Predictive power of T wave alternans in patients with left ventricular dysfunction and QRS prolongation
Presentation Time:	5/6/2005 8:45:00 AM
Author Block:	Sarah B. Levin, MD, J. Thomas. Bigger, MD, Richard C. Steinman, BA, Michael K. Parides, PhD, Daniel M. Bloomfield, MD, Pearila B. Namerow, PhD, TWA in CHF Investigators. Columbia Presbyterian Medical Center, New York, NY
Disclosures:	S.B. Levin, None; J.T. Bigger, None; R.C. Steinman, None; M.K. Parides, None; D.M. Bloomfield, Cambridge Heart Speakers Bureau; P.B. Namerow, None.
Background: T wave alternans (TWA) is associated with increased risk of arrhythmic events in patients with left ventricular ejection fraction (LVEF) < 0.41. However, it has been reported that TWA lacks predictive power in patients with QRS > 120 msec, i.e., that there is a significant interaction between TWA and QRS with respect to death and sustained ventricular arrhythmias (SVA).This analysis examines the relationship between TWA and QRS duration in 549 patients with LVEF < 0.41. Our hypothesis was that TWA predicts non-fatal SVA and death in patients with LVEF < 0.41 regardless of QRS duration. Methods: The 549 eligible patients had LVEF < 0.41, and had no history of SVA, class IV heart failure or recent myocardial infarction or coronary bypass graft surgery (CABG). All patients underwent TWA and results were categorized as normal or abnormal. Follow-up was for 2 years and the composite endpoint was all-cause mortality and non-fatal SVA. Cox models were used to determine the independent predictive accuracy of TWA and QRS and whether there was an interaction between the two. Results: Of the 549 patients; 49% had ischemic cardiomyopathy, 51% had non-ischemic cardiomyopathy, 27% had QRS > 120 msec, and 66% had an abnormal TWA test. During follow-up there were 51 events (11 non-fatal SVA and 40 deaths). Univariate Cox regression models demonstrated a HR for TWA of 6.5 (95% CI 2.4-18.1) and 1.6 (95% CI 0.9 -2.9) for QRS duration. A Cox model constructed to test for interaction between QRS duration and TWA showed no significant interaction (z-statistic 0.82, p = 0.33). Conclusion: TWA was a strong predictor of adverse events in patients with QRS ≤ 120 msec or > 120 msec. Additionally, TWA had a low false negative rate regardless of QRS duration. Lack of interaction between TWA and QRS indicates that TWA can risk stratify patients with left ventricular dysfunction, regardless of QRS duration.