Optimal cut-off point of left ventricular ejection fraction for prediction of death in end-stage hypertrophic cardiomyopathy with systolic dysfunction
J Hudelo, Y B Bohbot, T G Goncalves, S D Duhamel, T H Hovasse, A K Kante, J F Florence, J G D Dillinger, A H Hermida, C T Tribouilloy, V B Bousson, P G Garot, S T Toupin, J G Garot, T P PezelAbstract
Introduction
In hypertrophic cardiomyopathy (HCM), end-stage stage is traditionally defined by a left ventricular ejection fraction (LVEF) below 50%. This cut-off point is widely used and supported by observational data linking it to increased mortality. However, this binary definition may oversimplify the continuum of systolic dysfunction and fail to accurately capture its prognostic significance. We aim to determine the optimal LVEF cut-off point for predicting long-term all-cause mortality in a large cohort of end-stage HCM patients.
Methods
All patients referred for cardiovascular magnetic resonance (CMR) assessment of HCM at three French university hospitals between 2008 and 2024 were retrospectively screened. All patients with HCM and LVEF <50% were included. The primary endpoint was all-cause mortality using the National Registry of Death. A conditional inference tree (C-Tree) and Youden index method were applied to identify the optimal cut-off point of LVEF for predicting all-cause mortality. A propensity score matching approach was applied to optimize covariate balance between groups, ensuring standardized mean differences <0.1. Cox proportional hazards models were used both before and after propensity score matching, as well as after adjustment for traditional prognostic factors. A restricted cubic spline model assessed the continuous relationship between LVEF and all-cause mortality, adjusted for covariates used in the Cox model.
Results
Among 2,873 eligible patients with HCM, we included 691 patients (mean age 53 ± 7 years, 54% male) with a LVEF <50% for the study. The optimal cut-off point of LVEF for predicting all-cause mortality was 40% with both methods. Among those, 492 patients had a LVEF of 40–50%, and 199 a LVEF <40%. The patients with LVEF < 40% had more advanced structural and functional abnormalities: larger LV end-diastolic volumes (LVEDV indexed 99 ±11 vs 83 ±7 mL/m², p<0.001), higher prevalence of right ventricular dysfunction (28% vs 5%, p<0.001), and greater presence (66% vs 26%) and extent (2.4 ± 1.9 vs 0.6 ± 1.2 segments; p<0.001) of late gadolinium enhancement (LGE). After propensity score matching, two groups of 191 patients were obtained, with comparable baseline characteristics. After a median follow-up of 9 years (IQR 6–11), 226 patients died (32%). After adjustment for traditional prognostic factors, a LVEF value <40% was strongly associated with increased mortality before and after propensity score matching (HR: 23.6 [95% CI: 14.6–38.1]; p<0.001 and HR: 20.5 [95% CI: 12.4-33.9], p<0.001 respectively) (Figure 1). Annualized mortality rate was strongly higher with LVEF<40% (15.1 versus 0.7, p<0.001).
Conclusion
In this large multicenter cohort of end-stage HCM patients, a LVEF cut-off point of 40% delineated a subgroup with substantially increased mortality, which remained significant after adjustment for traditional prognostic factors and was confirmed following propensity score matching.Figure 1Figure 2