Download Risk stratification after myocardial infarction: a

Europace (2008) 10, 227–234
doi:10.1093/europace/eum302
Risk stratification after myocardial infarction: a new
method of determining the neural component of
the baroreflex is potentially more discriminative
in distinguishing patients at high and low
risk for arrhythmias
Pawel Ptaszynski1,2†, Thomas Klingenheben1,3*†, Bart Gerritse4, and Lilian Kornet4
1
Department of Cardiology, Division of Clinical Electrophysiology, J.W. Goethe University, Frankfurt, Germany;
Medical University Lodz, Lodz, Poland; 3Cardiology Practice Bonn, Im Mühlenbach 2B, Bonn D-53127, Germany;
and 4Medtronic Bakken Research Center, Maastricht, The Netherlands
2
Received 25 September 2007; accepted after revision 20 December 2007
KEYWORDS
Baroreflex;
Neural component;
Risk stratification
Aims We hypothesize that the neural component (NC) of the baroreflex sensitivity (BRS) is a better risk
stratifier for ventricular tachycardia/fibrillation (VT/VF) than conventional BRS itself, because it is both
independent of vessel wall stiffness and can be measured non-invasively.
Methods and results NC was determined by correlating spontaneous carotid artery diameter variations
with R–R interval variations using spectral analyses. In consecutive outpatient populations with
chronic coronary artery disease the ability of the NC to distinguish post-myocardial infarction (MI)
patients at risk for VT/VF (post-MIHIGH RISK) from post-MI less prone to arrhythmias (post-MILOW RISK)
was compared with the pressure-derived BRSphenyl and BRSspectral method. Ninety-six patients, i.e. 28
post-MILOW RISK, 28 post-MIHIGH RISK [a LVEF(left ventricular ejection fraction) ,30% and/or history of
VT/VF] and 40 healthy controls were enrolled. With NC, rather than with BRS methods, median
values for post-MIHIGH RISK were smaller than for post-MILOW RISK patients (NC, P = 0.03; BRSspectral, P =
0.35; BRSphenyl, P = 0.63). Variability of R–R interval (LF = 0.04–0.15 Hz) was significantly larger in the
control group than in the post-MIHIGH RISK and post-MILOW RISK group (P , 0.01 and P , 0.01). To separate
post-MIHIGH RISK from post-MILOW RISK patients, a linear combination of age and the logarithm of the NC
measurement was constructed as a risk index. By optimizing the intercept of this line, an optimal sensitivity and specificity pair was determined. The sum of optimal specificity and sensitivity was higher for
NC (155) than for BRSspectral (133) and BRSphenyl method (132). With all methods, values for post-MI
patients were significantly smaller than for controls.
Conclusion NC may be superior to conventional BRS measures in identifying post-MI patients at high risk
for VT/VF.
Introduction
The primary aim of this retrospective pilot study is to compare
the ability of the NC of the baroreflex with that of conventional
BRS measures to distinguish between post-myocardial infarction
(MI) patients with an accepted indication for primary or secondary preventive implantable cardioverter-defibrillator (ICD)
therapy [i.e. LVEF (left ventricular ejection fraction) 30%; or
†The first 2 authors contributed equally to the study.
*Corresponding author. Tel: þ49 228 623324; fax: þ49 228 616881.
E-mail address: [email protected]; [email protected]
survived ventricular tachyarrhythmic events] and post-MI
patients who did not fulfil these criteria. In addition it was examined: (i) whether neural control of the heart is decreased in postmyocardial patients compared with controls; (ii) whether this
decrease is greater in post-MI patients, who are vulnerable for
ventricular fibrillation than those who are not; and (iii)
whether the NC is better able to distinguish between post-MI
groups and controls than conventional measures of BRS. This
methodologic study was performed in consecutive coronary
artery disease (CAD) patients selected during the chronic phase
of the disease.
Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2008.
For permissions please email: [email protected].
228
Currently the best risk stratifier for post-MI patients is
LVEF.1,2 To further increase specificity and sensitivity, there
is a need for additional risk stratifiers, which can be used
in combination with LVEF. In particular, measures reflecting
cardiac autonomic activity such as heart rate variability
(HRV), and baroreflex sensitivity (BRS), as well as newer parameters such as heart rate turbulence and deceleration
capacity of heart rate have yielded promising results in
this regard.1,3–7
In most clinical studies, BRS has been defined as the linear
component of the relation between changes in pressure and
changes in R–R interval, expressed as change in R–R interval
per change in arterial pressure (ms/mmHg). First, arterial
pressure changes cause stretch and thus diameter changes
of the aortic arch and the carotid body. This relationship is
influenced by vascular wall stiffness.8,9 Secondly, these
diameter changes induce changes in the output of stretchsensitive nerve cells in these vessel walls. These changes
in output are translated by the autonomic nervous system
into vagal outflow to the heart influencing R–R interval
duration. BRS is smaller if the vessel wall is stiffer and
although the vessel wall becomes stiffer with age10 not all
aged persons will suffer arrhythmic events. Although
vessel wall stiffness is correlated with coronary heart
disease and stroke,11 it is not yet known to our knowledge
if it is correlated with arrhythmias. The NC is defined as
the relationship between the diameter changes (input) and
R–R interval changes (output). The NC is not influenced by
absolute diameter changes and vascular wall stiffness, if
assessed during spontaneous rather than drug induced
pressure changes, which is within the linear part of the
relation between stretch and BRS activity.8,9 It is known
that deterioration of the neural components (NCs) of the
baroreflex is responsible for the decrease of BRS in post-MI
patients and is correlated with the occurrence of ventricular
arrhythmias.4,12–15 Therefore, by excluding the mechanical
component of BRS (i.e. the pressure/stretch component),
one can postulate that the remaining NC (i.e. the stretch/
R–R component) is a more reliable predictor of arrhythmia
risk in post-MI patients. Furthermore, peripheral finger
pressures (FinapressTM ), used in most BRS methods, are not
a precise representation of carotid pressures, especially
with advancing age.16–18
In contrast to the conventional BRS measure, where an
injection of a vasoconstrictor is necessary, the assessment
of the NC can be performed non-invasively by extending
the routinely performed cardiovascular ultrasound measurements to assess spontaneous vessel wall diameter changes.
Using long recordings of carotid wall movements and simultaneous measurements of arterial pressure and ECG signal,
the relation between arterial pressure and carotid diameter
(vessel wall component BRS) and carotid diameter and R–R
interval (NC BRS) as well as total BRS can be determined
non-invasively8,9,19,20 in the low-frequency (LF) range
(0.04–0.15 Hz),21 where baroreflex modulation of heart
rate occurs.22–24
Methods
Study population
In 2003, consecutive patients scheduled for a regular follow-up visit
to undergo clinical assessment and risk stratification for chronic CAD
P. Ptaszynski et al.
were selected. During the same period, healthy volunteers were
asked to participate in the study protocol. Patients with chronic
CAD following prior MI could be included if they fulfilled the following criteria: (i) LVEF lower than 30% (primary indication for ICD prevention) and/or a history of VT or VF (secondary prevention for ICD
prevention); designated post-MIHIGH RISK group; (ii) LVEF . 30%
and no history of ventricular tachyarrhythmias; designated postMILOW RISK group. All patients had undergone extensive non-invasive
and invasive evaluation to establish the underlying heart disease.
From all patients and volunteers informed consent was obtained
before they entered the study. The study was approved by the
medical ethical committee of the hospital. In total 96 patients
were enrolled. Twenty-eight post-MIHIGH RISK patients (primary prevention n = 14, and secondary prevention n = 14; average = 19 + 13
months after the infarct), 28 post-MILOW RISK patients (average =
17 + 11 months after the infarct), and 40 healthy normotensive,
non-diabetic, non-hypercholesterolemic, non-smoking volunteers
(controls) were enrolled.
Protocol
Subjects were tested in the morning between 9:00 a.m. and 12:00
noon and refrained from exercise and caffeine consumption prior
to the tests. Beta blockers were withheld 2 days prior to testing.
In each subject common carotid artery distension waveform, systolic diameter and the peak of the R-waves of the ECG waves were
recorded simultaneously for 5 min three times without prior intervention and three times during an injection of a bolus of phenylephrine hydrochloride (Oxford technique, starting dose 200 mg) to
induce changes in pressure of at least 15 mmHg (31 + 17 mmHg).25
The cross-sectional common carotid artery diameter was measured
1 cm distal from the carotid body. The frequency of data acquisition
of the arterial wall movements was 200 Hz. During the ultrasound
recordings arterial finger pressure was assessed simultaneously by
means of a FinapressTM (Ohmeda).
Ultrasound method
The ultrasound echo system in combination with dedicated signal
processing (ATL Mark 9, HDI, Advanced Technology Laboratories,
Bothell, WA, USA) allowed the continuous assessment of systolic
diameter and distension waveform, i.e. the change in diameter
per cardiac cycle, and has been described in detail elsewhere.26
In brief, the common carotid artery was visualized in B-mode
using the C9-5 curved array probe (operating frequency 5–9 MHz).
After positioning the M-line 10 mm from the bifurcation and visualizing the intima-media transition to check for perpendicular positioning, the ultrasound system was switched to echo M-mode with a
pulse repetition frequency of 500 Hz. The spatial variation in enddiastolic diameter along this M-line, using raw radio frequency
data, is 40 mm.27 Recording was started synchronously with a
trigger derived from the simultaneously recorded peak of the
R-wave of the ECG signal. This facilitates the detection of distension
as well as systolic diameter.
Spectral analysis
The relationship between signals was evaluated by means of spectral analysis.28 For example, the relation between systolic diameter
and R–R interval (NC) is described as follows: First, the frequency
contents of the diameter and the R–R interval signal were obtained
by means of Fourier transformation. A very LF trend (,0.005 Hz),
which is the distensibility coefficient (DC) component, was
removed so that only variations were considered. Next the transfer
function H( f ) between the two signals was calculated as
Hð fÞ ¼ Sd;rr ð fÞ=Sd;d ð fÞ
ð1Þ
where Sd,d(f ) is the estimate of the autospectrum of the diameter
of the common carotid artery and Sd,rr( f ) the complex crossspectrum between the two signals. The transfer function magnitude
Risk stratification after myocardial infarction
229
jH( f )j, i.e. gain, reflects for a specified frequency the relative
amplitude of the relation between the systolic diameter of the
common carotid artery and R–R interval. The gain jH(f )j was
derived from the real part HR(f ) and the imaginary part HI( f ) of
the complex transfer function as
HðfÞ ¼ ð½HR ð fÞ2 þ ½HI ð fÞ2 Þ0:5
ð2Þ
The linearity and reliability of the transfer function between both
signals was estimated by a magnitude-squared coherence (C) function. In Figure 1 an example of power spectra of systolic diameter,
R–R interval and corresponding transfer function are given. The
peak between 0.04 and 0.15 Hz (between dashed lines) has been
attributed to the baroreflex.21 Additional details about data analyses have been previously published.9
Data analysis
The relation between change in pressure and change in R–R interval
as assessed after an injection of a bolus of phenylephrine hydrochloride (Oxford technique, starting dose 200 mg) was used to determine
BRSphenyl.25 The average transfer function between systolic pressure
and R–R interval variation (BRSspectral) was calculated in the LF
range between 0.04 and 0.15 Hz.21 In addition, the average transfer
function between systolic diameter and R–R interval variation (NC)
was calculated in the same frequency range. Furthermore, the LF
variation of systolic pressure, systolic diameter and R–R interval
was quantified between 0.04 and 0.15 Hz.9,28 A correction function
to exclude outliers in the systolic diameter, pressure and R–R interval signal due to artefacts, ventricular premature beats, and
measurement errors was applied.
Statistics
Median values of the three methods (BRSphenyl, BRSspectral and NC)
were calculated for the three patient groups by means of descriptive statistics. For further statistical analysis we performed log
transformations on the data (BRSphenyl, BRSspectral and NC) to
obtain normal distributions. In the primary analyses the effect of
age was not considered. We, therefore, first assessed the relation
to age using least squares linear regression on the measures vs.
age. Additionally, a risk index for each of the three BRS measurements was constructed, which separated the generally higher log
transformed values in post-MILOW RISK patients from lower log transformed values in post-MIHIGH RISK patients or higher values in controls
from lower values in post-MI LOW RISK as much as possible: an optimal
slope was determined with logistic regression for a linear combination of age and the logarithm of the measurement (Figure 2A).
Sensitivity and specificity pairs for distinguishing patient groups
were determined retrospectively. These sensitivity and specificity
values were used to construct a ROC curve, in which the optimal
sensitivity, specificity is visualized (Figure 2B). Wilcoxon test was
used to test for non-trivial area under the ROC curve. P-values
were adjusted for multiple comparisons using a Hochberg
correction.29
Results
Patient group description
Table 1 summarizes patient characteristics of controls, and
low and high risk patients. The patient groups did not
differ with respect to diastolic and systolic blood pressure,
resting heart rate (mean R–R interval), Q–T interval duration and its dispersion. Distension, i.e. the change in
Figure 1 Example of power spectra of systolic diameter, R–R interval and corresponding transfer function. The peak between 0.04 and
0.15 Hz has been attributed to the baroreflex.
230
P. Ptaszynski et al.
Figure 2 (A) Risk index: Log NC ¼ intercept20.0306 age. Optimal intercept is 22.144, which corresponds to 85% sensitivity and 70%
specificity. (B) Receiver operating characteristic (ROC) curve corresponding to Figure 2A.
Table 1 Patient characteristics
Controls
Age (years)
[median]
Sex (male/
female)
Site of MI
Post-MILOW
Table 2 Medication. Note that patients were omitted from beta
blockers 2 days before the tests
RISK
Post-MIHIGH
RISK
50
56
26/14
(65% / 35%)
None
24/4
22/60
(86% / 14%)
(79% / 21%)
15 anterior/13 15 anterior/11
inferior
inferior
posterior
posterior/2
unknown
Controls
Post-MILOW
0/40
0/40
0/40
0/40
0/40
0/40
25/28
26/28
2/28
23/28
27/28
0/28
RISK
Post-MIHIGH
RISK
61
diameter per cardiac cycle, decreased significantly with age
(linear regression) and this age-dependent decrease was not
significantly different between the groups. End-diastolic
diameter increased significantly with age in the three
groups and the increase was significantly higher in postMILOW RISK and post-MIHIGH RISK patients than in controls
(linear regression, slopes compared). Intima-media thickness (IMT) increased significantly with age and the increase
with age was not significantly greater in the post-MILOW RISK
and post-MIHIGH RISK group compared with the control
group. However, a significantly greater IMT was found in
the post-MILOW RISK and post-MIHIGH RISK group compared
with controls. Medication for the post-MILOW RISK and postMIHIGH RISK group is given in Table 2. Only the use of diuretics
was different between the post-MILOW RISK and post-MIHIGH
RISK
group. The controls did not use any medication
related to cardiovascular disease.
Baroreflex sensitivity methods and neural
component method
In Figure 3 plots with median, 25 and 75% percentile values
for the NC (A), BRSphenyl (B) and BRSspectral (C) are shown of
controls, post-MILOW RISK and post-MIHIGH RISK patients. A significant difference is present between post-MIHIGH RISK and
post-MILOW RISK patients if the NC is used (P = 0.03, Wilcoxon
test), but not if conventional BRS measures were used.
These differences were not found for the BRS methods.
Beta-blockers
ACE inhibitors
Diuretics
Statins
Aspirin
Nitrates
26/28
27/28
21/28
22/28
18/28
1/28
With all three measures median values of the control
group could be distinguished significantly from median
values of post-MI patients.
A significant decrease in the NC vs. age was found in the
control group [P (slope) = 0.0009] and the post-MILOW RISK
group [P (slope) = 0.006], but not in the post-MIHIGH RISK
group [P (slope) = 0.82]. For the BRSphenyl and BRSspectral
methods no significant relation to age was found.
Ability of the methods to distinguish patient groups
As described above, this study has shown that high and low
risk post-MI groups can be better distinguished using the
median NC than the median BRS values. To correct for the
influence of age and asses how good the individual data
could predict if a patient belongs to the high or low risk
groups, a risk index was calculated. In order to obtain
normal distributions, a log transformation on the measurements obtained with the three methods was performed.
Since we found a significant relation to age for the NC, a
linear combination of age and the logarithm of the measurement was constructed as a risk index for the three methods
using logistic regression. With this technique an optimal
slope of a line depending on age was determined, which separates the value of a parameter determined in high and low
risk post-MI patients as well as possible. By varying the intercept, various sensitivity-specificity pairs were obtained and
a ROC curve was calculated using these pairs. Table 3 presents the indices and performance characteristics: area
Risk stratification after myocardial infarction
231
arterial finger pressure (LF) and age or patient groups
was found.
Variability of systolic diameter (LF) increased with age
(linear regression). In the control group (P = 0.01) and postMIHIGH RISK group (P = 0.01) these changes were significant.
The increase with age was significantly greater in the postMILOW RISK group and post-MIHIGH RISK group compared with
that in controls. A significantly greater variability of systolic
diameter was found for controls than for post-MILOW RISK and
post-MIHIGH RISK patients, respectively.
Correlations between methods
A moderate correlation (R 2 = 0.29) was found between the
NC and BRSspectral values for the three groups [NC (ms/
mm) ¼ 0.05 BRSspectral (ms/mmHg) þ 0.003, N = 76] if a
linear regression was performed on all data of patients
and controls together. Furthermore, a moderate correlation
(R 2 = 0.27) was found between the BRSspectral and BRSphenyl
method for the three groups [BRSspectral (ms/mmHg) =
1.18 BRSphenyl (ms/mmHg) þ 0.95], N = 59.
Discussion
Figure 3 Median, 25 and 75% plots for the neural component (A),
BRSspectral (B), and BRSphenyl (C ) in controls, post-MIHIGH RISK and
post-MILOW RISK patients.
under ROC curve, P-value for non-triviality, the intercept
that achieved the highest sum of sensitivity and specificity,
the achieved sensitivity and specificity and the sum of
the two. The area under the ROC curve and the sum of
the optimal specificity and sensitivity to distinguish postMILOW RISK from post-MIHIGH RISK patients was greater for the
NC (sum = 155, areaROC = 81) than for the BRSspectral (sum =
133, areaROC = 65) and BRSphenyl method (sum = 132,
areaROC = 59) (Table 3).
Variabilities
Average variability of R–R interval (LF ¼ 0.04–0.15 Hz)
assessed from the short-time measurements was significantly larger in the control group than in the post-MIHIGH
RISK
and post-MILOW RISK group (P , 0.01 and P , 0.01). No
influence of age could be detected on variability of R–R
interval in post-MILOW RISK (P = 0.13), post-MIHIGH RISK patients
(P = 0.97) and controls (P = 0.09). Assessment of standard
HRV from Holter recordings was not part of the present
study protocol. No relation between variability of systolic
Although primary preventive therapy with an ICD is evidence
based in post-MI patients with reduced ejection fraction,1
many of these patients will never use their ICD. There is
thus a need for additional non-invasive methods to identify
those patients who may derive a particular benefit from
such treatment. In the past, several autonomic measures
have been investigated with respect to arrhythmia risk stratification after MI.3,5–7 The predictive value of measures
reflecting BRS seems to be superior to that of HRV.
However, conventional BRS3,5 has the disadvantage of
injecting a vasoactive agent. Newer methods such as heart
rate turbulence and deceleration capacity of heart rate –
both determined from Holter recordings – yield a high
predictive value in MI patients treated according to contemporary guidelines.6,7 In this retrospective pilot study, post-MI
patients at high risk for ventricular arrhythmias (LVEF lower
than 30% or previous ventricular tachycardia/fibrillation
(VT/VF) i.e. approved indication for primary or secondary
preventive ICD therapy)1,30 could be better distinguished
from post-MI patients at low risk for ventricular arrythmias
with the non-invasively determined NC than with the
BRSphenyl and BRSspectral method. The NC reflects BRS as
modulated by the sensing and vagal components without
the influence of changes in artery wall properties, since
the NC reflects the relationship between stretch and R–R
interval variations and not absolute vessel wall stretch
values. However, the concept of NC as a better predictor
of ventricular arrhythmias compared with the conventional
BRS measures and the additional value of NC in combination
with other risk stratifiers needs confirmation in a future prospective study. The added value of the NC compared with
total BRS methods may be explained by the elimination of
the vessel wall component, i.e. the transfer of pressure variations to diameter variations, which is influenced by the
stiffness of the vessel wall. Vessel wall stiffness increases
with age.10 Although high vessel wall stiffness is correlated
with coronary heart disease and stroke,11 its relation to ventricular tachyarrhythmias is not yet known. Another advantage of the NC is the elimination of the extrapolation of
232
P. Ptaszynski et al.
Table 3 For the methods a risk index is constructed using logistic regression, with respect to patients from the three groups. The table
presents the indices and performance characteristics: area under ROC curve, P-value for non-triviality, the cut-off that achieved the highest
sum of sensitivity and specificity, and the achieved sensitivity and specificity
Method
Risk index
Area under ROC
curve (%)
P-value for
non-triviality
Optimal
intercept
Sensitivity
(%)
Specificity
(%)
Sum
(specificity +
sensitivity)
NC
log(NC) ¼ intercept
20.0306 age
log(BRS) = intercept
20.0853 age
log(BRS) ¼ intercept
20.0682 age
81
0.011
22.144
85
70
155
65
0.69
6.354
92
41
133
59
0.36
4.686
70
62
132
BRSspectral
BRSphenyl
arterial finger pressure to central pressure. Prior work has
demonstrated that variations in carotid diameter correlated
significantly better with variations in R–R interval than variations in arterial finger pressure due to elimination of the
noise of incorrect pressure measurements.9
In this study, a smaller NC was found in post-MILOW RISK
patients than in controls. The NC was particularly low in
post-MI patients with a low ejection fraction or previous
VT/VF, i.e. patients prone to life threatening tachyarrhythmias.1,30 The pivotal role of deterioration of the NCs of
the baroreflex in the decrease of BRS, has been demonstrated in studies of myocardial infarction patients, where
areas of sympathetic denervation exceeding the limits of
myocardial necrosis have been demonstrated.15 This denervation has been considered a substrate for ventricular tachycardia and fibrillation. Furthermore, impaired reflex control
of heart rate is related to the central effects of altered
afferent input from the heart on the baroreceptors after
a myocardial infarct.12 In an experimental model in 191
dogs with a healed myocardial infarction, Schwartz and
coworkers4,14 demonstrated that reduced BRS (total
pathway) is associated with a greater susceptibility to ventricular fibrillation during subsequent ischemic episodes. In
these animals electrical vagal stimulation markedly
reduced the incidence of VF in a high-risk subgroup,
whereas muscarinic blockade increased VT/VF in low risk
animals.13 Furthermore, it was found that a muscarinic
agonist significantly reduced malignant arrhythmias during
acute myocardial ischaemia in 17 felines.13 The ATRAMI
(Autonomic Tone and Reflexes After Myocardial Infarction)
trial with 1071 post-MI patients demonstrated that markers
of reduced autonomic tone such as BRS and HRV, are
strong predictors of sudden cardiac death.3 The power of
BRS remained significant even after adjusting for LVEF and
number of premature ventricular complexes on Holter monitoring. Among patients with LVEF , 30%, BRS but not HRV
remained a powerful predictor.
Age
The trend observed in the decrease of R–R variability with
age is in agreement with the significant decrease in R–R
variability found in 2722 subjects by Tsuji et al.31 A significant increase in variability of carotid diameter with age
was found in this study. This increase can be explained by
the deterioration of the NC with age resulting in a decrease
in buffering capacity of the heart on blood pressure variations. An increase in blood pressure variability with age
leads to an increase in carotid diameter variability.
However, no relation was found between variability of arterial finger pressure and age. It should be noted that the bias
in the measurement of arterial finger pressure variability
(caused by the effect of wave reflections) changes with
age and, therefore, disturbs a possible relation between
pressure variability and age.
End-diastolic diameter increased significantly with age in
the control, post-MILOW RISK and post-MIHIGH RISK group. This
increase was more pronounced in post-MI patients than in
controls irrespective of their ejection fraction and history
of ventricular arrhythmias. Also intima-media thickess was
greater in post-MI patients than in controls. These findings
indicate that the effect of vessel wall aging is more pronounced in post-MI patients than in controls.
Comparison between spectral and
phenylepinephrine methods
A moderate correlation between the NC and BRSphenyl and
between BRSphenyl and BRSspectral was found. The latter corresponds to the results of Maestri et al.32 Furthermore, the
variation in BRSphenyl and BRSspectral in this study was comparable with that found by Maestri et al.32 This moderate
correspondence was found despite the possible influence
of phenylepinehrine on the baroreceptors and the possibility
that BRS, as determined by drug induced changes in
pressure, was measured in the non-linear part of the
stretch-BRS activity curve; whereas BRS as determined
with the spectral method using small spontaneous changes
in pressure was determined in the linear part of the curve.
Limitations and future studies
Although the NC performed better than the conventional
BRS measures in distinguishing post-MI patients at risk for
ventricular arrhythmias (as indicated by a low EF and/or a
prior ventricular arrhythmic event) from those less prone
to arrhythmias in this retrospective pilot study, the
concept of the NC being a better predictor of ventricular
arrhythmias than conventional BRS measures needs confirmation in a prospective study with a long duration of
follow-up. Groups were too small to determine a potential
relationship between NC and future arrhythmic events. Furthermore, patients were not consequently followed for the
purpose of this methodological study. The predictive value
of the NC combined with other risk factors on the occurrence of VT/VF thus needs further prospective investigation.
Risk stratification after myocardial infarction
Particularly, the value of the NC in comparison with ejection
fraction needs to be studied. In addition, it should be noted
that in the control group of this study the possibility of background arrhythmia could not be excluded.
The area under the ROC curve obtained with the NC
method had a significant P-value of 0.011. However, it
should be noted that the P-value for the area under the
ROC curve was not significant for the BRS methods. Thus,
to obtain a statistically significant difference between sensitivities and specificities of the various methods to distinguish patient groups, larger sample sizes would be
required. Cut-off values for identification of patients at
risk, and sensitivity and specificity estimates for the proposed risk-stratification are based on the data of the population used in this study. Thus, sensitivity and specificity
might be biased. Furthermore, the intra-individual reproducibility of NC measurements is not known and was not tested
in the present study.
With respect to the medication, beta blocker therapy was
withheld 2 days prior to the measurements, i.e. five
half-lives of drug action. This may not allow for all effects
to resolve. However, the procedure was in adherence to
the protocol of other studies investigating autonomic risk
stratifiers in post-MI patients. Although arrhythmia risk stratification should in clinical practice be performed in patients
on full protective medication, we decided to withhold beta
blockers in order to better compare the results between
patients and healthy volunteers in the present study.
Clinical implications
The results of the present study indicate that assessment of
NC of the baroreflex may more accurately reflect vagal
reflex activity. Furthermore, this measure is less invasive
than conventional BRS assessment since there is no need
to inject a vasoactive agent. In this pilot study, the NC
better separates post-MI patients at risk of ventricular
arrhythmia as indicated by a low LVEF and/or history of ventricular arrhythmia from those less prone to arrhythmia than
the conventional BRS measures. There is therefore a potential of NC to identify patients at particular high risk of
arrhythmic events. Future studies are warranted to define
the value of the NC in selecting patients who may benefit
from primary preventive ICD therapy.
Conflict of interest: B.G. and L.K. are employees of Medtronic Inc.
Acknowledgement
The authors would like to express their gratitude to Prof.
Richard Sutton, MD, for his expert advice and critical
review regarding this manuscript.
References
1. Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. Multicenter Automatic Defibrillator
Implantation Trial II Investigators. N Engl J Med 2002;346:877–83.
2. Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart
failure. N Engl J Med 2005;352:225–37.
233
3. La Rovere MT, Bigger JT Jr, Marcus FI, Mortara A, Schwartz PJ. Baroreflex
sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction. ATRAMI (Autonomic Tone and Reflexes
After Myocardial Infarction) Investigators. Lancet 1998;351:478–84.
4. Schwartz PJ, Vanoli E, Stramba-Badiale M, De Ferrari GM, Billman GE,
Foreman RD. Autonomic mechanisms and sudden death. New insights
from analysis of baroreceptor reflexes in conscious dogs with and
without a myocardial infarction. Circulation 1988;78:969–79.
5. La Rovere MT, Pinna GD, Hohnloser SH, Marcus FI, Mortara A, Nohara R
et al. Baroreflex sensitivity and heart rate variability in the identification
of patients at risk for life-threatening arrhythmias: implications for clinical trials. Circulation 2001;103:2072–7.
6. Bauer A, Kantelhardt JW, Barthel P, Schneider R, Mäkikallio T, Ulm K et al.
Deceleration capacity of heart rate as a predictor of mortality after
myocardial infarction: cohort study. Lancet 2006;367:1674–81.
7. Barthel P, Schneider R, Bauer A, Ulm K, Schmitt C, Schömig A et al. Risk
stratification after acute myocardial infarction by heart rate turbulence.
Circulation 2003;108:1221–6.
8. Kornet L, Hoeks AP, Janssen BJ, Houben AJ, De Leeuw PW, Reneman RS.
Neural activity of the cardiac baroreflex decreases with age in normotensive and hypertensive subjects. J Hypertension 2005;23:815–23.
9. Kornet L, Hoeks AP, Janssen BJ, Willigers JM, Reneman RS. Carotid diameter variations as a non-invasive tool to examine cardiac baroreceptor
sensitivity. J Hypertension 2002;20:1165–73.
10. Reneman RS, Van Merode T, Hick P, Muytjens AMM, Hoeks APG.
Age-related changes in carotid artery wall properties in man. Ultrasound
Med Biol 1986;12:465–71.
11. Mattace-Raso FU, van der Cammen TJ, Hofman A, van Popele NM, Bos ML,
Schalekamp MA et al. Arterial stiffness and risk of coronary heart disease
and stroke: the Rotterdam Study. Circulation 2006;113:657–63.
12. Minisi AJ, Nashed TB, Quinn MS. Regional left ventricular deafferentation
increases baroreflex sensitivity following myocardial infarction. Cardiovasc Res 2003;58:136–41.
13. De Ferrari GM, Vanoli E, Curcuruto P, Tommasini G, Schwartz PJ. Prevention of life-threatening arrhythmias by pharmacologic stimulation of the
muscarinic receptors with oxotremorine. Am Heart J 1992;124:883–90.
14. De Ferrari GM, Vanoli E, Cerati D, Schwartz PJ. Baroreceptor reflexes and
sudden cardiac death: experimental findings and background. G Ital
Cardiol 1992;22:629–37.
15. Minardo JD, Tuli MM, Mock BH, Weiner RE, Pride HP, Wellman HN et al.
Scintigraphic and electrophysiological evidence of canine myocardial
sympathetic denervation and reinnervation produced by myocardial
infarction or phenol application. Circulation 1988;78:1008–19.
16. O’Rourke MF, Nichols WW. Does reduced vascular stiffening fully explain
preserved cardiovagal baroreflex function in older physically active men.
Circulation 2002;105:E11.
17. Avolio A, Qasem A, Park YJ. Noninvasive estimation of baroreflex sensitivity using pressure pulse amplification. IEEE Eng Med Biol Mag 2001;
20:53–8.
18. Monahan KD, Dineno FA, Seals DR, Clevenger CM, Desouza CA, Tanaka H.
Age-associated changes in cardiovagal baroreflex sensitivity are related
to central arterial compliance. Am J Physiol Heart Circ Physiol 2001;
281:284–9.
19. Hunt BE, Fahy L, Farquhar WB, Taylor JA. Quantification of mechanical
and neural components of vagal baroreflex in humans. Hypertension
2001;37:1362–8.
20. Hunt BE, Farquhar WB, Taylor JA. Does reduced vascular stiffening fully
explain preserved cardiovagal baroreflex in older, physically active
men? Circulation 2001;103:2424–7.
21. Task Force of the European Society of Cardiology the North American
Society of Pacing and Electrophysiology. Heart rate variability: standards
of measurement, physiological interpretation, and clinical use. Circulation 1996;93:1043–65.
22. Kamath MV, Fallen EL. Power spectral analysis of heart rate variability:
a non-invasive signature of cardiac autonomic function. Crit Rev
Biomed Eng 1993;21:245–311.
23. Sleight P, La Rovere MT, Mortara A, Pinna G, Maestri R, Leuzzi S et al.
Physiology and pathophysiology of heart rate and blood pressure variability in humans: is power spectral analysis largely an index of baroreflex
gain. Clin Sci 1995;88:103–9.
24. Mancia G, Parati G, Castiglioni P, Di Rienzo M. Effect of sinoaortic denervation on frequency-domain estimates of baroreflex sensitivity in
conscious cats. Am J Phys 1999;276:1987–93.
25. La Rovere MDMT, Pinna MSGD, Mortara MDA. Assesment of baroreflex sensitivity. In: Clinical Guide To Cardiac Autonomic Tests 1998.
234
26. Brands PJ, Hoeks AP, Willigers J, Willekes C, Reneman RS. An integrated
system for the non-invasive assessment of vessel wall and hemodynamic
properties of large arteries by means of ultrasound. Eur J Ultrasound
1999;9:257–66.
27. Meinders JM, Brands PJ, Willigers JM, Kornet L, Hoeks APG. Assessment of
spatial homogeneity of artery dimension parameter with high frame rate
2D B-mode. Ultrasound Med 2001;27:785–94.
28. Zhang R, Zuckerman JH, Giller CA, Levine BD. Transfer function analysis of
dynamic cerebral autoregulation in humans. Am J Phys 1998;274:H233–41.
29. Hochberg Y.A. Sharper bonferroni procedure for multiple significance
testing. Biometrika 1988;75:800–2.
P. Ptaszynski et al.
30. Moss AJ, Hall WJ, Cannom DS, Daubert JP, Higgins SL, Klein H et al.
Improved survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. Multicenter Automatic
Defibrillator Implantation Trial Investigators. N Engl J Med 1996;335:
1933–40.
31. Tsuji H, Venditti FJ Jr, Manders ES, Evans JC, Larson MG, Feldman CL
et al. Determinants of heart rate variability. J Am Coll Cardiol 1996;
28:1539–46.
32. Maestri R, Pinna GD, Mortara A, La Rovere MT, Tavazzi L. Assessing baroreflex sensitivity in post-myocardial infarction patients: comparison of
spectral and phenylephrine techniques. J Am Coll Cardiol 1998;31:
344–51.