Usefulness of automated function imaging to detect myocardial ischemia during dipyridamole stress echocardiography.

Purpose: Dipyridamole stress echo (DSE) is currently used as an alternative to dobutamine stress echo in detecting coronary artery disease (CAD). However, the lower sensitivity, especially in single-vessel disease and the high inter-observer variability of wall motion (WM) analysis are two major drawbacks of DSE. We aimed in this study to investigate the usefulness of global longitudinal strain (GLS) by automated function imaging (AFI, Echopac GE Horten, Norway) to improve diagnostic accuracy and reproducibility of DSE in detecting myocardial ischemia. Methods: 37 patients (18 men, 67±9 years), with intermediate/high pre-test CAD probability, underwent DSE followed by coronary angiography within one week. Diagnostic accuracy in the identification of CAD, evaluated through sensitivity, specificity and positive/negative predictive values (PPV/NPV), was analyzed for wall motion score index (WMSI) and GLS. Optimal cutoff value to define normal GLS was -20%. Concordance between each diagnostic method and the reference standard, represented by coronary angiography, was evaluated by kappa score and Kendall's tau coefficient. Furthermore, the agreement between two observers with different experience in DSE was assessed by using Cohen's k coefficient. Results: Prevalence of significant CAD (more than 50% of luminal narrowing) was 70% and prevalence of single vessel disease was 60%. Mean GLS significantly decreased from rest (-17±4%) to peak DSE (–15±4%, p<0.001). Sensitivity, specificity, PPV and NPV for WMSI were respectively: 50%, 67%, 83% and 29%. However, combination GLS and WMSI had the highest sensitivity (70%), specificity (70%), PPV (87.5%) and NPV (40%). Furthermore GLS showed higher concordance with coronary angiography (k = 0.75; Kendall's tau = 0.78) than WMSI (k = 0.11; Kendall's tau = 0.14). In addition, there was a good agreement between a trainee and an expert observer by using GLS in comparison with WM analysis for images interpretation at rest (k = 0.61 for WM, k = 0.57 for GLS) whereas the agreement significantly improved for images interpretation at peak stress (k = 0.50 for WM, k = 0.70 for GLS). Conclusions: Combination of GLS and WMSI resulted in significant increase in the accuracy of DSE to detect myocardial ischemia, especially with regard to the test sensitivity. Besides, GLS analysis provides an increase of the agreement for images interpretations between experienced and non-experienced observer, especially at peak stress. Hence, adding routinely GLS analysis during DSE could probably be helpful for more accurate patient risk stratification.