Patients

Sixty-three patients were enrolled at four European centers, with balanced contributions. Patient enrollment was stratified at each center on the basis of results from angiographic ventriculography to achieve a balanced distribution within three predefined EF groups (>55%, 35%–55%, and <35%). An interpretable cine ventriculogram with the availability of at least two consecutive nonextrasystolic cardiac cycles during ventriculographic contrast administration was a prerequisite for inclusion in the study.

Echocardiography

At all sites, 2D echocardiography was performed using a commercially available ultrasound scanner (iE33; Philips Medical Systems, Andover, MA) using tissue harmonic imaging for unenhanced and contrast-specific imaging for contrast-enhanced echocardiography. Two-dimensional apical four-chamber, two-chamber, and three-chamber views as well as 3D full-volume data sets from the apical position were acquired without and with contrast enhancement. Five consecutive cardiac cycles of each view were acquired during breath hold and digitally stored. Great care was taken to avoid apical foreshortening and to maximize the length from base to apex. A 3D full-volume data set of the ventricle was obtained with gated (five beats) acquisition. Sector size and depth were optimized to obtain the highest possible volume rates, reaching 17 to 20 frames/sec in the contrast 3D full-volume mode.

For contrast-enhanced assessment of LV function, a 20-gauge intravenous catheter was introduced into the right antecubital vein. SonoVue (Bracco Imaging, Milan, Italy) was administered using a dedicated infusion pump (VueJect; Bracco Imaging) with continuous mixing of the contrast agent suspension at a starting infusion rate of 1 mL/min and subsequent adjustment to reach homogenous LV cavity opacification without attenuation. SonoVue is a commercially available ultrasound contrast agent consisting of sulfur hexafluoride microbubbles stabilized by a highly flexible phospholipid monolayer shell.

Ultrasound machine settings were optimized for contrast specific imaging. Transmit power was set to be low (mechanical index < 0.4), and dynamic range was adjusted to achieve optimal contrast between cardiac walls and the LV cavity.

Analysis of unenhanced and enhanced echocardiograms as well as 2D and 3D echocardiography was performed in random order. All acquired 2D and 3D data sets were transferred to a dedicated workstation (TomTec 4D; TomTec Imaging Systems, Munich, Germany). Considering 2D data sets, end-diastolic and end-systolic LV volumes and EF were determined by semiautomatic tracing of end-systolic and end-diastolic endocardial borders using apical four-chamber and two-chamber views, using Simpson’s method. According to the recommendations of the American Society of Echocardiography, the tracings were performed with the papillary muscles and trabeculations allocated to the LV cavity. The mitral annulus was to be traced as deeply as possible. Considering the 3D data sets, reconstructed 2D views of the four-chamber, two-chamber, and long-axis views were obtained using the TomTec system. Within the reconstructed views, endocardial border tracing of the end-diastolic and end-systolic images was performed to obtain the corresponding LV volumes. TomTec 4D LV-Analysis software was also used for semiautomatic tracing of the endocardial borders in the full-volume data sets. For this purpose, endocardial border tracings were semiautomatically performed in three long-axis images at end-systole and end-diastole. The contouring was verified on long-axis and short-axis cine images and modified as necessary to ensure optimal endocardial tracking including analysis of the valve plane.

Cine Ventriculography

Standard biplane cine ventriculography was performed using a 30° right anterior oblique projection and a 60° left anterior oblique projection with injection of ≥30 cm ³ contrast medium at a flow rate of 12 to 14 mL/sec using 5-Fr to 6-Fr pigtail catheters. The frame rate was set at 30 Hz. Semiautomatic border tracking was used to define the end-diastolic image on the basis of the R wave on electrocardiography and the end-systolic image on the basis of the frame with the smallest ventricular silhouette. Image calibration was performed with the use of a metal ball with a diameter of 5.0 cm with identical positions of the x-ray tubes. LV end-diastolic and end-systolic volumes were determined using Simpson’s method, according to well-defined standards and after formal training for biplane analyses, using CAAS II software with the LV biplane analysis module (Pie Medical Imaging, Maastricht, The Netherlands).

CMR

Electrocardiographically triggered CMR investigations at a field strength of 1.5 T during breath hold were performed using a special volume-adapted surface coil. To assess LV function, standard steady-state free precession cine imaging (spatial resolution, 1.4 × 1.4 × 8 mm; 35 phases per cardiac cycle; repetition time, 3.1 msec; echo time, 1.6 msec; flip angle, 55°) was performed during short repetitive end-expiratory breath holding. Four-chamber, two-chamber, three-chamber, and short-axis views with a slice thickness of 10 mm were acquired in the basal-apical direction.

Evaluations were performed using Siemens Argus software (syngoMMWP VE27A and syngo VE31H; Siemens Healthcare, Erlangen, Germany). Endocardial border tracings were performed automatically by the system, with manual correction if needed for each short-axis slice separately at end-diastole and end-systole to derive LV volumes and EF. The definition of the most basal slice required continuously visible myocardium, including its transition into the LV outflow tract. Tracings were performed with the papillary muscles and trabeculations allocated to the LV cavity as performed on echocardiographic images.

Statistical Analysis

Statistical analysis was performed using Medidata (Medidata Solutions, Konstanz, Germany). Continuous variables are presented as mean ± SD and were compared using Student’s t tests. The limits of agreement (defined as ±2 SDs from the mean difference) between readers 1 and 2 on echocardiographic analysis of global LV function without and with contrast administration as well as between echocardiographic and CMR measurements of global LV function were determined using Bland-Altman analysis. Linear regression analysis was performed to determine the correlations between readers 1 and 2 and between echocardiography and CMR in the assessment of volumes and EF. In addition, the interreader variability in the assessment of LV volumes and EF between the two readers was determined as a percentage of variability. The percentage of variability was calculated as the standard deviation between two measurements divided by their mean multiplied by 100. P values ≤ .05 were considered to indicate statistical significance.