Electrocardiographic characterization of myocardial scar in presence of conduction abnormality

Abstract: Presence, extent and localization of myocardial scar constitute prognostic clinical information that can influence therapeutical decisions. Cardiovascular magnetic resonance imaging with late gadolinium enhancement (CMR-LGE) is the preferred method to image myocardial scar in vivo but is resource intensive and has contraindications. The 12-lead electrocardiogram (ECG) is less accurate but inexpensive and widely available. Adaptations of the Selvester QRS score, an ECG analysis system for scar characterization, were developed in 2009 for use in presence of conduction abnormalities such as left bundle branch block (LBBB). Prior to the work described in this thesis, these newly developed Selvester QRS score adaptations had only been preliminarily validated against CMR-LGE. The overall aim of the thesis was to evaluate the diagnostic performance and clinical utility in predicting response to cardiac resynchronization therapy (CRT) of the 2009 Selvester QRS score for use in the presence of conduction abnormality, and to revise the LBBB version (2009 LBSS) based on empirical CMR-LGE data. We found that the ECG criteria comprising the 2009 LBSS had modest ability to localize CMR-LGE verified myocardial scar within the left ventricle (LV) (study I). Next, we evaluated the ability of the Selvester QRS score adaptations for use in LBBB, right bundle branch block (RBBB), left anterior fascicular block (LAFB) and combined RBBB+LAFB to identify and quantify myocardial scar as verified by CMR-LGE. The results revealed a tendency of the ECG method to overestimate scar burden and presence compared to CMR- LGE (study II). We then investigated the individual specificity of the 46 QRS morphology criteria that comprise the 2009 LBSS and found that certain criteria hade prohibitively low specificity, likely causing the overestimation of myocardial scar size and presence found previously (study III). We further evaluated the ability of all Selvester QRS score adaptations to predict response to CRT in a randomized cohort. Scar burden estimated by QRS scoring did predict clinical outcome in both study arms but could not distinguish who benefitted from CRT in terms of reduced risk of heart failure event or death (study IV). Finally, we assembled a large training dataset of 325 patients in which we performed careful continuous measurements on digital ECGs. We subsequently compared ECG measurements to myocardial scar quantified by CMR-LGE. We achieved an improved method of ECG scar detection in LBBB compared to the 2009 LBSS that remains to be tested in an independent population. The Selvester QRS score adaptations for presence of conduction abnormalities tends to overestimate scar and has limited correlation with CMR-LGE verified scar presence and extent. The 12-lead ECG likely contains insufficient information for accurate quantification of myocardial scar in LBBB. However, it may still be possible to distinguish between scar presence and absence using the 12-lead ECG in presence of LBBB.