ECG quantification of myocardial scar in cardiomyopathy patients with or without conduction defects: correlation with cardiac magnetic resonance and arrhythmogenesis

Abstract

Background: Myocardial scarring from infarction or nonischemic fibrosis forms an arrhythmogenic substrate. The Selvester QRS score has been extensively validated for estimating myocardial infarction scar size in the absence of ECG confounders, but has not been tested to quantify scar in patients with hypertrophy, bundle branch/fascicular blocks, or nonischemic cardiomyopathy. We assessed the hypotheses that (1) QRS scores (modified for each ECG confounder) correctly identify and quantify scar in ischemic and nonischemic patients when compared with the reference standard of cardiac magnetic resonance using late-gadolinium enhancement, and (2) QRS-estimated scar size predicts inducible sustained monomorphic ventricular tachycardia during electrophysiological testing.

Methods and results: One hundred sixty-two patients with left ventricular ejection fraction < or =35% (95 ischemic, 67 nonischemic) received 12-lead ECG and cardiac magnetic resonance using late-gadolinium enhancement before implantable cardioverter defibrillator placement for primary prevention of sudden cardiac death. QRS scores correctly diagnosed cardiovascular magnetic resonance scar presence with receiver operating characteristics area under the curve of 0.91 and correlation for scar quantification of r=0.74 (P<0.0001) for all patients. Performance within hypertrophy, conduction defect, and nonischemic subgroups ranged from area under the curve of 0.81 to 0.94 and r=0.60 to 0.80 (P<0.001 for all). Among the 137 patients undergoing electrophysiological or device testing, each 3-point QRS-score increase (9% left ventricular scarring) was associated with an odds ratio for inducing monomorphic ventricular tachycardia of 2.2 (95% CI, 1.5 to 3.2; P<0.001) for all patients, 1.7 (1.0 to 2.7, P=0.04) for ischemics, and 2.2 (1.0 to 5.0, P=0.05) for nonischemics.

Conclusions: QRS scores identify and quantify scar in ischemic and nonischemic cardiomyopathy patients despite ECG confounders. Higher QRS-estimated scar size is associated with increased arrhythmogenesis and warrants further study as a risk-stratifying tool.

Figure 1

Timing of electrical activation (depolarization) wavefronts in normal conduction (A) and LBBB (B), shown in sagittal view. For reference, two QRS-T waveforms are shown in their anatomic locations (V3 on the chest and aVF inferiorly). Electrical activation starts at the small arrows and spreads in a wavefront with each colored line representing successive 10 ms. In normal conduton, activation begins within both the LV and RV endocardium. In LBBB, activation only begins in the RV and must proceed through the septum before reaching the LV endocardium (i.e. this pattern in the septum is opposite to that seen in normal conduction). By taking into account the stereotypical LBBB activation, QRS-score criteria for scar can be in fact developed in LBBB, similar to that in normal conduction. Note that while scar in the septum causes Q-waves in V1-V3 when normal conduction is present, the same scar causes large R-waves in V1-V3 in the presence of LBBB because of unopposed electrical forces in the RV free wall (see Figure 2A).

Figure 2

ECGs with QRS scoring and short-axis CMR images from two patients with LBBB. For the CMR images, the core regions are shown in red and the gray zone in yellow (note that the corresponding 4 chamber long axis view is also shown with the arrow denoting the septal mid-wall LGE). For comparison with the QRS-score, total CMR scar was defined as core+1/2gray (see text). The complete LBBB QRS-score is shown in the Appendix. Patient A has a nonischemic cardiomyopathy with midwall anteroseptal scar comprising 7% of the LV by CMR-LGE and received 5 QRS points (ECG-estimated scar=15%). Note the large R-waves in V1-V2 that reflect anteroseptal scar. Patient B has an ischemic cardiomyopathy with inferior and posterolateral scar comprising 23% of the LV by CMR-LGE and received 8 QRS points (ECG-estimated scar=24%). Note the large S/S′ ratio in V1-V2 which reflects posterolateral scar.

Figure 3

ROC curves of QRS-scores to diagnose the presence of CMR-LGE scar for (A) all patients, (B) nonischemics, (C) no confounders, (D) only confounders, (E) LVH and (F) LBBB. QRS point cutoffs and areas under the curves (AUC) are shown.

Figure 4

Scatterplots and Bland-Altman plots of ECG-estimated vs. CMR-estimated scar quantification for (A) no confounders, (B) LVH, (C) LAFB and/or RBBB, and (D) LBBB. Regression equations and Spearman correlation coefficients are shown.