Can a simulation study of T-wave alternans resolve whether TWA is T-wave amplitude dependent?

Abstract

To the Editor: The study by Cuesta-Frau et al. [1] published online in the Journal [1] describes a preprocessing stage of curve alignment, based on continuous dynamic time warping, to improve the reliability of T-wave alternans (TWA) of modified moving average (MMA) analysis. Referring to the clinical significance of TWA, the authors emphasize that higher values of this index are associated with greater risk for malignant arrhythmias. Thus it is important not only to detect TWA, but also to assess it quantitatively. An outline of the different methodologies of spectral and timedomain analyses is offered. In reference to the MMA method the authors specifically indicate that it measures an averaged maximum absolute difference of even and odd heartbeats at various points of the T-waves or ST–T intervals. The MMA method is explicitly described, and its advantages over other methods are explained. The authors go on to elaborate in great detail on their enhanced MMA (EMMA) analytical method, which they have applied in simulation studies employing ECGs derived from a synthetic ECG generator. The EMMA appears to be more robust than the MMA against wave detection errors, noise, phase shifts, and baseline wandering. Recently it has been theorized that the TWA may be Twave amplitude dependent [2–5]. Although this alleged dependence applies to the entire J–T interval, for the sake of argument one can concentrate only on the TWA calculated from the peak of the T-waves. Implicit to the notion of dependence of the TWA magnitude on the corresponding T-wave amplitude is that, if the former is adjusted by the latter, a more physiologically meaningful TWA index could emerge. The above is a corollary of observations that the T-waves often change in morphology, amplitude, and polarity without an apparent reason, and thus by adjusting for the above, the calculated TWA would reflect more accurately the underlying arrhythmogenic risk, devoid of the ‘‘extraneous’’ elements of the T-wave (or J–T interval) volatility. Wide experience in the exercise laboratory and in ambulatory ECG recording, reveals that even normal subjects show large variations of T-waves related (but often unrelated) to heart rate changes. It should not be inappropriate to consider that even patients with underlying pathology could experience such changes in the T-waves. How then are such changes affecting the detection and quantitation of the TWA? Is it possible that the calculated TWA values in a patient reflect an expression of the underlying arrhythmogenicity, and other’’ non-arrhythmogenic elements’’, consequent to the T-wave changes. These possibilities would be more applicable to patients with serial TWA assessments. It is frequently observed that stable patients over short or long periods of time show remarkable changes in T-waves; how such changes impact the quantitative assessment of the TWA? Authors of papers, in which simulations have been employed, often discuss apologetically the non-clinical relevance of their contributions, and the need for clinical corroboration of their conclusions. However simulations have also their advantages, and in this spirit the MMA and the EMMA methods used by Cuesta-Frau et al., in conjunction with their generator of synthetic ECGs, provide an opportunity to evaluate whether TWA is T-wave dependent, by applying their methods of analysis to different sets of derived ECG data with altered characteristics (e.g., amplitude) of the T-waves. J. E. Madias (&) Division of Cardiology, Mount Sinai School of Medicine of the New York University, Elmhurst Hospital Center, 79-01 Broadway, Elmhurst, NY 11373, USA e-mail: madiasj@nychhc.org