Abstract 17052: Multiparametric Noninvasive Assessment of Left Ventricular Filling Pressures: Are We There Yet?
Background: Several echo parameters have been proposed as surrogates of LV filling pressures, but their comparative and additive values are unknown. Recent American Society of Echocardiography/European Association of Echocardiography (ASE/EAE) recommendations have proposed a multiparametric algorithm to predict LV filling pressures in the setting of preserved or decreased LV ejection fraction (EF). Our aim was to compare discriminative ability of individual parameters used in ASE/ESE algorithm and validate its use in a large cohort of patients.
Methods: We assessed 449 consecutive pts (181 women, age 63±13y) in whom echocardiography and invasive measurement of LV end-diastolic pressure (EDP) were performed within 24 hours. Elevated EDP was defined as ≥20 mmHg, and decreased EF was defined as <0.5. Predictive ability of individual parameters was compared by using area under the curve (AUC) obtained by receiver operator characteristic analysis. Finally, multiple parameters (see Table for list) were combined first in ASE/EAE algorithm, and then in a new Classification and Regression Tree (CART)-derived algorithm.
Results: LV EDP was elevated in out 80/331 pts with normal EF, and in 47/118 pts with decreased EF. AUC values were higher in decreased EF pts, with E/E', LAVI, and PASP being significant in both groups (Table 1). Sensitivity and specificity of ASE/ESE algorithm to detect elevated EDP were 74% and 57% in patients with normal EF, and 78% and 59% in patients with decreased EF, while they were 73% and 76% in patients with normal EF and 77% and 80% in patients if CART-derived algorithm was used.
Conclusion: While most echo parameters tested show discriminative ability to assess LV filling pressures, their predictive power is low. Their combined use in ASE/EAE algorithm shows moderate sensitivity but poor specificity in this series of pts. Specificity can be moderately improved by the use of the CART algorithm.
- © 2011 by American Heart Association, Inc.