Eighteen men with COPD and 17 healthy male control subjects were recruited from primary care and a hospital respiratory outpatient clinic. Subjects were matched for age and prior smoking habit. Men aged 40 to 80 years who were ex-smokers (smoking cessation for at least 3 mo) but with a smoking history of 10 pack-years or greater were included. Exclusion criteria were a history of pulmonary fibrosis, tuberculosis, bronchiectasis, lung cancer, or lung resection; conditions known to affect vascular function, including obstructive sleep apnea, cardiovascular, cerebrovascular, and peripheral vascular disease, uncontrolled hypertension, or diabetes; inflammatory conditions, such as rheumatoid arthritis or psoriasis; or taking drugs that affect vascular function, including statins, angiotensin-converting enzyme inhibitors, and β-blockers. Healthy control subjects had normal spirometry and no history of respiratory symptoms. Subjects with COPD had a history consistent with the disease, chronic airflow limitation on spirometry (post-bronchodilator FEV₁/FVC ratio ≤70%) (24), stable disease (no exacerbation of COPD within the previous 6 wk, defined as a sustained change in symptoms requiring antibiotic or steroid therapy), and were not prescribed regular oral steroid therapy or long-term oxygen therapy.

All studies were conducted at the Wellcome Trust Clinical Research Facility, Royal Infirmary, Edinburgh. Height, weight, and pre- and post-bronchodilator spirometry were measured (Alpha Spirometer; Vitalograph, Buckingham, UK) according to American Thoracic Society/European Respiratory Society standards (24). All studies were approved by Lothian Regional Ethics Committee and conducted with the written informed consent of all participants.

Studies were conducted as per the Expert Consensus Document on Arterial Stiffness, assessing both pulse wave velocity and pulse wave analysis using a high-fidelity micromanometer (Millar Instruments, Houston, TX) and the SphygmoCorTM system (AtCor Medical, Sydney, Australia) (25). After 30 minutes of supine rest, peripheral systolic and diastolic blood pressures were measured using an automated noninvasive oscillometric sphygmomanometer (Omron 705IT; Omron, Milton Keynes, UK). Studies were performed in the morning in a quiet, dimly lit, temperature-controlled room (22–25°C). Subjects had fasted overnight and abstained from coffee, tea, and alcohol for the 24 hours before the study. All medications were withheld on the morning of the study. Exhaled carbon monoxide measurements (<5 ppm) ensured no acute cigarette smoke exposure. Carotid-femoral (aortic) pulse wave velocity is increased with increasing arterial stiffness. Pulse wave analysis assesses both augmentation pressure and augmentation index, which are increased in the presence of stiff arteries, whereas time to wave reflection is reduced.

Under the same ambient conditions and subject restrictions, bilateral forearm blood flow was measured using venous occlusion plethysmography as described previously (26). The endothelium-dependent vasodilators, bradykinin (100, 300, and 1,000 pmol/min) and acetylcholine (5, 10, and 20 μg/min), and the endothelium-independent vasodilators, sodium nitroprusside (2, 4, and 8 μg/min) and verapamil (10, 30, and 100 μg/min), were infused via a 27-gauge intrabrachial needle incrementally for 6 minutes at each dose. Vasodilators were separated by 15-minute saline infusions and given in random order except for verapamil, which was administered last due to its slow offset of action. Plethysmograph traces were recorded using Chart 5 software (ADinstruments, Chalgrove, UK) and were analyzed by one investigator who was blinded to subject identity. The mean of the last five waveforms from each set of readings was used to calculate forearm blood flow.

Venous cannulae (17-gauge) were inserted into both antecubital fossae. Baseline blood samples were obtained for hemoglobin and hematocrit, and fasting blood samples were obtained for both glucose and lipid profile measured in the clinical laboratories of the Royal Infirmary of Edinburgh (Sysmex, Bornbarch, Germany, and Olympus Analyzers, Southend-on-Sea, UK). Arterial blood gases were measured at rest (Bayer Rapidlab; Bayer HealthCare, Uxbridge, UK). Blood was sampled at baseline, centrifuged, and serum stored at −80°C for subsequent analysis. Serum C-reactive protein (CRP) concentrations were measured using a highly sensitive immunonephelometric assay (Behring BN II nephelometer; Global Medical Instruments, Ramsay, MN).

Infusion of intrabrachial bradykinin (100, 300, and 1,000 pmol/min) not only causes endothelium-dependent vasodilatation but also stimulates endothelial tissue plasminogen activator (t-PA) release (26, 27). Venous blood (10 ml) was collected at baseline and during each dose of bradykinin into acidified buffered citrate (Stabilyte tubes; Biopool International, Newton, NC) for t-PA antigen, and into citrate (Monovette; Sarstedt, Bray, Ireland) for plasminogen activator inhibitor type 1 (PAI-1) antigen estimation. Samples were collected onto ice, centrifuged at 2,000 × g for 30 minutes at 4°C and plasma was stored at −80°C until analyzed. Plasma t-PA and PAI-1 antigen concentrations were determined by ELISA (TintElize t-PA, Biopool EIA, and Elitest PAI-1, Hyphen Biomed [Neuville-en-Oise, France], respectively).

Absolute t-PA release was calculated as the difference in the t-PA antigen concentration measured in the infused and noninfused arms. Estimated net release of t-PA antigen was calculated as previously described (27) as the product of the infused forearm plasma flow (based on the mean hematocrit [Hct] and the infused forearm blood flow [FBF]) and the concentration difference between the infused ([t-PA]_Inf) and noninfused ([t-PA]_Noninf) arms: Estimated net t-PA release = FBF × (1 − Hct) × ([t-PA]_Inf − [t-PA]_Noninf). Data were analyzed using two-way ANOVA with repeated measures and Student or Welch (for groups with unequal variances) unpaired t-tests as appropriate. Univariate comparisons were analyzed using Pearson correlations. CRP and PAI-1 were log transformed to correct for positive skewness and were presented as median (interquartile range). All analyses were performed using SPSS version 16.0 (Chicago, IL). Statistical significance was taken at P < 0.05.

Measures of arterial stiffness were higher in patients with COPD (Table 2, Figure 1). Carotid-femoral pulse wave velocity was higher in subjects with COPD compared with control subjects (mean ± standard deviation; 11 ± 2 vs. 9 ± 2 m/s, P = 0.003). When corrected for differences in heart rate, augmentation index was similarly increased in patients with COPD (27 ± 10 vs. 21 ± 6, P = 0.028). Consistent with these findings the time to wave reflection was reduced, although this difference was not statistically significant (142 ± 21 vs. 150 ± 12 milliseconds, P = 0.13).

Figure 1. (A) Aortic pulse wave velocity and (B) augmentation index are higher in patients with chronic obstructive pulmonary disease (COPD) when compared with age- and smoking habit–matched control subjects (t test, P = 0.003 and P = 0.03). Symbols represent individual values and lines represent the means.

Download Figure | Download Powerpoint

[More] [Minimize]

In all subjects, aortic pulse wave velocity correlated significantly with systolic blood pressure (r = 0.62, P < 0.001) and measurements of airflow obstruction (FEV₁, r = −0.38, P = 0.03; FEV₁/FVC, r = −0.45, P = 0.008), but there was no association with systemic inflammation (circulating leukocytes, r = 0.21, P = 0.24; hsCRP, r = 0.30, P = 0.10) or arterial oxygen tension (r = −0.05, P = 0.80).

Baseline blood flow was similar (P = 0.43; Table 3), and blood pressure and heart rate were unchanged throughout the studies in both groups. All vasodilators caused a dose-dependent increase in forearm blood flow (P < 0.001 for all; Figure 2) that was similar in both groups (bradykinin, P = 0.73; acetylcholine, P = 0.72; sodium nitroprusside, P = 0.31; verapamil, P = 0.80).

Figure 2. Vasomotor function in response to two endothelial-dependent and -independent vasodilators is preserved in patients with chronic obstructive pulmonary disease (closed symbols) in comparison to age- and smoking habit–matched control subjects (open symbols; repeated measures two-way analysis of variance: bradykinin, P = 0.73; acetylcholine, P = 0.72; sodium nitroprusside, P = 0.31; verapamil, P = 0.80). Symbols represent the means and bars represent the standard errors of the means.

Download Figure | Download Powerpoint

[More] [Minimize]

Baseline plasma t-PA antigen concentrations were similar in patients with COPD and control subjects (mean ± SD; 13.2 ± 5.0 vs. 13.0 ± 4.0 ng/ml, P = 0.92; Table 3). Bradykinin caused a dose-dependent increase in plasma t-PA antigen concentrations in both groups (P < 0.001 for both). There were no differences in absolute t-PA antigen release or stimulated net t-PA release after bradykinin infusion in subjects with COPD in comparison with control subjects (net t-PA release 63.2 ± 102.7 vs. 69.4 ± 71.9 ng/100 ml of tissue/min at 1,000 pmol/min, P = 0.90; Figure 3). Plasma PAI-1 antigen concentrations were similar in both groups (median [interquartile range], baseline 30.9 [24.4–41.6] ng/ml vs. 28.4 [18.1 to 53.9] ng/ml; P = 0.38) and were unchanged by bradykinin infusion (P = 0.99).

Figure 3. Acute endothelial release of tissue plasminogen activator (t-PA) in response to bradykinin in patients with chronic obstructive pulmonary disease (solid symbols) in comparison to age- and smoking habit–matched control subjects (open symbols; repeated measures two-way analysis of variance, P = 0.90). Symbols represent the means and bars represent the standard errors of the means.

Download Figure | Download Powerpoint

[More] [Minimize]

Patients with COPD have increased arterial stiffness and other abnormalities in the systemic vasculature. Increased large artery stiffness results in greater central aortic systolic pressures, increased left ventricular afterload, and reduced diastolic coronary artery filling (28), and as such may be an important determinant of cardiovascular risk in patients with COPD. Previously, we (29) and others (7) have shown increased aortic pulse wave velocity, elevated augmentation pressure, and reduced time to wave reflection in patients with COPD. However, these studies were limited by use of suboptimal measures of arterial stiffness (29) and inadequate matching of smoking exposure between study groups (7). In contrast, using recommended gold-standard measures of arterial stiffness (carotid-femoral pulse wave velocity), we have demonstrated that arterial stiffness is increased in patients with COPD compared with control subjects with similar cigarette smoke exposure. This suggests an association between arterial stiffness and COPD that is independent of the effects of cigarette smoke.

Although arterial stiffness of a conduit artery, such as the aorta, is influenced by the extracellular matrix, vascular smooth muscle, and the endothelium (12), regulation of blood flow in resistance vessels is governed primarily by vascular smooth muscle and the endothelium. We have performed detailed studies of endothelial function across resistance vessels in the forearm vascular bed using two endothelium-dependent vasodilators (bradykinin and acetylcholine) and two endothelium-independent vasodilators (sodium nitroprusside and verapamil). Using the robust and well-validated technique of forearm plethysmography, we found no differences in endothelium-dependent or -independent vasomotor function in patients with COPD when compared with control subjects matched for smoking status. This is not to say that patients with COPD do not have endothelial dysfunction. We have previously demonstrated marked endothelial dysfunction in smokers (20) and it is possible that the effects of chronic smoking or aging may dominate any effects of COPD on resistance vessel vasomotor function.

In a previous study of vascular function in COPD, Barr and colleagues (9) found that flow-mediated dilation (FMD), a noninvasive measure of arterial vasomotor function, was associated with both airflow obstruction and emphysema severity in former smokers with and without COPD. FMD measures vasodilatation of the brachial artery after reactive hyperemia of the forearm (30), which is partly endothelium- and nitric oxide–dependent (31). However, as the authors concede, no endothelium-independent vasodilator (e.g., nitroglycerine) was used as a control in this study, and therefore the abnormality described cannot be definitively localized to the endothelium and may be due to dysfunction of other components of the arterial wall, such as the vascular smooth muscle or the extracellular matrix. Indeed, a subsequent case-control study did use a nitroglycerin control and found that nitroglycerin-mediated dilatation was impaired in the COPD group (8). This finding implies that the vascular abnormality in COPD is not restricted to the endothelium. In addition, there is evidence that FMD is an unreliable measure of endothelial function in the presence of stiff arteries (32, 33). Interestingly, a pattern of vascular abnormalities similar to that presented in this study is seen in patients with Marfan syndrome who have large artery stiffness and preserved agonist-mediated vasodilatation despite impairment of FMD (34).

Both FMD and venous occlusion plethysmography are well-established techniques for assessing endothelial vasomotor function. Although the former technique examines vasomotion in a conduit vessel and the latter in resistance vessels, endothelial dysfunction is believed to be a systemic process and abnormalities are unlikely to be restricted to one vascular bed. Assessments of vascular function in the peripheral circulation, using either technique, closely relate to assessments of vasomotor function in the coronary circulation (35, 36) and are predictive of cardiovascular events, even in individuals with no known atherosclerosis (14, 37). Although there is evidence of selected abnormalities of either resistance or conduit vessels in specific circumstances, such as rare hereditary arteriopathies (38), we believe it is unlikely that COPD preferentially affects endothelial function in conduit vessels.

In addition to endothelial vasomotor function, we measured release of the endogenous fibrinolytic enzyme, tissue plasminogen activator (t-PA). Release of t-PA may be more sensitive than vasodilatation as a marker of endothelial function (39). One-third of patients with acute coronary events undergo spontaneous reperfusion of the occluded vessel within 12 hours of symptom onset (40–42), and t-PA release is believed to be the mechanism underlying this phenomenon. Impaired t-PA release has previously been described in cigarette smokers, in hypertension, and after acute exposure to air pollution (43, 44). However, consistent with our assessment of endothelial vasomotor function, we found that bradykinin-induced release of t-PA was similar in patients with COPD and in matched control subjects. This again suggests that COPD does not confer additional endothelial dysfunction above that observed with smoking.

We found that COPD caused no impairment of endothelial vasomotor or fibrinolytic function in addition to any abnormality that may be caused by age and cigarette smoke exposure. However, COPD was associated with increased large elastic artery stiffness, having controlled for age and smoking. This suggests that arterial stiffness in COPD may be due to a structural defect in the extracellular matrix in the vascular wall rather than a functional deficit in the endothelium.

The development of arterial stiffness is a complex and incompletely understood process wherein endothelial and smooth muscle cells interact with the extracellular matrix to modify vessel wall structure and function (45). There are a number of mechanisms that may contribute to increased arterial stiffness in COPD. Our group previously reported that arterial stiffness was associated with emphysema severity, and proposed that this may represent a systemic susceptibility to connective tissue degradation (46). Others have proposed that there may be systemic susceptibility to elastin degradation in COPD. Lee and colleagues (47) demonstrated that subjects with emphysema have increased anti-elastin antibodies compared with subjects without emphysema. In addition, skin wrinkling (characterized by elastin breakdown in the skin) is also associated with CT emphysema severity, suggesting a common mechanism of lung and systemic elastin degradation in COPD (48).

Chronic systemic inflammation may be an important determinant of the increase in large arterial stiffness in COPD. Systemic inflammation is an important risk factor for cardiovascular disease (49) and has been implicated as a contributing factor to the increased cardiovascular risk associated with COPD (50). Furthermore, arterial stiffness is positively associated with CRP in healthy individuals (51) and circulating interleukin-6 levels are independently associated with pulse wave velocity in a COPD population (7). Although our study was not powered to examine associations between arterial stiffness and inflammatory variables, patients with COPD had higher circulating leukocytes and levels of CRP than control subjects.

There were also differences in both heart rate and arterial oxygen tension between the COPD patients and control subjects. Sympathetic activation and subclinical autonomic dysfunction are established features of COPD and have been associated with reduced arterial compliance (52). However, if autonomic dysfunction was the principle cause of increased arterial stiffness in COPD, we would expect to have observed differences in basal vascular tone and vasodilatation. Hypoxemia has variable effects on vascular function (53, 54) and within a COPD population it is difficult to separate the roles of hypoxemia and severity of lung disease on vascular function.

Aging is associated with endothelial dysfunction (55, 56), and our subjects had a mean age of 65 years. Therefore, the effects of age may have superseded any effect of COPD on endothelial activity. It is possible that studies of younger patients with COPD may have identified abnormalities in endothelial function. Furthermore, we were not powered to look at the effects of disease severity within the COPD group. Given our subjects had a mean FEV₁ percent predicted of 47% we believe it is unlikely that we would have missed an abnormality of endothelial function associated with mild to moderate COPD. It is possible that there may be abnormalities of endothelial function in patients with more severe disease.

Additionally, we limited our study to male patients. Although vascular function differs between the sexes, the difference almost disappears after menopause, which is believed to result from the loss of the protective effects of estrogens (55, 57–59). However, given apparent differences in the natural history of COPD between sexes (60, 61), caution must be exercised when extrapolating these results to women. We did not include an age-matched healthy lifelong nonsmoking control group to confirm the presence of endothelial dysfunction in the patient and matched control subjects. However, vascular dysfunction in smokers has been widely described (19–22). The aim of our study was not to replicate previous work but to establish whether vascular abnormalities are attributable to COPD independently of other confounding factors, such as smoking.

We have shown that men with COPD have significantly increased arterial stiffness, but no impairment of systemic vasomotor or fibrinolytic endothelial function in comparison with control subjects well matched for age and smoking. We therefore conclude that although abnormal endothelial function may be present in COPD, it is likely due to the effects of age and smoking, whereas increased arterial stiffness in COPD is independent of these factors. Increased arterial stiffness may represent the mechanistic link between COPD and increased risk of cardiovascular disease associated with this condition.