related emphysema detected by computed tomography in smokers with early non–small cell lung cancer.

Patients. Between January 2000 and December 2003, 163 patients underwent surgery for primary lung cancer at our institution. The subjects of this study were a consecutive series of 100 of these patients who underwent lung lobectomy with hilar and mediastinal lymph node dissection for clinical stage I non–small cell lung cancer. All 100 patients had a smoking habit before surgery. Operability was determined according to the existing guidelines for pulmonary resection (16). The criteria for resection included a partial pressure of arterial carbon dioxide (pCO₂) of 50 mm Hg, a mean pulmonary arterial pressure <30 mm Hg, and a calculated predicted postoperative forced expiratory volume in 1 second >500 mL. The preoperative patient data included age, sex, performance status, presence or absence of breathlessness, smoking habit, spirometric variables, arterial blood gas values, tumor size, and the extent of emphysema. Smoking data were based on the pack-years smoked, with the smoking index calculated by the average number of packs of cigarettes smoked per day multiplied by the number of years the person had smoked. The extent of emphysema was determined by computed tomography densitometry. Spirometric variables were obtained within the preoperative month and included vital capacity and forced expiratory volume in 1 second (FEV₁). Vital capacity and FEV₁ were expressed as percentages of predicted values for age, sex, and height. Pathologic data obtained after surgery included histologic subtype, histologic grade, microvessel invasion, lymphatic vessel invasion, and pathologic tumor-node-metastasis status. The tumors were classified histologically as squamous cell carcinoma or non–squamous cell carcinoma and graded as well, moderately, or poorly differentiated carcinoma according to the WHO classification (17). Likewise, microvessel invasion and lymphatic vessel invasion were diagnosed histologically using H&E and Elastica Van Gieson stains. The patients' clinicopathologic characteristics are summarized in Table 1. The mean smoking index was 52 ± 27 pack-years (median = 50 pack-years), and the mean number of resected segments was 4.1 ± 1.1. Preoperatively, 29 patients complained of slight breathlessness, defined as inability to keep up with healthy persons of equivalent age on hills or stairs (18), and 23 patients had restricted physically strenuous activity, defined as Eastern Cooperative Oncology Group performance status grade 1 (19).

Diagnosis of emphysema. Computed tomography was done with a Siemens Volume Zoom Scanner (Siemens-Asahi Medical Technologies Ltd., Tokyo, Japan) in the helical mode, without i.v. contrast material. With the patient in the supine position, scans were done during full inspiration with the following measurements: 120 to 140 kVp, 280 to 320 mA, 10-mm collimation, and a pitch of 1.5. Scan volumes extended from the thoracic inlet to the lung base and were acquired in one-breath-hold periods. Lung images were reconstructed in a 512 × 512 matrix with a lung algorithm. We constructed volume-rendering three-dimensional models of the lungs using imaging software (M900 QUADRA, ZioSoft K.K., Osaka, Japan). Threshold limits of −600 to −1,024 Hounsfield units were applied to segment the lung parenchyma and to exclude the soft tissue surrounding the lung and the large vessels within the lung. The trachea, mainstem bronchi, and gastrointestinal structures were manually and selectively removed from the model, using volume-adding and multifusion imaging techniques. The total number of voxels with any selected specific attenuation number (in Hounsfield units) in the lung model was counted automatically by the computer. The percentage of voxels with attenuation values lower than −910 Hounsfield units among the total number of voxels in the entire lung was considered the extent of emphysema because the low-attenuation thresholds that have been used most often to identify emphysema on conventional 10-mm-thick computed tomography sections are −900 or −910 Hounsfield units (20–23). Emphysema was confirmed if it occupied >5% of the lung because 95% of nonsmokers in our preliminary series had lungs with <5% emphysematous involvement.

Postoperative follow up. All 100 patients were followed up in our outpatient clinic at 3- or 4-month intervals during the first 2 years after their operation, then at 6-month intervals, by chest computed tomography and if necessary, head and abdominal computed tomography.

Statistical analysis. We compared continuous variables by the unpaired Student's t test and categorical variables by the χ² test. Dependence between two continuous variables was tested by linear regression analysis. Survival analysis was done using a univariate Cox regression model. Variables with P < 0.1 at univariate analysis were then used as independent variables in a stepwise Cox regression analysis, with P = 0.05 as criterion for retention of variables in the final model. The multivariate procedure was validated by bootstrap bagging with 100 samples. In the bootstrap procedure, repeated samples of 100 observations were selected with replacement from the original set of observations. For each sample, stepwise Cox regression was done by entering the variables with P < 0.05 at the original analysis. Survival curves were drawn by the Kaplan-Meier's method, and the difference between the curves was determined by the log-rank test. All tests were two tailed, with significance defined as 0.05, and were done using the statistical software Statview 5.0 (SAS, Inc., Cary, NC) and STATA 9.1 (Stata Corp., College Station, TX).

There were 72 (72%) patients with pathologic stage I and 28 (28%) patients with pathologic stage II or more. We diagnosed emphysema in 58 (58%) of the 100 patients. There were no significant differences between the patients with and those without emphysema in age, pack-year smoked, performance status, incidence of breathlessness, gas exchange capacity, vital capacity, tumor size, clinical stage, surgical procedure, pathologic nodal status, pathologic T status, histologic subtype, histologic grade, and microvessel invasion; however, we noted male predominance and a poorer FEV₁ in the patients with emphysema (Table 2). The dependence between the extent of emphysema and the FEV₁ is shown in Fig. 1. The presence or absence of emphysema affected both overall survival (5-year overall survival rate of 39.4% for patients with emphysema and 83.9% for patients without emphysema; P = 0.001; Fig. 2) and disease-free survival (5-year disease-free survival rate of 44.0% for patients with emphysema and 73.6% for patients without emphysema; P = 0.0055; Fig. 2); whereas FEV₁ did not affect overall survival (5-year overall survival rate of 48.3% for patients with FEV₁ ≥ 70% and 58.4% for patients with FEV₁ < 70%; P = 0.9417; Fig. 3) and disease-free survival (5-year disease-free survival rate of 58.6% for patients with FEV₁ ≥ 70% and 53.5% for patients with FEV₁ < 70%; P = 0.7250; Fig. 3).

Using the univariate Cox regression model, tumor size, clinical stage, the presence or absence of emphysema, histologic subtype, histologic grade, microvessel invasion, and pathologic N status were significantly associated with overall survival; whereas tumor size, clinical stage, the presence or absence of emphysema, histologic subtype, histologic grade, microvessel invasion, pathologic T status, and pathologic N status were significantly associated with disease-free survival (Table 3). Using the stepwise Cox regression analysis, the presence of emphysema [hazard ratio (HR), 6.15; 95% confidence interval (95% CI), 2.16-17.53; P = 0.001], together with tumor size (HR, 1.08; 95% CI, 1.04-1.12; P < 0.001), clinical stage IB (HR, 3.85; 95% CI, 1.20-12.50; P = 0.022), nodal metastasis (HR, 4.01; 95% CI, 1.74-9.24; P = 0.001), and microvessel invasion (HR, 2.72; 95% CI, 1.25-5.92; P = 0.012), was considered an independent poor prognosticator for overall survival; whereas the presence of emphysema (HR, 5.12; 95% CI, 2.12-12.36; P < 0.001), together with tumor size (HR, 1.04; 95% CI, 1.02-1.07; P < 0.001), squamous cell carcinoma (HR, 2.27; 95% CI, 1.05-4.76; P = 0.037), nodal metastasis (HR, 3.22; 95% CI, 1.43-7.25; P = 0.005), and microvessel invasion (HR, 3.12; 95% CI, 1.46-6.68; P = 0.003), was considered an independent poor prognosticator for disease-free survival (Table 4). These results were stabilized by the bootstrap replication model, according to which the presence of emphysema remained an independent determinant for both overall survival (HR, 6.15; 95% CI, 1.72-21.93; P = 0.005) and disease-free survival (HR, 5.12; 95% CI, 1.87-14.04; P = 0.001).

Cigarette smoking is thought to influence the biological characteristics of lung cancer through cytogenetic damage caused by the oxidant inhalation (3–6). However, the smoking-related biological damage that develops differs individually, as shown by the fact that emphysema develops in only a proportion of smokers. The present study is the first to provide clear evidence that susceptibility to lung destruction caused by smoking is associated with the clinical aggressiveness of lung cancer in smokers. Determining the extent of emphysema using computed tomography densitometry may be useful for the accurate staging of lung cancer, as well as to study the pathogenesis of smoking-related lung cancer.

There is wide geographic variation in the incidence of emphysema (24, 25). Japan has a much lower incidence of emphysema than the United States or United Kingdom probably because emphysema is almost exclusively caused by smoking in Japan. Based on our preliminary results, only 5% of nonsmokers undergoing lung cancer surgery were found to have emphysema by computed tomography densitometry. Because the pathogenesis of emphysema is generally multifactorial (10), more effort should be made to clarify how the potential of an individual to develop emphysema affects survival, even in nonsmokers with lung cancer, especially in countries with a high incidence of emphysema.

The potential for the development of emphysema, but not emphysema itself, seemed to directly affect the prognostic outcome of smokers undergoing surgery for lung cancer because both the overall and disease-specific survival differed between the patients with and those without emphysema in this study. Moreover, two patients without emphysema and three patients with emphysema died without cancer recurrence, and only one of these three emphysema patients died of emphysema.

Major prospective studies have shown a rising trend in lung cancer death rates with increasing smoking exposure (26–28), suggesting that the degree of smoking exposure is proportional to the incidence of lung carcinogenesis but not the clinical aggressiveness of lung cancer. This is consistent with the fact that most investigators, including us, failed to find a definite association between the degree of smoking exposure and the therapeutic outcome of smoker's lung cancer (29).

Emphysema, together with chronic obstructive bronchitis, is a major component of chronic obstructive pulmonary disease (COPD). It is well recognized that persons with COPD have a shorter life expectancy (30) and an increased risk of lung cancer death (31, 32) than healthy persons. The progression of COPD is measured clinically by FEV₁ because limited FEV₁ is associated with an increased risk of death from respiratory failure (30) as well as that from lung cancer (31, 32). The risk of lung cancer in COPD patients has been attributed to impaired mucociliary clearance (33, 34). During the clearing process, particles tend to pool in areas with impaired mucociliary clearance. This pooling may allow carcinogens from the smoke in the mucous blanket to have longer exposure time at these sites, leading to development of lung cancer. In contrast to functional assessment, previous investigators have attempted to grade COPD by the morphologic aspect. Unfortunately, neither computed tomography quantification of emphysema nor pathologic measures of airway structural abnormalities correlated well with FEV₁ (35). Furthermore, according to a matched case-controlled study, computed tomography quantification of emphysema was not a risk factor for lung cancer (36). In the present study, we showed the prognostic significance of computed tomography–diagnosed emphysema, but not FEV₁, in patients with surgically resected, early-stage lung cancer. These findings suggest that although both computed tomography quantification of emphysema and FEV₁ can be used to grade COPD, these measurements may be linked differently with lung cancer.

Jiang et al. evaluated the prognostic significance of surfactant protein-A (SP-A) genetic aberration, detected by fluorescence in situ hybridization, in patients with stage I non–small cell lung cancer. They found that deletion of the SP-A gene in the underlying bronchial epithelial cells is associated with a poor prognosis (37). This finding indirectly supports our result because, according to the literature, the SP-A protein level is likely to be reduced in the bronchial lavage fluid in patients with emphysema (38).

There are some theories as to why computed tomography–diagnosed emphysema is associated with a poor prognosis in patients with lung cancer. First, patients with emphysema may have increased susceptibility to smoking-related biological damage, including damage to the DNA, which ultimately determines the aggressiveness of the tumor cells; and second, tumor progression is enhanced in emphysematous lungs where matrix metalloproteinases are rich (39). Ishikawa et al. reported the importance of matrix metalloproteinases in the underlying lung parenchyma on tumor progression (40). Thus, further studies should be done to clarify the direct link between computed tomography–diagnosed emphysema and the various biomarkers in smokers with lung cancer.

The user-friendly computer software available allows us to determine within a few minutes the proportion of voxels with attenuation values within the range of that representing emphysema, without need for extensive technical training, by creating a three-dimensional computed tomography lung model and imputing a defined threshold. The results are reproducible for viewers of varying expertise and experience and across institutions, allowing for accurate comparisons of results among different centers (14). In our series, the extent of emphysema was mildly dependent on the FEV₁ (r = 0.375) but was independent of the pack-year smoked (r = 0.091). This is consistent with the fact that there is little published evidence on the relationship between the degree of smoking exposure and the severity of emphysema (41). Thus, our patients with emphysema at the time of lung cancer development seemed to have increased susceptibility to smoking-related emphysema. Our previous study showed that measuring the extent of emphysema was also useful for predicting early-postoperative outcome with respect to the risks of hypoxemia, cardiopulmonary complications, and air leaks, in patients undergoing lung cancer surgery (42, 43). Furthermore, we found that patients with moderate to severe emphysema detected by computed tomography were likely affected by volume reduction effect after lung cancer surgery (44). Therefore, the routine determination of emphysema using this semiautomatic diagnostic tool should be mandatory for patients scheduled to undergo lung cancer surgery, considering that it incurs minimal cost and labor and does not subject the patient to invasive intervention.

In conclusion, susceptibility to emphysema, together with traditional clinicopathologic variables, affects the prognostic outcome of smokers with early-stage lung cancer independently. The routine use of computed tomography densitometry in smokers with lung cancer should be mandatory.