Test-retest reproducibility of three-dimensional left ventricular strain by transthoracic speckle-tracking echocardiography

Background Speckle-tracking echocardiography provides quantitative assessment of left ventricular (LV) myocardial strain and is widely applied in clinical practice. Conventional two-dimensional approaches are limited primarily to longitudinal strain assessment. Three-dimensional speckle-tracking echocardiography (3D-STE) enables multiplanar evaluation of LV mechanics, including longitudinal, circumferential, radial, area, and torsional strain. This technique has the potential to offer enhanced sensitivity and specificity for LV structural and functional assessment in clinical populations allowing for serial assessment and optimal patient management. However, data on the reproducibility of repeated 3D strain acquisitions remain limited. Purpose To assess the test-retest reproducibility from repeated acquisitions of multiplanar 3D-STE derived LV strain parameters using transthoracic echocardiography in healthy adults. Methods Twenty-eight healthy adult volunteers underwent two sequential transthoracic echocardiographic examinations performed two hours apart. Participants were instructed to avoid exercise and caffeine on the testing day, abstain from alcohol for at least 24 hours and recreational drugs for 3 days prior to the visit, and refrain from eating for ≥2 hours before arrival. Measurements were scheduled after 10:00 AM to reduce potential influence of the cortisol awakening response. Three-dimensional strain and volumetric datasets were acquired during each examination. Offline analysis was performed to derive global peak 3D longitudinal, circumferential, radial, area, and torsional strain. Image analysis was blinded, with scan pairing concealed until completion. Reproducibility was assessed using intraclass correlation coefficients (ICC), coefficients of variation (CoV), one-way ANOVA, and Bland–Altman analysis. Results Minimal systematic bias was identified for all of the 3D-STE derived measures of strain (P>0.05 and Bias -0.8 to 1.25). Good to excellent agreement was observed for global 3D longitudinal, radial, and area strain parameters (ICCs: 0.71-0.8; CoVs: 9-11%). Circumferential and torsional strain demonstrated weak to moderate agreement for test-retest reproducibility (ICCs: 0.49 and 0.3; CoVs: 13% and 46% respectively). Conclusion Repeated acquisition of multiplanar 3D longitudinal, radial, and area strain demonstrates strong reproducibility in healthy adults supporting their suitability for serial assessment in both clinical and research settings. In contrast, circumferential and torsional strain parameters showed comparatively lower reproducibility, warranting cautious interpretation and further methodological refinement before widespread application.