Abstract
Question addressed Echocardiography is not currently considered as providing sufficient prognostic information to serve as an integral part of treatment goals in pulmonary arterial hypertension (PAH). We tested the hypothesis that incorporation of multiple parameters reflecting right heart function would improve the prognostic value of this imaging modality.
Methods and main results We pooled individual patient data from a total of 517 patients (mean age 52±15 years, 64.8% females) included in seven observational studies conducted at five European and United States academic centres. Patients were subdivided into three groups representing progressive degrees of right ventricular dysfunction based on a combination of echocardiographic measurements, as follows. Group 1 (low risk): normal tricuspid annular plane systolic excursion (TAPSE) and nonsignificant tricuspid regurgitation (TR) (n=129); group 2 (intermediate risk): normal TAPSE and significant TR or impaired TAPSE and nondilated inferior vena cava (IVC) (n=256); group 3 (high risk): impaired TAPSE and dilated IVC (n=132). The 5-year cumulative survival rate was 82% in group 1, 63% in group 2 and 43% in group 3. Low-risk patients had better survival rates than intermediate-risk patients (log-rank Chi-squared 12.25; p<0.001) and intermediate-risk patients had better survival rates than high-risk patients (log-rank Chi-squared 26.25; p<0.001). Inclusion of other parameters such as right atrial area and pericardial effusion did not provide added prognostic value.
Answer to the question The proposed echocardiographic approach integrating the evaluation of TAPSE, TR grade and IVC is effective in stratifying the risk for all-cause mortality in PAH patients, outperforming the prognostic parameters suggested by current guidelines.
Abstract
The proposed comprehensive echocardiographic approach reflecting both RV pump function and systemic venous congestion is highly effective in risk stratification of PAH patients, outperforming the prognostic parameters suggested by current guidelines https://bit.ly/3fDiaUl
Introduction
The importance of risk stratification has gained significant momentum in the field of pulmonary arterial hypertension (PAH). In 2015, the European Society of Cardiology/European Respiratory Society guidelines recommended the use of a broad range of invasive and noninvasive parameters to assess the mortality risk of PAH patients [1]. In 2017, abbreviated versions of this model were evaluated retrospectively in newly diagnosed PAH cohorts, and further validated in systemic sclerosis associated PAH, demonstrating that stratifying patients into low-, intermediate- and high-risk groups based upon select measurements predicted mortality [2–6]. As suggested in the most recent recommendations from the 6th World Symposium on Pulmonary Hypertension, currently approved PAH therapies should be initiated based on risk stratification. Furthermore, achievement of a low-risk status should be the goal of therapy during follow-up assessment [7].
The National Institutes of Health Registry on Primary Pulmonary Hypertension first demonstrated that mortality best correlates with haemodynamic indices of right ventricular (RV) function, specifically right atrial pressure (RAP) and cardiac index [8]. Since then, two haemodynamic variables, i.e. one indicator of pump function (cardiac index or stroke volume index) and one of systemic congestion (RAP), proved to have pivotal roles in the multidimensional assessment of prognosis and treatment response in PAH, in association with demographic, clinical, functional and biochemical variables [2–6, 9–12].
However, despite extensive literature reporting that a number of echocardiographic variables may provide significant prognostic information, this imaging modality remains of limited application in risk stratification and guidance of treatment goals as it is felt to provide insufficient prognostic information [1]. It is noteworthy that a recent statement of the American Thoracic Society recommends that deriving and validating imaging biomarkers for an accurate evaluation of RV function is a major research priority to be addressed in the next 5 years [13]. We surmised that the present limited role of echocardiography might be at least in part related to the use of single best prognostic indicators rather than on the identification of a combination of parameters which could provide a comprehensive assessment of right heart pump function and of systemic venous congestion, similar to the established haemodynamic approach.
Accordingly, to test the hypothesis that a comprehensive echocardiographic assessment of RV function would be highly valuable in the risk stratification of PAH patients, we performed a patient-level pooled analysis of published studies conducted at our centres, focusing on RV systolic function, the degree of tricuspid regurgitation and the presence or absence of systemic venous congestion.
Methods
Study selection
We pooled individual patient data (IPD) from observational studies conducted at our centres that assessed the prognostic value of echocardiography in patients with PAH. To be eligible for inclusion, patients had to be characterised with the following echocardiographic parameters: an index of RV systolic function represented by the tricuspid annular plane systolic excursion (TAPSE), the degree of tricuspid regurgitation (TR) and a marker of systemic venous congestion represented by inferior vena cava (IVC) diameter [14–20]. The primary outcome of the study was all-cause mortality. This analysis is part of a study that was registered on PROSPERO (registration number: CRD42019141216). Two authors (FF, VM) extracted IPD from the selected studies including that for study design, sample size, patient characteristics, World Health Organization (WHO)/New York Heart Association (NYHA) functional class, functional capacity expressed as 6-min walking distance, PAH aetiology, PAH-specific therapy, echocardiographic assessment, follow-up and all-cause mortality. Data on scheduled lung transplantation were reported only in one study [15] where two patients underwent lung transplantation and their follow-up was censored at the time of surgery. Data on overlapping population between the studies were excluded.
Statistical analysis
Continuous variables are presented as mean±sd in case of Gaussian distribution, and as median (interquartile range) if not normally distributed. Categorical variables are presented as frequencies and percentages. To assess the hazard ratio change for all-cause mortality across a range of TAPSE at baseline, a spline curve analysis was performed. The cut-off value of TAPSE associated with excess mortality according to the literature was used to define RV systolic dysfunction [21]. According to the pathophysiological aim of the study, three echocardiographic variables were selected a priori: TAPSE, defined as abnormal when ≤17 mm; TR, defined as significant if moderate or severe [22]; IVC diameter, defined as dilated when >20 mm [21]. Based on the combination of these variables, four echocardiographic risk groups of patients were identified representing progressive degrees of right heart dysfunction based upon the presence/absence of RV dysfunction by TAPSE and degree of TR, and presence/absence of systemic venous congestion by IVC dilatation: group 1: normal TAPSE and nonsignificant TR; group 2: normal TAPSE in the presence of significant TR; group 3: impaired TAPSE without systemic congestion; and group 4: impaired TAPSE associated with systemic congestion.
Differences among the groups were analysed using the one-way ANOVA for continuous variables with Gaussian distribution, the Kruskal–Wallis test for non-normally distributed continuous variables and the Pearson's Chi-squared test for categorical variables. The Kaplan–Meier curves were used to estimate the 1-, 3- and 5-year survival rates and differences between groups were analysed using the Mantel–Cox log-rank test. A multivariable Cox proportional hazards regression analysis was performed to assess the clinical and echocardiographic factors that were independently associated with all-cause mortality. Possible confounders with a significant p-value (p<0.05) in the univariable analysis were selected (taking into account data availability and collinearity between variables) and included in the multivariable regression analysis. The proportional-hazards assumption was confirmed using statistics and graphs on the basis of the Schoenfeld residuals. p-values <0.05 were considered significant. All data were analysed with SPSS for Windows (version 21; SPSS: IBM, Armonk, NY, USA) and R environment 3.4.3 (R Foundation for Statistical Computing, Vienna, Austria) using the “stats” and “survival” packages.
Results
Study characteristics
Data from seven original articles were included in this study. Characteristics of the publications included in the analysis are presented in table 1. With the exception of two retrospective studies, all other studies had a prospective design. The median follow-up ranged from 15 to 95 months. The outcome measure defined as all-cause mortality was available in all studies.
General characteristics of publications included in the analysis
Patient population
After excluding overlapping patients between studies, patients with no PAH and those with incomplete echocardiographic data to be classified in one of the four echocardiographic groups, a total of 517 patients (mean age 52±15 years, 64.8% female) was included in the analysis (figure 1): 253 patients from Giessen University (Giessen, Germany) [20]; 123 from IRCCS Policlinico San Matteo (Pavia, Italy) [15, 16]; 71 from the Johns Hopkins Hospital (Baltimore, MD, USA) [14, 17, 19]; 70 from the University of Pennsylvania Hospital and Temple University Hospital (Philadelphia, PA, USA) [18]. The main characteristics of the patients included in the study were similar to those excluded due to missing echocardiographic data (supplementary table S1). Clinical and demographic characteristics of the study population are reported in table 1. Concerning the aetiology, 64.8% had idiopathic PAH, 8.9% had scleroderma-associated PAH and 8.4% had congenital heart disease associated PAH. 375 patients (72.5%) were on treatment with PAH-specific drugs (42.7% on monotherapy and 57.3% on combination therapy) at the time of echocardiographic assessment. Most patients (66.1%) were in WHO/NYHA functional class III–IV. To investigate the association between TAPSE and all-cause mortality, a spline curve analysis was performed. The assumption of linearity for all-cause mortality, predicted from the baseline TAPSE, was not violated (Chi-squared 2.56; p=0.29), i.e. demonstrating a nonlinear relation of TAPSE versus all-cause mortality. The spline curve analysis showed, after a slow rise, a steady increase in the hazard ratio for all-cause mortality for a TAPSE of ≤17 mm (figure 2). Therefore, based on previous [21] and current findings, we considered appropriate to identify RV systolic dysfunction as TAPSE ≤17 mm. There was a total of 129 (25%) patients in the echocardiographic group 1, 155 (30%) patients in group 2, 101 (19.5%) patients in group 3 and 132 (25.5%) patients in group 4. There was a statistically significant difference in WHO functional class and in 6-min walk distance (6MWD) across groups, with the highest percentage of patients in WHO/NYHA functional class III–IV (p<0.001) as well as the lowest values of 6MWD in group 4 (p=0.007). Heart rate was the lowest in patients in group 1 (p=0.004).
Patient selection. #: patients excluded due to overlap between studies: 37 from Mathai et al. [17] (25 overlapping cases with Forfia et al. [14] and 12 overlapping cases with Mukherjee et al. [19]) and nine from Ghio et al. [15] which overlapped with Ghio et al. [16]; ¶: from Forfia et al. [14]; +: patients excluded due to incomplete echocardiographic data to be classified according to the four groups: 74 with unavailable data on inferior vena cava, 12 with unavailable tricuspid annular plane systolic excursion and three with unavailable tricuspid regurgitation grade. WHO: World Health Organization; PH: pulmonary hypertension; PAH: pulmonary arterial hypertension.
Spline curve for tricuspid annular plane systolic excursion (TAPSE) versus hazard ratio of all-cause mortality. Changes in hazard ratio for all-cause mortality across the baseline TAPSE.
Echocardiographic data
Clinical and echocardiographic characteristics are presented in table 2. Overall, the study population had normal left ventricular size and function, while right heart chambers were increased in dimensions, with a median right atrial area of 21 (17–26) cm2 and a RV end-diastolic diameter of 38 (30–45) mm. TAPSE was impaired (≤17 mm) in 45.1% of patients, with an overall mean value of 18±5 mm for the entire cohort. Fractional area change was also impaired (≤35%) in 76.6% of patients, with an overall mean value of 27.1±12.2%. 333 (65%) patients had a significant TR (TR grade ≥moderate). Estimated pulmonary artery systolic pressure was 68±26 mmHg, and pulmonary flow acceleration time was 78±22 ms; pericardial effusion was only present in 27.6% of patients. Clinical and echocardiographic characteristics of the four echocardiographic risk groups are shown in table 2. All indices of right heart function and morphology were increasingly worse from group 1 to group 4.
Characteristic of the overall population and of the four pathophysiologically defined echocardiographic risk groups
Outcome data
During a median follow-up of 46 (21–99) months, 206 (39.8%) patients died. The Kaplan–Meier analysis showed lower survival rates in patients with more impaired RV systolic function and with systemic congestion (group 4) (log-rank Chi-squared 58.12; p<0.001) (figure 3a). The survival rates between groups 2 and 3 did not differ significantly (log-rank Chi-squared 1.68; p=0.195). Therefore, we considered it appropriate to merge patients belonging to these two groups into one, thus identifying three echo-derived risk categories which had significantly different survival rates at follow-up (figures 3b and 4): low risk (group 1), intermediate risk (group 2 and 3) and high risk (group 4). The cumulative 1-, 3- and 5-year survival rates were 92%, 87% and 82% for low-risk patients; 85%, 73% and 63% for intermediate-risk patients; and 68%, 57% and 43% for high-risk patients. Low-risk patients had better survival rates than intermediate-risk patients (log-rank Chi-squared 12.25; p<0.001) and the latter had better survival rates than high-risk patients (log-rank Chi-squared 26.25; p<0.001).
Kaplan–Meier curves for survival according to a) four and b) three echocardiographic-based risk groups in patients with pulmonary arterial hypertension. The Kaplan–Meier curves show significantly lower 5-year survival rates for high-risk patients (group 4) compared to both intermediate-risk (groups 2 and 3) (43% versus 63%, respectively; p<0.001) and low-risk (group 1) (43% versus 82%, respectively; p<0.001). TAPSE: tricuspid annular plane systolic excursion; TR: tricuspid regurgitation; IVC: inferior vena cava.
Echocardiographic risk stratification in patients with pulmonary arterial hypertension. Three risk categories were identified from four groups representing progressively increasing degrees of right heart dysfunction based upon presence/absence of right ventricular (RV) dysfunction and of systemic venous congestion: a) low risk: preserved tricuspid annular plane systolic excursion (TAPSE) and nonsignificant tricuspid regurgitation (TR) (group 1); b) intermediate risk: either preserved TAPSE and significant TR (group 2) or impaired TAPSE and nondilated inferior vena cava (IVC) (group 3); and c) high risk: impaired TAPSE and dilated IVC (group 4).
At univariable Cox regression analysis the following parameters were associated with all-cause mortality: older age, male sex, higher heart rate, PAH associated with pulmonary veno-occlusive disease, connective tissue disease aetiology, being an incident patient, being treated before 2011, WHO/NYHA functional class III–IV, lower 6MWD, impaired RV systolic function, RV dilation, shorter pulmonary flow acceleration time, higher pulmonary systolic pressure, greater right atrial (RA) area, systemic congestion, significant TR and lower left ventricular volume and ejection fraction (table 3). In the multivariable analysis after adjusting for potential confounders, intermediate- and high-risk categories were independently associated with 2.5- and 4.4-fold increased risk of all-cause mortality compared with low-risk patients. Moreover, these echocardiographically defined risk categories retained an independent association with all-cause mortality in several multivariable Cox regression models (supplementary table S2). As shown in supplementary figure S1, the Chi-square of a Cox regression model including a single echocardiographic parameter of RV systolic dysfunction (TAPSE ≤17 mm), systemic congestion (dilated IVC) or significant TR (TR grade 2–3), significantly increased when these parameters were combined (Chi-squared change 15.54 (p<0.001), 41.06 (p<0.001) and 47.96 (p<0.001), respectively). Thus, this analysis supports the echocardiographic-group risk stratification based on the combination of these three variables. Importantly, the presence of pericardial effusion was not associated with all-cause mortality at univariate analysis; in addition, although RA area showed an independent association with the primary outcome (supplementary table S3), its additive value was outperformed by the proposed echocardiographic risk stratification model (supplementary figure S2).
Univariable and multivariable Cox regression analysis
Discussion
The main finding of the present study is the demonstration that a comprehensive assessment of right heart function, easily obtainable through the combination of an echocardiographic indicator of the systolic function of the right ventricle associated with the degree of TR and an estimate of systemic venous congestion, can provide an accurate prognostic stratification of PAH patients, outperforming the prognostic value of the parameters suggested by current guidelines.
There is a sound physiological rationale for the proposed combination of echocardiographic parameters. First, it is important to acknowledge that the assessment of ventricular function using imaging methodologies is limited by the load-dependency of all parameters assessing cardiac motion, whether they are volume-based as in ejection fraction, or area-based as in fractional area change, or single plane-based such as TAPSE. Indeed, the presence of significant atrioventricular valve regurgitation inevitably leads to an overestimation of the systolic function of the left/right ventricle if this is based only on the analysis of ventricular motion. As a matter of fact, patients with normal TAPSE with absent or trivial TR had a significantly better prognosis than those with normal TAPSE associated with moderate to severe TR.
Patients with reduced TAPSE were further subgrouped according to the absence or presence of dilated IVC. The rationale for this subdivision is simply based on the fact that the presence of reduced TAPSE and dilated IVC identifies PAH patients having impaired RV systolic function associated with systemic venous congestion. This is consistent with the observation in heart failure from left heart disease where patients with severe left ventricular dysfunction have a better exercise tolerance and improved prognosis if their pulmonary artery wedge pressure is normal (i.e. if there is no pulmonary congestion) [23]. Our results clearly indicate that patients with reduced TAPSE without dilated IVC had a significantly better prognosis compared to patients with reduced TAPSE and dilated IVC.
The major strength of the present study is that the proposed methodology uses a combination of simply and widely available echocardiographic parameters that reflect a comprehensive assessment of right heart function, thus conforming with the results of multiple seminal invasive haemodynamic studies that have demonstrated the association of an indicator of pump function and one of systemic venous congestion is the best prognostic predictor in PAH [2–4, 7–12]. Importantly, the results of this study might pave the way for testing future, potentially more sensitive, parameters mirroring RV systolic function or the degree of TR or systemic venous congestion. In particular, three-dimensional analysis of RV volumes and ejection fraction, or two- or three-dimensional assessment of RV strain or indices of ventriculo-arterial coupling might be tested to replace TAPSE [24]. The quantification of TR regurgitation is known to be a challenging task for echocardiographists, and a most recent position statement specifies that new parameters should be identified and validated in prospective outcome studies [25]. Finally, the best modality to assess the prognostic relevance of right atrial function needs to be further clarified [26].
Importantly, this proposed comprehensive risk stratification model was superior to single echocardiographic parameters recommended by current PAH guidelines. Pericardial effusion was not associated with all-cause mortality in our study. In addition, while RA area was associated with all-cause mortality, this finding provided no added value to the comprehensive echocardiographic assessment.
Limitations
A major limitation of the present study is its retrospective nature with the inclusion of a limited amount of studies on the topic. As a matter of fact, this is a pooled analysis of seven studies conducted at five institutions, which may represent a source of selection bias. However, this limitation is partially overcome by the large number of patients enrolled in referral centres with great expertise in the echocardiographic evaluation of PAH, including prevalent and incident patients with different forms of PAH. In addition, we have included several variables in Cox regression analysis (i.e. study design, year of publication, incident/prevalent patients, site of enrolment) that partially allow taking into account between-studies heterogeneity. Another important limitation of the present study is the lack of integration of the proposed echocardiographic assessment, in a multidimensional tool including all the haemodynamic, clinical, functional and biochemical variables suggested by international guidelines. This should be addressed in future prospective studies. Similarly, since some studies suggest that treatment with PAH-specific drugs have beneficial effects on various echocardiographic parameters of RV morphology and function [24], the relevance of follow-up assessment using the proposed echocardiographic approach should be further explored. Echocardiographic variables are clearly preferable for serial evaluations compared to risk assessment tools incorporating invasive haemodynamic variables. They are also more practical and less costly, albeit less accurate, compared to other imaging techniques, such as cardiac magnetic resonance, that recently demonstrated to be useful in improving risk stratification of PAH patients [27].
Conclusion
Our study underlines the prognostic importance in PAH patients of combining individual echocardiographic parameters into a comprehensive physiological approach reflecting both RV pump function and systemic venous congestion. If validated in prospective studies and in cohorts of patients with different types of PAH, this approach could refine the current guideline recommendations related to echocardiography in assessing prognosis and stratification of PAH, which are currently limited to the assessment of the right atrial area and the presence of pericardial effusion.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-00513-2020.SUPPLEMENT
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00513-2020.Shareable
Footnotes
This article has an editorial commentary: https://doi.org/10.1183/13993003.02313-2020
The study protocol was registered on PROSPERO (registration number: CRD42019141216).
This article has supplementary material available from erj.ersjournals.com
List of contributors: Pavia: Michela Coccia, Federico Fortuni, Stefano Ghio, Alessandra Greco, Arianna Grelli, Letizia Mannucci, Claudia Raineri, Laura Scelsi; Baltimore: Rachel L. Damico, Todd M. Kolb, Paul M. Hassoun, Stephen C. Mathai, Valentina Mercurio, Monica Mukherjee; Philadelphia: Paul R. Forfia, Jeremy A. Mazurek; Giessen: Henning Gall, Ardeschir Ghofrani, Manuel Richter, Andreas Rieth, Werner Seeger, Khodr Tello; Naples: Valentina Mercurio, Alessandra Cuomo.
Conflict of interest: S. Ghio has nothing to disclose.
Conflict of interest: V. Mercurio has nothing to disclose.
Conflict of interest: F. Fortuni has nothing to disclose.
Conflict of interest: P.R. Forfia has nothing to disclose.
Conflict of interest: H. Gall has nothing to disclose.
Conflict of interest: A. Ghofrani has nothing to disclose.
Conflict of interest: S.C. Mathai has nothing to disclose.
Conflict of interest: J.A. Mazurek has nothing to disclose.
Conflict of interest: M. Mukherjee has nothing to disclose.
Conflict of interest: M. Richter has nothing to disclose.
Conflict of interest: L. Scelsi has nothing to disclose.
Conflict of interest: P.M. Hassoun has nothing to disclose.
Conflict of interest: K. Tello has nothing to disclose.
Support statement: V. Mercurio received a research fellowship grant from the European Respiratory Society. P.M. Hassoun was funded by NIH/NHLBI R01 HL114910.
- Received March 2, 2020.
- Accepted May 7, 2020.
- Copyright ©ERS 2020
References
