The aim of the present study was to explore the relation between right ventricular (RV) and left ventricular (LV) echocardiographic parameters with clinical outcome in patients with advanced heart failure referred for cardiac transplantation. Ninety-eight consecutive patients with advanced systolic heart failure, referred for cardiac transplant evaluation, were enrolled. All patients were prospectively followed for the development of new outcome events, which included hospitalization for acute heart failure, cardiovascular death, heart transplantation, intra-aortic balloon pump implantation, and ventricular assist device implantation. Conventional transthoracic echocardiography was performed in all subjects. RV longitudinal strain (RVLS) by speckle-tracking echocardiography was assessed by averaging all segments in apical 4-chamber view (global RVLS) and by averaging RV free-wall segments (free-wall RVLS). LV global longitudinal and global circumferential strains were also calculated. Of the 98 subjects at baseline, 46 had 67 new events during a mean follow-up of 1.5 ± 0.9 years. Free-wall RVLS, global RVLS, N-terminal fragment of the prohormone brain natriuretic peptide, RV fractional area change, and LV end-diastolic volume were independently predictive of combined outcomes (all p <0.0001). The overall performance for the prediction of cardiovascular events was greatest for free-wall RVLS (area under the curve free-wall RVLS: 0.87; global RVLS: 0.67; RV fractional area change: 0.60; N-terminal fragment of the prohormone brain natriuretic peptide, 0.62; global circumferential strain: 0.55; global longitudinal strain: 0.35; and LV ejection fraction: 0.26). Free-wall RVLS showed the highest adjusted hazards ratio. A graded association between the grade of RV dysfunction and the risk of cardiovascular events was only evident for free-wall RVLS and global RVLS. In conclusion, in patients referred for heart transplantation, RVLS is a stronger predictor of outcome than LV longitudinal strain and other conventional parameters, providing a stronger prognostic stratification.
The prognostic value of the ejection fraction of the left ventricle (LV) loses strength when applied to patients with advanced heart failure (HF). Recently, a number of studies have provided evidence that RV ejection fraction is an independent prognostic factor in patients with moderate-to-severe HF and that the status of the right ventricle (RV) is often a determinant in the success of using left ventricular assist devices in patients with end-stage HF. Thus, quantitative assessment of RV function, often neglected, may further stratify this high-risk population. Among the various echocardiographic parameters of RV systolic function, RV fractional area change (RVFAC), the systolic velocity (S′) of tricuspid annulus by tissue Doppler, and the M-mode measurement of tricuspid annulus systolic displacement are easy to obtain, reproducible, and have been validated in patients with HF but remain only a regional approach to the complex shape of RV.
Recently, RV systolic function can be more easily studied thanks to speckle-tracking echocardiography, which has been proposed as an alternative approach to define the RV longitudinal function, enabling the quantification of myocardial deformation from standard bidimensional echo images, without the problem of angle dependence. The aim of the present study was to explore the relation between RV and LV longitudinal strain measurements with clinical outcome in patients with advanced HF referred for cardiac transplantation.
Methods
Consecutive patients with advanced systolic HF, referred for cardiac transplant evaluation, were enrolled and underwent transthoracic echocardiographic study. All were in sinus rhythm and hemodynamically stable, without any episodes of acute decompensated state in the last 3 months. A previous cardiac resynchronization therapy with defibrillator was not an exclusion criterion. Patients were excluded if they had no sinus rhythm, mechanical ventilation, severe mitral/tricuspid regurgitation, any mitral/tricuspid stenosis, any prosthetic valve, heart transplantation, or an insufficient imaging quality of the RV and LV endocardial borders. All subjects gave their written informed consent for the participation in the study. All work was in compliance with the Declaration of Helsinki, and it was performed with the approval of the local ethics committee.
Echocardiographic studies were performed using an ultrasound machine (Vivid 7; GE, Milwaukee, Wisconsin) with the subjects in the left lateral recumbent position. In accordance with the recommendations of American Society of Echocardiography, LV and RV dimensions, RVFAC, and LV ejection fraction (LVEF) were calculated; longitudinal function was explored with pulsed tissue Doppler imaging by placing the sample volume at the level of mitral and tricuspid annuls, from the apical 4-chamber view. Peak systolic (S′), early diastolic (E′), and late diastolic (A′) annular velocities were recorded at the end of expiration to minimize translational motion. Also, the M-mode measurements of mitral annular plane systolic excursion and tricuspid annulus systolic displacement were calculated, with the M-mode cursor aligned through the mitral and tricuspid annuli in the apical 4-chamber view.
For speckle-tracking analysis, apical 2-, 3-, and 4-chamber view images were obtained using conventional 2-dimensional gray scale echocardiography, with a stable electrocardiographic recording. Care was taken to obtain the best visualization of LV and RV, from base to apex, and a more reliable delineation of its endocardial border in each view. Three consecutive heart cycles were recorded and averaged. The frame rate was set at 60 to 80 frames/s. The analysis of the files recorded was performed off-line by a single experienced and independent echocardiographer who was not directly involved in the image acquisition and had no knowledge of hemodynamic measurements, using a commercially available semiautomated 2-dimensional strain software (EchoPac; Vivid 7, GE, Waukesha, Wisconsin).
To calculate RV and LV global strains, endocardial border is manually traced in apical views, thus delineating a region of interest, composed by 6 segments for each view. Then, after the segmental tracking quality analysis and the eventual manual adjustment of the region of interest, the longitudinal strain curves are generated by the software for each ventricular segment. Global longitudinal strain (GLS) was assessed by the software as the average strain of all segments in 17-LV-segment model; the right ventricular longitudinal strain (RVLS) was calculated by averaging values observed in all RV segments.
In patients in whom some segments were excluded because of the impossibility of achieving adequate tracking, the longitudinal strain was calculated by averaging values measured in the remaining segments ( Figure 1 ). Free-wall RVLS was derived by precise and manually delineation of the endocardial border of the only free wall, obtaining a region of interest, divided in 3 segments, basal, medial, and apical. The corresponding strain curves, generated by the software, allowed to calculate RVLS for each free-wall segment, whereas the dashed curve, which represents the average value of the strain curves of other segments, was used to evaluate free-wall RVLS ( Figure 1 ). From the short-axis views at the basal apical levels, a global circumferential strain curve was obtained. The average of peak global circumferential strain (GCS) from the 2 short-axis views was calculated as LV GCS.
All patients were prospectively followed for development of new outcome events, which included cardiovascular death, hospitalization for acute HF, heart transplantation, intra-aortic balloon pump implantation, and ventricular assist device implantation.
To assess the reproducibility of LV GLS and RVLS, 10 patients were randomly selected and Bland-Altman analysis was performed to evaluate the intra- and interobserver agreements by repeating the analysis 1 week later by the same observer and a second independent observer.
Data are shown as mean ± SD. A p value <0.05 was considered statistically significant. Pearson correlation coefficients were calculated to assess the relations among continuous variables. Sensitivity and specificity were calculated using standard definitions. To evaluate the prognostic accuracy of the echocardiographic indexes, receiver operating characteristic curves were constructed and the area under the curve was calculated. Analyses were performed using the SPSS software (Statistical Package for the Social Sciences; SPSS Inc., Chicago, Illinois), release 12.0.
Results
Of 113 patients screened, 98 patients (37 women and 61 men) met eligibility criteria during the study period. The admitting diagnoses were coronary artery disease (67 patients) and nonischemic cardiomiopathy (31 patients). All patients were classified as New York Heart Association class III to IV with an LVEF of ≤30%. Seven patients were excluded for nonsinus rhythm, 4 for severe mitral valve disease, 3 for poor echocardiographic window, and 1 for difficulties in heart catheterization. Table 1 lists the clinical, echocardiographic, and catheterization data of the study population.
Characteristic | No Events (n = 52) | Events (n = 46) | p |
---|---|---|---|
Clinical data | |||
Age (yrs) | 59.0 ± 8.0 | 59.2 ± 6.9 | 0.77 |
Gender (% women) | 38 | 39 | 0.64 |
Body mass index (kg/m 2 ) | 25.6 ± 2.8 | 24.9 ± 3.0 | 0.62 |
Body surface area (m 2 ) | 1.94 ± 0.3 | 1.98 ± 0.2 | 0.86 |
Hypertension (%) | 86 | 87 | 0.75 |
Diabetes mellitus (%) | 36 | 34 | 0.59 |
Hypercholesterolemia (%) | 78 | 82 | 0.65 |
Current smoker (%) | 10 | 9 | 0.81 |
Ischemic cause (%) | 67 | 69 | 0.71 |
NYHA class (%) | III (77); IV (23) | III (74); IV (26) | 0.30 |
Echocardiographic data | |||
Left atrial volume indexed (ml/m 2 ) | 32.6 ± 8.1 | 31.9 ± 9.0 | 0.40 |
End-diastolic LV volume (ml) | 196.1 ± 30.1 | 215.0 ± 40.2 | 0.05 |
LV mass index (g/m 2 ) | 118.1 ± 26.0 | 126.2 ± 31.9 | 0.16 |
LVEF (%) | 26.4 ± 4.1 | 25.3 ± 4.9 | 0.71 |
Mitral E/A ratio | 2.03 ± 0.7 | 2.10 ± 0.8 | 0.31 |
LV global longitudinal strain (%) | −9.0 ± 2.5 | −8.0 ± 1.8 | 0.16 |
LV global circumferential strain (%) | −12.5 ± 3.7 | −10.8 ± 2.5 | 0.01 |
E/E′ ratio (cm/s) | 18.3 ± 8.4 | 16.8 ± 7.4 | 0.79 |
Tricuspid annular plane systolic excursion (mm) | 15.0 ± 2.1 | 14.5 ± 2.3 | 0.31 |
RV fractional area change (%) | 35.6 ± 4.6 | 32.2 ± 4.1 | 0.05 |
Tricuspid S′ (cm/s) | 12.0 ± 2.3 | 10.9 ± 2.1 | 0.01 |
Global RV longitudinal strain (%) | −14.2 ± 2.6 | −12.0 ± 1.7 | 0.001 |
Free-wall RV longitudinal strain (%) | −19.2 ± 2.7 | −15.0 ± 2.1 | <0.0001 |
Catheterization data | |||
Heart rate (beats/min) | 78.8 ± 9.1 | 76.1 ± 9.5 | 0.36 |
Systolic blood pressure (mm Hg) | 112 ± 22 | 110 ± 20 | 0.56 |
Diastolic blood pressure (mm Hg) | 77 ± 10 | 75 ± 11 | 0.57 |
Mean pulmonary artery pressure (mm Hg) | 29.7 ± 8.1 | 32.0 ± 6.8 | 0.18 |
Pulmonary artery occlusion pressure (mm Hg) | 17.0 ± 5.1 | 18.6 ± 7.0 | 0.15 |
Cardiac index by thermodilution (ml/min/m 2 ) | 2.15 ± 0.7 | 1.94 ± 0.9 | 0.001 |
RV stroke work index (mm Hg/L m 2 ) | 0.38 ± 0.16 | 0.24 ± 0.08 | <0.0001 |
During a mean follow-up period of 1.5 ± 0.9 years, of the 98 subjects at baseline, 46 had 67 new events (11 cardiovascular deaths, 31 hospitalizations for acute HF, 9 heart transplantations, 10 intra-aortic balloon pump implantations, and 6 ventricular assist device implantations). Patients who developed cardiovascular events presented reduced free-wall and global RVLSs, reduced RVFAC and RV stroke work index, and greater LV end-diastolic volume at baseline. Free-wall RVLS, global RVLS, N-terminal fragment of the prohormone brain natriuretic peptide, RVFAC, and LV GCS were independently predictive of combined outcomes (all p <0.0001). No significant difference was found for the analyzed parameters between the group of patients with and without cardiac resynchronization therapy.
To further investigate the value of these conventional and innovative indexes to predict cardiovascular events, we performed receiving operating characteristic curve analyses. The overall performance for the prediction of cardiovascular events was greatest for free-wall RVLS (area under the curve free-wall RVLS: 0.87; global RVLS: 0.67; RVFAC: 0.60; N-terminal fragment of the prohormone brain natriuretic peptide: 0.62; GCS: 0.55; GLS: 0.35; and LVEF: 0.26; Table 2 and Figure 2 ). A graded association between the grade of RV dysfunction and the risk of cardiovascular events was only evident for free-wall RVLS and global RVLS ( Figures 3 and 4 ). After adjustment for age, gender, severity of mitral regurgitation, systolic pulmonary arterial pressure, and severity of diastolic dysfunction, more depressed free-wall RVLS, RVFAC, and LV GCS predicted greater risks of cardiovascular events; free-wall RVLS showed the highest adjusted hazards ratio ( Table 3 ).
Variable | Cut-Off Value (%) | Sensitivity % (95% CI) | Specificity % (95% CI) | AUC |
---|---|---|---|---|
LVEF | <24.3 | 50.2 (42.0–80.3) | 51.0 (25.2–73.1) | 0.26 |
LV global longitudinal strain | >−8.1 | 58.2 (53.1–87.5) | 62.4 (26.5–76.5) | 0.35 |
LV global circumferential strain | >−11.4 | 57.9 (51.8–85.9) | 64.5 (40.2–80.2) | 0.55 |
RV fractional area change | <30.6 | 66.7 (58.3–88.6) | 78.9 (69.9–89.6) | 0.60 |
Global RV longitudinal strain | >−12.1 | 71.8 (59.5–87.2) | 80.8 (72.0–90.2) | 0.67 |
Free-wall RV longitudinal strain | >−15.0 | 85.1 (73.6–94.0) | 87.8 (80.1–98.9) | 0.87 |