Anger-induced T-wave alternans predicts future ventricular arrhythmias in patients with implantable cardioverter-defibrillators

Abstract

Objective

To determine whether T-wave alternans (TWA) induced by anger in a laboratory setting predicts future ventricular arrhythmias (VT/VF) in patients with implantable cardioverter-defibrillators (ICDs).

Background

Anger can precipitate spontaneous VT/VF, and induce TWA. Whether anger-induced TWA predicts future arrhythmias is unknown.

Methods

Sixty-two patients with ICDs underwent ambulatory ECG during a mental stress protocol, three months post-implant. TWA was analyzed using time-domain methods. After ≥ 1 year follow-up, ICD stored data was reviewed to determine incidence of ICD-terminated VT/VF.

Results

Patients with ICD-terminated arrhythmias during follow-up (N=10) had higher TWA induced by anger, 13.2uV (iqr 9.3-16), compared to 9.3uV (7.5-11.5) (p<0.01). Patients in the highest quartile of anger-induced TWA (>11.9uV, N=15) were more likely to experience arrhythmias by one year than those in the lower quartiles, (33% versus 4%), and during extended follow-up (40% versus 9%, p<0.01 for both.) In multivariable regression controlling for ejection fraction, prior clinical arrhythmia, and wide QRS, anger-induced TWA remained a significant predictor of arrhythmia, with likelihood in the top quartile 10.8 times that of other patients (CI 1.6-113, p<0.05.)

Conclusion

Anger-induced TWA predicts future ventricular arrhythmias in patients with ICDs, suggesting that emotion-induced repolarization instability may be one mechanism linking stress and sudden death. Whether there is a clinical role for anger-induced TWA testing requires further study.

Evidence linking strong emotion and sudden cardiac death continues to mount. Sudden death increases during emotionally devastating disasters such as earthquake.(1) In patients with implantable cardioverter-defibrillators (ICDs), ventricular tachycardia or fibrillation (VT/VF) increased in the emotional weeks following the World Trade Center attacks in 2001.(2) Further, anger precipitates spontaneous VT/VF among patients with ICDs.(3) The physiologic pathways through which emotion can trigger arrhythmia, however, remain incompletely understood.

Recently, we(4) and others(5) demonstrated that anger, induced in a laboratory setting, can increase T-wave alternans (TWA). TWA, a marker of repolarization instability,(6) immediately precedes development of VF in animal models,(6,7) and in patients,(8) suggesting that TWA may be mechanistically related to arrhythmia. Further, TWA measured during exercise(9) or atrial pacing(10) predicts vulnerability to ventricular arrhythmias. To test the hypothesis that TWA induced by anger will similarly predict future VT/VF, we measured TWA during a laboratory mental stress protocol in patients shortly after ICD implantation, and prospectively evaluated the incidence of spontaneous sustained VT/VF requiring ICD-termination in follow-up.

Methods

Patient Population

Patients were recruited from the Yale Electrophysiology practice who had coronary artery disease (≥ 75% narrowing of ≥ 1 artery or documented infarction) or dilated cardiomyopathy and a standard indication for ICD implantation. Patients with atrial fibrillation, diagnosed psychiatric or cognitive disorders, inability to speak English, Class IV congestive heart failure, or medical comorbidities associated with likelihood of death within two years, were excluded. Seventy-one appropriate patients receiving ICDs between 12/00 and 9/04 agreed to participate and underwent the protocol. In 7 patients, noise precluded TWA analysis in ≥1 stage. Two patients failed to complete one year of follow-up (1 death due to unknown cause, 1 heart transplant). Data from the remaining 62 are included. Clinical variables were collected through chart review. The study was approved by the Yale Human Investigation Committee and all patients provided written informed consent.

Mental stress protocol

Patients underwent laboratory mental stress testing in the morning, before their first scheduled ICD follow-up, three months after implantation. Patients took all usual medications. Pacing was programmed VVI, 40 bpm, with subsequent sinus rhythm with native AV conduction in all patients. Our mental stress protocol is described in detail previously.(4) Briefly, conditions included: resting baseline, mental arithmetic, (serial subtraction of 7,) second baseline, and anger recall.

Repolarization Analysis

Methods for repolarization analysis are described in detail previously.(8) Ambulatory ECGs with modified V1 and V5 leads were recorded (bipolar pairs) on GE Medical (Milwaukee, WI) Marquette Series 8500 direct (amplitude-modulated) recorders (flat frequency response; linear phase between 0.67 and 50 Hz (±3dB); digitized at 400 Hz sampling frequency, effective resolution 2.5 μV.) A single lead (with largest magnitude T-wave) was analyzed for each subject throughout.

A previously validated program for adaptive baseline correction(8) was applied to assure accurate detection of the isoelectrical line. The QRS complexes were classified using custom software and verified by an experienced technician. After exclusion of ectopics, series of consecutive sinus beats were processed to identify fiducial points.(8)

The dynamics of TWA were examined using two independent time-domain techniques, intra-beat average and modified moving average analyses.(7,8) Modified moving average reports a maximum difference during repolarization between beats, whereas intra-beat average reports the average differences and hence, results in lower absolute values for TWA. While not identical, these algorithms have performed similarly in previous validation studies.(8) Because the correlation coefficient between MMA-derived and IBA-derived TWA was 0.72, and results of analyses were similar, only intra-beat analysis results are presented.

Follow-up

Each patient's ICD was interrogated ≥ every three to six months, transtelephonically and/or in clinic. Review of stored electrograms and event details identified occurrence of ICD-treated sustained VT/VF episodes.

Statistical analysis

Patients with ICD-treated, confirmed VT/VF during follow-up were compared to those without on the magnitude of TWA during baseline, anger, and arithmetic, by t-test. The relationship between TWA at each stage and incidence of VT/VF in follow-up was evaluated with logistic regression. Subsequent models included clinical variables. All models included adjustment for heart rate during TWA analysis period, and length of follow-up.

Because anger-induced TWA showed the strongest associations with future VT/VF, in further analyses, we compared patients in the top quartile of anger-induced TWA to all other patients on incidence of ICD-treated, confirmed VT/VF during follow-up, using chi-square, and by construction of Kaplan-Meier survival curves. This approach was used in previous studies of TWA(11), and in analogous studies of other stress-induced cardiac changes.(12) Logistic regression was also performed to adjust for clinical variables demonstrating associations with higher anger-induced TWA and/or future VT/VF.

All analyses were performed with JMP 5.0 software, SAS Institute, Cary, NC.

Results

Patient population

Patient characteristics are shown in Table 1. During a median follow up of 37 months, (iqr 25-47, range 12-75), 10 patients experienced ICD-terminated VT/VF.

Table 1. Demographic and Clinical Characteristics in Top Quartile of Anger-Induced TWA and in Lower Quartiles.

	Top Quartile Anger-Induced TWA N=15	Lower Quartiles N=47
Age, years	65 ± 3	62 ± 2
Male gender	12 (80%)	41 (87%)
Coronary artery disease*	14 (93%)	42 (89%)
Ejection fraction, %	29 ± 3	32 ±2
Ejection fraction ≤35%	12 (80%)	32 (68%)
History of prior clinical VT/VF†	6 (40%)	17 (36%)
Use of beta blocker‡	13 (87%)	41 (87%)
Use of sotalol	0	4 (9%)
Use of amiodarone	1 (7%)	6 (13%)
Use of any beta-blocking medication§	14 (93%)	45 (96%)
QRS duration, msec	145 ± 10	133 ± 6
Heart rate with anger, bpm	68 ± 2	69 ± 1
Heart rate change with anger, bpm	4 ± 1	7 ± 1║
ICD rate cut-off for therapy, bpm	172 ± 4	175 ± 2
Months follow-up, months	40 ± 4	37 ± 2

Values expressed as mean + SE or N (%).

^*defined in text;

^†sustained ventricular tachycardia or fibrillation;

^‡excluding sotalol or amiodarone;

^§Amiodarone, sotalol, and/or other beta-blockers;

^║p<0.05. All other p-values non-significant (>0.15.) TWA, T-wave alternans.

Relationship between anger-induced TWA and arrhythmias in follow-up

Overall, TWA during anger was greater than baseline (Table 2). Patients with ICD-terminated VT/VF during follow-up had higher TWA than those without at all stages, a difference greatest during anger, with a median anger-induced TWA of 13.2uV (iqr 9.3-16), compared to 9.3uV (7.5-11.5) (p<0.01) in those without future VT/VF. In regression analysis, TWA induced with anger predicted incidence of ventricular arrhythmias requiring ICD-termination, R² 0.13, p=0.01. Baseline TWA and arithmetic-induced TWA also predicted future VT/VF, but less strongly, (R² 0.081, p<0.05 and R² 0.07, p<0.05 respectively.)

Table 2. T-wave alternans during Laboratory Mental Stress.

	Baseline 1	Arithmetic	Baseline 2	Anger
All Subjects	9.7 (7.6-11.2)	9.8 (7.2-12.4)	9.5 (6.5-11.6)	10.4 (8.1-11.9)*
With VT/VF	12.9 (8.5-16.4)†	13 (11.7-14.8)‡	12.4 (7.9-14)‡	13.2 (9.3-16)†
No VT/VF	9.2 (7.4-11.0)	9.3 (7.1-11.3)	9.1 (6.1-11.3)	9.3 (7.5-11.5)

Data are presented as median (interquartile range). VT/VF, ventricular tachycardia or fibrillation in follow-up.

^*p<0.001 for comparison anger versus baseline 2 for all subjects (paired comparison);

^†p<0.01,

^‡p<0.05 for between-group comparison of those with vs without VT/VF in follow-up,

Patients in the top quartile of anger-induced TWA (TWA ≥ 11.9uV, N=15) were more likely to experience ICD-terminated VT/VF at one year than other patients (33% versus 4%), and during extended follow-up (40% versus 9%, p<0.01 for both comparisons, Figure 1) After controlling for length of follow-up, the likelihood of ICD-terminated VT/VF for those in the top quartile was 11 times that of the other patients (95% CI 2-98, p<0.01). Kaplan-Meier analysis is shown in figure 2. Sensitivity of anger-induced TWA was 60%, specificity 83%, positive predictive value 40%, and negative predictive value, 92%. In analogous analyses of arithmetic-induced TWA, effects were similar but of smaller magnitude and/or statistical significance (data not shown).

Figure 1. Event Rates.

Event rates in patients with anger-induced TWA in the top quartile (N=15, TWA ≥ 11.9 uV) compared to the lower quartiles (N=47, TWA<11.9uV).

Figure 2. Event-free Survival.

Kaplan-Meier survival curves depicting survival free from ICD-treated ventricular arrhythmias in patients with anger-induced TWA in the top quartile compared to the lower quartiles

Relationship to heart rate and clinical variables

Heart rate during anger did not differ between those in the top versus lower quartiles of anger-induced TWA, (Table 1) and did not predict future arrhythmia. Controlling for this factor did not influence the relationship between anger-induced TWA and future VT/VF.

There were no significant clinical differences between top-quartile patients and all others (Table 1). Also, there were no significant associations between these variables and the magnitude of anger-induced TWA, although ejection fraction (EF) ≤ 35%, QRS width ≥ 120msec, and history of prior clinical VT/VF showed non-significant associations. Among clinical variables, only prior clinical VT/VF was associated with ICD-terminated VT/VF, (non-significant trend with lower EF.)

In multivariable regression analysis controlling for EF, VT/VF history, and wide QRS, (and heart rate and length follow-up), anger-induced TWA remained a significant predictor of ICD-terminated VT/VF (Table 3), with the odds of ICD-terminated VT/VF in follow-up in the top-quartile group 10.8 times that of other patients (CI 1.6-113, p<0.05.)

Table 3. Predictive Value of Anger-induced TWA and Clinical Variables for Occurrence of Ventricular Tachycardia or Fibrillation.

	Odds ratio	p-value	95% CI
Anger-induced TWA, top-quartile	10.8	<0.05	1.6-113
Ejection fraction ≤ 35%	2.4	0.49	0.3-52
QRS ≥ 120 msec	0.11	0.02	0.02-1.3
History of prior clinical VT/VF	5.8	0.15	0.6-83

The model included adjustment for heart rate and length of follow-up. TWA, T-wave alternans.

Discussion

In this study, higher magnitudes of TWA during laboratory anger-recall predicted future VT/VF requiring ICD-termination in a population at risk for sudden death. This association was independent of clinical factors predisposing patients to higher levels of TWA, and/or higher risk of VT/VF. Emotion can precipitate ventricular arrhythmias,(2,3) and these findings suggest that emotion-induced increases in repolarization instability may link psychological stress to sudden death.

Strong emotion increases sympathetic arousal,(4) which influences TWA. For example, stellectomy abolishes, while stellate ganglion stimulation increases, TWA.(13) While heart-rate is one important determinant of TWA, TWA induced with exercise,(14) or with anger,(15) which increase heart rate due to autonomic effects, is greater than that with atrial pacing, implying non-heart-rate dependent effects of autonomic activity on repolarization. Intravenous beta-blockade(16) decreases TWA magnitude, in particular in the setting of an anger-like state (as shown experimentally in dogs),(15) further emphasizing the role of beta-adrenergic receptors. In daily life, TWA on ambulatory monitoring peaks at 8am,(11) when catecholamine levels are highest. Cardiovascular reactivity to stress in the laboratory reflects reactivity in daily life,(17) thus those patients with highest anger-induced TWA in the laboratory are most likely to experience similar emotionally-mediated TWA increases during daily life. As TWA precedes spontaneously-occurring ventricular arrhythmias,(8) these patients may have increased likelihood of spontaneous VT/VF.

In many studies, exercise-induced TWA predicts subsequent arrhythmias,(9) and TWA has been proposed as an index for risk-stratification for ICD implantation.(18) This is the first study demonstrating the prognostic significance of anger-induced TWA, showing a negative predictive value similar to that for exercise-induced TWA.(9) Analogously, laboratory-mental-stress-induced ischemia predicts subsequent clinical ischemia with predictive values similar to exercise.(19) Thus, mental stress could provide an alternative to atrial pacing for patients unable to exercise.

TWA induced by atrial pacing shows poorer predictive value than exercise.(10,20) Coumel(21) describes that arrhythmogenesis involves the interaction of myocardial substrate, triggering factors, and modulation by the autonomic nervous system. Measurements of TWA involving autonomic stimulation such as exercise or anger, which characterize the myocardial substrate in the presence of these modulating factors, may thus more accurately reflect arrhythmic risk. The ability to mimic the real-life combination of emotional trigger and autonomic modulation with substrate may also be why TWA induced by anger, a common emotion previously shown to trigger spontaneous VT/VF,(3) was more closely associated with future VT/VF than baseline or arithmetic-induced TWA.

Few prospective studies have examined the predictive value of TWA using time-domain methodology. While absolute values differ between time-domain and spectral analysis, results correlate closely.(8,22) TWA measured in the time-domain during exercise-testing predicts arrhythmic, cardiovascular and all-cause mortality.(23) Time-domain analysis, robust in the presence of signal nonstationarity, heart-rate changes, and noise, can quantify TWA from ambulatory recordings,(7,8)and thus may broaden the scope of TWA testing.(24)

Limitations

Most patients were taking beta-blocking medications, possibly blunting the magnitudes of anger-induced TWA.(15,16) However, performing the anger protocol in the presence of chronic medications may more accurately characterize the daily-life interaction of trigger, milieu, and substrate. Also, the sample size and number of events was small, and these data should be viewed as hypothesis-generating. Whether patients who agreed to participate may differ from the overall population, creating a selection bias, cannot be determined. Larger studies may more definitively establish the predictive value of anger-induced TWA. Further, the programming of ICDs was not uniform. While the rate-cutoffs did not differ between high- and low-anger-TWA groups, the possibility that differences in detection-times or programmed treatment may have contributed to the findings can not be excluded.

Conclusions

Anger-induced TWA predicts future VT/VF in patients with ICDs, suggesting that emotion-induced repolarization instability may be one mechanism linking stress and sudden death. Whether there is a clinical role for anger-induced TWA testing requires further study.

Abbreviations

EF

ejection fraction

ICD

implantable cardioverter defibrillator

TWA

T-wave alternans

VT/VF

ventricular tachycardia or fibrillation