Abstract 12511: Left Ventricular Hypertrophy by Electrocardiogram versus Cardiac Magnetic Resonance Imaging as a Predictor for Stroke: The MESA Study

Background: Left ventricular hypertrophy (LVH) is an established risk factor for stroke and is a component of the Framingham Stroke Risk Score (FSRS). It is unknown, however, if LVH detected by the electrocardiogram (ECG-LVH) is equally predictive of stroke as LVH detected by cardiac magnetic resonance imaging (MRI-LVH). Methods: We analyzed 4716 participants (mean age 61±10 years, 53.4% women, 38.2% whites) from the Multi-Ethnic Study of Atherosclerosis (MESA) who were free of cardiovascular disease at baseline (2000-2002). ECG-LVH was defined as values > 95 th percentile of left ventricular mass (LVM) derived from an established race- and sex-specific ECG model for estimating LVM indexed to weight. MRI-LVH was obtained from LVM by cardiac MRI using the 95th percentile cut-off value of observed LVM/predicted LVM calculated as 1.31. Multivariable Cox regression was used to examine the association between 1-standard deviation (SD) increase in estimated ECG-LVM and MRI-LVM as well as ECG-LVH and MRI-LVH, separately, with incident ischemic stroke. Harrell’s C-index was used to estimate the predictive ability of the FSRS model when either ECG-LVH or MRI-LVH was included. Results: Over a median follow-up of 11.2 years, 119 participants developed stroke. In a model adjusted for demographics and components of the FSRS, ECG-LVM and MRI-LVM as well as ECG-LVH and MRI-LVH were all predictive of stroke with ECG-LVM and MRI-LVH showing stronger associations with stroke in terms of hazard ratios but with wider confidence intervals (Table). However, the ability of FSRS model to predict stroke was slightly better when MRI-LVH (C-index 0.748, 95% CI: 0.514-0.829) was used as one of the FSRS components instead of ECG-LVH (C-index 0.739, 95% CI: 0.503-0.823), likelihood ratio test p 0.008. Conclusion: In asymptomatic cardiovascular disease-free individuals, both ECG-LVH and MRI-LVH are predictive of stroke. Using MRI-LVH slightly improves the predictive ability of the FSRS model.