(E-187) A Correction to Ejection Fraction Measure for More Accurate Cardiac Function Representation and Prognosis

Introduction:: The ejection fraction (EF), or the percentage of blood ejected per heartbeat, is widely used to evaluate cardiac health, as low EF is indicative of poor outcomes and can effectively indicate treatment. However, EF is known to have shortcomings. First, it cannot indicate poor function during Heart Failure preserved EF (HFpEF). Secondly, EF is modulated by cardiac geometry, being elevated during ventricular hypertrophy (1) and depressed during ventricular dilation (2), which reduces its effectiveness in representing cardiac function and health. Here, we propose a corrected EF (EFc) to resolve these issues, and show that this improves the parameters' ability to represent cardiac contractile function, and its ability to predict outcomes in terms of re-hospitalization due to heart failure (HF).

Materials and Methods:: We developed a simple numerical model of the left ventricle (LV) to evaluate how myocardial strains converts to stroke volume, which is constrained by the incompressibility of myocardial tissue. the LV can be adjusted to have various wall thicknesses and dilation. Data from an animal model of HFpEF achieved via progressive aortic cuff constriction was used to evaluate EF and EFc. A retrospective clinical cohort of 2752 individuals was also used to evaluate EF and EFc. All patients were admitted with a request of troponin test and have an echocardiography scan close to the admission time. Cox proportional regression was applied to determine prognosis value of EF and EFc in terms of re-hospitalization within 3 years due to heart failure.

Results, Conclusions, and Discussions:: Numerical model showed that LV hypertrophy at constant myocardial strain increases EF, but did not change stroke volume, while LV dilation at constant strain decreases EF, but increases stroke volume. This demonstrated that EF departed from cardiac function with changes to geometry. This could be explained by a flawed reliance on the endocardial surface for quantifications, as the endocardial surface experienced additional strains due to radial displacements, causing it to depart from the mean strain behaviour, and this deviation directly depended on wall thickness and curvature. Our proposed EFc was equivalent to measuring EF at the mid-wall layer rather than the endocardial surface. Numerical modelling showed that EFc became independent of cardiac geometry, and more closely represented overall contractile strains.

In both our HFpEF animal model and our clinical data, EF could not distinguish between healthy and HFpEF LV, due to the hypertrophic geometry of the LV, but EFc showed a significant difference between the two, demonstrating that EFc could more effectively indicate heart failure.

The baseline cox proportional model was one that predicted the 3-year readmissions risk based on age, sex and creatinine levels, and additional models where EF or EFc were added as predictors were assessed. In the cohort where EF > 50%, using EFc rather than EF in the model increased true positive results by 12.2% and decreased false negative results by 16.6%. ROC analysis showed 18.6% reduced error in predicting readmissions with EFc rather than EF. This suggested that EFc had a better accuracy in predicting re-hospitalization outcomes.

Acknowledgements (Optional): : Funding was provided by British Heart Foundation (BHF) Centre of Research Excellence (RE/18/4/34215), and BHF fellowships (FS/20/18/34972 and FS/19/17/34172)

References (Optional): : 1. Katz DH, et al., Am J Cardiol 112, 1158-1164 (2013).
2. MacIver DH, et al. IJC Heart & Vasc. 7, 113-118 (2015).