Relative Predictive Value of T Wave Alternans and Left Ventricular
Ejection Fraction for Death and Sustained Ventricular Arrhythmias in Patients
with Left Ventricular Dysfunction
Category: 09 Signal Average ECG/T-Wave Alternans
Presentation Time: Friday, 8:30 a.m. - 8:45 a.m.
J. Thomas Bigger, MD, Michael K. Parides, PhD, Richard C. Steinman, BA, Pearila
B. Namerow, PhD, Daniel M. Bloomfield, MD, for the TWA in CHF Investigators.
Columbia University Medical Center, New York, NY
Presentation Number: AB27-2
Left ventricular ejection fraction (LVEF) is considered the “gold standard” for
predicting risk in heart disease. We conducted a prospective longitudinal study
to evaluate the relative predictive value of LVEF and T wave alternans (TWA)
alone or jointly in patients with LVEF <0.41.
Methods: Patients with ischemic (IHD) or non- ischemic (N-IHD) heart disease
were eligible if they had LVEF <0.41, sinus rhythm, NYHA class I-III, and no
atrial fibrillation or history of sustained ventricular arrhythmias (SVA). All
patients had a TWA exercise test and were followed for up to two years (average
20 months). Our composite endpoint included death of any cause and non-fatal
SVA. We used Cox regression to estimate the univariate and multivariate hazard
ratios (HR) relating LVEF and/or TWA to death/ SVA and to test for an
interaction between LVEF and TWA.
Results: We studied 549 patients: 49% had IHD, 74% had LVEF <0.31 (average
0.25), 66% had an abnormal TWA test; and 51 experienced an endpoint (40 deaths,
11 SVA). TWA and LVEF were only weakly associated (odds ratio = 1.25, 95% CI
0.84-1.85). The actuarial 2-year event rate was 11.4%. The univariate hazard
(HR) for death/ SVA was 1.8 (95% CI 0.88-3.73) for LVEF <0.31 vs ≥0.31
and 6.5 (95% CI 2.35-18.11) for TWA abnormal vs normal. In multivariate Cox
models, TWA added predictive value to LVEF, but LVEF did not add significant
predictive value to TWA; no significant interaction was found. The multivariate
HR when LVEF and TWA were both in a Cox model were 1.7 (95% CI 0.84-3.53) and
6.4 (95% CI 2.31-17.79) respectively. The false negative rate was 6.3% for LVEF
and 2.1% for TWA.
Conclusion: In both IHD and N-IHD, TWA was a significantly better univariate
and multivariate predictor of death and SVA than LVEF. TWA also had a
substantially lower false negative rate than LVEF. Moreover, TWA added
substantial predictive value to LVEF <0.31; the converse was not true. Thus,
TWA is a better predictor of death/SVA than LVEF among patients with LVEF
<0.41.