Echocardiographic Screening for Pulmonary Hypertension in Stable COPD Out-Patients and NT-proBNP as a Rule-Out Test

Abstract

Pulmonary hypertension (PH) worsens the prognosis in chronic obstructive pulmonary disease (COPD). The diagnosis of PH is established by right heart catheterisation (RHC), while echocardiography can be used for screening. We aimed to asses the outcome of echocardiographic screening for PH in a group of stable COPD out-patients, and to evaluate NT-proBNP as a first line screening tool. Criteria for PH on echocardiography were a tricuspid regurgitation pressure gradient > 40 mmHg, a tricuspid annular plane systolic excursion < 1.8 cm or right ventricular dilatation. Positively screened patients were asked to undergo RHC. Results (Mean ± SEM): 16 of 117 patients (14%) had PH on echocardiography. They had a higher mortality (hazard ratio for death: 2.7 ± 1.3, p = 0.037) and lower six minute walk test (224 ± 33 vs. 339 ± 15, p = 0.006). NT-proBNP below 95 ng/l excluded PH on echocardiography with a negative predictive value of 100 (95% CI: 89–100%). RHC was obtained in six patients screened positive. In three of these, PH was not confirmed. Conclusions: Signs of PH on echocardiography as defined here was found in 14% and had prognostic significance in COPD. A value of NT-proBNP less than 95 ng/l may be used to exclude signs of PH.

Keywords :

• biomarkers
• BNP
• chronic lung disease
• mortality
• 6-six minute walk test.

Introduction

Chronic obstructive pulmonary disease (COPD) is common and a leading cause of death in the Western World (1). COPD can be complicated by pulmonary hypertension (PH) which is important to diagnose, because it worsens the prognosis (2,3,4,5,6) and increases the risk of exacerbations (7).

In previous studies, the prevalence of PH in COPD varies between 20 to 91% (8,9,10,11,12,13). The studies are predominated by older (9,10,11), or retrospective studies(8,14,15,12,13), or studies performed in lung transplant candidates (14,16,13). Furthermore, a limit of a mean pulmonary artery pressure (MPAP) of 20 mmHg (9,10,11,12) or a systolic pulmonary arterial pressure (SPAP) ≥ 36 mmHg estimated by echocardiography (15) has been used to define the presence of PH in part of them. Thus, the prevalence of PH in COPD is unclear.

Challenges in investigating the prevalence and impact of PH in COPD are a large number of patients, and the invasive character of right heart catheterisation (RHC), which is the gold standard for diagnosing of PH, combined with the lack of evidence for intervention targeting PH (17). Echocardiography, with estimation of the SPAP and right ventricular function (18), remains the best non-invasive means of screening for PH (17), but it is subject to inaccuracies (19), and so far, it is not known if echocardiographic screening in COPD out-patients yields prognostic information.

Furthermore, there are currently no specific plasma biomarkers to help in the diagnosis of PH in COPD. Brain natriuretic peptide (BNP) is associated to PH in chronic lung disease (20), and the inactive split product NT-proBNP has been proven as a marker for pulmonary arterial hypertension (PAH)(21).

The aim of the present study was to asses the outcome and prognostic significance of echocardiographic screening for PH in a group of stable COPD out-patients, by using the criteria for PH that we use in daily practice at our hospital. Furthermore, we aimed to evaluate NT-proBNP as a first line screening tool.

Methods

The study was approved by The Central Denmark Region Committees on Biomedical Research Ethics (Issue no: M-20070206), registered at clinical trials.gov (ClinicalTrials.gov Identifier: NCT00614900) and conducted in accordance with the Helsinki Declaration.

Study subjects

Study subjects were recruited from March 2008 to March 2011. All patients referred to the out-patient clinic at Aarhus University Hospital after admission due to an exacerbation of COPD were considered potential participants and were asked to participate. Inclusion criteria were: age > 18 years, written consent, a diagnosis of COPD according to the GOLD criteria, and ≥ one hospitalisation due to an exacerbation of COPD. Exclusion criteria: exacerbation leading to treatment with intramuscular β₂-adrenoceptor agonist, systemic corticosteroids or antibiotics within the last 6 weeks of inclusion.

Participants in the study were followed up for mortality until February 12^th 2012. Based on previous studies indicating a prevalence of PH of 20–91% in COPD, we estimated that we would need 100 patients in order to identify more than 20 patients with indices of PH on echocardiography to perform a meaningful analysis between patients with and without PH on echocardiography. For example, to detect a clinically relevant difference of 50 m with a standard error of 50 m in the six minute walk test (6MWT) between groups with a power of 80% and a significance level of 0.05, 16 patients in each group would be needed based on the following calculation: n (each group) = (C_2α + C_β)² × 2 x S²/Δ² = (1,96 + 0,841) ² · 2 · 50² / 50² = 16, C_2α and C_β being the 2α fractile and β-fractile in the t-distribution, S is the standard deviation of the improvement, and Δ is the expected improvement.

Study protocol

All participants had an extra consultation by the authors for the purpose of the study during which informed, written consent was obtained. Interview, spirometry, echocardiography, 6-minute walk test (6MWT) and venous sampling for NT-proBNP measurements were performed. In a subset of patients, spirometry and 6MWT had been performed within four weeks of the consultation, and if the patients reported a stable level of symptoms, the values from these tests were used. At the end of the consultation, patients screened positive for PH on echocardiography were asked to undergo RHC. Advantages and risks were discussed with them, and they were given time to consider the offer.

Echocardiography

The velocity of the tricuspid regurgitation jet was measured by Doppler echocardiography in multiple projections, and the tricuspid pressure regurgitation gradient (TR) was calculated from the modified Bernouilli equation: TR = 4v² (v = peak velocity of tricuspid regurgitation, m/second). The dimension of each ventricle was evaluated from standard projections. The tricuspid annular plane systolic excursion (TAPSE) was assessed in the four chamber apical window with the M-mode cursor through the lateral tricuspid annulus ring. Respiratory compression of the inferior caval vein was evaluated from the right subcostal projection. Criteria for a positive screen for PH on echocardiography were: TR > 40 mmHg, right ventricular dilatation, or decreased TAPSE (<1.8 cm). If acoustic windows were too poor to measure at least one of these parameters, patients were excluded from further analysis.

RHC

Patients screened positive for PH on echocardiography were asked to undergo RHC performed with a Swan-Gantz catheter for measurements of MPAP, pulmonary capillary wedge pressure (PCWP) and cardiac output.

Mortality

Survival status was assessed using the electronic patient record system at Aarhus University Hospital in which all deaths in Denmark are registered on a week-to-week basis.

Spirometry

Forced expiratory volume in 1 second (FEV₁) and forced vital capacity (FVC) were measured by spirometry (Vitalograph compact, Vitalograph, Lenaxa, USA). Predicted values were calculated as recommended by the European Respiratory Society (22).

6MWT

6MWT was performed following guidelines from American Thoracic Society (23). The tests were conducted on a 40 m track using prescribed supplemental oxygen. Borg dyspnoea score, peripheral saturation and pulse rate were recorded before and immediately after the 6MWT.

NT-proBNP

Venous blood samples were analysed routinely for NT-proBNP (cobas 6000 E/C, Roche, West Succex, England) at the hospital's CME-accredited laboratory. The normal upper reference limit is 300 ng/l.

Use of medication and concomitant cardiovascular disease

The use of COPD medication and the presence of concomitant cardiovascular disease were assessed by interview and review of the patients’ records.

Data analysis

Data were analysed in Stata/IC 10 (StataCorp, College Station, Texas USA). Survival was analysed with time since inclusion as time scale. Follow-up was censored 10 months after ceasing recruitment. Mortality was estimated by Kaplan-Meier curves and differences between patient groups by the Cox-proportion hazard model. Proportionality was validated by log-log plots.

Parametric data were analysed using Student's t-test, non-parametric data by rank sum test, and proportions by proportion test. NT-proBNP values were log-transformed to obtain a normal distribution, and receiver operated characteristics analyses were used to evaluate the diagnostic value for PH on echocardiography.

Results from data following a normal distribution are expressed as means ± standard error of the mean (SEM), non-parametric data are expressed as median with 95% confidence intervals (CI), sensititivity, specificity, positive predictive values (PPV), negative predictive values (NPV) and hazard ratios for death (HR) are expressed with 95% CI, and proportions are expressed as% of patients.

A p-value below 0.05 was regarded as statistically significant.

Results

Patient demographics

A total of 117 patients agreed to participate in the study. During the period of enrolment, different strategies for referral to the out-patient clinic were followed and the precise number of potential participants was not available to us. Approximately 500 patients were invited, and the acceptance rate was thus less than 25%. Unfortunately, we were not able to obtain demographic data from patients not entering the study.

Demographic data from participants are summarised in Table 1. Patients were treated with standard COPD treatment resembling regimes listed in guidelines from global initiative for chronic obstructive lung disease(1) (Table 1).

Table 1.  Patient demographics

	All Pts (Total n = 117)	n	No PH (Total n = 97)	n	PH (Total n = 16)	n	P (PH vs No PH)
Age (years)	70 ± 1	117	69 ±1	97	75 ± 2 *	16	0.01
BMI (kg/m2)	25 ± 0.6	113	25 ± 1	93	25 ± 1	16	0.8
Female (% of pts)	61	117	64	97	50	16	0.3
Years since diagnosis (median (95% CI))	8 (5–10)	107	8 (5–10)	90	9 (5-12)	13	0.3
Pack-years	43 ± 3	109	43 ± 3	96	40 ± 6	15	0.7
Spirometry (% of expected)
FEV1	42 ± 2	116	43 ± 2	96	37 ± 4	16	0.2
FVC	72 ± 2	116	73 ± 2	96	63 ± 5*	16	0.04
COPD Class (% of pts)
Mild	2	117	2	97	0	16	0.6
Moderate	30	117	31	97	25	16	0.6
Severe	44	117	44	97	38	16	0.6
Very severe	25	117	23	97	38	16	0.2
COPD treatment (% of pts)
SABA	75	117	73	97	88	16	0.2
Ipratropium	17	117	16	97	25	16	0.4
LABA	86	117	86	97	88	16	0.8
Tiotropium	68	117	69	97	62	16	0.6
Inhaled glucocorticoids	86	117	85	97	94	16	0.3
Glucocorticoids p.o.	10	117	11	97	6	16	0.5
Theofylline	3	117	2	97	13*	16	0.03
Oxygen treatment	8	117	5	97	19 *	16	0.0495
Cardiovascular disease (% of pts)
Any cardiovascular disease	42	117	38	97	69*	16	0.02
Atrial fibrillation	10	117	7	97	31*	16	0.003
Ischemic heart disease	17	117	15	97	25	16	0.6
Former myocardial infarct	14	117	13	97	19	16	0.6
Left ventricular failure	6	117	5	97	13	16	0.25
Hypertension	35	116	34	96	50	16	0.2

Results are expressed as median with 95% CI, means ± SEM or proportions (e.g., 61% of patients were female). *p < 0.05 vs no PH. Total n indicates the number of patients in each group. n indicates the number of patients from whom data were available. Data for the 4 patients in which echocardiography was inconclusive are included in the data from all patients. No PH: No signs of PH on echocardiography. PH: patients screened positive by echocardiography. SABA: Short-acting β-adrenoceptor agonists, LABA: long-acting β-adrenoceptor agonists. Pts: Patients. Any cardiovascular disease describes the presence of one or more of the specific diseases listed.

Echocardiography

Values for TR were obtained in 86 and for TAPSE in 108 patients. Sixteen patients (14%) were screened positive for PH using the criteria defined for the present study (TR > 40 mmHg, TAPSE < 1.8 cm or right ventricular dilatation). In four, the echocardiogram was inconclusive. In patients screened negative, there were no significant differences in lung function, 6MWT and TAPSE between patients with and without a measurable TR (data not shown). Patients screened positive had a TR of 44 ± 2 mmHg and a TAPSE of 1.90 ± 0.02 cm. In patients screened negative TR was 21 ± 1 mmHg and TAPSE was 2.30 ± 0.03 cm.

RHC

RHC data from 6 patients screened positive for PH by echocardiography were obtained. In 3 of these, the diagnosis of PH was discarded, in 2 because of a MPAP < 25 mmHg (16 and 22 mmHg, respectively), and in 1 because of an abnormally high PCWP (MPAP = 32 and PCWP = 21). The time from echocardiography to RHC was 71, 75 and 101 days in these 3 patients. The three patients with COPD-related PH had a MPAP of 25, 26 and 28 mmHg, respectively. The time from echocardiography to RHC was 88, 110 and 188 days. Ten patients did not undergo RHC (1 died, 3 were considered ineligible due to their age and clinical condition, and 6 refused). These 10 patients had a TR of 43 ± 1.5 mmHg.

Demographic differences between patients without and with PH on echo

Compared to patients with no PH on echocardiography, patients screened positive were older, had lower FVC and were more likely to receive theophylline and long-term oxygen treatment. There were no differences with regard to BMI, disease duration or pack years (Table 1).

In patients with PH, cardiovascular disease in general and atrial fibrillation occurred more frequently than in non-PH patients, whereas ischemic heart disease, former myocardial infarct, left ventricular failure and hypertension did not differ in prevalence (Table 1).

Mortality

Median follow-up time from inclusion was 2.8 (95% CI: 2.0-3.0) years. Twenty-three deaths occurred. Survival curves for patients screened negative and positive for PH on echocardiography are depicted in Figure 1. The HR in the PH-positive group vs. the PH-negative group was 2.7 (1.06-6.9), p = 0.037. After adjusting for age HR was 1.7 (95% CI: 0.65-4.9, p = 0.26)

Figure 1.  Mortality. Kaplan-Meier survival curves for patients without and with PH on echocardiography defined as a tricuspid regurgitation pressure gradient > 40 mmHg, a tricuspid annular plane systolic excursion < 1.8 cm or right ventricular dilatation. * p < 0.05 vs. no PH on echocardiography.

6MWT

Compared to patients without signs of PH on echocardiography, 6MWT was significantly shorter in those screened positive (Table 2). PH patients also had a lower peripheral saturation and higher Borg dyspnoea score at rest, while these two parameters were not significantly different after the 6MWT (Table 2).

Table 2.  Six-minute walk test

	No PH (Total n = 97)	n	PH (Total n = 16)	n	P (PH vs No PH)
[spanname = “1to6”]6-minute walk test
Distance (m)	339 ± 15	88	224 ± 31 *	14	0.006
Saturation before (%)	94 ± 3	93	91 ± 2 *	15	0.01
Saturation after (%)	90 ± 0.6	88	87 ± 3	12	0.09
Borg score before	1.8 ± 0.2	85	3.5 ± 1.0 *	13	0.004
Borg score after	4.7 ± 0.3	82	5.5 ± 0.8	13	0.3

Results are expressed as mean ± SEM.*: p < 0.05 vs no PH. Total n indicates the number of patients in each group. n indicates the number of patients from whom data were available. No PH: No signs of PH on echocardiography. PH: patients screened positive by echocardiography.

NT-proBNP as a predictor of echocardiographic findings of PH on echocardiography

The area under the ROC curve for NT-proBNP and a positive screen for PH on echocardiography was 0.83 ± 0.05 (Figure 2A), and the sensitivity, specificity, NPV and PPV for PH of a cut-off value of 95 ng/l (given with 95% CI) was 100 (80–100)%, 36 (26-47)%, 22 (13–33)% and 100 (89–100%), respectively. 29 ± 4% of patients had NT-proBNP values below 95 ng/l (Figure 2B). Patients with NT-proBNP values below 95 ng/l had a decreased mortality compared to those with values above this limit (HR: 0.29 (0.09–0.97), p = 0.04).

Figure 2.  NT-proBNP as a first line screening tool. A: Receiver operator characteristics plot for NT-proBNP and the presence of PH on echocardiography. B: Dotplot of NT-proBNP in patients with and without PH on echocardiography. Dashed line indicates cut-off value used for calculation of sensitivity, specificity, positive and negative predictive values. Error bars indicate means ± SEM. * p<0.05 vs. no PH on echocardiography.

Discussion

The main findings of the present study were that echocardiographic signs of PH defined by a TR > 40 mmHg, TAPSE < 1.8 cm or right ventricular dilatation occurred in 14% of stable COPD outpatients formerly admitted to hospital with an acute exacerbation. Those screened positive for PH were older, had lower FVCs, a higher occurrence of atrial fibrillation, a higher mortality and a shorter 6MWT distance. Furthermore, a value of NT-proBNP below 95 ng/l excluded the presence of PH on echocardiography with a high NPV.

Echocardiographic screening for PH

The ability of echocardiography to distinguish between the presence and absence of PH has been evaluated by Arcasoy et al. in 374 lung transplant patients with COPD or ILD (19). They found that an echo-estimated SPAP of ≥ 45 mmHg had a PPV and NPV for PH established on RHC of 52% and 87%, respectively. In the same study, right ventricular dilatation and hypertrophy, were also assessed, and 96% of patients with no measurable TR, and a normal right ventricle, did not have PH (19).

As estimations of the right atrial pressure is often inaccurate (24), and because of the low PPV of SPAP ≥ 45 mmHg, we chose to rely in the present study on the TR of 40 mmHg combined with right ventricular dilatation or systolic function, which we generally consider indicative of PH at our hospital. No studies have evaluated the predictive values of TR, right ventricular abnormalities and TAPSE in patients with lung disease, but it is plausible that the combination of the parameters as used in the present study would yield a higher NPV for PH.

It is also important to notice that the NPV is dependent on the prevalence of PH in the cohort examined. In lung transplant candidates the prevalence of PH may be significantly higher than in cohorts of out-patients, and this would result in a lower NPV of echocardiography for PH in lung transplant candidates than in out-patients. Altogether, we believe that screening by echocardiography with the criteria used in the present study, is a fairly safe method for exclusion of PH in our population.

Prevalence of PH

A low PPV of echocardiography to detect PH has been observed previously,(19) and this was confirmed in our study by the 3 out of 6 positively screened patients in whom PH was not confirmed by RHC. Consequently, we are not able to establish the exact prevalence of PH in the present cohort, but can only speculate that it is somewhere below (16-3)/113 equal to less than 11.5%, even though it cannot be excluded that some cases of PH may have been missed using the criteria defining PH in the present study. This occurrence of PH is generally lower than reported by previous studies.

In older studies (9,10,11), the prevalence of PH in stable COPD patients was 20-35%. In at least one study (10), the patients’ degree of airway obstruction measured by FEV₁ was similar to that in the present study. Possible explanations for the differences in prevalence are that PH was defined by a MPAP of 20 mmHg in the prior studies, and that current guidelines for managing COPD are different. The influence of COPD treatment on the development of PH is not clear.

In other studies conducted in COPD lung transplant or lung resection candidates (14,12,16,13), the prevalence is also higher than in the present study, and this is probably due to the presence of more severe COPD, indicated by mean FEV₁ values of approximately 25% of expected or less in these studies.

In a recent study by Fayngersh et al. (15) stable outpatients with mean FEV₁ of 52–63% of expected were evaluated retrospectively and included in the study if a measurable TR was demonstrated on echocardiography. The criterion for PH was an SPAP ≥ 36 mmHg, and the prevalence of patients fulfilling this criterion was 60%. However, there is a risk of selection bias if the echocardiography was performed mainly in patients suspected for PH or other cardiac disease, and furthermore, a measurable TR naturally occurs more frequently in patients with PH, which may also overestimate the prevalence of PH.

The large difference between the results from the present study and that from Fayngersh et al. illustrates that study design is of major importance when comparing prevalence of PH between studies. We deliberately chose to include patients regardless of concomitant heart diseases. As these are common in COPD (1), we find that this approach may give a good picture of the outcome of echocardiographic screening in out-clinic COPD patients. However, our study may be biased by the low rate of patients accepting to participate. Unfortunately, we were not able to compare demographics between potential and actual participants, which is a limitation of the present study.

Prognostic value of echocardiographic screening

It was observed that echocardiographic screening using the criteria defined above was useful for risk stratification of the patients in this cohort, because those screened positive had an increased mortality and a reduced exercise capacity. This implies that even though not all patients with a positive screen for PH on echocardiography had true PH, the prevalence of PH is most likely much higher among these, and PH may have a large effect on mortality and exercise capacity which may be underestimated by grouping them together with patients who do not truly have PH. Furthermore, of the three patients screened positive, in whom PH was not confirmed, one had a MPAP of 22 mmHg which is above the upper limit of normal (25), and one had an abnormally high MPAP and PCWP due to left heart failure. Also, the prevalence of atrial fibrillation was higher in those screened positive. Atrial fibrillation can be caused by PH per se, but is more frequently associated with left ventricular dysfunction, which may then lead to PH. Taken together; the patients with echocardiographic signs compatible with PH represent a group with higher risk profile.

As shown before (15), higher age was also associated to PH on echocardiography in the present study. After adjusting mortality for age, the hazard ratio for death was 1.7, in those screened positive, but was no longer significantly different from those screened negative. This raises the question of whether screening for PH is superfluous, identifying individuals that will die due to age. However, with only 23 deaths during follow-up, the age-adjusted results should be interpreted with caution, and another argument that screening for PH may be worthwhile is that people die from diseases and not age per se. Furthermore, it has been shown that the pulmonary pressure increases with age in a normal population, but also that higher mortality is observed with increasing pulmonary pressures after adjustment for age, and it has been suggested that reducing pulmonary pressure might be a target for intervention (26). In line with this, other co-morbidities should also be identified and treated.

Thus, the results from the present study suggest that the identification of echocardiographic signs compatible with PH as defined by a TR > 40 mmHg, a TAPSE < 1.8 cm or right ventricular dilatation is an important prognostic parameter. For the individual patient, further examinations may be required to confirm the diagnosis of PH and to optimize treatment.

NT-proBNP

In a study by Goetze et al. (27), it was concluded that NT-proBNP was not able to detect mild to moderate PH in patients with chronic lung disease, because the NT-proBNP levels in these patients were within the normal limits. In accordance with these observations, the results from the present study show that only half of the patients with PH on echocardiography had NT-proBNP above 300 ng/l, which is the upper limit of normal at our hospital laboratory. However, levels of NT-proBNP below 95 ng/l, were able to exclude the presence of PH on echocardiography with a high NPV, and patients with values below this limit also had a better prognosis. This cut-off value is supported by results from Wieshammer et al.(28) in which a value below 93 ng/l could rule out cardiac disease including PH in a cohort of patients with pulmonary disease.

Approximately 30% of the population had NT-proBNP values below 95 ng/l and introducing NT-proBNP with this limit as a first line screening tool, may reduce the need for echocardiography in part of the COPD patients if PH is suspected. However, larger studies should confirm this before clinical implementation. On the other hand, the PPV of the cut-off value was low. This may in part be due to the occurrence of cardiovascular co-morbidities such as hypertension, ischemic heart disease and left ventricular dysfunction which can also result in higher levels of NT-proBNP (29).

Conclusions

In the present cohort of stable, well treated COPD out-patients, 14% were screened positive for PH by echocardiography using a TR > 40 mmHg, a TAPSE < 1.8 cm or right ventricular dilatation as criteria for PH. Data from RHC illustrated that not all of these had true PH. Consequently, we estimate the prevalence of PH to be below 11.5%.

PH on echocardiography was associated with a higher mortality and a lower exercise capacity. This suggests that echocardiographic findings of PH are important and should lead to more thorough diagnostic evaluation.

Finally, the results indicate that NT-proBNP values below 95 ng/l could be used as a first line screening tool to select COPD patients, who should be referred to echocardiography for PH.

Declaration of Interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. The Danish Lung Association supported the study financially.