Abstract
The present study aimed to explore the prevalence of pre-capillary pulmonary hypertension (PH) and characterise haemodynamic vascular responses to physical exercise in chronic obstructive pulmonary disease (COPD) outpatients, where left ventricular dysfunction and comorbidities were excluded.
98 patients with COPD underwent right heart catheterisation at rest and during supine exercise. Mean pulmonary artery pressure (Ppa), pulmonary capillary wedge pressure (Ppcw) and cardiac output (CO) were measured at rest and during exercise. Exercise-induced increase in mean Ppa was interpreted relative to increase in blood flow, mean Ppa/CO, workload (W) and mean Ppa/W. Pulmonary vascular resistance (PVR) and pulmonary artery compliance (PAC) were calculated. PH at rest was defined as mean Ppa at rest ≥25 mmHg and Ppcw at rest <15 mmHg.
Prevalence of PH was 5%, 27% and 53% in Global Initiative for Chronic Obstructive Lung Disease stages II, III and IV, respectively. The absolute exercise-induced rise in mean Ppa did not differ between subjects with and without PH. Patients without PH showed similar abnormal haemodynamic responses to exercise as the PH group, with increased PVR, reduced PAC and steeper slopes for mean Ppa/CO and mean Ppa/W.
Exercise revealed abnormal physiological haemodynamic responses in the majority of the COPD patients. The future definition of PH on exercise in COPD should rely on the slope of mean Ppa related to cardiac output and workload rather than the absolute values of mean Ppa.
Pulmonary hypertension (PH) is a serious complication of chronic obstructive pulmonary disease (COPD). An increase in mean pulmonary artery pressure (Ppa) at rest of 10 mmHg is associated with a more than fourfold increase in mortality [1]. The actual prevalence of PH in COPD classified according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) is not well known [2].
Increased pulmonary artery stiffness is demonstrated early in the course of PH, even in patients where PH is detectable only during exercise [3]. The resistance and elasticity of the pulmonary artery play an important role in facilitating the transition from right ventricular pulsatile flow to nearly steady state flow at the capillary level. There is, however, little knowledge of how the pulmonary haemodynamics, measured by right heart catheterisation (RHC), are affected during exercise in stable COPD patients.
Accordingly, the present study aimed to explore the prevalence of pre-capillary PH in a cohort of stable COPD patients, where left ventricular (LV) disease and comorbidities were systematically excluded. Furthermore, and based on the fact that alternation of the elastic properties of pulmonary artery is observed even in patients without established PH, we hypothesised that abnormal haemodynamic vascular responses to exercise are also present in COPD patients without PH.
METHODS
Study population
The present cross-sectional study was performed at Oslo University Hospital, Oslo, Norway, between 2006 and 2010. 98 consecutive COPD patients were prospectively enrolled from our pulmonary outpatient clinic. Prior to inclusion all the participants were assessed by one of three pulmonary physicians (I. Skjørten, M.N. Melsom or S. Humerfelt) and classified by severity of airway obstruction according to GOLD. COPD medical therapy was standardised according to the GOLD statement [2]. None of the subjects had noninvasive positive-pressure ventilation support. Long-term treatment with oxygen was administered in eight patients and ambulatory oxygen support in four patients. 15% of the patients had chronic respiratory failure and, of those, four had hypercapnic respiratory insufficiency. The study was approved by the Local Research Ethics Committee, and all subjects gave written, informed consent. The study complies with the Declaration of Helsinki.
Inclusion and exclusion criteria
Norwegian Caucasian subjects, aged 40–75 years, with spirometrically confirmed COPD in GOLD stages II–IV, all either current or former smokers, with a smoking history of at least 10 pack-years, were included. They had to have been free of COPD exacerbations during the last 2 months prior to inclusion. All participants underwent pre-inclusion screening, including resting ECG and an exercise test on cycle ergometer to screen for potential ischaemic heart disease. In addition, all subjects were screened by echocardiography at rest and during exercise, and moderate or severe tricuspid regurgitation jets were not observed. Patients with the following: LV disease, treated arterial hypertension with blood pressure >160/90 mmHg, arrhythmias, intracardiac shunts, sleep apnoea syndrome, previous pulmonary embolism, other chronic pulmonary disease (pulmonary fibrosis or combined fibrosis/emphysema), malignancy, metabolic conditions (except metabolically stable diabetes), hyperthyroidism, systemic inflammatory diseases or renal failure with estimated glomerular filtration rate <60 mL·min−1, were excluded. Patients using β-blockers, warfarin or clopidogrel or who could not perform bicycle-exercise testing or the 6-min walk test for any reason, were also excluded.
Physical and pulmonary function testing
Standardised 6-min walk distance (6MWD) without supplemental oxygen was obtained. Forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) were determined by spirometry in accordance with international guidelines [4]. FEV1 and FVC predicted values were calculated [5]. GOLD classification was performed according to their best spirometric values and arterial oxygen tension (PaO2) and arterial carbon dioxide tension (PaCO2) on the test day after administration of their regular COPD medication. PaO2 and PaCO2 were obtained from the radial artery at rest and at maximal exercise breathing room air. Static lung volumes and diffusing capacity of the lung for carbon monoxide were measured [4, 6]. Patients performed a standardised incremental maximal exercise test (the cardiopulmonary exercise test (CPET)), which involved 4 min of unloaded pedalling and 4 min of pedalling with 25 W workload (W), followed by a progressive increase of 10 W every 2 min until exhaustion. The ratio of cardiac output (CO) to oxygen uptake (V′O2) was calculated as the change in CO from rest to exercise divided by the change in V′O2. All patients underwent high-resolution computed tomography (HRCT) to evaluate the presence of emphysema. Pulmonary function measurements, HRCT and CPET were obtained within 1 day of the haemodynamic measurements.
RHC and haemodynamic measurements
A balloon-tipped 7 F Swan–Ganz catheter was inserted into the antecubital or femoral vein and advanced into the pulmonary artery during short-time fluoroscopy (ArcoScope; Siemens, Munich, Germany). In the supine position, the pressures were zeroed at the mid-axillary line at the right atrial level, and automatic calibration was performed prior to pressure measurements. CO was measured by the thermodilution technique, averaging three or five output determinations. One CO measurement was performed at maximum workload. Mean Ppa and pulmonary capillary wedge pressure (Ppcw) were measured at rest and during the last minute of each exercise level, and right atrial pressure (Pra) at rest and at peak exercise, while right ventricular pressures were measured only at rest. Pressures were measured during temporary breathhold at end-expiration and verified by flat respiration curve. Post-processing analyses were performed on pressure curves at end-expiration at rest and during exercise by manually corrected region of interest if necessary. Computer-generated algorithmic of mean pressures (mean Ppa, mean Pra and mean Ppcw) were used, and were averaged over three to six cardiac cycles. The ECG was monitored continuously.
The following haemodynamic variables were used: transpulmonary gradient (TPG)=mean Ppa-Ppcw; pulse pressure (PP)=systolic Ppa-diastolic Ppa (mmHg); stroke volume (SV)=(CO/cardiac frequency (fC))×1000 (mL·beat−1); pulmonary artery compliance (PAC)=SV/PP (mL·mmHg−1); pulmonary vascular resistance (PVR)=(mean Ppa-Ppcw)/CO (Wood units) and total pulmonary vascular resistance (TPR)=mean Ppa/CO (Wood units). Exercise-induced increase in mean Ppa was interpreted relative to the increase in blood flow (mean Ppa/CO) and workload (mean Ppa/W).
Dynamic supine leg exercise using a cycle ergometer (ERGOMED 840L; Siemens) was performed. A steady-state resting period was followed by a stepwise increment in workload, starting with a 4-min period of unloaded exercise (0 W) at 60 rpm, followed by 20 W for 4 min, and then a 10-W increment every 2 min until exhaustion while breathing room air.
As unloaded exercise at 0 W results in energy expenditure and augmentation in V′O2 and CO with increase in mean Ppa, this “internal work” was added to the performed external work [7]. A correction factor was thus applied in all 98 participants, based on the known relationship between the increment in V′O2 and workload during incremental exercise (10 mL·W−1) [8, 9]. Thus, V′O2 was calculated for the workload of 0 W. The increment in V′O2 (mL·min−1) from rest to 0 W was then used to calculate the “internal workload” and added to their external work during exercise, denoted as corrected workload [7]. The baseline and exercise pressure signals, respiration curves and the ECGs were digitally recorded using a Mac-Lab application (GE Healthcare Medical Systems, Milwaukee, WI, USA).
Definition of haemodynamic groups
Patients were divided into the following two main groups according to resting pressures: 1) non-PH, where mean Ppa was <25 mmHg and Ppcw was ≤15 mmHg; and 2) established pre-capillary PH, where mean Ppa was ≥25 mmHg and Ppcw was ≤15 mmHg [10]. A PVR >1.5 Wood units at rest (corresponding to upper limit of the 95% confidence interval for healthy controls (mean±1.96 sd)) was considered elevated in subjects aged >50 years [11].
Statistical analysis
The results of continuous variables are reported as mean±sd or mean±sem. Categorical variables are expressed as frequencies or percentages. An independent t-test was used to compare the mean of two different groups, and a paired t-test used to compare mean differences between rest and exercise within the same group. A p-value <0.05 were considered statistically significant. For categorical variables, the Chi-squared test was used. Non-normally distributed data were log-transformed (N-terminal pro-brain natriureteric peptide (NT-proBNP)). Multivariate linear regression was used to determine whether higher resting mean Ppa and exercise PVR were associated with a shorter 6MWD, independent of potential confounders. The statistical analyses were performed using SPSS version 15 (SPSS Inc., Chicago, IL, USA) and SigmaPlot version 12.0 (Systat Software, Inc., London, UK).
RESULTS
Demographic characteristics
Demographic and pulmonary function characteristics are listed in table 1. 33 (34%) patients of the study population were treated for essential systemic hypertension. No differences in systolic or diastolic blood pressure were observed between those treated for and those without hypertension. Emphysema was diagnosed in all except three patients. These three did not differ from the rest with respect to haemodynamic data; however, the group was too small to make any conclusion. Respiratory failure (defined as PaO2 <8 kPa) was observed in eight patients, all in GOLD stage IV and on chronic oxygen treatment. Four patients were on ambulatory oxygen, two each in GOLD stage III and IV, respectively. The oxygen-treated group comprised seven patients with PH (mean Ppa at rest 30±5, range 25–40 mmHg and PaO2 7.5±1.8 kPa), and five without (mean Ppa at rest 22±1, range 20–23 mmHg and PaO2 7.9±0.7 kPa).
Haemodynamic profile at rest
The haemodynamic variables and arterial blood gas profile in the two groups are summarised in table 2. The prevalence of PH was 27% (26 patients) and of non-PH was 73% (72 patients). Distribution of mean Ppa related to GOLD stages is shown in figure 1. PVR was elevated at rest in all patients with PH and in 50 (69%) patients in the non-PH group. PAC was significantly reduced in the PH group (p<0.01) compared with the non-PH group (table 2). Right ventricular function was normal at rest in both groups according to CO and Pra (table 2), and none had elevated Ppcw.
Haemodynamic profile at peak exercise
The relationship between CO and workload is illustrated in figure 2. A CO increase of 68% and 100% in the PH and non-PH groups was associated with a 66% and 105% increase in mean Ppa, respectively. There was no significant difference in the absolute mean Ppa increase in response to maximal exercise between the groups (fig. 3).
PVR increased significantly by 11% in the PH group during exercise (p=0.04). In the non-PH group, a nonsignificant increase of 5% in PVR was observed (p=0.08). However, considering percentage change in PVR from rest to exercise, adjusted for changes in CO, a similar increase in PVR of 6.3% in non-PH group as in PH group (6.9%) was observed (p=0.9). For both groups, TPR was significantly increased from rest to exercise (p<0.01 for both); however, there were no differences in TPR increases between the groups (p>0.8) (fig. 4).
The increase in PP from rest to exercise exceeded the increase in SV. Hence, PAC decreased in response to exercise by 47% and 44% in PH and non-PH, respectively (table 3). However, the increase in PVR was accompanied by a relatively larger drop in PAC in the non-PH group compared with the PH group (PVR/PAC ratio 0.6±0.4 versus 1.5±1.4, respectively; p<0.01). Adjusted for workload, Ppcw showed a similar increase in response to exercise in both groups. 14 patients (four and 10 in the PH and non-PH groups, respectively) showed a more pronounced Ppcw response to exercise than the rest (average maximal Ppcw 23±2, range 21–26 mmHg). In table 3, the slopes related to workload and CO for haemodynamic parameters are summarised. figure 5 illustrates how all 98 change (Δ) in mean Ppa/ΔCO slopes are related to mean Ppa at rest in non-PH and PH groups. There was only a weak correlation between mean Ppa at rest and Δmean Ppa/ΔCO (r=0.2; p<0.05).
Contribution of Ppcw and TPG to increase in mean Ppa by exercise
The ΔPpcw/ΔTPG ratio in the PH group was 0.5, and 0.8 in the non-PH group, reflecting a larger contribution of TPG than Ppcw to mean Ppa by exercise. The relative changes in mean Ppa, Ppcw, TPG, PVR and TPR from rest to exercise are shown in figure 4, and the slopes are presented in table 3.
Changes in blood gases by exercise and their relationship with haemodynamics
From rest to peak exercise a decline in relative arterial oxygen saturation (SaO2) of 2±3 and 8±7% was observed in non-PH and PH groups, respectively (p<0.01). The relative SaO2 decline was significantly lower for the non-PH group (0.06±0.1 % per W) than for the PH group (0.3±0.4 % per W; p<0.01), which showed a fivefold larger decline in SaO2 when related to maximum workload. During peak exercise, mean Ppa was negatively correlated to maximal PaO2 (r= -0.5; p<0.01) in the entire cohort. The non-PH group, however, showed only a weak negative correlation between mean maximal Ppa and maximal PaO2 (r= -0.4; p<0.01), while the PH group did not correlate at all (r=0.2; nonsignificant). PVR rest did not correlate with PaO2 at rest in the non-PH group, while in the PH group, a strong negative correlation between PVR at rest and PaO2 at rest (r= -0.6; p<0.01) was observed. PVR during peak load did correlate with maximal PaO2 in both groups (r= -0.5; p<0.01).
Functional outcomes
Adjusted for age, sex, height, weight, FEV1 and Ppcw, mean Ppa at rest was negatively associated with 6MWD (r= -0.55; p<0.01). A 9.5-m decline in 6MWD for every 1-mmHg increase in mean Ppa (95% CI -14.3– -4.5 m; p<0.01) was observed. Adjusted for the same confounding variables in addition to SaO2 measured by pulse oximetry, PVR was inversely related to 6MWD (r= -0.6; p<0.01). This multivariable linear regression model with 6MWD as dependent variable showed that for every Wood unit increase in PVR at peak exercise, the 6MWD dropped by 29.5 m (95% CI -48.9– -10.1 m; p<0.01). Furthermore, maximal workload was strongly correlated to PVR and PAC at peak workload (r= -0.7 and r=0.5, respectively; p<0.01 for both).
30 COPD patients showed a Δmean Ppa/ΔCO <3 mmHg·L−1·min−1 (2.1±0.6, range 1.05–2.92 mmHg·L−1·min−1), representing “true circulatory normal” (fig. 6).
DISCUSSION
In the present cohort of COPD patients, recruited from an outpatient population where LV dysfunction, comorbidities and exacerbations were thoroughly excluded, a prevalence of pre-capillary PH of 27% was found. The prevalence related to GOLD stage II, III and IV was 5%, 27% and 53%, respectively. As expected, abnormal exercise-induced haemodynamic responses were observed in the PH group. However, similar and comparable haemodynamic responses to exercise were also demonstrated in non-PH subjects. Based on these results and previous findings in healthy subjects, we consider the responses to exercise in non-PH subjects as abnormal.
Mean Ppa at rest
The majority of prevalence data are based on patients with severe to very severe COPD disease, i.e. awaiting lung transplantation, where RHC is a part of routine clinical evaluation. Previous studies have reported prevalence of PH in COPD patients to vary between 30% and 90% [15–19]. The inconsistency in prevalence of PH associated with COPD is based mainly upon dissimilarities in definition of PH, methods used to determine mean Ppa, the physiological characteristics of the underlying lung disease and patient population examined [20]. In the present study, great effort was made to ensure inclusion of COPD patients who were clinically stable and optimally treated and without evidence of LV dysfunction either at rest or during exercise. Our prospectively collected data showed a lower prevalence of pre-capillary PH than in the majority of previous studies. Only 27% of our patients met the current definition of PH (mean Ppa ≥25 mmHg) [10]. This lower prevalence compared with published estimates of PH reflects the present study's more restrictive definition and, secondly, a very strict selection process to avoid including comorbidities (i.e. both post- and other pre-capillary diseases) that could contribute to a rise in mean Ppa. Using the previous definitions of PH (mean Ppa >20 mmHg), the prevalence of pre-capillary PH would have been 46% [19]. The average 18 mmHg resting mean Ppa value in the non-PH group in the present study is higher than that reported for healthy subjects [13], but in accordance with the study in 1999 by Kessler et al. [21] on stable outpatients with COPD.
Mean Ppa response to increased W and CO
Workload and CO had a similar impact on the pressure increase of mean Ppa. The absolute increase in mean Ppa by exercise was also similar in those with and without PH. However, there was a larger increase in mean Ppa in the PH compared with non-PH group, which was related to the increase in workload and CO. Thus, our results clearly support that exercise-induced rise in mean Ppa should be interpreted relative to increase in workload or increase in CO, rather than by evaluating a single threshold or an absolute peak exercise value of mean Ppa, as highlighted by Saggar et al. [22]. In this context, we are in agreement with the decision of the Working Group on Diagnosis and Assessment of Pulmonary Arterial Hypertension that, in 2008, withdrew exercise mean Ppa >30 mmHg as a diagnostic criteria for PH.
Comparable haemodynamic data in healthy controls are limited due to ethical reasons, but some data have been reported, albeit with small numbers of subjects in all studies [7, 12, 13]. Lewis et al. [7] reported a Δmean Ppa/ΔCO slope of 1.4 mmHg·L−1·min (age 60±12 years), and in a slightly younger cohort (age 41±4.8 years); Degre et al. [12] demonstrated 1.5 mmHg·L−1. In a study by Kovacs et al. [13] on healthy controls aged >50 years, a Δmean Ppa/ΔCO of 2.8 mmHg·L−1·min was demonstrated, and in a study of healthy older males (age 71±6 years), Granath et al. [23] reported a slope of 2.5±0.8 mmHg·L−1·min. In the present study, we have shown slopes of 7.2 and 4.6 mmHg·L−1·min in PH and non-PH groups, respectively. Compared with the previous studies on healthy controls, our data indicate a pathological mean Ppa response to exercise also exists in a large proportion of the COPD patients without PH (fig. 5). A cut-off point for mean Ppa/CO of >3 mmHg·L−1·min, as proposed by Lewis [14], should reinforce this notion. Moreover, in contrast to mean Ppa at rest, mean Ppa during exercise has been shown to be strongly related to pulmonary arterial wall thickness in COPD; the authors suggested that this reflects reduced distensability and recruitability of pulmonary vessels in COPD, and that the degree of remodelling could not be estimated by mean Ppa at rest [24]. Our study also confirmed a poor relationship between mean Ppa at rest and Δmean Ppa/ΔCO.
Contribution of TPG and Ppcw to the exercise-induced increment in mean Ppa
Previous haemodynamic studies in healthy subjects have demonstrated a greater contribution of ΔPpcw compared with ΔTPG to exercise-induced increase in mean Ppa with a ratio of ∼2 [25–27]. In our study, an inverse ratio for both non-PH and PH groups was observed, consistent with a larger contribution of ΔTPG to the increase in mean Ppa compared with ΔPpcw, reflecting pre-capillary pathology in the non-PH and PH groups. In both groups, this increase of ΔTPG was related to pathological PVR response. In healthy subjects, PVR and also TPR are normally slightly reduced during exercise [11], probably due to passive recruitment and distension of a compliant pulmonary circulation and/or an active flow-mediated vasodilatation. Our patients without PH showed the same PVR and TPR response pattern as the PH group, indicative of early and significant pre-capillary vascular changes also existing in this group. Furthermore, during exercise, the non-PH group had a considerably larger drop in compliance relative to the change in PVR compared with the PH group, which is consistent with the findings of Saouti et al. [28], who suggested that a small increase in PVR accompanied by a relatively larger drop in PAC to be a hallmark of early changes in the pulmonary vascular bed.
Exercise decline in PaO2 and haemodynamics
In the present study, only a minor exercise-induced decline in PaO2 was observed, which is in contrast to data presented by Boerrigter et al. [29], but is comparable with the observations by Christensen et al. [30]. Augmented pressure responses during exercise partly occurred in our non-PH population in the presence of minor decline in Pa,O2 during exercise, and could indicate a mechanism other than hypoxaemia.
Clinical implications
GOLD classification is valuable in terms of differentiation of airway disease severity [2] but, in our study, showed a poor relationship with mean Ppa at rest (fig. 1). Its use as a predictor for PH is thus restricted. Furthermore, the results demonstrated that patients classified as GOLD III had mean Ppa distribution from low normal (mean Ppa 10 mmHg) to severe PH (mean Ppa 38 mmHg). A similar pattern was shown for GOLD stages II and IV, and half of the patients in GOLD stage IV did not have PH according to current guidelines (fig. 1).
The present study has also demonstrated that higher resting mean Ppa is associated with impaired functional capacity (6MWD) independent of demographics, Ppcw and GOLD classification. To our knowledge, this study is the first to report reduced functional capacity related to higher resting mean Ppa in an outpatient population. Similar findings have been reported in severe COPD patients listed for lung transplantation [31]. Interestingly, 6MWD, in our cohort decreased gradually by increasing PVR at peak exercise, even when controlling for confounding variables. A strong correlation between exercise afterload variables, PVR and PAC, and maximum workload was observed, and could further endorse the clinical relevance of exercise-induced PH. Patients with exercise-induced PH are particularly prone to developing persistent PH in the long term, as exercise mean Ppa >30 mmHg is a marker of disease progression [32]. Thus, the present non-PH group, and especially those with Δmean Ppa/ΔCO >3 mmHg·L−1·min, should be followed up longitudinally to see if they develop PH.
Study limitations
The local ethical committee in Norway does not approve invasive RHCs on healthy controls and, therefore, the present study did not provide haemodynamic data for a healthy age- and sex-matched control group. However, comparable data on haemodynamic exercise responses in historical controls have been published, albeit with a limited numbers of articles in the population aged >50 years.
14% of the patients showed exercise Ppcw >20 mmHg, and hence a post-capillary contribution to elevated mean Ppa during exercise cannot be entirely excluded. An upper limit for Ppcw was not adopted due to the paucity of data supporting the previously used cut-off point of 20 mmHg in a population aged >50 years. CO was measured by the thermodilution technique only once at maximal exercise at highest fC. Rapid decline in both Ppa and fC were observed when exercise was discontinued, and several CO measurements would thus have underestimated peak CO.
CPET was performed separately 24 h prior to RHC, which reflects the standard in our laboratory. A concomitant performance of these two procedures would have been preferable. It is, however, reasonable to assume that clinical and haemodynamic conditions did not differ in this short period of time, and the exercise levels were similar during CPET and RHC.
Upright positioning during exercise most closely mimics normal physical activity. However, the RHC exercise test in this study was performed in the supine position due to more stable and reliable pressure curves in this position. Despite the postural impact on haemodynamic variables, the relative changes between rest and exercise haemodynamics are most likely to be comparable.
It has been postulated that dynamic lung hyperinflation in COPD may contribute to the development of exercise-induced PH. At the time of inclusion, measurements of inspiratory capacity were not routinely performed in our exercise laboratory and, therefore, it is difficult to conclude with certainty that the exaggerated response of Δmean Ppa/ΔCO and PVR in COPD compared with historical controls is due solely to a vasculopathy.
Our patients represent a well-selected group of patients from an outpatient population. The main purpose was to study exclusively pre-capillary prevalence of PH and pre-capillary haemodynamic responses to exercise. Therefore, our findings are not necessarily applicable to all COPD patients.
Conclusions
The present study showed a prevalence of PH of 5%, 27% and 53% in a cohort of COPD outpatients in GOLD stage II, III and IV, respectively. Exercise revealed abnormal haemodynamic responses in the PH group. However, similar and comparable haemodynamic response patterns were also observed in a large section of the non-PH group. Resting haemodynamics, however, were inadequate to identify early pulmonary vascular disease. The future definition of PH on exercise should rely on the slope of mean Ppa, related to CO or workload, rather than the absolute values of mean Ppa, to account for the large interindividual variability in physical performance in COPD patients.
Footnotes
For editorial comments see page 1002.
Support Statement
Funding was supplied by Eastern Norway Regional Health Authority and Dept of Cardiology, Oslo University Hospital, Aker, Norway (grant number 2007006).
Statement of Interest
None declared.
- Received May 31, 2012.
- Accepted July 25, 2012.
- ©ERS 2013