Background
Right ventricular (RV) remodeling has been associated with outcomes in patients with pulmonary hypertension (PH). However, the additive prognostic significance of RV remodeling and left ventricular (LV) morphology in PH is unclear. The objective of this study was to test the hypothesis that the ratio of RV end-diastolic area to LV end-diastolic area is a biventricular index predictive of outcome in patients with PH.
Methods
In total, 139 patients with precapillary PH (mean age, 55 ± 15 years; 75% women) and 22 control subjects (mean age, 40 ± 17 years; 73% women) were studied. The apical four-chamber view was used to measure the RV-to-LV end-diastolic area ratio as an index of biventricular cardiac remodeling. RV free wall and global strain were measured using speckle-tracking echocardiography. The study design was prospective, with all-cause mortality over 5 years predefined as the outcome event.
Results
Patients with PH had significantly larger RV to LV end-diastolic area ratios than normal subjects, as expected (1.06 vs 0.67, P < .0001). There were 72 deaths over 5 years. Using a cutoff value of 0.93, patients with RV-to-LV ratios ≥ 0.93 had significantly higher all-cause mortality (hazard ratio,1.84; 95% CI, 1.14-2.96; P = .019). RV global strain was also significantly associated with survival using a cutoff of ≥−15% (hazard ratio, 1.66; 95% CI, 1.03-2.67; P = .044). In a multivariate analysis, only age and biventricular index were independent predictors of survival among other clinical and echocardiographic features.
Conclusions
The RV-to-LV end-diastolic area ratio is a simplified biventricular echocardiographic index of cardiac remodeling that is predictive of long-term survival in patients with PH.
Pulmonary hypertension (PH) is a progressive, often life-threatening disease but with a variable prognosis. Determining the appropriate individualized treatment for patients with PH, such as advanced therapies of continuous infusion of pharmacologic agents or lung transplantation, often depends on their prognosis. Right ventricular (RV) failure is the most common cause of death in patients with PH, and evaluation of RV function has become of great clinical importance. Although clinical practice frequently relies on an invasive right heart catheterization hemodynamic study, several noninvasive imaging markers are used for prognosis in PH. Echocardiography is the most common modality to evaluate RV function and to predict outcome in patients with PH. However, RV remodeling in PH is not isolated to the right ventricle alone and affects left ventricular (LV) structure and function through septal shift and ventricular interactions. High afterload from PH results in RV hypertrophy and enlargement, which may result in decreases of LV volume from right-to-left septal shift, as they share the common space in pericardium. RV volumes by cardiac magnetic resonance (CMR) imaging or three-dimensional echocardiography and RV speckle-tracking strain have emerged as having additional prognostic utility. However, CMR imaging or three-dimensional and strain echocardiography may not be as widely available in many clinical situations. Accordingly, the aim of this study was to test the hypothesis that a simplified two-dimensional echocardiographic biventricular remodeling index, the ratio of RV end-diastolic area to LV end-diastolic area, has prognostic utility for predicting mortality in patients with PH. An additional aim was to compare the prognostic utility of the RV-to-LV end-diastolic area ratio with that of other recently reported echocardiographic measures of outcomes, including RV speckle-tracking strain, in the same patients with PH.
Methods
Study Population
The study population was prospectively selected from 179 consecutive patients referred for the evaluation of confirmed PH who underwent routine two-dimensional echocardiography at our institution. The present research study was prospective in design and received institutional review board approval to access an existing deidentified database registry with waiver of consent. All patients had mean pulmonary artery ssures > 25 mm Hg confirmed by invasive right heart catheterization. To focus on prognosis in PH, we excluded six patients with LV ejection fractions < 50% and three patients with significant valvular heart disease. Ten patients with congenital heart disease in whom imaging of both the right and left ventricles were incomplete were excluded (single ventricles, transposition of great arteries, large ventricular septal defects, or uncorrected tetralogy of Fallot). Twenty-one patients (12%) were also excluded because they had echocardiographic data not suitable for quantitative analysis, including speckle-tracking strain analysis. Accordingly, the study cohort consisted of 139 patients with PH. All belonged to World Health Organization (WHO) class I classification of precapillary pulmonary arterial hypertension : 68 (49%) with idiopathic PH, 49 (33%) with collagen vascular disease, 11 (8%) with portal hypertension, eight (6%) with congenital heart disease (atrial septal defects or patent ductus arteriosus), two (1%) with familial PH, and one (1%) with drug-induced PH. Twenty-two normal healthy volunteers with normal history and physical examination findings and normal results on complete two-dimensional and Doppler echocardiographic examinations were included as a control group. The outcome event was predefined as all-cause death over 5 years from the date of echocardiography.
Routine and Speckle-Tracking Echocardiography
Two-dimensional echocardiography was done using GE Vivid 7 (GE Vingmed Ultrasound AS, Horten, Norway) or Philips iE33 (Philips Medical Systems, Andover, MA) systems with data exported as Digital Imaging and Communications in Medicine data for offline analysis. Standard echocardiographic parameters and speckle-tracking strain were analyzed using TomTec software (2D Cardiac Performance Analysis version 4.5.2.5; TomTec, Munich, Germany). LV ejection fraction was calculated using the biplane Simpson rule. A four-chamber view focused on the right ventricle including the RV free wall was used to measure end-diastolic and end-systolic RV area and to analyze RV strain. Endocardial RV tracing of the wall was performed to carefully exclude trabeculations. LV area was measured from the same four-chamber view if possible ( Figure 1 ) or from an alternative four-chamber view if the left ventricle was foreshortened and out of the imaging sector because of RV dilatation, which occurred in 29% of patients. The RV-to-LV end-diastolic area ratio was prespecified as the index of biventricular remodeling to be tested because of the previously validated association of RV end-diastolic volumes with invasively determined pulmonary vascular resistance (PVR). To compare other established measures of RV function in PH, tricuspid annular plane systolic plane excursion (TAPSE), RV fractional area change (FAC), and LV eccentricity index at end-systole were measured accordingly to guidelines of the American Society of Echocardiography. For speckle-tracking strain analysis, regions of interest were placed on the RV free wall and septum as previously described, with manual adjustment throughout the cardiac cycle to optimize tracking. Longitudinal strain of the RV free wall (averaged from three segments) and global RV (averaged line from six segments) were used as quantitative parameters of RV function.
To determine an appropriate cutoff value of RV-to-LV end-diastolic area ratio, a subgroup of 50 patients who underwent invasive hemodynamic studies mostly on the same day or within 1 day of their echocardiographic examinations was used. PVR > 3 Wood units was used as an established marker of disease severity using the formula PVR = (mean pulmonary artery pressure − pulmonary capillary wedge pressure)/cardiac output.
Reproducibility
Intraobserver and interobserver variability of RV-to-LV end-diastolic area ratio and RV speckle-tracking strain was evaluated in 10 randomly selected patients with PH. The variability was calculated using the intraclass correlation coefficient.
Statistical Analysis
Results are expressed as mean value ± SD if the distribution was normal. Unpaired Student’s t tests were used to compare values between the normal control and PH groups. Data with skewed distributions are expressed as median and interquartile range and were compared using Mann-Whitney U tests. The relationship among the indices was evaluated using appropriate regression analysis, either the Pearson correlation test or the Spearman test. Multiple regression stepwise analysis was used to identify the echocardiographic parameter most related to hemodynamics. We used receiver operating characteristic curve analysis to determine optimal cutoff values of continuous variables and defined the cutoff values as the point of highest sum of sensitivity and specificity. Kaplan-Meier survival analysis and the log-rank test were used to compare between two groups. Multivariate Cox regression analysis was used to identify the predictor associated with the death at 5 years. Age, gender, WHO functional class, and echocardiographic parameters with P values > .10 in univariate analysis were used in the multivariate Cox regression analysis. The incremental value of echocardiographic parameters to clinical features for evaluation of the risk for death at 5 years was calculated with the improvement in the global χ 2 statistic. A two-tailed P value < .05 was considered to indicate statistical significance. Statistical analysis was performed with standard software (MedCalc version 10.4.0.0; MedCalc Software, Mariakerke, Belgium).
Results
Baseline clinical and echocardiographic characteristics in 139 patients with PH and 22 normal subjects are shown in Table 1 . Patients with PH tended to be older, but the gender distribution was similar in both groups. As expected, RV end-diastolic area was larger and LV end-diastolic area was smaller in patients with PH than normal subjects (mean RV end-diastolic area, 28 ± 9 vs. 19 ± 4 mm 2 [ P < .0001] mean LV end-diastolic area, 25 ± 7 vs 29 ± 4 mm 2 [ P = .009]); therefore, the RV-to-LV end-diastolic area ratio was higher in patients with PH (1.03 vs 0.67, P < .0001). Regarding tricuspid regurgitation (TR) in patients with PH, 42 (30%) had no or trace TR, 36 (26%) had mild TR, 55 (40%) had moderate TR, and six (4%) had severe TR. All normal control subjects had only trace TR. Using the invasive hemodynamic subgroup ( n = 50), we examined our predefined criterion of RV-to-LV end-diastolic area ratio determined by receiver operating characteristic analysis and found that 0.93 was most closely related to PVR > 3 Wood units (area under the curve, 0.87; sensitivity, 74%; specificity, 86%).
| Variable | Normal subjects ( n = 22) | Patients with PH ( n = 139) | P |
|---|---|---|---|
| Age (y) | 40 ± 17 | 55 ± 15 | .050 |
| Women | 16 (73%) | 105 (76%) | .986 |
| Echocardiographic parameters | |||
| FAC (%) | 41 ± 7 | 30 ± 10 | <.0001 |
| TAPSE (mm) | 24 ± 4 | 17 ± 5 | <.0001 |
| LV eccentricity index | 0.99 (0.96-1.07) | 1.40 (1.11-1.80) | <.0001 |
| RV end-diastolic area (cm 2 ) | 19 ± 4 | 28 ± 9 | <.0001 |
| RV end-systolic area (cm 2 ) | 11 ± 3 | 20 ± 8 | <.0001 |
| LV end-diastolic area (cm 2 ) | 29 ± 4 | 25 ± 7 | .009 |
| LV end-systolic area (cm 2 ) | 16 ± 2 | 14 ± 4 | .022 |
| LV ejection fraction (%) | 63 ± 5 | 66 ± 7 | .135 |
| RV-to-LV area ratio, diastole | 0.67 (0.61-074) | 1.06 (0.76-1.57) | <.0001 |
| RV-to-LV area ratio, systole | 0.75 (0.59-0.85) | 1.40 (0.90-2.31) | <.0001 |
| Speckle-tracking echocardiographic parameters | |||
| RV global strain (%) | −21 ± 4 | −15 ± 5 | <.0001 |
| RV free wall strain (%) | −24 ± 5 | −15 ± 7 | <.0001 |
Inter- and Intraobserver Variability
The intraclass correlation coefficients for intraobserver reproducibility were 0.97 (95% CI, 0.88-0.99) for RV end-diastolic area, 0.99 (95% CI, 0.92-0.99) for LV end-diastolic area, and 0.92 (95% CI, 0.69-0.98) for RV global strain. The intraclass correlation coefficients for interobserver reproducibility were 0.95 (95% CI, 0.76-0.99) for RV end-diastolic area, 0.97 (95% CI, 0.88-0.99) for LV end-diastolic area, and 0.95 (95% CI, 0.82-0.99) for RV global strain.
Associations of Biventricular Remodeling with Clinical Outcomes
The median follow-up period was 4.5 years (interquartile range, 2.0-6.5 years). There were 69 deaths over 5 years. Sixty-one deaths (88%) were considered PH related from RV failure, sudden cardiac death, or perioperative death after lung transplantation. Of the 19 patients who underwent lung transplantation, three subsequently died of postoperative complications. Eight deaths (12%) were of noncardiac causes: five (7%) of primary liver disease and complications of liver failure, two (3%) of metastatic lung cancer, and one (1%) of complications related to renal failure. Four patients (3%) were lost to follow-up, with data censored at the time of last follow-up.
Using a biventricular remodeling index cutoff value of RV-to-LV end-diastolic area ratio of 0.93, there were 85 patients (61%) with ratios ≥ 0.93 and 54 patients (39%) with ratios < 0.93 at baseline. Kaplan-Meier analysis demonstrated a significant association of RV-to-LV end-diastolic area ratio ≥ 0.93 with all-cause mortality (hazard ratio [HR], 1.84; 95% CI, 1.14-3.09; P = .019) ( Figure 2 ). By post hoc analysis, RV end-diastolic area with 25 cm 2 as the cutoff value reached borderline statistical significance ( P = .046) (HR, 1.70; 95% CI, 1.01-2.85), and accordingly, the RV-to-LV end-diastolic area ratio was comparatively more closely associated with survival. RV global longitudinal strain in the same patients was also significantly associated with survival using a cutoff of ≥−15% (HR, 1.66; 95% CI, 1.03-2.67; P = .044) ( Figure 3 ). RV longitudinal free wall strain with a cutoff value of ≥−15% was not significantly associated with survival (HR, 1.26; 95% CI, 0.79-2.02; P = .338). The associations of all-cause mortality with age, gender, WHO functional class, TR grade, cardiac index, TAPSE, RV FAC, and LV eccentricity index using univariate and multivariate Cox regression analysis are shown in Table 2 . Using multivariate analysis, only age and RV-to-LV end-diastolic area ratio had associations with significantly higher risk for death (age: HR, 1.02 [95% CI, 1.01-1.04; P = .006]; RV-to-LV end-diastolic area ratio: HR, 1.83 [95% CI, 1.03-3.25; P = .040]). To examine potential differences in patients who died early versus later, we subdivided our patients into two groups on the basis of time: those who died within 2 years and those who died after 2 years (up to 5 years). The RV-to-LV end-diastolic area ratio was slightly greater in the patients who died within 2 years (1.33 ± 0.54 vs 1.23 ± 0.47), but this trend was not significant ( P = .426). There were no clinical characteristics or echocardiographic parameters that showed significant differences between these two subgroups of patients who died early versus later. The addition of RV-to-LV end-diastolic area ratio to the clinical features (age, gender, and WHO functional class) showed incremental prognostic value for determining risk for death over 5 years in patients with PH, while other the other RV functional parameters did not in our patient population ( Figure 4 ).
