Influence of the vectorcardiogram synthesis matrix on the power of the electrocardiogram-derived spatial QRS-T angle to predict arrhythmias in patients with ischemic heart disease and systolic left ventricular dysfunction

Abstract

Background and purpose: Several studies have demonstrated that the spatial mean QRS-T angle (SA) predicts cardiac events and mortality. Spatial mean QRS-T angle is a vectorcardiographic variable. Because in clinical practice, 12-lead standard electrocardiograms (ECGs) are recorded rather than vectorcardiograms (VCGs) according to Frank, VCGs are commonly obtained by synthesizing them from 12-lead ECGs, by using a VCG synthesis matrix. Hence, the thus computed SA is an estimate of the real SA measured in the Frank VCG. Recent studies have shown that Kors VCG synthesis matrix yields better estimates of SA than the inverse Dower VCG synthesis matrix. Our current study aims to compare the predictive power of these SA variants for the occurrence of potentially lethal arrhythmias.

Methods: The study group consisted of patients with ischemic heart disease and left ventricular systolic dysfunction who received an implantable cardioverter-defibrillator (ICD) for primary prevention. During follow-up, the occurrence of appropriate device therapy (occurrence of ventricular arrhythmia) was noted. Alternative SAs were computed in VCGs synthesized from standard 12-lead ECGs by using either the inverse Dower matrix (SA-Dower) or the Kors matrix (SA-Kors). Comparison of the predictive power of SA-Dower and SA- Kors was performed by receiver operating characteristic analysis, by Kaplan-Meier analysis, and by univariate and multivariate Cox regression analysis, using every 10th percentile of SA as a cutoff value.

Results: The study group consisted of 412 patients (361 men; mean ± SD age 63 ± 11 years), in which 56 patients had appropriate ICD therapy during follow-up. Receiver operating characteristic analysis revealed that the area under the curve of SA-Kors was significantly larger than area under the curve of SA-Dower (0.646 vs 0.607, P = .043). The discriminative power of SA-Kors for the absence/presence of appropriate ICD therapy in patients during follow-up was generally superior to SA-Dower over a wide range of cutoff values in the Kaplan-Meier analysis and generally yielded stronger hazard ratios in the univariate and multivariate Cox regression analyses.

Conclusion: If there is no specific reason to use the inverse Dower matrix, VCG synthesis from standard 12-lead ECGs should preferably be done by using the Kors matrix. It is likely to assume that already published studies in which the predictive value of SA-Dower was demonstrated would yield stronger results if the SA-Dower angles were substituted by SA-Kors angles.