Published in Volume
124, Issue 4
(April 1, 2014)J Clin Invest.
Copyright © 2014, American Society for Clinical
Sympathetic activity–associated periodic repolarization dynamics
predict mortality following myocardial infarction
1Medizinische Klinik III, Abteilung für Kardiologie und
Herz-Kreislauferkrankungen, Eberhard Karls University, Tübingen, Germany.
2Division of Cardiology, Helsinki University Central Hospital,
Helsinki, Finland. 31. Medizinische Klinik, Technische
Universität München, Munich, Germany. 4Department of
Clinical Physiology and 5Department of Biomedical Engineering, Tampere
University of Technology, Tampere, Finland. 6Department of Clinical
Chemistry, Fimlab Laboratories, School of Medicine at University of Tampere, Tampere,
Finland. 7Heart Centre, Department of Cardiology, Tampere University
Hospital, Tampere, Finland. 8Department of Thoracic, Cardiac and Vascular
Surgery, Eberhard Karls University, Tübingen, Germany.
Address correspondence to: Axel Bauer, Medizinische Klinik III, Abteilung
für Kardiologie und Herz-Kreislauferkrankungen,
Eberhard-Karls-Universität Tübingen, Otfried-Müller-Str. 10,
72076 Tübingen, Germany. Phone: 49.7071.29.82922; Fax: 49.7071.29.4550;
First published March 18, 2014
Submitted: October 23,
2013; Accepted: January 16,
Background. Enhanced sympathetic activity at the ventricular
myocardium can destabilize repolarization, increasing the risk of death. Sympathetic
activity is known to cluster in low-frequency bursts; therefore, we hypothesized that
sympathetic activity induces periodic low-frequency changes of repolarization. We
developed a technique to assess the sympathetic effect on repolarization and
identified periodic components in the low-frequency spectral range (≤0.1 Hz),
which we termed periodic repolarization dynamics (PRD).
Methods. We investigated the physiological properties of PRD in
multiple experimental studies, including a swine model of steady-state ventilation
(n = 7) and human studies involving fixed atrial pacing
(n = 10), passive head-up tilt testing (n = 11),
low-intensity exercise testing (n = 11), and beta blockade
(n = 10). We tested the prognostic power of PRD in 908 survivors
of acute myocardial infarction (MI). Finally, we tested the predictive values of PRD
and T-wave alternans (TWA) in 2,965 patients undergoing clinically indicated exercise
Results. PRD was not related to underlying respiratory activity
(P < 0.001) or heart-rate variability (P =
0.002). Furthermore, PRD was enhanced by activation of the sympathetic nervous
system, and pharmacological blockade of sympathetic nervous system activity
suppressed PRD (P ≤ 0.005 for both). Increased PRD was the
strongest single risk predictor of 5-year total mortality (hazard ratio 4.75, 95% CI
2.94–7.66; P < 0.001) after acute MI. In patients
undergoing exercise testing, the predictive value of PRD was strong and complementary
to that of TWA.
Conclusion. We have described and identified low-frequency rhythmic
modulations of repolarization that are associated with sympathetic activity.
Increased PRD can be used as a predictor of mortality in survivors of acute MI and
patients undergoing exercise testing.
Trial registration. ClinicalTrials.gov NCT00196274.
Funding. This study was funded by Angewandte Klinische Forschung,
University of Tübingen (252-1-0).
Sudden cardiac death (SCD) is the single most common cause of death in the
industrialized world (1). A substantial proportion
of SCD cases occur in patients after myocardial infarction (MI). Randomized trials have
demonstrated that in high-risk patients after MI, mortality can be effectively reduced
by prophylactic implantation of a cardioverter-defibrillator (ICD) (2). Consequently, identification of high-risk
individuals is a major objective in cardiology. Current guidelines recommend the
assessment of left ventricular ejection fraction (LVEF) as the gold standard risk
predictor (3, 4); however, this approach lacks both sensitivity and specificity (1, 5).
Therefore, development of novel risk markers is of great clinical interest.
Assessment of repolarization instability may more directly estimate the risk of fatal
cardiac arrhythmias (6). It is well known from
experimental and clinical studies that enhanced sympathetic activity is a key factor
leading to the destabilization of myocardial repolarization (7–14). However,
without directly recording neural activity, which is impractical in the clinical
setting, assessment of the sympathetic effect on myocardial repolarization has not been
possible to date. As sympathetic activity is organized in a series of low-frequency
bursts (15–19), we postulated that repolarization changes induced by the
sympathetic nervous system would exhibit low-frequency periodic features.
In the present study, we propose what we believe is a novel way to assess the
sympathetic effect on cardiac repolarization. We developed a technology and uncovered
periodic components of repolarization in the low-frequency spectral range (≤0.1
Hz), which we termed periodic repolarization dynamics (PRD). The first part of this
article focuses on the physiological properties of PRD, including activation and
blockade of the sympathetic nervous system. In the second part of this investigation, we
assess the prognostic meaning of enhanced PRD in patients surviving acute MI (post-MI
cohort; Figure 1A) and patients undergoing
clinically indicated exercise testing (stress-test cohort; Figure 1B). In the stress-test cohort we also tested the prognostic meaning
of exercise-induced T-wave alternans (TWA), which is presently considered to be the
strongest existing marker of repolarization instability.
CONSORT flow diagrams. Enrollment, follow-up, and analysis in the post-MI and stress-test cohorts.
Repolarization is subject to low-frequency periodic modulations. We developed a technique to dynamically track repolarization dynamics and to quantify
their periodic components. Details of the methodology are reported in Methods.
Briefly, we used standard, high-resolution, surface ECG recorded in or converted to
the orthogonal Frank lead configuration. As electrocardiographic repolarization is a
phenomenon occurring in both space and time, we integrated the spatiotemporal
information of each T-wave into a single vector, T°. We used
the angle dT° between successive repolarization vectors as
an estimate of the instantaneous repolarization instability (Figure 2, A–C). We observed characteristic
low-frequency oscillations in dT° in health and disease
(Figure 2D). In order to quantify these
low-frequency (≤0.1 Hz) periodic patterns, we employed wavelet analysis
Assessment of PRD. (A) Illustration of the weight-averaged vector of repolarization
(T°) for each T-wave from surface ECG recorded in the
Frank leads configuration. (B) Three-dimensional visualization of
successive T° vectors projected into virtual spheres. The
angle dT° between successive repolarization vectors was
used as an estimate of instantaneous repolarization instability. (C
and D) The dT° signal exhibits
characteristic low-frequency oscillations. C shows
dT° values for beats #219-223, corresponding to the
spheres in B. (E) Quantification of PRD using wavelet
analysis. PRD was defined as the average wavelet coefficient corresponding to
frequencies of 0.1 Hz or less.
PRD is not an epiphenomenon of underlying heart rate variability. We tested whether PRD was present in the absence of heart rate variability (HRV). We
studied 10 individuals (median age 52 [interquartile range (IQR) 32] years, 5
females), who underwent a clinically indicated electrophysiological (EP) study at our
institution. Patient characteristics are provided in Methods. We compared 5-minute
episodes of spontaneous sinus rhythm to 5-minute episodes during fixed atrial
stimulation, which was set above the spontaneous heart rate. Fixed atrial pacing
almost abolished HRV (P < 0.001; Supplemental Table 1;
supplemental material available online with this article; doi:
10.1172/JCI70085DS1), but exerted only minimal, nonsignificant effects
on PRD (ratio of PRD after provocation to PRD before provocation [PRD ratio] 0.75,
95% CI 0.50–1.17, P = 0.193; Figure 3A, Supplemental Figure 1A, and Supplemental Table 1).
Physiological and pharmacological provocations. Effects of fixed atrial stimulation (i), tilt-table testing (ii), exercise (iii),
and pharmacological beta blockade (iv) on PRD. PRD ratio was plotted on a
logarithmic axis and used to quantify the effect of each procedure.
PRD is not an epiphenomenon of underlying respiratory activity. To test whether PRD was present in the absence of spontaneous breathing, we performed
an experimental study in a swine model. Seven female domestic pigs were mechanically
ventilated and sedated with α-chloralose, which has been shown to induce only
minimal effects on the cardiac autonomic nervous system (20). Respiratory frequency and tidal volume were maintained
constant by means of volume-controlled ventilation. Details of the experimental
design are provided in Methods. PRD occurred independently of respiratory activity,
as illustrated in Figure 4A. There was no
interference between respiratory activity and PRD in any animal, as confirmed by
spectral and crossspectral analysis (Figure 4, B
and C; median coherence 0.044 [IQR 0.026]; P < 0.001 for the
difference from the threshold of 0.5).
Effect of respiration on PRD in a volume-controlled ventilated swine. (A) Signals of respiratory activity (green) and
dT° (blue). Respiratory activity was recorded by a
piezoelectric thoracic sensor. The dT° signal exhibits
typical low-frequency oscillations occurring independently from respiratory
activity. (B) Spectral analysis of respiratory activity and the
dT° signal. Power spectra were normalized by their
maximum value. (C) Cross-spectral analysis of respiratory activity
and the dT° signal showing a lack of interference between
PRD is enhanced by sympathetic activation and suppressed by sympathetic
blockade. We tested the effects of sympathetic activation on PRD in 11 healthy male volunteers
(median age 24 [IQR 3] years). Sympathetic activation was achieved by means of
head-up tilt testing and low-intensity exercise. Both tilt-table testing (PRD ratio
1.80, 95% CI 1.35–2.58, P = 0.005; Figure 3B, Supplemental Figure 1B, and Supplemental Table
1) and low-intensity exercise (PRD ratio 3.85, 95% CI 2.49–5.61,
P = 0.001; Figure 3C,
Supplemental Figure 1B, and Supplemental Table 1) led to substantial enhancement of
Conversely, we tested the effects of antiadrenergic intervention in 10 patients
(median age 57 [IQR 21] years, 7 females) undergoing an EP study at our institution.
Antiadrenergic intervention was achieved by pharmacological beta blockade. The
diagnostic protocol is described in Methods. Beta blockade caused a striking
suppression of PRD in all patients (PRD ratio 0.41, 95% CI 0.28–0.61,
P = 0.002; Figure 3D,
Supplemental Figure 1C, and Supplemental Table 1).
For comparison, the effects of sympathetic activation and blockade on the
low-frequency component of heart-rate variability are shown in Supplemental Table
Increased PRD predicts total and cardiovascular mortality after MI. We tested the prognostic significance of PRD in a cohort of 908 patients from the
Autonomic Regulation Trial (median age 61 [IQR 17] years, 174 females) who survived
an acute MI (Figure 1A and Table 1) (21,
22). Sixty-nine patients died within the
first 5 years of follow-up. Representative resting dT°
signals in a patient who survived the follow-up period and in a patient who suddenly
died 8 months after index MI are depicted in Figure 5, A and B, respectively. Although low-frequency oscillations in
dT° were evident in both patients, the amplitudes of PRD
were much higher in the nonsurviving patient. The level of PRD was significantly
associated with 5-year mortality (6.67 [IQR 8.58] deg2 vs. 2.66 [IQR 3.93]
deg2; P < 0.001). For subsequent survival analyses,
we dichotomized PRD at the upper quartile of the study population. The 227 patients
with PRD greater than or equal to 5.75 deg2 (Figure 5C) had a 5-year risk of death of 18.2% compared with 4.1% in the
681 patients with PRD of less than 5.75 deg2 (P <
0.001). Both uni- and multivariable analyses for the prediction of 5-year total
mortality indicated that PRD greater than or equal to 5.75 deg2 was the
strongest single risk predictor in the study cohort (Table 2 and Supplemental Figure 2). The predictive value of PRD greater
than or equal to 5.75 deg2 was independent of that of established risk
markers, including reduced LVEF of 35% or less (3, 4), the Global Registry of Acute
Coronary Events (GRACE) score (23), the
presence of diabetes mellitus, elevated mean heart rate, reduced HRV, and increased
QT variability index (QTVI) (24).
Subsequently, we assessed the incremental prognostic value of PRD to established
risk-prediction models (Supplemental Table 2). PRD significantly improved all tested
risk-prediction models based on the combination of LVEF, GRACE score, respiratory
rate, and HRV parameters. Subgroup analyses revealed that increased PRD was a
particularly strong predictor of mortality in the 179 post-MI patients who also
suffered from diabetes mellitus, identifying a group of 179 patients with a
cumulative 5-year mortality rate of 38.8% (Figure 5D).
PRD in post-MI patients. (A) Typical dT° signal (blue line) obtained
from a 50-year-old post-MI patient who survived the 5-year follow-up period. The
signal shows characteristic low-frequency oscillations. For better illustration of
these oscillations, a low-pass filter was applied and plotted on top of the
original signal (black line). (B) Typical
dT° signal (red line) from a 75-year-old post-MI
patient who suddenly died 8 months after MI. Compared with the survivor, the
amplitude of PRD was substantially enhanced. (C) Cumulative mortality
rates of patients stratified by PRD of 5.75 deg2 or more.
(D) Cumulative mortality rates of patients stratified by PRD of
5.75 deg2 or more and presence of diabetes mellitus.
Patient characteristics and treatment in the post-MI and stress-test cohorts
Univariable and multivariable association of risk markers with 5-year all-cause
and cardiovascular mortality in 908 survivors of acute MI (post-MI cohort)
Finally, we also investigated whether PRD predicted cardiovascular mortality. Of the
69 deaths, 36 were cardiovascular deaths. As shown in Table 2, PRD greater than or equal to 5.75 deg2 was also a
strong and independent predictor of 5-year cardiovascular mortality.
The predictive value of PRD is complementary to that of exercise-induced
TWA. To test the predictive values of PRD and TWA, we studied 2,965 patients (median age
57 [IQR 16] years, 1,187 females) from the Finnish Cardiovascular Study (median age
57 [IQR 16], 1,187 females; Figure 1B and Table
1) who underwent a clinically indicated
exercise test (25). During a median follow-up
of 6 years, 309 patients died. In all patients, TWA was measured during exercise by
the modified moving average (MMA) method (25–27). PRD was assessed in
the preexercise period with patients sitting on a bicycle ergometer. As expected, PRD
levels in this cohort were higher than those in the post-MI cohort, where PRD was
estimated in the supine resting position (9.2 [IQR 13.37] deg2 vs. 2.82
[IQR 4.34] deg2, respectively; P < 0.001). Both PRD
and TWA were significantly associated with mortality (8.96 [IQR 13.23]
deg2 vs. 12.04 [IQR 16.32] deg2, P <
0.001, and 23.00 [IQR 15] μV vs. 26.00 [IQR 17] μV,
P < 0.001, respectively). Univariable Cox regression analysis
showed that both markers were strong predictors of total mortality (standardized
coefficients 0.203, 95% CI 0.113–0.293, P < 0.001 for
PRD; and 0.256, 95% CI 0.164–0.348, P < 0.001 for TWA).
This remained true on multivariable analysis, which also included age, sex, previous
MI, presence of diabetes mellitus, and treatment with beta-blockers (Table 3). The significant crossterm between TWA and PRD
(TWA × PRD) indicated that the relationship between the outcome and one
predictor was dependent on the levels of the other predictor. We therefore tested the
additive prognostic value of PRD for different levels of TWA. As illustrated in
Supplemental Figure 3, PRD provided incremental prognostic information at all levels
Multivariable association of TWA and PRD with all-cause and cardiovascular
mortality in 2,965 patients undergoing a clinically indicated exercise test
Of the 309 deaths, 138 were cardiovascular deaths. Increased preexercise PRD was also
a significant predictor of cardiovascular mortality in univariable (standardized
coefficient 0.256, 95% CI 0.126–0.385; P < 0.001) and
multivariable analysis (Table 3).
In the present study, we identified periodic oscillations of repolarization that were
localized in the low-frequency spectral range and were detectable by conventional
surface ECG. PRD was evident in health and disease without provocations and occurred
autonomously from underlying HRV and respiratory activity. PRD was augmented by
physiological provocations leading to activation of the sympathetic nervous system and
was suppressed by pharmacological adrenergic blockade. Increased PRD obtained under
resting conditions was a very strong predictor of total and cardiovascular mortality in
survivors of acute MI and patients undergoing a clinically indicated exercise test. The
prognostic value of PRD was incremental to that of established risk markers, including
LVEF and TWA.
Noninvasive assessment of the sympathetic effect on myocardial repolarization is of
great clinical interest. A wealth of evidence supports the widely held belief that
increased sympathetic nervous system activity is associated with increased cardiac
vulnerability (11–14). In human subjects, noninvasively measured parameters, including
HRV and baroreflex sensitivity, have been employed to study sympathetic activity under
routine clinical conditions (28). This approach
is based on the principle that activation of the sympathetic nervous system evokes
several physiological effects, including increasing systolic contractility rate and
vasomotor tone as well as accelerating heart rate and atrioventricular conduction (10). However, these measurements provide only an
indirect probe of the sympathetic effect on repolarization; they reflect influences on
the sinoatrial node and blood vessels, not on the ventricular myocardium.
At the level of cardiomyocytes, stimulation of β-adrenergic receptors alters
intracellular calcium dynamics (28) and shortens
action potential duration (10). Importantly, the
effect of sympathetic stimulation on the 3 cell types of the ventricular myocardium
(epicardial cells, M cells, and endocardial cells) is nonuniform (29, 30). Adrenergic
activation abbreviates the action potential duration of epicardial and endocardial cells
to a greater degree than the action potential duration of M cells (31), leading to an increased transmural dispersion of repolarization
It is well known that sympathetic activity is organized in series of low-frequency
bursts (15–19). We therefore assumed that phasic sympathetic activation induces
phasic changes in repolarization localized in the low-frequency spectral range, and we
developed a method to track the sympathetic effect on transmural dispersion of
repolarization. Our findings confirmed this hypothesis. For what we believe is the first
time, we detected periodic changes in repolarization in the same range as those of the
sympathetic nervous system. PRD was significantly enhanced by sympathetic activation and
was substantially suppressed by sympathetic blockade. Moreover, increased PRD was a
strong predictor of total and cardiovascular mortality, which is in line with the
results of many studies showing that enhanced sympathetic activity is associated with an
increased risk of death (11–14).
In particular, increased PRD was identified as the strongest single risk predictor of
total and cardiovascular mortality in a large cohort of post-MI patients. The predictive
value of PRD was independent of established risk markers. PRD substantially improved
several multivariable models in prediction of total mortality, confirming its
incremental prognostic value. The mechanism by which PRD identifies high-risk patients
significantly differs from that of structural markers such as LVEF. Directly estimating
sympathetic activity at the level of myocardial repolarization may provide more accurate
information on cardiac risk. Post-MI patients with increased PRD had a very poor
prognosis when they also suffered from diabetes mellitus. Both MI (32) and diabetes mellitus (33) are characterized by spatially heterogeneous sympathetic innervation, which
is associated with negative prognosis (10).
We tested the prognostic significance of PRD in a large cohort of patients undergoing
clinically indicated exercise testing. Increased PRD was a strong predictor of total and
cardiovascular mortality and provided incremental prognostic information to that of
exercise-induced TWA. This indicates that PRD can be used to detect high-risk patients
who are not identified by TWA. The complementary prognostic information provided by PRD
and TWA implies that these 2 markers capture different aspects of repolarization
instability. While PRD most probably reflects low-frequency oscillations related to
sympathetic activity, TWA is mainly caused by high-frequency action potential
oscillations provoked by abnormal calcium handling (34). TWA is an important predictor of cardiovascular mortality, including
sudden death (6, 27, 35, 36). However, it needs to be provoked by exercise (37) or invasive procedures (38). Assessment of PRD is inexpensive, easily obtainable under
resting conditions, and noninvasive and significantly improves available
Our study has several limitations. First, high-resolution ECG is required in order to
measure PRD. It remains to be demonstrated whether our results are reproducible with
lower-resolution tracings. Second, in both cohorts, risk markers were only assessed at
enrollment. Therefore, we cannot comment on the immediate and long-term reproducibility
of PRD as well as the effect of treatment on PRD. Third, the prognostic value of PRD
needs to be validated in independent cohorts. Fourth, as PRD is dependent on the
patient’s body position and activity level, the proposed cutoff value is only
valid for recordings obtained in the supine resting position. Fifth, we confirmed the
prognostic value of PRD for prediction of total mortality and cardiovascular mortality.
Although it is plausible to assume that increased levels of PRD are also associated with
arrhythmic mortality, this needs to be tested in future studies. Finally, although we
have shown that increased PRD is a powerful risk predictor, we have no data to show that
specific treatments based on the use of this predictor will improve patient outcome.
In conclusion, PRD constitutes an electrocardiographic phenomenon that most likely
reflects the myocardial response to sympathetic activation. Increased PRD is a potent
risk predictor of total and cardiovascular mortality, and its use significantly improves
established risk-stratification concepts. Future studies are needed to test whether
high-risk patients identified by PRD benefit from prophylactic therapies.
Participants. The physiological properties of PRD were studied in 3 cohorts at the University
Hospital of Tübingen. We tested the effects of fixed atrial pacing
(atrial-pacing cohort) in 10 individuals (median age 52 [IQR 42] years, 5 females)
undergoing a clinically indicated diagnostic EP study. Indications for EP studies
were paroxysmal supraventricular tachycardia in 7 patients and evaluation of
unexplained syncope in 3 patients. We investigated the effect of beta blockade in 10
patients (median age 57 [IQR 21] years, 7 females) undergoing an EP study for
paroxysmal supraventricular tachycardia (adrenergic-blockade cohort). In both EP
studies, all patients were in sinus rhythm, had normal LVEF, were not suspected of
suffering from coronary artery disease, and had no significant valve stenosis or
insufficiency on echocardiography. We also studied the effects of passive head-up
tilt and low-intensity exercise in 11 healthy male volunteers (median age 24 [IQR 3]
years, adrenergic-activation cohort).
The prognostic power of PRD was tested in 908 survivors (median age 61 [IQR 17]
years, 174 females) of acute MI (post-MI cohort; Figure 1A and Table 1) and 2,965 patients
(median age 57 [IQR 16] years, 1,187 females) undergoing clinically indicated
exercise testing (stress-test cohort; Figure 1B
and Table 1). Patients in the post-MI cohort
were enrolled between May 2000 and March 2005 at 2 university centers in Munich,
Germany: the German Heart Centre and the Klinikum Rechts der Isar (21, 22).
Eligible patients had survived acute MI (<4 weeks), were aged 80 years or more,
had sinus rhythm, and did not meet the criteria for secondary prophylactic
implantation of ICD before hospital discharge. Patients in the stress-test cohort
were included between October 2001 and December 2008 at the Tampere University
Hospital (Finnish Cardiovascular Study) (25).
Eligible patients were aged 30–80 years, were in sinus rhythm, and underwent
a clinically indicated exercise test.
Procedures. We performed EP studies according to the hospital’s standard operating
procedures. No study-specific invasive procedures were performed on any patient. We
did not sedate the patients. All EP studies required placement of a pacing electrode
in the right atrium. Atrial pacing was performed at a fixed cycle length (CL), below
sinus rhythm CL and slightly above Wenckebach CL. In the atrial-pacing cohort, we
compared a 5-minute recording during undisturbed sinus rhythm to a 5-minute recording
during fixed atrial stimulation. In the adrenergic-blockade cohort, we compared
5-minute tracings before and after i.v. administration of 0.1 mg/kg metoprolol. Fixed
atrial pacing during the entire procedure was used to ensure constant heart rate.
Subjects in the adrenergic-activation cohort were not allowed to eat or drink coffee
for 12 hours before the tests. Vigorous exercise and alcohol were forbidden for 48
hours before the tests. All healthy volunteers lay in a supine position in a quiet
room for at least 15 minutes before data collection. We used 2 provocations: 2-minute
passive head-up tilt-test at 45° and 5-minute low-intensity exercise using a
bicycle ergometer. For the latter test, the individual workload was set to achieve a
constant heart rate of 110 bpm. For all physiological studies, we used
high-resolution (2,048 Hz) digital ECG (TMS; Porti System) recorded in Frank leads
configuration throughout the entire procedure.
For the post-MI cohort, we used 30-minute high-resolution (1,600 Hz) digital ECG
(TMS; Porti System) recorded in Frank leads configuration. Recordings were performed
within the second week after MI in resting conditions in the morning hours and in a
supine position. We additionally performed a 24-hour Holter recording (Oxford Excel
Holter System, Oxford Instruments; Pathfinder700, Reynolds Medical; and Mortara
Holter System, Mortara Instrument) within the second week after MI. For the
stress-test cohort, an upright bicycle was used for the exercise test. The workload
was increased from 20 to 30 W in a step-wise manner (10 to 30 W/min). For calculation
of PRD, a preexercise period of at least 2.5 minutes was recorded using 12-channel
digital ECG (500 Hz), which was converted into the Frank leads configuration by means
of the inverse Dower matrix (39).
Assessment of PRD. The technique used to calculate PRD is illustrated in Figure 2 and in Supplemental Figures 4–6. The prerequisite for
computing dT° is an ECG tracing recorded in or converted to
the 3 orthogonal axes X, Y, and Z. The time positions of the T-waves were identified
using previously published algorithms (40,
41). The end of each T-wave was set as the
reference point (amplitude = 0 mV). The first step in calculating the new parameter
was to transform the Cartesian coordinates X, Y, and Z (Supplemental Figure 4A) into
a time series of polar coordinates defined by 2 angles (elevation and azimuth) and
the resultant-force amplitude XYZ (Supplemental Figure 4B). For example, we selected
a time point t1 (Supplemental Figure 4B) and decomposed the XYZ
vector into 2 orthogonal vectors on the y axis and the transverse
(XZ) plane. The angle between the vector and the y axis was termed
the elevation (42) (Supplemental Figure 4D),
with an angle of 0° defined as the vector pointing to the caudal direction.
The angle between the vector on the transverse plane and the x axis
was termed the azimuth (42).
On the basis of the 3 new time signals of the polar coordinates, we defined the
weight-averaged direction of repolarization, which can be described by a set of 2
polar coordinates that we called the weight-averaged azimuth (WAA) and the
weight-averaged elevation (WAE). WAA and WAE can be calculated using Equations 1 and
2, respectively. For each time point t, the resultant-force
amplitude XYZ, represented as Amp(t) in Equations 1 and 2, was
multiplied by the corresponding values for azimuth and elevation at the same time
point. The (Amp(t) × Angle(t)) products
were initially summed for the entire duration of the T-wave and thereafter divided by
the sum of all resultant-force amplitudes. The result of each equation represents the
“with-the-amplitude weighted” average angle, which is measured by
means of the same units (deg rad) as the angle in Equations 1 and 2.
Using the repolarization wave of Supplemental Figure 4 as a backbone, exemplary WAA
and WAE values were calculated as illustrated in Supplemental Figure 5.
We used the angle dT° between successive repolarization
vectors as an estimate of the instantaneous degree of repolarization instability
(Figure 2, A–C). The angle
dT° was calculated using the dot product (scalar product)
equation (43), which by 2 vectors of the same
length r can be simplified to Equation 3 as illustrated in
Supplemental Figure 6. The dT° signal was linearly
interpolated with a sampling rate of 2 Hz and filtered using a low-pass filter to
remove artifacts. In order to quantify the periodic components of
dT°, we employed continuous wavelet transformation
(Figure 2D). The continuous wavelet
transformation provides wavelet coefficients for each scale at each time point. For
each scale, the average wavelet coefficient was computed. Finally, scales were
converted to pseudofrequencies using an established algorithm (44). PRD was defined as the average wavelet coefficient in the
frequency range of 0.1 Hz or less (Figure 2E).
Assessment of TWA. We assessed TWA by the time-domain MMA method according to previously established
technologies (26) (GE Healthcare version 5.2).
In brief, the MMA algorithm separates odd from even beats. The average morphologies
of both the odd and even beats were calculated separately and continuously updated by
a weighting factor of 1/8 or 1/32 of the difference between the ongoing average and
the new incoming beats. The update was calculated for every incoming beat, resulting
in continual moving averages of the odd and even beats. This approach makes the MMA
suitable for TWA analysis during the period of activity or during periods of
fluctuating heart rates (non–steady state periods) (45). In addition, algorithms were incorporated to reduce the
influence of noise and artifacts, such as those caused by pedaling and respiration
(46). The TWA values were calculated
continuously during the entire exercise test, from rest to recovery, using all
precordial leads. Finally, the maximum TWA value at heart rates of less than 125 bpm
Assessment of other risk predictors. We assessed LVEF by echocardiography or angiography. We obtained short-term and
24-hour HRV in time and frequency domains as previously proposed (47). Since the standard deviation of all
normal-to-normal intervals (SDNN) provided the strongest prognostic power of all HRV
measures, we used SDNN as a marker of HRV. We assessed QTVI from the resting ECGs
according to previously published technological methods (24). We calculated the GRACE score, which combines several
clinical risk factors, specifically patient age, history of previous MI and
congestive heart failure, ST-segment deviation, elevated cardiac enzymes, renal
impairment, systolic blood pressure and heart rate upon admission, and percutaneous
coronary interventions during the hospital stay (23).
Animal study. Seven female domestic pigs (60–78 kg) were preanesthetized with propofol (2
mg/kg i.v.) and anesthetized with α-chloralose (150 mg/kg i.v. with
supplemental doses of 600 mg in 60 ml saline as required), which has been shown to
induce only minimal effects on the cardiac autonomic nervous system (20). Immediately after induction of anesthesia,
the trachea was cannulated and the lungs were mechanically ventilated with room air.
Constant respiratory frequency and tidal volume were maintained by means of
volume-controlled ventilation with a fixed tidal volume (6 ml/kg) and a fixed
respiratory rate. In each animal, the respiratory rate was set individually
(respiratory rate 12–18 per minute corresponding to a frequency of
0.20–0.30 Hz) to maintain normal end-tidal CO2. Respiratory
activity was recorded by a piezoelectric thoracic sensor (48). Two hours after the administration of Propofol, a
high-resolution 30-minute ECG (2,048 Hz) was recorded in the Frank leads
Statistics. We present continuous variables as medians with IQRs. Categorical data are presented
as proportions. Results are presented as mean values with 95% CI. Results of
physiological and EP studies are presented as the ratio of the value after
provocation to the corresponding value before provocation with 95% CI. Differences in
the logarithmic ratios were assessed by means of a paired Wilcoxon signed-rank test.
To evaluate the effects of respiratory activity on PRD, we estimated the square
coherence function between respiratory and dT° signals using
the cross-spectral method (49). The coherence
function (range 0–1) expresses the linear coupling between 2 signals in the
frequency domain (50). A square coherence
function greater than 0.5 was considered significant (19, 51, 52). The end points of both prognostic studies were all-cause and
cardiovascular mortality. We used the standardized definition of cardiovascular death
(53, 54). The median follow-up time in the post-MI cohort was 5 years. PRD was
dichotomized at the upper quartile of the study population. For dichotomization of
other risk markers, we used established cutoff values of 35% or less for LVEF (55), –0.47 or more for QTVI (24), greater than 75 for mean heart rate (56), 70 ms or less for SDNN (56), and 120 or more for the GRACE score (23). We estimated survival curves using the
Kaplan-Meier method. Multivariable analyses were implemented by the adaptation of Cox
and multinomial logistic regression models. The latter method was used to calculate
integrated discrimination improvement (IDI) scores (57). The median follow-up time in the stress-test cohort was 6 years. The
prognostic powers of PRD and TWA were tested with univariable and multivariable Cox
regression analysis,including age, sex, previous MI, the presence of diabetes
mellitus, treatment with beta blockers, and the crossterm between TWA and PRD (TWA
× PRD). PRD, TWA, and age were normalized by subtraction of their mean value
and division by their SD and were included as scalar factors in the multivariable
model. Differences were considered statistically significant when the 2-sided
P value was less than 0.05. All statistical analyses were
performed using CRAN R, version 2.15.1.
Study approval. The ethics committees of Tübingen, Tampere, and Munich approved the studies
performed in the physiological, stress-test, and post-MI cohorts, respectively.
Written informed consent was obtained from each participant. The animal protocol was
in accordance with the German guidelines for use of living animals and was approved
by the local governmental commission for animal research (K 5/10,
Regierungspraesidium Tübingen, Baden-Wuerttemberg, Germany).
K. Rizas was awarded for this work with the ESC 2013 Young Investigator Award in
Clinical Science. The study was supported in part by grants from the program
“Angewandte klinische Forschung” (AKF) of the University of
Tübingen (252-1-0 to A. Bauer). No additional external funding was received for
this study. The Autonomic Regulation Trial (post-MI cohort) was supported by
Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie
(13N/7073/7), the Kommission für Klinische Forschung, and the Deutsche
Forschungsgemeinschaft (SFB 386). The Finnish Cardiovascular Study (stress-test cohort)
was supported by the Medical Research Fund of Tampere University Hospital (grants 9MO48
and 9N035), the Finnish Cultural Foundation, the Finnish Foundation for Cardiovascular
Research, the Emil Aaltonen Foundation, and the Tampere Tuberculosis Foundation. The
authors thank the staff of the Department of Clinical Physiology for collecting the
exercise test data.
Huikuri HV, Castellanos A, Myerburg RJ. Sudden death due to cardiac arrhythmias. N Engl J Med. 2001;345(20):1473–1482.
Moss AJ, et al. Prophylactic implantation of a defibrillator in patients with
myocardial infarction and reduced ejection fraction. N Engl J Med. 2002;346(12):877–883.
Hunt SA, et al. 2009 focused update incorporated into the ACC/AHA 2005 Guidelines for
the Diagnosis and Management of Heart Failure in Adults: a report of the American
College of Cardiology Foundation/American Heart Association Task Force on Practice
Guidelines: developed in collaboration with the International Society for Heart
and Lung Transplantation. Circulation. 2009;119(14):e391–e479.
Task Force for Diagnosis and Treatment of Acute and Chronic Heart Failure
2008 of European Society of Cardiology, et al. ESC Guidelines for the diagnosis and treatment of acute and chronic
heart failure 2008: the Task Force for the Diagnosis and Treatment of Acute and
Chronic Heart Failure 2008 of the European Society of Cardiology. Developed in
collaboration with the Heart Failure Association of the ESC (HFA) and endorsed by
the European Society of Intensive Care Medicine (ESICM). Eur Heart J. 2008;29(19):2388–2442.
Buxton AE. Risk stratification for sudden death: do we need anything more than
ejection fraction? Card Electrophysiol Rev. 2003;7(4):434–437.
Rosenbaum DS, et al. Electrical alternans and vulnerability to ventricular
arrhythmias. N Engl J Med. 1994;330(4):235–241.
Lown B, Verrier RL. Neural activity and ventricular fibrillation. N Engl J Med. 1976;294(21):1165–1170.
Verrier RL, Antzelevitch C. Autonomic aspects of arrhythmogenesis: the enduring and the
new. Curr Opin Cardiol. 2004;19(1):2–11.
Cao JM, et al. Relationship between regional cardiac hyperinnervation and ventricular
arrhythmia. Circulation. 2000;101(16):1960–1969.
Rubart M, Zipes DP. Mechanisms of sudden cardiac death. J Clin Invest. 2005;115(9):2305–2315.
Han J, Garcia de Jalon P, Moe GK. Adrenergic effects on ventricular vulnerability. Circ Res. 1964;14:516–524.
Maling HM, Moran NC. Ventricular arrhythmias induced by sympathomimetic amines in
unanesthetized dogs following coronary artery occlusion. Circ Res. 1957;5(4):409–413.
Kliks BR, Burgess MJ, Abildskov JA. Influence of sympathetic tone on ventricular fibrillation threshold
during experimental coronary occlusion. Am J Cardiol. 1975;36(1):45–49.
Butrous GS, Gough WB, Restivo M, Yang H, el-Sherif N. Adrenergic effects on reentrant ventricular rhythms in subacute
myocardial infarction. Circulation. 1992;86(1):247–254.
Pagani M, et al. Relationship between spectral components of cardiovascular
variabilities and direct measures of muscle sympathetic nerve activity in
humans. Circulation. 1997;95(6):1441–1448.
Malliani A, Pagani M, Lombardi F, Cerutti S. Cardiovascular neural regulation explored in the frequency
domain. Circulation. 1991;84(2):482–492.
Pagani M, et al. Power spectral analysis of heart rate and arterial pressure
variabilities as a marker of sympatho-vagal interaction in man and conscious
dog. Circ Res. 1986;59(2):178–193.
Montano N, et al. Spectral analysis of sympathetic discharge, R-R interval and systolic
arterial pressure in decerebrate cats. J Auton Nerv Syst. 1992;40(1):21–31.
Furlan R, et al. Oscillatory patterns in sympathetic neural discharge and
cardiovascular variables during orthostatic stimulus. Circulation. 2000;101(8):886–892.
Nearing BD, Verrier RL. Tracking cardiac electrical instability by computing interlead
heterogeneity of T-wave morphology. J Appl Physiol. 2003;95(6):2265–2272.
Barthel P, et al. Spontaneous baroreflex sensitivity: Prospective validation trial of a
novel technique in survivors of acute myocardial infarction [published online
ahead of print: April 16, 2012]. Heart Rhythm. doi:
Barthel P, et al. Respiratory rate predicts outcome after acute myocardial infarction: a
prospective cohort study [published online ahead of print: December 14,
2012]. Eur Heart J. doi: 10.1093/eurheartj/ehs420 .
Granger CB, et al. Predictors of hospital mortality in the global registry of acute
coronary events. Arch Intern Med. 2003;163(19):2345–2353.
Piccirillo G, et al. QT variability strongly predicts sudden cardiac death in asymptomatic
subjects with mild or moderate left ventricular systolic dysfunction: a
prospective study. Eur Heart J. 2007;28(11):1344–1350.
Nieminen T, et al. T-wave alternans predicts mortality in a population undergoing a
clinically indicated exercise test. Eur Heart J. 2007;28(19):2332–2337.
Nearing BD, Verrier RL. Modified moving average analysis of T-wave alternans to predict
ventricular fibrillation with high accuracy. J Appl Physiol. 2002;92(2):541–549.
Slawnych MP, et al. Post-exercise assessment of cardiac repolarization alternans in
patients with coronary artery disease using the modified moving average
method. J Am Coll Cardiol. 2009;53(13):1130–1137.
Verrier RL, Kumar K, Nearing BD. Basis for sudden cardiac death prediction by T-wave alternans from an
integrative physiology perspective. Heart Rhythm. 2009;6(3):416–422.
Tanabe Y, et al. Sympathetic stimulation produces a greater increase in both transmural
and spatial dispersion of repolarization in LQT1 than LQT2 forms of congenital
long QT syndrome. J Am Coll Cardiol. 2001;37(3):911–919.
Antzelevitch C. Transmural dispersion of repolarization and the T
wave. Cardiovasc Res. 2001;50(3):426–431.
Shimizu W, Antzelevitch C. Cellular basis for the ECG features of the LQT1 form of the long-QT
syndrome: effects of beta-adrenergic agonists and antagonists and sodium channel
blockers on transmural dispersion of repolarization and torsade de
pointes. Circulation. 1998;98(21):2314–2322.
Vaseghi M, Shivkumar K. The role of the autonomic nervous system in sudden cardiac
death. Prog Cardiovasc Dis. 2008;50(6):404–419.
Wei K, Dorian P, Newman D, Langer A. Association between QT dispersion and autonomic dysfunction in
patients with diabetes mellitus. J Am Coll Cardiol. 1995;26(4):859–863.
Narayan SM, Bayer JD, Lalani G, Trayanova NA. Action potential dynamics explain arrhythmic vulnerability in human
heart failure: a clinical and modeling study implicating abnormal calcium
handling. J Am Coll Cardiol. 2008;52(22):1782–1792.
Nearing BD, Huang AH, Verrier RL. Dynamic tracking of cardiac vulnerability by complex demodulation of
the T wave. Science. 1991;252(5004):437–440.
Nieminen T, et al. The Finnish Cardiovascular Study (FINCAVAS): characterising patients
with high risk of cardiovascular morbidity and mortality. BMC Cardiovasc Disord. 2006;6:9.
Verrier RL, et al. Microvolt T-wave alternans physiological basis, methods of
measurement, and clinical utility--consensus guideline by International Society
for Holter and Noninvasive Electrocardiology. J Am Coll Cardiol. 2011;58(13):1309–1324.
Tanno K, et al. Microvolt T-wave alternans as a predictor of ventricular
tachyarrhythmias: a prospective study using atrial pacing. Circulation. 2004;109(15):1854–1858.
Edenbrandt L, Pahlm O. Vectorcardiogram synthesized from a 12-lead ECG: superiority of the
inverse Dower matrix. J Electrocardiol. 1988;21(4):361–367.
Laguna P, Jané R, Caminal P. Automatic detection of wave boundaries in multilead ECG signals:
validation with the CSE database. Comput Biomed Res. 1994;27(1):45–60.
Pan J, Tompkins WJ. A real-time QRS detection algorithm. IEEE Trans Biomed Eng. 1985;32(3):230–236.
Shvilkin A, et al. Vectorcardiographic determinants of cardiac memory during normal
ventricular activation and continuous ventricular pacing. Heart Rhythm. 2009;6(7):943–948.
Lipschutz S, Spiegel M, Spellman D. Schaum’s Outline of Vector Analysis. 2nd ed. McGraw-Hill Professional; 2009
Abry P. Ondelettes et turbulences: Multirésolutions, algorithmes de
décomposition, invariance d’échelle et signaux de
pression. Paris, France: Diderot Editeur Arts Sciences, 1997.
Hostetler B, et al. Detect short run of TWA event with time-domain
algorithm. Compute Cardiol. 2005;2005:483–486.
Kaiser W, Findeis M, Young BJ. Improving T-wave alternans measurement quality by reducing noise and
artifacts. Compute Cardiol. 2004;2004:445–448.
Malik M, et al. Heart rate variability. Standards of measurement, physiological
interpretation, and clinical use. Task Force of the European Society of Cardiology
and the North American Society of Pacing and Electrophysiology. Eur Heart J. 1996;17(3):354–381.
Aminian K, Thouvenin X, Robert P, Seydoux J, Girardier L. A piezoelectric belt for cardiac pulse and respiration measurements on small
mammals. IEEE Engineering in Medicine and Biology Society, 1992 14th Annual
International Conference of the IEEE. 7:2663–2664
Saul JP, et al. Transfer function analysis of the circulation: unique insights into
cardiovascular regulation. Am J Physiol. 1991;261(4 pt 2):H1231–H1245.
Robbe HW, et al. Assessment of baroreceptor reflex sensitivity by means of spectral
analysis. Hypertension. 1987;10(5):538–543.
Song J-G, et al. Effects of bilateral stellate ganglion block on autonomic
cardiovascular regulation. Circ J. 2009;73(10):1909–1913.
Maslyukov PM, Nozdrachev AD. Rhythmic electrical activity in branches of the stellate ganglion in
the cat during postnatal ontogenesis. Neurosci Behav Physiol. 2007;37(5):505–508.
Cutlip DE, et al. Clinical end points in coronary stent trials: a case for standardized
definitions. Circulation. 2007;115(17):2344–2351.
Bardy GH, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive
heart failure. N Engl J Med. 2005;352(3):225–237.
Barthel P, et al. Reflex and tonic autonomic markers for risk stratification in patients
with type 2 diabetes surviving acute myocardial infarction. Diabetes Care. 2011;34(8):1833–1837.
Van Calster B, Van Huffel S. Integrated discrimination improvement and probability-sensitive AUC
variants. Stat Med. 2010;29(2):318–319.