Because of the unique geometry of the right ventricle, assessment of right ventricular (RV) volume and function is clinically challenging. The aim of this study was to investigate the feasibility of single-beat three-dimensional echocardiography (sb3DE) for RV volume and functional assessment in patients with dilated right ventricles.
Fifty-two patients with severe tricuspid regurgitation or atrial septal defects were enrolled. Fifty patients underwent sb3DE and cardiac magnetic resonance (CMR) within 24 hours under a euvolemic state, and the results of sb3DE were compared with those of CMR, the reference method. Fifteen normal subjects were also recruited for a broader validation of sb3DE.
Of the 67 individuals, data from 59 study participants (44 patients and 15 normal subjects) with adequate image quality were analyzed (mean age, 46.9 ± 19.3 years; 58% women). The correlation was excellent between sb3DE and CMR for measuring RV volumes and RV ejection fraction (RVEF) ( r = 0.96, r = 0.93, and r = 0.93 [ P < .001 for all] for RV end-diastolic volume, RV end-systolic volume, and RVEF, respectively). Bland-Altman analysis revealed that RV volumes, but not RVEF, tended to be slightly underestimated by sb3DE (−5.8 ± 9.6%, −3.8 ± 14.1%, and −1.2 ± 9.4% for RV end-diastolic volume, RV end-systolic volume, and RVEF, respectively). Intra- and interobserver variability was acceptable for all indices (4.9% and 6.1% for RV end-diastolic volume, 4.2% and 7.9% for RV end-systolic volume, and 5.7% and 2.8% for RVEF, respectively). Among patients with RV dilation, the difference in RVEF between sb3DE and CMR was more pronounced in patients with atrial fibrillation than those in sinus rhythm (−5.9% vs 0.9%, P = .041).
In patients with dilated right ventricles and in normal subjects, assessment of RV volume and systolic function by sb3DE is feasible in terms of accuracy and reproducibility. RV analysis using sb3DE can be performed in patients with atrial fibrillation, with the possibility of RVEF underestimation.
Single-beat 3DE may be an accurate method for RV volume and functional assessment.
It can be especially useful in patients with dilated right ventricles (i.e., those with severe TR or ASDs).
It is reproducible (ICC > 0.95 for volume and EF for inter- and intraobserver variability).
Certain measurement errors may be dependent on rhythm status (especially atrial fibrillation).
Accordingly, care is necessary when interpreting the results of sb3DE in patients with arrhythmia.
Recently, the importance of right ventricular (RV) structure and function has been emphasized in major cardiovascular disorders, such as heart failure, ischemic heart disease, pulmonary hypertension, and congenital heart disease. However, accurate measurement of RV indices has been challenging because of the complex and crescentic morphology of the right ventricle, whereas the left ventricle resembles an ellipsoid shape, which is relatively easy to analyze using the known volumetric model. Furthermore, the right ventricle is more prone to changes in volume status, and the change in RV geometry is extremely complicated following volume overload. For these reasons, two-dimensional approaches have failed to give accurate estimates of the size and systolic function of the right ventricle.
Although cardiac magnetic resonance (CMR) has been validated for the quantification of RV volume and function, its limited availability, high cost, and long image acquisition time are impediments to its wide use. For this reason, three-dimensional (3D) echocardiography (3DE) may be a promising option to quantify RV volume and function. However, the multiple-beat approach is mandatory for conventional 3DE, which requires regular heart rhythm, stable probe position, and good breath-holding capability to avoid stitching artifacts.
To overcome these limitations of multiple-beat 3DE (mb3DE), single-beat 3DE (sb3DE) has been proposed. Recent studies have shown that sb3DE may be a reliable technique to evaluate RV volume and systolic function in patients with normal right ventricles. However, the validation of sb3DE has not been performed in patients with pathologic right ventricles (i.e., RV dilatation), for which the role of 3DE will be more attractive in a clinical perspective. In this study, we sought to investigate the feasibility of sb3DE in measuring RV volume and function in patients with RV dilatation, such as atrial septal defect (ASD) or tricuspid regurgitation (TR).
We prospectively recruited patients with severe TR, ASD, or other diseases associated with RV dilatation between August 25, 2011, and October 15, 2014. Among 52 consecutive patients, two patients refused CMR examination after undergoing sb3DE. Additionally, six patients were excluded because of poor echocardiographic image quality for 3D analysis (five patients) and poor CMR image quality due to marked beat-to-beat variability in atrial fibrillation (one patient). The final analysis was performed in the remaining 44 patients, who had evaluable single-beat 3D echocardiographic and CMR studies ( Figure 1 A). All CMR and single-beat 3D echocardiographic images were acquired on the same day. Fifteen normal subjects were also studied for a broader validation of sb3DE. These normal control subjects were recruited from a database of volunteers who underwent sb3DE and CMR. The study was approved by the local institutional review board, and all study subjects provided written informed consent to participate in the study.
Single-Beat 3D Echocardiographic Image Acquisition and Analysis
Standard two-dimensional transthoracic echocardiography was performed in all subjects using a Siemens Acuson SC2000 echocardiographic system (Siemens AG, Healthcare Sector, Erlangen, Germany) with a 4V1c transducer (1.25–4.5 MHz). For sb3DE, the images were obtained using the same echocardiographic system, with a 4Z1c Instantaneous Full Volume transducer (1.5–3.5 MHz). The process of 3D echocardiographic image acquisition and analysis is demonstrated in Figure 1 B. In each patient, 3D full-volume data sets were acquired in real time from one cardiac cycle, and all single-beat 3D echocardiographic images were obtained at a stable end-expiratory phase, from a modified apical window, with the patient in the left lateral decubitus position. The ultrasound sector size and depth were adjusted to allow the highest temporal resolution and to encompass the entire RV. Single-beat 3D echocardiographic data sets were stored digitally for offline analysis, and the data were analyzed using the Right Ventricular Analysis Application (SC2000 workplace version 1.5; Siemens AG). We used the index-beat method to select beats to measure RV parameters, as previously described. Briefly, RV volumetric analysis was performed on the acquired images whose ratio of the preceding to the pre-preceding R-R interval was 1. For intraobserver analysis, one experienced sonographer was given a preselected data set of 20 images, which were randomly selected by a computer program, and analyzed these images on two different occasions ≥2 weeks apart. A second experienced sonographer examined the same set of images independently for interobserver analysis.
CMR Image Acquisition and Analysis
We performed CMR using a 1.5-T system (Magnetom Avanto; Siemens AG) equipped with adequate phased-array receiver coils under the standard protocols. Patients were positioned in the supine position. Steady-state free precession cine images were taken under a firm breath-hold to visualize both ventricular wall motions with the following sequence and parameters: TrueFISP; repetition time, 45 msec; echo time, 1.3 msec; flip angle, 80°; matrix size, 256 × 169; field of view, 330 × 330 mm. All short-axis images were acquired at a 6-mm interval with a 4-mm interslice gap from the base to apex to include the whole ventricular volume, and these images were used for analysis. In general, we used retrospective electrocardiographically gated segmented k -space filling for cine CMR. In this technique, nine to 13 heartbeat motions are averaged for each slice. To cover the entire right ventricle, nine or 10 short-axis cine images as a stack are needed. The beat-to-beat difference is averaged in RV volumetry. Because the averaged cine image can be blurred in patients with beat-to-beat differences in heart rate, image data were acquired during a relatively stable R-R interval to reduce the influence of the heart rate variability on RV volumetry. In patients with atrial fibrillation with pronounced beat-to-beat variation, prospective k -space filling was used for the acquisition of proper end-systolic image, because previous studies on cardiac timing intervals have shown that the R-T interval (systolic interval) is relatively constant regardless of heart rate variability compared with the T-R interval (diastolic interval). RV end-diastolic volume (RVEDV), RV end-systolic volume (RVESV), and RV ejection fraction (RVEF) were measured using dedicated software (Argus; Siemens AG). All measurements were performed by a single independent observer who was blinded to the results of sb3DE.
Continuous variables are expressed as mean ± SD, and categorical variables are presented as absolute values (percentages). To test for differences in continuous or categorical variables, we used Mann-Whitney U tests or Fisher exact tests as appropriate. Pearson bivariate correlation analysis was used to evaluate associations between variables estimated by sb3DE and CMR. Agreement between variables measured by sb3DE and CMR was analyzed using Bland-Altman analysis. The Pitman-Morgan test was used to assess the equality of variances between two repeated measurements in the same subject. Intra- and interobserver variability of performing sb3DE were tested using percentage difference, repeated-measures analysis of variance, and intraclass correlation coefficients. All analysis was performed using SPSS version 22 (IBM SPSS Statistics, Chicago, IL), Stata version 13 (StataCorp LP, College Station, TX), or R version 3.2.0 ( www.r-project.org ). A two-sided P value < .05 was considered to indicate statistical significance.
The population of this study consisted of 18 patients with severe TR, 22 patients with ASDs, and four patients with other diseases, as well as 15 normal subjects. Table 1 summarizes the baseline characteristics of patients and normal control subjects. Briefly, the mean age of the entire participants was 47 years, and 34 (58%) were women. Atrial fibrillation was observed in 41% of the patients (18 of 44) with RV dilation, whereas all normal subjects had sinus rhythm during measurements. In the entire study population, the average values of RVEDV, RVESV, and RVEF by CMR were 220.7 ± 80.3 mL, 114.1 ± 48.9 mL, and 49 ± 9%, respectively. The temporal resolution of sb3DE for RV full-volume data sets was 19 ± 6 volumes/sec (median, 19 volumes/sec; interquartile range, 14–21 volumes/sec). There were no adverse events caused by sb3DE or CMR. The ratio of the preceding to the pre-preceding R-R interval before sb3DE acquisition was 1.01 ± 0.21 (median, 0.98; interquartile range, 0.89–1.15) in 18 patients with atrial fibrillation. When we compared the variances of three consecutive R-R intervals during single-beat 3D echocardiographic and CMR acquisitions in 18 patients with atrial fibrillation, there was no significant difference in R-R interval variance between the two sets of measurements (Pitman-Morgan test, P = .712). The mean heart rate was also not significantly different during single-beat 3D echocardiographic and CMR acquisitions in patients with atrial fibrillation (71.9 ± 11.8 vs 73.8 ± 15.3 beats/min; Mann-Whitney U test, P = .646).
|Variable||Entire population ( n = 59)||Patients ( n = 44)||Normal subjects ( n = 15)|
|Sinus rhythm ( n = 26)||Atrial fibrillation ( n = 18)|
|Age (y)||46.9 ± 19.3||41.5 ± 17.1||65.6 ± 14.2||33.8 ± 9.4|
|Women||34 (58%)||15 (58%)||13 (72%)||6 (40%)|
|Height (cm)||165.1 ± 10.5||165.3 ± 8.1||157.7 ± 9.2||173.7 ± 9.3|
|Weight (kg)||62.4 ± 13.9||61.5 ± 14.4||56.1 ± 9.8||71.3 ± 13.2|
|BSA (m 2 )||1.70 ± 0.22||1.68 ± 0.20||1.59 ± 0.19||1.85 ± 0.22|
|SBP (mm Hg)||119.8 ± 13.3||117.3 ± 15.0||123.6 ± 13.4||119.7 ± 9.4|
|DBP (mm Hg)||70.2 ± 10.5||68.8 ± 11.4||69.6 ± 11.6||73.6 ± 6.5|
|During sb3DE||67.5 ± 11.9||65.9 ± 11.6||71.9 ± 11.8||65.1 ± 11.8|
|During CMR||70.4 ± 13.4||69.6 ± 14.0||73.8 ± 15.3||67.5 ± 9.0|
|TR||18 (41%)||2 (8%)||16 (89%)|
|ASD||22 (50%)||22 (85%)||0 (0%)|
|Other diseases ∗||4 (9%)||2 (8%)||2 (11%)|
|sb3D echocardiographic measurements|
|RVEDV (mL)||208.2 ± 78.9||230.9 ± 89.8||221.6 ± 59.3||152.8 ± 51.5|
|RVESV (mL)||109.1 ± 47.4||121.0 ± 53.9||122.2 ± 36.6||72.7 ± 24.7|
|RVEF (%)||48 ± 9||48 ± 10||45 ± 8||52 ± 6|
|RVEDV (mL)||220.7 ± 80.3||243.9 ± 88.6||234.3 ± 63.3||164.3 ± 56.1|
|RVEDV index (mL/m 2 )||132.0 ± 49.2||145.0 ± 48.7||149.6 ± 44.7||88.2 ± 24.3|
|RVESV (mL)||114.1 ± 48.9||128.7 ± 56.0||121.9 ± 36.9||79.5 ± 29.9|
|RVESV index (mL/m 2 )||68.3 ± 30.5||76.6 ± 32.9||77.7 ± 25.5||42.9 ± 14.1|
|RVEF (%)||49 ± 9||48 ± 11||48 ± 8||52 ± 7|
|Temporal resolution (volumes/sec)||19 ± 6||22 ± 8||17 ± 4||16 ± 5|
Evaluation of RV Volumes
In the entire study population, RVEDV and RVESV measured by sb3DE and CMR showed excellent correlation ( Figures 2 A and 2B). The differences obtained by the two modalities were well distributed around the mean of the difference. However, there was a systematic underestimation of RV volumes by sb3DE, especially with RVEDV (relative difference, −5.8 ± 10.0%; absolute difference, −12.5 ± 22.5 mL) compared with RVESV (relative difference, −3.8 ± 14.1%; absolute difference, −5.1 ± 17.9 mL) ( Figures 3 A and 4 A ).
In normal subjects, sb3DE and CMR were also highly correlated ( r = 0.98 for RVEDV and 0.98 for RVESV, P < .001 for all) ( Figures 2 A and 2B). There was a slight underestimation of volumes by sb3DE (−6.9 ± 6.2% and −11.5 ± 11.2 mL for RVEDV and −7.4 ± 9.7% and −6.8 ± 8.0 mL for RVESV) ( Figures 3 B and 4 B).
Because all echocardiograms were obtained from single beats, we analyzed the effect of the rhythm status on the accuracy of RV volume quantification in patients with RV dilation. The extent of correlation between RV volumes measured by sb3DE and CMR remained significant regardless of the rhythm status ( Figures 2 A and 2B). The bias was not increased when sb3DE was performed in patients with atrial fibrillation on Bland-Altman analysis ( Figures 3 C, 3D, 4 C, and 4D). The degree of underestimation was similar for RVEDV (−5.6 ± 11.1% and −13.0 ± 26.3 mL) and RVESV (−5.0 ± 16.6% and −7.8 ± 23.4 mL) in patients with sinus rhythm ( Figures 3 C and 4 C). However, in patients with atrial fibrillation, the direction of the volumetric measurement errors was not consistent for RVEDV (−5.2 ± 11.2% and −12.7 ± 24.7 mL) and RVESV (0.8 ± 12.6% and 0.3 ± 13.7 mL) ( Figure 3 D and 4 D).
Evaluation of RVEF
A strong correlation between sb3DE and CMR was observed for RVEF in the entire study population ( Figure 2 C). The measurement error for RVEF was small (relative difference, −1.2 ± 9.4%; absolute difference, −0.7 ± 4.2%) ( Figure 5 A), which was contrary to the relatively large underestimation observed in RV volume assessment. In 15 normal subjects, similar results were observed in terms of the correlation ( r = 0.92, P < .001) and the measurement error (0.6 ± 5.4% and 0.3 ± 2.8%) ( Figures 2 C and 5 B). When we stratified 44 patients with RV dilation into two subgroups according to rhythm status, a systematic underestimation of RVEF was noted in patients with atrial fibrillation, but not in those with sinus rhythm ( Figures 5 C and 5D). The measurement error for RVEF was significantly greater in patients with atrial fibrillation than in those in sinus rhythm (relative and absolute differences in patients with atrial fibrillation vs sinus rhythm, −5.9 ± 8.1% and −2.9 ± 3.3% vs 0.9 ± 10.9% and 0.3 ± 5.0%; P = .041), whereas measurement differences in RV volumes did not significantly differ by the rhythm status, as shown in Table 2 .
|Variable||Sinus rhythm||Atrial fibrillation|
|( n = 26)||( n = 18)||P ∗|
|RVEDV (mL)||243.9 ± 88.6||234.3 ± 63.3||.849|
|RVESV (mL)||128.7 ± 56.0||121.9 ± 36.9||.729|
|RVEF (%)||48 ± 11||48 ± 8||.849|
|sb3D echocardiographic measurements|
|RVEDV (mL)||230.9 ± 89.8||221.6 ± 59.3||.971|
|RVESV (mL)||121.0 ± 53.9||122.2 ± 36.6||.720|
|RVEF (%)||48 ± 10||45 ± 8||.384|
|Relative difference (%)|
|RVEDV||−5.6 ± 11.1||−5.2 ± 11.2||.659|
|RVESV||−5.0 ± 16.6||0.8 ± 12.6||.233|
|RVEF||0.9 ± 10.9||−5.9 ± 8.1||.041|