CT Emphysema Predicts Thoracic Aortic Calcification in Smokers with and Without COPD

ABSTRACT

COPD patients are at increased risk for cardiovascular morbidity and mortality independent of smoking habits. Recent studies suggest CT emphysema is an independent predictor of cardiovascular risk as evidenced by its association with arterial stiffness and impaired endothelial function. We examined the relationship between demographics, lung function, CT emphysema and airway wall thickness and thoracic aortic calcification, another marker of cardiovascular risk, in the National Lung Screening Trial. We hypothesized that CT emphysema would be independently associated with thoracic aortic calcification. Two hundred forty current and former smokers were enrolled. After CT examination, we recorded subjects’ demographics and they performed spirometry. Subjects were classified into COPD and non-COPD subgroups. CT emphysema was quantified as a percentage of lung volume and measurements of the right upper lobe airway were performed using standard methods and expressed as wall area (%). Total calcification scores for the thoracic aorta were computed using TeraRecon image analysis. Univariate and multivariate analyses were performed to determine the associations between calcium score and subject characteristics. Subjects with COPD were older, more often male, heavier smokers and had more CT emphysema and greater aortic calcification than those without COPD. Calcium score was associated with age, pack-years, CT emphysema, wall area%, and lung function on univariate testing but only with age and CT emphysema on multivariate analysis. We conclude that CT emphysema is independently associated with thoracic calcification and thus may be used to assess cardiovascular risk in smokers with and without COPD.

Keywords: :

• COPD
• emphysema
• CT scanning
• thoracic aortic calcification
• cardiovascular disease

BACKGROUND

Chronic obstructive pulmonary disease (COPD) is the fourth-leading cause of death in the United States and is the only major cause of death for which mortality continues to increase (1, 2). Although many COPD patients die from respiratory failure, cardiovascular disease is a leading cause of death in individuals with COPD regardless of the severity of airflow obstruction (3–5). Currently there is no COPD-specific marker of cardiovascular risk.

The mechanisms underlying the relationship between COPD and cardiovascular disease remain unclear, but appear to be independent of smoking history (3, 6). It has been hypothesized that the endothelial dysfunction responsible for insults to the coronary vasculature similarly affects the pulmonary vasculature and can lead to emphysema (7, 8). Others have suggested that the mechanistic link between COPD and cardiovascular disease is attributable to an inflammatory process whereby loss of elastin occurs in both the arterial wall and the alveolar architecture (9–11). These hypotheses are supported by recent studies demonstrating associations between CT emphysema and impaired brachial artery flow-mediated vasodilation (FMD), a marker of endothelial dysfunction, and increased pulse wave velocity (PWV), a measure of arterial stiffness (8, 9).

Alhough FMD and PWV are both highly predictive of cardiovascular events, there are limitations to the use of these markers in patients with COPD, as both are affected by cigarette smoking and other vasoactive drugs. Although the correlations between CT emphysema and these markers are intriguing, the findings are limited by the exclusion of active smokers in the former study (8) and the use of carotid-brachial PWV rather than the “gold standard” carotid-femoral PWV in the latter (9, 12).

Calcium deposits in the aorta signify the development of atherosclerotic disease, and populations with more extensive thoracic aortic calcification have an increased incidence of cardiovascular mortality independent of common risk factors such as age, smoking, hypertension, serum cholesterol, and diabetes (13–15).

To our knowledge, no study has examined the associations between the clinical and radiographic features of COPD and the presence of thoracic aortic calcification. The marker is attractive as it is unaffected by vasoactive medications or active cigarette smoking. The aim of this project was to examine the relationship between demographics, lung function, the presence of COPD, CT emphysema, and airway wall thickness and thoracic aortic calcification in the National Lung Screening Trial (NLST) population. We hypothesized that CT emphysema would correlate with the presence of aortic calcification. Some of the results of this study have been previously published in the form of an abstract (16).

MATERIALS AND METHODS

The study received approval from the University of Alabama at Birmingham and Partners Health Care Institutional Review Boards as well as the NLST as an associate study.

Patient population

The NLST is sponsored by the National Cancer Institute and was designed to compare annual chest radiographs with low-dose CT for the early detection of lung cancer. The primary outcome of the NLST is lung cancer mortality and approximately 50,000 subjects were recruited from 2002–2004. Participants are men and women aged 55 to 74 years with a minimum 30-pack-year history of cigarette smoking. Exclusion criteria included prior lung cancer, pulmonary resection, or recent respiratory infection. We approached consecutive NLST participants at the University of Alabama at Birmingham site for participation in the study at the time of one of three annual CT screenings.

Study procedures

After CT screening, subjects signed informed consent and their demographic and medical information was recorded. A smoking history was obtained and subjects performed pre-bronchodilator spirometry according to American Thoracic Society standards (17). FEV1 was recorded as percent predicted using race-adjusted reference values (18). Subjects with FEV1/FVC ratio ≥ 0.70 and FEV1% predicted < 80% were classified as restricted and excluded from analysis. COPD was defined by the presence of an FEV1/FVC ratio < 0.70.

CT analysis

NLST CT scans were obtained with a multidetector helical scanner at 120 kilovolt peak, 40 to 60 mA, and 1-s scan time using 2.5-mm collimation and contiguous reconstructions. Quantitative densitometric analysis was performed, and areas of CT emphysema were defined as low attenuation areas (LAA) (≤ −950 Hounsfield units). The percentage of LAA (LAA%) was then determined for the entire lung using open-source software (3D Slicer; Brigham and Women's Hospital; Boston, MA) as described previously (19, 20). Airway wall measurements of the right upper lobe bronchus were performed using standard methods and expressed as wall area% (WA%) (21, 22).

Measurements of aortic calcification were made using a post-imaging processing application (Aquarius Work Station, TeraRecon, San Mateo, CA) using the Calcium Analysis Tool (Figure 1). We determined the total aortic calcification score, or Agatston score (23), by obtaining the sum of each CT slice's area of calcification multiplied by a coefficient based on the peak CT number. Areas of calcification were defined by the standard scoring threshold (> 130 HU). To minimize the effect of CT image noise on the calcification measures, these values were divided by the standard deviation of the CT attenuation (HU) measured in the middle of the ascending aorta at the level of the origin of the left main coronary artery (24). Investigators determining the extent of aortic calcification were unaware of subjects’ lung function, smoking history and quantitative emphysema.

Figure 1.  Aortic calcification measurement. Shown is a representative screen image of the Aquarius Workstation highlighting areas of calcification within the thorax (>130 Hounsfield Units). The red arrow highlights an area of thoracic aortic calcification. The total aortic calcification score, or Agatston calcium score, was determined by obtaining the sum of each CT slice's area of calcification multiplied by a coefficient based on the peak CT number. Calcifications in the aortic valve and coronary arteries were excluded from analysis.

Statistical analysis

Differences in characteristics between subjects with and without COPD were examined using t-tests for continuous variables and homogeneity chi-squared analysis for categorical variables. Simple and multiple regression analysis was used to determine the associations of calcium score with age, sex, race, pack-years, LAA%, WA%, lung function and the presence of COPD. P values < 0.05 were considered significant. Statistical analyses were performed using SPSS for Windows version 15.0.

Table 1.  Subject characteristics

	Subjects with COPD (n = 159)	Subjects without COPD (n = 81)	p-value
Age, years	62.6 ± 5.0	60.7 ± 4.9	0.009
Sex,% male	69.8	51.9	0.006
Race,% Caucasian	89.3	93.8	0.36
Current smokers,%	56.6	51.9	0.57
Pack-years, No.	67 ± 32	54 ± 23	0.002
FEV1/FVC	0.58 ± 0.10	0.77 ± 0.05	< 0.001
FEV1,% predicted	66 ± 17	92 ± 8	< 0.001
WA%	78.1 ± 6.4	76.6 ± 5.4	0.08
LAA%	7.91 ± 6.95	3.34 ± 2.79	< 0.001
Calcium score	75.8 ± 109.7	47.6 ± 84.2	0.046

*Data are reported as mean ± SD unless otherwise indicated.

WA –wall area; LAA –low attenuation areas; FEV₁ –forced expiratory volume; FVC –forced vital capacity.

RESULTS

A total of 240 subjects were included in the study. The demographics, smoking history, lung function, LAA%, WA%, and thoracic aortic calcification scores for the COPD and non-COPD groups are shown in Table 1. Subjects were predominately male (64%) and Caucasian (91%) and had an extensive smoking history (55% were current smokers).

Approximately two-thirds of participants met spirometric criteria for COPD based on pre-bronchodilator spirometry. Subjects in the COPD group were older (62.6 vs. 60.7 years, p = 0.009), more often male (70% vs. 52%, p = 0.006), and were heavier smokers (67 vs. 54 pack-years, p = 0.002). Subjects with COPD also had more extensive CT emphysema (7.91% vs. 3.34%, p< 0.001) and higher calcium scores (75.8 vs. 47.6, p = 0.046). Subjects with COPD also exhibited a trend toward greater wall thickness (WA%) (78.1% vs. 76.6%, p = 0.08)

Table 2 displays the univariate correlations between subject demographics, lung function, CT parameters and calcium score. Age, pack-years, LAA%, and the presence of COPD were each positively associated with calcium score (p< 0.05 for all) while WA% and FEV1 (% predicted) were negatively correlated with calcium score (p<0.05). Gender, race and current smoking were not associated with calcium score.

Table 2.  Univariate analyses between subject characteristics and calcium score

	Calcium Score Regression Coefficient (95% Confidence Interval)	p-value
Age	9.74 (7.45 to 12.0)	< 0.001
Sex, male	15.8 (−11.3 to 42.8)	0.25
Race, Caucasian	10.7 (−34.6 to 55.9)	0.64
Pack-years	0.66 (0.24 to 1.09)	0.002
Current smoking	8.91 (−17.3 to 35.1)	0.50
WA%	−2.29 (−4.41 to −0.17)	0.03
LAA%	3.89 (1.86 to 5.92)	< 0.001
COPD (FEV1/FVC < 0.70)	27.9 (0.50 to 55.3)	0.046
FEV1,% predicted	−1.00 (−1.67 to −0.34)	0.003

WA –wall area; LAA –low attenuation areas; FEV₁ –forced expiratory volume; FVC –forced vital capacity

On multivariate analysis, age (8.63 (6.18–11.1); p<0.001) and LAA% (2.16 (0.10–4.22); p = 0.04) remained significant predictors for calcium score after adjustment for the other covariates (Table 3). The estimate for pack-years approached statistical significance (0.38 (−0.02–0.79); p = 0.06). The R² for the final model was 0.26. LAA% remained associated with calcium scores when the absolute FEV1/FVC ratio was used rather than the presence of COPD (2.36 (0.01–4.70); p = 0.049). Similar results were also obtained when FEV1 (% predicted) was included in the model, rather than the presence of COPD though the estimate for CT emphysema did not reach statistical significance (p = 0.10). When the presence of COPD and FEV1 (% predicted) were both included in the multivariate model the estimate for CT emphysema was again non-significant (p = 0.10). When uncorrected calcium scores were used we did not observe an independent correlation with LAA% (p = 0.35).

DISCUSSION

We have found that in a population of current and former smokers with and without COPD, CT emphysema (LAA%) is associated with the presence of thoracic aortic calcification, a well-established marker of cardiovascular risk (13–15). This association persists after adjustment for subject demographics, smoking history, a CT metric of airways disease and the presence of airflow obstruction. These data complement and extend prior reports demonstrating associations between LAA% and arterial stiffness and impaired flow-mediated vasodilation and argue that CT emphysema is a valid tool to stratify cardiovascular risk among smokers (8, 9).

In the current study, COPD subjects had higher thoracic calcium scores than healthy smokers and both the presence of airflow obstruction and FEV1 (%) were highly correlated with that measure of cardiovascular risk in univariate analyses. This is in keeping with prospective studies that have repeatedly shown that cardiovascular disease is a major cause of morbidity and mortality in COPD patients, accounting for between 25% and 50% of deaths (4, 5, 25).

In addition, it has been clearly demonstrated that reduced lung function as measured by FEV1 is highly associated with cardiovascular events (3, 26, 27) and that this correlation is independent of well-established cardiovascular risk factors such as smoking, hypercholesterolemia, and hypertension. Though these studies demonstrate the strong association between reduced lung function and cardiovascular morbidity and mortality, the mechanistic link remains unclear. It has been postulated that pathological events common to both emphysema and atherosclerosis, such as endothelial dysfunction and degradation of elastin, could explain this link and recent reports that demonstrate associations between CT emphysema and markers of cardiovascular disease support this assertion (8, 9).

McAllister has shown that CT emphysema is linearly associated with arterial stiffness as measured by PWV (9) extending the results reported by Sabit who demonstrated increased arterial stiffness in COPD patients as compared to control subjects matched for age, sex, blood pressure, and other cardiovascular risks (28). The results argue that elastin degradation may occur systemically due to exaggerated matrix-metalloproteinase driven inflammation and could lead to emphysema in the lung, stiffness of the arterial vasculature and accelerated atherosclerosis, and even loss of skin elasticity as suggested by animal models (11, 28–30).

Table 3.  Final multivariate model for predictors of calcium score

	Calcium Score Regression Coefficient (95% Confidence Interval)	p-value
Age	8.63 (6.18 to11.1)	< 0.001
Sex, Male	−14.6 (−40.4 to 11.1)	0.26
Race, Caucasian	2.44 (−37.7 to 42.6)	0.90
Pack-years	0.38 (−0.02 to 0.79)	0.06
Current smoking	2.68 (−20.5 to 25.8)	0.82
WA%	−0.92 (−2.90 to 1.06)	0.36
LAA%	2.16 (0.10 to 4.22)	0.04
COPD (FEV1/FVC < 0.70)	0.95 (−25.8 to 27.7)	0.94

WA –wall area; LAA –low attenuation areas; FEV₁ –forced expiratory volume; FVC –forced vital capacity.

Importantly, McAllister found no association between CRP and arterial stiffness, suggesting that systemic inflammation may not be the driver of accelerated vascular disease in COPD though many inflammatory markers were not examined (31, 32) and other authors have suggested such a relationship (33). Although these results are intriguing, the adoption of CT emphysema as a specific marker of cardiovascular risk is limited, as the correlation was observed between LAA% and carotid-brachial PWV rather than the preferred carotid-femoral PWV. Carotid-femoral measurements are thought superior as they are largely unaffected by changes in smooth muscle tone, better correlate with central artery stiffness, and predict cardiovascular risk in longitudinal studies (12).

In another recent study, Barr found that LAA% is linearly associated with endothelial dysfunction as measured by FMD of the brachial artery (8). The authors propose that the link between COPD and cardiovascular disease could be explained by vascular insults that affect both the coronary and pulmonary vasculature. It has been shown that apoptotic endothelial cells are present in the lungs of patients with COPD and that nitric oxide-mediated endothelial relaxation is reduced in their excised pulmonary vessels (8, 34, 35).

In their population of former smokers with COPD, Barr demonstrated a similar reduction in in vivo endothelial function as evidenced by reduced brachial FMD, which correlates well with intracoronary endothelial vasoactivity (36). Again, it is important to recognize that the use of CT emphysema as a marker of cardiovascular risk as supported by these data is limited by the exclusion of current smokers and by other factors such as medication use that may have confounded the observed associations. In addition, though no measures of CT emphysema were obtained, Maclay has recently reported that men with COPD, in fact, have normal endothelial function as compared to healthy controls raising questions about the utility of FMD as a marker of cardiovascular risk in this population (37).

Like PWV and FMD, aortic calcification is well-established marker of cardiovascular risk. Although the extent of aortic calcification has been shown to be associated with age and smoking (as we also observed); diabetes, and hypertension, it remains independently predictive of cardiovascular morbidity and mortality after adjustment for these traditional cardiovascular risk factors (13–15). In a population-based cohort study, Iribarren demonstrated that men and women with aortic arch calcifications were 25% and 22% more likely to develop coronary heart disease compared to those without aortic arch calcifications after adjustment for other risks (13). Other studies have shown similar associations between aortic calcification and an increased incidence of coronary artery disease and all-cause mortality (14, 15).

No prior study has examined the relationship between a metric of airways disease and a marker of cardiovascular risk in smokers. Though WA% was not an independent predictor of calcium score in the multivariate models it was negatively correlated when examined alone. As increasing WA% is associated with lower lung function, and therefore more advanced COPD, this may appear unexpected (21, 38); however, we observed an inverse correlation between LAA% and WA% (B = −0.15 (−0.28 to −0.02); p = 0.025) which may explain this finding. An inverse relationship between quantitative measures of airways disease and LAA% has also been described previously (39).

The independent correlation between LAA% and thoracic aortic calcification in our study strengthens prior reports and supports the concept that CT emphysema conveys additional information about cardiovascular risk in smokers that may also be associated with the underlying mechanism of disease. The findings are also important as measures of aortic calcification are not subject to instability caused by smoking or other vasoactive drugs or technical issues that limit the utility of FMD and PWV.

Our study has several limitations. The low-dose NLST CT protocol is not ideal for the measurement of airway wall thickness, and thus the determination of WA% could be impacted. We did observe a trend towards higher WA% among subjects with COPD, which would be expected. The low-dose CT protocol may also overestimate LAA% compared to higher dose scans though the technique has been used before and the effect is likely uniform across subjects (19, 20).

Measurements of aortic calcification are also subject to image noise due to the low-resolution scans, respiratory motion artifact, and cardiac pulsation artifact (23). We attempted to minimize the effect of image noise by standardization as described in the methods and indeed, an independent correlation between LAA% and aortic calcification was only observed when corrected calcium measures were used. In our preliminary analysis of these data, which was published as an abstract (16), a univariate association was observed between uncorrected calcium scores and LAA% but an independent association was not found on multivariate testing. Corrected calcium scores could not be reliably obtained on 26 of the 266 patients included in the abstract and thus these patients were excluded from the final analysis.

We did not examine the association between subject characteristics and coronary artery calcification, another powerful marker for cardiovascular risk, but a strong relationship between aortic calcification and coronary artery calcification as detected by CT has been previously shown (15, 23, 40). We were also unable to adjust our analysis for other cardiovascular risk factors not available in the database, such as blood pressure, lipids, diabetes, hypertension, and inflammatory markers. These factors may certainly be associated with the extent of aortic calcification though a larger study would be required to adequately correct for their effects. Lastly, though only pre-bronchodilator spirometry measures were available, which almost certainly led to an overestimation of the number of subjects with COPD (41), classification of COPD based on pre-bronchodilator spirometry measures has been previously shown to predict COPD-related hospitalization and death (42).

In conclusion, these results support the use of CT emphysema as a tool to stratify cardiovascular risk in smokers with and without COPD. Additional large-scale studies are needed to confirm these findings and to further examine the association between CT emphysema and other markers of cardiovascular disease as well as the occurrence of cardiovascular events.

Declaration of interest

The authors have no conflict of interests to declare. The authors are solely responsible for the content, writing and approval of the manuscript.