From the Departments of Pharmacology, Pediatrics and Medicine, College of
Physicians and Surgeons of Columbia University, New York, NY.
Correspondence to Michael R. Rosen, MD, Gustavus A. Pfeiffer Professor of Pharmacology, Professor of Pediatrics, College of Physicians and Surgeons of Columbia University, Department of Pharmacology, 630 W 168 St, PH 7West-321, New York, NY 10032. E-mail franeye{at}cudept.cis.columbia.edu
Methods and Results—Dogs were paced from the ventricles at rates
of 110 to 120 bpm for
Conclusions—CM is a dynamic process for which the final T vector
is predicted by the paced QRS vector and which is associated with
significant changes in epicardial and endocardial but not midmyocardial
cell action potential duration, such that the transmural gradient of
repolarization is altered. It is unaccompanied by evidence of altered
hemodynamics or flow, requires a change in pathway of
activation, and appears to require new protein synthesis.
Our purpose in the present study was to determine the kinetics and
stability of cardiac memory and its association with changes in the
cardiac action potential and to test whether it shares properties in
common with neuronal long-term potentiation. This is not necessarily a
far-fetched idea, given the frequent lessons in nature relating to
conservation of proteins and of signaling processes. To this end, we
used a model of pacing-induced long-term cardiac memory in the intact
dog to ask the following questions: (1) Can T-wave memory be verified
to occur in the absence of any manifestation of cardiac failure or
ischemia? (2) Are there means for its accurate quantification?
(3) Is it a purely HR-dependent process or is an altered activation
pathway essential to its genesis? (4) Does it manifest the property of
accumulation and if so, what are the kinetics of its evolution and
resolution? (5) Is its course modified by protein synthesis
inhibitors? and (6) Is it associated with action potential
changes that might alter the voltage gradients contributing to the T
wave?
Ventricular or atrial pacing was instituted (mode VVO;
rate, 110 to 120 bpm; amplitude, 3.3 to 5 V; pulse width, 0.35 to 0.5
ms) at a rate 10% to 15% faster than that of the animal's sinus
rhythm. All ECGs were recorded with animals standing quietly in a
sling. Every 2 to 3 days, the pacemaker was shut off for 1 hour and the
ECG was recorded at the end of that period. For all ECGs, six limb
leads and lead V10 were acquired at a 500- to
1000-Hz sampling rate on a computer using PO-NE-MAH data acquisition
software (Gould Instrument Systems, Inc) and the PC-EKG program (ISI,
Inc). Frontal vector images were reconstructed using the
pseudo-orthogonal lead system I-aVF-V10. The
distance from the origin of a vector loop to its most remote point was
considered to be the amplitude of the QRS or T-wave vector, and the
vector angle was calculated as the angle between this line and the 0
axis.
Twenty-four-hour monitoring was performed on four animals on random
days to ensure reliable pacemaker capture. At no time was the extent of
the capture <75%. Once cardiac memory developed and stabilized,
pacing was discontinued and some animals were allowed to recover for 3
to 5 weeks. The others were anesthetized with sodium
pentobarbital 30 mg/kg IV, and their hearts were removed and prepared
for cellular electrophysiological
study.
For studies of protein synthesis inhibition, cycloheximide (Aldrich
Chemical Co) was given to three dogs. A cumulative dose totaling 2400
mg was administered over a period of 2 days (two subcutaneous
injections) before pacing. We arrived at this dose on the basis of the
work of MacLean and Ungar4 and our own
(unpublished) work on dose ranging in dogs and rats.
For studies of hemodynamics and of myocardial flow,
four additional adult mongrel dogs weighing 24 to 27 kg were
anesthetized with isoflurane (2%) and mechanically ventilated.
A thoracotomy was performed in the left fifth intercostal space under
sterile conditions. Tygon catheters (ID, 0.04 to 0.05 in; OD, 0.07 to
0.09 in; Cardiovascular Instrument Corp) were placed in
the descending thoracic aorta, the apex of the LV, and the left atrium.
Instrumentation for cardiac pacing was as described above, and pacing
was performed from the anterior LV site.
After recovery from surgery, hemodynamic studies were
performed as a baseline before and immediately after pacing was
instituted. Recordings were made again at 3 weeks of continuous
ventricular pacing and after pacing was discontinued. All
hemodynamic measurements were made with the dog lying
on its right side in a conscious state. The previously implanted
fluid-filled Tygon catheters were connected to transducers (Statham
Instruments, Inc) to measure aortic pressure (MAP), LV pressure, and
left atrial pressure. The LV pressure signal was electronically
differentiated (Differentiator Signal Conditioner, Gould Electronics)
to measure LV dP/dt. The data were recorded on an eight-channel
recorder (Gould model 3800). Mean values were derived for aortic
pressure and left atrial pressure. To test ß-adrenergic
responsiveness, isoproterenol (Elkins-Sinn) was given as
intravenous bolus injections at doses of 0.1 and 0.5
µg/kg, and the maximum changes of LV dP/dt, HR, and MAP were
measured.
Regional myocardial blood flow was determined by injecting 2 mL of
colored microspheres within 15 seconds into the previously
implanted left atrial catheter. At the same time, a 15-mL sample of
aortic blood was obtained by a steady withdrawal (Harvard
Apparatus) at a rate of 7.5 mL/min for 2 minutes for
determination of blood microsphere concentration. Cardiac
pacing was then initiated. When hemodynamic
parameters achieved a steady state,
hemodynamic measurements and isoproterenol and colored
microsphere (of a different color) injections were repeated.
The pacer was then kept on for 3 weeks, after which the same
experimental protocol was repeated with the heart paced and then at the
spontaneous sinus rhythm (45 minutes after the pacemaker was turned
off). Two additional colors of microspheres were injected for
this protocol.
At the end of the entire experiment, the heart was removed, weighed,
and cut into small (
Microelectrode Methods
Transmembrane potentials were recorded with the use of 3 mol/L
KCl-filled glass capillary microelectrodes (tip resistances of 10 to 20
M
Statistical Analysis
To further explore whether 3 weeks of ventricular pacing
led to any detectable signs of heart failure,
hemodynamic responses to bolus infusions of
isoproterenol (0.5 µg/kg) were determined before the onset of pacing
and at the end of 3 weeks of pacing with the pacemaker turned off.
There were no significant differences in percent changes in
dP/dtmax (125±21% versus 131±21%) or in MAP
(-40±3% versus -36±0.2%). Peak HRs observed after the bolus
infusion were also insignificantly different between the two conditions
(237±10.2 versus 215±18.4 bpm).
As detailed in "Methods," regional myocardial blood flow was
assessed with the use of colored microspheres. Results obtained
from the basal, middle, and apical regions of the anterior LV wall,
which were further subdivided into endocardial, midmyocardial, and
epicardial zones, are summarized in Fig 2
ECG Studies
As the duration of ventricular pacing increased in all
animals, a gradual alignment of the sinus rhythm T-wave vector with the
paced QRS vector occurred. This resulted in rotation of the sinus
T-wave vector toward the paced QRS vector (ie, opposite to the paced
T-wave vector (Fig 4
To further test the association between the paced QRS-T axis and the
final T-wave vector, we performed additional experiments using pacing
sites on the posterior wall of the LV or the right
ventricular free wall. As expected, changes in pacing site
resulted in different paced QRS-T axis directions (Fig 5
The time course of T-wave vector change is plotted in Figs 6
Fig 8
As a control for the effects of increasing HR alone, we performed
atrial pacing in another group of four dogs (Table 2
The next set of experiments made use of cycloheximide to test the role
of new protein synthesis in the genesis of cardiac memory. In one of
the three animals studied, four sequences of ventricular
pacing were performed. Between each run, the animal remained in sinus
rhythm for a time sufficient for the T wave to return to control
levels. Before the second and fourth runs, cycloheximide was
administered as described in "Methods." Note in Fig 9
Cellular Electrophysiological Studies
The relationships between the changes in APD in epicardium and
endocardium are further explored in Fig 10
Figs 6 through 8
There are a number of determinants of the electrical gradient that
underlie the Q-T interval and T wave, including the
apico-basal13 14 and
epicardial-endocardial15 differences in
ventricular myocardial action potentials, whose timing
relative to one another is also influenced by the activation sequence
of the ventricle. In the present study, we have deliberately
focused on the epicardial-endocardial gradient in one region (the LV
anterobasal free wall) as well as on the role of activation. With
respect to the epicardial-endocardial gradient, the major effect of the
ventricular pacing protocol is to change the voltage-time
course of repolarization in endocardium and epicardium but not
midmyocardium, such that the differences in repolarization
time among the three are diminished. We must emphasize that these
action potential data can be interpreted in a limited context only.
Although we studied epicardium, endocardium, and
midmyocardium, this was at one site only in the LV. As
such, although it is clear that long-term memory is associated with
action potential changes, the extent to which gradients for
repolarization have been altered across the entire mass of the left and
right ventricles is not considered. Such information awaits detailed
mapping studies of ventricular repolarization in the
setting of memory.
Our earlier work11 12 and reports in the
literature16 suggest that the induction of memory
is based on fundamental alterations in a subset of ion channels that
determine the voltage-time course of repolarization. One of the
determinants of the action potential that is altered is the phase 1
notch.17 In preliminary studies, we have shown
that the transient outward current, Ito1, which
is a major determinant of the notch, has altered activation
kinetics11 and that its density and the mRNA for
its channel protein Kv4.3 are significantly reduced in long-term
cardiac memory (H. Yu, PhD, et al, preliminary data, 1997). Fig 8
The other factor contributing importantly to the gradient for
repolarization is the ventricular activation pathway. For
this reason, it was important to determine the extent to which cardiac
memory is influenced by changes in HR alone as opposed to the
combination of altered rate and activation pathway. It is apparent from
Table 2
It is of particular interest that the time course of onset of
cardiac memory is modified by cycloheximide. Our reason for using
cycloheximide was in part the potential analogy that exists between
memory in heart and memory in central nervous system. In the central
nervous system, memory is a process of long-term potentiation induced
by repetitive exposure to a signal that results in new protein
synthesis.21 22 The process is common to complex
mammalian life forms as well as to far simpler organisms such as
aplysia. Neuroscientists have relied on cycloheximide as an
inhibitor of new protein
synthesis,23 24 using it to differentiate the
mechanisms responsible for long-term memory (requiring new protein
synthesis) and those responsible for short-term memory (dependent on
channel phosphorylation).22 The
dose of cycloheximide we used was derived from experience by others in
the dog4 and from our own preliminary
dose-ranging experiments in rat and dog (data not reported). Although
we did not measure any independent descriptor of new protein synthesis,
the literature clearly provides evidence for the expectation of its
inhibition.23 24 When cycloheximide was
administered, the evolution of memory was significantly delayed.
Moreover, as shown in Fig 9
Given the absence of any structural remodeling or deterioration of
coronary flow or mechanical function, we assume that the
expression of new protein synthesis in memory is manifested uniquely at
the subcellular level. This is of particular importance in evaluating
memory-associated changes in action potential in light of our
preliminary description of changes in Kv4.3 mRNA and functional
properties of Ito1 that accompany the memory
process (Reference 1111 and H. Yu, PhD, et al, preliminary data, 1997).
These observations support the speculation that fundamental alterations
in channel density as well as protein structure may occupy the
induction of long-term cardiac memory.
In conclusion, long-term cardiac memory is a clearly defined alteration
in the T wave induced in settings in which ventricular
activation is changed in the absence of myocyte
hypertrophy, cardiac ischemia, or
hemodynamic deterioration. Its expression can be
delayed by inhibition of new protein synthesis, and it is associated
with changes in the action potential that appear to reflect alteration
in specific ion channels. Definition of the role of the memory process
in the modulation of the effective refractory period and of cardiac
arrhythmias and further definition of its transduction pathways
and molecular determinants are needed. Finally, we stress that results
of the present study relate entirely to the process of long-term
cardiac memory, a process requiring days to weeks for its induction and
weeks to months for resolution. They are not to be confused with the
changes seen in short-term memory, which requires minutes to induce and
minutes to hours for resolution.12 25
Received September 2, 1997;
revision received November 1, 1997;
accepted November 24, 1997.
© 1998 American Heart Association, Inc.
Basic Science Reports
Evolution and Resolution of Long-term Cardiac Memory
Abstract
Top
Abstract
Introduction
Methods
Results
Discussion
References
Background—Cardiac memory (CM)
refers to T-wave changes induced by ventricular pacing or
arrhythmia that accumulate in magnitude and duration with
repeated episodes of abnormal activation. We report herein the kinetics
of long-term CM and its association with the ventricular
action potential. 
3 weeks. CM characterized by gradual sinus
rhythm T vector rotation toward the paced QRS vector evolved in all
dogs regardless of pacing site (left ventricular [LV]
anterior apex or base, posterior LV, or right ventricular
free wall). Cardiac hemodynamics and myocardial flow
(microsphere studies) were unaltered by the pacing. Recovery
time for the memory T wave to return to control increased with duration
of the previous pacing. The protein synthesis inhibitor
cycloheximide markedly (P<.05) and reproducibly
attenuated evolution of CM. When pacing was performed from the atrium,
CM did not occur. Standard microelectrode techniques were used to study
action potential from the LV free wall of control and CM dogs. CM was
associated with increased action potential duration in epicardial and
endocardial but not midmyocardial cells, significantly altering the
transmyocardial gradient for repolarization.
Key Words: action potentials • electrocardiography • electrophysiology • T-waves • pacing
Introduction
Top
Abstract
Introduction
Methods
Results
Discussion
References
The term "cardiac
memory," used with respect to the ECG, was described and named by
Rosenbaum et al1 and had been reported
independently by Chatterjee et al.2 Cardiac
memory is a characteristic of the T wave during sinus rhythm that,
after periods of ventricular pacing or
ventricular arrhythmia, assumes a vector
approaching that of the paced or arrhythmic QRS
complex.1 3 A key aspect of the T-wave change
described by Rosenbaum et al1 is
"accumulation," ie, repeated periods of exposure to the inciting
event augment and sustain the T-wave change during subsequent sinus
rhythm. Equally important is that the T-wave change occurs in hearts
that have no demonstrable hemodynamic or structural
abnormalities or ischemia.
Methods
Top
Abstract
Introduction
Methods
Results
Discussion
References
Pacemaker Implantation
Control ECGs were recorded from 20 mongrel dogs of either
sex weighing 22 to 27 kg. Anesthesia was induced with
propofol 6 mg/kg IV, followed by inhalation of isoflurane (2%). Under
sterile conditions, the chest was opened, the heart suspended in a
pericardial cradle, and a Medtronic permanent pacing lead (model 6917)
attached to the epicardium of the left atrium or the right or left
ventricle. The lead was connected to a Medtronic MINIX 8340 pulse
generator that was placed in a subcutaneous pocket. The incisions were
closed, and the animals were allowed to recover for 2 to 3 weeks,
during which time they were laboratory trained and the ECG
stabilized. 1 g) samples from three regions of
myocardium: LV anterior wall, LV posterior wall, and
septum. LV anterior and posterior wall samples were further cut into
epicardial, midmyocardial, and endocardial layers. Tissue samples were
digested and the microspheres were retrieved and then trapped
in a polyester membrane filter (10 µm in pore size, 25 mm in
diameter; Poretics Corporation). The color dye was then digested from
the microspheres with the use of 100 µL of dimethylformamide.
The photometric absorption of each 100-µL sample was then measured by
a diode array ultraviolet/visible spectrophotometer (model 8452A;
Hewlett-Packard Co). The composite spectrum of each dye solution was
resolved at peak frequencies into the contributions from the individual
colored spheres by use of a matrix inversion
technique.5 We calculated regional myocardial
blood flow from the coronary circulation by determining
microsphere concentrations from aortic blood samples in a
similar manner for each colored sphere using the standard
equation5
where CBFreg is the regional
coronary blood flow, ABSreg is the
absorbance of the specific color, Fao is the
aortic withdrawal rate, ABSref is the absorbance
of the aortic blood sample, and mreg is the mass
of the tissue sample. The analysis was performed with the use
of a commercial software program (Triton Technology Inc) on an IBM
computer.
Dogs were anesthetized with sodium pentobarbital (30
mg/kg IV). Hearts were removed through a left lateral thoracotomy and
immersed in cold Tyrode's solution equilibrated with 95%
O2–5% CO2 and containing
(in mmol/L) NaCl 131, NaHCO3 18, KCl 4,
CaCl2 2.7, MgCl2 0.5,
NaH2PO4 1.8, and dextrose
5.5. Endocardial, epicardial, and transmural strips (1.5x1.0x0.1
cm) from the anterolateral LV, one third of the distance from base to
apex, were filleted with surgical blades either parallel or
perpendicular (transmural) to the surface.6 7
Endocardial and epicardial preparations were obtained from positions
directly opposite one another. The same region was studied in each
animal, and it was 
3 cm from the pacemaker site. The preparations
were placed in a tissue bath, superfused with Tyrode's solution warmed
to 37°C (pH 7.35±0.05), and allowed to equilibrate at a cycle length
of 2000 ms. Solutions were pumped through the bath at a rate of 12
mL/min, with chamber content changed three times per minute. The bath
was connected to ground with a 3 mol/L KCl/Ag/AgCl junction.
Experiments were not started until preparations had fully recovered and
displayed stable electrophysiological
characteristics, which required 2 to 3 hours for endocardial and
transmural strips and 4 to 6 hours for epicardial strips. 
) coupled by Ag/AgCl junction to an amplifier with high-input
impedance and capacity neutralization (model KS-700, World Precision
Instruments). Action potentials and Vmax were
displayed on a digital storage oscilloscope (model 4074, Gould) and
stored in digitized form in a personal computer for consequent
analysis. Vmax was obtained by electronic
differentiation with an operational amplifier, and the system was
calibrated as previously described.8 For
stimulation of preparations, standard techniques were used to deliver
1- to 2-ms square-wave pulses 2.0x threshold through bipolar
polytetrafluoroethylene-coated silver
electrodes.8
Data are expressed as mean±SE. Significance of differences
compared with control were determined by use of a one-way ANOVA
followed by the Duncan multiple range test or Bonferroni's test when
the F value permitted.9 Statistical significance
was determined at a value of P<.05.
Results
Top
Abstract
Introduction
Methods
Results
Discussion
References
Studies of Myocardial Pressures and Flows
Initially, we tested whether the pacing protocol induced cardiac
failure or myocardial ischemia. Hemodynamic
recordings made from dogs at baseline and after 3 weeks of
cardiac pacing are summarized in Fig 1
.
In the baseline setting, resting HR averaged 86±5.1 bpm. Pacing was
initiated at 122±1.7 bpm, which resulted in no statistically
significant alteration in LV end-diastolic pressure,
positive or negative dP/dtmax, or MAP. After 3
weeks of ventricular pacing, resting HR on return to sinus
rhythm was 101±6.7 bpm (P>.05 versus rate before pacing).
Importantly, there were no hemodynamic signs of
decompensation of ventricular function as evidenced by the
absence of changes in any of the hemodynamic values
with the pacemaker turned on or off.
View larger version (28K):
[in a new window]
Figure 1. Hemodynamic studies of four dogs
during a control period (Pacer off), immediately after the onset of
ventricular pacing (Pacer on), on the last day of pacing (3
wk Pacer on), and 45 to 60 minutes after the pacemaker off was turned
off (3 wk Pacer off). Data are presented for HR, LV dP/dt, LV
end-diastolic pressure (LVedp), and MAP. No significant
change was seen in any variable (P>.05). . As shown, flow was little affected 45
minutes after the pacemaker was turned on. After 3 weeks of pacing,
there was a nonsignificant trend for flow to increase in all regions.
Finally, flow was not further affected 45 minutes after the pacemaker
was turned off. Similar data were obtained from the posterior LV wall
and from the septa of these same hearts (data not shown). Thus, these
data indicate that myocardial ischemia did not develop during
the pacing period and therefore did not contribute to the
electrophysiological changes observed in
these hearts.
View larger version (41K):
[in a new window]
Figure 2. LV anterior wall flow measured in four dogs by use
of microsphere techniques. Data are presented for
endocardium (endo), midmyocardium (mid), and epicardium
(epi) from the basal, mid, and apical LV anterior wall. Studies were
done during control (Pacer off), immediately after the onset of pacing
(Pacer on), on the last day of pacing (3 wk Pacer on), and 45 to 60
minutes after the pacemaker was turned off (3 wk Pacer off). No
significant changes were seen (P>.05).
The time course of typical ECG evolution during cardiac memory
induction is shown in Fig 3. Direction,
magnitude, and evolution of the T-wave change differ in any given lead,
limiting the value of any form of quantification of a single lead. In
Fig 4, data from the same dog as in Fig 3
are presented using a frontal plane vectorcardiogram. Here, a
distinctive pattern of counterclockwise T-wave vector rotation is
readily seen. The record in Fig 4 is representative
of the changes seen in 13 dogs, all of which were paced from the
anterior wall of the LV. The distinctive pacing-induced ECG pattern had
a wide QRS complex directed upward and rightward (mean frontal angle,
-140°) and a wide QRS-T angle (approaching 172°); see Table 1 and Figs 3 and 4B.
View larger version (40K):
[in a new window]
Figure 3. Cardiac memory evolution over 21 days of
ventricular pacing (VP). Depicted are leads I and aVR from
one dog during control, during VP, and at the end of 1 hour after
pacing was discontinued on days 7, 14, and 21. In both leads, evolution
of the T wave is such that it tracks the vector of the paced QRS
complex.
View larger version (18K):
[in a new window]
Figure 4. Frontal plane vectorcardiographic depiction of
cardiac memory from the same dog presented in Fig 3 
. A shows P,
QRS, and T-wave vectors in control, before the onset of pacing. B, The
QRS and T vectors during ventricular pacing. C, The T-wave
vector alone (note enlarged scale as well as dotted line indicating the
change in vector) during sinus rhythm in control (identical to A) and
on days 14 and 21. Note the increase in amplitude of the vector as well
as the shift in vector angle from that in control to one that
approximates the QRS vector during pacing (B). D, A record made
within 1 hour of returning to sinus rhythm on day 21.
View this table:
[in a new window]
Table 1. ECG Sequelae of 21 Days of Ventricular
Pacing (13 Dogs Paced From the Anterior Wall of the LV) and Table 1). In addition to the T-wave vector
angle, the magnitude of the vector increased as memory evolved (Table 1
and Fig 4). The correlation between the magnitude of the paced QRS
vector and the magnitude of the T-wave vector was 0.48
(P<.05). Finally, whereas the T-wave vector was altered
considerably, no consistent changes in the QT or
QTc intervals occurred (Table 1). ). Nonetheless, in all animals, the
T-wave vector in sinus rhythm was still influenced by the vector of the
paced QRS complex.
View larger version (20K):
[in a new window]
Figure 5. Relationship between pacing site and direction of
the sinus rhythm T-wave vector change in the frontal plane
vectorcardiograph of four dogs, paced respectively from the anterior
and posterior LV and the right ventricular free wall. The
broken line and arrow depict the paced QRS vector angle in each
experiment. The solid circles represent the vector angle and
amplitude in control (indicated as 0) and on individual days during
which pacing was discontinued for 1 hour and T-wave vector angle and
amplitude were recorded. Hence, in the first animal, (LV anterior)
recordings from days 0, 4, 7, 14, 18, and 21 are
presented. Note that in the animals paced from the LV sites,
the T-wave vector increases in amplitude through day 21, and its angle
approximates that of the paced QRS complex by day 4 to 7. With right
ventricular pacing, vector amplitude again increases and
the vector angle approaches that of the paced QRS complex but not with
the same precision as seen during LV pacing. and 7.
Fig 6 demonstrates the absolute changes in T-wave vector amplitude.
Note that a steady state is reached in 10 days. In Fig 7, the
overall change in T-wave vector (the difference between the control
T-wave vector and that in sinus rhythm after a period of
ventricular pacing) is plotted. This variable
incorporates both vector angle and vector amplitude and requires 21
days to attain a steady state.
View larger version (16K):
[in a new window]
Figure 6. Time course of sinus rhythm T-wave vector
amplitude expressed as the absolute T-wave vector amplitude in 16 dogs.
Note the attainment of a steady state in 10 days.
View larger version (16K):
[in a new window]
Figure 7. Time course of sinus rhythm T-wave vector change
expressed as percent of the value at 21 days in the same 16 dogs as in
Fig 6 . The T-wave vector on day 21 was expressed as 100% and that
before pacing as 0%. The value on any day between control and the end
of the experiment was expressed as percent of the day 21 to 0 distance
value. Using Fig 4C as a reference point, the day 21 value is 100%,
the control value is 0%, and the day 14 value would be expressed as a
percent of the distance. Note that when this method, which incorporates
amplitude and vector angle in the measurement, is used, a steady state
is reached by 3 weeks. shows the time course of memory
dissipation in two groups of dogs. The first group was paced for 21 to
25 days. A steady state of memory was maintained for 2 days, and then
recovery to levels near control was achieved by 7 days. The second
group was paced for 42 to 52 days. Here, a steady-state level of memory
persisted for 4 to 5 days. Memory remained for the entire period of
observation such that at 30 days, the T vector still differed from
control. This stresses the important role of accumulation in the
persistence of the memory process. It also demonstrates that whereas
the magnitude of the peak pacing-induced T-wave changes was similar in
both groups, these changes persisted longer in the group that was paced
longer.
View larger version (15K):
[in a new window]
Figure 8. Time course of memory dissipation
presented as the change in T-wave vector amplitude from the day
of cessation of pacing (0) through 30 days of recovery. Results from
two groups of animals are presented. The reference (R) T-wave
value on the right is that recorded before the onset of pacing. See
text for discussion. VP indicates ventricular pacing. ). Atrial pacing at the same rates and
durations comparable to those in the ventricular pacing
protocol induced no significant change in vector angle and amplitude
and no memory. Hence, atrial pacing, inducing a change in HR without a
change in activation pathway, did not induce cardiac memory.
View this table:
[in a new window]
Table 2. ECG Sequelae of 21 Days of Atrial Pacing (4 Dogs
Paced From Right Atrial Appendage) the evolution of a reduced magnitude of
T-wave vector change with cycloheximide treatment compared with
control. Note as well that the two cycloheximide runs were entirely
reproducible, as were the two control runs. The other
cycloheximide-treated dogs underwent a single drug trial preceded and
followed by control trials. Results were similar to the above: at day
21 of pacing after cycloheximide administration, the magnitude of the
T-wave vector change for the three animals was 63±8% of that in the
control run (P<.05).
View larger version (19K):
[in a new window]
Figure 9. Effects of administration of cycloheximide (CHM)
on evolution of cardiac memory in one dog. T-wave vector amplitude
change is used as the form of measurement here. Four consecutive
experiments were done. The sequence was control 1–CHM 1–control
2–CHM 2. For control 1, a standard protocol was used to induce memory.
Note the evolution of a steady state in T-wave amplitude by about day
14. Complete resolution of the memory was then permitted, after which
the animal was administered CHM and pacing was reinstituted. The result
was only partial induction of memory, as seen in a curve that was
displaced downward and to the right of control. Resolution of the
memory process was permitted, after which a second control experiment
was done. The results were nearly identical to the initial control.
After resolution of the memory, a second course of CHM was given, and
the pacing protocol again was repeated. In this case, the result was
essentially identical to that of the initial CHM experiment.
Cellular electrophysiological data for
epicardial, endocardial, and midmyocardial cell preparations from
control and long-term–memory animals are presented in Table 3. Cardiac memory resulted in
prolongation of the epicardial and endocardial APDs in the absence of
any change in the midmyocardial cells. These changes would
significantly alter the transmural gradient for repolarization and
would be expected to contribute to the T-wave changes that characterize
memory on the ECG. Also noted were small but significant reductions in
maximum diastolic potential (MDP) in all three cell types.
View this table:
[in a new window]
Table 3. Epicardial, Endocardial, and Midmyocardial Cell
Recordings From Control Dogs and Dogs LV Paced for 21 Days and
Manifesting T-Wave Memory . Multiple recordings from an
LV epicardial and endocardial slab of a control animal appear in Fig 10A; Fig 10B shows records from slabs comparably located from an
animal with cardiac memory. In both panels, the effects of changing
pacing cycle length on APD50 and
APD90 are demonstrated. In the setting of memory,
the prolongation of APD was such that the endocardial and epicardial
durations become more similar to one another than was the case in the
control condition. In Fig 10, this change in gradient is indicated by
the cross-hatched area inscribed between the epicardial and endocardial
APD50 and APD90 curves. In
the control animal, the epicardial area only minimally overlapped that
from the endocardium. In the memory setting, the two areas were largely
overlapping, reflecting a lesser gradient than in control.
View larger version (31K):
[in a new window]
Figure 10. APD to 50% (APD50) and 90%
(APD90) of repolarization measured from epicardium (Epi)
and endocardium (Endo) of LV in a control dog (A) and a dog with
cardiac memory (B). Preparations were paced over a range of cycle
lengths (CL; horizontal axis) from 2000 to 400 ms. Three to five
minutes was required for equilibration at each cycle length.
Recordings were made at 10 minutes. Three sets of preparations
were used per heart, and multiple impalements were made per
preparation. The shaded areas are those inscribed between
APD50 and APD90 of the epicardial and
endocardial preparations. In A at almost all cycle lengths, almost all
values for endocardium differ from those at corresponding cycle lengths
in epicardium (P<.05 for all). In B, statistical
significance (P<.05) was seen only for
APD90 at the two longest cycle lengths. See text for
discussion.
Discussion
Top
Abstract
Introduction
Methods
Results
Discussion
References
The first question posed in this study related to the possibility
that myocardial ischemia or failure might contribute to the
occurrence of memory. It has been important to discriminate between
memory and ischemia clinically because the ST-T–wave changes
of ischemia can be mimicked by those of
memory.1 2 It is equally important to determine
whether failure or ischemia occurs in this canine model,
because prior studies10 have shown that chronic
pacing of the canine heart at critical rates does in fact induce
failure and repolarization changes. Had this been the case in the
present study, the conclusions concerning cardiac memory would have
been invalidated. Our microsphere studies demonstrate that
myocardial flow was not diminished to any area of the ventricles.
Moreover, there was no change in hemodynamics,
indicating that congestive heart failure was not of concern. In another
study,11 we measured the capacitance of
ventricular myocytes from dogs with cardiac memory induced
by use of the same protocol as in the present study (H. Yu, PhD,
and I. Cohen, MD, PhD, preliminary data, 1997). Capacitance was 107±17
pF in the control group and 105±18 pF in the memory group, indicating
that myocyte size had not changed. Hence, it is reasonable to conclude
that the pacing protocol induced a change in electrical function
unassociated with deterioration of flow or contractile
performance or with myocyte hypertrophy. These
findings emphasize that the T-wave changes of cardiac memory are
clearly distinguishable in origin from those in a variety of
pathological conditions that induce both structural and
electrophysiological changes. However, they
do not rule out the possibility that cardiac memory is an early marker
of a pathological process that is not yet expressed at a macroscopic
level, that is reversible, and whose identification may suggest the
need for early intervention and prevention. demonstrate that both the accumulation and resolution
of cardiac memory can be quantified accurately. Earlier studies,
including our own,12 have quantified alterations
in T-wave amplitude or area as the prime descriptors of memory.
However, it is clear from data such as those shown in Fig 3 that the
use of single (or even several) ECG leads can give varying results
concerning the magnitude and extent of the memory process. In contrast,
vectorcardiography provides an opportunity to consider the T-wave
vector angle and amplitude concomitantly in a way that is readily and
consistently measurable and that is concordant with the
original descriptions of cardiac memory in human
subjects1 ; that is, the vector of the "memory T
wave" during sinus rhythm tracks the vector of the paced QRS complex.
The latter either completely or partially predicts the former, with the
magnitude and direction of change apparently reflecting the site of
origin of ventricular activation. In all instances, the
increase in vector amplitude is a consistent phenomenon,
reaching a steady state in 3 weeks. The resolution of the process
clearly is slower than its accumulation, as shown in Fig 8.
suggests that these alterations, once in place, persist for long
periods. It is important to stress, however, that we do not believe
Kv4.3 is the one channel uniquely involved in the memory process. The
fact that repolarization is prolonged in epicardium (which has a large
Ito118 19 20 ) and endocardium
(which has a small Ito120 ) and
is unchanged in midmyocardium (which has a prominent
Ito1) together suggests that additional channels
play an important role. Candidates include IK,
ICa,L, IK1, and
INa. A reduction in IK1
could also explain the reduction in membrane potential demonstrated in
the memory setting (Table 3). Another possible explanation for the
reduction in membrane potential is the finding by our associates (J.
Gao, PhD, and I. Cohen, MD, PhD, preliminary results, 1997) that Na/K
pump current is reduced in the setting of memory. that a change in sinus rate alone is insufficient to
significantly alter the T-wave vector. When Tables 1 and 2 are
compared, it is clear that the changes induced by an altered pacing
rate plus an altered activation pathway significantly exceed those
resulting from a rate change alone. Therefore, we conclude that the
activation pathway is critical to the expression of long-term
memory. , both the effect of cycloheximide and the
recovery from its effect were reproducibly demonstrable in the same
animal. Therefore, it would appear that new protein synthesis does play
a role in the evolution of long-term memory. The caveat here is the
nonselective nature of the activity of cycloheximide: it inhibits
synthesis of a broad spectrum of proteins.
Selected Abbreviations and Acronyms
APD
=
action potential duration
HR
=
heart rate
LV
=
left ventricle, left ventricular
MAP
=
mean arterial pressure
Acknowledgments
This study was supported by United States Public Health
Service–National Heart, Lung, and Blood Institute grants (HL-28958,
HL-53956, and HL-51885) and by the Wild Wings Foundation. The authors
thank Dr Natalia Egorova for assisting in the performance of
tests of these studies and Eileen Franey for her careful attention to
the preparation of the manuscript.
References
Top
Abstract
Introduction
Methods
Results
Discussion
References
This article has been cited by other articles:
 |
|
 |
|
  D. Jeyaraj, L. D. Wilson, J. Zhong, C. Flask, J. E. Saffitz, I. Deschenes, X. Yu, and D. S. Rosenbaum Mechanoelectrical Feedback as Novel Mechanism of Cardiac Electrical Remodeling Circulation, June 26, 2007; 115(25): 3145 - 3155. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Coronel, T. Opthof, A. N. Plotnikov, F. J.G. Wilms-Schopman, I. N. Shlapakova, P. Danilo Jr., E. A. Sosunov, E. P. Anyukhovsky, M. J. Janse, and M. R. Rosen Long-term cardiac memory in canine heart is associated with the evolution of a transmural repolarization gradient Cardiovasc Res, June 1, 2007; 74(3): 416 - 425. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S.-k. Wei, A. M. Ruknudin, M. Shou, J. M. McCurley, S. U. Hanlon, E. Elgin, D. H. Schulze, and M. C.P. Haigney Muscarinic Modulation of the Sodium-Calcium Exchanger in Heart Failure Circulation, March 13, 2007; 115(10): 1225 - 1233. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. N. Obreztchikova, K. W. Patberg, A. N. Plotnikov, N. Ozgen, I. N. Shlapakova, A. V. Rybin, E. A. Sosunov, P. Danilo Jr., E. P. Anyukhovsky, R. B. Robinson, et al. IKr contributes to the altered ventricular repolarization that determines long-term cardiac memory Cardiovasc Res, July 1, 2006; 71(1): 88 - 96. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Rosen and I. S. Cohen Cardiac memory ... new insights into molecular mechanisms J. Physiol., January 15, 2006; 570(2): 209 - 218. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. W. Patberg, M. N. Obreztchikova, S. F. Giardina, A. J. Symes, A. N. Plotnikov, J. Qu, P. Chandra, D. McKinnon, S. R. Liou, A. V. Rybin, et al. The cAMP response element binding protein modulates expression of the transient outward current: Implications for cardiac memory Cardiovasc Res, November 1, 2005; 68(2): 259 - 267. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Janse, E. A. Sosunov, R. Coronel, T. Opthof, E. P. Anyukhovsky, J. M.T. de Bakker, A. N. Plotnikov, I. N. Shlapakova, P. Danilo Jr, J. G.P. Tijssen, et al. Repolarization Gradients in the Canine Left Ventricle Before and After Induction of Short-Term Cardiac Memory Circulation, September 20, 2005; 112(12): 1711 - 1718. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Madias, A. Shvilkin, K. K.L. Ho, M. E. Josephson, and M. R. Rosen Letter Regarding Article by Shvilkin et al, "T-Vector Direction Differentiates Postpacing From Ischemic T Wave Inversion in Precordial Leads" * Response Circulation, July 19, 2005; 112(3): e52 - e52. [Full Text] [PDF]
|
|
|
|
|
|
D. D. Spragg, F. G. Akar, R. H. Helm, R. S. Tunin, G. F. Tomaselli, and D. A. Kass Abnormal conduction and repolarization in late-activated myocardium of dyssynchronously contracting hearts Cardiovasc Res, July 1, 2005; 67(1): 77 - 86. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Shvilkin, K. K.L. Ho, M. R. Rosen, and M. E. Josephson T-Vector Direction Differentiates Postpacing From Ischemic T-Wave Inversion in Precordial Leads Circulation, March 1, 2005; 111(8): 969 - 974. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Nahlawi, M. Waligora, S. M. Spies, R. O. Bonow, A. H. Kadish, and J. J. Goldberger Left ventricular function during and after right ventricular pacing J. Am. Coll. Cardiol., November 2, 2004; 44(9): 1883 - 1888. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. N. Plotnikov, E. A. Sosunov, K. W. Patberg, E. P. Anyukhovsky, R. Z. Gainullin, I. N. Shlapakova, G. Krishnamurthy, P. Danilo Jr, and M. R. Rosen Cardiac Memory Evolves With Age in Association With Development of the Transient Outward Current Circulation, August 3, 2004; 110(5): 489 - 495. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Libbus, X. Wan, and D. S. Rosenbaum Electrotonic load triggers remodeling of repolarizing current Ito in ventricle Am J Physiol Heart Circ Physiol, May 1, 2004; 286(5): H1901 - H1909. [Abstract] [Full Text] [PDF]
|
|