Possibilities for predicting ventricular tachyarrhythmias in patients with heart failure with reduced ejection fraction based on surface electrocardiography. First results from a single-center prospective study

Abstract

According to current clinical guidelines, the risk of life-threatening ventricular tachyarrhythmias (VTAs) in patients with heart failure (HF) is determined by left ventricular ejection fraction (LVEF). The available clinical and experimental data indicate the imperfection of this one-factor approach, which specifies the need to search for new predictors of VTAs. In this prospective study, we performed a comparative analysis of surface electrocardiographic parameters in HF patients with LVEF ≤35% without syncope or sustained ventricular arrhythmias in history, who were implanted with cardioverter defibrillator as a primary prevention of sudden cardiac death. During the two-year follow-up, the primary endpoint (new-onset persistent VTA episode, or VTA/ventricular fibrillation that required electrotherapy) was recorded in 42 patients (25,5%). The secondary endpoint (an increase in LVEF by 5% or more of the initial level against the background of cardiac resynchronization therapy) was more often recorded in the group of patients without VTAs (41 (33%) vs 4 (9,5%), p=0,005). The studied cohort of patients was characterized by a left axis deviation (72%), LV hypertrophy signs (84%), impaired intra-atrial (P wave duration of 120 (101-120) ms) and intraventricular conduction (QRS duration of 140 (110-180) ms), ventricular electrical systole prolongation (QTcor — 465 (438-504) ms). Differences between the groups divided depending on reaching the primary endpoint in terms of the Cornell product, Cornell voltage index and ICEB, as well as the detection rate of complete left bundle branch block morphology had levels of significance close to critical (p=0,09; p=0,05; p=0,1; p=0,09, respectively). The multivariate predictive model included following factors: Cornell product, Tp-Te/ QRS, P wave duration (diagnostic efficiency of the model was 60%: sensitivity, 61,1%, specificity, 59,6%; p=0,007).