Feasibility, Reproducibility, and Prognostic Value of the Fully Automated Measurement of Left Ventricular Longitudinal Strain in Heart Transplant Recipients

Abstract

Abstract Aims Left ventricular longitudinal strain (LVGLS) is a robust parameter to predict adverse events of heart transplanted (HTx) patients. However, measuring LVGLS is time consuming and operator-dependent. Thus, we investigated whether automated strain software applied to measure LVGLS possess the feasibility, reproducibility and prognostic values in HTx patients.Methods 286 heart-transplanted patients who had comprehensive echocardiography was included. LVGLS was obtained from the same apical images by three different methods: fully automated LVGLS(Auto-Strain), semiautomated LVGLS (automated with manual editing) and manual LVGLS. Patients were followed up and primary composite end point (defined as all-cause death and rejection) was recorded.Results Fully automated measurements were feasible in 277subjects (96.8%). Analysis time for automated LVGLS (27.7 ± 2.8 s/patients) and semiautomated LVGLS (237.4 ± 41.0 s/patients) were shorter than manual LVGLS (440.4 ± 65 s/patients). Semiautomated LVGLS had a stronger correlation with manual LVGLS than automated LVGLS (r = 0.854 vs 0.654. P < 0.001), and there were smaller disagreements between Semiautomated LVGLS and manual LVGLS (bias:0.79, LOA:2.78) than automated LVGLS and manual LVGLS (bias:2.72, LOA:3.98). During a median follow-up of 51 months [35.0-66.5], 35 patients experienced end point events. Automated LVGLS can detect abnormal systolic function and predict adverse events of HTx patients, while the detecting and predicting ability of semiautomated LVGLS is greater.Conclusions Fully automated LVGLS could provide rapid and reproducible assessment of HTx patients’ graft function, but semiautomated LVGLS is superior to automated LVGLS in detecting abnormal systolic function and predicting adverse events, making it possible applying automated software in clinical routine.