Left ventricular stroke volume index outperforms ejection fraction in predicting 1-year mortality in patients with acute heart failure - data from the SwissHeart Failure Network (SHFN)

Abstract Background Left ventricular ejection fraction (LVEF) has prevailed as the gold standard for quantifying LV function in acute heart failure (AHF) for both clinical trials and routine clinical practice. Most landmark studies in AHF stratify patients according to LVEF-based cut-offs. AHF is a clinical syndrome resulting from insufficient cardiac output, defined as LV stroke volume (LVSV) x heart rate. Thus, LVSV and its derived value LVSV index (LVSVI = LVSV/body surface area) may be better suited than LVEF for assessing LV function and heart failure severity, as well as predicting outcome in AHF. Purpose This study compared the predictive power for 1-year all-cause mortality of readily available echocardiographic parameters related to LV function (LVEF, LVSV and LVSVI) in patients with AHF. Methods Automatic data extraction was performed in 966 patients fulfilling criteria for AHF, retrospectively included between 2015-19 in 4 tertiary centres in Switzerland constituting the SwissHeart Failure Network (SHFN). Patients were selected if they had a standardised TTE study performed upon diagnosis of AHF, wherein all necessary values were reported. Cut-off values were chosen in consideration of the cohort’s median values. Patient follow-up was censored at 365d. Values are displayed as medians and interquartile ranges. Hazard ratios (HR) for 1-year all-cause mortality were computed with univariate Cox Proportional Hazards models. Kaplan-Meier curves were derived along with log-rank p-values. Results Median values for the evaluated parameter were as follows: LVEF (%): 38 (28–54); LVSV (mL): 45 (35–58); LVSVI (ml/m2): 24 (19–31). 1-year all-cause mortality was 33.7% (326/966 patients). The highest predictive capacity for 1-year all-cause mortality was observed for LVSVI<25ml/m2 (HR = 1.57 [1.26–1.97], p<0.001), followed by LVSV<45ml (HR = 1.57 [1.26–1.96], p<0.001). The lowest predictive capacity was seen for LVEF<40% (HR = 1.35 [1.08–1.68], p = 0.007, Figure 1). Applying the rounded medians as cut-off values, Kaplan-Meier curves showed greater divergence for higher vs. lower LVSVI (log-rank p<0.001) than for higher vs. lower LVEF (log-rank p = 0.007, Figures 2A and 2B). For patients with higher LVSVI, mortality rates did not change irrespective of higher or lower LVEF (Figure 2C, blue vs. green curves: log-rank p = 0.885). Patients with lower LVEF showed significantly higher mortality when they also had lower LVSVI (green vs. orange curves: log-rank p= 0.001). Conclusions In patients with AHF, LVSV and its derived value LVSVI are superior in predicting 1-year all-cause mortality compared to LVEF. Requiring the same echocardiographic measurements and thus being as readily available as the currently prevailing LVEF, the assessment of LVSV and LVSVI deserves intensified attention, both for future studies and clinical routine.Figure 1:Hazard RatiosFigure 2:Kaplan-Meier Analyses