Abstract 4119: Utility of ECG to Detect Left Ventricular Hypertrophy Defined by Cardiac MRI in Different Ethnic Groups: The Multi-Ethnic Study of Atherosclerosis (MESA)

Abstract

Background: Left ventricular hypertrophy (LVH) is a strong predictor of cardiovascular event risk. We evaluated the ethnicity-specific diagnostic utility of ECG in the detection of LVH measured by MRI as a gold standard. Methods: 12-lead ECG and MRI of 4498 asymptomatic subjects in the MESA study (49% males, mean age 61 ± 10 years) with no prior cardiovascular disease were analyzed. Left ventricular mass (LVM) was determined from MRI. Body-size predicted LV mass (pLVM) was determined from an allometric height-weight index derived from a subpopulation of 822 normal-weight, normotensive, non-diabetic MESA participants with LVH defined as LVM/pLVM > 1.31 (>95 th percentile). Prediction of absolute LV mass by ECG was also assessed for whites and blacks based on previously validated prediction models for these groups. Results: The study population consisted of 39% Whites, 12% Chinese, 27% Blacks and 22% Hispanics. ECG detection rate for LVH as determined by MRI was significantly more in women than men (12.3% vs 3.9% respectively, p< 0.02). For a specificity >= 90%, the highest overall sensitivity for LVH was shown by Perguia score (28.7%) which increased from the first (17.9%) to fourth quartile (45.3%) of LV mass. At a fixed specificity of 90%, the highest sensitivity was shown by Sokolow-Lyon voltage duration product (SLP) (38.2%). Neither Perguia nor SLP differed significantly across the four ethnic groups with respect to sensitivity and area under the ROC curve (p< 0.27, p< 0.52; p< 0.35, p< 0.88 respectively). Presence of LV strain pattern on ECG showed the highest adjusted odds ratio for LVH across all ethnic groups. ECG predicted LVM correlated significantly with MRI-determined LVM for both whites and blacks (r=0.74, p< 0.001 and r=0.70, p< 0.001 respectively) with no significant difference in LVH detection rates (6.1% vs 9.4% respectively, p< 0.35) Conclusion: Based on LVM prediction models, ECG correlates well with MRI. ECG sensitivity to detect LVH is low and it does not differ significantly between whites and blacks. Based on individual criteria like Perguia and SLP, sensitivity and diagnostic performance of ECG did not vary significantly across different ethnic groups.