Background
Three-dimensional echocardiography (3DE) is a reliable and reproducible tool for assessing left ventricular (LV) function but remains sensitive to patient echogenicity. Contrast-enhanced 3DE (C3DE) has the potential to improve quantification in challenging patients. The aim of this study was to evaluate the impact of temporal resolution, spatial resolution, and image dynamic range on LV function assessed using C3DE compared with cardiac magnetic resonance imaging (MRI) in patients with poor echogenicity.
Methods
Forty-one patients with poor echogenicity who underwent two-dimensional echocardiography (2DE), 3DE, C3DE, and MRI were retrospectively investigated.
Results
Before contrast injection, 24 patients had three or more nonvisible segments. Three cases of 2DE and 12 cases of 3DE were not suitable for quantification. LV end-diastolic volumes were systematically underestimated by 2DE (142 ± 58 mL), 3DE (146 ± 69 mL), and C3DE (172 ± 61 mL) compared with MRI (216 ± 85 mL) ( P < .001). Similar results were found for LV end-systolic volumes (81 ± 65 mL for 2DE, 82 ± 69 mL for 3DE, and 102 ± 80 mL for C3DE vs 129 ± 94 mL for MRI; P < .001). C3DE provided the best agreement with MRI (Lin concordance correlation coefficients of 0.67, 0.93, and 0.99, respectively, for end-diastolic volume, end-systolic volume, and ejection fraction) as well as the best measurement reproducibility. Finally, ultrasound settings had no significant effect on LV volumes and ejection fraction measurements.
Conclusions
In these patients with poor ultrasound image quality, C3DE, regardless of instrument settings, outperformed 2DE and 3DE to assess LV volumes and ejection fraction and can thus be proposed as an acceptable alternative when MRI cannot be performed in this subgroup.
The assessment of left ventricular (LV) function remains central to clinical decision making in a wide range of cardiac patients. Because two-dimensional echocardiography (2DE) is widely available and offers high temporal and spatial resolution, this modality is prominent for this task. In case of poor ultrasound windows, contrast-enhanced 2DE yields a higher definition of endocardial borders, hence increasing both the accuracy and the reproducibility of LV systolic functional measurements.
Recently, three-dimensional (3D) echocardiography (3DE) has shown significant improvements in the assessment of LV volumes and ejection fraction (EF), leading to reproducibility similar to that observed with cardiac magnetic resonance imaging (MRI). The benefit of contrast-enhanced 3DE (C3DE) remains unclear compared with 3DE. C3DE requires specific probes with harmonic capacity and a variety of algorithms enhancing the signal coming from the microbubbles and canceling tissue signals. Temporal resolution is lower with C3DE than with 3DE because of the emission of two pulses and the postprocessing of the received signal to enhance microbubble echoes. While microbubbles enhance the detection of blood within the LV cavity, they also induce attenuations of the ultrasound signal. Apical destruction, signal attenuation in the LV basal region, and heterogeneous concentrations between subvolumes contribute to heterogeneities in echo intensities. In addition, dynamic range settings and spatial resolution may dramatically influence image quality. To be relevant, C3DE requires at least four cardiac cycles, sampled at 25 frames/cardiac cycle.
Only a few studies have focused on the feasibility and the clinical value of C3DE, and none, to our knowledge, has explored the impact of temporal and spatial resolution or image dynamic range on C3DE-based measurements accuracy. The aim of this study was to evaluate the impact of these factors on the ability of C3DE to assess LV function in comparison with MRI in a subset of patients with poor echogenicity.
Methods
Study Design
We retrospectively reviewed the data acquired from 54 consecutive patients referred to our institution who underwent LV functional assessment using contrast echocardiography and MRI as part of routine medical care. Among these patients, 13 were excluded because of extended delays between MRI and echocardiography (>90 days). All patients were in sinus rhythm and sequentially underwent 2DE and 3DE. According to American Society of Echocardiography guidelines, contrast was injected in these patients before the acquisition of CD3E because of poor image quality. All patients with known contraindications to MRI or MRI contrast agents were excluded from the study. All LV measurements were performed in a blinded manner. Different observers quantified MRI and echocardiographic modalities (A.P. and F.L. for echocardiography, N.P. and E.S. for MRI).
Conventional 2DE
All echocardiograms were acquired by two experienced sonographers (E.S., A.P.) using a 1.0- to 5.0-MHz S5.1 probe (Philips Medical Systems, Andover, MA). All subjects were examined in a lateral decubitus position using color Doppler tissue imaging and two-dimensional M-mode echocardiography. All images were recorded using harmonic imaging after the patient fully exhaled.
Data stored on a hard drive were reviewed and analyzed offline using commercially available software (Xcelera; Philips Medical Systems) by investigators blinded to any other information. Five two-dimensional loops of three cardiac cycles were recorded in parasternal long- and short-axis views and in two-, three-, and four-chamber apical views. Image quality of two-dimensional loops was evaluated (E.S., B.A.) by grading all 17 LV segments at rest without contrast on a three-point scale as follows: 0 = invisible border, 1 = inadequate or partially visible borders, or 2 = fully visible borders. A quality score was calculated for each acquisition and each patient by the sum of all segmental scores divided by 17. The study population was split into two groups according to the median value of the population quality score: group A for values less than and group B for values greater than or equal to the median quality rank.
LV EF was computed from LV end-diastolic volume (EDV) and end-systolic volume (ESV) using the biplane Simpson’s rule on two- and four-chamber apical views. The endocardial border was manually delineated by an experienced sonographer according to the recommendations of the American Society of Echocardiography, including the papillary muscles and trabeculae within the cavity. LV EDV time was determined as the frame immediately preceding mitral valve closure and ESV time as the frame with the smallest LV volume.
3DE
Three-dimensional echocardiography was performed using an iE33 echocardiographic system (Philips Medical Systems) equipped with a broadband, wide-angle matrix-array transducer designed for harmonic contrast imaging (X3-1 ultrasound probe). Electrocardiographically triggered 3D echocardiographic data sets were acquired from the apical window. All 3D images were optimized by modifying the gain, brightness, compression, and time-gain compensation settings. Three-dimensional echocardiographic data sets were then acquired using four subvolumes. The spatial extent was limited to the smallest volume covering the left ventricle to maximize spatial and temporal resolution. Quantifications of 3D volume were performed offline with dedicated semiautomated software, as previously described (QLAB version 6; Philips Medical Systems). QLAB generates a first subendocardial contour at the first interface between the cavity and the myocardium. We did not modify the automated segmentation of the endocardium when it correctly fit the first visible interface. Otherwise, the endocardial edge was corrected by the operator. All volumes with atrial premature complexes and ventricular premature complexes were rejected. All measurements were performed by an experienced sonographer blinded to any other data.
C3DE
The echo enhancer used in this study was a suspension of microbubbles of sulfur hexafluoride (SonoVue; Bracco, Milan, Italy). Real-time C3DE was performed using second-harmonic imaging. The same equipment (including the transducer) was used for C3DE and standard 3DE. SonoVue was delivered as an initial bolus of 0.5 mL, followed by a slow saline flush until contrast appeared in the right ventricle. Additional doses of contrast agent were titrated to achieve optimal LV chamber visualization, as recommended. According to previous reports, the mechanical index was set between 0.2 and 0.3 with the highest possible frame rate and a gain level of 60% to 70%. Attention was paid to acquire data when flow was visually stable and when no or little apical swirling was observed. All contrast-enhanced 3D echocardiographic acquisitions were characterized by three indices: spatial resolution, temporal resolution, and compression.
Images obtained from eight and four subvolumes were indexed respectively as high– and low–spatial resolution data sets. Temporal resolution was indexed according to the number of frames per cardiac cycle, calculated as [frame rate (Hz) × 60]/heart rate. In our study group, the median value was 10 frames/cardiac cycle (range, 8–17 frames/cardiac cycle). All volumes were labeled as high or low temporal resolution using a threshold of 10 frames/cardiac cycle. Image volumes with <20% compression were labeled as low-level dynamic range, and image volumes with compression values equal to 80% were labeled as having high dynamic range. Spatial resolution was labeled (low vs high) according to the number of subvolumes (four vs eight) used to reconstruct the image. One acquisition was performed for each setting.
Ventricular volume and EF calculations were repeated by an experienced sonographer, using the same offline software and technique as for standard 3DE. Finally, volume and EF measurements were evaluated according to ultrasound machine settings: dynamic range, temporal resolution, and spatial resolution.
Reproducibility
The reproducibility of C3DE-based EDV and ESV assessment was measured according to interobserver variability. To achieve this, 15 of the 41 patients were randomly selected. Analysis of the contrast-enhanced 3D echocardiographic images obtained in these patients was repeated twice by the first reader and by an additional expert reader, who were both blinded to all previous measurements, including volumes obtained on MRI. Interobserver reliability for the measurement of LV volumes and EF for each technique was assessed using two-way random, single-measure intra class correlation. Intraobserver reliability was assessed using one-way random, two-measure intraclass correlation.
Cardiac MRI
Magnetic resonance images were obtained with a 1.5-T scanner (SignaHdx; GE Healthcare, Milwaukee, WI) with a phased-array cardiac coil. Electrocardiographically gated localizing MRI-echocardiographic sequences were used to identify the long axis of the heart to allow imaging of the left ventricle in the anatomically correct short-axis plane. Steady-state free precession (fast imaging employing steady-state acquisition) cine loops were obtained during 10- to 15-sec breath-holds with a temporal resolution of 25 frames/cardiac cycle. In all patients, a stack of short-axis cine loops was acquired from the atrioventricular ring to the apex (slice thickness, 8–10 mm; no gap). Offline calculations of EDV, ESV, and EF were performed using CardioReport (GE Healthcare, Buc, France). In every short-axis slice, contours were manually traced at end-systole and end-diastole, including the papillary muscles in the LV cavity. All tracings were performed by an experienced researcher in the interpretation of magnetic resonance images without knowledge of the echocardiographic measurements. The traced contours were used to calculate EDV and ESV, which served as the reference for comparison against noncontrast two-dimensional echocardiographic, 3D echocardiographic, and contrast-enhanced 3D echocardiographic data.
Statistical Analysis
All statistical analyses were performed using R version 2.15.0 (R Foundation for Statistical Computing, Vienna, Austria). LV volumes and EF values are expressed as mean ± SD. The agreement between echocardiography and MRI was evaluated using both the Lin concordance correlation coefficient and Bland-Altman analysis. The Lin concordance correlation coefficient is essentially equivalent to the κ coefficient but is applicable to continuous data. This coefficient evaluates both accuracy and precision, indicating how far measurement pairs fall from the line of identity, ranging from +1 (perfect agreement) through 0 (no agreement) to −1 (perfect inverse agreement). The significance of the biases was tested using a nonparametric paired t test with a two-tailed distribution. P values ≤ .05 were considered significant. The significance of the biases for each ultrasound modality according to an independent factor (volume enlargement, image quality, and unavailability of 3D echocardiographic data set) was tested by using an unpaired nonparametric t test. The impact of the contrast agent on interobserver agreement was assessed using the nonparametric Wilcoxon signed-rank test (volumetric values for 3DE, volumetric value of each combination, and mean value of all combinations for C3DE).
Results
Study Population
The mean delay between MRI and echocardiography was 4.4 ± 7.5 days. Thirty-two patients (78%) underwent MRI and echocardiography on the same day. Six patients underwent MRI between 6 and 17 days and three patients between 17 and 28 days. In the 41 patients (29 men and 12 women; mean age, 57.8 ± 15.4 years), 3D and contrast-enhanced 3D echocardiographic analysis was feasible. Thirteen patients (32%) had ischemic cardiomyopathy (ICM), 13 (32%) had dilated cardiomyopathy (DCM), and 15 (36%) had hypertrophic cardiomyopathy (HCM). SonoVue tolerance was excellent in all patients. The baseline characteristics of the study population are summarized in Table 1 .
| Variable | Subgroup | Total | P | ||
|---|---|---|---|---|---|
| ICM | DCM | HCM | |||
| Number of patients | 13 | 13 | 15 | 41 | |
| Age (y) | 56 ± 2.8 | 59 ± 4.3 | 58 ± 5.0 | 58 ± 2.4 | .72 |
| Men/women | 9/4 | 9/4 | 11/4 | 29/12 | .96 |
| BMI (kg/m 2 ) | 27 ± 1.0 | 27 ± 1.2 | 26 ± 1.0 | 27 ± 0.6 | .26 |
| Weight (kg) | 76 ± 13 | 81 ± 20 | 75 ± 16 | 78 ± 16 | .65 |
| Height (m) | 1.7 ± 0.1 | 1.7 ± 0.7 | 1.7 ± 0.1 | 1.7 ± 0.1 | .26 |
| Heart rate (beats/min) | 63 ± 7 | 67 ± 12 | 56 ± 5 | 62 ± 10 | .014 |
| MRI | |||||
| EDV > 250 mL | 8/13 | 6/13 | 0/13 | 14/41 | .04 |
| EDV < 150 mL | 1/13 | 1/13 | 8/15 | 10/41 | .05 |
| EDV (mL) | 269 ± 76 | 253 ± 68 | 138 ± 36 | 216 ± 85 | <.001 |
| ESV (mL) | 197 ± 81 | 171 ± 66 | 33 ± 11 | 129 ± 94 | <.001 |
| EF (%) | 29 ± 13 | 33 ± 12 | 77 ± 5 | 48 ± 24 | <.001 |
| 2DE | |||||
| EDV (mL) | 178 ± 52 | 177 ± 35 | 87 ± 26 | 142 ± 58 | <.001 |
| ESV (mL) | 123 ± 67 | 119 ± 37 | 20 ± 7 | 81 ± 65 | <.001 |
| EF (%) | 30 ± 12 | 34 ± 9 | 76 ± 6 | 49 ± 24 | <.001 |
| 3DE | |||||
| 3D Quality Score | 22.9 ± 1.6 | 27.0 ± 1.7 | 29.7 ± 0.9 | 26.9 ± 0.9 | .013 |
| No 3DE ∗ | 5/13 | 4/13 | 3/15 | 12/41 | .72 |
| EDV (mL) | 179 ± 73 | 192 ± 45 | 89 ± 32 | 146 ± 69 | <.001 |
| ESV (mL) | 133 ± 80 | 121 ± 30 | 20 ± 8 | 82 ± 69 | <.001 |
| EF (%) | 29 ± 14 | 38 ± 6 | 78 ± 5 | 52 ± 24 | <.001 |
| C3DE | |||||
| EDV (mL) | 205 ± 49 | 197 ± 65 | 124 ± 29 | 172 ± 61 | <.001 |
| ESV (mL) | 140 ± 55 | 153 ± 64 | 27 ± 8 | 102 ± 80 | <.001 |
| EF (%) | 33 ± 13 | 33 ± 13 | 78 ± 5 | 47 ± 24 | <.001 |
∗ Non-contrast-enhanced 3DE not available for postprocessing.
Echocardiography
Image Quality
The mean quality score was 26 (range, 12–34), with a median value of 28. Half of our population had more than three nonvisible segments before contrast injection. Patients with poor quality indices (group B) had increased LV volumes: mean EDV on 2DE was 164 ± 14 mL in group B ( n = 22) and 122 + 13 mL in group A ( n = 19) ( P = .043), and mean EDV on C3DE was 206 ± 18 mL in group B and 168 ± 16 mL in group A ( P = .10).
Image Processing
Of the 41 patients included in the study, image quality was not suitable for quantitative analysis in three cases of 2DE (two patients with DCM and one with ICM) and in 12 cases of 3DE (four patients with DCM, five with ICM, and three with HCM). All contrast-enhanced 3D echocardiographic data sets were suitable for image processing and quantification. In patients with nonsuitable 3DE, image quality indices were significantly impaired (22.7 ± 2.2 vs 28.5 ± 0.8, P = .015). Body mass index and MRI diastolic volumes were higher (28.0 ± 1.4 vs 25.9 ± 0.7 kg/m 2 and 242 ± 23 vs 205 ± 16 mL, respectively) but without reaching statistical significance ( P = .122 and P = .24, respectively).
Volume and EF Measurements
LV volumes were statistically different depending on the ultrasound modality. EDV values were significantly underestimated by 2DE: 142 ± 58 mL with 2DE versus 146 ± 69 mL with 3DE and 172 ± 61 mL with C3DE ( P < .0001). The same results were found for ESV: 81 ± 65 mL with 2DE, 82 ± 69 mL with 3DE, and 102 ± 80 mL with C3DE ( P < .0001). EF values were not statistically different between echocardiographic modalities: 49 ± 24% for 2DE, 52 ± 24% for 3DE, and 47 ± 24% for C3DE ( P = .44).
Analysis According to Cardiomyopathy Subgroup
C3DE showed statistically significant differences from 2DE and 3DE for all EDV measurements ( P < .0001) and in the DCM and ICM subgroups for ESV measurements ( P < .0001). The only statistically significant difference between 3DE and 2DE was found in the DCM subgroup for EDV ( P = .031). For all other comparisons between 3DE and 2DE, the difference did not reach statistical significance. Three-dimensional echocardiography and C3DE were not statistically different for ESV measurement in the DCM and HCM subgroups ( P = .10 and P = .32, respectively).
MRI Results
MRI-based LV functional parameters are listed in Table 1 for all subgroups. There were no differences between the ICM and DCM subgroups in terms of EDV ( P = .62), ESV ( P = .29), and EF ( P = .25).
Depending on image quality, LV volumes and EF were different in the two groups. The group with poor quality (group B) was associated with higher volumes and lower EF (EDV, 243 ± 84 mL vs 190 ± 82 mL in group A [ P = .051]; ESV, 161 ± 100 mL vs 97 ± 84 in group A [ P = .032]; EF, 39 ± 24% vs 57 ± 23% in group A [ P = .024]).
Comparison of Echocardiography and MRI
LV EDV
All ultrasound techniques significantly underestimated EDV compared with MRI ( Figure 1 , Table 2 ). However, this underestimation was dramatically reduced when using C3DE. Taking into account the cardiomyopathy subgroup did not affect these findings.
| Factor | 2DE | 3DE | C3DE | MRI Volume (mL) |
|---|---|---|---|---|
| EDV bias (mL) (% of MRI EDV) | EDV | |||
| All patients | 76 ± 40 (35 ± 19) | 59.8 ± 89 (27 ± 22) | 28 ± 62.6 (13 ± 15) | 216 ± 85 |
| EDV > 250 mL | 119 ± 23 (37 ± 7) | 109 ± 39 (34 ± 13) | 43 ± 35 (14 ± 11) | 315 ± 40 |
| EDV < 150 mL | 42 ± 17 (36 ± 15) | 39 ± 20 (34 ± 18) | 9 ± 14 (7 ± 12) | 114 ± 23 |
| Group A | 92 ± 38 (37 ± 35) | 68 ± 36 (28 ± 15) | 37 ± 34 (15 ± 14) | 243 ± 84 |
| Group B | 62 ± 40 (32 ± 29) | 57 ± 50 (30 ± 26) | 22 ± 30 (11 ± 16) | 190 ± 82 |
| HCM | 51 ± 24 (37 ± 19) | 42 ± 21 (30 ± 15) | 17 ± 22 (12 ± 16) | 138 ± 36 |
| ICM | 99 ± 39 (36 ± 15) | 88 ± 58 (32 ± 22) | 35 ± 38 (13 ± 14) | 269 ± 76 |
| DCM | 85 ± 44 (35 ± 18) | 60 ± 46 (23 ± 18) | 35 ± 32 (14 ± 13) | 253 ± 68 |
| 3DE | 68 ± 38 (33 ± 19) | 60 ± 45 (29 ± 22) | 28 ± 29 (14 ± 14) | 206 ± 86 |
| No 3DE ∗ | 99 ± 40 (41 ± 17) | NA | 29 ± 35 (12 ± 16) | 242 ± 80 |
| ESV bias (mL) (% of MRI ESV) | ESV | |||
| All patients | 45 ± 45 (35 ± 35) | 33 ± 37 (25 ± 29) | 17.76 ± 23 (14 ± 18) | 129 ± 94 |
| EDV > 250 mL | 92 ± 21 (39 ± 9) | 78 ± 32 (33 ± 14) | 33 ± 27 (14 ± 12) | 234 ± 61 |
| EDV < 150 mL | 11 ± 6 (30 ± 16) | 11 ± 5 (31 ± 14) | 4 ± 5 (12 ± 14) | 36 ± 21 |
| Group A | 66 ± 53 (41 ± 33) | 43 ± 33 (26 ± 20) | 26 ± 28 (16 ± 17) | 162 ± 100 |
| Group B | 27 ± 28 (27 ± 29) | 28 ± 40 (29 ± 42) | 11 ± 16 (11 ± 17) | 97 ± 84 |
| HCM | 12 ± 9 (38 ± 27) | 12 ± 5 (35 ± 14) | 5 ± 6 (15 ± 17) | 33 ± 11 |
| ICM | 80 ± 50 (41 ± 26) | 66 ± 43 (33 ± 22) | 27 ± 29 (14 ± 15) | 197 ± 82 |
| DCM | 52 ± 36 (30 ± 22) | 32 ± 38 (19 ± 22) | 24 ± 21 (14 ± 12) | 171 ± 63 |
| 3DE | 35 ± 33 (31 ± 28) | 33 ± 37 (28 ± 32) | 16 ± 20 (14 ± 18) | 115 ± 92 |
| No 3DE ∗ | 77 ± 63 (47 ± 39) | NA | 22 ± 29 (14 ± 18) | 162 ± 96 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
