Abstract
Survival in patients with pulmonary arterial hypertension (PAH) is determined by right ventricular (RV) function adaptation to afterload. How altered RV function impacts on exercise capacity in PAH is not exactly known.
104 idiopathic PAH (IPAH) patients aged 52±14 years underwent a diagnostic right heart catheterisation, a comprehensive echocardiography including two-dimensional speckle tracking for RV dyssynchrony evaluation and a cardiopulmonary exercise test. Multivariate analyses were performed to identify independent predictors of peak oxygen uptake (peak V′O2).
A first multivariate analysis of only resting haemodynamic variables identified cardiac index, right atrial (RA) pressure and pulmonary arterial compliance as independent predictors, with low predictive capacity (r2=0.31; p<0.001). A second multivariate analysis model which considered only echocardiographic parameters but without RV dyssynchrony, identified RV fractional area change (FAC) and RA area as independent predictors with still low predictivity (r2=0.35; p<0.001). Adding RV dyssynchrony to the second model increased its predictivity (r2=0.48; p<0.001). Repetition of the three multivariate analyses in patients with preserved RVFAC confirmed that inclusion of RV dyssynchrony results in the highest predictive capability of peak V′O2 (r2=0.53; p=0.001).
A comprehensive echocardiography with speckle tracking-derived assessment of the heterogeneity of RV contraction improves the prediction of aerobic exercise capacity in IPAH.
Abstract
Speckle tracking assessment of RV dyssynchrony improves the prediction of aerobic exercise capacity in IPAH http://ow.ly/yHyh30bedtD
Introduction
It has been better realised in recent years that symptomatology, functional state, exercise capacity and survival in pulmonary arterial hypertensions (PAH) are essentially to be accounted for by right ventricular (RV) function [1, 2]. Studies relying on pressure–volume loops measured by high-fidelity manometer-tipped catheters and advanced imaging with cardiac magnetic resonance (CMR) have shown that coupling of RV function to the increased resistive state of the pulmonary circulation basically relies on an increased contractility (or end-systolic elastance, Ees) to match the increased afterload (or arterial elastance, Ea) [2, 3]. However, bedside assessment of RV arterial coupling with Ees and Ea determined by high-fidelity catheters and CMR is not possible in daily clinical practice. A number of simpler surrogate right heart catheterisation- and echocardiography-derived parameters of RV function have been shown to be of prognostic relevance [3], but the independent impact of each of them on symptomatology and exercise capacity has not been exactly defined.
Standard echocardiography in patients with pulmonary hypertension includes estimates of pulmonary artery pressure (PAP) combined with measurements of RV structure and global function [4]. However, recent imaging by tissue Doppler, magnetic resonance imaging (MRI) or speckle tracking echocardiography have shown that increased PAP may be associated with a considerable heterogeneity of RV regional function, or dyssynchrony, prolonged contraction and eventual post-systolic shortening [5–12]. Although RV dyssychrony has been shown to be an independent predictor of disease severity and outcome in PAH [11, 12], how it affects cardiac output adaptation to metabolic demand during exercise has not yet been explored.
We hypothesised that RV dyssychrony would decrease the efficiency of RV contraction and thereby decrease exercise capacity. Therefore, the aim of this study was to analyse the determinants of exercise capacity as defined by peak oxygen uptake (peak V′O2) during a cardiopulmonary exercise test (CPET) among right heart catheterisation and standard echocardiographic measurements with and without assessment of RV dyssynchrony.
Methods
Study population
108 consecutive patients with idiopathic PAH gave informed consent to the study, which was approved by the Institutional Review Board for human studies of the Policlinico Umberto I – Sapienza University of Rome (Protocol no. 42412). Four patients were excluded because of poor echocardiographic imaging quality. Thus, 104 idiopathic PAH patients were enrolled in the study, 65 women and 39 men, aged 52±14 years, with a body mass index of 25±4 kg·m−2, a WHO functional class of 2.8±0.4 and a 6-min walk distance of 430±60 m. The diagnosis of idiopathic PAH rested on dyspnoea–fatigue symptomatology, right heart catheterisation showing a mean PAP (mPAP) ≥25 mmHg and a wedged PAP (PAWP) ≤15 mmHg, and a step-by-step approach including lung function tests, a computed tomography scan of the chest, a ventilation/perfusion scan, echocardiography and laboratory tests to exclude left heart conditions with increased pulmonary venous pressure, lung diseases, thromboembolism, connective tissue disease and other associated PAH conditions, following recently updated guidelines [13].
Right heart catheterisation
The patients underwent right heart catheterisation with a triple lumen balloon-tipped thermodilution Swan-Ganz catheter in the supine position with zero at the mid-chest level and pressures measured at end-expiration. The measurements included PAP, PAWP, right atrial pressure (RAP) and cardiac output (CO). Pulmonary vascular resistance (PVR) was calculated as (mPAP – PAWP)/CO, and PA compliance as (systolicPAP – diastolicPAP)/stroke volume.
Echocardiographic assessment
A comprehensive echocardiography was performed within 24 h after the right heart catheterisation. All echocardiographic data were acquired by the same operator using commercially available equipment (Vivid S6, GE), with a 3.5-MHz transducer at a depth of 16 cm in the standard views; the patient was in the left lateral decubitus position. Standard M-mode, 2D and Doppler images were obtained during breath-hold at end expiration and measurements were obtained from the mean of three consecutive beats in accordance with the American Society of Echocardiography Guidelines [4]. The following standard parameters and derived measures were considered in the analysis: right atrial (RA) area, RV end-diastolic area (RVEDA), RV end-systolic area (RVESA), RV fractional area change % (RVFAC=(RVEDA−RVESA)/RVEDA×100), tricuspid annular plane systolic excursion (TAPSE), left ventricular systolic and diastolic eccentricity index (LV-EIs and LV-EId, respectively) and presence of pericardial effusion. Tricuspid regurgitation was semiquantitatively graded considering the regurgitant jet area using colour Doppler imaging. The transmitral flow velocity curve was obtained using pulsed Doppler imaging, positioning the sample volume between the tips of the mitral leaflets. E- and A-wave peak velocities and the ratio of early transmitral flow velocity to atrial flow velocity were measured.
2D Speckle tracking echocardiography
Acquisition
For speckle tracking analysis (EchoPAC workstation 7.0.1, GE Medical Systems), standard greyscale 2D images in the RV dedicated four-chamber apical view were acquired (frame rate >60 fps) and digitally stored in a 3 beats cine-loop format.
Analysis
To assess the segmental characteristics of the RV, we adopted the six-segment model, excluding the apical segments for the analysis because of the high variability of these segments observed even in normal subjects, as previously reported and discussed [11, 12]. For the quantification of RV dyssynchrony we considered longitudinal strain and calculated the standard deviation of the times to peak-systolic strain for the four mid-basal RV segments corrected to the R–R interval between two QRS complexes, according to the Bazett's formula, and called RV-SD4. Using the upper 95% limit of normal (mean+2 sd) of healthy subjects previously reported by our group [12], we defined a cut-off value of 18 ms as the criterion for RV dyssynchrony.
For RV-SD4 measurement, intra-observer and inter-observer variability were assessed for 20 randomly selected patients by the Bland–Altman method, with results similar to previously published data [11, 12]: 0.00±2.05 (95% CI –3.94 to 3.94) (average percentage variability 6.4%) and −0.37±1.89 (95% CI –4.08 to 3.34) (average percentage variability 5.9%), respectively, which can be considered acceptable for our clinical purpose.
Cardiopulmonary exercise test
All patients performed a symptom-limited incremental cycler ergometer CPET with 10–15 W·min−1 workload increments. Oxygen uptake (V′O2), carbon dioxide output (V′CO2) and minute ventilation (V′E) were measured breath-by-breath (Quark CPET, Rome, Italy) and averaged every 5 s for subsequent analysis. Oxygen pulse was calculated as V′O2/HR. Heart rate (HR) was monitored via a 12-lead ECG. Tests were considered maximal if the peak respiratory exchange ratio (RER) was greater than 1.1.
Statistical analysis
Continuous data were expressed as mean±standard deviation, and categorical data were expressed as counts and proportions. Two-group comparisons were done with unpaired, two-tailed t-tests for means if the data were normally distributed or with Wilcoxon's rank-sum tests if the data were not normally distributed. Chi-squared or Fisher's exact tests were used to analyse the categorical data.
Linear regression analysis was performed to assess the relations between RVFAC and peak V′O2 and expressed as a Pearson correlation coefficient. Patients were further divided according to the tertile of RV dyssynchrony to build a scatterplot.
The median value of RVFAC (38%) was used to divide the population into two groups, low and high systolic function, as this value resembles the lower limit of normal suggested by guidelines [4].
Multivariate regression analysis was used to identify the variables that were associated with peak V′O2 by a stepwise variable selection method with significance level to entry 0.1 and significance level to stay 0.05. Three models were constructed for the overall population: Model-1 was limited to haemodynamic variables, Model-2 was limited to echocardiographic variables excluding RV dyssynchrony, Model-3 was limited to echocardiographic variables including RV dyssynchrony. Another six models were constructed for subgroups of patients with RVFAC above and below 38%: Model-4 and Model-7 were limited to haemodynamic variables; Model-5 and Model-8 were limited to echocardiographic variables excluding RV dyssynchrony; Model-6 and Model-9 were limited to echocardiographic variables including RV dyssynchrony. Predictive accuracy of the models was compared using r2, and details for each model are reported, including corresponding r2, p-value, constant, and regression coefficient.
All statistical analyses were performed using SPSS software (version 20.0, IBM) and Stata 13 (StataCorp, College Station, TX, USA). All statistical tests were two-sided, and p<0.05 was considered statistically significant.
Results
As shown in table 1, patients presented with a haemodynamic profile compatible with severe pulmonary hypertension, echocardiography showing increased RV dimensions and depressed systolic function and a CPET showing markedly decreased peak V′O2, workload, O2 pulse and maximum HR with increased V′E/V′CO2 slope. There was marked RV dyssynchrony, with RV-SD4 values ranging from 0 to 124 ms, and exceeding 18 ms in 60% of the patients.
There was a significant correlation between RVFAC and peak V′O2 (r2=0.39; p=0.0001). The correlation between TAPSE and peak V′O2 was also significant, but weaker (r2=0.18; p<0.001).
Analysing the characteristics of the two groups of patients based on the median value of RVFAC (38%) (table 2), those ≤38% had a worse WHO functional class, shorter 6-min walk distance, worse haemodynamics and more advanced RV remodelling and dyssynchrony. Considering the CPET, all patients reached maximal exercise from a metabolic point of view (peak respiratory exchange ratio >1.1 in both groups). Patients with RVFAC ≤38% had lower peak workload, V′O2 and O2 pulse, and reduced ventilatory efficiency (higher V′E/V′CO2 slope).
In the overall population (table 3), a first multivariate analysis model considering resting hemodynamic variables exclusively identified CI, RAP and pulmonary artery compliance as independent predictors of peak V′O2, but with low predictive capability (Model-1, r2=0.31; p<0.001). A second model considering only echocardiographic variables (RVEDA, RVESA, RVFAC, TAPSE, RA area, pericardial effusion, LV-EId and LV-EIs) without RV dyssynchrony evaluation disclosed RVFAC and RA area as independent predictors, but with low predictive capability (Model-2, r2=0.35; p<0.001). A third model with RV dyssynchrony added to echocardiographic variables disclosed RVFAC, RA area and RV dyssynchrony as independent predictors, with much stronger predictive capability (Model-3, r2=0.48; p<0.001).
In the group of patients with preserved RVFAC (table 4), RAP and pulmonary artery compliance emerged as the independent factors in the haemodynamic model (Model-4, r2=0.18; p<0.01) and confirmed a low predictive capability of peak V′O2 from haemodynamic measurements. Among the echocardiographic parameters excluding RV dyssynchrony evaluation, RVEDA and RA area remained independent predictors of peak V′O2 (Model-5, r2=0.21; p<0.01). Adding RV dyssynchrony to the echocardiographic evaluation a final model was generated, in which the only independent variables remaining were RA area and RV dyssynchrony, but with incremental predictive capability compared with all the other models (Model-6, r2=0.53; p<0.001).
A scatterplot of RVFAC versus peak V′O2 (figures 1 and 2) highlights how the presence of RV dyssynchrony may play a role among those patients with low peak V′O2 and preserved systolic function, as documented by regression modelling analysis. As shown in figure 1, patients with the intermediate and the upper tertile of RV dyssynchrony distribution are more common among the group of patients with low peak V′O2 and preserved systolic function (lower-right quadrant) than those with high peak V′O2 and preserved systolic function (higher-right quadrant).
Indeed, the upper and intermediate tertiles of RV dyssynchrony distribution were associated with a more impaired WHO functional class, exercise capacity, haemodynamic condition and RV sizes and systolic function than the lower tertile (table 5, figure 3).
Finally, in the group of patients with RVFAC ≤38% (table 6), CI and RAP emerged as independent variables in the haemodynamic model (Model-7, r2=0.48; p<0.001) confirming the high predictive capability of peak V′O2 in the subset of patients with more impaired RV dysfunction. Among the echocardiographic parameters excluding RV dyssynchrony, RVEDA and LV-EId remained independent predictors of peak V′O2 (Model-8, r2=0.26; p<0.01). Adding RV dyssynchrony to the echocardiographic evaluation, the RVEDA and LV-EId remained independent factors associated with RV dyssynchrony, presenting incremental predictive capability compared with Model-8 (Model-9, r2=0.46; p<0.001). Interestingly, the later model was not able to improve peak V′O2 predictive capability compared with the haemodynamic model.
Discussion
The present results show that echocardiographic imaging of RV function with speckle tracking quantification of regional heterogeneity of contraction is of added value with respect to right heart catheterisation-derived indices to account for decreased aerobic exercise capacity in patients with idiopathic PAH.
Exercise capacity is impaired in PAH, with typically a marked decrease in peak V′O2 and O2 pulse and enhanced ventilatory responses [14]. This is thought to be due to a decreased cardiac output response to increased metabolic demand, with the addition of increased ventilation explaining exercise-induced dyspnoea and fatigue. Decreased maximum cardiac output in severe pulmonary hyptertension is essentially explained by the uncoupling of RV systolic function to the pulmonary circulation (i.e. impaired homeometric adaptation) [15] with eventual compensation by increased RV dimensions (heterometric adaptation) [16, 17]. Increased RA and RV surface areas, increased eccentricity indices and depressed FAC and TAPSE indicate decreased systolic function and increased dimensions, and thus suggest RV failure as a cause of limitation of maximum cardiac output and exercise capacity. It is of interest that RVFAC was tightly correlated with peak V′O2 and clinical/haemodynamic indicators of severity of PAH, further underscoring the importance of preserved systolic function adaptation to afterload in these patients.
In the present study different multivariate analysis models had to be applied to consider the independent impact of physiologically relevant invasive and noninvasive variables on exercise capacity. A single analysis combining right heart catheterisation and echocardiography was not possible owing to the limited number of patients. Both the invasive and the noninvasive approaches capture determinants of RV function. Right atrial pressure is an estimate of RV end-diastolic volume, the cardiac index is the indirect result of RV contractility at any given loading condition, and PA compliance is an estimate of afterload [3]. On the other hand, FAC is a load-dependent estimate of contractility and RA surface a partial and indirect estimate of preload. It is not therefore surprising that the prediction capability of exercise capacity was significant but equally low for both right heart catheterisation and standard comprehensive echocardiography.
RVFAC was chosen over TAPSE for description of the systolic function as it allows a more clear and continuous distribution of patients in respect of increased afterload. In contrast to TAPSE, RVFAC does not present a floor effect in cases of severe RV dysfunction [18] and is not affected by the overall heart motion [19].
Tissue Doppler imaging, speckle tracking echocardiography and also magnetic resonance studies have demonstrated that pulmonary hypertension may be associated with a prolonged RV contraction into LV diastole, resulting in “post-systolic shortening” or asynchrony [5–7], but also with a considerable regional heterogeneity of RV contraction or dyssynchrony [8–12]. Recent studies have shown that RV dyssynchrony is associated with a worse functional state, more severely impaired haemodynamics and echocardiography [10, 11], and poor outcome [12]. Correction of dyssynchrony by RV pacing has been shown to restore RV function and cardiac output in rats with monocrotaline-induced PH [20]. Regional heterogeneity is thus an important component of altered RV function in PH. If one assumes that the addition of a measure of regional synchrony of contraction improves the evaluation of RV systolic function, it may not be surprising that it also improves the prediction capability of exercise capacity by echocardiography or right heart catheterisation alone.
Speckle tracking-derived strain of the RV in patients with PH has been shown to be depressed in proportion to haemodynamic severity and decreased ejection fraction or 6-min walk distance [21] and to be sensitive to therapeutic interventions [21, 22] as well as to the occurrence of cardiovascular events [23]. More recently, this novel imaging modality allowed disclosure of regional differences between apical, mid and base segments which were prominent yet not detected by conventional measures of function such as TAPSE in patients with systemic sclerosis compared with controls [24]. Thus regional inhomogeneity of speckle tracking measurements of RV strain also appears to be a sensitive marker of early or occult myocardial disease. The approach was further developed in the present study with quantification and determination of functional relevance with respect to exercise capacity.
There are several limitations to the present results. First, RV dyssynchrony by 2D speckle tracking was evaluated using software validated for the LV. Whether software specifically developed for the RV might perform better for the prediction of V′O2 max is not known. Second, 2D speckle tracking is dependent on plane selection and thus may be methodologically inferior to 3D speckle tracking [10]. However, there has been no comparison between these different approaches, and 2D speckle tracking in the present study showed-up as a sensible predictor of exercise capacity. Third, echocardiography and right heart catheterisation were performed at rest only, although exercise measurements would be more pertinent to exercise capacity. However, there are still technical limitations to speckle tracking of the RV during exercise. Fourth, the multivariate analyses had to rely on multiple separate models because of a large number of variables with respect to the patient population. However, idiopathic PAH is an orphan disease and recruiting a much larger population would require a multicentric effort in the long term. Finally, a multiparametric approach combining advanced imaging techniques with invasive assessments would probably improve the characterisation of the functional state, exercise capacity and survival in PAH [25]. However, this will be for further studies after rigorous identification of all the independent predictors to incorporate in composite scores.
In conclusion, speckle tracking echocardiography of RV dyssynchrony is of added value to right heart catheterisation and standard echocardiography in the evaluation of disease severity of patients with idiopathic PAH.
Disclosures
Footnotes
Conflict of interest: Disclosures can be found alongside this article at erj.ersjournals.com
- Received July 15, 2016.
- Accepted March 7, 2017.
- Copyright ©ERS 2017