Spatial indices of repolarization correlate with non-ST elevation myocardial ischemia in patients with chest pain

Abstract

Background Mild-to-moderate ischemia does not generally result in ST segment elevation on the ECG, but rather it frequently results in non-specific, dynamic changes in the T wave. These repolarization changes are frequently considered non-diagnostic for ischemia, warranting further diagnostics that are costly and time-consuming. Robust methods to quantify such T wave heterogeneity can have immediate clinical applications. We sought to evaluate the effects of spontaneous ischemia on the evolution of spatial T wave changes in patients undergoing nuclear cardiac imaging for evaluating intermittent chest pain. Methods We recruited non-traumatic chest pain patients (> 30 years old) referred for 2-stage, stress-rest, single-photon emission computed tomography (SPECT) scans to rule out focal myocardial ischemia. At baseline, consented subjects underwent 5-min, 12-lead ECG recordings using high-fidelity, digital ECG Holters (H12 + recorders, 1,000 s/s, 0.05–300 Hz, Mortara Instrument). Subjects then underwent a symptom-limited cardiac stress test with an abrupt recovery protocol (lying down supine when reaching peak test), where we recorded another 5-min peak-stress/recovery ECG while in supine position. After excluding those with LBBB or pacing, spatial variations in repolarization were quantified using T wave complexity (TWC) based on the eigenvalues of the spatial correlation matrix of the ECG (ratio of second to first eigenvalues of repolarization). Calculations were based on 10-second averaged beats, and we computed an ischemia change index based on the relative variation in TWC during peak stress to the normal variability observed in baseline recording. Results Our sample included 30 males and 20 females aged 63 ± 11 years. One in five patients developed mild-to-moderate ischemia on their stress SPECT scans. Although there were no differences in TWC values between groups at baseline, significant changes in TWC were seen during the entire peak-stress/recovery period in patients with active ischemia (n = 10), but not in those with negative ischemia (n = 40). Using a support vector machine model based on the ischemia change index, we predicted pre-SPECT probability of ischemia with a sensitivity of 0.8571 and specificity of 0.8000. Interestingly, we also observed that the magnitude of change in the ischemic index followed a pattern that depends on the severity of ischemia (greater dispersion with moderate ischemia) and the anatomic distribution of ischemic region (greater dispersion with LAD and LCX disease). Conclusions Our findings demonstrate that ischemia-induced changes in T wave morphology can be meaningfully quantified from the surface 12-lead ECG, suggesting an important opportunity for improving diagnostics in patients with chest pain.