8A.02

Abstract

Objective: Several allometric methods for indexing cardiac structures to body size have been proposed but the optimal way for normalization of cardiac structures is still controversial. We aimed to estimate the allometric exponents that best describe the relationships between cardiac dimensions and body size and propose normative values. We also explored how different scaling metrics influence the associations of left heart size with cardiovascular risk factors and outcome in the general population. Design and method: We measured left ventricular end-diastolic dimension (LVEDD), end-diastolic volume (LVEDV), left ventricular mass (LVM) and left atrial volume (LAV) in randomly recruited population cohorts (n = 1509; 52.8% women; mean age, 47.8 years). After determining optimal scaling metrics in a healthy reference population (n = 656) and proposing normative values, we analyzed how the different scaling metrics influence predictive models for left ventricular hypertrophy (LVH) and left atrial enlargement (LAE) as well as cardiovascular outcome. Results: The allometric exponents that described the relationships between LVEDD and body size were 1, 0.5 and 0.33 for body height (BH), body surface area (BSA) and estimated lean body mass (eLBM), respectively. With regards to LVEDV, LVM and LAV the allometric exponents for BH were 2.9, 2.7 and 2.0, respectively; for BSA they ranged from 1.7 to 1.8; for eLBM all exponents were around 1. These exponents were used to appropriately scale the cardiac dimensions to body size and derived sex-specific cut-off limits for different indexed cardiac dimensions. Indexation of LVM to height2.7 better detected LVH in overweight and obese subjects. The hazard ratios of cardiovascular outcome were highest for LVH defined by LVM/height2.7. Conclusions: Our current study resulted in a proposal for thresholds for various indexed cardiac dimensions. LVM indexed to height has the advantage of being more sensitive in detection of LVH associated with obesity and slightly better for prediction of outcome.