Comparison of Three-dimensional and Two-dimensional Left Ventricle Strain With Speckle-tracking Echocardiography in Heart Failure With Reduced and Mildly Reduced Ejection Fraction (3V2-STEM-HF)

Left Ventricular Myocardial Strain Assessed by Three-Dimensional Speckle-Tracking Echocardiography in Heart Failure with Reduced or Mildly Reduced Ejection Fraction: Comparison with Two-Dimensional Assessment in a Prospective Observational Study

Background

Global longitudinal strain (GLS) derived from speckle-tracking echocardiography (STE) is a robust, well-validated, and reproducible technique for assessing left ventricular (LV) longitudinal deformation and is available on most modern echocardiography systems. GLS appears to provide superior prognostic value compared with ejection fraction (EF) for predicting major adverse cardiac events across various clinical settings, including chronic heart failure (CHF).

Three-dimensional (3D) STE represents a further advancement in myocardial deformation imaging, enabling rapid and comprehensive assessment of all LV segments and measurement of global 3D GLS from a single dataset. However, in healthy individuals, reference values for 3D and 2D GLS differ significantly, likely because 3D GLS more accurately captures the complex mechanics of LV contraction, whereas 2D GLS provides higher spatial and temporal resolution.

Although 3D STE is increasingly being used to evaluate LV systolic function in patients with CHF, data regarding the agreement and differences between 2D and 3D measurements in this clinical setting remain limited. This issue is particularly relevant in patients with systolic heart failure, as reduced EF appears to weaken the correlation between 2D and 3D GLS measurements, according to available studies with relatively small sample sizes.

Objectives

This prospective multicenter study aims to evaluate the agreement between 2D and 3D GLS measurements obtained using single-vendor ultrasound systems and analysis software in patients with heart failure with reduced ejection fraction (HFrEF) or mildly reduced ejection fraction (HFmrEF).

Secondary objectives are to identify factors associated with differences between 2D and 3D GLS measurements and to determine whether 3D GLS is a stronger independent predictor of all-cause mortality than 2D GLS.

Methods Study Population

This multicenter prospective observational study will enroll patients with HFrEF and HFmrEF, defined as a left ventricular ejection fraction (LVEF) ≤40% and 41-49%, respectively, from the echocardiography laboratories of the Cardiovascular Department of Istituti Clinici Scientifici (ICS) Maugeri across seven tertiary cardiac rehabilitation hospitals in Italy.

The study protocol has been approved to the local Ethics Committee. Written informed consent will be obtained from all participants before eligibility screening.

Image Acquisition and Analysis

Echocardiographic datasets will be acquired using Vivid E95 ultrasound systems (GE Vingmed Ultrasound AS, Horten, Norway) equipped with M5S and 4V probes for 2D and 3D imaging, respectively, or a 4VC probe for both 2D and 3D acquisitions.

The acquisition protocol will include 2-, 3-, and 4-chamber apical views, as well as parasternal short-axis views at the mid-papillary level for 2D imaging, together with 3D LV datasets for the assessment of myocardial strain parameters using 2D and 3D STE. If necessary, two- to six-beat full-volume 3D acquisitions will be performed to ensure complete LV coverage with a frame rate of at least 37 volumes/s.

Two-dimensional and three-dimensional datasets will be analyzed offline using the Q-Analysis and 4D AutoLVQ software packages (EchoPAC BT12 and BT13; GE Vingmed Ultrasound AS) by investigators with at least three years of experience.

Peak global 2D longitudinal strain (2DLε) will be calculated from the three apical views, whereas peak global 2D circumferential strain (2DCε) will be derived from the parasternal short-axis view. Three 3D strain parameters will be obtained from the 3D LV dataset: longitudinal strain (3DLε), circumferential strain (3DCε), radial strain (3DRε), and area strain (3DAε).

Datasets will be excluded from analysis under the following conditions:

1. poor image quality of either the 2D or 3D datasets, defined using a 4-point scale (poor = 1, fair = 2, good = 3, excellent = 4) because of inadequate signal-to-noise ratio, poor blood-tissue contrast, stitching artifacts or incomplete visualization of the LV wall;
2. inadequate tracking of more than two LV segments in a single apical 2D view for global 2DLε analysis or in the parasternal short-axis view for 2DCε and 2DRε analysis;
3. inadequate tracking of more than three LV segments during 3D dataset analysis.

Sample Size

Agreement between 2D and 3D GLS measurements will be assessed using Bland-Altman analysis. Based on published data reporting a mean difference of 0.1% and a standard deviation of differences of 2.9%, the required sample size was calculated to be 419 participants, assuming a maximum acceptable difference between methods of ±6.5%, a type I error (α) of 5%, and a type II error (β) of 20%.

Statistical Analysis

Normally distributed continuous variables will be summarized as mean ± standard deviation (SD), whereas non-normally distributed continuous variables will be reported as median and interquartile range (IQR). Categorical variables will be expressed as counts and percentages.

All LV strain parameters will be analyzed as absolute values, with lower values indicating worse myocardial deformation and higher values indicating better deformation.

Differences between groups will be assessed using unpaired t-tests for normally distributed variables and Mann-Whitney U tests for non-normally distributed variables. Agreement between 2D and 3D LV strain measurements will be evaluated using Bland-Altman analysis. Differences between paired 2D and 3D strain measurements within predefined subgroups will be assessed using paired statistical tests, as appropriate.

Pearson correlation analysis will be used to evaluate associations between the differences in 2D and 3D GLS measurements and demographic, cardiac, and technical variables. Stepwise multivariable linear regression analysis will be performed to identify independent predictors of the differences between 2D and 3D GLS measurements, including age, sex, weight, height, blood pressure, body surface area (BSA), LV volumes, LV mass, temporal resolution, and 3D image quality.

The area under the receiver operating characteristic curve (AUC-ROC) will be used to assess the predictive performance of 2D and 3D GLS for all-cause mortality. The DeLong test will be used to compare the AUCs of the two methods.

All hypothesis tests will be two-sided, and a p-value <0.05 will be considered statistically significant. Statistical analyses will be performed using SPSS software (IBM, Armonk, NY, USA).