Sudden Cardiac Death in Patients with Previous Acute Coronary Syndrome

Cardiovascular diseases remain the leading cause of global mortality across all income levels. Sudden cardiac deaths (SCD) or sudden cardiac arrests (SCA) are believed to be the cause of roughly half of all cardiovascular deaths and are primarily attributed to coronary artery disease (CAD). SCAs are characterized by an abrupt cessation of the heartbeat, leading to unconsciousness, hemodynamic collapse, and death if not treated. SCDs are a subgroup of SCAs but are defined as a sudden death of cardiac origin occurring within an hour of symptom onset or within 24 hours of the individual having been last seen to be asymptomatic. Ventricular fibrillation (VF), often preceded by ventricular tachycardia (VT), is a common underlying mechanism. Preventive measures for SCD include medical therapy, catheter ablation therapy, and implantable cardiac defibrillators (ICD). While a low left ventricular ejection fraction (LVEF) is a known risk factor for SCD, most victims have a preserved LVEF, making risk stratification challenging. The role of a prior SCA as a long-term risk factor is debated, particularly when occurring during an episode of acute coronary syndrome (ACS). Other established risk factors include male sex, diabetes, and smoking. Active smoking is considered to be a risk for SCD in the general population, whereas former smoking is not. However, in patients with established CAD, the studies are few and the results have been somewhat disputable. In addition, multiple electrocardiogram-based (ECG-based) risk factors for SCD have been identified. These include heart rate, myocardial scarring, and abnormal depolarization/repolarization. Clinical SCD risk assessment using the ECG is still under further investigation. The methodological challenges in identifying novel ECG risk markers include the heterogeneity of ECG parameter interpretation, small sample sizes of previous studies, and a lack of standardized ECG data. For these reasons, our study aimed to clarify the current epidemiology of SCA and SCD, smoking and VF/VT as risk factors for SCD, and the possibility of utilizing ECG risk markers in clinical SCD risk assessment.

The results of this dissertation are based on a retrospective analysis of over 10,000 consecutive patients undergoing invasive evaluation for ACS between 2007 and 2018 at Tays Heart Hospital in Pirkanmaa, Finland. Data were collected from electronic health records, written death certificates, and patient records to study the incidence of SCA and SCD.

The purpose of this study was I) to assess the incidence of SCD (and that of SCA) and review the epidemiology of these incidents in patients with a previous ACS in the modern setting; II) to re-evaluate whether a hemodynamic collapse due to VF/VT during ACS is a risk factor for SCD in the future in patients without guideline-implicated preventive ICD implantation (LVEF > 40%); III) to clarify the role of smoking as a risk factor in patients with established CAD; IV) as well as to investigate whether computerized ECG interpretation (CIE) and machine learning (ML) algorithms could improve the SCD risk prediction.

We found a 9.8% cumulative incidence of SCAs (0.8% annually) and a 5.4% incidence of SCDs (0.5% annually) over 12 years of follow-up, with SCAs and SCDs accounting for 30.5% and 18.3% of all cardiovascular deaths, respectively (Study I). Although these estimates are generally lower than in previous studies, SCD is still one of the most important single causes of death in these patients.

When focusing on patients with ACS and a preserved LVEF of ≥ 40%, we found that acute-phase VF/VT is, contrary to most previous studies, a risk factor for SCD long-term. Among 211 patients who survived acute-phase VF/VT, the long-term risk of future SCD was over three times higher, with most life-threatening arrhythmias occurring in patients without ICDs (Study II). Thus, VF or VT that results in a hemodynamic collapse during an ACS episode should be considered as a significant risk factor for subsequent events in these patients. Furthermore, our result reinforced the notion that active smoking is a major risk factor of SCDs, with an approximately 79% higher adjusted risk when compared to non-smokers and ex- smokers (Study III).

Finally, we employed three machine learning algorithms to predict the five-year occurrence of SCDs or equivalent events in ACS patients, achieving a modest performance with area under the curve (AUC) values ranging from 0.652 to 0.693 and identifying QRS duration, nonspecific intraventricular conduction delay (NIVCD), incomplete left bundle branch block (LBBB), heart rate-corrected QT time (QTc), lateral ST segment depression, and premature ventricular contractions as significant predictors of SCD (Study IV).

In conclusion, SCD is a significant contributor to mortality among patients with ACS, although the incidence is lower than previously published estimates (approximately 0.5% annually). Moreover, active smoking is an important risk factor for SCD, and VF/VT during ACS is associated with a notably high long-term risk of SCD in patients with a LVEF of ≥40%, and ML and CIE successfully identified risk features for SCD. This proof-of-concept study demonstrates a potential tool for characterizing high-risk individuals with clinical risk markers and ECG parameters.