T wave (repolarization) alternans: Overview of technical aspects and clinical applications

INTRODUCTION — T-wave alternans (TWA) refers to periodic beat-to-beat variability in the timing, shape, and/or amplitude of T-waves on the surface electrocardiogram (ECG) [1,2]. TWA therefore reflects sudden changes in temporal heterogeneity in ventricular repolarization, which is an important mechanism underlying reentrant arrhythmias [3,4].

Despite the term "T-wave" alternans, alternating behavior may also involve the ST segment and U wave (figure 1) [2,5-8]. The term repolarization alternans is thus more accurate.

TWA of sufficient magnitude to be seen with visual inspection is uncommon. With the development of computerized filtering and spectral analysis tools, TWA on the order of microvolts can be detected (figure 2), and microvolt TWA (MTWA, often used synonymously with TWA) is both sensitive and specific for predicting ventricular arrhythmias in a number of clinical scenarios [1,2,7,9-16].

Clinical risk stratification for sudden cardiac death typically focuses on the presence of a reduced left ventricular ejection fraction and the presence of heart failure. However, these parameters identify only a minority of individuals who will die suddenly, and also identify others who will not suffer lethal ventricular arrhythmias [1,2]. As a result, there is continued interest in additional electrophysiologic indices for risk prediction, of which one of the most promising remains TWA [2].

The pathophysiology of TWA, along with the technical aspects and clinical utility of TWA testing, will be reviewed.

PATHOPHYSIOLOGY OF TWA — Originally, microvolt- and macrovolt-level TWA were considered to be distinct entities. However, studies from isolated guinea pig and rabbit hearts now suggest that these forms of TWA may differ only quantitatively and represent the same pathophysiologic processes of dispersion of repolarization [4,17]. (See "Cardiac excitability, mechanisms of arrhythmia, and action of antiarrhythmic drugs".)

Distinct hypotheses have been proposed to explain TWA [4,18]:

●Tissue heterogeneity – According to this hypothesis, dispersion of repolarization results in differences in electrical recovery time and conduction among myocytes [19,20]. Longer recovery time, or delayed activation time, in some myocytes may alter their depolarization from beat-to-beat [17,21,22].

●Dynamic functional tissue factors – TWA may be due to dynamic functional tissue factors. These include the rate-dynamics (restitution) of repolarization and conduction velocity [19,20].

●Mechanical and ionic abnormalities – The mechanism for TWA, and its arrhythmic link in patients at risk for sudden cardiac arrest may be influenced by mechanical and ionic abnormalities in heart failure. Acute volume overload exaggerates TWA, suggesting a role for mechanical stretch [23]. At the ionic level, abnormal calcium cycling is emphasized in the etiology of TWA, and may underscore the prevalence of TWA in patients with heart failure and cytosolic calcium loading [4,24].

Even though the listed mechanisms appear to clearly explain repolarization alternans in animals, some evidence suggests that the above mechanisms may not fully contribute to clinical TWA. A major source of this discrepancy is that clinical TWA must be measured at relatively slow heart rates in order to be predictive (≤109 beats/minute) [25]. However, these heart rates do not engage steep action potential duration or conduction restitution, and action potential duration therefore does not fluctuate at these relatively slow heart rates [26].

An important issue relates to the relationship between T-wave and mechanical alternans. Studies have shown that visible TWA accompanies mechanical alternans and also that microvolt-level TWA yields a mechanical effect [27-30].

OUR APPROACH TO TWA TESTING — TWA testing has been evaluated in a variety of settings. The greatest utility of TWA lies in its high negative predictive value and in identifying low-risk individuals. Our approach is as follows:

●We perform TWA testing in patients who meet established criteria for a primary prevention ICD but who are reluctant to proceed. Positive (abnormal) TWA results in this setting may help to stratify the patient to a higher risk group, while negative (normal) TWA results may reassure the patient of a lower risk of SCD. (See "Primary prevention of sudden cardiac death in patients with cardiomyopathy and heart failure with reduced LVEF".)

●We perform TWA testing in patients with known or suspected long QT syndrome (LQTS) as part of the evaluation for diagnostic and risk stratification purposes. A 2016 study demonstrated that the presence of microvolt TWA is more common in patients with LQTS than previously reported and may be associated with a prior history of torsade de pointes [31]. This supports historical data of macroscopic TWA reported in LQTS patients. (See "Congenital long QT syndrome: Diagnosis".)

●We do not perform TWA testing to screen for the risk of ventricular arrhythmias in the population without known coronary heart disease (CHD). We also do not routinely perform TWA testing in patients with known CHD.

●We do not routinely perform TWA testing in patients who meet criteria for a primary prevention ICD. It is not known if the negative predictive value of TWA is high enough to withhold ICD implantation in such patients [32].

EVIDENCE FOR TWA TESTING — TWA is primarily used as a tool for the risk stratification for sudden cardiac death (SCD) [33]. Most of the focus of such efforts has been on patients with prior myocardial infarction (MI), reduced left ventricular (LV) ejection fraction, and/or symptomatic heart failure (HF) [34-36].

TWA was first linked to ventricular arrhythmias in the early 1900s in cases that were evident on visual inspection of the ECG [37,38]. Visually apparent TWA was subsequently reported in a variety of settings including ischemia and long QT syndrome (LQTS) [39,40]. TWA may accompany pulsus (mechanical) alternans in severe cardiac failure, but may occur without mechanical alternans, and vice versa. (See "Congenital long QT syndrome: Epidemiology and clinical manifestations".)

Historically, studies of TWA may be divided into three phases:

●Studies showing that TWA is concordant with the results from programmed ventricular stimulation at electrophysiologic (EP) testing.

●Studies showing that TWA predicts spontaneous ventricular arrhythmias in patients with LV dysfunction, with or without coronary heart disease (CHD) and with or without symptomatic HF.

●Studies evaluating TWA for predicting spontaneous ventricular arrhythmias in other patient populations, including those with left ventricular hypertrophy or with end-stage kidney disease.

The greatest utility of TWA lies in its high negative predictive value and therefore in identifying low-risk individuals. Studies show that the negative predictive value of TWA for arrhythmic events occurring within one year exceeds 90 percent (range 90 to 99 percent) [41].

Screening unselected populations — Several attempts have been made to extend current implantable cardioverter-defibrillator (ICD) risk-stratification guidelines to unselected patients. Accordingly, studies have examined the predictive value of TWA in patients undergoing exercise stress testing and not traditionally identified as being at high risk for ventricular arrhythmias [42-45]. Based upon the available data, we do not routinely proceed with TWA testing as a screening tool in unselected populations of patients without known CHD.

●In the largest study of 3598 patients undergoing stress testing, the risk of cardiovascular mortality and sudden cardiac death increased progressively with each 20-microvolt increase in TWA in lead V5 (hazard ratios [HR] 1.5, 95% CI 1.2-2.1 and 1.6, 95% CI 1.0-2.4, respectively, for each 20-microvolt increase) [45].

●In a smaller study of nearly 2000 patients undergoing stress testing from the same population, abnormal TWA and heart rate recovery were assessed [43]. Both abnormal TWA and abnormal TWA combined with abnormal heart rate recovery were associated with an increased risk for cardiovascular death (relative risk [RR] 5.0, 95% CI 2.1-12.1 and 6.1, 95% CI 2.8-13.2, respectively). (See "Prognostic features of stress testing in patients with known or suspected coronary disease", section on 'Heart rate recovery after exercise'.)

Patients with CHD — Several patient populations with CHD can be considered for TWA testing. In general, such patients (with or without prior MI) represent the population most extensively studied for ventricular arrhythmia risk stratification using TWA [46-48]. Based upon the available data, we do not routinely proceed with TWA testing in all patients with CHD. However, we consider TWA testing for additional risk stratification in high-risk patients who are candidates for an ICD for primary prevention of sudden cardiac death but who are reluctant to proceed with ICD implantation. (See 'Further risk stratifying candidates for ICD' below and "Primary prevention of sudden cardiac death in patients with cardiomyopathy and heart failure with reduced LVEF".)

The following observations illustrate the range of findings:

●In a prospective study of 102 patients with a recent (7 to 30 days) MI (follow-up 13 months), the risk of sustained ventricular tachycardia (VT) or ventricular fibrillation (VF) was increased in patients with TWA, particularly those who also had late potentials (figure 3 and figure 4) [47]. The sensitivity and specificity of TWA for predicting arrhythmic events were 93 and 59 percent, respectively.

●In a multicenter study of 850 consecutive MI survivors, TWA was assessed at 2 to 10 weeks in 701 patients, and from several months to two years in 149 [46]. The risk of death or resuscitated VF was predicted best by TWA (RR 5.9) and a left ventricular ejection fraction (LVEF) <40 percent (RR 4.4). The sensitivity and specificity of TWA for predicting arrhythmic events were 92 and 61 percent. TWA was also predictive of arrhythmic events in a larger series limited to post-MI patients with a preserved ejection fraction (LVEF ≥40 percent) [44].

●In a large observational study of 768 high-risk patients with ischemic cardiomyopathy (over 80 percent had a prior MI, average LVEF was 27 percent, and 51 percent had an ICD) with a mean follow-up of 18 months, a non-negative TWA result was associated with increased all-cause and arrhythmic mortality (adjusted HR 2.24 and 2.29, respectively, compared with patients with a negative TWA) [49].

These results provide evidence that TWA retains independent prognostic value when applied to a high-risk population. In theory, there is considerable evidence that alterations in autonomic balance, with a sympathetic predominance evidenced by heart rate variability analysis, may precede spontaneous ventricular arrhythmias and hence augment TWA [50]. This has been demonstrated by some authors, while beta blockers are known to suppress TWA [51,52]. A general discussion of risk stratification for arrhythmic death after MI is presented separately. (See "Incidence of and risk stratification for sudden cardiac death after myocardial infarction".)

LV systolic dysfunction and/or heart failure — Left ventricular systolic dysfunction in patients with HF is the primary mode of risk stratification for ventricular arrhythmias. Systolic dysfunction (LVEF ≤40 percent) identifies patients with ischemic and nonischemic cardiomyopathy who are at risk for serious ventricular tachyarrhythmias and SCD. We do not routinely perform TWA testing in all patients who otherwise meet criteria for a primary prevention ICD, as it is not known if the negative predictive value is high enough to withhold ICD implantation in patients who otherwise meet ICD implantation criteria. The data supporting these conclusions are presented separately. (See "Incidence of and risk stratification for sudden cardiac death after myocardial infarction", section on 'LV dysfunction' and "Ventricular arrhythmias: Overview in patients with heart failure and cardiomyopathy".)

Multivariate analyses have shown that TWA predicts ventricular arrhythmias independent of reduced LVEF [2,13,46,47], albeit in varying populations.

●In a series of 850 post-MI patients, the sensitivity and specificity of an LVEF <40 percent for ventricular tachyarrhythmias were 56 and 83 percent, respectively [46]. Sensitivity and specificity were not significantly different (52 and 92 percent) in the smaller subset of patients who had both a low LVEF and TWA.

●The Alternans before Cardioverter-Defibrillator (ABCD) trial compared the effectiveness of ICD implantation guided either by a positive T-wave alternans test or by an EP study in 566 patients with coronary disease and reduced systolic function (LVEF <40 percent) [53]. TWA-guided therapy achieved one-year positive (9 percent) and negative (95 percent) predictive values comparable with those from electrophysiology study (EPS) guided therapy (11 percent and 95 percent, respectively). Arrhythmic event rates were significantly higher in patients with a positive TWA (HR 2.1) or positive EPS (HR 2.4), although with positive predictive values of around 10 percent. Moreover, event rates in patients with negative TWA and EPS were lower than in those with two positive tests (2 versus 12 percent). The authors concluded that ICD implantation was equally effective whether guided by noninvasive MTWA or invasive EPS at one year, and complementary when both tests were combined.

●In the TWA in CHF study, which included 549 patients with ischemic and nonischemic cardiomyopathy who were followed for an average of 20 months, patients with abnormal TWA were more likely to die or experience sustained VT (HR 6.5, 95% CI 2.4-18.1) [54].

Several studies have suggested that TWA predicts spontaneous ventricular arrhythmias and mortality in patients with LV dysfunction that is nonischemic or out of proportion to the extent of CHD. In a report of 137 patients with nonischemic cardiomyopathy, the presence of TWA predicted the combined endpoint of sudden death, resuscitated VF, or documented hemodynamically unstable VT [55]. Additionally, the presence of HF predicts worse outcomes independent of LV function, and TWA predicts arrhythmic events in patients with HF [56-59]. As examples:

●In a study of 650 ambulatory patients with mild to moderate HF who underwent 24-hour ambulatory monitoring and were followed for a median of 48 months, patients without TWA had significantly higher survival [59].

●The ALPHA trial tested the prognostic utility of TWA for predicting the combined endpoint of cardiac death and life-threatening arrhythmias in patients with nonischemic dilated cardiomyopathy (LVEF <40 percent) and New York Heart Association (NYHA) class II and III HF over 18 to 24 months of follow-up [16]. TWA retained a high negative predictive value for the combined endpoint, although the overall event rate of the population was also very low [58].

Further risk stratifying candidates for ICD — There is ongoing interest in further risk stratification of patients who meet criteria for ICD therapy based upon MADIT II (LVEF ≤30 percent due to a prior MI) and SCD-HeFT (LVEF ≤35 percent due to any cause with associated NYHA class II or III HF) criteria [60,61]. Although each of these trials demonstrated mortality benefit with ICD therapy in their respective populations, the absolute benefit was lower than that seen in the earlier MUSTT and MADIT I trials, suggesting that subsets of patients with lower individual risk are included within the population encompassed by MADIT II and SCD-HeFT inclusion criteria. (See "Primary prevention of sudden cardiac death in patients with cardiomyopathy and heart failure with reduced LVEF".)

Although the data from several studies are conflicting, the bulk of the evidence suggests that TWA predicts mortality from arrhythmic events. This is particularly true for studies that utilized clinical endpoints, but less true for studies that used ICD endpoints (with their tendency to overestimate the incidence of arrhythmic events) [62]. TWA testing has high negative predictive value for ventricular arrhythmias in patients with coronary disease; but from the ABCD study, TWA may have to be repeated on an annual basis to be used in this way [53]. Moreover, the negative predictive value, which is sufficient to withhold ICD implantation in an otherwise indicated patient, has yet to be resolved. As such, we do not routinely proceed with TWA testing for risk stratification purposes in patients who are otherwise candidates for an ICD.

Data regarding the utility of TWA for further risk stratifying potential ICD recipients are mixed in terms of the potential benefit of TWA:

●A post hoc analysis of two prospective studies included 129 patients meeting MADIT II criteria who underwent TWA testing [63]. Over a mean follow-up of 17 months, there was significant increase in SCD in patients with TWA (15.6 versus 0 percent without TWA). The absence of SCD in patients with a negative TWA study supports the hypothesis that this test may select low-risk patients from within an otherwise high-risk cohort.

●In a subset of patients from a prospective multicenter trial of TWA, among 177 patients who met MADIT II indications for ICD placement, 32 percent had a normal TWA study [54,64]. At an average follow-up of 20 months, total mortality was lower in patients with a negative TWA (3.8 versus 17.8 percent in patients with an abnormal or indeterminate TWA, HR = 4.8).

●In a prospective cohort study of 768 patients with an ischemic cardiomyopathy (LVEF ≤35 percent) and no history of sustained ventricular arrhythmias, only the subset of patients with a non-negative (ie, abnormal or indeterminate) TWA study derived survival benefit from ICD placement [65].

●The Master study was a prospective observational cohort study that evaluated the correlation between TWA testing and event rates in patients receiving an ICD for routine indications. TWA failed to predict the primary endpoint of arrhythmic events, although it did predict total mortality [66].

●The TWA substudy of the Sudden Cardiac Arrest in Heart Failure trial (SCD-HeFT) also showed that TWA failed to predict arrhythmic events in a subset of 490 patients from the parent trial [67].

Although the data from several studies are conflicting, the bulk of the evidence suggests at least a modest association between TWA and mortality from arrhythmic events [68]. However, studies of this association show considerable heterogeneity. One factor influencing this heterogeneity is that TWA was more predictive of events in studies that utilized clinical endpoints but was less predictive in studies that used ICD endpoints (which may overestimate the incidence of arrhythmic events) [62]. TWA testing has high negative predictive value for ventricular arrhythmias in patients with coronary disease; but from the ABCD study, TWA may have to be repeated on an annual basis to be used in this way [53]. Moreover, the negative predictive value, which is sufficient to withhold ICD implantation in an otherwise indicated patient, has yet to be resolved.

Predicting inducible VT/VF at electrophysiology study — Historical studies focused on the ability to induce sustained VT and, in certain circumstances, VF during EP study, particularly in patients with coronary artery disease [69]. As a result, initial studies of TWA used inducible VT and VF as surrogate endpoints for arrhythmic risk. These studies demonstrated that the presence of TWA during atrial or ventricular pacing or exercise correlated with the ability to induce ventricular arrhythmias in patients undergoing EP testing [2,11,13,14,16].

Abnormal TWA results have been associated with relatively high sensitivity and specificity (both between 75 and 85 percent) for the prediction of inducible VT or VF during invasive EP testing [2,13,16]. (See 'Accuracy of TWA testing' below.)

Long QT syndrome — Some of the earliest reports of macroscopic TWA (ie, visually apparent on the conventional ECG) were in patients with the long QT syndrome (LQTS), and macroscopic TWA is part of the LQTS score that is most often used for diagnosis (figure 2) [70]. We perform TWA testing in patients with known or suspected long QT syndrome as part of the evaluation for diagnostic and risk stratification purposes. (See "Congenital long QT syndrome: Diagnosis", section on 'Other ECG features'.)

Macroscopic TWA is, however, an uncommon finding in congenital LQTS, occurring in only 1.2 percent of patients in a large series [71]. Microvolt TWA is much more common, occurring in 44 percent of patients [72].

The mechanism of TWA in LQTS patients may differ from that in patients with CHD, and the following mechanisms have been proposed:

●LQTS-associated TWA may represent early after depolarizations during phase 2 or phase 3 of the action potential that occur in alternate beats due to alternation in the duration of the plateau (phase 2) associated with increased susceptibility to this type of electrical instability [8]. TWA of this nature has been shown to precede the onset of tachycardia-dependent torsade de pointes.

●In a canine model of LQTS, TWA was induced by abrupt increases in heart rate [73]. This was accompanied by significant increases in the dispersion of repolarization between epicardial and midmyocardial regions compared with baseline. A similar mechanism has been observed in patients with congenital LQTS following abrupt heart rate acceleration, in which it was shown that successive beats may encroach upon refractoriness resulting from repolarization dispersion to cause functional conduction block and reentry [74].

Further study is required to delineate the mechanisms underlying TWA in LQTS, the predictive value of TWA for arrhythmic events in each genotypic form, and whether the disappearance of TWA is a surrogate for effective therapy such as beta blockers. (See "Congenital long QT syndrome: Treatment", section on 'Beta blockers'.)

METHODS FOR PERFORMING TWA ANALYSIS — TWA was originally detected by visual inspection, allowing the detection of ECG fluctuations of 50 to 100 mV to be resolved; resolution beyond this is limited by biologic, electrode, and electronic noise [37,38]. Large TWA is rarely seen in patients at risk for sudden cardiac death (SCD), and TWA of a few microvolts may identify patients at risk for SCD [75].

Initial steps — Beta blockers should be held for 24 to 48 hours before testing. (See 'Limitations of analysis methods' below.)

ECG data are first collected from the 12 leads of a standard ECG, or orthogonal Frank XYZ vector leads [1,7]. Concentric Laplacian electrodes have been used commercially to reduce movement noise via spatial averaging and "far-field" subtraction [15,75,76]. ECG data are subsequently digitized for TWA [1,7].

Spectral methods — Spectral decomposition can detect low amplitude ECG fluctuations by exploiting the concept that TWA represents a periodic process occurring at a frequency of half the heart rate. Frequency decomposition represents consecutive ECG beats as a voltage time series. The Fourier transform generates a series of sine waves that describe this periodicity, in which alternans represent alternate-beat periodicity.

ECG sequences are first identified excluding ectopy [7]. Beats are then baseline corrected and registered to allow corresponding time points to be compared between beats (figure 1). Although several alignment strategies may be used, we have found that cross-correlation alignment is superior to other methods [7]. TWA is then computed using multidimensional spectral analysis (figure 1).

The T-wave alternans ratio (TWAR) represents the difference between alternans and non-alternating periodicity (spectral noise), defined within a bandwidth such as 0.33 to 0.48 cycles/beat (figure 5) [16]. TWAR greater than 0 indicates detectable TWA; TWAR ≥ 3 has been shown to predict ventricular arrhythmias [2]. The mean absolute voltage difference of alternation (V_alt), across the JT interval, may also be estimated (figure 1) [2,4,7].

Time-domain methods of analysis — Time-domain methods quantitate beat-to-beat ECG alternation without relying on a mathematical description of periodicity and may be used to assess time-varying TWA and have lower computational overhead. There are limited direct head-to-head comparisons with spectral methods [53,77].

Modified moving average (MMA) method — This approach creates a weighted rolling average of beats in alternate odd and even bins [4], whose subtraction yields TWA magnitude. This method, applied to routine ambulatory recordings, predicted clinical events [78]. MMA amplifies TWA magnitude at the cost of specificity [79]. Conversely, two different studies found that the MMA method may not add predictive value in post-myocardial infarction (MI) patients with mild to moderate systolic dysfunction and unselected patients undergoing stress testing [80,81].

Simple averaging method — In a multicenter study of 63 ICD recipients, alternans were significantly greater before VT/VF (62.9±3.1 microvolts) than during baseline rhythm (12.8±1.8 microvolts), rapid pacing (14.5±2.0 microvolts), before supraventricular tachycardia (27.5±6.1 microvolts), or during time-matched controls (12.3±3.5 microvolts) [82]. The odds of VT/VF increased by a factor of 2.2 for each 10-microvolt increment in TWA/V. This relatively simple method holds promise as a dynamic index of vulnerability to VT/VF.

Correlation method — The correlation method (CM) has primarily been applied to uncover TWA in patients with long QT syndrome [83]. The method first involves filtering and baseline correction of the ECG, followed by alignment of consecutive ECG beats to T waves. Corresponding time samples from successive T waves (Tj) and the median T wave (Tm), are cross-correlated to yield a dimensionless alternans correlation index (ACI) at each time sample (j) [78]. In the presence of TWA, ACI for consecutive time samples will alternate above and below unity.

The best method for determining TWA clinically is yet to be established and may theoretically vary with populations (eg, for spectral analysis, it may be best applied in patients in whom TWA is stable over time) [84].

Limitations of analysis methods — All methods of TWA analysis have limitations.

●The requirement to withhold beta blockers for 24 to 48 hours prior to testing may be difficult for patients who require beta blockers (eg, ischemic cardiomyopathy with angina, hypertrophic cardiomyopathy with symptomatic left ventricular outflow tract [LVOT] obstruction, certain long QT syndrome patients, etc).

●The comparative sensitivity and specificity of TWA methods require further study.

Measurement noise ultimately limits the sensitivity of any detection method for microvolt-level TWA, although the spectral method is relatively noise insensitive.

POSSIBLE OUTCOMES FROM TWA TESTING

Positive (abnormal) TWA results — TWA testing is positive (abnormal) if >1.9 microvolt amplitude for >1 minute in the absence of noise. Frequently, positive and indeterminate TWA tests (ie, non-negative) are grouped together. For modified moving average analysis (MMA), TWA in upper quartiles for that population are used [85].

Indeterminate TWA results — Approximately 20 to 40 percent of TWA tests are classified as "indeterminate" (ie, results do not demonstrate or exclude TWA) [12,13,55,86]. It has been recommended that patients be immediately retested if results are indeterminate.

Outcomes are similar in patients with abnormal TWA and those with indeterminate studies [46,49,87], if the indeterminacy is caused by patient-related rather than technical factors [86].

Negative (normal) TWA results — TWA testing is most useful for identifying patients at a low risk of ventricular arrhythmias via its negative results (see 'Further risk stratifying candidates for ICD' above). TWA using the spectral method is defined as below 1.9 microvolts for >1 minute in the absence of noise.

Accuracy of TWA testing — A 2006 meta-analysis included 2608 patients from 19 prospective studies of TWA (measured spectrally) in a variety of clinical settings: seven studies of nonischemic heart failure (HF), three studies of unspecified HF, two studies of ischemic HF, three studies of post-myocardial infarction (MI) patients, one study of implantable cardioverter-defibrillator recipients, one study of athletic heart patients, and two studies of healthy individuals [36]. The following findings were reported:

●The overall positive and negative predictive values of TWA for ventricular arrhythmic events were 19 and 97 percent.

●Positive predictive value was lower and negative predictive value was slightly higher in post-MI patients (6 and 99 percent, respectively). However, the patients in the post-MI studies had average left ventricular ejection fractions (LVEFs) between 45 and 56 percent.

●Predictive accuracy of TWA was not significantly different in ischemic and nonischemic HF.

Differential diagnosis of TWA — TWA does not involve the QRS complex and is distinct from electrical alternans (also referred to as electrical alternans totalis). This term refers to an alternating axis and/or voltage of all ECG components and is typically seen with the cardiac motion associated with large pericardial effusions, often in conjunction with sinus tachycardia [88]. QRS-T alternans may also occur with paroxysmal supraventricular tachycardias. (See "Pericardial effusion: Approach to diagnosis", section on 'ECG findings' and "ECG tutorial: Miscellaneous diagnoses", section on 'Electrical alternans'.)

Fluctuations in T-wave amplitude from respiratory fluctuations are typically not on an alternate beat basis. Non-alternating fluctuations in the T-wave may have prognostic value in some studies, but this is less well-demonstrated than for TWA [51].

SUMMARY AND RECOMMENDATIONS

●T-wave alternans (TWA) refers to periodic beat-to-beat variability in the timing, shape, and/or amplitude of T-waves on the surface electrocardiogram (ECG). TWA therefore reflects abrupt temporal heterogeneity in ventricular repolarization, which is an important mechanism underlying reentrant arrhythmias. (See 'Introduction' above.)

●Several distinct hypotheses have been proposed to explain TWA. (See 'Pathophysiology of TWA' above.)

●TWA reflects periodic changes in temporal heterogeneity or dispersion in ventricular repolarization, which is an important mechanism underlying reentrant arrhythmias. Thus, TWA is a noninvasive method for assessing risk for sudden cardiac arrest using the surface ECG. (See 'Evidence for TWA testing' above.)

●While historically detected by visual inspection of the surface ECG, TWA can now be measured using commercial systems that use frequency or time domain analyses. ECG data are first collected from the 12 standard leads of a surface ECG, or the ECG data may be obtained from orthogonal Frank XYZ vector leads. ECG data are subsequently digitized and analyzed for TWA, using either commercial TWA analysis systems or a variety of custom, user-written software packages. (See 'Methods for performing TWA analysis' above.)

●All methods for T-wave alternans detection have some limitations but have been validated empirically in different populations. Future advances may come from methods that analyze other aspects of repolarization variability including non-alternating periodicity and regional variations in T-wave oscillations. (See 'Limitations of analysis methods' above.)

●TWA testing may yield positive (abnormal), negative (normal), or indeterminate results. (See 'Possible outcomes from TWA testing' above.)

●Our approach to TWA testing in contemporary practice is presented in the body of the topic. (See 'Our approach to TWA testing' above.)