Inter-vendor agreement and test-retest reproducibility of right ventricular and left atrial strain measurements using two-dimensional speckle-tracking echocardiography
A E Balinisteanu, A Youssef, J Duchenne, A Puvrez, L Wouters, S Bezy, L Van Langenhoven, J Bogaert, D Vinereanu, J D Thomas, L Badano, J U VoigtAbstract
Background
Right ventricular (RV) and left atrial (LA) strain derived from two-dimensional speckle-tracking echocardiography (2D-STE) are increasingly used for risk stratification and therapeutic decision-making. However, concerns regarding inter-vendor variability and reproducibility continue to limit their routine clinical implementation.
Purpose
To evaluate inter-vendor agreement, and test–retest reproducibility of RV and LA strain in a standardized, real-world test–retest study design.
Methods
Sixty-two subjects (50 men; age 56 ± 17 years) underwent echocardiographic acquisitions using six different ultrasound systems. To assess test–retest reproducibility, two image datasets were acquired in each subject by the same experienced echocardiographer under strictly standardized conditions. We performed in total 3348 RV and LA strain analyses from dedicated views using 3 vendor-specific and 6 vendor-neutral 2D-STE software platforms.
Results
Mid-/full-wall RV free-wall longitudinal strain (RVFWLS) showed good inter-vendor agreement with contemporary software solutions (bias <1strain%), while endocardial RVFWLS demonstrated larger discrepancies (Figure 1). Depending on the software used, test–retest variability for RVFWLS was high to very high (11.3% to 18.3%). Four-chamber LA longitudinal reservoir and contraction strain had a clinically relevant inter-vendor bias (9.3 strain% and 3.7 strain%, respectively, Figure 2). LA contraction strain showed a high test–retest variability (34%). Calculating LA strain as average of both apical views reduced variability significantly (p < 0.05).
Conclusions
In a strictly controlled, real-world test–retest setting, RV free-wall longitudinal strain showed good to moderate inter-vendor agreement. Measurement variability was 2 to 3 times higher than for left ventricular global longitudinal strain. LA strain, especially during atrial contraction, had only moderate inter-vendor agreement and substantial variability. These findings highlight the need for methodological standardization before widespread implementation of these parameters in routine practice, particularly for left atrial strain.Figure 1Figure 2