The right ventricle is a complex structure that is challenging to quantify by two-dimensional (2D) echocardiography. Unlike disk summation three-dimensional (3D) echocardiography (3DE), single-beat 3DE can acquire large volumes at high volume rates in one cardiac cycle, avoiding stitching artifacts or long breath-holds. The aim of this study was to assess the accuracy and test-retest reproducibility of single-beat 3DE for quantifying right ventricular (RV) volumes in adult populations of acquired RV pressure or volume overload, namely, pulmonary hypertension (PH) and carcinoid heart disease, respectively. Three-dimensional and 2D echocardiographic indices were also compared for identifying RV dysfunction in PH.
A prospective cross-sectional study was performed in 100 individuals who underwent 2D echocardiography, 3DE, and cardiac magnetic resonance imaging: 49 patients with PH, 20 with carcinoid heart disease, 11 with metastatic carcinoid tumors without cardiac involvement, and 20 healthy volunteers. Two operators performed test-retest acquisition and postprocessing for inter- and intraobserver reproducibility in 20 subjects.
RV single-beat 3DE was attainable in 96% of cases, with mean volume rates of 32 to 45 volumes/sec. Bland-Altman analysis of all subjects (presented as mean bias ± 95% limits of agreement) revealed good agreement for end-diastolic volume (−2.3 ± 27.4 mL) and end-systolic volume (5.2 ± 19.0 mL) measured by 3DE and cardiac magnetic resonance imaging, with a tendency to underestimate stroke volume (−7.5 ± 23.6 mL) and ejection fraction (−4.6 ± 13.8%) by 3DE. Subgroup analysis demonstrated a greater bias for volumetric underestimation, particularly in healthy volunteers (end-diastolic volume, −11.9 ± 18.0 mL; stroke volume, −11.2 ± 20.2 mL). Receiver operating characteristic curve analysis showed that 3DE-derived ejection fraction was significantly superior to 2D echocardiographic parameters for identifying RV dysfunction in PH (sensitivity, 94%; specificity, 88%; area under the curve, 0.95; P = .031). There was significant interobserver test-retest bias for RV volume underestimation (end-diastolic volume, −12.5 ± 28.1 mL; stroke volume, −10.6 ± 23.2 mL).
Single-beat 3DE is feasible and clinically applicable for volumetric quantification in acquired RV pressure or volume overload. It has improved limits of agreement compared with previous disk summation 3D echocardiographic studies and has incremental value over standard 2D echocardiographic measures for identifying RV dysfunction. Despite the ability to obtain and postprocess a full-volume 3D echocardiographic RV data set, the quality of the raw data did influence the accuracy of the data obtained. The technique performs better with dilated rather than nondilated RV cavities, with a learning curve that might affect the test-retest reproducibility for serial RV studies.
Quantification of right ventricular (RV) size and function is prognostic in congenital and acquired heart disease. The most convenient imaging modality for assessing the right ventricle is two-dimensional (2D) echocardiography (2DE). However, this is limited by the crescentic RV chamber shape and complex geometry, with inflow and outflow portions in different planes. Thus, cardiac magnetic resonance imaging (CMRI) has become the gold-standard imaging modality for RV quantification. Unfortunately, CMRI is expensive, time consuming, and of limited availability compared with echocardiography.
One possibility to overcome the limitations of 2DE is three-dimensional (3D) echocardiography (3DE), compared against CMRI in a range of congenital and acquired diseases for RV volumetric quantification. Three-dimensional echocardiography traditionally uses the disk summation method to reconstruct the right ventricle after sequential slice acquisition over consecutive electrocardiographically gated heartbeats. This technique, however, is limited by breath holding throughout successive cardiac cycles, stitching artifacts during acquisition, and difficulties identifying inlet and outflow regions in the basal slices during postprocessing.
More recently, ultrasound transducer technology allows the real-time acquisition of a 90° × 90° data set in a single cardiac cycle. We therefore compared RV volumetric quantification by single-beat full-volume 3DE against CMRI in homogenous patient populations of acquired RV pressure and volume overload, namely, pulmonary hypertension (PH) and carcinoid heart disease, respectively. We also sought to determine the potential incremental value of 3DE versus 2DE in PH and to evaluate the test-retest reproducibility of 3DE for both the acquisition and postprocessing components.
We performed a prospective cross-sectional study that enrolled 100 participants in sinus rhythm with no contraindications to magnetic resonance imaging, all of whom underwent comprehensive 2DE, single-beat 3DE of the right ventricle, and CMRI within 2 hours of one another. The participants were divided into four subgroups:
A group of 49 consecutive patients with PH (diagnosed by right heart catheterization as a mean pulmonary artery pressure >25 mm Hg and a pulmonary capillary wedge pressure <15 mm Hg ) who presented for diagnosis and/or follow-up of PH by clinical evaluation and/or right heart catheterization as a disease model of RV pressure overload. The etiologies of PH included idiopathic ( n = 9), connective tissue disease associated ( n = 32), and chronic thromboembolic disease ( n = 8). Exclusion criteria comprised clinically significant restrictive or obstructive lung disease identified by pulmonary function tests, arrhythmia, and known independent left-sided cardiac disease unrelated to PH.
A group of 20 consecutive patients undergoing 2DE for diagnosis and/or follow-up of carcinoid heart disease were studied as a disease model of RV volume overload.
A control group of 20 healthy volunteers affiliated with our institution who were age and sex matched to the PH group.
A control group of 11 age- and sex-matched patients with metastatic neuroendocrine tumor who were screened as negative for carcinoid valvular heart disease.
All control participants were eligible for study inclusion if they had no cardiac symptomatology, had no medical histories of cardiac disease including hypertension, and were not taking any cardiac medications. Normal 2D transthoracic echocardiographic findings were also required to exclude any occult structural cardiac disease before study inclusion.
The institutional research ethics committee approved the study, and informed written consent was obtained from all patients and control subjects.
All patients underwent comprehensive 2D and Doppler transthoracic echocardiography in the left lateral decubitus position using the Acuson Siemens SC2000 cardiac ultrasound system (Siemens Healthcare, Erlangen, Germany), with a 4V1c transducer (frequency bandwidth, 1.25–4.5 MHz). A standard study protocol was followed in conjunction with American Society of Echocardiography guidelines for chamber quantification and the British Society of Echocardiography guidelines for PH assessment as appropriate. RV function was assessed using M-mode tricuspid annular plane systolic excursion (TAPSE); RV fractional area change, calculated as [(end-diastolic area) − (end-systolic area)/end-diastolic area] × 100; and mean RV free wall peak systolic strain using syngo Vector Velocity Imaging (Siemens Medical Solutions USA, Inc, Mountain View, CA). A three-beat 2D echocardiographic digital clip of an apical four-chamber view optimized for RV visualization was acquired and exported to Velocity Vector Imaging, and 10 to 15 endocardial points were plotted in end-systole from the lateral to the medial tricuspid annulus. The adequacy of speckle-tracking was visually checked and manually adjusted as required.
Single-beat full-volume 3D echocardiographic RV data sets were acquired using the 4Z1c matrix-array transducer (frequency bandwidth, 1.5–3.5 MHz; maximum depth, 30 cm; maximum field of view, 90° × 90°). Probe position started from the apical four-chamber view with the patient in the left lateral decubitus position. Both the patient and transducer positions were subsequently modified for optimal simultaneous visualization of the tricuspid valve, cardiac apex, infundibulum, and RV outflow tract (RVOT) as assessed by the real-time 2D four-chamber, basal sagittal, and coronal views, and by inclusion of the RV chamber in the pyramidal data set. In our experience, a more lateral apical window with posterior tilt of the probe tail was beneficial to visualize the infundibulum and RVOT in the coronal window. Image depth and sector width were adjusted for maximal visualization of the right ventricle at the highest volume rate. At least three 3D echocardiographic RV data sets were acquired during a breath-hold to ensure optimal image quality, which was subjectively graded on a five-point scale ranging from zero (very poor) to four (perfect). A score of two or less was attributed if ultrasound dropout was evident in greater than half of the RVOT border.
Full-volume 3D echocardiographic RV data sets were imported into the on-cart RV Analysis application. Manual adjustment of the RV data set was initially required to ensure the correct orientation of four-chamber, sagittal, and coronal slices; maximize the RV cavity area and identify the most apical RV view on visual assessment of the four-chamber window; and allow the identification of cardiac landmarks. This process was performed in a stepwise approach by rotation and angulation of the four-chamber window, with manipulation of this plane causing the simultaneous adjustment of the other two (sagittal and coronal) orthogonal planes ( Figure 1 A). Both atrioventricular valves followed by the left ventricular apex were identified as anatomic landmarks. When the apex of a dilated right ventricle overrode that of the left ventricle, the most apical cardiac point was identified with the left ventricular apex marker. End-diastolic and end-systolic frames were assigned by visual identification of the largest and smallest RV four-chamber areas, respectively.
Endocardial RV borders were traced at end-diastole and end-systole in four-chamber, basal sagittal, and coronal views. The software algorithm obliges the operator to intersect the endocardial border tracing in sequential views with crosshair reference markers that are positioned in response to endocardial border traces from a preceding view. Therefore, correction of a previous slice tracing was undertaken when a crosshair position suggested a prior tracing error. Trabeculae were included in the blood pool volume. To assist with RVOT delineation in the basal sagittal view, the insertion point of the RV myocardium at the interventricular septum was routinely included in the endocardial tracing.
At the final stage, the algorithm presents the results of semiautomated contour tracking for the four-chamber, coronal and basal, middle and apical short-axis views. Misalignment of endocardial contours prompted identification of the region of suboptimal tracking followed by manual correction of the original tracing. Automated volumetric reconstruction was accepted only once the semiautomated endocardial border tracking was visually satisfactory and represented meaningful RV shapes in all views ( Figure 1 B), as optimization of this final reconstruction stage significantly affects the results generated. The algorithm from which the final RV volume is generated has been previously described.
Test-Retest Reproducibility of 3DE
Reproducibility was studied in 20 randomly selected subjects (14 with PH, one with carcinoid heart disease, and five healthy volunteers) for both the 3D echocardiographic acquisition and postprocessing stages by two independent sonographers (D.S.K. and A.E.G.), as described previously. The two sonographers had equal experience with 2DE but differing levels of experience with 3D echocardiographic RV full-volume acquisition (10 and 3 months, respectively). Sonographer 1 (D.S.K.) obtained a 3D echocardiographic RV data set, after which sonographer 2 (A.E.G.) independently obtained a 3D echocardiographic RV data set. Then, sonographer 1 acquired a second separate 3D echocardiographic RV data set. The sonographers, who were blinded to each other’s results, performed postprocessing of their own 3D echocardiographic RV data sets. Data sets for intraobserver test-retest reproducibility were postprocessed separately at time intervals of >2 weeks.
All cardiac magnetic resonance images were acquired using a 1.5-T magnetic resonance scanner (Avanto; Siemens Healthcare) using a 12-element phased-array coil for signal reception and the body coil for signal transmission. A vector electrocardiographic system was used for cardiac gating. Ventricular volumes and great vessel flow were measured in all patients. Volumetric RV data were obtained using either retrospectively gated balanced steady-state free precession ( n = 19) cine imaging of contiguous short-axis slices or real-time radial k – t sensitivity encoding imaging ( n = 81) of contiguous transaxial slices depending on the pathology under investigation and the patient’s ability to hold his or her breath. Real-time radial k – t sensitivity imaging allows the collection of high spatiotemporal resolution real-time images during free breathing and is part of the standard clinical CMRI work flow at our institution in the pediatric PH population. Transaxial RV slices were preferred for the PH cohort and their respective control population because of the relative preservation of longitudinal versus radial RV function that is manifest in this condition. Blood flow data were acquired in the ascending aorta, in the right and left branch pulmonary arteries, and at the level of the atrioventricular valves using a velocity-encoded prospectively triggered spiral phase-contrast magnetic resonance flow sequence. This provided an internal check for the RV volumetric data.
All image postprocessing was performed using “in-house” plug-ins for the open-source OsiriX Digital Imaging and Communications in Medicine software. Endocardial RV borders were traced manually at end-diastole and end-systole, the time points of which were identified by the largest and smallest RV cavity areas, respectively. The inclusion of RV trabeculae was the same as that performed by 3D echocardiographic postprocessing. Ventricular stroke volume (SV) was the difference between end-diastolic volume (EDV) and end-systolic volume (ESV), and ejection fraction (EF) was calculated as (SV/EDV) × 100. Phase-contrast magnetic resonance flow data were segmented using a semiautomatic vessel edge detection algorithm with manual operator correction. The CMRI data sets for the patients who underwent 3D echocardiographic test-retest reproducibility scans were also tested for interobserver (D.S.K. and M.A.Q.) and intraobserver postprocessing reproducibility.
Statistical analysis was performed using SPSS version 21.0 (IBM Corporation, Armonk, NY) and Prism 6.0b for Mac (GraphPad Software, Inc, La Jolla, CA). Normally distributed continuous data were expressed as mean ± SD. Systematic differences between measurements were evaluated with Student paired t tests (two tailed), with Pearson correlation coefficients used to assess the relationship between 3DE- and CMRI-derived RV volumes and EF. Differences between the four participant subgroups were analyzed using one-way analysis of variance, with the Tukey post hoc tests identifying which specific means differed. P values < .05 were considered statistically significant. Image scoring data were nonparametrically distributed, represented by medians with 25th and 75th percentiles. Rank sum tests were used for comparisons of image scoring data, with the Mann-Whitney U test and the Kruskal-Wallis test used for comparisons of two and three independent groups, respectively.
Intermodality, interobserver, and intraobserver agreement was studied using the Bland-Altman method, whereby the mean difference was presented as the bias and 95% limits of agreement around the bias expressed as the mean difference ± 1.96 SDs. Differences between test-retest measurements were analyzed by one-way repeated measures analysis of variance, with the Bonferroni post hoc test identifying which specific means differed. The Greenhouse-Geisser correction was used if the assumption of sphericity had been violated. Test-retest variability was expressed using intraclass correlation coefficients (ICC), relative differences, and coefficients of variation (COVs). The ICC was quantified by the two-way random-effects model with absolute agreement. An ICC > 0.85 was considered excellent. Relative differences were calculated by taking the absolute difference between two observations divided by the mean of the repeated observations and expressed as a percentage. COVs were calculated as the standard deviation of the difference between two acquisitions divided by their mean value and expressed as a percentage. A COV ≤ 10% was considered excellent.
Receiver operating characteristic (ROC) curves were derived for 2D and 3D echocardiographic parameters to identify CMRI-derived RV EFs of <50% in patients with PH and healthy volunteers. Patients with carcinoid disease were excluded from this analysis to avoid the confounding effects of severe valvular regurgitation on ventricular function. The area under the ROC curve for an echocardiographic parameter is presented together with the optimal cutoff threshold for detecting CMRI-derived RV EF < 50%, defined as the value of the parameter that corresponded to the highest sum of sensitivity and specificity. The Delong method was used to compare the areas under the curve between ROC curves (Analyse-it Software, Ltd, Leeds, United Kingdom).
Population Characteristics and 3DE Technical Data
Of 100 individuals who were recruited, four had unobtainable RV echocardiographic windows. The clinical characteristics and 3D echocardiographic technical data of the final cohort of 96 subjects are presented in Table 1 . Patients with PH had significantly larger and impaired right ventricles than controls, whereas the right ventricles of patients with carcinoid heart disease were also significantly dilated but with preserved EFs. The dilated right ventricles of the patient groups resulted in a significantly lower mean volume rate compared with controls because of the greater 3D sector angles ( P < .001), but the median image quality score was significantly higher among patients (3.00; interquartile range, 2.00–3.00) than controls (2.00 interquartile range, 1.00–3.00) ( P < .001). The image quality among three successive, equally populated subgroups of patients significantly improved with increasing experience with 3DE ( Figure 2 ; P = .031). There was a trend, albeit not statistically significant, for greater differences in SV between modalities with worse subjective image scores ( Figure 3 ; P < .13 for percentage intermodality difference in SV for image score groups 1 and 2 combined vs groups 3 and 4 combined).
|Variable||PH ( n = 46)||Carcinoid heart disease ( n = 19)||Healthy volunteers ( n = 20)||Carcinoid (no valvulopathy) ( n = 11)||P ∗|
|Age (y)||56 ± 13||63 ± 8||50 ± 12||59 ± 10|
|Women||35 (76%)||7 (37%)||15 (75%)||7 (64%)|
|Height (cm)||164 ± 9||171 ± 10||169 ± 8||168 ± 10||.035|
|Weight (kg)||69 ± 17||72 ± 18||72 ± 12||77 ± 20||.54|
|Body surface area (m 2 )||1.8 ± 0.2||1.8 ± 0.3||1.8 ± 0.2||1.9 ± 0.3||.37|
|Heart rate (beats/min)||74 ± 14||67 ± 13||68 ± 9||69 ± 12||.19|
|Mean PASP on RHC (mm Hg)||44 ± 16|
|Endothelin antagonist||21 (46%)|
|PDE-5 antagonist||31 (67%)|
|Prostanoid infusion||2 (4%)|
|Oral prostanoid||1 (2%)|
|Prostaglandin receptor agonist||1 (2%)|
|Carcinoid heart disease: affected valves||TV = 19 (100%), PV = 13 (68%), MV = 3 (16%), AV = 3 (16%)|
|EDV (mL/m 2 )||87 ± 26||100 ± 35||64 ± 14||52 ± 8||<.0001|
|ESV (mL/m 2 )||52 ± 25||33 ± 15||22 ± 7||16 ± 5||<.0001|
|EF (%)||43 ± 14||68 ± 7||65 ± 7||71 ± 7||<.0001|
|3D echocardiographic temporal resolution (volumes/sec)||34 ± 5||32 ± 7||40 ± 5||45 ± 6||<.0001|
Volumetric Analysis by 3DE versus CMRI
Correlation coefficients showed good to excellent correlations between modalities for RV metrics in patient groups and moderate to good correlations for control subjects ( Table 2 ). RV volumes and EFs by 3DE showed differences with CMRI in both patient groups, with a bias for underestimating SV and EF but with overall acceptable limits of agreement ( Figure 4 ). By contrast, 3DE underestimated EDV for control subjects ( Table 3 ), with a consequent negative bias for quantifying SV in this group ( Figure 5 ).
|PH||EDV (mL)||158 ± 53||154 ± 52||.043|
|ESV (mL)||100 ± 44||92 ± 47||<.0001|
|SV (mL)||58 ± 18||63 ± 17||.011|
|EF (%)||39 ± 11||43 ± 14||.00029|
|Carcinoid heart disease||EDV (mL)||182 ± 69||185 ± 71||.21|
|ESV (mL)||67 ± 28||62 ± 3||.01|
|SV (mL)||115 ± 42||124 ± 45||.014|
|EF (%)||64 ± 5||68 ± 7||.001|
|Healthy volunteers||EDV (mL)||105 ± 26||117 ± 27||<.0001|
|ESV (mL)||41 ± 12||41 ± 14||.80|
|SV (mL)||65 ± 16||76 ± 18||<.0001|
|EF (%)||61 ± 5||65 ± 7||.014|
|Carcinoid (no valvulopathy)||EDV (mL)||88 ± 21||98 ± 27||.05|
|ESV (mL)||32 ± 13||30 ± 14||.24|
|SV (mL)||56 ± 10||68 ± 14||.009|
|EF (%)||64 ± 7||71 ± 7||.004|
|Group||Measurements||Bias ± SD||Limits of agreement||r||P ∗|
|All subjects||EDV (mL)||−2.3 ± 13.7||−29.1 to 24.5||0.97||<.0001|
|ESV (mL)||5.2 ± 9.5||−13.4 to 23.9||0.98||<.0001|
|SV (mL)||−7.5 ± 11.8||−30.6 to 15.7||0.94||<.0001|
|EF (%)||−4.6 ± 6.9||−18.2 to 9.0||0.91||<.0001|
|PH||EDV (mL)||4.0 ± 13.1||−21.6 to 29.7||0.97||<.0001|
|ESV (mL)||8.4 ± 10.6||−12. 3 to 29.1||0.98||<.0001|
|SV (mL)||−4.3 ± 10.8||−25.5 to 17.0||0.82||<.0001|
|EF (%)||−4.8 ± 8.3||−21.1 to 11.5||0.81||<.0001|
|Carcinoid heart disease||EDV (mL)||−3.1 ± 10.1||−22.9 to 16.8||0.99||<.0001|
|ESV (mL)||5.4 ± 8.2||−10.6 to 21.4||0.96||<.0001|
|SV (mL)||−8.6 ± 13.9||−35.9 to 18.6||0.95||<.0001|
|EF (%)||−3.8 ± 4.1||−11.9 to 4.2||0.82||<.0001|
|Healthy volunteers||EDV (mL)||−11.9 ± 9.0||−29.5 to 5.8||0.94||<.0001|
|ESV (mL)||−0.4 ± 6.7||−13.6 to 12.9||0.88||<.0001|
|SV (mL)||−11.2 ± 10.1||−31.0 to 8.7||0.84||<.0001|
|EF (%)||−3.9 ± 6.5||−16.6 to 8.8||0.51||.021|
|Carcinoid (no valvulopathy)||EDV (mL)||−10.1 ± 15.0||−39.6 to 19.4||0.84||.001|
|ESV (mL)||2.1 ± 5.5||−8.7 to 12.9||0.92||<.0001|
|SV (mL)||−12.2 ± 12.3||−36.3 to 11.9||0.53||.096|
|EF (%)||−6.2 ± 5.6||−17.1 to 4.7||0.69||.019|