ECG-based Risk Stratification of Ventricular Arrhythmia in Implantable Cardioverter Defibrillator (ICD) Patients

Aim: Implantable cardioverter defibrillators (ICDs) are life-saving devices invented in the 20th century that are now the standard of care for patients at high risk of ventricular arrhythmia, largely defined by either previous malignant arrhythmic event or depressed left ventricular ejection fraction (LVEF). However, nearly half of the current ICD population never receive appropriate therapy or benefit from their device, while still facing the risks and complications associated with it. Improvement in our ability to determine risk of ventricular arrhythmia is needed to allow for better patient selection for ICD implantation. This thesis examined the use of ECG-based parameters and their value in ventricular arrhythmia risk stratification in ICD patients. Methods and Results: Firstly a retrospective cohort of 138 ICD patients was reviewed. Over an average follow-up of 7 years, 55% of patients received appropriate shock therapy, and 25% received inappropriate therapy. In Chapter 3, manual assessment of 12-lead ECGs was undertaken in a retrospective cohort of 108 ICD patients. Increased QRS dispersion and presence of premature ventricular complex over 10sec were found to be associated with higher risk of appropriate therapy with a mean follow-up of 29±11 months. Manual assessment of paper 12-lead ECGs was limited in terms of the parameters that could be derived and measurement accuracy. In Chapter 4 and Chapter 5 the use of digital ECG analysis was explored. In Chapter 4, custom written software was developed to compute a large number of ECG parameters that characterise the depolarisation and repolarisation processes and autonomic function. In Chapter 5, the set of ECG parameters computed from Chapter 4 was reduced to a smaller subset where only measures that were relatively independent from conventional ECG measures and ones that distinguished ICD patients from normal volunteers were retained. In Chapter 6, the selected subset of ECG measures was correlated to structural indices of LV global remodelling and scar formation assessed by cardiac MRI. While some ECG parameters were significantly related to structural remodelling, others were relatively independent, implying the ECG is a useful tool in providing unique electrophysiological information. In Chapter 7, the ECG measures were related to the primary endpoint of appropriate therapy in a prospective ICD cohort of 201 patients. Two novel ECG measures, TWR rel (relative T wave non-dipolar component) and QMD (QRS morphology dispersion) were found to be independent predictors of arrhythmic events in addition to clinical variables of LVEF, secondary prevention and male gender. Conclusion: ICDs are currently implanted in patients considered at high risk of ventricular arrhythmia; however, our ability to characterise this risk is limited. Two digitally derived novel ECG measures were identified that held incremental prognostic value to the traditional risk markers of LVEF and secondary prevention indication and these were also independent of structural indices. This suggested that digital analysis of the ECG provides information on the electrical vulnerability of the heart to arrhythmia. These results need to be confirmed with larger prospective cohorts and ideally in subgroups of patients of specific aetiologies. Combination of multiple risk factors to assess arrhythmic risk is likely to be the desired approach for better risk stratification in the future