Objective To investigate clinical outcomes of exercise-induced pulmonary hypertension (PH) and implications of an increase in left ventricular (LV) filling pressure during exercise in subjects with preserved LV ejection fraction.
Design Longitudinal follow-up study.
Setting Subjects who were referred for diastolic stress echocardiography.
Patients and methods The ratio of transmitral and annular velocities (E/Ea) and pulmonary artery systolic pressure (PASP) at rest and during exercise were measured in 498 subjects (57±11 years; 201 male). Exercise-induced PH was defined as present if PASP ≥50 mm Hg at 50 W of exercise, and an increase in LV filling pressure during exercise was present if E/Ea ≥15 at 50 W.
Main outcome measures A combination of major cardiovascular events and any cause of death.
Results During a median follow-up of 41 months, there were 14 hospitalisations and four deaths. Subjects with exercise-induced PH had significantly worse clinical outcomes than those without (p=0.014). Subjects with exercise-induced PH associated with an increase in E/Ea during exercise had significantly worse outcomes than other groups (p<0.001). However, prognosis was similar between subjects with exercise-induced PH without an increase in E/Ea and those without exercise-induced PH. In subjects with exercise-induced PH, E/Ea at 50 W was an independent predictor of adverse outcomes (HR 1.37; 95% CI 1.02 to 1.83; p=0.036).
Conclusions Exercise-induced PH provides prognostic information in subjects with preserved LV ejection fraction. The excess risk of exercise-induced PH is restricted to subjects with an increase in estimated LV filling pressure during exercise.
- diastolic dysfunction
- pulmonary arterial hypertension (PAH)
Statistics from Altmetric.com
- diastolic dysfunction
- pulmonary arterial hypertension (PAH)
Exercise-induced pulmonary hypertension (PH) is a poorly understood entity, but it is common in subjects with reduced as well as preserved left ventricular (LV) ejection fraction.1 It is considered an important cause of exertional dyspnoea and exercise intolerance.1 2 As exercise-induced PH has been suggested as an early, mild and clinically relevant phase of PH, screening and early intervention for exercise-induced PH has been recommended.2 Surprisingly, however, although disease progression has been documented in patients with heart failure with diastolic dysfunction,3 there are no systematic longitudinal follow-up data regarding clinical outcomes of exercise-induced PH. PH can be related to an increase in LV filling pressure4 or pulmonary resistance and pulmonary vascular remodelling.5 Doppler echocardiography has been shown to be able to estimate pulmonary arterial systolic pressure (PASP) and LV filling pressure at rest6–8 and during exercise.9 10 Therefore we can detect exercise-induced PH or exercise-induced increase in LV filling pressure by using diastolic stress echocardiography.1 11–13 In this study, we hypothesised that: (1) subjects with exercise-induced PH would have poorer long-term outcomes than those without exercise-induced PH; and (2) exercise-induced PH particularly associated with an increase in estimated LV filling pressure during exercise would have poorer outcomes.
We retrospectively analysed 1347 consecutive subjects (592 male; mean±SD age 57±11 years) referred for diastolic stress echocardiography11 from January 2003 to 2009. Subjects with LV systolic dysfunction (LV ejection fraction <50%, n=50), hypertrophic cardiomyopathy (n=136) and end-stage renal disease (n=34) were excluded. All subjects were in sinus rhythm. None had significant valvular disease, defined as moderate or greater severity. Those with echocardiographic evidence of inducible myocardial ischaemia (n=25) were excluded. Subjects whose tricuspid regurgitation (TR) velocity or E/Ea could not be measured either at rest or up to 50 W of exercise (n=465 and 139, respectively) were also excluded. The remaining 498 subjects (201 male; mean±SD age 57±11 years) constituted the study population. No subjects were diagnosed with lung disease or connective tissue diseases such as systemic lupus erythematosus, rheumatoid arthritis or systemic sclerosis. All subjects underwent symptom-limited exercise testing on a supine bicycle ergometer with simultaneous respiratory gas analysis. Exercise-induced PH was defined as PASP ≥50 mm Hg at 50 W of exercise. An increase in estimated LV filling pressure during exercise was defined as E/Ea ≥15 at 50 W of exercise. Subjects were classified according to the absence (group I, n=327) or presence (n=171) of exercise-induced PH. Those with exercise-induced PH were further subdivided into two groups according to the absence (group II, n=122) or presence (group III, n=49) of exercise-induced increase in estimated LV filling pressure. All subjects gave informed consent, and study approval was obtained from the institutional review board of Yonsei University College of Medicine.
Diastolic stress echocardiography
Standard two-dimensional measurements were obtained as recommended by the American Society of Echocardiography in the left lateral position.14 From the apical window, a 1–2 mm pulsed Doppler sample volume was placed at the mitral valve tip. Mitral inflow velocities were traced, and E and A peak velocities and deceleration time of E velocity were measured. TR jet velocity was also obtained to estimate PASP using continuous-wave Doppler.15 PASP was estimated from TR velocity by adding the right atrial pressure of 10 mm Hg. Peak velocities of mitral annulus during systole (Sa), early diastole (Ea) and atrial contraction (Aa) were measured from the apical four-chamber view, with a 2–5 mm sample volume placed at the septal corner of the mitral annulus. Stroke volume was measured from the LV outflow tract diameter and time velocity integral as previously described.16 Cardiac output was calculated from stroke volume × heart rate. Pulmonary vascular resistance was measured as previously described.17
After standard rest images had been obtained, multistage supine bicycle exercise testing was performed with a variable load bicycle ergometer (Medical Positioning, Kansas City, Missouri, USA). Subjects pedalled at a constant speed, starting with a 25 W workload; speed was increased by 25 W every 3 min. A standard segmental model was used to measure wall motion scores index, in accordance with American Society of Echocardiography guidelines.14 In the apical four-chamber view, colour flow imaging was performed at rest and during peak exercise to assess mitral regurgitation. A 12-lead ECG was continuously recorded to exclude significant myocardial ischaemia. Blood pressure was recorded every 3 min with a cuff sphygmomanometer. Exercise workload was defined as the total metabolic equivalents (METs) achieved. Oxygen uptake (Vo2), carbon dioxide production (Vco2), respiratory exchange ratio and minute ventilation (VE) were measured using breath-by-breath gas analysis (Medical Graphics, St Paul, Minnesota, USA). Peak Vo2 was defined as the highest 10 s average of oxygen uptake in the last minute of exercise. The respiratory exchange ratio represents the amount of CO2 produced divided by Vo2. VE and Vco2 responses throughout exercise were used to calculate the VE/Vco2 slope by least squares linear regression (y=mx+b, m=slope). During exercise, oxygen saturation was recorded continuously by finger pulse oximetry (Solar 8000M Patient Monitor; GE Medical Systems, Milwaukee, Wisconsin, USA).
Follow-up data were obtained by a review of medical records, telephone interviews, or information from the Korea National Statistical Office by an independent investigator who was unaware of the subjects' clinical and echocardiographic data. The primary end point was a combination of death from any cause or unplanned hospitalisation due to heart failure or myocardial infarction. The follow-up period was initiated on the day of diastolic stress echocardiography. Clinical follow-up was achieved in 481 of 498 (96.6%) subjects for death and major cardiovascular events.
Continuous variables are presented as mean±SD, and categorical variables as absolute and relative frequencies (%). Categorical data were analysed using the χ2 statistic. To assess the significance of continuous data, one-way analysis of variance corrected by the Scheffe procedure for multiple comparisons was used. The correlation between exercise capacity and echocardiographic variables was analysed using the Pearson correlation coefficient.
Survival curves were constructed using Kaplan–Meier estimates and compared with the log-rank test according to the presence of exercise-induced PH. After subgrouping of the subjects with exercise-induced PH according to the presence of exercise-induced increase in estimated LV filling pressure, survival curves were compared among the three groups. Cox univariate analysis was used to determine relationships between baseline and exercise variables and clinical outcomes in subjects with exercise-induced PH. Adjusted survival rates were compared using multivariate Cox proportional-hazards regression analysis. Variables selected for entry into the analysis were those with p<0.1 on Cox univariate analysis and important variables. The incremental prognostic value of an increase in E/Ea during exercise was assessed in three modelling steps. The first step consisted of fitting a multivariate model to the clinical data. Then echocardiographic variables at rest and during exercise were added sequentially. The model χ2 value was used to determine whether there was a difference between various nested models for predicting the end point. In addition, to reduce the effect of potential confounding between group II and group III, we performed rigorous adjustment for the differences in baseline characteristics of subjects using propensity score matching.18 19 The propensity scores were estimated without regard to outcome variables through multiple logistic-regression analysis. To develop the propensity score-matched pairs without replacement (a 1:1 match), the Greedy 5→1 digit match algorithm was used as described previously.20 Propensity score matching for the groups with exercise-induced PH yielded 39 matched pairs of subjects. After propensity score matching, the baseline covariates were compared with the paired t test for continuous variables and the McNemar test for categorical variables. In the propensity score-matched cohort, the risks of clinical end points were compared using Cox regression models. All reported p values are two-sided, and values of p<0.05 were considered significant.
Clinical and echocardiographic characteristics and haemodynamic variables at rest and during exercise
Tables 1 and 2 present the clinical and haemodynamic comparisons among the three groups. Compared with subjects without exercise-induced PH, those with exercise-induced PH were older, more often female, and had higher systolic blood pressure during exercise. Subjects in group III showed shorter exercise duration, lower METs and lower peak Vo2 than those in groups I and II. Nearly all (97%) subjects reached a peak respiratory exchange ratio >1.0, and there was no significant difference between the groups. None had echocardiographic or ECG evidence of myocardial ischaemia at rest and during exercise. Only three (0.6%) of 498 subjects showed dynamic mitral regurgitation.
Echocardiographic variables at rest and during exercise suggest that subjects in group III had worse diastolic function at rest (higher LV mass index, left atrial volume index, higher E, prolonged deceleration time, lower Ea, and higher E/Ea) and further worsening of diastolic function with exercise (lower Ea and higher E/Ea at each stage of exercise). Stroke volume and cardiac output were not significantly different at rest and during exercise. Subjects with exercise-induced PH showed significantly higher TR velocities and estimated PASP at rest as well as with exercise. The mean PASP at rest, 25 W of exercise and 50 W of exercise in group III were 28.6±6.5, 44.8±8.3, 58.2±7.0 mm Hg, respectively. These were not significantly different from those in group II, but significantly higher than those in group I.
Figure 1 shows Doppler findings for two subjects presenting with exercise-induced PH with different responses of exercise E/Ea.
Relationship between exercise capacity and PASP at rest and during exercise
In simple correlation analysis, the PASP at 50 W of exercise showed a better correlation with peak Vo2 (r=−0.21, p<0.001) than the PASP at rest (r=−0.10, p=0.033). The change in PASP from rest to 50 W exercise also significantly correlated with peak Vo2 (r=−0.22, p<0.001). E/Ea at rest and at 50 W of exercise showed similar correlation coefficients (r=−0.35, p<0.001; r=−0.32, p<0.001) with peak Vo2, reflecting exercise capacity.
During the follow-up (median duration 41 (range 5–93) months), 18 (3.6%) of 498 subjects had adverse events (four deaths and 14 unplanned hospitalisations due to heart failure).
The subjects with exercise-induced PH had significantly worse clinical outcomes than those without exercise-induced PH (p=0.014 by log-rank test, figure 2A). When subjects were classified into three groups, the subjects with exercise-induced PH associated with an increase in E/Ea during exercise had significantly worse outcomes than those in the other groups (p<0.001 by log-rank test, figure 2B). However, the prognosis was similar in subjects with exercise-induced PH without an increase in E/Ea and those without exercise-induced PH.
The results of Cox hazards regression analysis for subjects with exercise-induced PH are presented in table 3. In univariate analysis, diabetes mellitus, LV mass index, left atrial volume index, E velocity at rest, Ea velocity at 50 W, E/Ea at rest and E/Ea at 50 W of exercise were found to have statistically significant associations with adverse clinical outcomes. However, in a multivariate Cox hazards regression analysis, the only independent predictor of adverse outcomes in subjects with exercise-induced PH was an E/Ea at 50 W of exercise (HR 1.37; 95% CI 1.02 to 1.83; p=0.036). For the 37 propensity score-matched pairs, the risk of clinical end points was significantly higher in the group of exercise-induced PH associated with an increase in E/Ea during exercise than in the group of exercise-induced PH without an increase in E/Ea (HR 8.68; 95% CI 1.09 to 68.17; p=0.042).
Incremental prognostic value of an increase in E/Ea in subjects with exercise-induced PH
Incremental value of an E/Ea at rest and with exercise is shown in figure 2C. The addition of a resting E/Ea significantly improved the prognostic utility of a model containing clinical data (age, gender, history of hypertension or diabetes mellitus) and an echocardiographic parameter (LV mass index). Furthermore, addition of E/Ea at 50 W of exercise to the clinical data, LV mass index and resting E/Ea significantly increased the χ2 value from 17.2 to 25.9 (p=0.035), indicating the incremental prognostic value of estimated LV filling pressure during exercise in subjects with exercise-induced PH.
The principal findings of this study are as follows: (1) Subjects with exercise-induced PH have worse outcomes than those without exercise-induced PH even those with preserved LV ejection fraction. (2) However, when subjects were classified into three groups, the subjects with exercise-induced PH associated with an increase in E/Ea had significantly worse outcomes than those in the other groups (p<0.001 by log-rank test, figure 2B), whereas no significant differences in clinical outcomes were found between the subjects with exercise-induced PH without an increase in E/Ea and the subjects without exercise-induced PH. Therefore exercise-induced PH when associated with an increase in E/Ea during exercise portends poorer clinical outcome. This study is the first to demonstrate the prognostic effect of exercise-induced PH in subjects with preserved LV ejection fraction and the implication of an increase in E/Ea as a prognosticator of exercise-induced PH.
Exercise-induced PH in subjects with preserved LV ejection fraction
PH is common in patients with LV systolic dysfunction, contributes to exercise intolerance,21 and is associated with worse outcomes.22 The degree of PH is not independently related to the severity of LV systolic dysfunction, but is associated with LV diastolic filling abnormalities.4 During exercise, a significant increase in PASP occurs, which is the result of the exercise-induced increase in left atrial pressure (that is, transmitted backwards into the pulmonary vasculature) and cardiac output. Despite the low pulmonary vascular resistance under resting conditions, a further decrease in pulmonary resistance occurs during exercise.23 Therefore the level of increase in PASP is highly variable, and an exercise-induced increase in left atrial pressure and blunted pulmonary vasodilation during exercise can influence the level of the increase in PASP during exercise.
In a previous study, we found that exercise-induced PH is common even in subjects with preserved LV systolic function and that it is strongly associated with E/Ea, TR velocity, age, gender, and systolic blood pressure during exercise.1 Although an increase in LV filling pressure, estimated by E/Ea ratio, is strongly related to exercise-induced PH, exercise-induced PH can occur in the absence of a significant increase in LV filling pressure. There was some evidence that exercise-induced PH was associated with an increase in pulmonary arterial constriction, remodelling of the pulmonary arteries, and alveolar hypoxia.1 23–25 Thus some believe that exercise-induced PH is an early and more treatable phase that precedes resting PH.2 On the other hand, Bossone et al26 suggested that exercise PASP could increase to a greater extent even in physiological conditions, probably related to enhanced cardiac output response. In this study, the subjects with exercise-induced PH did not show significantly enhanced cardiac output response during exercise.
In this study, we first showed that exercise-induced PH in subjects with preserved LV ejection fraction was associated with worse outcomes, and an increase in estimated LV filling pressure during exercise was the most important prognosticator. These results are consistent with the conclusion of Tolle et al2 that exercise-induced PH is an important pathophysiological entity. Furthermore, our study provides more in-depth information for predicting clinical outcomes in subjects presenting with exercise-induced PH. Regarding the relatively good clinical outcomes in subjects with exercise-induced PH without an increase in exercise E/Ea, there are a few things to consider. First, in this study, there were no subjects with connective tissue disease, who may be highly susceptible to exercise-induced PH and future PH. Second, the primary end point in this study was cardiovascular events, and the major one was unplanned hospitalisation due to heart failure. Therefore fewer cardiovascular events in subjects with exercise-induced PH without an increase in exercise E/Ea can expect results.
Prognostic value of estimated LV filling pressure in subjects with exercise-induced PH
Exertional dyspnoea is a common and important symptom that has multiple aetiologies.27 The dynamic changes in cardiac function and pulmonary pressure can lead to conditions such as an exercise-induced increase in LV filling pressure12 27 28 and exercise-induced PH.1 Diastolic stress echocardiography is widely used because a reliable estimation of both LV filling pressure and PASP can be obtained non-invasively during exercise.9–13 Furthermore, recent studies using diastolic stress echocardiography have shown a dynamic change in the LV diastolic elastance29 and a combination of systolic and diastolic abnormalities of ventricular function during exercise as underlying pathophysiologies of heart failure with preserved LV ejection fraction.30
The prognostic significance of LV filling pressure at rest is well recognised in different clinical settings and populations, including heart failure31 and acute myocardial infarction.32 In recent research, Holland et al33 showed the incremental prognostic implications of LV filling pressure with exercise. Our study, however, is quite different because it focused on the effect of exercise-induced increase in LV filling pressure on the outcome of exercise-induced PH, whereas that of Holland et al focused on the prognostic implication of LV filling pressure with exercise for inducible myocardial ischaemia. Furthermore, the study aims and baseline characteristics of the study subjects were also different. None of the subjects from our study had inducible myocardial ischaemia during exercise, whereas nearly half (47.9%) of the subjects in the study by Holland et al did. In addition, our study had a longer follow-up (41 months vs 13 months).
In our study population, exercise-induced PH was not always associated with poor clinical outcomes during the 41 months of the follow-up period. When it was associated with an increase in estimated LV filling pressure with exercise, it revealed poorer clinical outcomes in subjects with exercise-induced PH. In a recent invasive study, a potential role for exercise PASP in screening for heart failure with preserved ejection fraction in subjects with exertional dyspnoea was emphasised.34 In that study, exercise PASP correlated highly with exercise LV filling pressure and was shown to have superior discriminative ability for heart failure than resting PASP.34 Therefore attempts to prevent or treat exercise-induced PH should focus on improving LV diastolic properties and decreasing the LV filling pressure at rest and during exercise in preserved LV ejection fraction.
Limitations of this study need to be addressed. First, PASP and LV filling pressure were calculated non-invasively by Doppler echocardiography and not measured by catheterisation. Therefore subjects with immeasurable TR velocity were excluded from analysis, and subjects in whom E/Ea could not be measured during exercise were also excluded. Non-invasive methods show moderate correlations with invasively measured PASP.35 In addition, there is ongoing controversy regarding the accuracy of exercise E/Ea as a non-invasive measure of LV filling pressure.36 Therefore the applicability of the results in this study to individual subjects may not be as robust. However, we suggest that the combined increase in PASP and E/Ea during exercise is certainly abnormal and can predict clinical outcomes. For calculating E/Ea, we obtained Ea at the septal mitral annulus. The septal Ea could be influenced by septal motion abnormalities in subjects with PH. However, correct alignment of Doppler angle tends to be difficult during exercise when we measure Ea at the lateral mitral annulus. There was a potential limitation of right atrial pressure estimation with exercise. That assumption of right atrial pressure in our study can probably be violated, and further study is needed to identify non-invasive ways to measure right atrial pressure with exercise. Second, the baseline characteristics between groups showed significant differences. To analyse the prognostic importance of an increase in LV filling pressure during exercise in subjects with exercise-induced PH, we developed the technique of propensity score matched pairs through multiple logistic-regression analysis to reduce the effect of potential confounding in an observational study. We also found that the risk of clinical end points was significantly higher in the group with exercise-induced PH associated with an increase in LV filling pressure during exercise than in the group with exercise-induced PH without an increase in LV filling pressure. However, there are inherent limitations to using an observational population because of unmeasured confounding factors. Third, in this study, there were no subjects with connective tissue diseases, who could be highly susceptible to exercise-induced PH and future PH. Therefore the results from our study cannot be generalised to subjects with connective tissue diseases.
In subjects with preserved LV ejection fraction, exercise-induced PH provides prognostic information. The increased risk of exercise-induced PH is restricted to subjects with an increase in estimated LV filling pressure during exercise.
The authors thank the staff at the Medical Research Support Section of Yonsei University for statistical support.
Funding This work was supported by a Korean Science and Engineering Foundation (KOSEF) grant funded by the Korean government (M10642120001-06N4212-00110).
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the institutional review board of Yonsei University College of Medicine.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.