Agreement of 2D transthoracic echocardiography with cardiovascular magnetic resonance imaging after ST-elevation myocardial infarction

Abstract

BackgroundThis study was designed to investigate the agreement of 2D transthoracic echocardiography (2D TTE) with cardiovascular magnetic resonance imaging (CMR) in a contemporary population of ST-elevation myocardial infarction (STEMI) patients. MethodsIn this subanalysis of the GIPS-III trial, a randomized controlled trial investigating the administration of metformin in STEMI patients to prevent reperfusion injury, we studied 259 patients who underwent same-day CMR and 2D TTE assessments four months after hospitalization for a first STEMI. Bland-Altman analyses were performed to assess agreement between LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV ejection fraction (LVEF), and LV mass measurements. Sensitivity and specificity of 2D TTE to detect categories of LVEF (≤35%, 35–50%, ≥50%) was determined. Linear regression of absolute differences in measurements between imaging modalities was used to investigate whether patient characteristics impact measurement bias. ResultsPairwise difference (bias) and 95% limits of agreement between CMR and 2D TTE measurements were +84 (37, 147) ml for LVEDV, +39 (6, 85) ml for LVESV, -1.1 ± 13.5% for LVEF, and -75 (-154, -14) g for LV mass. Sensitivity and specificity of 2D TTE to detect subjects with moderately depressed LVEF (35–50%) as measured by CMR were 52% and 88% respectively. We observed a significant effect of enzymatic infarct size on bias between 2D TTE and CMR in measuring LVESV and LVEF (P = 0.029, P = 0.001 respectively), of age and sex on bias between 2D TTE and CMR in measuring LV mass (P = 0.027, P < 0.001) and LVEDV (P = 0.001, P = 0.039), and of heart rate on bias between 2D TTE and CMR in LV volume measurements (P = 0.004, P = 0.016). ConclusionsWide limits of agreement, underestimation of LV volumes and overestimation of LV mass was observed when comparing 2D TTE to CMR. Enzymatic infarct size, age, sex, and heart rate are potential sources of bias between imaging modalities.