T Wave Spectral Variance for Noninvasive Identification of Patients with Idiopathic Dilated Cardiomyopathy Prone to Ventricular Fibrillation Even in the Presence of Bundle Branch Block or Atrial Fibrillation

Conventional methods using Holter ECG recordings for noninvasive risk stratification are limited in patients with idiopathic dilated cardiomyopathy (IDC) prone to ventricular fibrillation (VF) having atrial fibrillation (AF) or bundle branch block (BBB). We therefore investigated, whether spectral assessment of beat-to-beat alternations of repolarization is associated with VF in these patients. Twenty-four-hour Holter ECG recordings in 462 patients with IDC were used. The VF group comprised of 64 consecutive patients who survived cardiac arrest, the no VF group consisted of 398 consecutive patients without a history of malignant ventricular arrhythmia. One hundred patients with ischemic cardiomyopathy (ICM) served as a control group. In each patient, 1,024 consecutive T waves were aligned using cross correlation methods. Two-dimensional Fourier transform (2D FFT) used the data matrix of 1,024 consecutive 200-ms segments centered to the T wave peak. Power spectra of the 2D FFT revealed the frequency content of the T wave in the first dimension and the periodicity of this frequency content in the second dimension. The ratio between periodic frequency contents and the sum of nonperiodic and periodic frequency contents between 0.5 and 50 Hz is equal to the T wave spectral variance (TWSV) index. Thus, TWSV index = 0 would mean that all 1,024 T waves are identical and TWSV index = 1 would mean that the 1,024 T waves are totally variable. The TWSV index was significantly higher in the VF group (0.93 +/- 0.14) than in the no VF group (0.53 +/- 0.13, P < 0.01). The best cutoff between the VF and the no VF group was achieved by using a TWSV index of 0.75 (sensitivity = 89%, specificity = 78%). No significant differences were observed between patients with and without AF or with and without BBB, and between patients with IDC and ICM. Even in the presence of BBB or AF spectral assessment of T wave alternations by TWSV index using 2D FFT in Holter ECG recordings, allows the identification of patients with IDC at risk for VF.