Pilot study using 3D-longitudinal strain computation in a multi-parametric approach for best selecting responders to cardiac resynchronization therapy
Résumé
BACKGROUND: Almost all attempts to improve patient selection for cardiac resynchronization therapy (CRT) using echo-derived indices have failed so far. We sought to assess: the performance of homemade software for the automatic quantification of integral 3D regional longitudinal strain curves exploring left ventricular (LV) mechanics and the potential value of this tool to predict CRT response. METHODS: Forty-eight heart failure patients in sinus rhythm, referred for CRT-implantation (mean age: 65 years; LV-ejection fraction: 26%; QRS-duration: 160 milliseconds) were prospectively explored. Thirty-four patients (71%) had positive responses, defined as an LV end-systolic volume decrease ≥15% at 6-months. 3D-longitudinal strain curves were exported for analysis using custom-made algorithms. The integrals of the longitudinal strain signals (I L,peak) were automatically measured and calculated for all 17 LV-segments. RESULTS: The standard deviation of longitudinal strain peak (SDI L,peak ) for all 17 LV-segments was greater in CRT responders than non-responders (1.18% s(-1) [0.96; 1.35] versus 0.83% s(-1) [0.55; 0.99], p = 0.007). The optimal cut-off value of SDI L,peak to predict response was 1.037%.s(-1). In the 18-patients without septal flash, SDI L,peak was significantly higher in the CRT-responders. CONCLUSIONS: This new automatic software for analyzing 3D longitudinal strain curves is avoiding previous limitations of imaging techniques for assessing dyssynchrony and then its value will have to be tested in a large group of patients.