Background
Transthoracic echocardiographic (TTE) imaging is the mainstay of clinical practice for evaluating right ventricular (RV) size and function, but its accuracy in patients with pulmonary hypertension has not been well validated.
Methods
Magnetic resonance imaging (MRI) and TTE images were retrospectively reviewed in 40 consecutive patients with pulmonary hypertension. RV and left ventricular volumes and ejection fractions were calculated using MRI. TTE areas and indices of RV ejection fraction (RVEF) were compared.
Results
The average age was 42 ± 12 years, with a majority of women (85%). There was a wide range of mean pulmonary arterial pressures (27–81 mm Hg) and RV end-diastolic volumes (111–576 mL), RVEFs (8%–67 %), and left ventricular ejection fractions (26%–72%) by MRI. There was a strong association between TTE and MRI-derived parameters: RV end-diastolic area (by TTE imaging) and RV end-diastolic volume (by MRI), R 2 = 0.78 ( P < .001); RV fractional area change by TTE imaging and RVEF by MRI, R 2 = 0.76 ( P < .001); and tricuspid annular plane systolic excursion by TTE imaging and RVEF by MRI, R 2 = 0.64 ( P < .001). By receiver operating characteristic curve analysis, an RV fractional area change < 25% provided excellent discrimination of moderate systolic dysfunction (RVEF < 35%), with an area under the curve of 0.97 ( P < .001). An RV end-diastolic area index of 18 cm 2 /m 2 provided excellent discrimination for moderate RV enlargement (area under the curve, 0.89; P < .001).
Conclusions
Echocardiographic estimates of RV volume (by RV end-diastolic area) and function (by RV fractional area change and tricuspid annular plane systolic excursion) offer good approximations of RV size and function in patients with pulmonary hypertension and allow the accurate discrimination of normal from abnormal.
The assessment of right ventricular (RV) size and function plays an important role in the management of patients with pulmonary hypertension (PH). On two-dimensional (2D) transthoracic echocardiographic (TTE) imaging, RV volume is often estimated using linear dimensions or RV area in the apical four-chamber view. Commonly used 2D indices of RV systolic function include RV fractional area change (RVFAC) and tricuspid annular plane systolic excursion (TAPSE). In contrast to the left heart, volumetric estimates using 2D methods are more difficult because of the complex geometry of the right ventricle. We aimed to examine the value of simple 2D RV indices as measures of RV systolic function in patients with PH. Although often used clinically in PH and associated with outcomes, these estimates of RV size and function have not been well validated compared to magnetic resonance imaging (MRI) reference standards.
In a recent study, Sato et al . found TAPSE superior to RVFAC, which is in contrast to the MRI-based findings of Kind et al ., who showed that RVFAC is superior. The controversy surrounding the utility of these parameters prompted us to examine the value of 2D echocardiography in patients with PH and compare these conventional simple measures with RV function by MRI. We used receiver operating characteristic (ROC) curve analysis to identify echocardiographic thresholds to discriminate the presence of moderate RV systolic dysfunction.
Methods
Patient Selection
A total of 45 patients with PH who underwent MRI and echocardiography between May 2007 and May 2012 were retrospectively considered for inclusion in the study. The diagnosis of PH was established using the World Health Organization criteria defined by a mean pulmonary arterial pressure > 25 mm Hg and a pulmonary capillary wedge pressure < 15 mm Hg. The majority had World Health Organization group I pulmonary arterial hypertension ( n = 42), and three had World Health Organization group IV chronic thromboembolic PH. The etiology of PH was determined per published guidelines. A total of five patients were excluded from the study, three because of complex congenital heart disease and prior cardiac surgery, one because of atrial fibrillation, and one because of suboptimal quality of echocardiographic images. Patients with coronary artery disease and prior infarctions were not specifically excluded, although they were not represented in this sample, because all were regularly followed in our PH center and none had risk factors for coronary artery disease. Thirty-six patients (90%) underwent echocardiography on the same day as MRI (within 3 hours), and four patients underwent echocardiography within 3 weeks (stable clinical status without any change in therapy). After obtaining approval from the institutional review board, we also reviewed patients’ medical records for evaluation of patient characteristics and right heart catheterization data.
MRI
Cardiac MRI scans were completed on the hospital’s clinical scanner at 1.5 T (TwinSpeed; GE Healthcare, Little Chalfont, United Kingdom). Images were acquired supine using an eight-channel cardiac phased-array coil. All acquisitions were electrocardiographically gated and obtained during breath holding. Using initial three-plane-localizer sequences, a short-axis stack of transverse slices spanning the volume of the left and right ventricles was obtained. Standard two-chamber, three-chamber, and four-chamber views as well as RV two-chamber and three-chamber views were prescribed.
Image assessment was performed offline on an independent workstation using commercially available software. Medis software (Medis Medical Imaging Systems, Inc, Leiden, The Netherlands) was used for volume and function quantification. Ventricular volumes were calculated using manual contour tracing of short-axis slices from the apex to the base, including the left ventricular (LV) outflow tract and outflow tract but excluding all trabeculations and papillary muscles. RV three-chamber, LV three-chamber, and four-chamber views were used for verification of the exact localization of the valves to allow accurate segmentation of the tricuspid valve plane and inclusion of the outflow tracts. Multiphase scans in movie play mode were used for contour tracing guidance. The ejection fractions of the right and left ventricles were then derived from the chamber volumes at end-diastole and end-systole, as defined by the largest and smallest volumes, respectively, before the septal shift, when this was observed.
AquariusNet (TeraRecon, Inc, Foster City, CA) was used for chamber areas and TAPSE measurements. Using the standard four-chamber view, LV area was manually contoured at end-diastole, and RV area was contoured at both end-diastole and end-systole, as defined above (see Figure 1 ). TAPSE was performed by placing the cursor on the junction of the tricuspid valve plane and the RV free wall in end-diastole, changing the view to the predefined end-systolic phase, and extending a line to the same anatomic location. RVFAC was calculated as the percentage change in RV area between systole and diastole. Three consecutive measurements of RVFAC were averaged. All MRI measurements were performed by a single experienced cardiologist who was blinded to the TTE data and measurements. A second experienced cardiologist blinded to the initial measurements evaluated 10 patients selected at random for interobserver variability. The measurements were derived independently without knowledge of which frames were analyzed by the primary cardiologist.
Echocardiography
All TTE studies were performed at the Stanford Hospital and Clinics echocardiography laboratory using a Philips iE33 ultrasound system (Philips Medical Systems, Andover, MA) and a Philips S5-1 transducer (5-MHz to 1-MHz extended operating frequency range, harmonic imaging). Standard TTE views were obtained, and digitized images were analyzed by a reader blinded to the MRI data and measurements. A second experienced cardiologist blinded to the initial measurements evaluated 10 patients selected at random for interobserver variability, as for MRI. All measurements were averaged over three cardiac cycles.
LV end-diastolic and end-systolic diameters were measured according to American Society of Echocardiography recommendations using the 2D linear method from the parasternal long-axis view. Because of the presence of interventricular dyssynchrony observed in PH, the LV end-diastolic diameter measures were standardized to the beginning of the QRS complex, and LV end-systolic diameter was standardized at aortic valve closure, which in the majority of cases does not correspond to the smallest LV diameter.
Measures of the RVOT were performed in the parasternal long-axis view and the basal parasternal short-axis view ( Figure 2 ). Additional linear measures of the right ventricle were performed in the apical four-chamber view, including basal radial diameter, mid radial diameter, and RV length ( Figure 2 ). RV and LV end-diastolic and end-systolic areas were measured from a standard apical four-chamber view using the largest and smallest area, before the septal shift ( Figure 2 ). Three consecutive measurements of RVFAC were averaged. RVFAC was calculated as relative systolic-to-diastolic RV area change. All measurements excluded trabeculations. To ensure that a standard four-chamber view was used, maximal RV diastolic diameter was measured in the apical four-chamber view and found to be within 15% of maximal diastolic diameter in the short axis. TAPSE was measured using a 2D method as the total displacement of the tricuspid annulus from end-diastole to end-systole. For TAPSE, the 2D apical four-chamber view was used, the cursor was placed on the junction of the tricuspid valve plane and the RV free wall in end-diastole, and after scrolling to end-systole, a line was extended to the same anatomic location. This was done in the same manner as for MRI. We used the 2D method for TAPSE measurements because measures based on M-mode imaging were not acquired in five patients. In the remainder of patients, the correlation between 2D and M-mode-derived measures of TAPSE was 0.94 ( P < .001). Fractional TAPSE was calculated as TAPSE divided by RV length. Fractional radial shortening of the right ventricle was measured using the relative change in end-diastolic to end-systolic diameter change for the apical, mid, and basal measurements.
Tricuspid regurgitation was graded as severe if at least two of the following criteria were met: vena contracta width > 0.7 cm, jet area > 10 cm 2 for central jets at a Nyquist limit of 50 to 60 cm/sec with a relative jet to atrial size > 40%, and systolic reversal in the hepatic veins. RV systolic pressure was estimated by maximal velocity of the tricuspid regurgitation Doppler signal with right atrial pressure estimations made on the basis of the collapse index and size of the inferior vena cava.
Statistical Analysis
All statistical analyses were performed using PASW Statistics version 18 (SPSS, Inc, Hong Kong, China). Multiple regression analysis was used to determine the relationship between echocardiographic and MRI-derived indices of RV and LV size and function. Bland-Altman analysis was used to determine the mean difference between MRI-derived and echocardiographic measures of LV volumes and LV ejection fraction (LVEF) and linear and 2D measures of RV size and function. ROC curves were constructed using MedCalc (MedCalc Software, Mariakerke, Belgium), with cutoff values selected on the basis of existing guidelines. P values < .05 were considered statistically significant. Interobserver variability was analyzed using intraclass correlation coefficients (ICCs).
Results
Patient Characteristics
A total of 40 patients with PH were included in the study. The clinical characteristics, etiology of PH, relevant laboratory values, invasive hemodynamics, and medication profile are noted in Table 1 . MRI and TTE parameters are listed in Table 2 .
| Variable | Value |
|---|---|
| Age (y) | 42 ± 12 (19–70) |
| Men | 6 (15%) |
| Etiology | |
| Idiopathic | 18 (45%) |
| Associated with connective tissue disease | 8 (20%) |
| Associated with congenital heart diseases | 5 (13%) |
| Drugs and toxin associated | 5 (13%) |
| Chronic thromboembolic PH | 3 (8%) |
| Associated with HIV infection | 1 (3%) |
| Hemodynamics | |
| Systolic systemic blood pressure (mm Hg) | 114 ± 19 (78–172) |
| Heart rate (beats/min) | 74 ± 14 (49–107) |
| Mean right atrial pressure (mm Hg) | 8.6 ± 5.4 (1–21) |
| Pulmonary capillary wedge pressure (mm Hg) ( n = 39) | 9.7 ± 4.9 (3–15) |
| Mean pulmonary artery pressure (mm Hg) ( n = 38) | 46.6 ± 15.0 (27–81) |
| Cardiac index (L/min/m 2 ) | 2.02 ± 0.57 (0.98–3.66) |
| MRI | |
| RVEDV (mL) | 231 ± 101 (111–576) |
| RVEF (%) | 35 ± 15 (8–67) |
| LVEF (%) | 57 ± 10 (26–72) |
| LV stroke volume | 61 ± 25 (13–152) |
| Laboratory ( n = 39) | |
| Serum creatinine (mg/mL) | 1.0 ± 0.6 (0.4–6.0) |
| NT-BNP (ng/L) | 2,027 ± 4,253 (30–24,790) |
| Hemoglobin (g/dL) | 13.5 ± 2.3 (8.4–18) |
| Medications | |
| Prostacycline | 25 (63%) |
| Phosphodiesterase inhibitors | 18 (45%) |
| Endothelin receptor antagonists | 6 (15%) |
| Relationship between RVEDV and echocardiographic parameter | Coefficient of determination ( R 2 ) | P | Regression equation (selected relationship) |
|---|---|---|---|
| Linear dimensions | |||
| RVOT1 | 0.57 | <.001 | RVEDV = 11 × RVOT − 205 |
| RVOT2 | 0.47 | <.001 | — |
| RVOT3 | 0.25 | <.001 | — |
| RVEDD basal | 0.57 | <.001 | RVEDV = 8.5 × RVEDD basal – 87 |
| RVEDD mid | 0.65 | <.001 | RVEDV = 8.4 × RVEDD mid − 147 |
| RV height | 0.38 | <.001 | — |
| RV areas | |||
| RVEDA | 0.78 | <.001 | RVEDV = 8.2 × RVEDA − 40 |
| Relative RVEDA (RVEDA/LVEDA) | 0.40 (R2= 0.82 for relative RVEDV) | <.001 | RVEDV/LVEDV = 1.5 × (RVEDA/LVEDA) |
Relationship between Echocardiographic and MRI Measures of RV Size
Table 2 and Figure 3 summarize the relationship between echocardiographic and MRI measures of RV size. The best echocardiographic estimation of RV end-diastolic volume (RVEDV) was the RV end-diastolic area (RVEDA) measurement ( R 2 = 0.78, P < .001). Basal and midventricular measures of the right ventricle also provide reasonable estimation of RVEDV. Relative RV area provides a strong approximation of relative RV to LV end-diastolic volumes but only a moderate relationship with RVEDV. Correlations between RVEDA and RVEDV were statistically different compared with correlations between RVEDV and RVOT2 ( P = .019), RVOT3 ( P < .001), RV height ( P = .005), and relative RVEDA ( P = .006). See Figure 2 for details on RVOT measures. There was also good correlation of RV end-systolic volume and RV end-systolic area ( R 2 = 0.75, P < .001). To illustrate the ability of RVEDA to detect RV dilatation (MRI RVEDV index > 104 mL/m 2 for women and > 115 mL/m 2 for men) by echocardiographic RVEDA index, a ROC curve was constructed ( Figure 4 A), with an area under curve of 0.89 ( P < .001). An indexed RVEDA of 17.7 cm 2 /m 2 had the highest specificity (87%) and sensitivity (77%) for detecting RV dilation.
